Livv
Décisions

EC, June 27, 2007, No M.4439

COMMISSION OF THE EUROPEAN COMMUNITIES

Decision

Ryanair/Aer Lingus

EC n° M.4439

27 juin 2007

THE COMMISSION OF THE EUROPEAN COMMUNITIES,

Having regard to the treaty establishing the European Community, Having regard to the Agreement on the European Economic Area, and in particular Article 57 thereof, Having regard to Council Regulation (EC) No 139-2004 of 20 January 2004 on the control of concentrations between undertakings (1), and in particular Article 8(3) thereof, Having regard to the Commission's decision of 20 December 2006 to initiate proceedings in this case, Having regard to the opinion of the Advisory Committee on Concentrations (2), Having regard to the final report of the Hearing Officer in this case (3), Whereas:

1. INTRODUCTION

1. On 30 October 2006, the Commission received a notification of a proposed concentration pursuant to Article 4 of Regulation (EC) No 139-2004 ("the Merger Regulation") by which the undertaking Ryanair Holdings Plc ("Ryanair", Ireland), acquires within the meaning of Article 3(1)(b) of the Merger Regulation, control of the whole of the undertaking Aer Lingus Group Plc ("Aer Lingus", Ireland), by way of public bid announced on 23 October 2006.

2. After its initial examination of the notification, the Commission concluded that the concentration fell within the scope of the Merger Regulation and, even taking into account commitments offered by Ryanair on 19 November 2006 as modified on 14 December 2006, raised serious doubts as to its compatibility with the common market and with the EEA Agreement. It therefore decided on 20 December 2006 to initiate proceedings pursuant to Article 6(1)(c) of the Merger Regulation.

3. In the interest of the investigation and with the agreement of Ryanair, the deadline for the decision in this case was extended by 20 working days on 22 February 2007, pursuant to Article 10(3), second subparagraph, of the Merger Regulation.

4. On 27 March 2007, a Statement of Objections was sent to Ryanair pursuant to Article 18 of the Merger Regulation.

5. On 17 April and 3 May 2007, Ryanair offered further commitments with a view to rendering the proposed concentration compatible with the common market.

6. The Advisory Committee discussed a draft of this Decision on 11 June 2007.

2. THE PARTIES

7. Ryanair is an airline offering point-to-point scheduled air transport services on more than 400 routes across 24 European countries. Ryanair operates more than 75 routes between Ireland (mainly Dublin, but also Shannon, Cork, Kerry and Knock) and other European countries. The company has a fleet of 120 aircraft (with firm order of 161 new aircraft to be delivered over the next six years) (4) and currently 20 bases across Europe, the most important ones being London-Stansted and Dublin. Ryanair is not a member of an airline alliance and does not have interlining agreements (5) with any other airline. It is an Irish public limited company listed on the Dublin, London and New York (NASDAQ) stock exchanges.

8. Aer Lingus is an Irish-based airline. As a publicly listed company, Aer Lingus offers essentially point-to point scheduled air transport services on more than 70 routes connecting the Irish airports of Dublin, Shannon and Cork with a number of European destinations. In addition, Aer Lingus offers long-haul flights, mainly to the United States, and cargo transport services. Aer Lingus is based principally at Dublin Airport (and to a smaller extent in Cork and Shannon) with a total fleet of currently 28 short-haul and 7 long-haul aircraft (and a further 4 short-haul and 2 long-haul aircraft on firm orders to be delivered by the end of 2007). Aer Lingus was previously a member of the OneWorld alliance but has terminated the membership and left the alliance as of April 2007.

9. Ryanair and Aer Lingus are referred to together as "the Merging Parties" in this Decision.

3. CONCENTRATION

10. The proposed transaction concerns an acquisition of sole control by Ryanair of Aer Lingus by way of a public bid for all outstanding shares not already acquired. Ryanair started to acquire a substantial number of shares of Aer Lingus on 27 September 2006. Ryanair acquired 43.7 million shares on 27 September 2006, 25.05 million shares on 28 September 2006, 8.3 million shares on 29 September 2006, 7.775 million shares on 4 October 2006 and 16.56 million shares on 5 October 2006. These shares amounted to 19.16% of the share capital of Aer Lingus.

11. On 5 October Ryanair announced a public bid for the entire share capital of Aer Lingus. The offer document was sent to Aer Lingus shareholders on 23 October 2006 with a deadline for acceptance of 13 November 2006, which was subsequently extended by Ryanair first until 4 December 2006 and then until 22 December 2006. During the bid period, Ryanair acquired further shares of Aer Lingus and, by 28 November 2006, held 25.17% of the share capital in Aer Lingus. Ryanair has confirmed that its acquisition of shares in Aer Lingus since 26 September 2006 was part of its plan to acquire control of Aer Lingus (6).

12. As Ryanair acquired the first 19% of the share capital of Aer Lingus within a period of less than 10 days before launching the public bid, and the further 6% shortly thereafter, and in view of Ryanair's explanations of the economic purpose it pursued at the time it concluded the transactions, the entire operation comprising the acquisition of shares before and during the public bid period as well as the public bid itself is considered to constitute a single concentration within the meaning of Article 3 of the Merger Regulation.

4. COMMUNITY DIMENSION

13. In its decision of 20 December 2006 pursuant to Article 6(1)(c) of the Merger Regulation the Commission concluded that the notified concentration has a Community dimension pursuant to Article 1(3) of the Merger Regulation. For the sake of clarity, the arguments used in that decision are recalled in this section..

14. The concentration does not have a Community dimension within the meaning of Article 1(2) of the Merger Regulation since the combined aggregate worldwide turnover of Ryanair and Aer Lingus is less than EUR 5 000 million (7). It therefore needs to be assessed whether the concentration has a Community dimension within the meaning of Article 1(3) of the Merger Regulation.

15. The undertakings concerned have a combined aggregate world-wide turnover of more than EUR 2 500 million (8) and both Ryanair and Aer Lingus have a Community-wide turnover in excess of EUR 100 million (9). The conditions of Article 1(3)(a) and (d) of the Merger Regulation are therefore met. Furthermore, it is clear that Ryanair and Aer Lingus do not achieve more than two-thirds of their aggregate Community-wide turnover within one and the same Member State. Whether or not both Ryanair and Aer Lingus achieve a combined aggregate turnover of more than EUR 100 million in at least three Member States and each of them achieves at least EUR 25 million in these Member States, as required under Article 1 (3) (b) and (c) of the Merger Regulation, depends on the geographical allocation of the turnover of these undertakings.

16. Article 5(1) of the Merger Regulation provides, with regard to geographic allocation of turnover: "Turnover, in the Community or in a Member State, shall comprise products sold and services provided to undertakings or consumers, in the Community or in that Member State as the case may be."

17. The aim of the turnover-based thresholds is to provide a simple and objective mechanism that can be easily handled by the companies involved in a merger in order to determine if their transaction has a Community dimension and is therefore notifiable (10). At the same time, the calculation of turnover should reflect as accurately as possible the economic strength of the undertakings involved in a transaction in the Member State concerned (11). The Notice on the calculation of turnover presumes that, in general, the geographical allocation of turnover is determined by the location of the customer at the time of the transaction (paragraphs 45 et seq.). In particular, paragraph 46 of the Notice presumes that this will correspond, for practical purposes, to the place of provision of services.

18. In previous airline cases (12), the Commission identified the following three possibilities for geographical allocation of turnover:

(1) to allocate revenue from individual routes to the country of destination (this option was specifically mentioned in some cases for transatlantic routes (13) and was abandoned in the later decisions not involving transatlantic routes such as M.616 - Swissair /Sabena (II), M.857 - British Airways/Air Liberté and others; therefore it is listed only for completeness sake but will not be discussed further below);

(2) to allocate the turnover in a 50%/50% ratio to the country of origin and the country of final destination so as to take into account the cross border character of the service provided ("50/50 method");

(3) to allocate the turnover to the country where the ticket sale occurred (referred to also as "point of sale method").

19. As previous transactions had a Community dimension under all possible methods, the Commission left open which one would be the most appropriate one. It must be noted that most of the decisions pre-dated the Notice on the calculation of turnover.

20. Ryanair has notified the merger to the Commission in the belief that it has a Community dimension on the basis of the so called 50/50 methodology. It stresses that this methodology was used by the Commission in a number of previous decisions and that it is the appropriate method to be used in this case, given the cross-border character of the routes where the activities of Ryanair and Aer Lingus may overlap. Further, they argue that it is more in line with the industry practice as Ryanair itself monitors the national turnover on a 50/50 basis for its accounting and operational purposes. Ryanair also considers this methodology as sufficiently simple and easy to use without necessity of complex calculations.

21. Aer Lingus, on the other hand, submits that the merger does not have a Community dimension on the basis that the turnover is allocated to the place of departure of the customer. It argues that the general rule for geographic allocation of turnover according to Article 5(1) of the Merger Regulation and the Notice on the calculation of turnover is to allocate the turnover to that Member States where the customer is located. Taking into account the fact that a vast majority of tickets is sold by Aer Lingus via the Internet (14), Aer Lingus is not able to identify the location of the customer at the moment of the sale of the ticket. Therefore, as it is not practically possible to use the location of customer at the time of sale, Aer Lingus rather allocates the revenue to the Member State where the place of departure of the journey is located as it reflects the location of the customer at the moment at which the provision of the service commences. Aer Lingus also refers in this respect to paragraph 170 of the Draft Consolidated Jurisdictional Notice (15). Aer Lingus further argues that under this place of departure methodology, return tickets are to be treated as having only one place of departure, namely the place where the first leg of the journey started. Although Aer Lingus does not sell traditional return tickets in which both legs of a journey are combined in a single ticket for a specific price, they argue that the return ticket is a service sold together with the outbound ticket in one place and therefore the revenue from both the outbound and inbound leg of the journey should be allocated to the original place of departure and not split for the two legs of the journey. According to the turnover data submitted under this methodology, Aer Lingus would have turnover in excess of EUR 25 million only in two Member States (16) and therefore the transaction would not have a Community dimension.

22. The Commission observes that neither Ryanair nor Aer Lingus is relying on the point of sale methodology mentioned in past cases. This method is also increasingly difficult to apply in the air transport sector, given the constantly growing shares of direct Internet sales (in particular in case of point-to-point, low-cost airlines such as Ryanair and, to a large extent, Aer Lingus with the majority of tickets sold over the Internet (17)). The absence of a physical transaction at a brick-and-mortar airline counter or travel agent makes it more difficult to physically locate customers purchasing tickets. Further, as the customer could buy the ticket on the Internet from practically any place in the world, this information would, even if available, not necessarily support the assumption of the Notice on the calculation of turnover that the location of customers when purchasing services will normally reflect where the parties to the transaction provide their services and the economic strength of the parties in a specific Member State. Both Aer Lingus and Ryanair have confirmed that they are not in a position to allocate their revenue on the basis of the location of the customer as they do not track the addresses or locations of customers at the time of sale. Therefore, the point of sale principle cannot serve as a methodology in the context of the current transaction.

23. The 50/50 method proposed by Ryanair has been accepted as a possible approach in a number of past cases and seems, as noted in M.157 - Air France/Sabena, close to the spirit of the Merger Regulation since it takes into account the two places between which the service is actually provided, reflecting the cross-border character of the service in question. This method also provides a simple and clear criterion, and in this respect appears therefore in line with the very purpose of the system of thresholds of the Merger Regulation, that is, to provide a simple and efficient method to determine the authority that is competent to review a merger. (18) Finally, the 50/50 method appears not to be contrary to the Notice on the calculation of turnover, to the extent that the basic principle of the Notice, that is to say, the location of customer at the time of the transaction, cannot be applied in a meaningful way in the case at hand, taking into account the difficulty of establishing that location and the very specific nature of scheduled air transport services.

24. Aer Lingus argues that considering the nature of the service provided, the relevant methodology in the air transport industry is to allocate the revenue from a flight to the Member State where the place of departure of the flight is located (place of departure methodology). It submits that this methodology reflects the wording of Article 5 of the Merger Regulation, does not contradict the Notice on the calculation of turnover and is proposed as a possible appropriate methodology by the Commission in the recently published Draft Consolidated Jurisdictional Notice (19).

25. As indicated above in paragraph 16, Article 5(1) of the Merger Regulation refers to the Member State where the service is provided. The place of departure seems to be a good proxy for determining where the service is provided as it is clear that at the commencement of the provision of the service, the customer is indeed located at the place of departure. In circumstances where the conditions of purchase are unlikely to be influenced by the place at which the customer conducts the transaction, the allocation of the turnover to the place of departure may also be said to reflect, in principle, where the airlines compete for customers to provide their services and corresponds to the economic strength of the airline in a certain Member State. (20) Further, the place of departure is simple to identify for each journey and customer, which is an important factor for the determination of jurisdiction, thereby ensuring legal certainty. This is particularly important in circumstances where the location of the customer at the time of sale cannot be identified and where satisfaction of certain of the subsidiary criteria identified in paragraph 46 of the Notice on calculation of turnover (where a deal was made, where the turnover for the supplier in question was generated) is equally difficult to determine.

26. For these reasons, the argument that the place of departure principle may be an appropriate basis for geographic allocation of turnover in the case of air transport would appear to be in line with Article 5(1) of the Merger Regulation and with the underlying approach of the Notice on calculation of turnover, having regard also to the evolution of business practices in the airline sector and, in particular, in the case of the two undertakings concerned by the concentration in this case (21).

27. The application of the place of departure principle would also raise questions about how the tickets for roundtrips bought at the same time should be treated. Such tickets, on the one hand, could be treated as "one service" with only one place of departure; then the total revenue from the return ticket should be allocated to one country, namely that where the place of departure of the original outbound flight is located. On the other hand, such tickets could be split and treated separately as two flights from different places of departure, with the distinct revenue from each leg of the journey being allocated to the country from which that leg departed.

28. In order to resolve this issue, the practice of Aer Lingus and Ryanair needs to be examined to determine whether, if the place-of-departure methodology were to be applied, the return tickets sold by these two airlines could be regarded as constituting a single service or rather as two separate services, each commencing at a different place of departure. The examination of this issue shows that neither Ryanair nor Aer Lingus sells "traditional" return tickets, whereby a return flight "bundle" is more advantageous than two one-way flights. On the contrary, they both sell one-way tickets, together or separately as the case may be, and simply sum up their individual prices in the case of simultaneous booking of tickets comprising a round trip, without any price or other advantages for the customer buying the tickets for such a round trip. Therefore, these are not "traditional" return tickets but rather two one-way tickets bought simultaneously in one transaction for a round trip. The customer, however, always has the possibility to buy such tickets in two different transactions without being penalised, including the possibility to buy the outbound ticket with one airline and the inbound ticket with another airline depending on the most advantageous price for each leg of the roundtrip. For the purposes of the turnoverallocation process, it could therefore be assumed that, in principle, the airlines compete with each other for each leg of the journey and not on the sale of round trip tickets. This argument would be supported by the fact that the customer is in a position, in the case of the predominant Internet sales, easily to compare the prices of the individual one-way flights and decide to fly with two different airlines if this is more attractive.

29. From the above, it can be argued that the airlines in this case provide two connected but distinct services, one provided at the place of departure of the outbound flight and the other at the place of departure of the inbound flight. In this situation, to the extent that it is concluded that the place of departure methodology should be applied, it would appear most appropriate, at least given the nature of the business activities of the undertakings concerned by the present concentration, to split the two one-way flights of a round trip. Therefore, the place-of-departure methodology as proposed by Aer Lingus must be rejected insofar as it applies the rule foreseen for return tickets also to the (simultaneous) purchase of two single tickets.

30. Therefore, it is concluded that, of the possible alternative methodologies for geographic allocation of turnover in respect of transactions for which the location of the customer at the time of purchase cannot be identified and cannot affect the conditions of such purchase, in particular the 50/50 methodology, as well as the methodology based on place of departure with splitting the two one-way flights of a round trip tickets bought at the same time, seem to be the most appropriate, especially in the case of point-to-point airlines such as Ryanair or Aer Lingus (22).

31. Under the place of departure methodology with splitting the two one-way flights of a round trip, the turnover of both Ryanair and Aer Lingus exceeds EUR 25 million at least in three Member States (Ireland, the United Kingdom and Spain) whereas the combined turnover of these two airlines in those three Member States exceeds EUR 100 million (23). Further, the relevant thresholds would also be exceeded in these three Member States if the 50/50 methodology were to be used (24). It is therefore not necessary in this case to decide which of these two possible methodologies is the more appropriate.

32. The notified operation therefore has a Community dimension pursuant to Article 1(3) of the Merger Regulation.

5. INVESTIGATION OF THE CASE

33. Given the complexity of the case, the Commission has sought to make use of all available means of investigation pursuant to Article 11 of the Merger Regulation. It has not only analysed questionnaires which were sent inter alia to competing scheduled airlines, charter airlines, airports and (corporate) customers but also other written and oral contacts with these and other third parties such as slot coordination authorities, civil aviation authorities and transport authorities.

34. Further, in view of the various economic and econometric submissions, in particular by the Merging Parties, the Commission has decided to address these submissions to the extent possible within the constraints (in particular timing) of a merger investigation. The Commission has reviewed the submitted data by generating descriptive statistics to better understand the factors affecting competition in the affected markets. It then conducted two sets of regression analysis, one based on comparisons of fares across routes (the "crosssection" analysis, Annex IV) and an assessment of price variations over time and across routes (the "fixed-effects" analysis, Annex IV). The objective in both cases was to identify the level of competitive constraints exercised between the Merging Parties as well as by their competitors. Further, a price correlation analysis for individual airport pairs and city pairs provided input in particular for the market definition section (see for further details Annex IV).

35. The Commission also had to cope with a specific problem concerning the gathering of relevant evidence in this case: although the transaction is likely to have an effect on more than 14 million passengers travelling with the Merging Parties' airlines, these are largely individual customers (25) that could not be contacted by the Commission by way of the classic investigative techniques (questionnaires, telephone interviews) in a meaningful way.

36. Although the Commission tried to have a representative view of the affected customers by contacting large corporations (including Aer Lingus' largest customers) in Phase I of the investigation, the responses showed that the replies of these "business customers" could only to a limited extent be regarded as representative of the preferences of the customers affected by this merger of two "low-frills" airlines. Indeed, corporate customers are by nature business customers and are likely to be more time-sensitive and less price-sensitive than the average "low-frills" customer (26). Due to their specific needs, corporate customers are also less appropriate as a source of information on how and on the basis of which parameters Ryanair's and Aer Lingus' "typical" (low-frills) customers choose an airline. The Commission therefore assigned an independent consultant to carry out a Customer Survey at Dublin Airport to obtain a representative sample of responses from customers who departed from Dublin. The questions were sent to Ryanair and Aer Lingus for consultation before the Customer Survey was carried out during ten days in the month of February. The results of the Customer Survey cover 12 of the 35 overlap routes (that is to say, routes on which Aer Lingus and Ryanair both provide services), representing different characteristic types of all the overlap routes. Details of the survey and tables which contain the main results are set out in Annex I.

37. Ryanair has, notably in its response to the Statement of Objections, criticised the Commission's method of fact-finding in the present case. It has not only questioned the results of the Customer Survey (27), but is also of the view that the Commission quoted "selectively" from the results of the market investigation.

38. In this context it is important to stress that the Commission's assessment of the competitive impact of this transaction involves a complex legal and economic analysis, the result of which is not based only on certain parts of the collected evidence, but on the totality of all the available evidence. The fact that single pieces of evidence (answers to questions, result of econometric studies (28)) may not support a certain conclusion, cannot as such put into question the Commission's assessment, since the Commission cannot base its decision on one single piece of evidence, but must collect as many pieces of evidence as possible, analyse all available facts and opinions and weigh all the available evidence when deciding on the compatibility of a transaction with the common market.

39. In particular with a view to written questionnaires, it is important to note that the Commission's market investigation is by no means an "opinion poll". For instance, the fact that the majority of answering third parties may have a certain opinion can only be an indication for the Commission's own investigation. Nor is the Commission required to carry out a "representative" customer poll within the limited timeframe of a merger procedure and under the constraints of often narrow markets with third parties who are often reluctant to provide an answer (29). Nor would it be appropriate to assume that the answers to the Commission's questionnaires can always be regarded as an objective and well-reflected response to the respective question. As Ryanair itself notes (30), the knowledge of the respondent on the subject matter can vary, he can have misunderstood the question, he can be more or less representative, and the answer can also be "biased" in order to influence the Commission's decision-making process (31). Like in any other merger investigation, the Commission has therefore carefully analysed, interpreted and weighed all answers to the market investigation. The quotes from customers selected by Ryanair in its response to the Statement of Objections are indeed part of the large number of statements from third parties in this procedure. However, the Commission believes that the few selected quotes are neither representative of the majority of the answering customers, nor do they give a meaningful picture of the result of the Commission's market investigation.

40. In this respect it is important to stress that the Commission's market investigation is an ongoing process, in the course of which the Commission usually refines and narrows down the issues it analyses (32) and uses the opportunity to clarify unclear and contradictory answers with third parties (33). In the present case, for example, the Commission has tried to take into account the views of as many actual and potential competitors as possible. The Commission has therefore not only sent various written questionnaires to these competitors, but has also carried out detailed interviews with the most important competitors in order to clarify unclear answers and contradictions from the written responses and to learn more about some key facts of the case (for example, entry barriers). The minutes that were taken of these interviews were sent to the interviewees in order to give them the opportunity to correct them (and to delete business secrets) before they were added to the case file (34).

6. RELEVANT MARKETS

6.1. Introduction

41. Ryanair's and Aer Lingus' activities overlap in the field of supply of scheduled passenger air transport services within the European Economic Area. A large number of airlines (35) are currently offering such services within the EEA. However, these airlines form a heterogeneous group, with significant differences between each airline. Differences between carriers relate mainly to (i) the operating model of the respective airline (in principle hub & spoke or "network" carriers as opposed to point-to-point models) and (ii) the level of service that is offered to passengers (full service as opposed to low-frills model).

6.1.1. Operating model

42. Certain carriers, usually referred to as "network carriers", operate a so-called "hub-and-spoke" system. Network carriers direct ("feed") traffic into their specific hub airport(s), from where they disperse the passengers via connections to numerous other destinations (often long haul destinations). By directing passengers through their hubs, network carriers can ensure a connection from every origin to every destination ("network") with a limited number of routes (36) and can fill their aircraft more easily. On the other hand, a hub-and-spoke network requires a significant level of co-ordination and harmonisation of schedules of the "feeder" and the respective (long-haul) services at the hub airport. Hub-and-spoke operations are characterised by a succession throughout the day of waves of incoming aircrafts (which bring connecting passengers (and cargo) onto the hub airport), followed by waves of departing aircrafts with sufficient time in between the two to allow passengers (and cargo) to connect. Between an incoming and an outgoing wave, there are usually few departures or arrivals. This makes the hub-and-spoke airline operating model more inflexible and complex than a simple point-to-point connection, where no connection has to be taken into account. Network carriers normally refrain from entering routes which are not connected to their hubs (for example, point-to-point services without connection to their hub) (37). Network carriers, often former national "flag carriers", typically hold a relatively large slot portfolio at their main hub or "home airport".

43. In contrast, other airlines, usually referred to as "point-to-point" carriers, concentrate on providing point-to-point services. In a point-to-point operation, each individual route is in principle operated independently from the others. The point-to-point model simplifies significantly the airline operation. In general, point-to-point airlines are more flexible as regards maximisation of their aircraft utilisation or fixing schedules and destinations. This is because point-to-point airlines do not need to harmonise their schedules and to take into account connecting flights within the network; they are also more flexible to choose new destinations, without having to assess the potential contribution in terms of feeder traffic of a new city and without considering to what extent serving a particular destination fits into the rest of the network. Planning and operating routes on a point-to-point basis is not incompatible, however, with allowing passengers to connect, where such connections are possible and suit passengers' needs.

44. Although point-to-point carriers are not "hub" carriers, since they do not operate according to the hub-and-spoke system, most point-to-point carriers also concentrate their traffic at certain airports, the so-called base airports or "bases". The concept of a "base" has to be distinguished from the concept of a "hub". While the term "hub" relates to the "hub-and-spoke" system and the system of connecting "feeding" traffic into a network, the term "base" is used to characterise airports on which airlines base their aircraft and on which they concentrate their operations, offering mainly flights from and to these "base" airports. The concept of a base will be explained below in greater detail (38).

45. The actual operating model used by the airline may differ also in terms of the geographic area that is covered by the operations and the type of aircraft used. While network carriers often cover a world-wide or at least trans-continental network of destinations (39), European point-to-point carriers tend to operate on a regional, national or European-wide basis. However, differences in particular between point-to-point carriers can be significant. While some point-to-point carriers (regional carriers) offer two or three routes with one single plane, others (such as Ryanair or easyJet) cover almost every European country and offer hundreds of routes through a number of bases. Smaller regional and larger point-topoint carriers can also operate with significantly different aircraft: While the regional point-to-point carriers could operate with turboprop or with smaller jet aircrafts with only 20-100 seats, larger point-to-point carriers may operate with aircraft that can transport up to 200 passengers on short haul routes. The differences between the smaller regional and larger point-to-point carriers may not justify defining two separate markets, but the Commission will take into account the differences between the models in its competitive Assessment (40).

46. There is a high degree of differentiation between the different airlines as concerns their operating model. While some airlines can clearly be characterised as network carriers (for example, KLM, Lufthansa, Air France or British Airways) or as true point-to-point airlines (for example, Ryanair, easyJet or most regional carriers), others have opted for an intermediate Model (41).

47. In recent years, Aer Lingus has transformed its European operation into a point-to-point operation and marketed its services on European routes as such. As part of this strategy, Aer Lingus decided to leave the global airline alliance One World as of 1 April 2007. In doing so however, Aer Lingus has maintained the connectivity of its services from Dublin with some partner airlines at a few of Europe's main airports (London Heathrow, Amsterdam Schiphol and Frankfurt am Main in particular). However, a large majority of Aer Lingus' passengers also travel point-to-point on these routes (42).

48. Ryanair is a pure point-to-point carrier. This is because, further to providing and marketing their passenger air transport services as point-to-point, they actively discourage passengers from connecting or interlining (43).

6.1.2. Type of airline service offered

49. In addition to the differences in the operating model, airlines can be distinguished according to the level of services they offer to passengers. Indeed, airlines do not only compete on routes and prices, but also on a number of qualitative features, in particular the level of services they offer to their customers. Airlines have traditionally not only offered "basic" services ("frills") to their customers (for example, transport by air between two points), but also ancillary services such as complimentary drinks and food, complimentary seat reservation, offering of different cabin classes, complimentary luggage transport, complimentary newspapers etc. Many airlines, however, have changed their "full service" model in the wake of the appearance of various so-called "low frills" or "no frills" airlines. These "low frills" airlines, such as Ryanair, were able to offer significantly cheaper fares than the established carriers, at least partly because they reduced the level of (complimentary) services drastically. Today, many established former "flag carriers" have reduced their free service offerings and introduced some elements of the "low frills" model. The distinction between "full-service" carriers (that is to say, carriers which offer a higher/upmarket level of service) and "low-frills" carriers (which, as the name suggests, offer a rudimentary level of service and compete mainly on the price of the flight) therefore characterises only the extreme ends of possible service levels. Most airlines offer an individual combination of services, without clearly falling into the category of a "no frills" or a "full service" carrier. The aspects, according to which airlines try to distinguish themselves more or less from their competitors include booking services (for example, seat reservation, on-line check in, last-minute bookings, differentiation in services and prices for different types of customers (44) (for example, unrestricted that is to say, flexible tickets, restricted roundtrip tickets), customer loyalty schemes such as frequent flyer programs ("FFPs")), services offered on the ground (for example, free luggage handling, availability of a business lounge, free newspapers), in the air (for example, availability of premium cabin classes, free drinks and food, number of crew, quality of the interior), or the destination airport (for example, "primary" airports close to city centres or more remote "secondary" airports).

50. Network carriers which operate a hub-and-spoke model normally fall into the category of "full-service" carriers. Similarly, low-frills carriers tend to be point-to-point carriers. However, point-to-point carriers can also be more low-frills or more full-service. For example Ryanair, easyJet and Aer Lingus (on its European services (45)) have, with some variances between them, the typical attributes of low-frills point-to-point carriers (46), whereas carriers like Aer Arann or CityJet (on its Dublin - London City service (47)) are more "full-service" Operators (48).

51. For the purpose of the assessment of the proposed transaction it is not appropriate to define separate markets according to the type of airline operations or the level of service offered. However, all of these differences between carriers are relevant when assessing the competitive impact of the proposed transaction. This is because the constraints that a competitor will exercise on the merged entity are determined by its business model and thus its incentives and/or its possibilities to compete effectively with the merged entity.

6.1.3. Ryanair is not in a market of its own

52. Ryanair has argued (49) that it is not constrained by any competitor but, due to its business model, rather by the overall price sensitivity of its customer base. It argued that its business model is to target discretionary passengers whose alternative is not so much to fly with another airline as not to fly at all, claiming that only Ryanair would be able to target such a customer base due to its low cost base.

53. This would, however, imply that Ryanair effectively operates in a different market than other airlines. The Commission has analysed the arguments put forward by Ryanair and found that Ryanair does react to competition on the routes from and to Ireland (50). The evidence provided by Ryanair therefore does not allow the Commission to conclude that Ryanair acts independently from other competitors. Hence Ryanair cannot be considered to belong to a market of its own but rather competes as other carriers within differentiated markets for scheduled point-to-point passenger air transport services.

6.2. Definition of individual markets for each route (origin and destination) versus definition of a market for short-haul flights out of Dublin

54. Ryanair submits that the relevant product market is "point-to-point scheduled air transport passenger services (51)" whereby each route between a point of origin and a point of destination should be defined as a separate market ("Origin & Destination"- or "O&D-approach").

55. This is in line with the Commission's findings in previous cases involving scheduled air transport services (52). The Court of Justice and the Court of First Instance of the European Communities have confirmed that markets for passenger air transport can be defined on the basis of individual routes or a bundle of routes, to the extent that there is substitutability between them according to the specific features or the case (53).

56. Due to the specific structure of the present transaction, which would combine two airlines having large scale operations from and to Ireland, the Commission has also examined whether aspects of substitutability, in particular for routes out of Dublin, might militate for the definition of a joint market.

6.2.1. Possible definition of a market for short-haul flights from/to Ireland

57. Aer Lingus and Ryanair are the two leading airlines in Ireland, with by far the largest number of aircraft based in Ireland, most of which in Dublin (around 41 out of in total around 48 shorthaul aircraft of all airlines based in Dublin belong either to Ryanair or Aer Lingus). Moreover, unlike previous airline merger cases:

(a) this case involves two point-to-point low frills carriers as opposed to two full service network carriers;

(b) the overlap between the Merging Parties is concentrated on traffic out of one Member State (Ireland) and one airport (Dublin) in particular, as opposed to a collection of individual routes between various points located in the respective home countries of the Merging Parties as was, for example, the case in the Air France/KLM merger case (54).

58. Instead of a more "fragmented" market definition following the O&D approach, it could therefore be argued that the "bundle of routes" which form the relevant market in this case are all flights from or to Ireland. Indeed, customers and competitors (including Ryanair) do often refer to an "Irish market" and claim that the conditions in this market are at least to a certain extent different from conditions in other markets (55).

59. Aer Lingus and Ryanair hold a share of about 80% of all scheduled European traffic from and to Dublin. (56) Having a significant presence in particular in Dublin allows both airlines to switch between routes and to add other routes out of the airport more easily than other competitors without such a significant base (57). From a supply-side perspective, it could therefore be argued that the "bundle" of routes out of Dublin forms one market, since suppliers operating from Ireland can switch between the different routes (supply-side substitutability).

60. The market investigation has confirmed that non-Irish airlines believe that there are significant barriers to entering the market as a non-Irish airline (or to "importing" flights). Non-Irish competitors refer to the difficulty of not having a base in Dublin from which they can easily operate on various routes out of Ireland. They also indicate that it is more difficult for non-Irish airlines to win sufficient customers for their Irish routes than for the two well-established leading Irish airlines (58). It should be noted that Ryanair and Aer Lingus enjoy by far the best brand recognition in Ireland. Third parties also argued that routes from/to Ireland differ from other European routes with respect to its relatively remote geographic position (59). The low share of non-Irish companies in flights from/to Dublin on a number of routes seems to support the view. Finally, one could argue that defining an "Irish" market would have the advantage of taking into account those factors of competition which go beyond the single "O&D"-approach (60).

61. Some third parties argued that even from a demand-side perspective, the relevant markets could be defined wider than to one specific destination. They argue that a significant proportion of the groups of non time-sensitive or "leisure" passengers could be regarded as "destination insensitive" customers. These passengers would be looking for short breaks or extended holidays only in a certain area or even throughout Europe, without having in mind a specific destination. For this customer group, some third parties argued that it would be appropriate to define a market for "city-breaks" or "holiday flights" from Dublin (61).

6.2.2. Market definition according to the O&D-approach

62. However, the results of the Commission's market investigation suggest that defining a market for a "bundle" of all flights from or to Ireland is not the most appropriate way to define the market in the present case. Indeed, the arguments mentioned in Section 6.2.1. relate almost exclusively to supply-side considerations. They disregard the fact that, from the demand side, passengers are in principle flying a given route to a given destination rather than any route to anywhere. This is of particular importance given that the demandside is, in principle, the Commission's starting point for the definition of relevant markets (62).

63. Customers normally wish to fly from a specific origin to a specific destination. While the "point" of origin and the "point" of destination may in reality be defined as an "area" from and to which customers are ready to fly, and while the O&D-approach also may include other means of transport in specific cases, the vast majority of airline customers book their flights according to plans to get from a specific city or region to another specific city or region. Following a small but significant and non-transitory price increase, these customers would not change their travel plans and choose another destination from Ireland (63). Although the Commission acknowledges that some customers might consider flying to different city or holiday airports without having a clear preference for one destination ("destination insensitive customers"), it appears unrealistic to assume that a significant proportion of passengers would not care whether they flew to Rome, to Faro or to Riga or even to Turin. On the contrary, there are obvious differences between each destination which are highly relevant for the customers' choice of destination, even in the case of leisure passengers (language spoken, tourist infrastructure, climate and other individual features of the country/city). Furthermore, in the case of business passengers or passengers visiting friends and relatives, any substitutability of different destinations is unlikely as the purpose of their journey is itself connected with a specific destination (place of a business meeting or place of residence of friends or relatives). For the vast majority of passengers, therefore, a flight from Ireland to one destination is not simply substitutable with a flight to another destination. On the contrary, from a demand-side perspective, every combination of a point of origin and a point of destination forms a separate market from a customers' viewpoint.

64. Moreover, it is questionable whether the competitive constraints from the supply-side, that is to say, the possibilities for competitors to react to a price increase on a given route by entering into competition on this route, are sufficiently immediate and effective. In contrast to taxi-companies, which can easily serve any destination a customer wants to go to, the Commission's market investigation showed that there are a number of barriers which can effectively prevent airlines from reacting to competition by opening new routes and that opening routes requires investments, strategic decisions and time. If a route from Ireland is not connected to a "base" of a competitor, this competitor is less likely to enter this new route (64). Opening a new route also requires sufficient airport capacity both at the origin and the destination, which may not be available (65). A potential competitor might also have insufficient access to customers, due to a lack of brand awareness in the destination country, to fill his aircraft on a new route (66). Further, opening a new route also involves opportunity costs as the aircraft and crew needs to be taken from another existing route which then needs to be abandoned or serviced with lower frequencies. As a result of these barriers, it cannot be expected that Irish-based airlines could immediately switch to any destination out of Ireland or that non-Irish competitors could easily fly to any Irish destination should they wish to. Therefore, the effects of supply-side substitution cannot be regarded as equivalent to those of demand substitution in terms of effectiveness and immediacy (67).

65. Finally, it should be noted that the O&D approach is in line with the Commission's established practice in a number of airline mergers and antitrust cases and that it has been approved by the Court of Justice and the Court of First Instance on various occasions; it was also supported by a large majority of competitors (68) in the Commission's market investigation.

66. For all these reasons, the O&D approach appears to be the most appropriate approach to define the relevant markets in the present case (69).

6.3. Analysis of the relevant routes (airport pairs versus city-pairs)

67. The Commission has, as set out in Section 6.2. above, based its competitive assessment on an analysis of individual routes from one origin to one destination (O&D). Since airlines operating on the identified overlap routes normally offer connections in both directions of this route, no distinction between the directions of the flight is made on the single routes.

6.3.1. Connecting flights are not part of the same market

68. Moreover, it follows from the O&D approach that the Commission will assess the effects of the proposed transaction on some but not all passengers on certain flights. On a flight between Dublin and London Heathrow, some passengers will be flying point-to-point between Dublin and London, whilst others will be taking a connecting flight in London Heathrow to another destination like, for example Tokyo, Sidney or Moscow. Although Dublin-London is an O&D pair which is affected by the proposed transaction, passengers on Dublin-London flights who connect to Tokyo, Sidney or Moscow are in principle not affected by the proposed merger. This is because there is no overlap between the services of Aer Lingus and Ryanair on the routes Dublin-Tokyo, Dublin-Sidney or Dublin-Moscow. In contrast, those passengers who are, for example, on a Dublin to London Heathrow flight and who travel point-to-point between Dublin and London are likely to be affected by the proposed transaction to the extent that Dublin-London is an O&D pair on which the Merging Parties' services overlap. Indeed, customers who book a flight, for example, from Dublin to Tokyo using "connecting" or "feeder" services (such as partly offered by Aer Lingus), usually pay a price for the entire route and do not know the separate "price" for the Dublin-London limb of this route. Connecting passengers can in such a situation not compare prices, and airlines can price discriminate between connecting passengers and "classic" point-to-point customers. Although there are certainly passengers who book their "connecting" flight with a different airline and pay it separately, these passengers are regarded as "point-to-point" passengers for the purpose of this decision, even if their ultimate destination is different. The above distinction is of material importance with a view to assessing the effects of the proposed transaction on passenger air transport services in particular between Dublin and cities in which any of the carriers operates hub airports (for example London, Frankfurt, Paris, Madrid) (70).

6.3.2. Definition of the relevant "O&D" airport and/or city pairs

69. To establish whether an O&D pair forms a relevant market, the Commission considers the different possibilities offered to consumers to travel between these two points. Since many cities are connected to two or more airports, the Commission has not only considered the direct flights between the two airports concerned, but also alternative airports to the extent that they are regarded sufficiently substitutable to these direct flights (71). The Commission therefore had to determine which "bundle of routes" between different airports belonging to two cities are substitutable and which are not (that is to say, which airports can be considered to belong to the same catchment area from the consumers' point of view). This analysis by the Commission mainly concerned the question whether the "main" airport of a city or region is substitutable with a "secondary" airport by which the same city or region may be served. The secondary airports which the Commission considered as potential substitutes are usually smaller airports (often former regional or military airports) in cities more or less remote from the "marketed" destination city (for example, "Paris/Beauvais" or "Frankfurt/Hahn"). In some other cases, the question of substitutability concerned two or more main airports of a city (for example, London).

70. Ryanair contends that secondary airports are, as a matter of principle, not substitutes for primary airports. In support of this view, it refers to Commission decisions that found that certain secondary airports were not substitutable with primary airports, particularly for time sensitive passengers (72). It argues that, for Aer Lingus, time-sensitive passengers play an important role, hence secondary airports would not be an alternative for them. Ryanair also claims that the radius of an individual airport's catchment area should be drawn rather small given the overall short travelling time of point-to-point short haul flights.

71. The question of substitutability of scheduled air transport services from different airports is relevant with a view to determining to what extent the activities of the Merging Parties overlap in the present case. The activities of Ryanair only overlap with Aer Lingus on 16 routes on which Aer Lingus and Ryanair fly between the two same airports ("airport pair" approach). The Commission has also identified additional 19 cities ("city pairs"), to which Ryanair or Aer Lingus fly from Ireland, using different airports (in most cases Aer Lingus using "primary" and Ryanair "secondary" airports). Further, in the case of four airport overlaps Ryanair also operates flights to additional destination airports belonging to the relevant route which are also taken into account in this Section (73). Moreover, the question of substitutability between airports is relevant with a view to determining to what extent the merged entity would be constrained on services to/from Dublin by services operated from either of the Belfast airports. For all these city pairs, the Commission has carried out a detailed analysis (74) in order to establish whether or not the respective airports are substitutable for flights between Dublin (Shannon and Cork) and the relevant destination (75).

6.3.3. Analytical framework

72. When analysing the substitutability of scheduled air transport services from different airports pairs, the Commission analysed whether passengers would consider passenger air transport services to/from neighbouring airports as reasonable alternatives. Do customers who wish to fly for example between Dublin (or its region) and Venice (or its region) consider as alternatives the services of Aer Lingus between Dublin Airport and Venice Marco Polo airport (the primary airport serving Venice), on the one hand, and the services of Ryanair between Dublin Airport and Treviso airport (a regional airport some 20 km North from Venice), on the other? If the answer to these questions is positive, then the air transport services concerned belong to the same market for the purposes of this Decision, that is to say, they form part of the same O&D pair. A contrario, if the answer to that question is negative, the air transport services concerned belong to distinct markets, that is to say, they do not form part of the same O&D pair.

6.3.3.1. Relevant parameters

73. In order to analyse the substitutability of scheduled air transport services from different airports, the Commission has sought to identify the main factors which are relevant for individual customers when it comes to choosing between air transport services out of different airports. The results of the Commission's investigation show that the customers take into account mainly the following elements (76):

(i) Travel time: All customers have a preference to minimise the travel time (and costs) and prefer, other things being equal, the closer airport to the more remote one. However, for the majority of all passengers on the analysed routes, time is not the decisive criterion when considering different airport alternatives (77).

(ii) Travel cost: Customers have a general preference for the cheapest solution for their journey. It should be noted that customers consider their total travel costs and not only the transfer/parking costs at a specific airport. Even a more expensive transfer to the alternative airport can therefore be a viable alternative if the total costs of the trip (flight ticket plus transfer & parking) are comparable to the costs for a trip from the closer main airport. Lower prices at a secondary airport may therefore outweigh the disadvantages of a longer and more expensive transfer.

(iii) Flight times/schedules/frequencies: Most customers also have a preference for a specific departure and return time and date and will chose the airline (and the airport from which it is operating) which corresponds most to their preferences.

(iv) Quality of service: As explained in Section 6.1. above, air carriers offer different levels of service. Similarly, airports offer different levels of service. By way of example, shopping facilities at large main airports may be relevant for the airport choice of some customers, while shorter check-in times at some airports might be considered as an advantage by other customers.

74. The criteria above are not necessarily listed in order of importance. It is the combination of these factors that drives passengers' choice for the one or the other airline service. For example, some passengers value convenience. This does not depend only on journey time, however, but also on timing and frequency. Passengers wanting to fly to an afternoon business meeting may find an afternoon flight to a secondary airport more suitable (even with the additional travel time) compared to a morning city flight. In summary, the relative importance of each of these criteria may vary from one consumer to the other when combining them depending on the consumer's individual preferences or the specifics of his journey.

75. In principle, cheap airline fares reduce the total cost of a passenger's journey. However if these cheap fares are only available from a distant airport, additional costs have to be added to the airline fare for reaching the airport. These costs reduce the passenger's incentives to opt for the cheap airline services. In this regard, Ryanair refers to the statements of Aer Lingus in its Initial Public Offering Prospectus, where Aer Lingus believes that "its customers have been willing to pay a premium over its low-cost competitors for its enhanced service offering, including seat allocation and flying to centrally located city airports (78)." Moreover, the additional travel time to the distant airport represents an inconvenience that the passenger is willing to accept in principle only if this is reflected in a lower total cost for the journey. Similarly, very cheap airline fares are commonly associated with a lower level of airline service: this is another type of inconvenience, in comparison to higher levels of airline service that may be available from another airport, which the passenger is willing to accept in principle only if this is reflected in a lower total cost for the journey. On the other hand, the shorter turnaround times and less congestion at the secondary airports may also shorten the overall travelling time. The way in which these factors combine to result in a passenger's choice for the one or the other airline service depends on this consumer's individual preferences and financial constraints (consumers seek to maximise their personal utility under their personal budgetary constraints). Further, the catchment area of low-frills carriers may be larger than those of network carriers as customers are prepared to travel further to an airport to fly on their low cost flights (79).

76. The Commission is not in a position to assess the decision of every individual passenger according to all the criteria indicated in paragraph 73 above for the purposes of defining the relevant markets. The Commission, however, in its assessment whether services from Dublin to two or more neighbouring airports belong to the same market took into account these criteria to the extent possible. The assessment was undertaken, inter alia, on the basis of the elements described in sections 6.3.3.2 to 6.3.3.5 below.

6.3.3.2. Definition of catchment areas

77. For most passengers, the more distant the airport from the point of departure or arrival, the more inconvenient the airport and the cheaper the airline fare has to be for an equivalent level of airline service. This suggests that services from an airport within a reasonable distance from the point of departure or arrival are likely to exert a greater competitive constraint than services from an airport which is distant or only accessible with difficulty.

78. The Commission has compared the distances in kilometres and the travelling time from an airport to the city centre, but also the transfer time by car, bus/coach and, if available, by public transport to the respective city centres.

79. It should be noted that the relevant time to consider with regard to determining the catchment area is not the time it takes to transfer from an airport to the destination city, but the difference in the time between transferring to the city from one airport and another candidate substitute airport.

80. The Commission observes in this regard that the extra time it takes to get from the adjacent airport to the city is often relatively little in comparison to the total door to door travelling time (for example, 10-20% longer travelling time (80)). This is because the total door to door travelling time is not simply a matter of flight time plus journey time from the destination airport, but also includes time to travel to the departure airport, time for check-in, security and customs clearance and for collecting baggage. In this regard, total travelling time is probably more important than the distance travelled (see paragraphs 73 and 74 above).

81. The Commission's analysis whether passenger air transport services between Dublin (and Shannon or Cork as relevant), on the one hand, and two or more distinct airports, on the other, is for the following city pairs and airports:

<emplacement tableau>

82. The Commission asked the airports (81) listed in table 1 about the commercial arguments and material that they use for the purposes of marketing airport services towards air carriers and attracting them on their tarmac. In all cases, whether primary or secondary airports, the "catchment area" that airports present to airlines is at least either 100 km or 1 hour driving time. In most instances, airports argue or suggest that their catchment area exceeds these limits, sometimes by far.

83. The Commission considers, therefore, that 100km or 1 hour driving time is a conservative estimate of an airport's typical minimum catchment area (82). Within such travelling distances or times to the airports, most passengers would not consider that flying from the one or the other airport is manifestly inconvenient. As a result, most passengers would openly consider flying from the one or the other airport, to the effect that competing air transport services between a point in Ireland, on the one hand, and these airports on the other may exercise a competitive constraint on each other. It should, however, be noted that the Commission uses the 100km/1 hour-"rule" only as a first "proxy" to define a catchment area. Due to the specificities of the respective airport and other evidence, the catchment area may be wider in reality and will therefore be discussed in greater detail on a case by case basis in the individual airport pair analysis (83).

84. In its response to the Statement of Objections Ryanair argued that the use of the 100 km/1 hour measure, although a useful proxy, is arbitrary and that the determination of substitutability of air transport services to different airports it is much more complex. Ryanair considers that it is more important to consider whether the competing airlines have similar route networks or are totally differentiated. This is particularly so when an important share of passengers on the route are time-sensitive or connecting (and Aer Lingus and Ryanair serve different airports).

85. The Commission notes in this regard that the 100 km/1 hour benchmark is a proxy based on the results of the view of airports on what they consider to be a reasonable catchment area. The Commission has also taken into account in its assessment the view of competitors and customers and additional evidence that is available to it (see further in airport-to-airport analysis). As regards the issue of time vs. non time-sensitive passengers, both Aer Lingus and Ryanair appear to have a comparable proportion of business customers and the evidence on the file further does not allow the Commission to distinguish time-sensitive passengers (see further on the matter of market definition below (84)).

6.3.3.3. City Centre criterion

86. Ryanair argues that the Commission's determination in the field of airport substitutability would be flawed if it relied on distances or travel times to city centres. This is because, Ryanair explains, city centres are not the final destination point of all passengers (85).

87. If the centre of a city is included in the catchment area of an airport, however, then it can be presumed that the city itself or a substantial part of this city at least is included in this catchment area. Any suburbs or other urban areas located between the city and the airport would also be included in this catchment area. Where the catchment areas overlap over densely populated areas, the number of residents, potential airline customers, who would consider flying from either airport is substantial. If this number is high enough, carriers serving one of two or more overlapping airports will take this into account when setting the level of their own fares. This applies in particular to large cities, in which the majority of passengers do not necessarily live in the city centre but in a larger area around the city centre (such as London). In such cities, the Commission has not only looked at the distance to the city centre, but also whether the catchment areas of two airports overlap in a densely populated region.

88. Therefore, the so-called city centre criterion is not to say that all passengers depart from or end their journey at the centre of a city. It is a benchmark with a view to determining whether customers would consider services to/from a neighbouring airport as an alternative. This also therefore largely determines why air carriers would consider services to/from neighbouring airports as a competitive constraint on the services that they operate. It could be further argued that in the case of some holiday destinations (for example, Tenerife, Alicante/Murcia, Bologna/Forlì or Lyon/Grenoble) the city centre is not the relevant criterion as the passengers' final destination is probably rather a tourist resort in the vicinity. However, the Commission has used the city centre criterion in these cases as an indicative benchmark showing the relative distance of the airports from the local centre and thus also the difference in their ability to serve the tourist resorts in the vicinity. Further, in all these cases but for Tenerife, Ryanair itself markets its services with reference to the relevant larger city in the vicinity, that is to say, as Bologna (Forlì), Grenoble Lyon and Vitoria (Bilbao).

89. The Commission's approach to airport substitutability from the customers' perspective is consistent with evidence from Ryanair's own media releases or marketing activity, which suggest that Ryanair views secondary airports as substitutable for primary airports. Ryanair has provided a substantial number of media releases promoting either existing or new destinations. These advertisements provide a valuable indication of both the passengers Ryanair considers as buyers of flights, and the airports it considers that its routes are in competition with.

90. Ryanair has consistently positioned its routes as substitutes for routes to adjacent city destination airports. For example Aer Lingus flies to Bologna. On the establishment of its Dublin to Forlì (Bologna) route, Ryanair announced that: "[this] will end Aer Lingus' high fare monopoly...Bologna. Irish customers previously had no alternative to Aer Lingus" (86). This shows that Ryanair clearly positions Forlì airport vis-à-vis its customers as a substitute for Bologna main airport as regards passenger air transport services to/from Dublin. Similarly Ryanair makes direct price comparisons between routes that fly to adjacent airports (catchment area).

91. Moreover, Ryanair often lists its airport with the destination city in brackets or vice versa. For example by listing "Frankfurt (Hahn)" Ryanair is identifying to the passenger that Hahn airport serves the city of Frankfurt. Further, Ryanair provides a description of the distance and the methods to reach the primary city. For example, it provides a description of how to travel to Frankfurt from Hahn airport. Ryanair also provides a description of activities in "Top 5 things to do" of the city destination since it understands that buyers of flights to adjacent airports intend to visit the city destination. For example, it lists things to do in Frankfurt under its Hahn destination.

92. In its response to the Statement of Objections, Ryanair takes the view that the way in which it markets its services to certain airports is not at all relevant for market definition purposes. The Commission does, however, believe that the way Ryanair markets its services is an important element for the definition of the relevant markets, not least because the name under which an airport is marketed can be a decisive factor for customers when they purchase a ticket. In particular new customers on a route may not know much about the town at which the secondary airport is located87 and decide to fly on this route because they know the "main" city in the vicinity of the airport to which most of them intend to fly. Marketing activities are in general designed to reflect and/or shape perceptions of customers about the offered services and thus are relevant for assessing whether the customers would consider the air transport services to different airports as substitutable.

93. Finally, the Commission's approach to airport substitutability from the side of passengers is also consistent with the information gathered by the Commission from competing airlines (88), and Member States' Civil Aviation Authorities in the course of the investigation in this case. The results of the Commission's approach are also largely confirmed by an empirical Analysis (89) conducted by the Commission with respect to the routes for which sufficient price data supplied by the Merging Parties is available (see further Annex III).

6.3.3.4. The views of individual customers

94. The Commission also conducted a Customer Survey at Dublin Airport (90) primarily for the purpose of testing Ryanair's claim that, from the perspective of the customer, Aer Lingus and Ryanair do not compete with each other (see Annex I for details). The sample of routes included all London airports and a limited number of routes where the Merging Parties travel to different airports close to the same city. Where passengers consider an airline travelling to a different airport as an alternative it can be inferred that the passenger air transport services in question are likely to exert a competitive pressure on each other. However, such evidence is of an indirect nature since respondents to the Commission's questionnaire were not asked to explicitly state whether they would consider travelling to a different airport (91).

95. Ryanair, in its response to the Statement of Objections (paragraphs 404 to 411), argues that the Customer Survey shows that the percentage of respondents that indicate that they would consider switching to a carrier serving another airport is in general lower than those that would switch to another carrier serving the same airport. According to Ryanair this is evidence that these airports are not substitutable. The Commission does not dispute that whereas up to 52% of respondents consider the other merging carrier when it serves the same airport at destination, this percentage declines when in reference to other routes. The Commission does not deny that the competitive constraint exercised by the Merging Parties on each other may be less significant when they serve different airports compared to when they serve the same airports. This does, however, not imply that such a constraint is absent or insignificant. On the contrary, the Commission's investigation showed that a large number of customers of lowfrills airlines compare and substitute the offers of airlines flying from primary and secondary airports.

96. As concerns the econometric evidence put forward by Ryanair, the "benchmarking" proposed by Ryanair in its response to the Statement of Objections mirrors the technique followed by the Commission for price correlations (see Annex III). However this approach is not relevant in the context of a survey. This is because prices of goods or services in the same geographic market necessarily move in parallel since significant differences would create opportunities for profitable arbitrage (it is this correction that induces prices to comove). On the other hand to consider an alternative carrier is the expression of an individual preference and is not influenced by market interactions. Hence heterogeneity across routes, which is not controlled for, naturally would lead to different ratios of responses to the survey. These cannot be directly compared. In contrast, in the case of price correlations, market dynamics that reflect substitutability drive the price correlation towards a common value (duly taking into consideration variations due to seasonality and common costs).

97. With respect to the answers of corporate customers in Ireland and travel agencies received by the Commission in Phase I of its investigation, these answers proved to be of limited value for certain aspects of the investigation. This is, by way of example, the case as regards airport substitutability issues. The responses that were obtained were relatively heterogeneous and a fairly high proportion of respondents replied that they were not competent to give any answer or that they did not have the information to take a position regarding airport substitutability issues. On the other hand, answers by competitors and competitors and airlines were often much more substantiated (92).

6.3.3.5. Conclusion

98. To assess and conclude on whether scheduled passenger air transport services between two or more distinct airports are substitutable from the customers' point of view, that is to say, whether such services belong to the same market from the demand-side, the Commission relied on various pieces of qualitative and quantitative evidence. Further, the Commission, in its assessment, also considered whether the different types of information available to it went in the same direction and corroborated each other.

99. In light of the above, the below listed pieces of evidence are used and considered before the Commission concludes on the substitutability of the airports concerned from the demand side. For some routes not all of the evidence listed was available.

(1) Distances and travelling times are compared to the indicative benchmark of 100km/1 hour driving time. To this effect, the Commission has relied first on the results of the market investigation (responses to airport substitutability questionnaires). Where these results were incomplete or manifestly questionable, the Commission searched for the information on reputed mapping and route planning websites (e.g. www.viamichelin.com for continental Europe and www.rac.com for the United Kingdom);

(2) The views of competitors to the merged entity;

(3) The views of airports and of Member States' civil aviation authorities to the extent such views were expressed during the course of the Commission's investigation, and results of reports of such authorities made independently of the proposed transaction;

(4) The estimated proportion of leisure passengers on a route. It is generally acknowledged that leisure passengers are rather price sensitive. They are more inclined to trade a longer total journey time in exchange for a lower total journey cost than business passengers. Therefore, on routes where the estimated proportion of leisure passengers is high, it is reasonable to assume that airport substitutability from the standpoint of passengers is wider rather than narrower. Accordingly, the indicative benchmark of 100km/1 hour driving time should be applied in a wider rather than a narrower way (see also paragraph 84).

(5) The Commission also notes that when airports form part of a so-called "airport system" pursuant to Annex II to Council Regulation (EEC) No 2408-92 of 23 July 1992 on access for Community air carriers to intra-Community air routes (93) this is used as supplementary evidence which supports the finding that airports listed under that Regulation belong to the same conurbation and can be considered substitutable from the demand side;

(6) Marketing practices and in particular the way in which Ryanair markets its services, and the fact that certain airports are presented as serving a specific city/conurbation is considered as additional evidence as this forms part of the elements that customers take into consideration when purchasing an airline ticket over the Internet;

(7) Whether transport services exist between the secondary airports and certain cities and whether these are marketed through Ryanair's own website or perhaps organised by Ryanair itself (94); (8) The result of the Commission's price correlation analysis for seventeen city-pair routes out of Dublin as available (95); and

(9) Finally, the Customer Survey conducted at Dublin Airport which provides indirect evidence as to the substitutability of certain airport pairs from the demand side.

100. The results of this analysis are further discussed in Section 6.3.4. below.

6.3.4. Detailed city-by-city analysis

6.3.4.1.Dublin - Belfast

101. The city of Belfast in Northern Ireland is served by two airports: Belfast International Airport and George Best Belfast City Airport. In the Notification (96), Ryanair stated that "there is no secondary airport to Dublin. Dublin Airport is the gateway to Ireland and airlines wishing to offer services to Dublin and the east coast of Ireland have no alternative but to use this airport". In its response to the Commission's decision to open proceedings in this case, Ryanair argued that scheduled passenger air transport services between Dublin Airport and a point located in the EEA could be constrained by similar services from either of the airports that serve Belfast. However, Ryanair also submits that the Belfast airports are "somewhat at the margins of constituting alternatives to Dublin Airport." It stated further that "in the event Dublin Airport becomes congested, or if prices are raised in the future, it is possible that the Belfast Airports may become an alternative to Dublin" (97). Hence, also Ryanair acknowledges that today these airports are not substitutable.

102. The Commission observes that the distance between Dublin and Belfast is in excess of 160 km, that is to say, substantially above the 100 km indicative benchmark. The journey by rail or the driving time is above two hours (98), which is more than double the indicative 1 hour driving time benchmark. These findings indicate that passenger air transport services out of either of Belfast airports are unlikely to constrain similar services out of Dublin Airport.

103. The Commission's market investigation also supports the finding that Dublin and Belfast should not be considered substitutable for the purpose of the notified operation. As noted neither of the Merging Parties currently fly out of Belfast airport. Further, none of the third parties the Commission has contacted during its market investigation indicated that they see Belfast and Dublin as substitutable airports. Competitors confirmed that they do not regard Belfast and Dublin as belonging to the same market, since the distance between the two airports is too long (99). The Dublin Airport and George Best Belfast City Airport agree in considering that they are not substitutes but attract distinct customer groups in different regions (100). Although Belfast International airport claims in its response that it competes with Dublin Airport, this regards in particular competition for new carriers. Indeed, the reply of Belfast International indicates that its catchment area is practically limited to Northern Ireland and also that the number of passengers with residence in Ireland is negligible (amounting to only 2.24% of all passengers at Belfast International airport) (101).

104. The Customer Survey conducted by the Commission at Dublin Airport further suggests that most passengers would not consider flying out of either of Belfast airports if prices for flights out of Dublin increased (102). In response to the hypothetical question whether customers would consider Belfast as an alternative airport a total of 16.6% of all passengers stated that they would consider using the Belfast airports in the future, whilst some two thirds of all passengers responded that they would never consider Belfast as an alternative. The remaining customers did not answer this particular question.

105. Ryanair, in its response to the Statement of Objections, argues that this question was flawed because it was not sufficiently thorough. In particular it did not explore the conditions under which substitutability may be possible (103). The Commission notes that the question in point was hypothetically phrased and as such broad. It was not drafted in a manner to catch substitutability for a specific flight (unlike the questions of the survey related to whether the passengers had in fact considered another airline). That allowed the respondent to consider any possible reason (whether it actually materialises or not) which might lead him to consider Belfast as an alternative to Dublin. The respondents were not prompted to answer whether they considered Belfast as an alternative for the route on which they were flying at that time. This thus increases the likelihood that passengers would answer this hypothetical question positively.

106. Neither party to the notified operation flies from Belfast. In this regard it should be noted that easyJet serves from Belfast some of the destinations covered by the survey.. The Commission verified the percentage of passengers that indicated easyJet as an alternative (in spite of the fact that easyJet do not fly from Dublin) to the Merging Parties. Of these, only between 0% and 4.7% (depending on the relevant route) of all passengers flying from Dublin did actually consider flying with easyJet from Belfast. As regards London in particular, Ryanair is the only carrier offering scheduled services between Dublin and London Stansted, Gatwick or Luton airports. easyJet flies to these three London airports, albeit from Belfast, whilst Aer Lingus serves only London Heathrow from Dublin. The survey shows that 24-46% of passengers flying between Dublin and London Stansted, Gatwick or Luton airports would consider flying to London Heathrow with Aer Lingus whilst only 0-4.7% of these passengers would consider flying with easyJet from Belfast. These results show that easyJet's services from Belfast are clearly not a substitute to the respective services provided by the Merging Parties from Dublin (see further Annexes I and II).

107. Finally, the Commission notes that the alleged competitive pressure from services operated out of Belfast airports was not mentioned by Ryanair in the Notification. Had the alleged competitive pressure been substantial, Ryanair would have presumably mentioned it in the notification. On the contrary, Ryanair argues in the Notification that "the DAA, owner and operator of Dublin Airport, is an absolute monopoly which possesses enormous bargaining power over all airlines. Airlines wishing to service the Dublin market have no alternative but to use Dublin Airport because there is no other airport in the Dublin area so airlines (however big) have no way of meaningfully threatening to leave" (104). In the same vein, the Commission observes the striking contrast between the vigorous competitive reaction from Ryanair in response to new entrants serving points located in the Republic of Ireland on the one hand, and the relative absence of such a reaction in response to carriers operating out of either of the Belfast airports on the other (105). Further, the internal documents of Ryanair do not indicate that the airports in Belfast would be considered as substitutable for Dublin and that carriers operating from them would be considered to provide any competitive constraints. [...]*??106).

108. In the light of the above, the Commission takes the view that scheduled point-to-point passenger air transport services from either of the Belfast airports to a point in the EEA are in a different market from such services from Dublin Airport to the same point in the EEA.

6.3.4.2. London airports

109. The table below summarises the travelling distances and times between airports and the centre of London.

<emplacement tableau>

110. In the notification, Ryanair states that the London airports are not substitutes for one another from a demand-side point of view, at least not as regards time-sensitive passengers, or from a supply-side point of view. Ryanair's arguments concerning these airports are threefold. First, Ryanair invokes the decision of the Commission in the British Midland/Aer Lingus case (113). In that decision, the Commission took the view that London Heathrow was not substitutable with other London airports in regard to flights to Dublin for customers affected by that merger. Secondly, Ryanair submits that each of the airports serves a distinct "core" catchment area and is different in terms of its size and capacity. Thirdly and finally, Ryanair claims that Heathrow is the only airport that is connected to the London Underground. This would, according to Ryanair, make Heathrow clearly different from any of the other London airports in terms of passengers' total journey time to destination.

111. Firstly the Commission can agree that the London airports are not substitutable from a supply side point of view. The Commission, however, notes that for the purpose of identifying the relevant routes it should assess whether the passenger air transport services in question are in competition with each other with respect to point-to-point passengers.

112. The British Midland/Aer Lingus case primarily concerned interlining traffic originating from Dublin and connecting in London onto the services of carriers other than Aer Lingus or British Midland. It was not about O&D point-to-point traffic between London and Dublin, but about the ability of British Midland as an entrant to capture a sufficient share of the interlining traffic between Dublin and London in addition to point-to-point passengers on this route. It is not disputed that London Heathrow, as a major hub airport, is in a special position for passengers connecting in London (114). However, this is not necessarily the case for passengers travelling point-to-point between London and Dublin. Since the overlap between the services of Ryanair and Aer Lingus concerns point-to-point traffic, the Commission can therefore not rely on a decision which was taken 15 years ago in a case that concerned a different type of transaction (notably without the involvement of low-frills carriers), but has to make its assessment, taking into account the present competitive situation in London with a particular focus on the competitive conditions for the passengers relevant in this decision (115). Further, in other previous cases the Commission found evidence that suggested that there is a substitutability of the London airports, at least for non-time sensitive passengers (116).

113. In its reply to the Statement of objections, Ryanair refers to a study carried out by York Aviation on the basis of data from passenger surveys conducted by the UK Civil Aviation Authority in 2005. The data is analysed by county outside of Greater London and by Metropolitan District within Greater London. The study summarises in a table the counties/districts which provide over 80% of the passengers using each of the London airports. On this basis, Ryanair criticises the "city-centre criterion" used by the Commission and suggests that in most instances London airports are not substitutable. The Commission notes, however, that this study does not include London City Airport. More importantly, the geographical scope of the study appears to extend way beyond the borders of London: it includes counties such as East and West Sussex, Suffolk, Kent, Norfolk and Northamptonshire. It is only normal to find, for example, that passengers from East or West Sussex will have a very strong preference for London Gatwick Airport. Accordingly, the Commission would not dispute that flights between Dublin and London Heathrow are unlikely to exert an effective competitive constraint on flights between Dublin and London Gatwick for these passengers. However, the more or less 7 million inhabitants in inner and outer London (117) are evidently in a different position when it comes to choosing between London airports. It is misleading, therefore, to mix counties such as West Sussex or Norfolk with London districts such as Kensington or Westminster. Moreover, the mere fact that no data is available concerning London City airport may introduce a bias in the results (118). Accordingly, the Commission does not agree that York Aviation's study puts into question the Commission's finding that London Airports are substitutable for a large number of Ryanair's and Aer Lingus' customers.

114. The Commission observes that all five London airports are within substantially less than 100 km from the centre of London, one of the most densely populated agglomerations in Europe. Driving times by car between the city centre and the airports are below or not significantly above 1 hour (1 hour and 25 minutes maximum). It should also be noted that many customers use rail connections to London airports due to the difficult traffic situation in greater London. Journey times by rail are an hour or less from the city centre in all cases. The fact that all the London airports are largely within the indicative 100 km/1 hour driving time benchmark thus suggests that these airports are substitutable.

115. Studies (119) conducted by the UK Civil Aviation Authority (CAA) also suggest that the catchment areas of airports located in the London area are wider than Ryanair argues. These studies also show that the profile of passengers travelling through the London airports is diverse, with substantial proportions of passengers travelling for private as well as for business purposes at primary as well as secondary airports. Contrary to Ryanair's assertions therefore, it appears that the London airports are, albeit to a varying degree depending on the exact location of the passenger and the availability in particular of air transport services at an airport, substitutes for one another.

116. In its own analyses, the CAA has adopted a two-hour surface access time as a relevant benchmark for the airport catchment areas in which to consider competition for leisure passengers. For business passengers, CAA has adopted a 1 hour drive time to represent a conservative assumption that should avoid overstating the extent of the airport's catchment area. Whilst different catchment areas may be defined for business and leisure passengers, it is important to consider whether this degree of disaggregation by passenger type is appropriate. This will depend upon the degree to which the airline is able to price discriminate effectively between different types of passengers. If price discrimination is not practical (due to the airline carrying a broad mix of passengers on all of its flights), the Civil Aviation Authority explains, a broader catchment area might be appropriate. The Commission notes in this particular respect that neither Ryanair nor Aer Lingus price discriminate business from leisure passengers.

117. The Commission observes that all of Ryanair's direct competitors (120) on the Dublin to London route (Aer Lingus, British Airways, BMI and Cityjet) take the view that London airports are substitutable for a majority of the point-to-point customers. This has been confirmed by many other carriers and by the affected airports (121).

118. The mere fact that Heathrow Airport is connected to the London Underground is not sufficient (as argued by Ryanair) to distinguish it from other London airports for many point-to-point passengers and to put it into a separate market. A tube ride from Heathrow to Westminster or Leicester Square would typically take over 45 minutes (122). This journey time would be longer for destinations in the City. Even though Heathrow is also connected by an express rail service to Paddington, this is not particularly convenient for access to Central London, and the City in particular. Journey times from Heathrow to Central London are thus comparable to times from other London airports. In any event, irrespective of transport mode travelling time is within the benchmark.

119. It should also be noted that London Heathrow, Gatwick and Stansted airports form part of an 'airport system' as indicated in Annex II to Regulation (EEC) No 2408-92 that is to say that they are grouped together as serving the same city or conurbation. This is an additional element which contradicts Ryanair's view that London airports are not substitutes for one another and that the transaction should be assessed on an airport to airport basis.

120. The Commission also notes that Ryanair's services to London Gatwick, Stansted and Luton airports are marketed on its website as services to London with the name of the airport added in brackets. Furthermore, Ryanair suggests that the fact that an airport is designated by the International Air Transport Association (IATA) for a particular city is irrelevant for market definition purposes (123). However, the Commission observes that a recent promotion by Ryanair relies on the IATA-city designation. Under their "lowest fare guarantee" (see, for example, a recent promotion between 9 May and 12 June 2007), Ryanair commits to offering the lowest fares on certain routes. The terms and conditions applicable to this promotion include a provision whereby the guarantee is only "valid for flights where there [...] is a direct city pair comparison. This includes only airports that have the same city designation as defined by IATA (International Air Transport Association). For example, claims will be accepted from competitor flights operating to/from the five IATA designated London Airports (Heathrow, Gatwick, Stansted, Luton and City)" (124). Both elements suggest that Ryanair considers that all London airports in question are substitutable for its passengers. Finally, the minutes of Ryanair's Board of Directors [...]* (125). In this context, the market share and performance of Ryanair on the routes is consistently compared with that of competitors, [...]*, irrespective whether these competitors fly to different London airports. [...]* (126) (127).

121. The empirical analysis (price correlation analysis - see further Annex III) conducted by the Commission on this route further confirms that services between Dublin Airport and the airports of Stansted, Gatwick and Luton belong to the same market. The analysis is not conclusive as to whether services between Dublin and London Heathrow airport, on the one hand, and Dublin and other London airports are in the same market. (128) However, it should be noted that the price correlation between Ryanair and Aer Lingus was very strong for Cork - London and Shannon - London routes up to the end of 2004, that is to say, before the entry of easyJet (see further Section 7).

122. Also the Customer Survey conducted by the Commission at Dublin Airport (129) provides indirect evidence that Heathrow and Gatwick are in the same market from the point of view of point-to-point scheduled customers. It also, albeit to a lesser extent, supports the view that Heathrow, Stansted and Luton are in the same market. In particular:

(1) 45% of Ryanair and 49% of British Airways passengers (flying to Gatwick) considered flying Aer Lingus (serving Heathrow). This suggests that Gatwick and Heathrow are in the same market.

(2) This is further supported by the fact that 33% of the passengers of British Midland (flying to Heathrow) considered flying with Ryanair (flying to Stansted, Gatwick and Luton).

(3) The proportion of British Airways passengers who considered Aer Lingus' services to Heathrow to be an alternative was higher than the proportion of British Airways passengers who considered Ryanair's services to be alternative. This finding is noteworthy to the extent that Aer Lingus, on the one hand, and British Airways and Ryanair, on the other, fly on to different airports.

(4) However, 18% of Aer Lingus' passengers considered Ryanair to be an alternative (indeed around 20% of passengers on Aer Lingus flights mentioned that convenient airport connections was a reason for the choice Aer Lingus). This can be at least partly explained by the fact that a certain proportion of Aer Lingus' passengers to Heathrow are connecting passengers. These passengers would not have the choice to fly to any other London airport and could therefore not consider Ryanair as an alternative. Further, it should be noted that the question concerned alternative carriers and not alternative airports. Therefore, a high percentage of those considering a carrier flying to a different airport indicate that the customer regards, among other factors, also the airport used by that carrier as an alternative for its actual flight. However, if the share is low it cannot be simply assumed that it is because the passenger views the airports as not substitutable as there are many other factors (130).

(5) 46% of the passengers of CityJet considered Aer Lingus to be an alternative. This would support the view that Heathrow and London City are in the same market

(6) On services to Stansted and Luton, the proportion of Ryanair passengers who consider Aer Lingus as an alternative is lower. However with 25-30% of Ryanair passengers who consider this to be the case, there is some support for the view that services to Stansted and Luton belong to the same market as Heathrow.

(7) Closeness of the airport to the final destination is not a factor mentioned more often by passengers flying to Heathrow than to Stansted or Gatwick (31%, 26% and 21% respectively). In contrast this factor is important for 74% and 72% of passengers flying to Luton and City airports respectively.

123. Ryanair, in its reply to the Statement of Objections (paragraph 70, first indent) points out that Aer Lingus passengers travelling to Heathrow were only surveyed on weekends. Ryanair then concludes in paragraph 375 that this "will have significantly skewed results towards carrying a disproportionately high percentage of leisure passengers on the Heathrow route. This will impact substantially on the validity of any conclusions drawn about the substitutability of Aer Lingus and Ryanair on the London route".

124. The Commission acknowledges that passengers on Aer Lingus flights to Heathrow were surveyed on a Saturday and a Sunday. This may have induced some bias in the responses for such customers but it is unlikely that the bias should be "significant". Further, it is noted that a significant proportion of Ryanair customers have considered Aer Lingus although they fly into different London airports (45%) (131). In fact even if Ryanair does not impose a significant constraint on Aer Lingus on Heathrow it is clear that Aer Lingus does exert a competitive constraint on Ryanair, because passengers, whether business or leisure, recognise that to the extent Heathrow is at a more convenient location this may justify a higher price (see further below). Hence if Ryanair were to raise its prices on the Dublin routes to the other three London airports, customers might consider switching to Aer Lingus. This also explains why even if Ryanair passengers are more price sensitive than Aer Lingus passengers (132) the latter constrain the former in that if Ryanair raises prices passengers may switch to Aer Lingus if it flies to a more convenient airport.

125. In the light of the foregoing, the Commission takes the view that scheduled point-to-point passenger air transport services between Dublin (as well as Cork and Shannon as there are no indications that the situation would be different) and London Heathrow, Gatwick, Stansted, Luton and City airports belong to the same market.

6.3.4.3.Manchester, Liverpool and Leeds-Bradford airports

126. The table below summarises the travelling distances and times between airports and the city centre of Manchester.

<emplacement tableau>

127. On the basis of the 100 km or 1 hour driving time benchmark, Liverpool and Leeds-Bradford airports appear prima facie to be substitutable for Manchester Airport from the demand side for point-to-point scheduled passenger air transport services to/from Dublin. There exists a bus service between Liverpool Airport and the city of Manchester. Tickets for the bus service are available through the website of Ryanair.

128. The UK Civil Aviation Authority concurs that Manchester and Liverpool airports are substitutable from the demand side for passenger air transport services to/from Ireland, at least for non-time sensitive passengers (138).

129. The Commission's market investigation (139) showed that the two competitors to Ryanair on this route (Aer Lingus and Luxair) operating on the affected routes or from Manchester and other third parties share the view that both airports are substitutable for a majority of point-to-point passengers.

130. This view is not put into question by arguments that were discussed in the recent decision of a national competition authority (the "Flybe/BA Connect (140)" decision by the Office of Fair Trading, "OFT"). That case concerned different markets, different carrier types and different customer groups from the merger in this case. It is true that the OFT did not consider Liverpool or Leeds-Bradford airports to be substitutable for Manchester in respect of scheduled passenger air transport services between Manchester and Belfast in its analysis of the Flybe/BA Connect-case. However, the OFT took this view on the basis that a large proportion of passengers on this route were travelling for business purposes (141) and with a view to ensuring that the merged Flybe-BA Connect could not raise prices for business travellers. The Commission's assessment of the present transaction is not focused on business passengers only. Rather it involves low-frills carriers catering mainly for passengers travelling for private purposes, whether on holidays or visiting friends and relatives. Further, the OFT assessed the substitutability for that specific route (Belfast - Manchester) and also took into account the fact that the competing carrier active in Liverpool also flies from Belfast to a different airport from that served by the Merging Parties, which further reduces the extent of competitive constraints from that competitor (142).

131. The Commission notes further in this respect that, in a report of December 2006 (143) concerning airports within the United Kingdom, the Office of Fair Trading concludes in respect of the substitutability between Manchester, Liverpool and Leeds Bradford airports for passengers: "There is therefore strong evidence that the set of potential substitutes to Manchester airport includes Liverpool and Leeds Bradford airports but could potentially include others. This accords with the views expressed to the OFT that there is an area of overlap between Manchester, Liverpool and Leeds-Bradford airports, in which all three airports compete." The view taken by the OFT in the particular circumstances of the Flybe/BA Connect case can therefore not be construed as a firm conclusion that these airports could not be substitutes in the context of a merger of airlines following a different business model, and in particular in the absence of a specific focus on business passengers.

132. Manchester airport also considers that Manchester, Liverpool and Leeds Bradford are substitutes for such services. Moreover, Manchester airport considers that it competes with the other airports when it comes to attracting air carriers. Leeds Bradford airport considers that the city of Manchester falls outside its catchment area. However, Leeds Bradford recognises that (i) a high proportion of the passengers originating in West Yorkshire or North Yorkshire (which the airport considers to be in its core cachment area) have historically used Manchester airport and that (ii) Leeds Bradford competes with Manchester airport with a view to attracting/retaining air carriers.

133. Ryanair's internal documents also suggest that Ryanair regards at least flights from Manchester and Liverpool as competing with each other as they state the following [...]* (144).

134. The Commission's empirical analysis (price correlation) also provides complementary evidence that air transport services between Dublin on the one hand, and Liverpool or Manchester airports, on the other, belong to the same market (see further Annex III).

135. In the light of the above, the Commission takes the view that scheduled point-to-point passenger air transport services between Dublin (as well as Cork and Shannon as there are no indications that the situation would be different) on the one hand, and Liverpool, Leeds and Manchester airports on the other, belong to the same market.

6.3.4.4.Birmingham International and East Midlands airport

136. The table below summarises the travelling distances and times between airports and the city centre of Birmingham.

<emplacement tableau>

137. On the basis of the 100 km or 1 hour driving time benchmark, Birmingham International and East Midlands airports appear prima facie to be substitutable from the demand side for pointto- point scheduled passenger air transport services to/from Dublin.

138. The UK Civil Aviation Authority concurs that these airports are substitutable from the demand side for passenger air transport services to/from Ireland, at least for non-time sensitive passengers (149). Birmingham International airport does not agree that they are substitutes for such services, citing poor public transport connections and doubts as to the capacity of the road network around East Midlands airport.

139. The Commission's market investigation (150) showed that the competitor operating on the affected route (Aer Lingus) and other third parties share the view that the airports are substitutes from the customers' point of view for a large majority of point-to-point passengers.

140. Ryanair's internal documents also suggest that Ryanair regards flights from Birmingham and East Midlands as competing with each other as they state the following [...]* (151).

141. The Commission's empirical analysis (price correlation) suggests that the two airports do belong to the same market (see further Annex III).

142. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Birmingham International or East Midlands airports, on the other, belong to the same market.

6.3.4.5.Newcastle and Durham Tees Valley airports

143. The table below summarises the travelling distances and times between airports and the city centre of Newcastle.

<emplacement tableau>

144. On the basis of the 100 km or 1 hour driving time benchmark, Newcastle and Durham Tees Valley airports appear prima facie to be substitutable from the demand side for point-to-point scheduled passenger air transport services to/from Dublin.

145. Newcastle airport considers that there is limited substitutability between the two airports in respect of leisure customers. Newcastle airport considers that it is in competition with other airports in the immediate area to attract airlines, as well as other airports across Europe, given the flexibility airlines have when deploying aircraft.

146. The Commission's market investigation (156) further supported the view that the two airports are substitutable from the demand side. In particular, it showed that the competitor operating on the affected route (Aer Lingus) and other third parties share the view that the airports are substitutes from the customers' point of view for a large majority of point-to-point passengers.

147. The Commission's empirical analysis (price correlation) also suggests that the two airports lie within the same market (see further Annex III).

148. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Newcastle and Durham Tees Valley airports, on the other, belong to the same market.

6.3.4.6.Glasgow International and Glasgow Prestwick airports

149. The table below summarises the travelling distances and times between airports and the city centre of Glasgow.

<emplacement tableau>

150. On the basis of the 100 km or 1 hour driving time benchmark, Glasgow International and Prestwick airports appear prima facie to be substitutable from the demand side for point-topoint scheduled passenger air transport services to/from Dublin.

151. The two airports agree that they are substitutes for such services. Moreover, the airports consider that they compete with each other when it comes to attracting air carriers.

152. On its website, Ryanair markets its service to Prestwick as a service to "Glasgow (Prestwick)". Although Ryanair suggests that the fact that an airport is designated by the International Air Transport Association (IATA) for a particular city is irrelevant for market definition Purposes (162), the Commission observes that a recent promotion by Ryanair relies on the IATAcity designation. Under their "lowest fare guarantee" (see, for example, the recent promotion between 9 May and 12 June 2007), Ryanair commits to offering the lowest fares on certain routes. The terms and conditions applicable to this promotion include a provision whereby the guarantee is only "valid for flights where there this is a direct city pair comparison. This includes only airports that have the same city designation as defined by IATA (International Air Transport Association). For example, claims will be accepted from competitor flights operating to/from [...] designated airports in Glasgow which are Glasgow International and Glasgow Prestwick" (163). These elements suggest that Ryanair considers that air transport services to the airports in question are substitutable for its passengers.

153. The Commission's market investigation (164) showed that the competitor operating on the affected route (Aer Lingus) and a clear majority of other third parties share the view that the airports are substitutes from the customers' point of view for a majority of point-to-point scheduled passengers (165).

154. The Commission's empirical analysis (price correlation) is not conclusive as to whether services between Dublin and either of the two airports lie within the same market. However, this does not rule out the possibility that services between Dublin and these airports belong to the same markets (see further Annex III) (166).

155. The Customer Survey conducted by the Commission at Dublin Airport provides an indication that there is substitutability between Glasgow International and Glasgow Prestwick airports for scheduled passenger air transport services to and from Dublin. In particular, 14.8% of Ryanair passengers to Glasgow Prestwick considered flying with Aer Lingus to Glasgow International airport, whilst 33.8% of Aer Lingus passengers considered flying with Ryanair to Glasgow Prestwick airport.

156. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Glasgow International or Glasgow Prestwick airports, on the other, belong to the same market.

6.3.4.7.Paris Charles de Gaulle and Paris-Beauvais-Tillé airports

157. The table below summarises the travelling distances and times between airports and the city centre of Paris.

<emplacement tableau>

158. On the basis of the 100 km or 1 hour driving time benchmark, Beauvais and Paris Charles de Gaulle (CDG) airports appear prima facie to be substitutable from the demand side for pointto- point scheduled passenger air transport services to/from Dublin in terms of travelling time for public transport as regards Beauvais.

159. On its website, Ryanair markets its service to Beauvais as a service to "Paris (Beauvais)". This suggests that it takes the view that the two airports are appropriate for passengers heading for Paris, the largest nearby conurbation. There is a bus service between Beauvais Airport and the city of Paris. Tickets for the bus service are available on Ryanair's website.

160. The Commission's market investigation (171) showed that Ryanair's two competitors (Aer Lingus and Cityjet) operating on the affected route and other third parties share the view that the airports are substitutes from the customers' point of view for a large majority of point-to-point passengers.

161. The Commission's price correlation analysis provides further evidence that the two airports belong to the same market (see further Annex III).

162. The Customer Survey conducted by the Commission at Dublin Airport also suggests that there is substitutability between Paris CDG and Beauvais airports for point-to-point scheduled passenger air transport services to and from Dublin. In particular, 37.2% of Ryanair passengers to Beauvais considered flying with Aer Lingus to Paris CDG airport, whilst 25.9% of Aer Lingus passengers considered flying with Ryanair to Beauvais airport.

163. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Paris-Beauvais-Tillé or Paris CDG airports, on the other, belong to the same market.

6.3.4.8.Lyon St Exupéry and Grenoble airports

164. The table below summarises the travelling distances and times between airports and the city centre of Lyon.

<emplacement tableau>

165. On the basis of the 100 km or 1 hour driving time benchmark, Lyon St Exupéry and Grenoble airports appear prima facie to be substitutable from the demand side for point-to-point scheduled passenger air transport services to/from Dublin.

166. As explained in Section 7.10.3 in respect of the competitive assessment of this route, the Dublin to Lyon route is predominantly a leisure route. It follows that the 100 km or 1 hour driving time benchmark should be applied in a wider rather than a narrower way.

167. The operators of the two airports agree that they may be substitutes for such services.

168. On its website, Ryanair markets its service to Grenoble as a service to "Grenoble Lyon".

169. The Commission's market investigation (176) showed that the Ryanair's competitor operating on the affected route (Aer Lingus) and other third parties share the view that the airports are substitutes from the customers' point of view for a large majority of point-to-point passengers.

170. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Lyon St Exupéry or Grenoble airports, on the other, belong to the same market.

6.3.4.9. Toulouse Blagnac and Carcassonne airports

171. The table below summarises the travelling distances and times between airports and the city centre of Toulouse.

<emplacement tableau>

172. On the basis of the 100 km or 1 hour driving time benchmark, Toulouse Blagnac and Carcassonne airports appear prima facie to be substitutable from the demand side for point-topoint scheduled passenger air transport services to/from Dublin.

173. As explained in section 7.9.3 in respect of the competitive assessment of this route, the Dublin to Toulouse route is predominantly a leisure route. It follows that the 100 km or 1 hour driving time benchmark should be applied in a wider rather than a narrower way.

174. Toulouse Blagnac airport does not consider that it competes with Carcassonne when it comes to attracting air carriers, given that it is not part of Toulouse Blagnac airport's strategy to attract carriers such as Ryanair. However, Toulouse Blagnac airport considers that the two airports are substitutes for services between Dublin and Toulouse-Carcassonne, from the passenger's point of view (182).

175. The Commission's market investigation (183) showed that the competitor operating on the affected route (Aer Lingus) and other third parties share the view that the airports are substitutes from the customers' point of view for a large majority of point-to-point passengers.

176. The Commission's price correlation analysis further suggests that the two airports belong to the same market (184).

177. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Toulouse Blagnac or Carcassonne airports, on the other, belong to the same market.

6.3.4.10.Rennes and Nantes Atlantique airports

178. The table below summarises the travelling distances and times between airports and the city centre of Rennes.

<emplacement tableau>

179. On the basis of the 100 km or 1 hour driving time benchmark, Rennes and Nantes Atlantique airports are slightly outside the benchmark set by the Commission as regards substitutability from the demand side for point-to-point scheduled passenger air transport services to/from Dublin.

180. Aer Lingus takes the view that the airports are substitutes. However, the Commission's overall market investigation (190) showed no clear results as to whether the airports are substitutes from the customers' point of view.

181. The Commission has found that Ryanair does not market its service to Nantes with reference to Rennes (for example, by advertising Nantes as "Nantes (Rennes)" or vice versa (191)). Also, there are no significant differences between the airports of Rennes and Nantes Atlantique which would be an incentive for passengers located in Rennes to travel to Nantes for the flight. Unlike, for example, in Frankfurt, where the main airport is highly congested and more expensive and where a large proportion of passengers consider the secondary airport to be a substitute for the main airport, the Commission could not detect such a pattern for Rennes.

182. Unlike for other city-pairs, the available data was not sufficient for the Commission to conduct a meaningful price correlation analysis for this city-pair. Further, there was no customer survey conducted for this route.

183. In the light of the above, the Commission finds that it cannot be concluded from the result of its investigation that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Nantes Atlantique or Rennes airports, on the other, belong to the same market.

6.3.4.11.Brussels Airport and Charleroi Brussels South

184. The table below summarises the travelling distances and times between airports and the city centre of Brussels.

<emplacement tableau>

185. On the basis of the 100 km or 1 hour driving time benchmark, Brussels Airport and Charleroi Brussels South airports appear prima facie to be substitutable from the demand side for pointto- point scheduled passenger air transport services to/from Dublin.

186. The competent authorities in Belgium concur that these airports are substitutes from the demand side for passenger air transport services to/from Ireland, at least for less time sensitive passengers (196). Likewise, the two airports agree that they are substitutes for such services. Moreover, the airports consider that they compete with each other when it comes to attracting air carriers.

187. On its website, Ryanair markets its service to Charleroi Brussels South as a service to "Brussels (Charleroi)", strongly suggesting that it takes the view that Charleroi Brussels South Airport is appropriate for passengers heading for Brussels, the largest nearby conurbation. There exists a bus service between Charleroi Brussels South Airport and the city of Brussels. On its website, Ryanair informs passengers where to find information on the bus service to Brussels.

188. The Commission's market investigation (197) showed that the only competitor to Ryanair on this route (Aer Lingus) and other third parties take the view that the two airports can be considered substitutes from the customers' point of view for a large majority of point-to-point passengers.

189. The empirical analysis (price correlation) is not conclusive as to whether Brussels Airport and Charleroi Brussels South airports should be considered substitutes. However, as is further explained in Annex III this does not rule out the possibility that services between Dublin and these airports belong to the same markets (198).

190. The Customer Survey conducted by the Commission at Dublin Airport (199) suggests that services between Dublin, on the one hand, and Brussels Airport or Charleroi Brussels South, on the other, are in the same market. According to the survey, 38.4% of Ryanair passengers have considered Aer Lingus and 60.6% of Aer Lingus passengers have considered Ryanair. As the two carriers fly to different airports, this is a strong indication for substitution between these airports.

191. Ryanair itself refers regularly to the Dublin/Brussels route without distinguishing between Charleroi and Brussels and [...]* (200).

192. In its response to the Commission's Statement of Objections, Ryanair asserts that the Commission fails to distinguish between the airports served by Ryanair and Aer Lingus, that is to say, between primary and secondary airports. Ryanair refers in particular to the decision of the Commission in the Charleroi State aid case (201). In that decision it is stated (202) that "[airlines] are sensitive to the type of airport services on offer, which correspond to clearly defined commercial strategies: main airports and secondary airports do not offer the same type of airport services to airlines, as the former offer full services, while the services offered by the latter are generally more basic. [...].Secondary airports present inconveniences for passengers in comparison with main airports (distance from main cities, problem of connections between main cities and secondary airports, absence of airline connections) [...] In addition, it appears that when an airline sets up at a secondary airport, the passenger traffic will not necessarily be diverted from the main airport to the secondary airport. On the contrary, it appears that this could be a factor in stimulating the market for both airports. [...] The total [...] traffic [...] from Brussels National and Charleroi to Dublin doubled in volume very quickly and has continued growing for the past seven years."

193. The Commission disagrees with the suggestion made by Ryanair on the basis of the above that point-to-point scheduled passenger air transport services between Dublin and neighbouring primary and secondary airports belong to separate markets. As discussed in Section 6.3.3, the Commission does not dispute that secondary airports may present certain inconveniences for passengers. In the same vein and as noted in Section 6.1.2, the type of airports served is precisely one of the elements according to which airlines differentiate their services. Moreover and as explained in Section 6.4 regarding airport substitutability from the standpoint of air carriers, the fact that primary and secondary airports offer different services to air carriers does not imply that consumers will not consider air transport services from Dublin to the primary or secondary airports of their destination as alternatives for one another.

194. Moreover, the Commission observes that Ryanair's quote from the Charleroi decision is selective since the first sentence of paragraph 300 of the Charleroi decision, where the Commission notes that carriers "operating out of a regional airport could attract passengers who initially travelled out of a main airport", was omitted by Ryanair. Thus the Commission recognised explicitly, as a starting point, that air transport services to primary and secondary airports may be substitutes from the standpoint of the passenger. In any event, the analysis of the impact on competition in the State aid sector is different from the one applied to concentrations between airlines, and the definition of the relevant geographic market is specific to the competition rules applying to undertakings. (203)

195. Accordingly, the Commission does not find that its assessment of airport substitutability issues in the present decision contradicts the findings of the Charleroi decision.

196. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Brussels Airport or Charleroi Brussels South airports, on the other, belong to the same market.

6.3.4.12.Amsterdam-Schiphol and Eindhoven airports

197. The table below summarises the travelling distances and times between airports and the city centre of Amsterdam.

<emplacement tableau>

198. On the basis of the 100 km or 1 hour driving time benchmark, Amsterdam-Schiphol and Eindhoven airports may not prima facie be considered substitutable from the demand side for point-to-point scheduled passenger air transport services to/from Dublin.

199. Aer Lingus takes the view that the airports are substitutes. However, the Commission's overall market investigation (208) showed no clear indication that the two airports are substitutes from the customers' point of view.

200. The Commission notes that Ryanair does not market its services to Eindhoven as services to Amsterdam. Further, Eindhoven Airport does not consider that it is substitutable with Amsterdam-Schiphol. The Dutch authorities also have no strong indications that would suggest that Amsterdam-Schiphol and Eindhoven airports are substitutes.

201. In a previous decision the Commission also suggested that (unlike in the case of London airports), there is a low degree of substitutability between Amsterdam Schiphol and other airports including Eindhoven airport (209).

202. The Commission's own empirical analysis (price correlation) is not conclusive as to whether services between Dublin and either of the two airports lie within the same catchment area. No customer survey was conducted for this route.

203. In the light of the above, the Commission considers that it cannot be concluded from the investigation that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Amsterdam-Schiphol or Eindhoven airports, on the other, belong to the same market.

6.3.4.13.Frankfurt am Main and Frankfurt/Hahn airports

204. The table below summarises the travelling distances and times between airports and the

<emplacement tableau>

205. On the basis only of the 100 km or 1 hour driving time benchmark, Frankfurt am Main and Frankfurt Hahn airports would not appear to be substitutable from the demand side for pointto- point scheduled passenger air transport services to/from Dublin. However, despite the distance, Ryanair itself markets its service to Hahn as "Frankfurt (Hahn)", strongly suggesting that it takes the view that Frankfurt/Hahn Airport is appropriate for passengers heading for Frankfurt. It should also be noted that Ryanair has argued in a court case on the appropriateness of the use of the name "Frankfurt" for Frankfurt (Hahn) that it serves, to a large extent, customers in the Frankfurt region (214). Further, it is noted that Ryanair and Aer Lingus target in particular the low-frills customer group prepared to travel further to an airport to fly on low-cost flights.

206. This is supported by internal documents of Ryanair. Ryanair systematically refers to the "Frankfurt market" and [...]* (215). This is additional evidence showing that Ryanair considers carriers serving Frankfurt am Main airport as competitors for its services in Frankfurt/Hahn.

207. The Customer Survey conducted by the Commission at Dublin Airport also suggests that Frankfurt am Main and Frankfurt/Hahn are in the same market for the purpose of the assessment of the current case (216). This is, in particular, because 66.7% of Ryanair passengers considered Aer Lingus as an alternative whilst 33.3% and 20% of Aer Lingus passengers respectively considered Ryanair and Lufthansa. Furthermore, 31.5% of Aer Lingus' business customers considered Ryanair as an alternative for flying to Frankfurt. At the same time, 47.2% and 5.7% of Lufthansa's passengers respectively considered Aer Lingus and Ryanair as a possible alternative. This supports the view that:

(8) The services of Aer Lingus and Ryanair are substitutable for a large number of pointto- point customers. This is the case in spite of the fact that they fly into different airports;

(9) Therefore, for the particular case of the Dublin - Frankfurt route, Frankfurt/Hahn and Frankfurt am Main airports can be considered part of the same market for a significant part of Aer Lingus and Ryanair passengers and their customers consider flying with the other carrier as an alternative despite the different airports served;

208. The Commission considers that the evidence provided by the customer survey is of importance, in particular because the proportion of Ryanair's passengers who considered the services of Aer Lingus as an alternative, and vice versa, (i) is high and (ii) is amongst the highest compared to the proportions observed on other routes to and from Dublin.

209. Ryanair's direct competitors on this route have diverging views: whilst the low-frills/point-topoint airline Aer Lingus considers that its passenger air transport services from Dublin to Frankfurt main airport are substitutable with services from Dublin to Hahn airport from the customers' point of view, the full-service/network carrier Lufthansa takes the opposite standpoint (Lufthansa is flying into the Frankfurt/Main airport (217)). The Commission's market investigation showed that for non time-sensitive passengers only, other competitors operating to/from points in Germany or Ireland share Aer Lingus' view that airline services from both airports are substitutable (218).

210. The Commission's empirical analysis (price correlation) is not conclusive as to whether services between Dublin and either of the two airports lie within the same market. Although the price correlation indicated that the two airports would be in the same market but inter alia given the few observations available (219), the result is not robust enough to draw firm conclusions (see further Annex III). This does, however, not rule out that services between Dublin and these airports belong to the same market (220).

211. In the light of the above, for the purpose of this case, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Frankfurt am Main or Frankfurt/Hahn airports, on the other, belong to the same markets.

6.3.4.14. Hamburg and Lübeck Blankensee airports

212. The table below summarises the travelling distances and times between airports and the city centre of Hamburg.

<emplacement tableau>

213. On the basis of the 100 km or 1 hour driving time benchmark, Hamburg and Lübeck Blankensee airports appear prima facie to be substitutable from the demand side for point-topoint scheduled passenger air transport services to/from Dublin.

214. As explained in Section 7.9.3 in respect of the competitive assessment of this route, the Dublin to Hamburg route is predominantly a leisure route. It follows that the 100 km or 1 hour driving time benchmark should be applied in a wider rather than a narrower way.

215. On its website, Ryanair markets its service to Lübeck airport as a service to "Hamburg Lübeck", strongly suggesting that it takes the view that Lübeck Airport is appropriate for passengers heading for Hamburg.

216. The Commission's market investigation (227) also showed that the competitor operating on the affected route (Aer Lingus) and the majority of other third parties share the view that the two airports are substitutable from the customers' point of view for a large majority of point-topoint passengers (228). The operators of the two airports also agree that they are substitutes for such services.

217. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Hamburg or Lübeck airports, on the other, belong to the same market.

6.3.4.15.Vienna Schwechat International and Bratislava airports

218. The table below summarises the travelling distances and times between airports and the city centre of Vienna.

<emplacement tableau>

219. On the basis of the 100 km or 1 hour driving time benchmark, Vienna Schwechat International and Bratislava airports appear prima facie to be substitutable from the demand side for pointto- point scheduled passenger air transport services to/from Dublin (within the 100 km benchmark and slightly above the 1 hour benchmark for Bratislava to Vienna city centre).

220. As explained in Section 7.10.3 in respect of the competitive assessment of this route, the Dublin to Vienna route is predominantly a leisure route. It follows that the 100 km or 1 hour driving time benchmark should be applied in a wider rather than a narrower way.

221. On its website, Ryanair markets its service to Bratislava airport as a service to "Bratislava (Vienna)", strongly suggesting that it takes the view that Bratislava Airport is appropriate for passengers heading for Vienna. In addition, there exists a bus service between Bratislava airport and the city of Vienna. Tickets for this bus service are available through the website of Ryanair.

222. The Commission's market investigation235 showed that the two competitors of Ryanair on this route (Aer Lingus and Sky Europe) and the majority of other third parties share the view that the two airports are substitutable from the customers' point of view for a large majority of point-to-point passengers.

223. As regards further evidence on the substitutability of the two airports the Commission notes the findings of the Austrian and Slovak national competition authorities as regards a proposed concentration involving Vienna and Bratislava airports in 2006. The proposed transaction was approved subject to conditions by the Austrian authorities on 1 February 2006 (236) and prohibited by a first-instance decision of the Slovak national competition authorities (237). Even though both authorities assessed the transaction primarily from the supply-side perspective - that is to say, whether these airports are substitutable from the carriers' point of view, the available information suggests that they also considered these airports substitutable from the passengers' point of view (238). Vienna airport also takes the view that Bratislava and Vienna airports are not substitutes. However, it could be questionable whether they may have taken a different view taking into consideration the on-going appeal against the decision of the Slovak national competition authority.

224. Moreover, the Customer Survey conducted by the Commission at Dublin Airport (239) suggests that services between Dublin, on the one hand, and Vienna or Bratislava, on the other, are in the same market. According to the survey, 28.4% of Ryanair passengers and 23.5% SkyEurope passengers (both flying to Bratislava) considered Aer Lingus (flying to Vienna) as an alternative and 29.3% and 16.2% of Aer Lingus passengers respectively considered Ryanair and SkyEurope.

225. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Vienna or Bratislava airports, on the other, belong to the same market.

6.3.4.16.Barcelona and Girona and Reus airports

226. The table below summarises the travelling distances and times between airports and the city centre of Barcelona.

<emplacement tableau>

227. On the basis of the 100 km or 1 hour driving time benchmark, Barcelona and Girona airports and Barcelona and Reus could be considered substitutable from the demand side for point-topoint scheduled passenger air transport services to/from Dublin, as these indicative benchmarks are only slightly exceeded in the case of travelling time and the overall findings concerning the substitutability of these airports confirms this view.

228. As explained in Section 7.9.3 in respect of the competitive assessment of this route, the Dublin to Barcelona route is predominantly a leisure route. It follows that the 100 km or 1 hour driving time benchmark should be applied in a wider rather than a narrower way.

229. The airport operator, Aena, agrees that the airports are substitutes for passenger air transport services to/from Dublin for consumers. Aena states that it does not consider that the airports compete with each other when it comes to attracting air carriers, given that they all belong to same operator.

230. On its website, Ryanair markets its service to Girona and Reus as services to "Barcelona (Girona)" and "Barcelona (Reus)" respectively. There exist bus services between Girona and Reus airports, on the one hand, and the city of Barcelona, on the other. The web sites of the bus service providers are accessible through the website of Ryanair.

231. The Commission's market investigation (245) showed that Ryanair's three competitors operating on the affected route (Aer Lingus, Iberia and Spanair) and the majority of other third parties share the view that the two airports are substitutable from the customers' point of view for a large majority of point-to-point passengers.

232. The Commission's empirical analysis (price correlation) provides further evidence that services between Dublin, on the one hand, and Barcelona, Girona or Reus airports, on the other, belong to the same market (see further Annex III).

233. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Barcelona, Girona or Reus airports, on the other, belong to the same market.

6.3.4.17.Alicante and Murcia airports

234. The table below summarises the travelling distances and times between airports and the city centre of Alicante.

<emplacement tableau>

235. On the basis of the 100 km or 1 hour driving time benchmark, Alicante and Murcia San Javier airports appear prima facie to be substitutable from the demand side for point-to-point scheduled passenger air transport services to/from Dublin.

236. As explained in Section 7.9.3 in respect of the competitive assessment of this route, the Dublin to Alicante route is predominantly a leisure route. It follows that the 100 km or 1 hour driving time benchmark should be applied in a wider rather than a narrower way.

237. On its website, Ryanair markets its service to Murcia as a service to "Murcia (Alicante)", strongly suggesting that it takes the view that Murcia Airport is appropriate for passengers heading for Murcia.

238. The Commission's market investigation (251) showed that the competitor operating on the affected route (Aer Lingus) and the majority of other third parties share the view that the two airports are substitutable from the customers' point of view for a large majority of point-topoint passengers. Aena, the operator of the two airports, also considers that the airports are substitutable for such services.

239. In its response to the Statement of Objections Ryanair argues that the two airports are, for example, not substitutable for time-sensitive/business passengers. In this regard the Commission notes that the route in question is a leisure type route with a limited proportion of business customers. Further, very few customers (around [0-10]*% of Ryanair customers) book less than seven days prior to departure (252).

240. The Commission's empirical analysis (price correlation) further suggests that the two airports belong to the same market.

241. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Alicante or Murcia airports, on the other, belong to the same market.

6.3.4.18. Bilbao Sondica and Vitoria airports

242. The table below summarises the travelling distances and times between airports and the city centre of Bilbao.

<emplacement tableau>

243. On the basis of the 100 km or 1 hour driving time benchmark, Bilbao Sondica and Vitoria airports appear prima facie to be substitutable from the demand side for point-to-point scheduled passenger air transport services to/from Dublin.

244. As explained in Section 7.9.3 in respect of the competitive assessment of this route, the Dublin to Bilbao route is predominantly a leisure route. It follows that the 100 km or 1 hour driving time benchmark should be applied in a wider rather than a narrower way.

245. On its website, Ryanair markets its service to Vitoria as a service to "Vitoria (Bilbao)", strongly suggesting that it takes the view that Vitoria Airport is appropriate for passengers heading for Bilbao.

246. The Commission's market investigation showed that the competitor operating on the affected route (Aer Lingus) takes the view that the two airports are substitutable from the customers' point of view for a large majority of point-to-point passengers (257).

247. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Bilbao Sondica or Vitoria airports, on the other, belong to the same market.

6.3.4.19.Tenerife North (Norte Los Rodeos) and South (Sur Reina Sofia) airports

248. The table below summarises the travelling distances and times between airports and the city centre of Santa Cruz.

<emplacement tableau>

249. On the basis of the 100 km or 1 hour driving time benchmark, Tenerife North and South airports appear prima facie to be substitutable from the demand side for point-to-point scheduled passenger air transport services to/from Dublin. Further, as the two airports are located on opposite sides of the island, it can be presumed that any other location on the island (for example, different sea-side resorts on the coast) would be within the indicative thresholds as well.

250. As explained in Section 7.9.3 in respect of the competitive assessment of this route, the Dublin to Tenerife route is predominantly a leisure route. It follows that the 100 km or 1 hour driving time benchmark should be applied in a wider rather than a narrower way.

251. The Commission's market investigation (262) showed that the competitor operating on the affected route (Aer Lingus) and the majority of other third parties share the view that the two airports are substitutable from the customers' point of view for a large majority of point-to- point passengers.

252. The Customer Survey conducted by the Commission at Dublin Airport (263) suggests that services between Dublin, on the one hand, and Tenerife North or Tenerife South, on the other, are in the same market. According to the survey, 32.3% of Ryanair passengers (flying to Tenerife North) considered Aer Lingus (flying to Tenerife South) as an alternative and 33.3% of Aer Lingus passengers considered Ryanair as an alternative.

253. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Tenerife North or South airports, on the other, belong to the same market.

6.3.4.20.Rome Fiumicino and Rome Ciampino airports

254. The table below summarises the travelling distances and times between airports and the city centre of Rome.

<emplacement tableau>

255. On the basis of the 100 km or 1 hour driving time benchmark, Rome Fiumicino and Rome Ciampino airports appear prima facie to be substitutable from the demand side for point-topoint scheduled passenger air transport services to/from Dublin.

256. The two airports form part of an 'airport system' as indicated in Annex II to Regulation (EEC) No 2408-92. That is to say that they are grouped together as serving the same city or conurbation.

257. On its website, Ryanair markets its service to Ciampino as a service to "Rome (Ciampino)".

258. The Commission's market investigation (269) showed that the competitor operating on the affected route (Aer Lingus) and the majority of other third parties share the view that the two airports are substitutable from the customers' point of view for a large majority of point-topoint passengers. The operator of the two airports further considers that they may be considered as substitutable, albeit catering for different market segments.

259. Further, the Customer Survey conducted by the Commission at Dublin Airport (270) suggests that services between Dublin, on the one hand, and Rome Fiumicino or Rome Ciampino, on the other, are in the same market. According to the survey, 50.5% of Ryanair passengers (flying to Rome Ciampino) considered Aer Lingus (flying to Rome Fiumicino) as an alternative and 53.8% of Aer Lingus passengers considered Ryanair as an alternative.

260. The Commission's empirical analysis (price correlation) also suggests that passenger air transport services between Dublin, on the one hand, and Rome Fiumicino or Rome Ciampino airports, on the other, belong to the same market.

261. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Rome Fiumicino or Rome Ciampino airports, on the other, belong to the same market.

6.3.4.21.Milan Linate, Malpensa and Bergamo (Orio al Serio) airports

262. The table below summarises the travelling distances and times between airports and the city centre of Milan.

<emplacement tableau>

263. On the basis of the 100 km or 1 hour driving time benchmark, Milan Linate, Malpensa and Bergamo airports appear prima facie to be substitutable from the demand side for point-topoint scheduled passenger air transport services to/from Dublin.

264. The airports form part of an 'airport system' pursuant to Annex II to Council Regulation (EEC) No 2408/92, that is to say that they are grouped together as serving the same city or conurbation.

265. On its website, Ryanair markets its service to Bergamo as a service to "Milan (Bergamo)" or "Milan (Orio Al Serio)", strongly suggesting that it takes the view that Bergamo Airport is appropriate for passengers heading for Milan. In addition, there exists a bus service between Bergamo Airport and the city of Milan. Tickets for this bus service are available through the website of Ryanair.

266. The Commission's market investigation278 showed that the competitor operating on the affected route (Aer Lingus) and the majority of other third parties share the view that the three airports are substitutable from the customers' point of view for a large majority of point-topoint passengers. The operators of the three airports also take the view that they all serve the city of Milan for such passengers.

267. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Milan Linate, Milan Malpensa or Bergamo airports, on the other, belong to the same market.

6.3.4.22.Venice and Treviso airports

268. The table below summarises the travelling distances and times between airports and the city centre of Venice.

<emplacement tableau>

269. On the basis of the 100 km or 1 hour driving time benchmark, Venice and Treviso airports appear prima facie to be substitutable from the demand side for point-to-point scheduled passenger air transport services to/from Dublin.

270. As explained in Section 7.9.3 in respect of the competitive assessment of this route, the Dublin to Venice route is predominantly a leisure route. It follows that the 100 km or 1 hour driving time benchmark should be applied in a wider rather than a narrower way.

271. The airports form part of an 'airport system' pursuant to Annex II to Regulation (EEC) No 2408-92, that is to say that they are grouped together as serving the same city or conurbation.

272. On its website, Ryanair markets its service to Treviso as a service to "Venice (Treviso)". In addition, there exists a bus service between Treviso Airport and the city of Venice. The bus schedule, as apparent from the website of the service provider, is tailor made to the flight schedule of Ryanair, with direct references to Ryanair flight numbers (285).

273. The Commission's market investigation (286) showed that the competitor operating on the affected route (Aer Lingus) and the majority of other third parties share the view that the two airports are substitutable from the customers' point of view for a large majority of point-topoint passengers. The two airports also consider that they are substitutable for such services.

274. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Treviso or Venice airports, on the other, belong to the same market.

6.3.4.23.Bologna and Forlì airports

275. The table below summarises the travelling distances and times between airports and the city centre of Bologna.

<emplacement tableau>

276. On the basis of the 100 km or 1 hour driving time benchmark, Bologna and Forlì airports appear to be substitutable from the demand side for point-to-point scheduled passenger air transport services to/from Dublin, even though the travelling time from Forlì is slightly above the 1 hour benchmark, as other evidence supports this view.

277. As explained in Section 7.9.3 in respect of the competitive assessment of this route, the Dublin to Bologna route is predominantly a leisure route. It follows that the 100 km or 1 hour driving time benchmark should be applied in a wider rather than a narrower way.

278. On its website, Ryanair markets its service to Forlì as a service to "Bologna (Forlì)", strongly suggesting that it takes the view that Forlì Airport is appropriate for passengers heading for Bologna.

279. The Commission's market investigation (292) showed that the competitor operating on the affected route (Aer Lingus) and the majority of other third parties share the view that the two airports are substitutable from the customers' point of view for a large majority of point-topoint passengers.

280. In the light of the above, the Commission considers that scheduled point-to-point passenger air transport services between Dublin, on the one hand, and Bologna or Forlì airports, on the other, belong to the same market.

6.3.4.24. Berlin airports

281. In its response to the Statement of Objections, Ryanair argues that the Commission has not considered the substitutability of the three Berlin airports: Schönefeld, Tegel and Tempelhof. However, this is not necessary as Ryanair and Aer Lingus both fly to Schönefeld and no other carrier serves Dublin from any of the three Berlin airports (see further Section 7.9.3).

6.4. Airport substitutability from the standpoint of air carriers (supply-side substitution)

282. It is important to underline the fact that although passengers may consider flying between, for example, Dublin and a given city via two or more destination airports serving this city, this demand side substitutability does not imply that these airports are substitutable from the standpoint of air carriers who would consider launching an air service between Dublin and the city in question. This can be explained by supply-side considerations pertaining either to operational efficiency or to the differentiation in airlines' passenger air transport services. The investigation of the proposed transaction, in line with previous Commission decisions such as AF/KLM and Lufthansa/Swiss, has also found that the fact that an airport is substitutable from the demand side does not imply that it is substitutable from the supply side. The following summarises the main reasons for this.

283. Operational efficiency: As explained briefly above (293), air carriers tend to concentrate their operations at some airports. Network carriers who operate a hub & spoke model do so because it is inherent in their model that they will concentrate traffic on their hub with a view to maximising network connectivity. Point-to-point carriers tend to do so because it allows them to achieve cost savings (294). It follows, for example, that a hub carrier may not be interested in operating from a secondary airport close to its main hub, whilst a low frills carrier with existing operations at a secondary airport may not be interested in launching new services from another nearby airport.

284. Airline service differentiation: As explained above (295), the type of airports (primary versus secondary airports) that air carriers fly to constitutes one of the dimensions according to which air carriers seek to differentiate their services from those of competitors. For example, network carriers are in general full service airlines. As such, they would in principle have a strong preference for flying to primary airports, that is to say, to airports which are more centrally located and offer a higher quality of service to passengers. Similarly, one of the ways in which certain low-frills carriers like Aer Lingus differentiate their offer from other carriers is by flying preferably to primary airports. Conversely, other low frills operators like Ryanair would not hesitate in lowering the quality of their service if this resulted in lower costs and prices. Consistent with this policy, such carriers would show a clear preference for flying to secondary airports.

285. It is worth noting in this regard that the ability of a new entrant to differentiate its services from that of incumbents may be a critical factor when it comes to assessing the viability of entry in a highly competitive market.

286. The Commission market asked (296) the airports listed in Table 1 and the Civil Aviation Authorities of Member States whether the factors which passengers use are the same as those which airlines use when choosing to fly from one or more airports. Consistent with the replies from competitors (297) in the first phase of the investigation, the results show clearly that airport substitutability issues should be looked at differently from the standpoint of passengers or of air carriers. The responses suggest that the criteria that are relevant for airlines with a view to choosing to serve one or another airport relate mainly to the airport infrastructure itself (capacity and congestion, operational infrastructure, airport opening statistics (hours of opening, airport equipment adapted to local weather conditions), airport charges and aircraft handling costs) or to the market served by an airport (size of the city, intensity of competition at the airport, type of traffic (business or leisure)).

287. In the light of the foregoing, the Commission concludes that the substitutability between airports from the standpoint of carriers cannot be presumed from such substitutability from the standpoint of passengers. Whilst the latter is the most relevant criterion to define the relevant markets, the first is important in particular within the framework of the competitive assessment to evaluate the likelihood of air carriers to enter certain routes (298).

6.5. Joint market for direct flights and indirect flights?

288. Customers can travel to a given destination either by way of a direct flight between the point of departure and the point of destination or using an "indirect" flight via an intermediate destination. Indirect flights take significantly longer than direct flights, since they require the passenger to change plane. This requires the passenger to get back to the terminal, to wait for at the terminal, to check in for the second flight and to walk to the plane and to board the aircraft. Depending on whether the considered airlines offer connecting services, this may involve long waiting periods of several hours in between the two flights. Some airlines, in particular smaller ones, offer so-called "stop-over" flights, which have to be distinguished from indirect flights, since "stop-over" passengers can remain in the aircraft when it stops at the intermediate destination to pick up more passengers. However, also for these flights, flight times are longer, since they involve one stop more (landing, boarding of new passengers and take-off), and since the intermediate destination is in most cases not directly in between the point of origin and the enddestination.

289. As concerns the market definition for indirect flights, the Commission has in previously cases distinguished between short-haul and, to a lesser extent, medium-haul intra-European flights and long-haul inter-continental routes (299): For the purpose of the current transaction the two parties' activities only overlap with regard to intra-European flights. As regards the substitutability of direct and indirect flights within Europe, the Commission has found in previous cases that indirect services do not normally provide a constraint to direct services on short-haul flights (300), mainly due to the short travelling time as customers are less likely to consider indirect flights when the additional stop could easily double or triple the flight time. Indirect flights appear to be even less of a substitute to direct flights in the present case, since the main operators on the affected routes are point-to-point low frills carriers and not hub carriers (which commonly feed traffic indirectly through hubs). Ryanair offers no and Aer Lingus very limited connecting services, which leads to long waiting times in between two flights and makes indirect flights even more unattractive to customers.

290. Ryanair follows this approach and submits that indirect flights in the context of short-haul flights do not exert a competitive constraint over direct short-haul flights, given the limited duration of the journey (301). The Commission therefore concludes that for the assessment of the proposed transaction indirect flights cannot be considered part of the same market as direct flights on the identified O&D pairs.

291. Although there are good reasons to also exclude indirect "stop-over" flights from the relevant market (in particular the longer travel time), the Commission can finally leave this question open, since there are only a very few "stop-over" flights offered by airlines on the relevant O&D pairs, hence, the competitive assessment would not materially change under either definition. In cases where relevant, "stop-over" flights will be discussed in more detail within the framework of the competitive analysis of individual routes.

6.6. Possible inclusion of other means of transport

292. The Commission has in its previous practice not only considered air transport services when defining the relevant O&D markets, but also other transport alternatives to the extent that they are substitutable to a flight (intermodal competition) (302). This has been considered in cases where alternative modes of transport on the respective O&D market, can be considered comparable in terms of price, quality and travel time and can therefore be considered as substitutes by customers. The Commission's investigation of the proposed transaction has shown that intermodal competition does not impose any significant constraints on the scheduled air transport services offered by the parties to the concentration.

293. In particular, since the notified transaction combines two airlines operating from an island and since all overlapping O&D pairs in the present case concern routes from/to Ireland, the Commission's investigation found that customers have very limited possibilities to substitute a flight from Ireland by other means of transport, which would all require significantly more travel time than a flight.

294. The most pertinent transport alternative to air transport identified in previous cases was travel in high speed trains, which can come close to a flight in terms of price, quality and travel time (303). Due to the geographic characteristics of the present case, however, this transport alternative is not available to customers wanting to fly from/to Ireland. In the absence of a tunnel between Ireland and the United Kingdom/the continent, the existing train connections are indirect and require a crossing of the Irish Sea by ferry. Even to the closest destinations, a journey with train and ferry would take significantly more time (for example, more than 6 hours for Dublin-Manchester (304)) than a flight. Moreover, the differences in terms of quality of the offered service between a flight and a combined train/ferry trip which involves a sea crossing of one or more hours across the Irish Sea appear to be more significant than the differences between a (high speed) train connection and a flight.

295. The same applies to the possible transport alternative of combined bus and ferry travel. This appears evident for most continental routes, which often bridge distances of more than 1000 km, and which would require travelling in buses and ferries for more than 24 hours. Even on the shorter routes between Dublin and the United Kingdom, the competitive constraints exercised on airlines by bus companies appear to be too limited to justify the definition of a joint product market encompassing flights and bus/ferry travel. Although it cannot be excluded that some customers who are extremely time-insensitive may consider both alternatives when planning a journey, the Commission has not found evidence that a significant proportion of the flight passengers on the affected routes would be ready to switch to the bus and the ferry transfer should flight prices increase (305). Even on the shortest routes, the travel times by bus and ferry remain manifestly longer than those by flight (for example, four and a half hours for Dublin-Manchester), and the product itself (involving crossing the Irish Sea) is hardly comparable to a flight in a large aircraft such as those operated by Aer Lingus and Ryanair. Neither Ryanair nor Aer Lingus or any other competitor takes into account ferry or bus prices when setting prices for their products or feels constrained by ferry routes (306).

296. It can thus be concluded that in the present case the limited possibilities of inter-modal substitution do not exercise any appreciable competitive pressure on the airlines operating on the affected routes and that these services can therefore be disregarded for the analysis of the competitive structure of the relevant city pairs.

6.7. Joint market for scheduled flights and charter flights?

297. Ryanair further suggests including not only so-called "scheduled" airline services (that is to say, air transport services which operate according to regular, published schedules), but, at least on certain routes, also so-called "charter" or non-scheduled air transport services in the relevant product market. Charter air transport services are usually defined as air transport services that take place outside normal schedules, normally by a hiring arrangement with a particular customer (for example, a tour operator). Indeed, charters airlines previously did not sell tickets directly to passengers, but to holiday companies who had chartered the flight. These companies would offer the flight as (transport) part of a package holiday, in which the price paid includes flights, accommodation and other services. However, in recent years some charter airlines have started selling some of their seats directly to the end customer (307), for example, via the internet (hereinafter referred to as "dry seat" sales (308)).

298. Charter airlines often fly to destinations where no scheduled airline is active and usually operate on a seasonal basis with a relatively low frequency of flights, in response to the requirements of tour operators (for example, once a week on Saturday, only during the summer or only during the ski season). Different types of activities by charter airlines can be distinguished: Sales of seats to tour operators (6.7.1), sales of "dry-seats" to end customers (6.7.2) and sales of "package holidays" to end customers (6.7.3).

6.7.1. Sales of seats to tour operators

299. Charter airlines sell seats (or entire flights) to holiday companies, who then integrate the flight into a package holiday (309). In previous decisions, the Commission regarded the supply of airline seats to tour operators ("wholesale") as a distinct market from the supply of scheduled air transport services to end customers (310). Indeed, from a demand-side perspective, market conditions on the market for sales of seats to tour operators differ significantly from those on the market for sales to individuals. The market for sales of seat packages to tour operators is a market which is upstream to the market for seat sales to individuals. Accordingly, the competitive conditions in this market are manifestly different, since tour operators have different requirements from individual customers (for example, buying of large seat packages, negotiation of rebates, taking into account of customers' needs in terms of flight times etc.). It can thus be concluded that "wholesale" sales of seat packages to tour operators are not in the same market as scheduled air transport services for end customers.

6.7.2. Market for "dry seat" sales to end customers

300. When arguing that there is a competitive constraint from charter flights on Ryanair/Aer Lingus on certain routes, Ryanair refers particularly to sales of "dry seats (311)". Indeed, the differences between "dry seat" sales of charter airlines and "classic" sales of seats on scheduled flights appear to be very limited. Some differences might nevertheless justify arguing that sales of dry seats offered by charter airlines are not in the same market as sales of scheduled air transport services to end customers (for example, the limited frequency offered by charter airlines, frequent requirement of a "minimum stay" of 7 days since flights are usually offered only at weekends, limited access of charter airlines to "traditional" point-to-point customers and business customers, etc.). However, since "dry seats" can be booked individually (for example, via the internet) for a given date, these "charter" seats can, from a customer's point of view, hardly be distinguished from other scheduled (312) flights. It can therefore not be excluded that dry seat sales exert a competitive pressure on scheduled services in countries where charter airlines offer significant numbers of dry seats on their charter flights.

301. The Commission, however, notes that while Ryanair seems to imply that a number of charter carriers offer a significant number of "dry seats" on the overlapping "leisure" routes, the Commission has found no evidence of charter carriers or tour operators offering "dry seats" on a larger or even significant scale in Ireland which could effectively constrain the Merging Parties' behaviour.

302. Ryanair has not put forward any convincing evidence to support its view that it faces a significant competitive constraint from charter carriers operating out of Ireland for "ski or sun" holiday destinations and on services to other destinations (313). In particular, while pointing the Commission at some websites which also contain dry-seat offers by charter airlines (314), Ryanair, did not submit evidence that other charter airlines than those identified by the Commission are indeed offering "dry seats" to end customers on the relevant routes to/from Dublin, Shannon or Cork to a significant larger extent than estimated by the Commission, based on its results of the market investigation.

303. The various minutes of board meetings Ryanair referred to (315), in which the participants discussed, among others, issues related to charter operators, do not relate to any of the relevant O&D routes out of Dublin (316) but to [...]* where, in contrast to Ireland, many charter airlines offer a significant number of dry seats.

304. The Commission does not dispute that Ryanair may be subject to certain competitive constraints from charter carriers on some of its UK or, for example, German routes. However, the present decision concerns exclusively routes from and to Ireland. The evidence submitted by Ryanair on the effects of charter airlines on its Glasgow - Barcelona route is therefore of limited relevance for this decision (317). Also the recent decision of the OFT in the "Flybe/BA Connect" case (in which the OFT took the view that charter flights were likely to constrain the merged Flybe-BA Connect on the Birmingham to Geneva route) does not imply that charter flights exert a similar constraint on Irish routes. Indeed, neither "Glasgow- Barcelona" nor "Birmingham-Geneva" are relevant markets for the purposes of this Decision. Both routes are routes to/from the United Kingdom where the Commission does not contest that a number of charter airlines offers "dry seats" for sale in substantial quantities. For example ThomasCook, My Travel, First Choice Airways and Monarch all offer flights from a number of UK cities to various holiday destinations on their websites. In Ireland, however, dry seat sales play a much smaller role according to the Commission's market investigation, where only a few charter airlines offer any dry-seat flights at all (318).

305. The Commission's market investigation has shown that there is very little competitive constraint from the few charter carriers selling "dry seats" to individuals on the relevant routes from or to Ireland. Indeed, most of the charter airlines mentioned as competitors by Ryanair who answered the Commission's questionnaire have indicated that they do not sell any "dry seats" for Irish flights to end customers on any of the overlapping routes. Most of the responding charter airlines indicated that they sell their seats for flights from Ireland exclusively to tour operators, who then sell them on to end customers as part of a holiday package. As set out in Section 6.7.1., the conditions in this "wholesale" market are fundamentally different from the market for sales to end customers. Some tour operators do, however, sell small numbers of dry seats to end-customers.

306. The Commission has identified just a few charter airlines (for example, XL Airways, MyTravel and FirstChoice) who offer directly or via tour operators a limited number of "dry seats" on a limited number of frequencies for three of the affected routes from Dublin (319). In no market does the number of such dry seats exceed 0-5%.The competitive pressure from "dry seat" sales on all routes out of Ireland appears therefore to be negligible. Indeed, no charter airline answering the Commission's market investigation replied that it would enter a new route on which Ryanair and/or Aer Lingus are active, should the merged entity decide to increase their fares (320). Finally, the Commission notes that both Ryanair and Aer Lingus have confirmed that they do not consider charter airlines when setting prices for their flights out of Dublin (321). In view of the limited presence of dry seat sales in Ireland, the question whether to include "dry seat" sales in the relevant market or not can ultimately be left open for the purpose of the present decision, since the competitive assessment would not change under either delineation.

6.7.3. Market for tour operating or for "package holidays"

307. The Commission has also verified to what extent Ryanair and Aer Lingus might face competition from "package holiday" operators. Since Ryanair and Aer Lingus provide the possibility to book a hotel and to rent a car on their websites, it could be argued that Ryanair and Aer Lingus are, at least to a certain extent, active in the market for "package holidays" and exercise pressure on tour operators and charter carriers and vice versa. The market for package holidays has undergone a change in recent years. Customers, who a few years ago could only book a package holiday if they wanted to fly cheaply to some destinations, can now book a cheap flight with a scheduled airline as an isolated product without any holiday package. Charter and scheduled airlines (322) as well as independent studies confirm that the rise of the low-frills carriers had a certain (negative) impact on the business-model of tour operators and charter airlines (323) that now face competition from low frills-carriers.

308. The fact that tour operators are negatively affected by low-frills carriers does, however, not mean that these tour operators exert any competitive pressure on the Merging Parties. Indeed, Ryanair does not argue that it is active in the market for package holidays. Ryanair and Aer Lingus do not sell and invoice accommodation or rental cars, but receive a fee if such products are booked via their website. In that respect, their service is rather similar to that of a travel agency or retailer. It should also be noted that the service of tour operators goes well beyond the product that can be booked on Ryanair's and Aer Lingus' website (flight plus hotel), since they offer a number of additional services (transfer from/to the hotel, travel guides, animation, insurance, etc.).

309. For the assessment of the question whether charter airlines/tour operators exercise a significant competitive constraint on scheduled airlines (beyond selling "dry seats"), it is important to stress that such a constraint would only affect those customers (i) who want to book a flight and accommodation and (ii) who would have considered booking a package holiday from a tour operator/charter airline. Many passengers flying with Ryanair or Aer Lingus, however, do not need a hotel (that is to say, passengers flying to holiday homes, passengers flying to visit friends and family, etc.). Even those customers who book a hotel via Ryanair or independently would not necessarily consider package holidays as an alternative, since they might not want to pay for "extra services" of a tour operator such as hotel transfer, travel guides, animation etc. The results of the Commission's Customer Survey carried out at Dublin Airport also showed that no passengers flying to "leisure" or so-called "ski & sun" holiday destinations considered charter airlines as an alternative. This confirms that package holidays are a manifestly different product.

310. The Commission therefore concludes that sales of package holidays offered by charter airlines and tour operators are not in the same market as the scheduled air transport services provided by Ryanair and Aer Lingus on the affected routes.

6.7.4. Conclusion on competition from charter airlines

311. The Commission therefore concludes that most of the services offered by charter airlines are not in the same market as scheduled air transport services (package holiday sales, seat sales to tour operators). It can be left open whether or not dry seat sales are part of the same relevant market, because this would not change the competitive assessment.

6.8. Separate market for non time-sensitive and time-sensitive passengers?

312. The Commission has, in several previous decisions concerning passenger air transport, defined separate markets for "time-sensitive" and "non time-sensitive" passengers (324). Time-sensitive passengers were defined as passengers whose main concern is to reach their destination in the shortest possible time, who are not flexible in terms of time of departure/arrival, and who require that the airline offers them the possibility to change their reservation at short notice (so-called "unrestricted tickets") (325). In order to distinguish between the two types, the Commission has grouped passengers travelling on "unrestricted" tickets in the category of time-sensitive and the remaining passengers in the category of non time-sensitive passengers.

313. All of the previous Commission decisions concerned at least one network carrier. This is not the case here in so far as Ryanair and Aer Lingus are both point-to-point low frills carriers. This difference is of particular importance because one of the typical features of low-frills carriers, in comparison to network full service airlines, is that their fare structure is much simpler. This is also the case in the present transaction where both Ryanair and Aer Lingus (326) have chosen to market their services only in the form of one-way restricted tickets for which, at any given point in time and for any one flight, there is only one price. The implication of this policy is that both Aer Lingus and Ryanair have foregone the possibility of price discriminating between customers on the basis of, for example, restricted/unrestricted tickets or the so called "Saturday night rule (327)".

314. It is indisputable that there exist different types of passengers, as explained by Ryanair in the Notification and many of its submissions. Various distinctions can be made in this regard. For example:

(10) the distinction which the Commission found relevant in previous cases between time-sensitive and non time-sensitive passengers;

(11) the distinction between business and leisure passengers;

(12) customers who are less price-sensitive and more price-sensitive ("low-frills") customers;

(13) within leisure passengers, the distinction between passengers on holiday and those visiting friends or relatives; or

(14) on certain routes, in particular between Ireland and the Eastern European countries, passenger traffic associated with increasing numbers of migrants into the dynamic Irish economy.

315. The proportion of passengers in the above groups or subgroups may vary from one route to the other and the requirements or preferences of passengers in these groups or subgroups may also vary. For the present case therefore, the key question is whether different markets reflecting different types of passengers should be identified with a view to assessing the competitive effects of the proposed transaction and, if so, according to which distinction. Indeed, the fact that the present transaction, unlike previous merger cases, concerns in particular point-to-point customers flying with low-frills airlines, is an important difference to previous merger cases assessed by the Commission. It can be assumed that low-frills airline customers have, to a certain extent, different preferences from customers of more expensive network carriers. For instance, the Commission's market investigation has shown that customers of low-frills airlines are usually more ready to travel to remote "secondary" airports in order to obtain cheaper tickets (328).

6.8.1. The distinction between time sensitive and non time-sensitive passengers

316. One possibility to address the preferences of the customer group mainly concerned by this merger could be to classify them as "time-insensitive" in line with the distinction in previous merger cases. However, in its previous decisions the Commission has already observed that the formerly "clear" distinction between time-sensitive and non timesensitive passengers had become more and more blurred in recent years, not least because the price-sensitivity of business customers has increased over time due to new possibilities to book cheaper flights with low-frills carriers (329). These carriers operate on certain routes between city pairs with a relatively high number of frequencies and have therefore become more attractive for business customers. In fact up to 40% of the "business" customers responding to the Commission's survey mentioned "best price" as a reason why they had purchased the ticket. This proportion is less than the 62% of leisure passengers that mention this same factor but it is still significant. Furthermore it is the second most often cited factor, after "Best time" (48%). Closeness of airport to final destination was mentioned only by 30% of business passengers and "loyalty programs" by only 5%. This underscores the limited importance of loyalty schemes for business customers on the shorthaul routes assessed in this case.

317. Moreover, neither Ryanair nor Aer Lingus price discriminate between time-sensitive and non-time sensitive passengers by offering unrestricted alongside restricted tickets. The consequences of this are twofold. Firstly, it raises the question how, if appropriate, to identify and separate time sensitive from non time- sensitive passengers for the purposes of this assessment. Secondly, it is less obvious why a separate assessment should be relevant in the absence of a tailor made service with the corresponding price discrimination between these two categories of passengers. This is because, in the absence of this discrimination, passengers would be affected across the board, whether time sensitive or non time-sensitive, by any price increase or decrease by either Ryanair or Aer Lingus (330).

318. A separate assessment for time-sensitive and non time-sensitive appears all the less relevant in this case as Ryanair and Aer Lingus have both deliberately chosen not to explicitly discriminate between these categories of passengers. Aer Lingus changed its pricing policy, moving away from a complex pricing structure similar to those network carriers continue to apply today towards a simplified pricing structure with one-way restricted tickets only.

319. In the light of the foregoing, the Commission considers that the distinction between timesensitive and non-time sensitive passengers is not relevant with a view to the assessment of the present transaction.

6.8.2. The distinction between business and leisure passengers

320. In the notification and various submissions thereafter, Ryanair argues that the distinction between business and leisure passengers is relevant in the assessment of this case. In support of its arguments, Ryanair refers to the recent Flybe/BA Connect decision of the Office of Fair Trading, where the OFT assessed to some extent separately the effects for business passengers of the Flybe/BA Connect merger.

321. The Commission observes that, similar to the distinction between time sensitive and nontime sensitive passengers, neither Ryanair nor Aer Lingus seek to discriminate between business and leisure passengers. Tickets are for sale under the same terms for both groups of passengers (331). Both airlines have chosen to sell their tickets on a one-way basis, to the effect that they cannot discriminate passengers by means of the Saturday night rule. Similar to the distinction between time sensitive and non-time sensitive passengers, this has two fold consequences. Firstly, it raises the question how, if appropriate, to identify and separate leisure from business passengers for the purposes of this assessment. Secondly, it is questionable why a separate assessment should be relevant in the absence of a tailor made service with the corresponding price discrimination between these two categories of passengers. This is because, in the absence of this discrimination, passengers would be affected across the board, whether business or leisure, by any price increase or decrease by either Ryanair or Aer Lingus.

322. The Commission further notes in this regard that both Ryanair and Aer Lingus carry business and leisure passengers. The Customer Survey conducted by the Commission confirms that substantial proportions of both leisure passengers and business passengers consider Ryanair's services as a possible alternative for those of Aer Lingus and vice versa. Over the full sample 25.6% and 20.7% of Aer Lingus and Ryanair passengers, respectively, travel for business purposes. Furthermore 51.9% of Ryanair business customers had considered Aer Lingus as an alternative. Conversely 43.2% of Aer Lingus business customers considered Ryanair as an alternative. In fact this proportion is slightly higher than the proportion of leisure (holiday) and "visiting relatives" customers that considered Ryanair as an alternative, 42% and 40%, respectively. For Ryanair's passengers, the percentage that considered Aer Lingus is broadly similar, amounting to 43% for leisure (holiday) passengers and 33% for visiting friends/relatives passengers. There does not appear to exist, therefore, a clear relationship between differences in the needs or preferences of business and leisure passengers, on the one hand, and the differentiation between the services of Ryanair and Aer Lingus, on the other.

323. In the light of the foregoing, the Commission considers that it is not appropriate for the purpose of this case to define separate markets for business and leisure passengers (332).

6.8.3. Market for "late booking customers"?

324. Confronted with the difficulties regarding time sensitive versus non-time sensitive passengers and business versus leisure passengers described in Sections 6.8.1. and 6.8.2., the Commission explored whether any meaningful passenger groups could be identified on the basis of time lapsed between booking and departure for market definition purposes. In particular, the Commission considered the proportion of passengers who booked their flight seven days (333) or less before departure, hereinafter "late bookers".

325. It appears that the proportion of late bookers varies between continental "leisure" routes (late bookings usually between 5 and 15%) and the nine Ireland-United Kingdom routes (late bookings between 10% and 30%). Further, there are no major differences in the share of late bookings for Ryanair and Aer Lingus on individual overlap routes (see in more detail the competitive analysis of the individual routes).

326. However, it is unclear if differences in ticket prices, whether for early or late bookers, reflect real price discrimination by the carriers (334). For Ryanair as well as for Aer Lingus, cheap fares remain available on certain flights almost up until departure. This is in particular the case for flights where demand (load-factor) is lower. Conversely on other flights, several weeks ahead of departure, only fairly expensive fares are available for booking. This is the case of flights where demand is high (335).

327. Accordingly, the Commission concludes that no relevant distinction can be made with a view to assessing the competitive effects of the proposed transaction on the basis of the "time between booking and departure" approach.

6.8.4. Separate market for "price insensitive" ("low-frills") customers?

328. Ryanair and Aer Lingus' customer appear to be more price-sensitive than, for example, customers of network carriers or business-customer oriented airlines (such as CityJet or Aer Arann), who offer a higher service level but charge higher prices than the Merging Parties. The Commission has therefore also considered distinguishing customers who are rather price insensitive (one typical group being again business customers) and customers who are more price sensitive ("low-frills" customers). Indeed, the fact that the present transaction, unlike previous merger cases, concerns mainly customers flying with low-frills airlines, is an important difference to previous merger cases. The Commission's market investigation has, for instance, shown that customers of low-frills airlines are usually more ready to travel to remote "secondary" airports in order to obtain cheaper tickets (336).

329. However, although the Commission again acknowledges that "price-sensitive" customers (typically flying with low-frill airlines) and "price-insensitive" customers (typically flying with network carriers) may have different preferences and needs, it is not appropriate to define two separate markets along this distinction in the present case. Indeed, it is questionable whether "price insensitive" customers stricto sensu exist at all, and it is in any event not possible to define a "dividing line" which distinguishes rather "price-sensitive" from rather "price-insensitive" customers.

6.8.5. Conclusion

330. In the light of the above, the Commission concludes, that, although an overall majority of Ryanair and Aer Lingus' "low frills" customers on the overlap routes are rather non timesensitive and fly for private purposes (337), it is not appropriate to define separate markets for different categories of passengers, whether according to the time sensitive/non-time sensitive distinction, the business/leisure passenger distinction or the "time between booking and departure" approach.

331. This conclusion, however, is without prejudice to the actual existence of different groups of passengers. Any differences between passenger groups are considered as appropriate in the competitive assessment for each of the relevant O & D pairs in Section 7.9.

6.9. Conclusion

332. On the basis of the above the Commission finds that for the purpose of the assessment of the proposed transaction the relevant markets are the markets for direct (338) scheduled passenger air transport services between a given point (or region) of origin and a given point (or region) of destination (339).

333. In view of the services currently offered by the Merging Parties, the proposed transaction would lead to actual overlaps between the Merging Parties on the following markets:

(i.) scheduled passenger air transport services between Dublin and Alicante, Barcelona, Berlin, Bilbao/Vitoria, Birmingham, Bologna, Brussels, Edinburgh, Faro, Frankfurt, Glasgow, Hamburg/Lübeck, Krakow, London, Lyon, Madrid, Malaga, Manchester, Marseille, Milan, Newcastle, Paris, Poznan, Riga, Rome, Salzburg, Seville, Tenerife, Toulouse/Carcassonne, Venice, Vienna/Bratislava and Warsaw;

(ii.) scheduled passenger air transport services between Cork and London as well as Cork and Manchester;

(iii.) scheduled passenger air transport services between Shannon and London.

334. Further, the proposed transaction may raise concerns on the following markets where currently only one of the Merging Parties operates:

(i.) scheduled passenger air transport services between Dublin and Aberdeen, Alghero, Almeria, Amsterdam, Athens, Biarritz, Bournemouth, Bremen, Bydgoszcz, Billund, Blackpool, Bordeaux, Bristol, Budapest, Cork, Doncaster, Dubrovnik, Düsseldorf, Eindhoven, Friedrichshafen, Gdansk, Geneva, Gothenburg, Jersey, Karlsruhe, Kaunas, Lanzarote, La Rochelle, Lisbon, Lodz, Malmo, Malta, Munich, Nantes, Naples, Nice, Oslo, Palma de Mallorca, Pisa, Porto, Prague, Pula, Rennes, Rzeszow, Santiago de Compostela, Shannon, Stockholm, Tampere, Trapani, Valencia, Vilnius, Wroclaw, Zurich;

(ii.) scheduled passenger air transport services between Shannon and Alicante, Barcelona, Biarritz, Birmingham, Bournemouth, Bristol, Brussels, Düsseldorf, Edinburgh, Faro, Frankfurt, Glasgow, Krakow, Lodz, Madrid, Malaga, Manchester, Milan, Nantes, Paris, Rome, Toulousse/Carcassonne, Venice, Wroclaw; flights between Cork and Alicante, Amsterdam, Barcelona, Berlin, Birmingham, Faro, Lanzarote, Malaga, Madrid, Nice, Paris, Prague, Rome, Tenerife, Warsaw.

7. COMPETITIVE ASSESSMENT

7.1. Introduction

335. The Commission has dealt with a number of airline mergers in the past years. However, the present case differs in some significant aspects from these past cases. All previous cases concerned mainly mergers of two carriers which had their main centres of operations at different airports, often in different countries and raised concerns on a relatively limited number of overlapping routes. In contrast, this merger concerns the two main airlines in Ireland with a significant base at the same airport, namely Dublin Airport. As a consequence, the merger raises concerns on a significant number of overlap routes, all with a common point of origin. Indeed, all affected routes in the present case are from or to Ireland.

336. Moreover, while past cases concerned airlines with complementary networks and operating models (340), the present transaction is the first case in which the Commission has to assess a merger combining two airlines which operate according to the low-frills/lowfares business model (341) and which are focused on point-to-point intra-European services. On a European level, the merger combines two of Europe's largest low-frills carriers. According to the Association of European Airlines, Ryanair and Aer Lingus ranked number one and three respectively in terms of weekly "no-frills" seats sold in summer 2006 (342).

337. Previous airline cases dealt mainly with mergers which combined complementary networks and operating models, e.g. network carriers operating with a hub and spoke model and charter/ tour operators (343), network carriers and smaller regional airlines (344) or, in some cases, two (smaller and larger) network carriers (345).

338. Both the fact that Ryanair and Aer Lingus have a large base at Dublin Airport, where they are the two largest airlines with operations of a similar scale, and the fact that they operate according to a similar model of point-to-point/low-frills services have a significant effect on the nature of competition that they exert on each other and that would be lost following the proposed merger. As set out below in detail, the strength of both airlines at the same airports combined with their low frills/low cost business model would increase the already high barriers to entry or expansion for actual and potential competitors (346). Since the nature of competition in the airline industry in general and between Ryanair and Aer Lingus in particular is not static but dynamic, this applies not only to those markets where both Merging Parties are currently active, but also to other routes that are operated only by Ryanair or Aer Lingus out of Dublin today (347).

339. In the competitive assessment set out below, the Commission first analyses the market shares on the actual routes on which Ryanair's and Aer Lingus' overlap (Section 7.2), and examines the degree of closeness of competition between Ryanair and Aer Lingus on the affected routes (Section 7.3) and whether both companies actually compete with each other in practice (Section 7.4). The Commission then describes the consequences of the merger for the actual competition between Ryanair and Aer Lingus (Section 7.5) as well as the consequences for routes where they are potential competitors (Section 7.6). The Commission then assesses the switching possibilities and buyer power of Aer Lingus' and Ryanair's customers (Section 7.7), as well as the existing barriers to entry and the likelihood of the existing competitors to compensate for the expected loss of competition (Section 7.8). The Commission then analyses the individual effects of the merger on each individual route-pair ("route-by-route analysis", Section 7.9) and finally discusses whether efficiencies are likely to outweigh the competitive harm (Section 7.10).

7.2. The merger leads to very high market shares on a large number of routes

340. Ryanair and Aer Lingus account for the largest share of passenger traffic at the airports of Dublin, Shannon and Cork. Ryanair and Aer Lingus accounted for [30-40]*% and [30- 40]*%, respectively of all passengers ex-Dublin in 2005, whereas British Midlands accounted for [less than 5]*% and Lufthansa for [less than 5]*%. Taking into account only intra-European traffic, the market shares of Ryanair and Aer Lingus amount to [40-50]*% and [30-40]*% respectively. Ryanair and Aer Lingus also serve the largest number of destinations ex-Dublin with over 110 destination airports in Europe for both airlines together and 80% of all intra-European traffic. The combined entity would be the largest short-haul carrier at Cork and Shannon airports. From both airports, they operate in total to around 20 destination airports, being by far the largest operator.

7.2.1. The merger creates a monopoly on 22 routes and leads to market shares above 60% on all remaining routes

341. The operations of Ryanair and Aer Lingus overlap currently on 35 routes (348). Table 2 summarises all overlap routes together with market shares (349) of the Merging Parties and their main competitors.

Table 2: Routes with existing overlaps between Ryanair and Aer Lingus with market shares of the Merging Parties and all existing competitors based on planned weekly seat capacity for Summer 2007 season (unless specified otherwise) (350).

<emplacement tableau>

342. On these 35 (354) overlapping routes, the transaction would be a merger to monopoly (355) on 22 routes and lead to very high combined market shares above 60% on further 13 routes. In the absence of competitors with considerable market shares on almost all routes, the merger would lead to very high Herfindahl-Hirschman-Index (hereinafter "HHI") levels. Even on those few non-monopoly routes where a number of different competitors are active, such as the London-Dublin route, the HHI level would reach [6000-6500]*, with a delta of [3000-3500]*. The concentration level created through this merger on all 35 overlapping routes would therefore be very high.

343. On 16 of the overlap routes, there are direct airport-to-airport overlaps between the Merging Parties' activities as both Merging Parties fly from the same airports at both ends of the O&D-pair (356). On the remaining 19 routes, there is a city-to-city overlap with Ryanair and Aer Lingus using different airports on the non-Irish side of the leg as listed in Table 1 in the Market Definition section above (paragraph 82) (357).

344. The high market shares do not only relate to "thin" routes, but to a number of important routes from and to Dublin. According to the report of Airport Coordination Limited (hereinafter "ACL") on distribution of slots at Dublin Airport for the Summer 2006 season ("Start of Season Report"), the following routes from Dublin (on an airport-to-airport basis) are the most important as regards their capacity: London - Heathrow, London/Stansted, London/Gatwick, Manchester, Paris CDG, New York JFK, Malaga, Birmingham, Amsterdam and Faro. The markets where the Merging Parties' activities overlap cover 8 out of these 10 most important routes to/from Dublin (while one of the remaining two most important routes is a long-haul route to New York) and their combined market shares are very high in all these markets (London - [70-80%]*, Manchester - [90-100%]*, Paris - [60-70%]* on the basis of total capacity but up to [80- 90%]* when connecting passengers are excluded, Malaga - [90-100%]*, Birmingham - 100%, Faro - 100% (358). These figures further illustrate the effects of the transaction on short-haul air traffic to/from Dublin as the Merging Parties' activities overlap not only on a large number of routes, but these routes include a number of the most important routes to/from Dublin.

345. The overall importance of the overlap routes is also demonstrated by the fact that the total number of passengers on only the 32 "overlap routes" to/from Dublin represented in 2006 around 70% of all passengers carried on intra-European routes to/from Dublin (359). The total passenger number on all overlap routes affected by the proposed merger amounts to more than 14 million (360).

346. Although the number of passengers and the frequency of operations on the overlap routes may vary (361), and although some of the routes are only operated by one airline on a seasonal basis (362), it remains that the merger will eliminate or significantly reduce customers' choice on all these routes, even thinner or seasonal ones. Even on "thinner" routes, customers can currently choose between two airlines which successfully operate on these routes. The Commission therefore has to ensure that also customers of "thinner" or seasonal routes are not harmed by the elimination of the main competitor.

347. It should also be noted that on the routes to important hub destinations, in particular London, Paris and Frankfurt, the market shares based on capacity underestimate the importance of Ryanair and Aer Lingus for point-to-point flights to the extent that the main competitors on these routes usually have a higher proportion of connecting/transfer passengers than the Merging Parties. Ryanair serves only point-to-point customers and even discourages their customers from using their flights to transfer to other flights. The change in business model of Aer Lingus has also led to a focus on point-to-point passengers, which means that the share of transfer passengers on these routes is usually significantly lower than for traditional network carriers (363). The assessment of the individual routes below provides more details on this issue.

7.2.2. The very high market shares on all overlap routes constitute evidence of a dominant position

348. According to the Horizontal Merger Guidelines, market shares and concentration levels provide useful first indications of the market structure and of the competitive importance of both the merging parties and their competitors. (364)

349. The Horizontal Merger Guidelines also state that according to well established case-law, very large market shares - 50% or more - may in themselves be evidence of the existence of a dominant market position (365). The Court of First Instance recently confirmed that although the importance of market shares may vary from one market to another, very large shares are in themselves, and save in exceptional circumstances, evidence of the existence of a dominant position (366). In a recent case concerning the airline sector, the Court of First Instance also stated that 'market shares held by the parties to the merger led the Commission to conclude that commitments should be offered on the markets affected and on which those parties enjoyed a market share of almost 50%, thereby respecting the presumption of dominance as laid down by the case-law' (367).

350. The Merging Parties would have very high market shares on all 35 routes on which their activities overlap. For 22 routes the merger would create a monopoly, depriving customers of the only available alternative, but also for all remaining overlap routes. Also on the other 13 routes, given the very high market shares of the merged entity, convincing evidence is necessary to conclude that the merger would not result in the creation or strengthening of a dominant position.

351. The Commission has carefully analysed whether there are circumstances which might exclude a dominant position in the present case despite the high market shares. The relevant factors for the Commission's assessment will be first discussed in general terms in the following sections (see Sections 7.3 - 7.8). Many of the factors considered by the Commission apply to a greater or lesser extent to all overlap routes.

352. In order to take due account of the particularities of each route, the Commission has also analysed the competitive situation and notably the main potential factors that might counteract the effects of the merger on a route-by-route basis (see Section 7.9).

7.3. Ryanair and Aer Lingus are closest competitors on all affected routes

353. Ryanair argues that the product offering of the Merging Parties are significantly different in that Ryanair offers a pure low-cost no frills service and focuses on generating demand from low fares whereas the product offering of Aer Lingus is more "mid-frills". In particular, Ryanair notes that Aer Lingus' services are generally more expensive and that it offers additional services such as seat allocation. In addition, Aer Lingus flies into primary airports, while Ryanair mainly serves secondary airports. A service between primary airports can be perceived as offering higher quality with respect to, for example, reduced overall travelling time for customers who wish to go to the city centre or the availability of business lounges.

354. The concept of "closeness of competition" may play an important role to better understand the competitive constraint different competitors exert on each other in oligopolistic markets (368).This concept is however of limited added value on markets in which from the outset only two competitors are active (such as on the 24 current duopoly markets affected by this transaction), since both competitors in a duopoly are "by nature" each other's closest competitors.

355. Even on those routes on which Ryanair and Aer Lingus do face one or more competitors, the distance in terms of market share to this competitor(s) is in most cases significant. Similarly, even on those markets, where one or more smaller competitors are active, the transaction would have a significant competitive impact, including in cases where the Merging Parties would not be the closest substitutes in terms of product characteristics. In other words, even if customer diversion (switching) between Ryanair and Aer Lingus were somewhat less than proportional to their market shares on some routes, the transaction would still significantly impede effective competition.

356. It is also important to stress that although an existing airline (such as easyjet) operating outside the affected markets might be "overall" closer to Ryanair than Aer Lingus as concerns certain elements (e.g. the service model) this does not put into question that Ryanair and Aer Lingus are each other's closest competitors on the routes actually affected and analysed in this decision (i.e. the routes from and to Ireland). That is to say, for the analysis of closeness of competition, the Commission must focus on those competitors who actually compete with Ryanair and Aer Lingus on the affected markets. However, it does not only compare the two merging firms (369), but also the remaining competitors or "rivals" in the affected markets (370).

357. It should be particularly noted that for two airlines to be considered as offering services which are close substitutes to each other it is not necessary that the two services are identical. Some services may be closer substitutes than others. What matters from a substantive competition perspective is the (high) degree of substitutability between the services of Aer Lingus and Ryanair. The fact that there are certain differences between Ryanair and Aer Lingus in terms of overall size, cost structure, average fares, service offering, served regions and brand, does therefore not put into question that both airlines are each others' closest competitors on routes from and to Ireland. This is further explained in the next sections.

358. The only current competitors of Ryanair and Aer Lingus on the overlap routes offering scheduled services are Aer Arann, Air Baltic, Austrian Airlines, BMI/bmibaby, CityJet (Air France), Clickair (Iberia), LOT, Lufthansa, Luxair, Norwegian Airline Shuttle, Sky Europe. It is noted that these competitors are in most cases network carriers or (equally business-customer and feeder-traffic oriented) regional airlines. On some routes also charter airlines (e.g. XL, My Travel, FirstChoice) are active (371). Hence, even those routes where competitors would remain post-merger, Ryanair and Aer Lingus are each other's closest competitors. Even if on some routes a competitor with a more similar service and operating model is active (such as e.g. bmibaby, SkyEurope, Clickair or Norwegian), the market shares of these carriers on the overlap routes are in most cases significantly smaller, and none of these competitors has significant operations or even a large base in Dublin. Hence, also on that basis, Aer Lingus and Ryanair can also be regarded as closer competitors than these carriers (372).

359. The Commission will nevertheless explain in more detail why Ryanair and Aer Lingus are regarded as the two closest competitors on those affected routes where they face a competitor.

360. Apart from Ryanair's and Aer Lingus similar size and market position in Ireland (1) and their similar business model (2) and cost structure (3), the fact that they both have a significant base in Dublin (4) makes Aer Lingus and Ryanair the closest competitors on routes out of these Irish airports. Both airlines are also viewed by customers as the two closest competitors (5). These individual factors will be analysed below in the given order.

7.3.1. Both companies have equally strong position in Ireland and face few competitors

361. Ryanair and Aer Lingus have a particularly strong position in Ireland373. On almost all routes, they are the only two or the two main carriers active, often with comparably high market shares. Even on routes where both airlines face competition, the position of their competitors is in most cases significantly weaker than the Merging Parties' position. Only on three of all 35 overlap routes, is there a competitor with a higher "nominal" market share than one of the Merging Parties. On at least two of these three routes, however, the "larger" competitor carries a high proportion of connecting passengers. If only looking at "point-to-point"-services, the Merging Parties would be the largest "point-to-point" carriers also on these routes (374).

362. Ryanair and Aer Lingus are also each other's closest competitors as regards their ability to serve routes from/to Dublin. In Dublin, Aer Lingus and Ryanair currently base 23 and 20 larger short-haul aircraft respectively, while the other Dublin based airlines, Aer Arann and CityJet, have only 4 and 3 small aircraft (less than 130 seats) based in Dublin, and other airlines overnight only one aircraft at Dublin Airport (375). As described in more detail below, as these latter two carriers concentrate on a different type of services (regional and feeder/business traffic), Aer Arann's and CityJet's fleets consist of airplanes which are much smaller than those operated by the Merging Parties.

7.3.2. Both companies' business model has become increasingly similar and is markedly different from most of their competitors

363. As concerns the main characteristics of an airline, a distinction can be made between different service and operating models. As pointed out above (376), airlines are highly differentiated and do not often fulfil all characteristics of a particular category ("full service" or "low frills" model or pure network carriers and point-to-point carriers).

364. Ryanair argues that the product offerings of the Merging Parties are significantly different in that Ryanair offers a pure low-cost no frills service and focuses on generating demand from low fares, whereas the product offering of Aer Lingus is more focused on quality than on offering low prices. Ryanair notably stresses that Ryanair's average fare on shorthaul flights is significantly lower than Aer Lingus' average fare, that Aer Lingus offers additional services such as seat allocation, and that Aer Lingus flies into primary airports, while Ryanair mainly serves secondary airports. According to Ryanair, a service between primary airports can be perceived as offering higher quality with respect to, for example, reduced overall travelling time for customers who wish to go to the city centre or the availability of business lounges.

365. The Commission does not contest that there are a number of elements which may differ in the service from Aer Lingus and Ryanair. Aer Lingus offers some additional services which Ryanair does not offer (such as business lounges or a frequent flyer programme). In addition, Aer Lingus flies into primary airports, while Ryanair mainly serves secondary airports.

366. The market investigation has also clearly confirmed (377) that there is still some differentiation in terms of brand and service offering between Aer Lingus and Ryanair. Indeed, Aer Lingus' is generally perceived as offering a higher service level, and business and corporate customers seem to have a certain preference for Aer Lingus, while also regularly flying with Ryanair (378).

367. However, it should be noted that despite being a former flag carrier, Aer Lingus has undergone a major restructuring process, thereby changing its business model on its intra- European operations (379) and repositioning itself as a low-frills airline focused on point-topoint services. Indeed, for their short-haul operations, the main elements of the service and business model of Ryanair and Aer Lingus have become increasingly similar to each other. Both companies present themselves as low cost/low fares carriers (380), and operate a single ticket class, with no or little on-board frills or services on intra EU flights. They apply a similar unidirectional pricing model for one-way, one-class and not refundable tickets. They both realise most of their bookings from the web rather than via travel agents. Recently, Aer Lingus for example also followed Ryanair in charging an extra fee (€ 5 to 8 per piece) from its customers for check-in luggage, and Aer Lingus no longer offers free meals on short-haul flights. Ryanair's newly introduced express check-in makes in turn Ryanair more attractive for business customers. Although Aer Lingus' services might be regarded as "mid-frills" when compared to its direct competitor Ryanair (mainly because of the remaining service differentiation in terms of serving primary airports) and although Ryanair's services might be called "no-frills", the Commission considers, for the purposes of this decision, that both airlines can be considered "low-frills" carriers, since following the latest adaptations of Aer Lingus' service model, the services included in the Aer Lingus base fare are broadly in line with those included in the Ryanair base fare and in stark contrast to traditional "full service" airlines such as British Airways or Lufthansa.

368. Indeed, although Aer Lingus's fares are usually higher and Aer Lingus customers pay a certain "premium" for the additional services (381), it must be stressed that the business model of other competitors on the overlap routes is overall more remote from Ryanair's model than Aer Lingus' business model.

369. The competitors Ryanair faces on its routes are mainly (382) carriers focused on "feeder- oriented" traffic and on business customers (e.g. CityJet, British Airways or Lufthansa) and smaller airlines with regional activities (Luxair, Loganair). Although not identical, Aer Lingus' business model is therefore closer to Ryanair's model than to the business customer/feeder-traffic oriented model of almost all its competitors on the overlap routes. 370. The Commission observes that Ryanair and Aer Lingus face only on a few routes other low-frills/point-to-point airlines as competitors (383). Europe's largest no-frills airlines (such as easyJet, Air Berlin or the regional (384) low-frills carrier FlyBe) are not present on any of the overlap routes.

371. The same consideration applies to the fact that Aer Lingus' average prices are higher than Ryanair's prices. Even if Aer Lingus' prices are usually (but not always (385)) higher than Ryanair's prices (386), it should be noted that for a comparison it is important to know exactly which taxes and charges are included in the respective prices (e.g. airport charges, government taxes, credit card fees, baggage fees, insurance fees, "passenger service charges" and other charges (387)). According to the fare data made available to the Commission by Ryanair and Aer Lingus, the difference between Ryanair's and Aer Lingus' average fares amounted only to around €30 in 2006. In any event, both carriers' fares are usually well below the price level of the competitors they face on the respective routes. According to Ryanair's own estimates, the prices of full-service network carriers (such as Air France, British Airways or Lufthansa) are much higher, and far above Aer Lingus prices (according to Ryanair, between € 216 for Air France, € 225 for Lufthansa and € 268 for British Airways (388)). Moreover, when comparing Ryanair's and Aer Lingus' fares, it must be taken into account that Aer Lingus' often higher price reflects some quality advantages, such as the fact that Aer Lingus flies to primary airports, which saves transfer costs for the customer (389). The price correlation analysis conducted by the Commission provides additional evidence that, at least with respect to a majority of overlap routes, Aer Lingus and Ryanair prices have also systematically moved together over time. As the price difference reflects (perceived) differences in quality, a price change by one carrier would affect demand for services provided by the other on the same route. Likewise at given prices, increases in frequencies or improved flight schedules by either of the Merging Parties can be expected to divert passengers to the other (390).

372. The Commission further observes that Aer Lingus' higher prices also reflect the remaining somewhat higher service level of Aer Lingus described above (e.g. serving primary airports, business lounges, higher service orientation and less rigid interpretation of customer service rules (391)). For the customer who takes into account the time and price for the transfer to the airport when purchasing tickets, the effective price difference may therefore be lower than the mere difference of the sales price.

373. The Commission finally notes that also independent websites which serve to compare and inform about airlines regularly describe Aer Lingus' transformation process into a "lowfrills" (or "low cost") carrier (392).

7.3.3. The Merging Parties have lower business costs than most of their competitors on the overlap routes

374. Ryanair claims that Aer Lingus cannot be close to Ryanair because of its much higher cost Basis (393). The Commission acknowledges that Aer Lingus, until recently a state-owned flag carrier, has not reached Ryanair's unmatched low operating costs. However, this appears also to be true for the other main "low cost" airlines in Europe. Compared to the airlines it faces on the relevant markets, Aer Lingus costs are generally significantly lower and rather within the group of "low cost carriers" than within the group of network carriers.

375. While providing some level of product differentiation relative to Ryanair (higher service level resulting in higher costs), Aer Lingus has lower operating costs than other major airlines in Europe, and than the other airlines operating on the affected routes. An independent study, the IATA's "Airline Cost Performance" report (394) provides a comparison of the operating costs of major network carriers and a number of low-cost operators. The study was published in July 2006 and estimates operating costs per available seat kilometre (ASK) in 2004 for network airlines, as well as for easyJet, Ryanair and Virgin Express. Comparing costs between airlines may be difficult because of the significant share of fixed and shared costs in airlines' operating expenses, which means that longer routes tend to generate lower costs per kilometre. At the same time, many assets are shared for short- and long-haul flights. IATA, in collaboration with McKinsey, has therefore adjusted its cost estimates for stage-length (i.e. flight length) and for the higher seat density of low-cost airlines. Aer Lingus is not included in the IATA analysis; however, the airline reports total unit costs exclusive of depreciation and lease costs of 5.28 €ct/ ASK in its published accounts (2005). Allowing for 11% ownership costs (395) leads to approximately 5.9 €ct operating costs per ASK. The IATA cost data and the Aer Lingus cost estimate are shown in Graph 1 below:

<emplacement tableau>

376. Hence, while Ryanair is the clear cost leader among European airlines, Aer Lingus, together with easyJet and Virgin Express (396), have significantly lower operating costs than the major network airlines, based on the IATA/Aer Lingus data. It is important to note that, in contrast to Aer Lingus, neither easyJet nor VirginExpress (or its successor Brussels Airlines) compete against Ryanair on any overlap route out of Ireland.

377. In response to the Statement of Objections Ryanair points out that Aer Lingus' cost data is not comparable to other European LCCs as it includes also long-haul flights, which generally have lower costs per ASK (397). This is a justified caveat, although it should be noted that 87% of Aer Lingus passengers travel on short-haul routes (398). Ryanair includes in its reply to the Statement of Objections a chart, according to which Aer Lingus' shorthaul costs are "somewhat along the lines of" 8 ct/ASK, although no details are given as to how this number is calculated (399). More generally, the IATA report cautions that no widely accepted cost reporting standards exist and that cost structure reporting may not always be consistent between network airlines and LCCs. However, even when allowing for a margin of error or, for that matter, assuming that Ryanair's estimate was accurate, the data indicate that Aer Lingus has successfully adopted a low-cost business model (albeit with elements of product differentiation) and now operates with unit costs well below traditional network airlines.

378. The evolution of Aer Lingus unit costs over time further highlights its gradual migration from a "traditional" to a low-cost business model:

<emplacement tableau>

379. In conclusion, the cost data available to the Commission suggest that Aer Lingus is among the closest competitors to Ryanair even in terms of unit costs. Its restructuring towards a low-cost business model has enabled it to operate profitably and on a sustained basis in head-to-head competition with Ryanair in particular from its Dublin base.

7.3.4. Both airlines operate with a large a base at Dublin.

7.3.4.1.General advantages from operating with a "base"

380. As set out above401, the term "base" is used in the airline industry to characterise airports where airlines base a certain number of aircraft and on which they concentrate their operations, and from which they operate several routes (402). At a "base" airport airlines typically concentrate a certain number of aircraft overnight; in addition to "overnighting" these aircraft; the airlines usually entertain additional facilities and services at the base airport, such as maintenance services, customer care services, ground-handling services or stand-by planes. A base also allows airlines to avoid accommodating their staff and crews in a hotel. Establishing a "base" is therefore more than just parking an aircraft overnight. It requires the commitment of infrastructure, personnel and equipment (with the concomitant capital expenditure) at an airport, which are used to operate several routes from that airport.

381. Concentrating aircraft and traffic at certain bases can generate considerable cost savings and increased flexibility. In comparison to providing a point-to-point service on a single route without a base at either end, base operations provide numerous competitive advantages for the routes operated from that base. There are mainly two effects which arise from base operations. First, the base allows realising cost savings due to economies of scale and scope. Secondly, there is a "competition effect", in that the airline can react more easily and more quickly to changes in supply and demand on routes out of this base. Dynamic reaction to the behaviour of competitors is therefore easier when operating from a base than outside a base.

Economies of scale and scope

382. The Commission's market investigation (403) has confirmed that carriers benefit in particular from the following economies of scale and scope by operating from a base:

* Sales and marketing costs: the costs associated with sales and marketing are largely fixed. Recovering these costs over a larger fleet of aircraft and a wider portfolio of routes therefore also translates into lower unit costs.

* Customer service/ground handling facilities: Sufficient scale allows carriers to operate their own dedicated facilities such as own check-in and customer information counters, etc. at an airport, thereby providing passengers with higher levels of service at a lower cost (404).

* Disruption costs: a larger pool of aircraft allows a carrier to reduce the cost associated with cancellations and to increase dispatch reliability by giving the carrier greater schedule recovery options. For example Aer Lingus has a stand-by aircraft at Dublin. Ryanair has pointed out that it does not base stand-by aircraft at a certain airport. Instead, it operates [2-6]* aircraft rotating across all of its 20 bases (405).

* Slots and stands: To start serving a particular route an airline requires departure and arrival slots as well as stand/gate facilities at the origin and the destination airports. When there are capacity constraints, airlines that have many routes from the airport will be able to substitute slots from one route to another, contrary to airlines with few or no existing routes, which will find entry by shifting from an existing route from the same airport more difficult (406).

383. Large scale carriers may also benefit from a stronger negotiation position of competitors with a base (407):

* Ground handling: A base with multiple aircraft allows a carrier to obtain better terms for ground handling reflecting the greater volume.

* Negotiations with airport and regulators: A strong negotiation position, due to the overall amount of traffic a carrier brings to an airport may allow the carrier to gain access to preferential facilities, e.g. gates in close proximity to main terminals, lower airport charges and ultimately lead to an increase in product quality at little or no additional cost (408).

* Procurement: Sufficient scale both absolute as well as relative to other carriers at the airport may give a carrier more bargaining power in negotiations with suppliers, such as catering and maintenance companies, leading to cost advantages.

384. By having a number of aircraft based at the same airport, airlines can also benefit from economies of scope which may derive from the ability to easily switch assets between routes, thereby increasing the competitiveness of the routes operated from the base.

* The more flights a carrier operates from an airport, the easier it will be for that carrier to maximise the aircraft utilisation because there are more combinations of flights available to fill an eight or ten hour schedule. For example, Aer Lingus' short-haul network from Dublin varies in duration from approximately 50 minutes (e.g. Dublin- Manchester) to three hours and 10 minutes (e.g. Dublin-Rome). Operating rules set a 10 hour limitation on its crew utilisation. Therefore, as Aer Lingus has a base at Dublin it can easily operate a flight to Rome and Manchester with the same crew and equipment (assuming a turnaround time of 40 minutes). A larger fleet also allows the carrier to pool reserve requirements for pilots and cabin crew and to maximise the working hours of crew members, who can only work eight to ten hours per day and return to their home airport in the evening (thereby avoiding hotel costs). For the same reason (limited working-hours for crew members), airlines operating from a base airport can more easily operate longer than ten or eleven hours a day thereby maximising their aircraft utilisation time, since they can easily exchange the "morning crew" during the day at the base, which is not possible for airlines operating without a base.

* A portfolio of aircraft also allows improved aircraft utilisation by reducing overall fleet downtime in order to meet punctuality requirements since any temporary delays can be absorbed by using the ground time allowed for other aircraft in the portfolio (409).

* Carriers operating from a base can use the early morning slots at its base airport which are economically important since the demand for flights is particularly high at that time, in particular for routes where a significant majority of passengers originate at this end of the route. If an aircraft had to fly to Dublin from another airport or base in the morning, the precious early morning slots would be lost (410).

385. Savings from operating from a base are usually higher at the beginning (e.g., vis-à-vis a point-to-point operation or when comparing a base with one or two aircraft with a base with 10 to 15 aircraft) and can decrease when a certain size is reached (411).

386. Ryanair does not deny the existence of economies of scale as such. However, in its view they do not arise from base operations but rather from overall size. Cost efficiencies are, according to Ryanair, determined by the airline's overall operations and scale (412). Ryanair, however, acknowledges that operating with base airports allows operational complexity to be reduced, as technical support can be concentrated at such bases (413). In its submission on the efficiency gains which Ryanair expects from the merger, it points to the benefits which would arise from base operations. For instance, Ryanair argues that the merger would "alter the present base structure and [Aer Lingus'] schedule so that no crews are overnighted away from base. This would also reduce the crew ratio per aircraft." (414)

387. A study submitted by Ryanair (RBB) that allegedly shows that for Ryanair operating costs do not decrease with a base does not put into question that such advantages exist for airlines in general, as evidenced by the market investigation (415).

388. In any case, Ryanair's submission reflects its own business model of operating a network of different bases across Europe. The density of this network and its overall size may indeed give Ryanair the flexibility to spread (some) of the cost across its entire network. However, other low-cost carriers (notably the few competing with Ryanair on the overlap routes) do not operate with a similar network of base airports (416). As regards such competitors, scale economies would only arise at their respective base airports. Finally, the fact that some of Ryanair's bases are rather small does by itself not imply that base advantages would not arise at all at such airports. As such secondary airports often grant strong incentives to attract new airlines, entering even at small scale (for instance during a trial period) is encouraged (417).

Increased flexibility to adjust to fluctuations in demand

389. Apart from economies of scale and scope, larger operations at one base airport also allow to react more quickly to shifts at the demand side. The dynamic nature of competition in the airline sector is shown by the fact that airlines constantly adjust the capacity (and routes operated) in their network. In particular Aer Lingus and Ryanair shift their frequencies and routes from their base in Dublin regularly, leading to significant fluctuations in daily frequencies and routes (418).

390. Two types of "demand shocks" (that is, a sudden increase or decrease in demand for the services provided) may be distinguished. Certain shifts in demand, for instance changed customer preferences or an increased overall income, can affect all routes out of an airport. Ryanair has indeed pointed out that the high growth rate of the Irish economy had a positive effect on overall demand for flights out of Ireland. This would also be a reason for the fast increase in the frequencies of existing routes and opening of new routes over recent years. On the other hand, it should be acknowledged that once investments have been made at a particular airport, a base operator could also be more vulnerable to negative demand shocks (i.e. a sudden overall demand decrease) (419).

391. The second type of changes in demand is route specific and does not affect all routes in the same way. Route specific demand fluctuations can arise due to seasonal demand for holiday destinations (sun and ski destinations). In the case of route specific "demand shocks", having a wider route network at a base allows the airline to adjust capacity to match such fluctuations in demand. The airline can more easily reallocate planes, slots and crew, between the different routes in response to changes in relative demand or profitability without incurring significant additional costs or risks. An airline which does not have a significant presence at the airport would find it more difficult and timeconsuming to react to such an increase in demand on a certain route. Increasing frequencies or opening a new route from an origin where the airline has no such presence would require higher start-up investments (in particular for marketing and the necessary access to customers in the region) and more time than from an origin where the airline has already established a base (420). Thus, a non-base carrier could not react quickly to a sudden increase in demand on a particular route in the same way as the Merging Parties.

392. A carrier with large scale operations at an airport has also advantages in generating demand on routes originating at that airport. The main source for this advantage appears to be inherent marketing advantages. If a carrier operates more than one route out of an airport, the more likely it is that consumers are generally aware of the carrier, and that they will consult its website when considering who to fly with (421). Carriers with many routes from an airport can reduce costs of advertising, as these costs can be spread over more routes.

Most carriers use base airports

393. The Commission's investigation has shown that non-base operations (e.g. operating triangular or so-called "W-flights") are rare in the airline industry. Airlines have confirmed that it is uncommon and in most cases an economic disadvantage to operate routes which are not connected to a base (422). For example, there are only very few routes from or to Dublin on which Aer Lingus faces a competitor that does not operate from a base (423).

394. Ryanair itself operates very few "non-base routes" (i.e. routes without a base at one end). In fact, of the almost 450 routes operated by Ryanair, there are only four non-base routes (i.e. less than 1% of Ryanair's entire network) (424). Similarly, from easyJet's more than 250 routes, only two are not connected to one of easyJet's bases (425). Ryanair has stated that it seeks over time to expand the number of bases which it has in Europe and to link those bases with as many of the destination airports as is commercially sensible (426). This leads to a geometric growth pattern, linking bases with each other (427). Ryanair states that it aims "to benefit from the cost savings and economies of scale arising from our route development strategy 'connecting the dots'". (428) Base advantages can increase with the size of the base

395. A large majority of competitors confirmed that operating from a base has significant advantages compared to non-base operations (429). The investigation has also confirmed the view that a larger base may in general increase the flexibility and cost savings (430). However, the Commission's market investigation has shown that different airlines have different estimates about the "optimal scale" of operations at a base (431). This confirms the Commission's own assessment that in practice the minimum viable size for a base will depend on a number of factors, including the geographical proximity of that base to other established bases of the same carrier, the density of the airline's network, and the nature of the airport where the base is established. For example, for a secondary/regional airport, the absolute minimum number of aircraft will be lower if the airport offers incentives to the carrier to induce it to establish a base there.

7.3.4.2.Both competitors have similar advantages from their large base in Dublin

396. As noted above, the fact that Ryanair and Aer Lingus operate from a large base at Dublin Airport further supports the fact that both airlines are each other's closest competitors. Ryanair has concentrated its traffic at 20 "base airports" all over Europe. After London- Stansted, Dublin is, with 20 aircraft, Ryanair's second most important base. Ryanair's base accounts for about [30-40]*% of all passengers in Dublin. Dublin is also Aer Lingus' principal airport where about 90% of its entire fleet is based (432). Aer Lingus has currently based 23 short-haul aircraft at Dublin from which it serves around 70 routes to 15 countries across Europe. Its base accounts for approximately [30-40]*% of all passengers in Dublin (433).

397. Given that both carriers have a significant base at Dublin Airport, with a similar fleet size, a similar share of passengers and roughly a comparable number of routes out of that airport, it can be assumed that they both benefit to the same extent from the described advantages of having large scale operations at Dublin. In particular, their bases at Dublin Airport increase Aer Lingus' and Ryanair' abilities and incentives to offer new routes or adjust capacity on their existing routes, in order to develop an even wider portfolio with more options for capacity redeployment and allows them to compete more effectively, in a way which a carrier with only one (or a few) route(s) into Dublin would not be able to match.

398. At Cork, the same advantages apply in principle, however on a significant smaller scale. Aer Lingus has based 4 Airbus A 320 and operates 17 scheduled destinations. Ryanair operates four routes out of Cork with one Boeing 737-800 based aircraft. Ryanair has indicated that it would base more aircraft at Cork if an agreement can be reached with the airport. In the case of Shannon only Ryanair operates with a base as it has based four Boeing 737-800 serving 26 destinations from that airport. Aer Lingus operates into Shannon from Dublin and London without having based an aircraft at Shannon for the purpose of short-haul flights.

399. Ryanair strongly contests that the fact that Aer Lingus and Ryanair operate from strong bases in Ireland confer them any advantages over other airline and can be regard as an element that qualifies them as closest competitors. Ryanair notably argues that other competitors also have bases at Dublin (see Section 7.3.4.3) and that competitors operating from bases outside Dublin are equally close to the Merging Parties (see Section 7.3.4.4).

7.3.4.3.Competitors with bases in Dublin exert a limited constraint

400. Ryanair has argued that various airlines base aircraft at Dublin overnight, including Aer Arann, BMI/bmibaby, Lufthansa, CityJet/Air France, LOT, Malev, SAS and British Airways. Thus, according to Ryanair, these carriers should be considered as equally close to competitors who operate with a base. They also argue in their reply to the Statement of Objections that BMI should be considered to have a base at Dublin Airport (434). However, as explained above, while overnighting one aircraft at Dublin allows an airline to offer an early morning departure, this is not akin to establishing a dedicated "base" at Dublin. An airline which only overnights aircraft does not benefit from the other scale and scope advantages of a base nor from the flexibility of switching between routes, and opening new routes (435) which is a further reason for airlines to operate from a base. In line with these results from the market investigation and from competitor's own replies (436), the Commission therefore considers the parked aircraft of BMI, Lufthansa, LOT, Malev, SAS and British Airways do not offer the same advantages as airlines operating with a "base" would have.

401. The competitors which might be considered as having a base in Dublin are Aer Arann and CityJet, who operate with 4 and 3 smaller jets respectively from Dublin and consider themselves as having a base in Dublin (437). However, given their small scale of operations to and from Dublin, it is unlikely that they can realise similar cost savings as the Merging Parties. CityJet and Aer Arann have 3 BAe 146 aircraft (93 seater) and 4 ATR 42 and 72 turboprop (up to 66 seats) based in Dublin, respectively. With a maximum flight length of up to around 1.5 hours Aer Arann cannot compete effectively on routes to Continental Europe (438). Smaller aircraft lead to higher cost per passenger and - in the case of turboprop aircraft - to a shorter maximum travel distance (439). They are not suitable for many services offered by Aer Lingus and Ryanair out of Dublin. Both carriers have specialised on routes and services where they face limited competition from the Merging Parties. Aer Arann does not operate on any of the overlap routes. As an Air France subsidiary, CityJet provides mainly feeder services into Air France's Paris hub. The only other service out of Dublin which is not directly linked to Air France's network is the London-City route. CityJet's "full service" operation is specialised in time-sensitive business customers (440).

402. Moreover, Aer Arann's and CityJet's limited number of services do not allow to re-deploy capacity from one route to another in reaction to demand shifts as easily as Aer Lingus and Ryanair can do at Dublin Airport. Both companies account only for a fraction of total passengers at Dublin (441). Apart from Ryanair and Aer Lingus, no other airline has a share of passengers from Dublin of more than 5%.

7.3.4.4.Competitors with a base at the destination airport are not equally close competitors

403. Ryanair has argued that even if base advantages existed, it would not be relevant whether the base is located in Ireland or at the other end of the route. Such an operator would be equally close to the Merging Parties since it would equally benefit from the cost advantages of a base and would have a similar flexibility to switch capacity to a route from this airport to Dublin, if that is commercially attractive.

404. The Commission does not contest that the scale advantages of a base are, in principle, the same at both ends of a given route. However, for the assessment of the competitive situation, the Commission observes that notably the "competition effect" (increased flexibility in particular in Dublin) confers a specific advantage to the Merging Parties which is relevant for the competitive assessment in the present case.

405. It must first be observed that the competitor on all of those 12 overlap routes does not always operate from a base at the other end. According to Ryanair (442), the competitors of Ryanair/Aer Lingus do, for example, not operate from a base on the routes from Dublin to Faro, Seville, Poznan, Tenerife, Salzburg and Bologna.

406. Secondly, while it is true that cost advantages from a base may also accrue if the base is at the destination airport, for the routes covered by this investigation, some of the cost savings are more significant for a base in Dublin. In part this is because there is an asymmetry of traffic origination on many of the overlap routes. On at least 15 of the 35 routes, the majority of the customer's originate in Ireland (443). Accordingly, a competitor without a base in Dublin is likely to exercise a more limited competitive constraint even if it had a base at the destination airport, as most foreign carriers are likely suffer from a disadvantage in terms of brand recognition as explained in detail in Section 7.8.4 below. Ryanair's brand is usually well-known to low-frills customers at both ends of any route, which is not the case for most competitors.

407. Thirdly, unlike Ryanair and Aer Lingus, carriers with a base at the destination end would normally only operate one route into Dublin. The destination-based carriers are therefore unlikely to show the same degree of commitment to routes from/to Dublin or the same ability and incentive to adjust capacity to react to shifts in demand or to competition. For a destination-based carrier the Dublin route represents just one route from its destination base; it would have many other routes in which it would not face Ryanair/Aer Lingus as a competitor. Their "commitment" to routes to/from Dublin is therefore likely to be limited in the face of aggressive competition. As a result Ryanair may be more prone to compete more aggressively on a specific route since it may expect such "one-route" competitors to leave the route for good (and refrain from entering other Dublin markets). There has been significant exit on ex-Dublin routes by both network and low-frills carriers operating from bases at destination airports as further analysed below in Section 7.8.5.

408. Carriers with a base at the destination airport have indeed performed poorly on ex-Ireland Routes (444). On at least 9 of the 35 overlap routes a competing carrier has left the route as of January 2006, unable to compete with Aer Lingus and Ryanair (Alitalia to Rome and Milan, Hapag Lloyd Express to Hamburg, TAP Portugal to Faro, Spanair to Barcelona, Madrid and Tenerife, easyJet on routes to London from Cork and Shannon). As further analysed in the route-by-route section, the market shares (point-to-point passengers) of network carriers based at destination flying on routes also flown by the Merging Parties are generally low. If these carriers were strong competitive constraints, one might expect their share of these routes to be at least as high as Ryanair's or Aer Lingus'. On all routes on which Ryanair was an incumbent carrier and a competitor entered over the last five years, no airline has been successful - apart from Aer Lingus (445). Seven airlines have attempted to enter Ireland - UK routes in competition with Ryanair. They have all since exited. Apart from the Dublin - London route, Ryanair faces no significant competition on any of its Ireland - Great Britain routes from any carrier other than Aer Lingus.

7.3.5. Customers consider Aer Lingus and Ryanair as the closest competitors

409. The Commission's market investigation has shown that customers overall consider Aer Lingus and Ryanair as the closets competitors in terms of product offering on routes to/from Ireland. In this regard it should be noted that the Commission generally places significant weight on the views and perceptions of customers.

410. In contrast with previous airline cases, any anti-competitive effects of the proposed merger would be felt primarily by individual passengers as opposed to corporate customers.

411. Ryanair has repeatedly argued during the course of the Commission's investigation that "there is clear, consistent and cogent evidence from customers that they do not perceive FR [Ryanair] and EI [Aer Lingus] as close competitors, evidence that the Commission has chosen to ignore"(446). In this regard Ryanair refers in particular to the responses of corporate customers to the Commission's initial market investigation.

412. Further, Ryanair has argued that the Merging Parties do not constrain each other because Ryanair serves customers that would otherwise not fly with Aer Lingus but in the event of a price increase would choose not to fly. In spite of these strong claims Ryanair has not submitted concrete evidence, for example in the form of surveys or marketing studies which could provide some credible support to the claim that "customers do not perceive FR and EI as close competitors".

413. The factors detailed above clearly indicate that Ryanair and Aer Lingus are close competitors. The Commission nevertheless acknowledges that Ryanair has on average lower prices than Aer Lingus, and that, at least hypothetically, it would be possible that Aer Lingus and Ryanair serve two entirely separate consumer segments.

414. As regards Ryanair's arguments that the Commission ignores the results of its own market investigation it should be noted that the views of large corporate customers are not necessarily representative of the views of the customers of the merging parties. Large corporate customers may generally be more time-sensitive and many have special deals (corporate deals) with network carriers that make them biased vis-à-vis such carriers. As the Commission's investigation had indicated that it would not be appropriate to consider mainly the views of so-called time-sensitive passengers as was the case in for example the Air France/KLM and Lufthansa/Swiss transactions, it was important for the Commission to try to ascertain, to the extent possible within the constraints of the Commission's investigation, the views of individual customers directly. As there was little information available as regards the views of individual customers on the substitutability between the two carriers, the Commission decided to conduct a Customer Survey at Dublin airport. The main purpose of the survey was to validate or refute, as the case might be, the claims made by Ryanair that passengers do not regard the Merging Parties as offering services which are close substitutes on routes out of Dublin (447).

415. Ryanair argues that the Customer Survey is deficient because "it was undertaken on a selfcompletion basis" and "the questions posed are ambiguous and often leading "questions" requiring knowledge on the part of the respondent, particularly in the case of the Consumer (passenger) survey, which it was not reasonable to assume it existed" (448). The Commission cannot agree with these statements for the reasons set out in more detail in Annex I. In particular self-completion questionnaires are a standard technique for gathering information from consumers in all sectors, including air travel. Second, all questions refer to the respondents own actions, perceptions and beliefs so it is reasonable to assume that the respondent had sufficient knowledge (449). Ryanair further claims that the survey was biased and was full of errors in many other aspects that would inflate and invalidate the results. The detailed rebuttal to these claims is found in Annex I to this decision.

7.3.5.1. Results based on unweighted data

416. The result of the Commission's Customer Survey shows that passengers flying on routes to/from Dublin consider Aer Lingus and Ryanair to be the closest substitutes. In particular, when customers have been asked which other airlines they have considered when planning their journey, the survey shows that overall the main alternative considered by both Ryanair and Aer Lingus customers are the other party (450).

417. The Commission also notes that there is a certain symmetry in the responses by passengers of Ryanair and Aer Lingus. In particular passengers travelling with other airlines other than the merging parties considered Aer Lingus more often than Ryanair. This is consistent with the hypothesis that Ryanair is less constrained by airlines other than Aer Lingus. This further suggests that the competitive constraint that both carriers impose on each other is symmetric. This is relevant because, in contrast to the Commission's regression analysis that focuses on the impact of presence of one firm on the prices of the other, respondents to the survey were not asked whether they considered the other carrier on the basis of a particular dimension (e.g. price).

418. Ryanair argues that the survey results are not conclusive proof of substitutability because the figures are inflated for two reasons. First "respondents were prompted to tick FR and EI (but not other carriers)" and second "many consumers may "consider" both FR and EI only to conclude that the two airlines are not substitutable" (451).

419. First, it should be emphasised that the primary goal of the survey was to test (i.e. validate or refute) Ryanair's claim that Ryanair and Aer Lingus are not considered as substitutes by their respective customers on routes out of Dublin. The Commission considers that asking passengers directly whether they considered Aer Lingus or Ryanair when flying in a particular route out of Dublin is the best way to avoid bias in the response to the critical question that serves to test Ryanair's claim that Merging Parties do not constraint each other.

420. In this regard, it is noted that both Aer Lingus and Ryanair were consulted on the questionnaire design before the survey was launched. With respect to the question in point, neither party expressed any concern or insisted in a reformulation of the question to ensure a balanced result (452).

421. Second, in order to assess the substitutability between the parties (and other parties), the Commission needed to understand which airline/s the customer had considered at the time of purchasing the ticket.

422. In this regard the Commission notes that of those Ryanair passengers that indicated that they always use Ryanair to travel on the route, a minority (22%) responded that they had considered Aer Lingus as an alternative (but only 5% considered any other airline) (453). In comparison a clear majority (66.1%) of Ryanair passengers that do not only fly Ryanair to travel on the route had considered Aer Lingus as an alternative (and only 24.5% considered any other airline). This implies that for passengers that have not made up their mind to always travel with the same airline 2/3 of them had considered Aer Lingus (454).

7.3.5.2. Results based on unweighted data by route category

423. The Commission's survey showed that where both airlines fly into the same airport at the destination end, more than half Aer Lingus and Ryanair's passengers have considered the other carrier. On routes where Aer Lingus and Ryanair compete with a third carrier around 40% of Ryanair's passengers considered Aer Lingus as an alternative, while around 17% considered any other competitor. Similarly, for Aer Lingus, 32.5% of its passengers considered Ryanair as an alternative whilst only 15.7% considered any other competitor (see table A.4.1 in Annex II). Thus, this is additional evidence to support the findings that the two companies are perceived by customers as the closest substitutes.

424. These results hold also when distinguishing between different customer groups (as business or leisure travellers) or between different reasons why passengers bought a ticket (see table A.4.2 in Annex II). A further indication is the number of Ryanair passengers that have selected Best price, Best time, Close Airport and Punctuality (the four most popular reasons) and whether they had considered Aer Lingus and other airlines (all aggregated). Of the 860 passengers that indicated they had chosen Ryanair because it offered the best price 44.4% considered Aer Lingus as an alternative, as opposed to 14.7% who considered carriers other than Aer Lingus (see table A.4.3 in Annex II). The same pattern is apparent whether the respondent's preference for Ryanair is due to the fact that it offered a good departure time, the destination airport was conveniently located or punctuality.

425. The results are similar when one looks at Aer Lingus customers. Aer Lingus passengers also seem to have a strong preference for Ryanair as an alternative again irrespective of the reason why they have selected Aer Lingus in the first place.

7.3.5.3. Results based on weighted data

426. Ryanair argues in its Response to the Statement of Objections (455) that the Customer Survey data should be weighted by route. In particular, it argues that since different routes will have different characteristics, data should therefore have been weighted by the number of passengers carried on each flight as a proportion of all passengers carried by the relevant carriers from Dublin over the period (456). Notwithstanding this concern, Ryanair did not make any attempt to produce results based on weighted data although it had full access to the survey data (457).

427. The Commission does not disagree with Ryanair that in some instances the results based on weighted data may produce very different, if not opposite results than if the data is left unweighted. In order to test Ryanair's claim, the Commission therefore also re-calculated the results on the basis of weighted data as suggested by Ryanair.

428. The re-calculation showed that the percentage of Aer Lingus passengers that considered Ryanair as an alternative is lower than in the case of the unweighted data (34.5% as opposed to 41.3%, see Annex I, second table in Section 3). However, it can still be regarded as significant that slightly more than 1/3 of Aer Lingus passenger had considered Ryanair as an alternative. It is also still above the 8.8% of Aer Lingus passengers that considered any other carrier.

429. All other percentages do not differ significantly from the results with unweighted data. It can therefore be concluded that the results do not differ significantly when based on weighted or unweighted aggregate data.

430. The closeness of Aer Lingus and Ryanair is also confirmed by another survey, conducted for Aer Lingus. The majority of respondents, irrespective of gender, age, social class or region of residence cited either Ryanair or Aer Lingus as their preferred airline of choice in routes departing out of Ireland for both UK and European destinations. Moreover they were also the number one and number two airlines in terms of overall spontaneous awareness and awareness of routes offered to UK and European destinations. Indeed the evidence led the survey analysts to compare both airlines with each other with respect to the most relevant factors that influence choice, without making such comparisons against any other airline (458).

7.3.6. Conclusion

431. On the basis of the above, it can be concluded that Aer Lingus and Ryanair are each others' closest competitors on the relevant routes out of Ireland.

7.4. The Merging Parties are currently competing with each other

432. It seems obvious that the Merging Parties compete with each other on a large number of routes. Ryanair claims, however, that it is not in competition with Aer Lingus. On the contrary, Ryanair argues that it behaves independently of Aer Lingus (and any other competitor) when setting prices and deciding on frequencies for its routes and does not consider the prices of its competitors (459). Ryanair claims that it is constrained only by the price sensitivity of its customers and not by the pricing behaviour of its competitors.

433. A price increase as a result from the merger would, according to Ryanair, be excluded by the fact that such price increases would contradict Ryanair's strategy to optimise the number of passengers on its flights ("load factor"). In support of its view, Ryanair submitted economic evidence (price regression analysis) that was, inter alia, intended to show that Ryanair's prices on "monopoly routes" were not higher or even lower than on routes where it faces competition460.

434. The Commission found, however, that Ryanair and Aer Lingus exercise a competitive constraint on each other and that they react directly on each other's pricing behaviour.

7.4.1. The Merging Parties' prices are directly influenced by the prices of its main competitor

435. The Commission has found that the pricing policy in the airline sector involves most often close monitoring of the other airlines flying on a route. Indeed, the Merging Parties regularly monitor each other's pricing policy and react regularly to price changes and to price-driven variations in demand of the other competitor.

7.4.1.1. Prices in the airline sector are transparent

436. One specific feature of the markets for scheduled short-haul flights is the transparency of prices the different competitors charge on their routes. Indeed, an absolute majority of the tickets of Aer Lingus ([70-80]*% (461)) and Ryanair ([90-100]*%) are sold via the internet, and also their competitors also sell a large part of their tickets through the web. Given this, and with the help of specialised software (462), airlines are able to observe the fares their competitors apply on any given flight. Prices are therefore transparent in the airline sector, notably in the low-frills airline segment with few connecting passengers.

7.4.1.2. Aer Lingus and Ryanair take into account the prices of the other when setting prices on a respective route

437. The Commission has found evidence that shows that both Aer Lingus and Ryanair closely follow the behaviour of their main competitors. In particular, they use specific software to adjust their capacities and lower their prices on a daily basis in reaction to the competitive behaviour of its main competitor. Without the main competitor, the need to do so would immediately disappear. The mechanism by which the Merging Parties observe and react to each other will be explained in the following paragraphs.

438. Like many other carriers, Ryanair and Aer Lingus use a so-called "yield management system" for their pricing policy and the determination of individual prices. The Commission has found that Ryanair's and Aer Lingus' yield management systems are similar, notably since they use two similar types of software tools to set their prices.

439. The first software type which both companies use is a system that tracks the booking status of each flight, provides forecasts for the further development and makes proposals for the pricing pattern (463). This software allows the responsible "yield manager" or "analyst" to verify the booking status for any given Ryanair or Aer Lingus flight. With the help of the program, the analyst can compare the actual booking status (or "load factor") of a flight with the booking forecast which is provided by the system. This booking forecast is based on previous experience with the same route at the relevant dates or on similar routes. The forecasts are adjusted according to forecasted growth and other relevant changes of the factors affecting supply and demand on the route. At the same time, the Revenue Management System (RMS) makes a proposal for the pricing pattern of each flight (the socalled "template"), i.e. for the number of places that should be available in a given price category ("booking class"). Both Ryanair and Aer Lingus have a certain number of "standard" templates, each with a different distribution of cheaper and more expensive seats. On a route with a relatively low demand, a template with more seats at lower prices would be used than on routes with high demand.

440. Should the load factor fall behind forecast, the analysts of both companies try to stimulate the demand, normally by making more seats available in the cheaper price categories. [CONFIDENTIAL - description of yield management system]*.

441. A decline or increase in the load factor can have different reasons, e.g. increased demand due to special events (e.g. football games) or holidays, which are not related the behaviour of a competitor. However, one important reason can also be the behaviour of the main competitor. If Ryanair or Aer Lingus lower their fares on a specific route (e.g. for a promotion or on a permanent basis), this can be assumed to have an immediate effect on the load factor of the respective competitor, since more customers will choose the cheaper airline than before. [...]*.

442. According to Ryanair and Aer Lingus, [CONFIDENTIAL - description of yield management system]* (464).

443. It is clear that Merging Parties, although not necessarily directly comparing prices of each others flights, react to each others pricing behaviour on a daily basis. Should the load factor of either Ryanair or Air Lingus fall behind expectations, the competitor would reduce its fares for a number of seats, in order to attract more customers in such a situation. The Commission was shown at the occasion of its site visits of the headquarters of Aer Lingus [...]* that a number of specialised yield managers monitor the single routes throughout the day and adjust the templates regularly in reaction to variations in demand. The reason for such a decrease of the fare could well be a promotion of a competitor or an entry of a new competitor on a route (465).

444. Ryanair and Aer Lingus do not only react "indirectly" (via the observation of a decreasing load factor) to each other. Both companies also use a price comparison software tool ("QL2") which allows them to track and compare air fares for every flight of all competing scheduled carriers that is distributed over the Internet. With QL2, Ryanair and Aer Lingus monitor regularly the competitive behaviour of their main competitors (466).

445. Ryanair explained to the Commission that [CONFIDENTIAL - description of yield management system]* (467).

446. The fact that Aer Lingus is constrained in its pricing by Ryanair is also confirmed by internal documents of Aer Lingus where their pricing strategy on intra-European routes is described in a presentation from April/May 2006 as follows: "Our strategy is to be within €[...]* of Ryanair on UK routes and within €[...]* on European routes. (468)". Also as concerns Ryanair, the Commission's investigation has found manifold evidence (including Ryanair's yield management system (469)) that Ryanair reacts to price changes of competitors. This is also supported by Ryanair's reaction to entry or to promotions of competitors (470), internal documents (471) and econometric evidence (472) that shows that also Ryanair is constrained by the pricing behaviour of other airlines, including Aer Lingus. 447. It can thus be concluded that both Ryanair and Aer Lingus regularly monitor and take into account each other's fares. Each of the Merging Parties reduces its fares (notably by offering more tickets at lower fares) when fewer seats are sold than expected (low "load factor"), in particular as a result of lower fares of the other (e.g. general price reductions, special promotions). Furthermore, Ryanair typically reduces its fares when it observes that it charges lower fares on a specific route (473), and also Aer Lingus reduces its fares if it observes that the difference between its fares and those of Ryanair exceeds a certain amount.

7.4.2. The Merging Parties react to each other's promotions and advertising campaigns

448. It should also be noted that Aer Lingus' and Ryanair's advertising strategy indicates that each considers the other as their closest competitor in Ireland and that they monitor and react to each other's promotions and advertising campaigns. Both Ryanair (474) and Aer Lingus (475) routinely publish advertisements in which they compare fares and services offered by the other. Ryanair and Aer Lingus also both carefully monitor and seek to react to each other's special promotional offers (476).

449. Aer Lingus has explained to the Commission that on a number of occasions, Ryanair reacted to Aer Lingus' promotions by reducing its fares. Once Aer Lingus' promotion was over, Ryanair raised its fares again (477). As set out in the previous paragraph, Ryanair's internal documents also confirm that Ryanair reacts to promotion campaigns of Aer Lingus.

7.4.3. The Commission's price regression analysis confirms significant competitive interaction between Ryanair and Aer Lingus

450. The Commission believes that the factors described in the previous sections provide sufficient proof that Aer Lingus and Ryanair compete with each other. Nonetheless, Ryanair and Aer Lingus have presented additional econometric evidence using data sets provided by the respective carriers. The Commission has verified and assessed these submissions and has undertaken its own regression analysis intended to estimate the likely impact of the Merging Parties on each other's prices.

451. This section summarises the main findings from the various econometric reports and derives conclusions relevant to the competitive assessment. The Commission takes into account in its analysis the comments made by the Merging Parties in their response to the Statement of Objections. A more in-depth analysis is included in Annex IV. Annex IV (i) motivates the use of econometric analysis in this case, (ii) examines the strengths and limitations of particular methodologies, (iii) reviews the econometric evidence submitted by the Merging Parties, (iv) presents the results from the Commission's own price regression analysis and (v) discusses a number of technical issues in more detail in reaction to the comments made by both Ryanair and Aer Lingus in their response to the Statement of Objections.

452. The Commission's price regression analysis tests the following hypothesis: ???Whether the presence of one of the Merging Parties in the route is associated with a statistically and economically significant reduction in the fares of the other; ???Whether the Merging Parties exert on each other a stronger competitive constraint than any other existing competitor; ???Whether the existence of an actual or potential competitor with a significant presence at the destination airport on a route originating in Dublin has an impact on the Merging Parties' prices; ???Whether a stronger presence of one of the Merging Parties (in terms of number of frequencies) has a more pronounced effect on the other's fares.

453. Ryanair first proposed a regression technique known as cross-section analysis. This technique examines differences in prices across a number of affected routes at a given point in time (for example October 2006). The Commission considers this technique illsuited for this case. This is because the routes served by the Merging Parties are highly heterogeneous. Routes differ significantly in ways that likely affect prices but are difficult to observe or measure. Without controlling for such differences, cross-section regressions are meaningless or biased. Notwithstanding these concerns the Commission has decided to test the usefulness of cross-section analysis, correcting to the extent possible some of the methodological difficulties identified in Ryanair's submission.

454. The Commission constructed a data set combining data provided by the Merging Parties and the Dublin Airport Authority ("DAA"). It consists of data where the Merging Parties' average prices in multiple routes are observed every month for 5 years. Data in such format is commonly known as panel data. Panel data allows using a regression technique known as fixed-effects. This technique exploits the variation in market structure at individual routes over time to assess the extent to which the merging firms exert a competitive constraint on each other (holding constant competition from other airlines or other factors that vary over time).

455. The motivation for using fixed-effects is as follows: the entry of Ryanair on a route served by Aer Lingus may affect the latter's price (or vice-versa). The fixed-effect procedure compares the level of Aer Lingus prices on a route after Ryanair entered, with the level before Ryanair entered. This before-and-after comparison is done systematically over time and for all routes where Aer Lingus operates. In this way the regression generates the average effect of Ryanair's presence on Aer Lingus fares.

456. It should be noted that the fixed-effects procedure compares the incumbent's price beforeand- after the entry of a rival within the given route. This offers a significant advantage over cross-section regressions. By focusing on a within-route comparison of prices, the fixed-effects procedure keeps constant any time-invariant factors that might affect prices, significantly reducing the risk of obtaining biased results478. Consequently, the Commission regards the panel data regression analysis with fixed-effects as the most suitable for testing all the hypotheses identified above.

7.4.3.1. Cross-section regression analysis

457. In the course of the market investigation Ryanair submitted three reports based on its own data. Ryanair's first report is based on a data set of 313 city pairs for the month of October 2006 (479). Ryanair estimates several cross-section regressions that examine differences in its fares across a number of EU routes at a point in time. The regression specification (480) includes a variable to indicate the presence or absence of Aer Lingus in a Ryanair route. The presence of Aer Lingus is measured by a dichotomous variable (or dummy) that takes the value 1 when Aer Lingus was present in the route on that particular month and zero if it was absent. This variable is intended to capture differences in prices across route that may be associated with Aer Lingus' presence. Ryanair also considers evidence on whether the absence of "base competition" leads to a statistically significant increase in its fares.

458. Ryanair in its Response to the Statement of Objections of 17 April 2007, paragraph 140 argues that "... it is astonishing that the 265 pages of the Commission's SO did not find the time or space to consider the most relevant evidence of how FR actually sets prices on routes where it faces no actual or potential competition from El. Such evidence was presented to the Commission in the report by RBB Economics which analysed the fares charged by FR across 300 different routes, and specifically analysed the question of whether FR charged higher fares on monopoly routes than on routes where it faces active competition". The Commission notes that a full section in Annex IV was dedicated to reporting, discussing and extending Ryanair's cross-sectional analysis (section 5). The Commission's conclusion was reported in the main body of the SO - which incidentally is quoted by Ryanair further below in the same paragraph (140).

459. Ryanair finds that there is no statistically significant association between the presence of Aer Lingus and Ryanair prices in the selected sample of routes. According to Ryanair this implies that Aer Lingus is "not an effective competitive constraint on Ryanair on existing overlap routes".

460. The Commission cannot accept these results as evidence that Ryanair's prices are unconstrained by Aer Lingus. First, the Commission has verified that Ryanair's results are not robust to small changes in the specification used. Further technical problems concern the "market power dummy" and the "number of competitors" variable (see Annex IV for further details).

461. Second, as explained in more detail in Annex IV, the cross-section technique cannot control for unobserved or unmeasured factors that influence price but vary across routes. Failing to control for such factors likely leads to biased estimates (481). For example, prices may be higher on monopoly routes not because there is no competition but because in this particular route, demand is relatively inelastic or costs are relatively high (e.g. when high entry barriers are correlated with high operation costs). The opposite is also true, prices on monopoly routes may be relatively low because demand is weak (or very elastic) in such routes.

462. Third, the data set is not constrained to routes out of Dublin (or even routes out of Ireland). Instead it includes more than 300 of Ryanair routes across Europe. Ryanair's analysis is thus based largely on routes where Ryanair and Aer Lingus do not compete and are unlikely to ever do so (such as, for example routes from Germany to Italy). Ryanair argues that this argument "is entirely formalistic and out of keeping with normal good practice in empirical economic analysis [...] observations of how prices are formed in situations where concentration is already high provide a basis for simulating the impact that a merger will have if it is allowed to go ahead, thus creating a similarly high level of concentration". (482) The Commission cannot agree for two reasons:

* First, price differences between, say, routes from Germany to Italy and routes from Ireland to France are likely to be a result of many factors other than competition between the Merging Parties. While this does not in itself invalidate Ryanair's crosssection analysis, effectively controlling for the structural differences between the routes in Ryanair's sample would be a complex task. By choosing a particularly heterogeneous sample of routes, Ryanair has thus reduced the likelihood that its analysis yields statistically significant results.

* Second, Ryanair's cross-section analysis does not provide any indication of whether Aer Lingus' presence has a significant effect on Ryanair prices in overlap routes, relative to Ryanair's price in non-overlap routes out of Ireland. This would be a more informative comparison to provide "a basis for simulating the impact of a merger".

463. In any event, to assess the robustness of Ryanair's cross-section regressions and assess the suitability of this technique in this case the Commission has also run a number of crosssection regressions. The Commission's initial specification tests whether the presence of one of the Merging Parties on a given route is negatively related with the fares charged by the other (this is labelled the presence specification). An alternative specification tests whether the number of frequencies of one merging party in a given route is associated with the other party charging lower fares (this is labelled the frequency specification). Following Ryanair, the Commission considers the two shoulder months of April 2006 and October 2006.

464. The Commission's cross-section regressions correct, to the extent possible, some of the problems identified in Ryanair's submission.

* First, for the reasons mentioned above, the Commission considers that in this case it is important to focus on routes out of Dublin since at least one end of the route is common to all routes included in the regression, and unobserved differences across routes can thereby be partially reduced. This approach also allows to directly assessing the primary question on interest to be directly assessed, namely the competitive constraint that the Merging Parties mutually exert on each other.

* Second, the Commission's specification includes dummy variables for the presence of both flag and non-flag carriers on the route and also for flag and non-flag carriers with a strong presence at the destination airport. A demand control and a cost control are also included.

465. The Commission's cross-section analysis does not confirm the results reported by Ryanair. This is not surprising since the Commission focuses on routes out of Dublin whereas Ryanair considers routes all across the EU. On the basis of the retained market definition (city-pairs database) the Commission's results indicate that Aer Lingus' prices are negatively related with Ryanair's presence. This association is statistically and economically significant despite the very small sample (48 observations). The same negative relationship is obtained when considering routes where the Merging Parties serve the same airport. However, it is statistically significant at least at the 10% level for both months only when the baseline regression is augmented to control for flag and non-flag carriers with a strong presence at the destination airport.

466. Conversely, the Commission's cross-section regressions indicates that Ryanair's fares are negatively associated with Aer Lingus' presence on both months only when they serve the same airport. When considering city-pairs the relationship is also negative but statistically not significant, so limited, if any, weight should be placed on this result.

467. The frequency specification would also suggest that the strength of presence of one of the Merging Parties has an effect on the prices of the other. This effect is statistically significant in all the regressions based on the city-pair database. However, as pointed out by Ryanair, omitting the variable distance in these regressions is likely to lead to biased estimates. The Commission agrees that the results from the regressions based on the frequency specification cannot be relied on.

468. In fact, as explained in Annex IV in more detail, the Commission considers that neither Ryanair's nor its own cross-section analysis is robust to the necessary standard. Indeed, no definite conclusions can be derived from cross-section regressions in this case given the impossibility to control for unobserved factors that affect prices and differ across routes, the small number of observations, the sensitivity of the results to the month considered or the fact that the inclusion of statistically insignificant explanatory variables sometimes affects the coefficients of other variables.

7.4.3.2. Ryanair's two stage regressions

469. Ryanair submitted two further reports to test the hypothesis that Aer Lingus' presence has an effect on its fares.

470. In the first report (483) Ryanair examines whether the presence of Aer Lingus has an effect on Ryanair prices on routes out of Dublin. Ryanair's dataset consists of a panel of Ryanair one-way fares from Dublin (i.e. single fares from Dublin to other European destinations) over the period April 2002-December 2006, as well as other potential explanatory variables. In this dataset, Ryanair identifies only a few instances of entry or exit of Aer Lingus over time into routes where Ryanair is already present. Ryanair argues that in such case the standard fixed-effects procedure is not appropriate. Instead they use a two step procedure that according to Ryanair "still benefits from the panel nature of the data" and "is more robust than the standard fixed effect approach" (484).

471. Ryanair reports that for all the considered specifications the presence of Aer Lingus has no statistically significant impact on its prices, either on a city pair or on an airport pair basis. According to Ryanair this finding supports its claim that its business model is to target a passenger base that other (higher cost) airlines cannot profitably target. It follows that the strongest competitive constraint on Ryanair is not Aer Lingus but the price sensitivity of its customer base. Ryanair concludes that attempts by Ryanair to sustain higher fares would not be profitable - not because of switching to Aer Lingus but due to passengers switching to alternative activities.

472. Ryanair submitted a complementary report (485) to show that it does not set higher fares on "monopoly" routes (on either a city or airport pair basis). Ryanair follows the same twostage approach described in the previous submission. The main difference is that this report considers average monthly yield on all origin-destination pairs from Frankfurt Hahn, not Dublin, or Ireland. The variable of interest separates routes in which Ryanair is a monopolist and others where it faces actual or potential competition - note that Aer Lingus is not present on routes out of Frankfurt (except to Dublin). Ryanair's reported results indicate that the presence of carriers in routes out of Frankfurt-Hahn has no statistically significant effect on its prices.

473. The Commission acknowledges that in light of the reduced number of instances of Aer Lingus entry into Ryanair routes a standard fixed-effects regression may be uninformative as to the effect that Aer Lingus entry or exit might have on Ryanair's prices. However, the Commission considers that the two-step methodology proposed by Ryanair cannot resolve this difficulty. It is also inadequate to assess the effect of Aer Lingus on Ryanair over time, for the following main reasons: * First, Ryanair's two-step methodology is effectively a cross-section analysis (see Annex IV, section 6.3 for a detailed explanation). Hence, the Commission's concerns with respect to the cross-sectional regressions in this case also apply here. In particular, the small number of observations, which leads to high standard errors and the impossibility to control for unobserved factors that affect prices and differ across routes which results in omitted variable bias.

* Second, the second stage regression only includes Aer Lingus' presence. It does not include presence variables for other competitors. It is thus likely misspecified.

* Third, the use of the "matched-route" as a proxy variable for demand implies that the fixed-effects estimated in the first stage do not in reality correspond to the average fares. This source of misspecification further implies that Ryanair's twostage regressions are unlikely to generate reliable estimates.

474. Further problems include inconsistencies in the quality of the data used by Ryanair and the wide variation in the estimates due to slight changes in the specification. Moreover, it should be noted that even Ryanair's preferred specification fits very poorly and explains only 18 percent of the variance in average fares.

475. With respect to the analysis comparing prices on monopoly routes and non-monopoly routes out of Frankfurt Hahn the above mentioned concerns apply since Ryanair uses the same two-step analysis. Additionally, the analysis of routes out of Frankfurt-Hahn does not provide any direct evidence regarding the competitive constraints that Ryanair faces on the affected routes - that is routes out of Ireland.

476. It should be remembered in this context, that the regression techniques used by Ryanair, Aer Lingus and the Commission involve a one-sided test: A regression can either establish a statistical link between two parameters (say, simultaneous presence of Ryanair/Aer Lingus and price levels) or it can fail to do so. However, failure to prove a statistical link is not equivalent to proving that no such link exists. Alternative explanations for an "unsuccessful" regression include, in particular, unsuitable data sets or misspecified regression equations. It would appear that Ryanair, by choosing a complex two-stage stage approach and a data set with routes outside Ireland, increased the likelihood that its regression would not yield statistically significant results. Because Ryanair had a vested interest in seeing its regressions "fail", the probative value of such an outcome is low and cannot be taken as evidence that Ryanair is not constrained by Aer Lingus on routes out of Ireland.

7.4.3.3. Fixed-effects regressions

477. A panel regression with route specific fixed-effects can mitigate the omitted variable bias that affects cross-section regressions. This is because unobservable cost or demand factors, whose variation across routes would be likely to affect fares are more likely not to vary over time at any single route. The Commission regards this methodology as the most suitable to assess the competitive constraint exerted by Ryanair on Aer Lingus.

Aer Lingus' fixed-effects regression analysis

478. The economic consultants engaged by Aer Lingus submitted several econometric reports using data provided by Aer Lingus and the Dublin Airport Authority (DAA). They relied on a fixed-effects procedure to test the hypothesis that Ryanair's entry or exit on Aer Lingus routes had an effect on Aer Lingus' average fare and load factor.

479. Aer Lingus considers three alternative measures of Ryanair's presence on a route served by Aer Lingus: (i) share of total seat capacity, (ii) absolute monthly seat capacity and (iii) a dummy indicating presence of Ryanair in the route. Ryanair has criticised the inclusion of "capacity shares" as a explanatory variables for two reasons. First, to the extent this variable is endogenous it can lead to biased estimates. Second it imposes strong restrictions on the estimated coefficients which are not consistent with theory. The Commission agrees with Ryanair that the use of capacity shares in the regression is problematic.

480. However, the Commission also notes that the results based on regressions that include a "presence" dummy (similar to the Commission's own fixed effect regressions) are not affected by this criticism. Aer Lingus' reported results for this specification indicate that Ryanair's presence on a route is associated with Aer Lingus charging [5-10]* percent lower prices. These estimates are in line with the results from the Commission's fixed-effects regression reported below.

Commission's fixed-effects regression analysis

481. The Commission's panel data has been constructed on a city pair basis according to the geographic relevant market for each route as defined by the Commission. The data set tracks on a monthly basis all the information submitted by Aer Lingus, Ryanair and the DAA for routes out of Dublin.

482. Using this merged data set which covers the period January 2002 to December 2006, the Commission has used a standard fixed-effects procedure to assess the competitive constraint exerted by the Merging Parties on each other. The Commission's initial specification tests whether the presence of one of the Merging Parties in a given route is negatively related with the fares charged by the other (this is labelled as the presence specification). An alternative specification tests whether the number of frequencies of one merging party in a given route is associated with the other party charging lower fares (this is labelled as the frequency specification).

483. The Commission's analysis confirm Ryanair's claim that there is not sufficient variation, within a reasonable time period, in the presence of Aer Lingus in routes operated by Ryanair. As a result the fixed-effects regression does not provide reliable estimates of the possible impact of Aer Lingus' presence on Ryanair prices486. In contrast, there are many instances of Ryanair entering/exiting routes in which Aer Lingus was already present. Hence the fixed-effects procedure is very well-suited to assess whether Ryanair's presence is negatively associated with Aer Lingus prices.

484. The presence specification includes dummy variables for the presence of (i) Ryanair, (ii) one or more flag carriers and (iii) one or more non-flag carriers. It also includes dummy variables for the presence of Aer Arann and CityJet, which according to Ryanair also exercise a strong constraint on the Merging Parties on Dublin routes. The Commission extends the baseline specification to include various control variables for demand and costs and a dummy variable to indicate whether flag or non-flag carriers have a strong presence at the destination airport. In all cases, the presence of Ryanair is statistically significant and is related to Aer Lingus charging lower prices.

485. The Commission's fixed-effects regressions validates the following hypothesis set out exante:

* First, depending on the specification, the Ryanair's presence is associated with Aer Lingus charging around 7-8% lower prices when considering city-pairs reflecting the Commission's retained market definition and around 5% lower prices when considering airport-pairs. This effect is economically and statistically significant in all tested regressions. This result is also robust, correcting for the presence of outliers, heteroskedasticity and serial correlation. It is also highly robust to the use of alternative specifications including alternative demand and supply controls. Notably, in practically all cases the control variables in the different regressions have the expected signs and are statistically significant. The explanatory power of the regression is also high with R2 consistently above 80%.

* Second, comparing the coefficients of Ryanair with that of flag-carriers and non-flag carriers, as well as Aer Arann and CityJet, Ryanair's presence or number of frequencies have a much stronger economic impact (at least double) than that of any other type of carrier. In most cases the regressions indicate that the presence of other carriers has no economic or statistically significant effect on Aer Lingus fares.

* Third, destination-based flag carriers exert only a very limited constraint on Aer Lingus. Destination-based non-flag carriers exert a higher constraint than flag based carriers. However, their constraint is around half or less than the constraint exerted by Ryanair on Aer Lingus retaining the Commission's market definition. Moreover flag carriers are only present on 8 of the 37 overlap routes upon which Aer Lingus and Ryanair competed in May 2007, and tend to be much smaller than either Ryanair or Aer Lingus where they are present (especially for point-to-point passengers). Thus, contrary to Ryanair's claim, it cannot be expected that the merged entity would be effectively constrained by flag or other non-flag carriers post-merger.

* Fourth, measuring the strength of Ryanair's presence using number of frequencies in the route as a proxy provides further confirmation that Ryanair constrains Aer Lingus. It is possible to examine the price change in overlap market only or across all markets under various assumptions. For example one can focus on the price effect on the last month for which data is available or the price effect on average over the full sample period. Depending on the specification the price effect of the merger implied by the Commission's frequency regressions is around 5-6% (on average over all routes) or 10-12% (if only overlap routes are considered). This adds to the robustness of the results derived from the presence specifications. It is also worth noting that, as expected, Ryanair appears to impose a more significant constraint on Aer Lingus when it serves the same airport.

486. The fixed-effects regressions with Ryanair's prices as the dependent variable do not allow conclusions to be reached with respect to the impact of Aer Lingus on Ryanair prices. This is because there are insufficient instances of Aer Lingus exiting or entering into a route where Ryanair was already present. In other words there is little variation in the presence of Aer Lingus on Ryanair routes. It should be emphasised, however, that this neither validates nor refutes the hypothesis that Aer Lingus exerts a competitive constraint on Ryanair's prices. Indeed the evidence presented in the previous section makes it clear that Ryanair as well as Aer Lingus permanently monitor their own load factor and each other's prices and adjust prices accordingly. If Aer Lingus were to lower its prices this would tend to attract customers that would otherwise have flown Ryanair. If a sufficient number of customers switch to Aer Lingus (switching costs are very low and in most overlap routes there are no closer alternatives) this would lead to lower load factors. According to its pricing policy Ryanair would react by reducing its prices in order to meet its load factor targets. If Ryanair were not constrained by Aer Lingus there would be little logic in permanently monitoring their prices (487).

487. Moreover both economic theory and qualitative evidence suggest that Ryanair might also be constrained on parameters of competition other than price. As a result of the merger the ability to increase Aer Lingus fares would further relax the competitive constraint faced by Ryanair, allowing it to increase fares of ancillary services or to reduce quality without risking to lose as many customers as in the absence of the merger.

488. Finally, it should be noted that the estimated effect of Ryanair on Aer Lingus prices is likely to be underestimated. As pointed out below (Section 7.6), the presence of Ryanair in Dublin exerts a potential competitive constraint on Aer Lingus. On routes out of Dublin where it is the only carrier, it can be expected that Aer Lingus sets prices which are lower than what it would charge if Ryanair had no Dublin base. Since the regression analysis considers only fares' overtime variations within each route and only captures price reductions subsequent to Ryanair's entry, this potential competition constraint does not show up in the empirical results.

7.4.3.4.Conclusion

489. The Commission's regression analysis confirms and complements the conclusions derived from qualitative evidence that Ryanair and Aer Lingus are close competitors. Moreover the results from the regression analysis are also in line with the majority of respondents to the Customer Survey that consider the Merging Parties to be the closest competitors when other carriers are present on the route.

490. The fixed effects regressions therefore provide clear and convincing evidence that Aer Lingus' prices are currently constrained by competition from Ryanair.

7.5. The merger eliminates actual competition between the Merging Parties to the detriment of customers

491. As shown in the above Sections (notably Section 7.4), Aer Lingus and Ryanair are in competition with each other and constrain each other when it comes to determining fares for flights on the overlap routes (488). The proposed merger would eliminate this actual competition between the Merging Parties, giving the merged entity significantly increased market power with the likely consequence of increased fares and/or a reduction of the number flights for passengers wishing to travel to or from Ireland. Indeed, post-merger, as predicted by standard ''non-coordinated effects" analysis (see paragraph 24 in the Horizontal Merger Guidelines), both carriers would internalise the effects of setting higher fares on each other. In particular, the merged entity would have the incentive to set higher fares for Aer Lingus since most of the customers lost would be captured by Ryanair.

492. The likely consequences of the elimination of competition on the overlap routes are, however, not limited to increased prices. Effective competition does not only bring low prices to consumers but also other benefits such as high quality products, a wide selection of goods, and services and innovation (489). The elimination of competition between the Merging Parties would also deprive customers of these benefits of effective competition.

493. Another consequence of the combination of the two strongest short-haul airlines in Ireland may be that less new routes that would be developed without a merger. Indeed, the Commission observes that the presence of two base operators in Dublin, Ryanair and Aer Lingus, has in the past stimulated both to engage in competition for opening new routes, in particularly out of Dublin. Indeed, by 2001, Ryanair's operations from Ireland were mainly directed to Great Britain, having a share of approximately 50% of the overall Ireland-Great Britain traffic. At that time, it served only two destinations in Continental Europe from Ireland (Paris Beauvais and Brussels Charleroi). Aer Lingus then started its transformation process into a point-to-point airline, modernising its fleet and reducing significantly its costs. As a result, Aer Lingus increased significantly the number of destinations served from Dublin. While in 2001 Aer Lingus operated 29 routes out of Dublin to European destinations, this number increased to 52 in 2004 and more than 70 in 2006. It can be observed that at the same time also Ryanair increased significantly the number of destinations served out of Dublin, operating 14 routes in 2002, which increased to 22 in 2004. It serves today more than 50 routes to Continental Europe only from Dublin (490).Given the expansion of both airlines in recent years, Irish customers therefore could benefit from a significant increase of new destinations by more than 100% . During this period, Dublin Airport also witnessed a 29% growth in terms of overall passenger numbers. Both Merging Parties have clearly documented plans to continue expanding their Irish network (491).

494. Although it is true that every airline has an incentive to develop new routes and that Ryanair has expanded its route network even from those bases where it does not face direct Competition (492), the incentive to quickly develop new routes is higher in a situation in which an airline faces an equally strong competitor at the same base. This incentive would be reduced post-merger, since competition between Ryanair and Aer Lingus will be eliminated.

495. Indeed, the airline which first develops and opens a route may benefit from "first-mover advantages" which may make it more difficult for the late entrant to operate as profitably on the route as the "first mover". This may be the case for instance for small or "thin" routes which generate only limited traffic (493). The presence of another strong airline is therefore likely to stimulate the fast development of new routes, since both airlines have an interest in securing the first-mover advantages. After the merger, the need to develop new routes as fast as possible would disappear, since the merged entity would not need to secure a "first mover advantage".

496. Furthermore, as shown inter alia by the Commission's regression analysis and internal documents of the Merging Parties, fares and margins are in general lower when an airline faces competition on a route than when it operates a monopoly route. At present, both airlines have therefore a strong incentive to discover such monopoly routes on which entry of the other airline is less likely (e.g. particular thin routes). Assuming that Aer Lingus and Ryanair deploy their aircraft to the most profitable use, the average yield required for opening a new route will depend, inter alia, on the yields available on existing routes. The more competitive (and, hence, less profitable) the existing routes are, the more attractive, in relative terms, it becomes to discover and open viable new routes (494).

497. The Commission therefore concludes that the proposed merger would eliminate actual competition between the Merging Parties on all overlap routes, giving the merged entity significantly increased market power with the likely consequence of increased fares (495) and/or a reduction of the number of flights for passengers wishing to travel to or from Ireland. The elimination of competition may also result in a reduction of service quality, of choice between different service models and in a reduced incentive for the merged entity to develop new routes. Thus the merger would eliminate actual competition between the Merging Parties to the detriment of consumers.

7.6. The merger would eliminate potential competition between the Merging Parties to the detriment of customers

498. There is a significant number of routes to/from the Irish airports where both Ryanair and Aer Lingus have short-haul activities (in particular Dublin as well as Shannon and Cork) where only one of the Merging Parties is present. It is therefore necessary, in line with the previous practice of the Commission (496), to analyse whether the transaction would affect competition by eliminating the other carrier as a potential competitor on these routes.

7.6.1. The analysis of potential competition reflects the dynamic character of competition in this case

499. Indeed, as set out in Section 6.2, competition on the individual routes operated by Ryanair and Aer Lingus from the same base cannot be regarded in isolation. Such an isolated analysis would imply that the respective product markets were entirely independent from each other. As set out in detail above (e.g. Section 6.2 on a possible market for flights from/to Ireland and Section 7.3.4 on the advantages of a base), the single routes or markets are linked by many elements to each other, notably by the fact that most of them are operated from a large base in Ireland. In particular, airlines operating pursuant to the lowfrills/ point-to-point model from a large base have the necessary flexibility to shift and add routes from their existing base in reaction to changes in the competitive structure of the different markets/routes operated from their base (497). The Commission can therefore not only analyse in a "static" manner the situation on the 35 overlap routes which happen to exist at the time of the Commission's decision. Such an analysis would disregard the dynamic character of competition in the affected markets. This is already evidenced by the fact that the number of routes the Merging Parties operate from/to Dublin, Cork or Shannon is constantly changing and that even the number of the overlap routes is far from static, and has changed even during the time of the Commission's investigation (498). It is therefore necessary that the Commission also analyses to what extent the disappearance of the Merging Parties' closest and most important competitor might eliminate potential competition that would have constrained the Merging Parties in the absence of the transaction.

500. The Horizontal Merger Guidelines acknowledge that concentrations where an undertaking already active on a relevant market merges with a potential competitor may have anticompetitive effects and thus lead to significant impediment of effective competition. For this, two basic conditions must be fulfilled: (1) the potential competitor must already exert a significant constraining influence or there must be a significant likelihood that it would grew into an effective competitive force and (2) there must not be a sufficient number of other potential competitors, which could maintain sufficient competitive pressure after the merger (499).

501. As regards the first criterion, it is relevant to consider (i) the closeness of competition between Ryanair and Aer Lingus and the extent to which they constrain each other, (ii) the ability of Ryanair and Aer Lingus to enter into markets (routes) where currently only one of them is active, (iii) past evidence of entry by Ryanair and Aer Lingus is to these routes. As regards the second criterion (ability to enter), it is necessary to demonstrate that entry by other competing carriers on these routes is significantly less likely and thus would not provide sufficient competitive pressure on the merged entity. All these elements are analysed below, first in general, and then applied to the individual markets (routes).

7.6.2. The Merging Parties are close competitors that constrain each other and are capable of entering into each other's routes

502. Ryanair has put forward that it behaves independently of Aer Lingus as it is only constrained by the price sensitivity of its customers. This argument has been refuted in detail in Sections 7.3 and 7.4. The Commission also notes that the Horizontal Merger Guidelines acknowledge that a potential competitor significantly constraints the behaviour of firms active in the market in particular if it possesses assets that could easily be used to enter the market without incurring significant sunk costs (500).

503. As explained in detail in Section 7.3.4 above, both Ryanair and Aer Lingus enjoy a number of competitive advantages derived from their strong presence at the three Irish airports. In particular both airlines, unlike their competitors (501), enjoy significant flexibility to reschedule their operations across routes from/to Ireland without incurring significant sunk costs of entry and within a reasonably short time frame due to their strong presence in Dublin. The same flexibility also mitigates the costs of failed entry since again it allows both carriers to shift back capacity to another route out of the relevant Irish airport - for example as soon as it is identified that demand is weaker than expected or if costs unexpectedly increase, thereby reducing profitability (502).

504. Further, both Ryanair and Aer Lingus plan to expand their operations which enforces the likelihood that they would enter one or more routes served currently only by one of them. As set in the previous Section (7.5), there is strong evidence that both Aer Lingus and Ryanair have significantly expanded their operations in Dublin in the past and that they would continue to do so without the merger. In 2001, Aer Lingus operated 29 short-haul routes which increased to 71 by 2005. Similarly, Aer Lingus' passenger numbers grew from 6.6 million in 2001 to 8 million in 2005. The total number of short-haul passengers carried by Aer Lingus increased only in 2006 by 9.3% (503). Aer Lingus has also replaced its fleet with larger aircraft. Ryanair acknowledges that "the introduction of the A-320 increased [Aer Lingus'] capacity on European routes by 22%..." (504). Furthermore, the Aer Lingus IPO was intended precisely to deliver additional capital and access to the capital markets for Aer Lingus in order to allow it to expand its fleet. This expansion is on-going as Aer Lingus currently has firm orders for 4 short-haul aircraft (two to be delivered in May 2007 and the remaining two in November and December 2007). Further, Aer Lingus plans to expand its short-haul fleet to a total of 42 aircraft by 2012. Ryanair argues that Aer Lingus has already confirmed that its firm orders for four short-haul aircraft will be committed to the launch of a new base in Continental Europe rather than to strengthen its presence in Dublin, Shannon and Cork (505). Aer Lingus states that they intend to set up a new base with three aircraft outside Ireland (most likely in one of the UK airports) which would serve Dublin and destinations in Continental Europe (506). Therefore, at least one of the additional aircraft would be still based in Dublin, Cork or Shannon. Further, as the aircraft at the new base would also serve Dublin, this would allow Aer Lingus to either increase frequencies between the new base and Dublin (and possibly also Shannon or Cork) and thus enable to reschedule the added capacity to other routes within its network out of the three Irish airports. This would further enhance the ability of Aer Lingus to compete with Ryanair which has a strong presence in the UK.

505. Also Ryanair envisages a further rapid expansion with firm orders for a further 170 new aircraft to be delivered over the next six years which should allow Ryanair to double the number of passengers carried to over 80 million passengers per annum by 2012 (507). Ryanair claims in the Notification that they have opened overall 170 routes in the last year and 42 new routes out of Dublin Airport over the last two years (508). Further, Ryanair's CEO in his comments to the Q1 2006 results stated as follows: "Ryanair will increase its fleet by 27 aircraft this Winter (compared to last year's net increase of just 15) during which we will launch many more new routes and bases." (509) The growth plans of Ryanair specifically for Ireland are also confirmed by its internal documents (510) as well as published information (511). Ryanair argued that their expansion reflects the "joining the dots" approach when they open new routes to destinations already served in their network. They argue that if Aer Lingus' non-overlap routes from Dublin are considered, there are only two airports in Ryanair's current network - Düsseldorf and Palma de Mallorca (512). This would however be a very static approach to Ryanair's possible expansion. As the total number of destinations served by Ryanair has grown steadily in recent years (513) and in view of the overall expansion plans, it cannot be presumed that there will be no other "dots" added to the Ryanair's network in the near future. Furthermore, there are a number of Aer Lingus' nonoverlap routes to/from Cork to destinations already in the current Ryanair's network (see below in the analysis of potential competition on routes to/from Cork).

7.6.3. Past evidence shows that in particular Ryanair enters into Aer Lingus' Routes (514)

506. Past evidence shows that in particular Ryanair has in the past systematically entered on routes where Aer Lingus was active (515). There are also a number of examples of Aer Lingus entering routes served by Ryanair.

507. The entry of the competitor on routes where previously only one of the Merging Parties was active led to a sharp increase in the number of overlap routes in previous years: Over the last 6 years the number of overlap routes has increased from 8 to 35 (or even to 37 if the situation in May 2007 is regarded (516)). This past evidence of entry is particularly compelling for Ryanair.

508. Indeed, Ryanair entered the following routes on which Aer Lingus operated previously: it commenced operations in March 2003 on Dublin-Faro where Aer Lingus operates since June 2002; in April 2006 on Dublin-Krakow where Aer Lingus operates since November 2005; in April 2006 it entered Dublin-Milan where Aer Lingus operates since March 2005, in March 2003 it entered Dublin-Malaga; in January 2006 it entered Dublin-Riga where Aer Lingus commenced operations in November 2005, in April 2006 it entered Dublin- Salzburg, where Aer Lingus operates since December 2005; on Dublin-Tenerife it entered in December 2006 where Aer Lingus operates since November 2003; in 2003 it entered Dublin-Barcelona on which Aer Lingus operates since March 2002; in April 2006 it entered Dublin-Marseille where Aer Lingus operates since March 2005; In April 2006 Dublin-Valencia, where Aer Lingus operated since March 2004; in April 2006 on Dublin- Bratislava competing with Aer Lingus which flies to Vienna since October 2002; in April 2005, it entered Dublin-Frankfurt-Hahn, competing with Aer Lingus which flies to Frankfurt main airport; it entered Dublin-Luebeck in April 2006 competing with Aer Lingus' operations to Hamburg; in April 2004, Ryanair entered Dublin-Murcia competing with Aer Lingus which operates on Dublin-Alicante; in 2006 it started operating on Dublin-Berlin where Aer Lingus operates since 2004; it commenced operations on Dublin - Carcassonne in December 2005 competing with Aer Lingus which flies to Toulouse; in December 2006 it entered Dublin-Madrid; in March 2006 it started operating Dublin - Treviso in competition with Aer Lingus' services to Venice operated as of March 2004; in December 2006 it started operating Dublin - Fuerteventura where Aer Lingus operated from November 2005 to May 2007; in December 2006 it started operating Dublin - Warsaw where Aer Lingus is active as of June 2004; in January 2007 it started operating Dublin - Vitoria in competition with Aer Lingus' services to Bilbao; in February 2007 it started operating Dublin - Forlì in competition with Aer Lingus' services to Bologna; in December 2006 it started operating Dublin - Grenoble in competition with Aer Lingus' services to Lyon; in January 2007 it started operating Dublin - Seville where Aer Lingus is active as of March 2005.

509. Therefore, there is a considerable evidence of entry by Ryanair into routes previously operated by Aer Lingus. In view of Ryanair's continuing expansion and to the extent that they fit into Ryanair's business model, also for the future, entry into routes served by Aer Lingus is very likely.

510. Also Aer Lingus has entered routes on which Ryanair operated previously as the Incumbent (517). This applies to Dublin-Bristol on which it resumed operations in March 2004 (but exited in March 2007), to Dublin-Liverpool where it entered in October 2004 (but left in June 2006, even though it still operates to Manchester airport which can be considered substitutable for Liverpool) and to Dublin Newcastle where Ryanair commenced operation in January 2003 and Aer Lingus entered in October 2006. Further, both entered at the same time Dublin-Poznan (April 2006). On Dublin-Rome, in spite of the fact that Ryanair entered in April 2005, Aer Lingus has increased its number of frequencies since 2002. Further, Aer Lingus starts as of 25 March 2007 a new route from Cork to Manchester in competition with existing operations of Ryanair from Cork to Liverpool.

511. In view of these more limited instances of actual entry by Aer Lingus, Ryanair claims in its response to the Statement of Objections claims that Aer Lingus is not a potential competitor to Ryanair. The above summary (paragraphs 508-512) shows in fact that there have been many more instances of Ryanair entry on routes where Aer Lingus operated than vice versa. However, this may at least partly be explained by the historic development of the Irish air transport market. Until a few years ago; Ryanair concentrated only on Ireland - UK routes (in 2001, Ryanair operated only 2 routes between Dublin and Continental Europe). As part of its restructuring, Aer Lingus instead focused more on routes to Continental Europe, taking away capacity from the UK routes. As Ryanair started its expansion to Continental Europe only after Aer Lingus expanded its operations on these routes, it is only normal to find more instances of Ryanair's entry against existing operations of Aer Lingus than the other way round. It should be noted that the situation is different at present since also Ryanair has substantial operations to Continental Europe and it may be expected that there will be more examples of Aer Lingus' entry into Ryanair's routes from/to Ireland in the near future (518).

512. It is also noteworthy that on routes where Aer Lingus commenced services first and Ryanair followed later, Aer Lingus has remained in the market in nearly all cases. Thus, in contrast to other competitors, Aer Lingus has "survived" on a number of routes against Ryanair.

513. It can thus be concluded that there is a pattern of entry on routes on which previously only one of the two Merging Parties was active. This pattern is particularly strong for Ryanair.

7.6.4. Entry by other airlines on routes out of Dublin is less likely

514. Over the same period, one does not observe any similar pattern of entry on ex-Dublin routes by carriers with a base at the destination airport. For instance, on the Dublin-Berlin route two low-cost carriers easyJet and Germanwings have a base in Berlin. However, easyJet has tried and failed to enter other Irish routes in competition with Ryanair before. Further, neither easyJet nor Germanwings, each have tried to enter this specific route against Ryanair Thus, even on such a route where there are low cost carriers established at the other end, only the two Irish operators fly. The situation is similar for the route Dublin- Brussels; where Brussels Airlines has a base in Brussels (519). Prior to its merger with SN Brussels, VirginExpress served Brussels-Shannon but it withdrew in the face of entry by Ryanair on that route. Brussels Airlines has not entered the Dublin-Brussels route. On the contrary, where such entry took place in competition with the Merging Parties, it often failed. For instance, on Dublin-Doncaster, Ryanair and Thomsonfly started services in 2005 but the latter has since withdrawn from the route. On Dublin-Hamburg, Hapag-Lloyd Express commenced operations to Hamburg in April 2004 but withdrew from the route in January 2006. On Dublin-Jersey, British Airways CitiExpress withdrew 5 weekly flights when Aer Lingus commenced operations in April 2003; on Dublin-Prague Smartwings commenced operations in October 2004 but has withdrawn since.

515. Moreover, where third carriers already operated, subsequent to entry by one of the Merging Parties there has been a tendency of such carriers to exit or reduce operation on ex-Dublin routes. For instance, Air France operated on Dublin-Bordeaux since 2002 but reduced frequencies following Aer Lingus entry in September 2005 and withdrew from the route in January 2006. On Dublin-East Midlands, bmibaby withdrew in March 2005 following Ryanair's entry in March 2004; on Dublin-Malaga CityJet withdrew 2 weekly frequencies in January 2006; on Dublin-Newcastle following Ryanair's entry British Airways CitiExpress withdrew in 2003 all its 21 weekly frequencies; following Ryanair's entry, Alitalia withdrew 7 weekly services in October 2005 on Dublin-Rome; on Dublin- Stockholm FlyNordic entered in March 2005 but withdrew in October 2005 and Finnair in 2006. In 2001, Go Fly entered Dublin-Glasgow and Dublin-Edinburgh with 19 weekly frequencies each. After an aggressive fares war by Ryanair, which operated on Dublin- Edinburgh and Dublin-PIK, Go Fly left both routes in March 2002 (520).

516. Other cases of entry are mainly a defensive reaction to increased competitive pressure exerted by the Irish airlines. For instance, on Dublin-Barcelona, the low-cost carrier Clickair, which is partly owned by Iberia, announced that it will take over Iberia's place. On Dublin-Faro, TAP Air Portugal recently commenced operations, taking over Air Luxor's place, but exited in February 2007. Even though there are two instances when the new entrants stayed on the route, it should be noted that these new entrants do not exercise significant competitive constraint on the Merging Parties. First of these entrants is CityJet on the Dublin - London route which entered in November 2003 against existing operations of both Ryanair and Aer Lingus. CityJet however focuses on business passengers by serving the specific London City airport by smaller jet aircraft and thus cannot be considered as a significant competitive constraint on Ryanair - see further below in Section 7.8. Its market share on the route is around [0-10]*%. The second case is Spanair on the Dublin - Malaga route which entered in February 2005 also against existing operations of both Ryanair and Aer Lingus. On that route Spanair operates only during the summer season with around 2 weekly rotations on weekends. This service therefore resembles more a charter operation than the Merging Parties' daily services which are operated throughout the year on this route. Spanair's market share on the route is around [0-10]*%. For more details, see the analysis of the Dublin - London and Dublin - Malaga routes in Section 7.9.

517. In comparison to the above, the list of successful entry in competition with the Irish carriers is rather short: On Dublin-Krakow SkyEurope started operating in parallel to Aer Lingus in 2005 and on Dublin-Vienna SkyEurope commenced in October 2005 with 3 weekly frequencies in competition with Aer Lingus which has been on the route since 2002. On Dublin-Madrid Iberia and on Dublin-Prague Czech Airlines increased frequencies in competition with Aer Lingus. There has been no example of entry by any third carrier on a Dublin route where Ryanair already operated.

518. Given past experience, such entry by third carriers on routes to/from Dublin operated by Ryanair can not be expected in the foreseeable future. As discussed in more detail in section 7.8.4., carriers without a base in Dublin have been vulnerable to strategic entry deterrence by Ryanair. In comparison to such airlines, due to its base in Dublin, Aer Lingus is better placed to compete with Ryanair. It would be much more costly for Ryanair to attempt to cause Aer Lingus to exit the affected markets than other carriers which operate only on a few routes to/from Ireland. Such an attempt would require aggressive competition (a "price war") not only on a few routes, but at least on most of the existing overlap routes. Such a "simultaneous attack" on a multitude of routes would be expensive for Ryanair and - given Aer Lingus' additional commitment to its long-distance services from Ireland - of uncertain outcome. If Ryanair instead entered aggressively against Aer Lingus only on a few Dublin routes, Aer Lingus, operating on about 70 routes only from Dublin, might leave these routes, but would use the free capacity to re-enter other routes from the same airport where it has a strong base. Thus, unlike other competitors who were subject to aggressive pricing on individual routes and have withdraw from overlap routes in the past, Ryanair cannot reasonably expect that Aer Lingus would withdraw from the affected markets or even from Dublin in the foreseeable future. As a potential competitor, Aer Lingus exerts therefore a considerably stronger constraint on Ryanair than any other carrier.

519. This also holds for the two smaller carriers which already have significant operations in Dublin (CityJet and Aer Arann). Given their different business model and scope of activities in Ireland, these carriers cannot be considered to be likely entrants on Ryanair's or Aer Lingus' routes who could provide sufficient competitive pressure post merger (see also Sections 7.3. on closeness of competition and the individual analysis of the respective competitors in Section 7.8.).

520. In conclusion, while there has been a dynamic pattern of entry in particular by Ryanair on routes where Aer Lingus operates and to some extent also of entry by Aer Lingus on Ryanair-routes, the same is not true for third carriers. With two exceptions, the latter have not entered routes on which any of the two Irish airlines already operated. Moreover, where they faced entry by the latter they withdrew their services more often than Ryanair and Aer Lingus.

7.6.5. Dublin routes on which only one of the Merging Parties is present

521. The analysis of the development of the overlap routes has shown that both Ryanair and to a certain extent, Aer Lingus have the ability to enter onto each other's routes. They both derive important cost advantages from their strong presence at Dublin, including significant flexibility to reschedule capacity between routes. Also, there is a pattern of dynamic entry and capacity reallocation in routes out of Dublin, and both airlines have plans to continue expanding such their capacity. However, the Commission acknowledges that the threat of entry is not necessarily equally significant on all the routes where only one of the Merging Parties is present. Taking a prudent and "conservative" approach, the Commission has identified some routes on which the likelihood of entry is high (521) and has established certain criteria for such routes. Thus on the route in question and in light of the discussion in the above paragraphs, the merging party not present is a potential entrant.

522. The criteria on which the Commission has based its selection are as follows: (i) there is an "attractive" airport in the route for the potential entrant (this might differ for Aer Lingus and Ryanair in view of their focus on primary and secondary airports respectively) without any significant congestion hindering entry; (ii) the route is likely to have some growth potential (522); (iii) there is no other carrier currently active on the route (523) (the presence of an actual competitor is likely to constitute a stronger competitive constraint) (v) entry by other competitor is relatively unlikely and not imminent (e.g. as there is no significant potential entrant based at the destination airport or the existing entrants based at the other end are not likely to provide sufficient competitive pressure). There may, of course, be other reasons why a particular route may be particularly attractive for entry by either Ryanair or even Aer Lingus - for example the seasonality pattern complements well the existing networks and opening such route may increase overall profitability. Further, Ryanair is more likely to enter a route if it has already a base at the other end. Similarly, it is more likely that Ryanair or Aer Lingus enter the route if the destination is already served by them from other airport(s).

523. In May 2007, Aer Lingus operated on 20 intra-European routes to/from Dublin where Ryanair is not present (524). Out of these 20 routes, Aer Lingus faced existing competitors on the following 6 non-overlap routes to/from Dublin: Budapest (BUD) where Malev operates around 7 weekly rotations (525) and Aer Lingus up to 7 weekly rotations; Düsseldorf (DUS) where Aer Lingus operates around 11 weekly rotations while Germanwings operate around 3 weekly rotations to Köln airport (CGN) which can possibly be regarded as substitutable with DUS; Lanzarote (ACE) where Spanair operates around 1 weekly rotations and Aer Lingus around 3 weekly frequencies in the winter season only; Prague (PRG) where Czech Airlines operate around 13 rotations and Aer Lingus around 6 weekly rotations; Vilnius (VNO) where Air Baltic operates around 6 weekly rotations, FlyLAL around 4 weekly rotations and Aer Lingus up to 3 weekly rotations; Zurich (ZRH) where Swiss operates up to 5 weekly rotations and Aer Lingus around 7 weekly rotations. On these routes, the existence of an actual competitor is likely to be a stronger competitive constraint on Aer Lingus than the threat of potential Ryanair's entry.

524. On the other hand, Aer Lingus is the only carrier currently active on 14 of these routes as there are no competing carriers active on the same airport-pair or to any destination airport which could be prima facie considered substitutable for the airport served by Aer Lingus. These Aer Lingus monopoly routes connect Dublin with the following destinations: Amsterdam (AMS), Athens (ATH), Bordeaux (BOD), Dubrovnik (DBV), Geneva (GVA), Jersey (JER), Lisbon (LIS), Naples (NAP), Nice (NCE), Palma de Mallorca (PMI), Rennes (RNS), Munich (MUC), Santiago de Compostela (SCQ) (526) and Shannon (SNN). According to the information provided by Ryanair, there are no competing carriers based at the destination airport at least in case of 7 of these destinations (BOD, DBV, NAP, NCE, RNS, SCQ as well as SNN where only the Merging Parties are based). Further, a number of these 14 routes are predominantly holiday destinations (in particular ATH, DBV, GVA, NAP, NCE, PMI, and SCQ) for which in general a large proportion of customers originate in Ireland and where a destination based carrier would be a less strong competitor compared to Ryanair (this would apply in particular in case of PMI where according to Ryanair Spanair, Iberia/Iberworld, LTE Volare and Air Europe are based) (527). These holiday routes also in general have a potential for further growth of the scheduled traffic as evidenced by the actual overlap routes to holiday destinations (see in the Section 7.9. below the growth or even new emergence of the scheduled traffic in the recent years on the leisure routes to/from Dublin to sea & sun destinations such as Alicante, Faro, Malaga or Tenerife; ski destinations such as Lyon and Salzburg; or city break destinations such as Seville, Toulouse/Carcassonne and Venice). Further, Ryanair already operates flights from other destinations to at least 3 of these destinations (PMI, SCQ as well as SNN where Ryanair even has a base). However, even the other destinations cannot be excluded in view of the continuing expansion of the Ryanair's network of destinations (528), possibly with exception of capacity constraint and/or primary airports such as AMS, ATH, LIS or MUC.

525. In view of the above, Ryanair appears to be a likely entrant to a number of these routes served currently only by Aer Lingus. However, in determining and applying the criteria of the Horizontal Merger Guidelines, the Commission is taking a cautious and conservative approach to identify those concrete routes where it is most likely that entry has a significant constraining influence. Therefore, by applying the above discussed criteria in a restrictive manner, it is concluded that (i) Ryanair exerts a significant constraining influence on Aer Lingus or is significantly likely to grow into an effective competitive force for Aer Lingus, and (ii) there are no other potential competitors which could maintain sufficient competitive pressure after the merger (529), on the Aer Lingus' routes between Dublin and DBV, NAP, NCE, PMI and SCQ (530).

526. As regards Ryanair, it operates currently on 33 intra-European routes to/from Dublin where Aer Lingus is not present (531). Ryanair currently faces existing competitors on the following 8 non-overlap routes to/from Dublin: Bournemouth (BOH) where Ryanair operates around 7 weekly rotations while Flybe operates around 20 weekly rotations to Southampton airport (SOU) which can possibly be regarded as substitutable with BOH; Malmo (MMX) where Ryanair operates up to 3 weekly rotations while SAS operates around 14 weekly rotations to Copenhagen airport (CPH) which can possibly be regarded as substitutable with MMX; Oslo (TRF) where SAS operates around 5 weekly rotations and Ryanair around 7 weekly rotations; Stockholm Skavsta (VST) where Ryanair operates up to 6 weekly rotations while SAS operates around 7 weekly rotations and Flynordic around 1 weekly rotation to Stockholm airport (ARN) which can possibly be regarded as substitutable with VST; Wroclaw (WRO) where Centralwings operate around 4 weekly rotations and Ryanair around 4 weekly rotations; Bristol (BRS) where Ryanair operates around 24 weekly rotations while Aer Arann operates around 16 weekly rotations to Cardiff airport (CWL) which can possibly be regarded as substitutable with BRS; Gdansk (GDN) where Centralwings operate around 4 weekly rotations and Ryanair around 2 weekly rotations; Cork (ORK) where Aer Arann operates around 40 weekly rotations and Ryanair around 34 weekly rotations. On these routes, the existence of an actual competitor is likely to be a stronger competitive constraint on Ryanair than the threat of potential Aer Lingus' entry.

527. On the other hand, Ryanair is the only currently active carrier on 25 of these routes as there are no competing carriers active on the same airport-pair or to any destination airport which could be prima facie considered substitutable for the airport served by Ryanair. These Ryanair monopoly routes connect Dublin with the following destinations: Aberdeen (ABZ), Alghero (AHO), Biarritz (BIQ), Billund (BLL), Blackpool (BLK), Doncaster (DSA), Friedrichshafen (FDH), Gothenburg (GSE), Karlsruhe (FKB), Kaunas (KUN), La Rochelle (LRH), Lodz (LCJ), Malta (MLA), Nantes (NTE), Pisa (PSA), Pula (PUY), Rzeszow (RZE), Tampere (TMP), Trapani (TPS), Eindhoven (EIN), Bydgoszcz (BZG), Porto (OPO), Almeria (LEI), Valencia (VLC) and Bremen (BRE). According to the information provided by Ryanair, there are no competing carriers based at the destination airport at least in case of 16 of these destinations (AHO, BIQ, FDH, GSE, KUN, LRH, LCJ, NTE, PUY, RZE, TMP, TPS, BZG, LEI, VLC and BRE). On the contrary, Ryanair is also based in Pisa and Bremen (532). Further, a number of these routes are predominantly holiday destinations (in particular AHO, BIQ, LRH, MLA, PSA, PUY, TPS, OPO, LEI and VLC) for which in general a large proportion of customers originate in Ireland and where a destination based carrier would be less strong competitor compared to Ryanair (533). These holiday routes also in general have a potential for further growth of the scheduled traffic as evidenced by the actual overlap routes to holiday destinations (see in the Section 7.9. below the growth or even new emergence of the scheduled traffic in the recent years on the leisure routes to/from Dublin to sea & sun destinations such as Alicante, Faro, Malaga or Tenerife; ski destinations such as Lyon and Salzburg; or city break destinations such as Seville, Toulouse/Carcassonne and Venice). Further, there are several destinations to the Central and Eastern Europe (KUN, LCJ, RZE and BZG) which are rather dynamically growing routes as evidenced by the similar actual overlap routes analysed in Section 7.12. (e.g. Dublin - Krakow, Dublin - Warsaw, Dublin - Poznan or Dublin - Riga).

528. In view of the above, Aer Lingus appears to be a likely entrant to a number of these routes served currently only by Ryanair. However, in determining and applying the criteria of the Horizontal Merger Guidelines, the Commission is taking a cautious and conservative approach to identify those concrete routes where it is most likely that entry has a significant constraining influence. As indicated above, the evidence of the past entry by Aer Lingus to Ryanair's routes is not as strong. Therefore, even though this may change in the near future, the Commission concludes that it is not able to establish to the necessary legal standard that Aer Lingus exerts a significant constraining influence or is significantly likely to grow into an effective competitive force on the above identified routes to/from Dublin operated by Ryanair.

7.6.6. Cork routes where only one of the Merging Parties is present

529. As regards Cork, Aer Lingus has four A320 aircraft based at that airport. It has increased the number of routes served from Cork from 1 in 2002 to 17 today. Ryanair opened a base in Cork in September 2005 and operates today four routes with a B737-800 (189 seater) also based at Cork. As discussed above, though Aer Arann has based 2ATR-72 (66 seater) and one ATR-42 (50 seater), it only exerts a limited constraint, given that it operates turboprop aircraft. It operates six routes to small regional airports in the UK. Several destination-based carriers have recently ceased operations at Cork. The remaining ones are bmibaby (Manchester/Liverpool and Birmingham) and Jet2 (Newcastle).

530. Presently, Aer Lingus operates on 15 intra-European routes to/from Cork where Ryanair is not present (534). Only on two routes to/from Cork Aer Lingus currently faces competition from other carriers: Warsaw (WAW) where Centralwings operates up to 3 weekly rotations and Aer Lingus around 4 weekly rotations; Birmingham (BHX) where bmibaby operates around 7 weekly rotations and Aer Lingus around 4 weekly rotations.

531. On the other hand, Aer Lingus is the only currently active operator on 13 of these routes as there are no competing carriers active on the same airport-pair or to any destination airport which could be prima facie considered substitutable for the airport served by Aer Lingus. These Aer Lingus monopoly routes connect Cork with the following destinations: Tenerife (TNS), Lanzarote (ACE), Faro (FAO), Malaga (AGP), Madrid (MAD), Alicante (ALC), Barcelona (BCN), Nice (NCE), Rome (FCO), Paris (CDG), Prague (PRG), Amsterdam (AMS) and Berlin (SXF). According to the information provided by Ryanair, there are no competing carriers based at the destination airport in case of at least 4 of these destinations (TNS, ACE, FAO, NCE). It should also be noted that in particular full-service network carriers based at the destination airports are less likely to enter the less important Cork Airport. Ryanair is also based in Madrid, Barcelona and Rome (in the latter two cases at airports which can be considered as substitutable for BCN and FCO - see above Section 6.3) and they could thus use the aircraft based in these destinations to enter these routes to/from Cork. Further, it also operates to other destinations including TNS, FAO, AGP, ALC (Ryanair operates to a substitutable airport Murcia - see Section 6.3 above), Paris Beauvais (which can be considered as substitutable for CDG - see above Section 6.3) and SXF. Similarly as in case of the overlap routes to/from Dublin analysed in detail in Section 7.9 (route-by-route analysis), the carriers based at the destination airports on the Cork routes to these destinations are significantly less likely to be potential competitors providing significant competitive constraint as compared to Ryanair. Further, a number of these routes are predominantly holiday destinations (in particular TNS, ACE, FAO, AGP, ALC, NCE) for which in general a large proportion of customers originate in Ireland and where a destination based carrier would be less strong competitor compared to Ryanair (535). These holiday destinations also in general have a potential for further growth of the scheduled traffic as evidenced by the actual overlap routes to holiday destinations (see Section 7.9). Further, as according to Section 7.9. Dublin - Barcelona route is steadily growing and Ryanair entered the Dublin - Madrid and the Dublin - Berlin routes (against existing operations of Aer Lingus) only in December 2006 and June 2006 respectively, it can be argued that there is a potential for Ryanair's entry also with respect to the routes from Cork to these two destinations. As regards FCO, even though the Dublin-Rome route operated by both Merging Parties is rather mature, it could be expected that there would be scope for Ryanair's entry from Cork to Rome, as Rome is an important destination currently served from Cork only by Aer Lingus. On the other hand, Ryanair is a less likely entrant on routes to capacity constraint and/or primary airports not served by them such as AMS, or PRG.

532. It should be also acknowledged that Ryanair currently bases only one aircraft in Cork. However, in view of the fact that Ryanair have a significant number of aircraft on order they could easily increase their presence in Cork. Further, on the routes to Madrid, Barcelona and Rome, Ryanair could use the aircraft based at the destination airport to serve the route, as they do in several cases of the current overlap routes (including the Dublin - Rome route (536)). As regards a constraint from potential competition, given their already existing operations at Shannon airport, their brand recognition in Ireland and their rapid expansion on the Irish market (537), the Merging Parties can be considered to be each other's closest competitor on the routes to/from Shannon.

533. In view of the above, it can be considered that Ryanair would be a likely entrant to a number of these routes to/from Cork served currently only by Aer Lingus. However, in determining and applying the criteria of the Horizontal Merger Guidelines, the Commission is taking a cautious and conservative approach to identify those concrete routes where it is most likely that entry has a significant constraining influence. Therefore, by applying the above discussed criteria in a restrictive manner, it is concluded that (i) Ryanair exerts a significant constraining influence on Aer Lingus or is significantly likely to grow into an effective competitive force for Aer Lingus, and (ii) there are no other potential competitors which could maintain sufficient competitive pressure after the merger (538), on the Aer Lingus' routes between Cork and TNS, ACE, FAO, AGP, ALC, NCE, MAD, SXF, FCO and BCN.

534. As regards Ryanair, there are currently no Ryanair routes to/from Cork on which Aer Lingus is not present (apart from the Dublin - Cork route discussed already above as a non-overlap route to/from Dublin).

7.6.7. Shannon routes where only one of the Merging Parties is present

535. Four airlines operate short-haul operations to Shannon. FlyLAL offers a service from Shannon to Vilnius and Centralwings operates to Gdansk and Warsaw. Aer Lingus operates from Shannon to Dublin and London, with aircraft based at the other airports. Ryanair has a small base at Shannon with four B 737-800 aircraft, operating to more than 20 destinations. Several other carriers have failed to successfully enter routes to/from Shannon (539).

536. Presently, Ryanair operates on 24 intra-European routes to/from Shannon where Aer Lingus is not present (540). On all these routes Ryanair is the only currently active operator as there are no competing carriers active on the same airport-pair or to any destination airport which could be prima facie considered substitutable for the airport served by Ryanair. These Ryanair monopoly routes connect Shannon with the following destinations: Faro (FAO), Malaga (AGP), Madrid (MAD), Murcia (MJV), Barcelona Girona (GRO), Biarritz (BIQ), Carcassonne (CCF), Rome (CIA), Milan (BGY), Venice-Treviso (TSF), Nantes (NTE), Paris Beauvais (BVA), Brussels Charleroi (CRL), Frankfurt Hahn (HHN), Düsseldorf Weeze (NRN), Krakow (KRK), Wroclaw (WRO), Lodz (LCJ), Glasgow Prestwick (PIK), Edinburgh (EDI), Liverpool and Manchester (LPL and MAN), East Midlands (EMA), Bristol (BRS) and Bournemouth (BOH). According to the information provided by Ryanair, there are no competing carriers based at the destination airport at least in case of 5 of these destinations (FAO, BIQ, NTE, WRO, LCJ). On the contrary, Ryanair is also based in Madrid, Barcelona, Rome, Milan, Brussels, Frankfurt, Düsseldorf Weeze, Glasgow, Liverpool and East Midlands (541). Further, a number of these routes are predominantly holiday destinations (in particular FAO, AGP, MJV, BIQ, CCF, TSF) for which in general a large proportion of customers originate in Ireland and where a destination based carrier would be less strong competitor compared to Ryanair (542). These holiday destinations also in general have a potential for further growth of the scheduled traffic as evidenced by the actual overlap routes to holiday destinations (see Section 7.9.). There are also several destinations to the Central and Eastern Europe (KRK, WRO and LCJ) which are rather dynamically growing routes as evidenced by the similar actual overlap routes analysed in Section 7.12. Further, Aer Lingus is already active to some of these destinations (to the same airports or airports substitutable in line with the conclusions of the Section 6.4.) from Dublin (FAO, AGP, MAD, MJV, GRO, CCF, CIA, BGY, TSF, BVA, CRL, HHN, KRK, PIK, EDI, LPL/MAN and EMA). Similarly as in case of the overlap routes to/from Dublin analysed in detail in Section 7.9., the carriers based on the destination airports on the Shannon routes to these destinations are significantly less likely to be potential competitors providing significant competitive constraint as compared to Aer Lingus.

537. As regards Aer Lingus, there are no routes to/from Shannon on which Ryanair is not present (apart from the Dublin - Shannon route discussed already above as a non-overlap route to/from Dublin).

538. In view of the above, Aer Lingus might enter some of these routes served currently only by Ryanair. However, in determining and applying the criteria of the Horizontal Merger Guidelines, the Commission is taking a cautious and conservative approach to identify those concrete routes where it is most likely that entry would have a significant constraining influence. As Aer Lingus is currently not based in Shannon with any shorthaul aircraft, its flexibility to enter these routes is lower. Even though it could be presumed that should the opportunity arise, Aer Lingus would be able and willing to base aircraft in Shannon and enter new routes (in view of its experience with the Irish market as well as existing activities in Shannon), this entry appears less imminent than in the case of an existing base. Further, as described above, the evidence of the past entry by Aer Lingus to Ryanair's routes is not as strong. Therefore, the Commission concludes that it is not able to establish to the necessary legal standard that Aer Lingus would exert a significant constraining influence or is significantly likely to grow into an effective competitive force in case of the above identified routes to/from Shannon.

7.6.8. Conclusion on the elimination of potential competition

539. Against the backdrop of the pattern of entry in competition with each other and the limited impact of entry by other carriers on the two Merging Parties, the Commission concludes that Ryanair and Aer Lingus exert a competitive constraint on each other on a number of non-overlap routes. This applies in particular to Dublin and Cork, but also to a lesser extent to Shannon.

540. However, as regards the identification of the concrete routes where it is most likely that entry has a significant constraining influence, the Commission adopted a cautious and conservative approach. The Commission therefore concludes that the notified merger is likely to significantly impede effective competition as a result of the creation or strengthening of a dominant position by the elimination of a credible potential entrant on the above identified 5 current Aer Lingus' routes to/from Dublin (Dubrovnik - DBV, Naples - NAP, Nice - NCE, Palma de Mallorca - PMI and Santiago de Compostela - SCQ) and 10 further Aer Lingus' routes to/from Cork (Tenerife - TNS, Lanzarote - ACE, Faro - FAO, Malaga - AGP, Alicante - ALC, Nice - NCE, Madrid - MAD, Berlin - SXF, Rome - FCO and Barcelona - BCN).

7.7. The Merging Parties' fragmented customers have no countervailing buyer power and limited or no possibilities of switching suppliers

541. According to the Commission's practice set out in the Horizontal Merger Guidelines (543), a merger leading to high market shares is particularly likely to lead to price increases (544) when customers of the Merging Parties are likely to have difficulties or to be unable to switch to other suppliers, because there are few or no alternative suppliers. Such customers are, according to the Guidelines, "particularly vulnerable to price increases (545)".

542. The proposed transaction would significantly reduce customers' ability to switch between different suppliers, since it would not only reduce the number of alternative airlines to whom they could turn in case of a price increase post-merger, but remove the possibilities to switch for customers entirely on many routes. Indeed, on 22 routes, the merger would create a monopoly of the merged entity on these routes. On 9 other routes, the merger would reduce the number of alternatives from currently 3 to 2. Only on 4 of 35 markets, the merged entity would face competition from more than one competitor post merger.

543. It is also obvious that airline customers have no countervailing buyer power to offset Merging Parties the anti-competitive effects of the merger. This is particularly true since sales to larger corporate customers or tour operators are negligible. The vast majority of almost 15 million passengers per year who are potentially affected by the transaction cannot be expected to have any countervailing buyer power vis-à-vis the merged entity, since their individual ticket value is of no commercial significance for the Merging Parties and since they have few or no possibilities to switch to alternative suppliers (546).

544. It can thus be concluded that customers are particularly vulnerable to price increases in the present case since they would have no or very limited possibilities to switch to other suppliers on almost all affected routes and no buyer power to offset such price increases. There is indeed a high risk of anti-competitive effects for those customers (in particular price increases, loss of choice and service quality), in particular because both competitors are currently the only or main source of competition on the affected routes, which would immediately disappear after the merger.

7.8. Entry is unlikely to defeat the anticompetitive effects of the mergers

7.8.1. Principles

545. The Commission has examined to what extent entry is likely to constrain the behaviour of the Merging Parties post-merger. For entry to be considered as a sufficient competitive constraint on the Merging Parties, it must be shown to be likely, timely and sufficient to deter or defeat any potential anti-competitive effects of the merger (547).

546. Ryanair submits that there are no or only insignificant barriers to enter the overlapping markets; competitors wishing to enter on routes from and to Dublin/Ireland could easily do so (548). In particular, Ryanair argues that there are no significant barriers to entry on the overlapping routes for either low-frills or traditional network airlines, in particular as regulatory barriers and obstacles have been removed within the EU. Ryanair therefore considers that the mere perspective of entry should be sufficient for the Commission to dismiss any competition concerns and claims that the Commission had already established a case practice in this respect.

547. Barriers to entry can be defined as features of the market, which give the incumbent firms an advantage over potential competitors (549). According to the Court of First Instance, barriers to entry "(...) may consist in elements of various natures, in particular economic, commercial or financial elements, which are likely to expose potential competitors of the established undertakings to risks and costs sufficiently high to deter them from entering the market within a reasonable time or to make it particularly difficult for them to enter the market, thus depriving them of the capacity to exercise a competitive constraint on the conduct of the established undertakings (550)."

548. Hence, the Commission has to make sure that potential entry is not only a theoretical and remote possibility, but will be an immediate and actual threat for the Merging Parties, which will exert a competitive constraint on them, and this to an extent that they would refrain from any merger-induced anti-competitive behaviour. It should be noted that it is not necessary to show that entry is impossible on a given route in order to dismiss the likelihood of entry as a countervailing factor. It is sufficient that the Commission finds that the degree of likelihood for entry is low or that the scope of the expected entry is not sufficient to defeat the anticompetitive effects of the merger in all affected markets. This holds in particular true in the present case, where the Merging Parties hold very high market shares on all affected markets, often amounting to a monopoly position. In such a situation, new entry by potential competitors would need to be particularly likely, substantial and timely to reverse the competitive impact of a merger leading to nearmonopoly market shares.

549. The Commission's market investigation (551) has, however, confirmed that there are a number of important barriers to enter and/or expand on all affected routes which altogether make it unlikely that any competitor would replace the competitive constraint exerted by the Merging Parties today after the merger.

550. The Commission also underlines that in markets in which the Merging Parties reach very high market shares or create even a monopoly on not less than 35 routes, affecting more than 14 million customers, a remote "likelihood" of entry which could offset the anticompetitive effects cannot be sufficient to dismiss the Commission's competition concerns on the relevant route-pairs. The Commission cannot dismiss the competition concerns in such a situation without clear indications that actual or potential entry would constrain the Merging Parties (552).

7.8.2. Regulatory barriers do not play an important role as an entry barrier

551. The Commission agrees with Ryanair (553) that regulatory barriers do not play an important role as a barrier to entry in the affected markets (554). All overlaps are on routes within the European Union, in which regulatory barriers have been largely removed for EU-based Airlines (555). Although opening a new route may still require some regulatory and administrative approvals, these approvals can be obtained within a relatively short timeframe of several weeks or even days. The market investigation has confirmed that competitors do not consider the need for regulatory approvals as an important barrier to entry (556).

7.8.3. Entry barriers related to Ryanair's and Aer Lingus' strong position with large bases in Ireland

552. As described above, the fact that Ryanair and Aer Lingus operate from the same base in Dublin, and are also in the two other main airports in Ireland, Cork and Shannon, is one of the reasons why they are the closest competitors on routes from and to Ireland. This constraint on each other goes beyond the fact that they compete on a large number of routes. The "additional" competitive constraint comes from the fact that by operating a number of routes from bases in Ireland, the Merging Parties constrain each other also with a view to potential competition.

553. The Commission has also in the past passenger air transport cases recognised that economies of scale and scope stemming from flexibility of assets at a base and from ability to spread fixed costs over many markets (routes) constitutes a barrier to entry and that "a carrier with an established base of operations at a particular airport will benefit from clear cost advantages" (557). None of these past cases, however, involved a combination of two carriers with such a strong presence at one airport as is the case in the notified concentration. The role of the base-related barriers to entry was thus less significant in those past cases.

554. The Commission has assessed under what conditions other carriers could replace the competitive constraint on the merged entity and to what extent the absence of a base in Dublin might constitute a barrier. Three different forms of entry can be distinguished. First, there may be an entrant which establishes a base at Dublin (or at the other Irish airports) to compete on a number of routes ex-Ireland. This could be either a carrier establishing a new base in the Irish airports, or a carrier which has already a base at these airports and would expand that base. Second, one may consider entry on a particular route by an operator which has already a base at the other end ("destination-based entrant"). The third option is an entrant which enters on a route-by-route basis without having a base at either end. The three options are discussed below.

7.8.3.1. Competitors which have a base at Dublin are not likely to efficiently constrain the Merging Parties

555. As set out above (Section 7.2), there exist two main advantages from having a significant operation at one airport (a base). First, it allows a carrier to benefit from certain cost advantages. Second, it raises the carrier's flexibility to react quickly to shifts in demand and competition on routes out of the same airport. The larger the number of operations from a particular airport, the larger is the slot portfolio, the number of aircraft and crew available, thus, the greater is the carrier's ability to react to changes in the market place.

556. The Merging Parties' main base in Ireland is Dublin. There are two other carriers with a base in Dublin - CityJet with 3 aircraft and Aer Arann with 4 aircraft. This compares with 22 short-haul aircraft of Aer Lingus (planned to be increased to 23 in May 2007) and 20 aircraft of Ryanair at Dublin. However, these two other carriers would exert only a limited competitive constraint on the merged entity on routes out of Dublin (558), in particular due to their different type of business model. Due to their large route portfolio, Ryanair and Aer Lingus can quickly react to entry, expansion or exit of a competitor on almost any route ex-Dublin. For a carrier such as Aer Arann, that operates only few routes served by small aircraft out of the airport, the scale and flexibility advantages are more limited. While each of the two Irish airlines already has a significant advantage over the other two base operators in Dublin, this advantage would increase post merger by merging their respective fleets (559).

557. In this respect it is noted that neither of the two smaller carriers in Dublin have expanded their Dublin base significantly in the last years. They have also refrained from developing new routes or - with a few exceptions - from switching to other Dublin routes. CityJet has stated that it would only be interested in opening some routes to the UK, notably the Dublin-London route (if some of the precious Heathrow slots were available). There is no indication that CityJet could and would enter on any of the continental "leisure" routes. Moreover, even if CityJet, currently operating with a limited number of smaller aircraft, were to significantly expand their operations in Dublin, the competitive constraint on the Merging Parties would still be limited, since CityJet is focused on business passengers and needs higher fares to cover the higher costs of their smaller aircraft (560). As a subsidiary of Air France, CityJet mainly feeds the latter's network (561). If CityJet wished to compete effectively with the merged entity, it would not only have to operate many more aircraft, it would also have to operate with a different type of aircraft, i.e. models which have a similar seat capacity as the ones employed by the Merging Parties (A320 or the B737NG). This would not only require the acquisition of new aircraft but also generate further cost of integrating different fleets, as for instance for the training of staff (pilots) etc.

558. Aer Arann defines itself as a regional carrier. It is positioned towards a full service offering focusing on business customers on shorter routes to and from Ireland. Aer Arann operates with a fleet of smaller aircraft and has stated that it would not enter any route on which Ryanair is present (562).

559. Any expansion would furthermore increase problems of slot congestion at Dublin Airport. Depending on the route they wish to enter, this competitor carrier may also face slot congestion at some of the destination airports.

560. Other charter and scheduled airlines that have an aircraft parked overnight at Dublin for a season or throughout the year do not enjoy the same advantages as the Merging Parties. While they may be able to use the early morning and late evening slots at Dublin, they are not able to reap the full benefits of flexibility of switching between routes, redeployment of aircraft, minimising disruption costs, exchange of crews, customer care, brand awareness etc.

7.8.3.2. There are not sufficient indications that a new carrier would enter with a base in Dublin or Cork and Shannon

561. Setting up a base in Dublin would require an even more significant upfront investment. In order to realise the flexibility of supply that Ryanair and Aer Lingus currently enjoy and even more so that the merged entity would enjoy post merger, a new carrier wishing to set up a base in Dublin would need a sizeable fleet of modern, fuel efficient jets with more than 130 seats (such as A320 or B737NG). In addition it would require investments e.g. for maintenance facilities, customer care facilities, etc.

562. The Commission has closely analysed whether there are indications that one of the airlines listed by Ryanair as potential competitors or any other airline would open a base in Ireland in the event of a post-merger price increase (563). However there is no indication from the Commission's market investigation that any carrier has the incentive to establish or develop a significant base at Dublin. Indeed, most of the airlines cited by Ryanair are network carriers who are likely to be primarily interested in spoke development - feeding passengers to their respective home country hubs. The market investigation has confirmed that none of these carriers would be interested in setting up a base at Dublin (564). Given their incentive to feed their existing hub, they would also consider it unprofitable and not in their business model to open and develop individual routes from Dublin to destinations such as Seville, Bologna, Milan, Brussels, Glasgow, Hamburg, Marseille, etc. Moreover, the Commission notes that a number of such carriers have pulled out of Dublin in the last years. For example, British Airways, cited by Ryanair consistently as one of its main potential competitors (together with CityJet (Aer France) and easyJet (565)), does not serve Dublin from its Heathrow hub (but only from London-Gatwick) anymore and Alitalia ceased its Dublin-Milan services in October 2006. Further, the aggressive behaviour with which Ryanair has reacted against entry on Irish routes is likely to be an additional barrier that will be considered by competitors before considering entry (see further below, Section 7.8.5.).

563. In addition, the existing congestion problems at Dublin (566) make Dublin an unattractive choice to establish a base for many potential competitors. Competitors may fear that they would not get the necessary contact stands or would not be able to use the runway at the commercially important peak-times are unlikely to choose Dublin as a new base if they have the choice of entering other destinations (567).

564. With regard to Cork and Shannon, entry by establishing a base could take place on a smaller scale. Moreover, at these airports there does not appear to exist any major congestion problems. However, due to the general entry barriers discussed further below, in particular the need to establish a brand and the entry deterring behaviour by Ryanair, it appears that any significant entry (or expansion) is unlikely. Due to this behaviour a number of carriers which were operating at these airports have withdrawn their services in the last years (568). In any event, any entry or expansion at these airports has to take into account the strong position of the merged entity post-merger. In both of these airports the merged entity will be the only carrier with a base of large jet aircraft that serves a significant number of European destinations.

565. On the basis of the above, the Commission concludes that any significant entry through the establishment of a base in Dublin, Cork and Shannon is unlikely in the short to medium term. Even if small scale entry were to take place, it is unlikely that such entry would be sufficient to replace the constraint that the two Merging Parties have had on each other until today.

7.8.3.3. Potential entrants with a base at the other end of a route face disadvantages

566. Ryanair has argued that even if some advantages from having a base existed, these advantages would not be decisive since on many routes there exists a competitor which operates already with a base at the other end. Such an operator ("destination-based competitor") would equally benefit from the cost advantages of a base and it would have a similar flexibility to switch capacity to a route from this airport to Dublin, if that is commercially attractive.

567. While the Commission agrees that the base advantages may, in theory, relate to both ends of a given route, it has found that in the present case competitors with a base outside Ireland face certain disadvantages on routes from and to Ireland.

568. The cost advantages the merged entity would enjoy by having a large base at Dublin are more significant for Dublin than Cork and Shannon. There is an asymmetry of origination on a number of the overlap routes (in particular continental destinations) as will be discussed in the route-by-route assessment. On these routes, carriers based at the destination end are in a less favourable position than carriers with a base at Dublin to attract Irish-originating passengers (569). At the same time, as it operates only one route to Dublin, it would not be able to spread marketing expenses directed at Irish-origin passengers over several Irish routes. This implies that the unit marketing costs are lower in the case of a base competitor at Dublin, by virtue of the greater volumes that would be achieved on routes with predominantly Irish origination.

569. Further, destination based carriers are unlikely to show the same degree of commitment as a Dublin-based carrier to maintain a service to Dublin. For a destination base carrier the Dublin route represents just one route of several alternative routes out of its base. It would make a decision on which route to enter based on the likelihood of successful entry. For example, it may have the potential to enter other routes in which it would not face a competitor as aggressive as Ryanair/the merged entity. In this knowledge, the merged entity may be more prone to react aggressively against entry in the knowledge that those carriers could leave the route for good (and refrain from entering other Dublin markets) if faced with strong competition. As described above, there has been significant exit on ex- Dublin routes by both network carriers and low-frills/low-cost carriers operating from bases at destination airports. In addition, on many important routes the merged entity will already have a base also at the destination end, which gives them the advantage of having a base at both ends of the routes which potentially further increases their strength and flexibility vis-à-vis their competitors (570).

570. In addition many carriers based at the destination end operate a business model which is different from Aer Lingus and Ryanair. Most of them provide feeder services concentrating on transfer traffic for which the Merging Parties do not compete. Others provide full-frills services focussing on business customers. Such carriers exert a much smaller competitive constraint on the merged entity than Ryanair and Aer Lingus are exerting on each other today on routes ex-Ireland. Further, they would not be an alternative for a large part of Ryanair's and Aer Lingus' customers who are looking for a low fare alternative that would replace Aer Lingus.

571. In any case, it should also be noted that there exist today around 30 routes from Dublin on which Ryanair or Aer Lingus, or both of them, operate and where no competitor is based at the other end. On these routes, at least in the short run, the merger's elimination of actual or potential or actual competition could not be replaced by a destination-based carrier (571).

7.8.3.4. New point-to-point entry on a non-base route is unlikely

572. Ryanair has put forward two other alternatives for entry. First, a carrier could enter a route without a base at either end on a mere "point-to point" basis. Indeed, on all routes where there is no other competitor with a base on either end of the city pair, subsequent to the merger the competitive constraint would only arise from the possibility that another carrier enters point-to-point. This carrier would have to dedicate an aircraft to that route only. This competitor may fly to the route from its own base or just park the aircraft over night without having a base at the respective airport.

573. The most common option (572) of (non-base) point-to-point flights are so called "W" or "triangular" flights. On such a "W"-flight, the aircraft first flies from its base to a "nonbase" destination. Instead of returning to the base, the aircraft then flies one or several times to another (non-base) destination (e.g. a shorter destination to use the "weak demand" over lunchtime) during the day and only returns to the base in the evening. There are some limited examples of carriers operating according to such a "W" or "triangular" model (573). The Commission has, however, found that this operating model is not regarded as optimal by most carriers as they usually prefer operating from a base (574).

574. Operating between two points which are not connected to a base entails an increased risk. Such risk, arises for instance, if a replacement aircraft is needed in case of technical problems on the route. Moreover, such entry would only be commercially viable for routes which generate sufficient traffic for 3-4 roundtrips a day since otherwise the dedicated aircraft would not be used sufficiently. According to the market investigation, such simple point-to-point entry is therefore very rare (575). As pointed out above, with only four exceptions, Ryanair itself does not operate in such a manner. Neither does e.g. easyJet (576).

575. The Commission's investigation has not supported the view that entry or entry on a sufficient scale can be expected through the use of W or triangular flights. Indeed, competitors have explained to the Commission that such an operation has significant disadvantages. Third parties mentioned, for instance, that it is difficult to operate on "W"- routes because the crews cannot be simply exchanged in the middle of the day on a non base-connected destination. W-routes would therefore only allow fewer return flights and reduce the total operation time of an aircraft. Although some smaller regional carriers operate some W-routes/triangular routes, these can be operated efficiently only under specific circumstances. The majority of carriers try to do avoid it or to operate such services only on a transitional basis (577).

576. "W" flights or triangular can be a targeted solution on certain routes and if the carrier wishes to make use of some idle aircraft capacity. "W" entry would mainly be an add-on to a route which is connected to the carrier's base. Maintaining such flights would depend inter alia on developments of the other leg of the "W" operation.

577. Thus, "W" or triangular flights provide an even less competitive constraint than destination and base-entry described above. This would normally only be used for certain routes which can be combined with another operation. The cost advantages which arise from operating from a base and the flexibility as regards route-by-route market entry do not, or only to a smaller extent, arise in the case of "W" flights. For instance, the carrier would not be free to add capacity on the route, in reaction to demand or supply shifts. This would only be commercially viable if at the same time it also wishes to add capacity to the other leg of the "W" operation.

578. This conclusion is supported by the evidence. The Commission has verified to what extent airlines actually operate flights between points which are not connected to a bases. The result clearly showed that such connections are the exception (578).

579. Already destination-based entry has been considered insufficient to replace the competitive constraint removed by the proposed merger. This is even more so for carriers which do not have a base at either end or provide "W" services. Such punctual entry may appear on certain routes, however, it is even less likely to replicate the dynamic competitive environment which presently exists on routes out of Dublin and to a lesser extent out of the other two Irish airports.

580. Finally, even more than a destination-based carrier, a non-base competitors entering on a point-to-point basis faces the risk of a strategic response from the incumbent. This amounts to an additional entry barrier for such carriers:

581. After the merger, a non-base entrant on a particular route from Dublin may encounter a more aggressive response than an entrant which establishes a base in Dublin. A basecarrier has invested sunk cost (marketing, brand, service contracts etc) in establishing itself at a particular airport. As these investments would be lost if the base is given up, expelling such a competitor from its base is more costly/difficult. It would require attacking simultaneously this competitor on a number of markets. Indeed, if the merged entity is forced to leave only some routes, as it is dedicated to Dublin, it may use the free capacity to re-enter another route out of the same airport - on which it may then compete more fiercely with the other incumbent. This is different with regard to a carrier which enters a new route from Dublin without a base. A non-base-entrant may face a more severe response from the incumbent airline since the latter may expect the entrant to leave the route for good (and refrain from entering other Dublin markets).

582. Moreover, the merger itself could increase the scope for such strategic behaviour as the merged entity could use the base advantage for entry deterrence. For instance, if threatened by entry on a particular route, the base airline can quickly re-allocate slots such that it can offer the most attractive schedules on the challenged route. This flexibility can also be used for entry deterrence, for instance by arranging departure times close to the ones of the entrant ("sandwiching"). Without the merger, the scope for such strategic measures is reduced as the two base carriers control each other at Dublin Airport. If, for instance, today Ryanair withdrew capacity from one route to deter a non-base entrant on another, it would face the risk that Aer Lingus would enter (or expand on) the route from which Ryanair has withdrawn this capacity.

583. Thus, it can be concluded that non-base entrants are even more vulnerable and less likely to exert a competitive constraint than destination-based carriers. This conclusion applies as well to Shannon and Cork where the potential competition constraint exerted by the Merging Parties on each other today is more significant than the constraint which comes from non-base entrants.

7.8.3.5. Conclusion on base-airport related barriers to entry

584. Thus, on the basis of the above the Commission concludes that destination-based entry and non-base (triangular, W or point-to-point) entry are in general not sufficient to replace the competitive constraint exerted by the Merging Parties on each other pre-merger. On routes from Dublin, Aer Lingus and Ryanair have a number of advantages over other competitors due to their strong base in particular in Dublin in addition to their strong position in Cork and Shannon. Even though entry may occur on particular routes, such selected entry cannot replace the competitive pressure exerted by the two Merging Parties today. 585. In this regard it is noted that competitors with bases at the destination airports have performed badly on ex-Dublin routes compared to Ryanair or Aer Lingus. Both full service and low-cost carriers operating from destination airports have significantly lower passenger shares on routes from Dublin compared to Aer Lingus and Ryanair and there has been significant exit on ex-Dublin routes of both legacy and low-cost carriers operating from bases at destination airports.

586. In the case of Cork and Shannon, entry barriers linked to a base are of a smaller scale than for Dublin. In order to replicate the existing competitive constraint, it would suffice to establish a small base in these airports. However, the entry barriers mentioned above also arise in the case of routes operated out of these airports. In particular since the merged entity would be the only carrier with a base with large aircraft offering services to a large number of European destinations at these airports.

7.8.4. Entry costs and risks would be significant in a market already served by two strong airlines with well-established brands

587. All potential competitors would have to take into account that post-merger the Irish market is served by an airline which combines two well-established brands. Today, the Merging Parties account for 68% of the total passenger traffic (and 80% of the relevant intra- European short-haul passenger traffic) from and to Dublin, against which any new entrant would have to compete.

588. Ryanair submits that reputation, brand-awareness and the high market share held by the incumbent do not play an important role for the decision of potential entrants to open new routes from or to Ireland. Ryanair mainly argues that any airline can easily open a new route in any country today without incurring significant costs or risks, e.g. by using the internet as a distribution channel. Ryanair particularly refers to its own example which, according to Ryanair, proved that opening new routes in countries where Ryanair was not active before can be done without major difficulties within a reasonably short period of time (579).

589. According to the Horizontal Merger Guidelines, "(...) barriers to entry may also exist because of the established position of the incumbent firms on the market. In particular, it may be difficult to enter a particular industry because experience or reputation is necessary to compete effectively, both of which may be difficult to obtain as an entrant. Factors such as consumer loyalty to a particular brand, the closeness of relationships between suppliers and customers, the importance of promotion or advertising, or other advantages relating to reputation will be taken into account in this context (580)."

590. Further the Commission has also in the past in the air transport sector indicated that branding and promotion costs represent a barrier to entry. It was acknowledged that "any new entrant would face significant sunk costs in terms of marketing expenditures which will not be recovered upon exit" and that "customer information (or lack thereof) and promotional campaigns remain critical to the individual success of suppliers" (581). Further, the importance of the brand name of established operators developed over many years was recognised as a barrier to entry for new carriers (582).

591. Ryanair and Aer Lingus have such an "established position" in Ireland. They are by far the two most important players on the markets for flights from/to Ireland, with the most experience with Irish customers and Irish routes, and both enjoy by far the highest brandawareness of all airlines in Ireland as confirmed by a study prepared for Aer Lingus (583).

7.8.4.1. Advantages from Ryanair's and Aer Lingus' experience with the Irish market

592. As concerns experience and know-how on the specific market conditions in Ireland, Ryanair and Aer Lingus have been able to build up experience and knowledge on the particularities of the Irish market for decades (Aer Lingus) or more than 15 years (Ryanair) respectively. Ryanair itself stresses the importance of intimate knowledge of the Irish environment for potential competitors in its Notification (584). The Merging Parties' experience with the Irish market constitutes therefore an advantage vis-à-vis other potential entrants, however it is in itself certainly not an insurmountable entry barrier.

7.8.4.2. Advantages from Aer Lingus' and Ryanair's well-known brand

593. More importantly, many third parties and potential competitors (585) pointed at the fact that Aer Lingus and Ryanair enjoy a manifest advantage as concerns reputation and brandawareness, which makes it more difficult for potential competitors to compete for Irish customers. As one competitor put it, the issue is not whether it is "possible" to open a route, but to win sufficient customers for it (586).

594. Ryanair acknowledges in principle that a strong reputation of the incumbent competitors can be a barrier to enter a market (587), while it contests that this would be the case here.

595. The Commission's investigation has shown that it is much easier for Aer Lingus and Ryanair as the two best-known airlines in Ireland with an established marketing and distribution system to win new Irish customers for existing and new routes than for their competitors. In particular on routes where the majority of customers originate in Ireland, competitors face a significant disadvantage vis-à-vis Ryanair and Are Lingus. It should be noted that the overlap markets are to a large extent "outbound" markets, i.e. on most routes affected by the merger, a majority of customers buy their ticket in Ireland. On a number of routes (notably continental routes to holiday destinations), more than [75-85]*% of the traffic is outbound (588). There are no routes where a comparable share high of passengers is "inbound" to Ireland (589). 596. To run "outbound" routes successfully, airlines must be able to sell tickets to Irish customers. It is therefore crucial for potential competitors that want to enter such routes to have access to Irish customers. Non-Irish potential competitors on those routes cannot rely at all on their on customer base at the other end of the route, since these customers account only for a minority of the business.

597. Although on a number of other routes customers flying from Ireland seem to represent only 50% or slightly less of the customers (590), potential competitors even on these routes cannot economically offer flights without winning the 50% Irish customers for their routes. Further, as indicated in detail in Section 7.12. below, most of the routes where the shares of inbound and outbound customers are more balanced are routes between Ireland and the UK. In this regard it should be noted that Ryanair is not only very strong in Ireland but also in the UK (Ryanair has 5 base airports throughout the UK (591) with more than 50 based aircraft) and would thus still have an advantageous position. Most airlines currently not present in Ireland would have more difficulties in winning customers in Ireland than a merged Ryanair/Aer Lingus in their country after a possible merger. Although this disadvantage may play a more important role for smaller airlines from the continent than e.g. for larger UK carriers, the Commission notes that even Europe's second largest lowfrills carrier, easyJet, has indicated that it faces a disadvantage in competition with Ryanair on UK-Irish routes. According to easyJet "Ryanair had the advantage that it had a strong brand name in both, the UK and Ireland, while easyJet was only well-established in the UK (592)." Indeed, Ryanair is already well known in a large number of countries and has a well established brand throughout Europe (593). Ryanair is present today on more than 400 routes in 24 European countries. Ryanair offers therefore already today more destinations than e.g. the UK's largest carrier, BA, offers. The same is not true for most of the other potential competitors (594), who would need to establish a distribution system and, above all, to get sufficient brand-awareness to fill their flights with sufficient Irish customers, always in competition with the two well-known incumbents (595).

598. Ryanair argues that any company from any third country could have easily access to Irish customers, since tickets can be sold via the internet today. The Commission's investigation has, indeed, confirmed that both Ryanair and Aer Lingus, as well as many other low-frills carriers, sell the vast majority of their tickets via their web pages.

599. While the increased importance of the internet distribution channel has certainly reduced distribution costs, new competitors still face the problem that their web pages must be visited and used by Irish customers in order to win such customers. In fact, competitors have confirmed that brand-awareness and reputation still play an important role in the distribution of flight tickets and are key to winning new customers for a given route (596). It would take a significant investment and a lot of time to build up a web site which would be as recognised and regularly visited as Aer Lingus' and Ryanair's pages are currently in Ireland.

600. Indeed, today most Irish customers wanting to fly from Ireland to another destination almost "automatically" turn to the web pages of either Ryanair or Aer Lingus. Although it is true that there are a number of websites available which collect fares of different competitors, it appears that the majority of flights are still booked via the companies' own websites. Ryanair's internet pages ("Ryanair.com" and "Bookryanair.com") are, according to independent third party research, the most popular airline internet pages in Ireland (ranking no 44 and no 64 respectively of Irelands most visited websites. Aer Lingus' internet page ("Flyaerlingus.com") is also in the group of the 50 most visited web sites in Ireland (ranking number 33). No web page from any actual or potential competitor appears in the "top 100" list of Irish web pages (597). By contrast, Ryanair's webpage is among the most popular airline webpages not only in the UK, but also in almost all countries in Europe (598).

601. Also the Customer Survey at Dublin Airport confirmed that customers on all routes out of Ireland mainly consider either Ryanair or Aer Lingus and to a much lesser extent carriers based outside Ireland. On routes where they face also a third (or more) competitor, about 40% of Ryanair customers considered Aer Lingus, but only 17% considered the other competitor(s) on the route. In the case of Aer Lingus passengers, 33% considered Ryanair and 16% the other competitor(s) on the route (599).

602. As the previous "flag carrier" which used to have a quasi monopoly on the Irish market before the entry of Ryanair, Aer Lingus has traditionally a close relationship to their customers and a significant brand recognition in Ireland, where its service is perceived as particularly adapted to national culture (600). Aer Lingus has build up its brand in Ireland over decades and can profit from that brand recognition today.

603. Ryanair started its operations more than 15 years ago from Dublin and has developed into a strong brand in Ireland, in particular for price-sensitive customers. Ryanair therefore benefits not only from the advantage of being an Irish company, but also from its strong reputation as the largest (and most profitable) European "low-frills" carrier with a clear price-aggressive airline profile.

7.8.4.3. Entry would require high marketing costs

604. Potential competitors considering entering the Irish market would have to weigh the costs and the potential benefits of such entry. These costs can include not only costs for ground handling, customer care or offices at the respective airport where they do not yet have operations, but notably marketing and advertising costs. These may vary, e.g. dependent on whether the potential entrant is already active at the other end of the route and just expanding it operations ("connecting the dots") or not.

605. Any new entrant would have to consider incurring expenditure to market its service in the relevant catchment areas at both ends of the respective route. Potential entrants to routes from or to Ireland will therefore face significant sunk costs in terms of marketing expenditure, which cannot be recovered upon exit. The success of entry depends very much on the airlines ability to win sufficient customers for its service. Customer information (or lack thereof) and promotional campaigns are therefore critical to the individual success of airlines601. This holds particularly true for low-fills airlines, because the success of the low-frills business model depends to a large extent on a high load factor, i.e. the ability to fill the individual flights with as many passengers as possible.

606. Ryanair claims in its submissions to the Commission that winning customers for new routes is not very costly and can be done within a relatively short time period of three to four months or even weeks. In its reply to the Statement of Objections, it further quotes single competitors who have indicated in a general manner that there were "no entry barriers" (602). Ryanair stresses that customers in every country can easily be reached via the internet today. It would therefore in principle be sufficient to launch a press conference, to offer some promotions (free seats) to radios and newspapers and to spend a limited amount of money for advertising in newspapers at both ends of the route. Altogether, the costs would, according to Ryanair, not amount to more than around EUR [60,000-80,000]* for a new route (603). Ryanair refers to its own example with advertising costs of only EUR 0.40 per passenger, despite being in full expansion, which would be an "inconsequential issue" compared with many other cost airlines incurred in their operation (604).

607. The Commission's market investigation, however, showed that the costs for potential entrants to build the necessary brand-awareness in Ireland are perceived by many potential competitors named by Ryanair as significant and as an important entry barrier (605).

608. The required level of marketing expenditure is related to the brand recognition that each airline has developed over the past years in the targeted catchment area and to the brand recognition of the competitors operating in that area. As set out above, the Irish market is characterised by the existence of already two strong national airline brands. Having established a brand over many years and disposing of a joint customer base of 80% of Irish short- or mid-haul passengers on intra-European routes, the necessary efforts for Ryanair and Aer Lingus to raise attention for new flights and to attract potential passengers in Ireland via their web sites are much lower than for any new entrant.

609. Ryanair does not contest that it has a very strong brand throughout Europe and it rightly points out that Ryanair has been known as the "pioneer" of low-frills flights in Europe. Ryanair operates in 24 countries, offering flights to around 450 destinations. Apart from easyJet (active on about 250 airports), there is hardly any airline which could build on such a strong reputation on a European-wide scale, notably amongst the relevant "low frills" customers. It seems thus plausible that many competitors describe the marketing investments as much more significant than Ryanair does it from its own perspective.

610. The Commission notes that many relevant competitors perceive their own marketing costs necessary to establish a new route or even a base in Ireland as important. While the fact that Ryanair's own marketing costs are lower is not contested by the Commission, the relevant criterion for the Commission's analysis is whether from the perspective of a potential competitor the entry costs are so significant that they might contribute to deterring new entry. The fact that Ryanair has opened new routes elsewhere does not prove that other airlines could easily do the same on routes from or to Ireland in competition with Ryanair.

611. Marketing costs necessary for a new entrant to enter into competition with Ryanair in Ireland may differ significantly from Ryanair/Aer Lingus' marketing costs for a new route from or to Dublin. This is evident for marketing efforts to win customers in the Irish market, where Ryanair and Aer Lingus are already established and can have easily access to their large customer base, while the necessary marketing costs for any foreign competitor without such a customer access and an equally strong brand in Ireland would be significantly higher. While Ryanair and Aer Lingus would not incur any costs for getting enough web traffic on their web sites, any new competitor would have to invest in a marketing campaign for his website in order to attract customers.

612. Aer Lingus and Ryanair therefore have an advantage over potential competitors in terms of marketing due to the fact that marketing costs for entry on new routes would in most cases are "asymmetric". As set out above, on most overlap routes the majority of customers flies from Ireland to other countries and not the other way round. On these "outbound" routes (examples being continental leisure routes), the majority of the necessary marketing expenditure would relate to Ireland and not the destination country, which makes it much easier for Ryanair and Aer Lingus to fill these routes.

613. Even on those routes which are not predominantly used by Irish-based customers, but service more or less equally Irish and foreign passengers, competitors will normally face significantly higher marketing costs than Ryanair for several reasons.

614. First, Ryanair's figures are calculated on the basis of an average of more than 40 million Ryanair passengers. Ryanair has referred to costs of € 0.40 per passenger for marketing. For smaller airlines with passenger numbers of 1 million or less passengers, the relative investment costs to open a new route to Ireland or even a base at Dublin would therefore be higher.

615. Second, as set out above, Ryanair has easier access to customers throughout Europe than most competing airlines, in particular smaller local carriers, since Ryanair is Europe's oldest and largest low-frills carrier, established and well-known in almost every European country. As a result, the Ryanair brand can benefit today from a high brand-awareness throughout Europe, and Ryanair also enjoys a strong European-wide reputation for offering flights at very low prices.

616. Finally, Ryanair's business model differs significantly from its potential competitors as regards marketing expenditure. Unlike its competitors, Ryanair flies mainly to secondary airports. These secondary airports have an interest to develop the airport and are therefore willing to support Ryanair (in most cases their most important customer) in terms of marketing, be it by own advertising, brochures or direct financial contributions to Ryanair's marketing expenses. Due to the contribution of secondary airports to Ryanair's marketing costs, these costs are naturally lower than those of its potential competitors on the overlap routes. Only very few competitors, however, would consider flying to secondary airports at all.

617. By way of example, the Commission has found that when establishing a new base in Charleroi, Ryanair itself had calculated marketing expenses not of 40 cent, but of EUR 8 per passenger. Under an agreement concluded in 2001 between Ryanair and Brussels South Charleroi Airport (BSCA), Ryanair and BSCA have formed a joint advertising and publicity company which contributes to Ryanair's publicity and marketing in relation to its activities at Charleroi. For example, the company financed parts of the price of Ryanair's promotional tickets. BSCA and Ryanair contributed in the same proportions to the socalled "Promocy" operation, which includes a marketing contribution (for each) of EUR 4 per passenger (606).

618. Ryanair rightly points at the fact that also the Dublin Airport Authority tries to attract new airlines by way of an incentive scheme, which, inter alia, provides for moderately reduced airport charges for new entrants. However, none of the competitors who answered to the Commission's market investigation regards these incentives as sufficient to "neutralise" the entry costs which are necessary for a new entry with a new route or even a base in Dublin. It should also be noted that the financial incentives Ryanair receives from secondary airports (which include active marketing participation of up to EUR 4 per passenger, see the "Charleroi" case) are much greater than the incentives the Dublin Airport Authority provides. In full knowledge of these incentives, many potential competitors still describe the necessary financial investment for an entry to Ireland as an important entry barrier and as "a major investment (607)".

619. Potential entrants have confirmed that entering against the merged entity would require a significant marketing investment.. By way of example, one competitor indicated that its costs for establishing operations from Manchester amounted to around EUR 4 million, mainly spent on advertising. Establishing a brand in a new country, and in particular entering to an airport like Dublin, where, unlike in Manchester, already two strong national low-cost airlines account for 80% of all short-haul passengers, is likely to require an even larger investment than entry into Manchester (608).

7.8.4.4. Entry entails a high risk of sunk costs for competitors

620. The likelihood of entry on a given route depends on the profitability the potential competitor would expect on this route. One important element for the decision of a potential entrant whether to enter or not a market is the level of sunk costs which he would have to write off in case of a failed entry. In this context, the Horizontal Merger Guidelines provide that: "(...) a high risk and costs of failed entry may make entry less likely. The costs of failed entry will be higher, the higher is the level of sunk cost associated with entry (609)."

621. The significant marketing expenditure which are necessary for any new entrant to establish its brand in Ireland and have access to new customers (and which represent the largest part of expenditure for opening a new route (610)) can be considered as sunk costs should the entry fail (611). Competitors have also explained to the Commission that opening of a new route from and to Dublin would require additional investments in new or at least newly leased aircraft or for deploying an aircraft from another route (with the related opportunity costs of losing that plane for services on another route). Additional costs are necessary for installing and operating the necessary ground services (boarding personnel, customer care personnel (612), technical staff, office buildings, but also costs for luggage handling, security, and other airport related costs (613) etc.) in the new airport and often in a new country where no such facilities existed before. These costs, in particular for those competitors who were not active in Ireland before the entry at all, do indeed further increase the necessary investment to open a new route and, at the same time, the risk of incurring additional sunk costs (614).

622. A number of competitors mentioned that a further disincentive to enter routes from and to Ireland in competition with Ryanair and Aer Lingus were the preferential airport charges and treatment, in particular for Ryanair on many airports. With the exception of Dublin Airport, Ryanair was able to negotiate preferential airport charges with many airports in Europe, including for Shannon, which gives them an important financial advantage over potential competitors who would not be able to take advantage from the same discounts. This is particular the case for carriers based at main airports that are less likely to give significant discounts due to higher demand for flying from/to their airports than the secondary airports used by Ryanair. This aggravates further the existing cost disadvantage for competitors.

7.8.4.5. Conclusion on entry costs

623. As set out above, entering in competition with Ryanair/Aer Lingus in Ireland would require a significant investment in marketing, brand-building, and other start-up costs for any potential competitor who would wish to compete head-to-head with the merged entity. Since these costs cannot be recouped should the attempt to enter the market fail, any potential competitor takes a considerable financial risk. This is likely to deter entry.

7.8.5. The risk of aggressive retaliation by Ryanair/Aer Lingus is high

624. Another important element for the assessment of the likelihood of potential entry is the ability of the merged entity to react aggressively to the entry in order to prevent the potential competitor from successfully entering the market.

625. The Commission's investigation showed that Ryanair has a reputation of engaging in aggressive competition in case of new entry to Ireland, notably by temporarily lowering prices and expanding its capacity in order to drive out the new entrant on routes to or from Ireland (615).

626. The likelihood of aggressive retaliation is relevant because it has the factual consequence that potential entrants are likely to be deterred from entry and that it is less likely that a dominant firm will be constraint by the threat of new entry. This is also reflected in the text of the Horizontal Merger Guidelines, which provide in paragraph 69:

"For entry to be likely, it must be sufficiently profitable taking into account the price effects of injecting additional output into the market and the potential responses of the incumbents."

627. Ryanair claims that the risk of aggressive retaliation by Ryanair/Aer Lingus is not relevant for potential competitors in the present case, since Ryanair would in any event charge the same low prices regardless of whether a competitor is active on the same route or not ("no higher prices on monopoly routes") and since many examples of successful entry showed that competitors are not deterred by the presence of Ryanair on a market (616).

628. However, as shown in detail in Section 7.4.1 above, Ryanair takes into account the pricing policy of its competitors, whether by adjusting its load factor when demand should decrease as a reaction of new entry or a price decrease or by direct comparison of Ryanair's prices with the prices of its competitors. Ryanair has expressly confirmed to the Commission that it has lowered prices in reaction to entry in previous case (617). The examples in Sections 7.8.5.1-4 below further illustrate this behaviour.

629. The Commission's market investigation has shown (618) that Ryanair in the past successfully reacted to entry by competitors on Irish routes by systematically undercutting the competitors' prices and adding new capacity on the market which was, depending on the case, afterwards reduced when the competitor had left the route (619).

630. There are numerous examples where Ryanair has attacked the entrant by increasing frequencies and capacity. Ryanair has used its base-advantage by rapidly expanding capacity on routes in order to force new entrants out. Below are set out examples for routes where Ryanair has been the incumbent and a new carrier has entered, during the period 2000-2005. On each analysed route the result has been the same - the new entrant has left the route within a period of, on average, one and half years and there have been no new carriers entering the route thereafter.

631. Furthermore, on several of the routes detailed below, after the exit of the new entrant, Ryanair reduced its frequencies on the route to fewer frequencies than it had when the new carrier entered the route. This suggests that the frequency increase was rather a strategic one than a reaction to changes in demand. The Commission has also found evidence that indicates that following the exit of the competitor, Ryanair increased prices again by reducing capacity.

632. It should be noted that in contacts with actual or potential competitors, none of these carriers mentioned an example of successful entry in the past; if the history of entry to Ireland was mentioned, it was to express that potential entrants were further discouraged by past experience (620). The analysis of past entry cases confirms the existence of high entry barriers (621).

633. Although Ryanair argues that there is "extensive evidence" of entry by competing carriers against Ryanair (622), the Commission has found that these examples do not show that the barriers to entry on the identified overlap routes against Ryanair/Aer Lingus are low. First, it should be noted that only 11 of these examples of entry actually were on routes to/from Ireland (623). In two of these cases the entrant was Aer Lingus itself (624). In the remaining 9 cases, Ryanair itself indicates that 6 entrants exited the relevant routes the year following their entry (625). In fact this is also the case for one additional entrant (TAP Portugal on Dublin - Faro), which exited in February 2007, after around 5 months of operation. Therefore, only in the two remaining cases the new entrants remained longer on the market: the first one is CityJet on the Dublin - London route. CityJet however focuses on business passengers by serving the specific London City airport with smaller jet aircraft and cannot be considered as a significant competitive constraint on Ryanair - see further below in Section 7.9. Its market share on the route is around [0-10]*%. The second example is Spanair on the Dublin - Malaga route. On that route, Spanair operates only during the summer season with around 2 weekly rotations on weekends. This service therefore only exerts a negligible constraint on the Merging Parties' daily services which are operated throughout the year on this route. Spanair's market share on the route is around [0-10]*%. Therefore, these examples of entry do not provide evidence of sustained entry against Ryanair on the routes to/from Ireland that would exercise significant competitive constraint

634. In its response to the Statement of Objections, Ryanair does not contest the conclusion that entrants have not been successful on routes where Ryanair is present. However, it argues that this is due to its lower cost base (626) and that the Commission should not penalise Ryanair simply because it is more efficient and has lower fares than other entrants. However, in the context of the analysis of objective entry barriers, the absence of successful entry and Ryanair's reputation of aggressively driving entrants out of the market is relevant for the Commission's assessment. Being in a position of doing so due to its lower cost base and its financial strength discourages new entrants from entering routes on which Ryanair operates (627).

7.8.5.1.The example of easyJet's entry attempt

635. The example most cited for unsuccessful entry against Ryanair is probably the failed entry of European's second largest low-frill carrier, easyJet, to Ireland in 2005 (628). EasyJet, having discovered a "niche" on the UK-Ireland market (629), started to offer flights from Gatwick to Cork, Knock and Shannon in January 2005, which triggered, according to easyJet, an "immediate reaction" of Ryanair. Following Ryanair's response, easyJet left these routes in 2006 and no longer flies to Ireland. Ireland is the only Member State that easyJet has abandoned after an attempt to enter. Ryanair's reaction in this case is documented both by public and internal statements. The internal documents for the Board Meeting of Ryanair for example state [...]* (630).

636. The Irish Department of Transport has submitted a detailed analysis of the examples of previous entry of new competitors on Irish routes and on Ryanair's reaction (631). The Commission has analysed this detailed description to the extent possible in the limited time available during the investigation of the case and carried out its own analysis of Ryanair's reaction on the basis of fares & frequencies data submitted by Ryanair.

London Gatwick-Cork-Gatwick (LGW- ORK)

637. Before easyJet's entry on Gatwick-Cork, in Summer 2004 Ryanair operated on the Cork- Stansted route with 21 and Aer Lingus on Cork-Heathrow with 28 weekly frequencies. EasyJet announced on 23 September 2004 its planned entry on the Gatwick-Cork route and started flying on this route on 28 January 2005. In Summer 2005 easyJet operated 14 weekly frequencies on this route which it increased to 21 in Winter 2005/6 (632). While maintaining 21 frequencies in Cork-Stansted, in Winter 2005/6 Ryanair in addition started operating also on Gatwick-Cork with 14 weekly frequencies. Thus, only after easyJet's entry into Gatwick Ryanair opened a second route from Cork to London-Gatwick. This service was therefore in direct competition to easyJet. EasyJet announced that it would leave this route in July 2006 and stopped operations on 1st October 2006. Subsequently Ryanair reduced its frequencies from Cork to Gatwick from two to one per day.

638. In its response to the Statement of Objections, Ryanair has pointed out that it had announced its capacity reduction already in July 2006, i.e. 3 months before easyJet had announced its withdrawal. The frequency reduction was therefore, according to Ryanair, not a reaction to easyJet's exit (633). Ryanair has, however, not provided further evidence in support of this assertion. The Commission notes that Ryanair's claim contrasts with its own statements that Ryanair did react to easyJet's entry (634). Furthermore, according to easyJet, easyJet announced its exit from the routes Gatwick to Cork, Shannon and Knock already in July 2006 (635), the same month, in which Ryanair reduced its capacity. Ryanair's claim has therefore no been confirmed.

Graph 3: Entry effects Cork - London

639. The graph above illustrates the fare and frequency development following easyJet's entry on the Gatwick-Cork route (636). It uses data on Ryanair monthly frequencies and average fares. The horizontal axis depicts time on a monthly basis from 2001 to 2007. On the vertical axis there is the number of monthly frequencies (on the left) and the average fare (on the right). STN refers to Stansted and LGW to Gatwick. Finally, the vertical bars (in red) indicate the moment of entry and exit by easyJet on this route.

640. First, as on many other routes, one observes a seasonal effect. Due to demand fluctuations, frequencies and average prices decrease in the Winter season while they increase in the Summer season. Second, one observes that in the Summer period 2004, i.e. before easyJet's entry, this pattern changed. Frequencies did not increase as previously (this also holds for Aer Lingus). As a result average prices were higher than in Summer 2003. This corresponds to easyJet's remark that they had observed high prices on the route before entry (637).

641. At the time of easyJet's entry, Ryanair's frequencies in Stansted went up. However, that may be considered a reaction to the seasonal effect. In Summer 2005, also Aer Lingus' number of frequencies went back to the level of Summer 2003. Nevertheless, given that due to easyJet's entry the total number of frequencies had increased, while maintaining a high load factor, it would be more profitable for Ryanair to reduce frequencies (or increase them by less than in previous years). As Ryanair did not adapt its frequencies, due to the overall increase in capacity in the market, its prices dropped significantly.

642. A significant change then happened in Winter 2005/6 when its total number of frequencies on the Cork-London market increased significantly due to the opening of the Gatwick route. That increase happened at a point in time when - according to the seasonal effect - one would have expected a capacity reduction. Moreover, there was a shift of frequencies from Stansted to Gatwick. Thus, Ryanair went closer to the competitor from a city pair overlap to an airport overlap. At that point Ryanair charged an average price of EUR [10- 20]* (according to its own data) in Gatwick (638). Finally, subsequent to easyJet's exit, one observes a frequency reduction by Ryanair on the Gatwick route.

643. Ryanair has pointed out that it is normal that Ryanair's monthly fare fell on the Cork- Stansted route following the arrival of easyJet, as it is its policy to always sell the lowest fares in every market (639). Admittedly, a price reduction in the face of entry would be normal business behaviour. However, apart from the price reaction, one also has to look at capacity. Given that entry leads to an increase in overall capacity on the route, normally the incumbent would reduce its own capacity (640). However, it is the reaction in capacity which brings the development on the Cork-London market in line with the above description of an entry deterrence strategy. Not only did Ryanair substantially increase capacity in the market, it also moved the capacity directly into the airport of its competitor. As a result, average prices went down more than they would have done in case of a capacity adjustment. Finally, subsequent to the competitor's exit, prices increased.

644. In its response to the Statement of Objections, Ryanair points out that in spite of its frequency reduction after easyJet's exit, Ryanair's total capacity between Cork and London post- exit was [25-35]*% higher and its average fare [15-25]*% lower than for the corresponding months prior to easyJet's entry (641). This observation is in line with the data in Graph 3. However, Graph 3 also shows that prior to easyJet's entry, Ryanair had reduced capacity and increased prices significantly (642), which had originally encouraged easyJet's entry. Thus, if post-exit capacity was higher and prices lower than during the pre-entry period, this change most likely resulted from easyJet's entry. In any case, it remains the fact that Ryanair has reacted aggressively against an entrant and caused the entrant to exit the route, which is relevant for the Commission's assessment of the likelihood of successful entry.

Shannon-London Stansted

645. On the Shannon-London route, prior to easyJet's entry, Ryanair operated to Stansted and Aer Lingus to Heathrow. Over the period 2003-2006, Aer Lingus' weekly frequencies have remained stable at 27 in the Summer and 21 in the Winter season. In Summer 2003 Ryanair operated 29 frequencies which were cut to only 14 in Summer 2004. Ryanair's average price went up from € [40-50]* to € [50-60]* (643). In its response to the Statement of Objections, Ryanair points out that this capacity reduction was due to a cost dispute with Shannon airport.

646. EasyJet announced that it would start operating from Gatwick to Shannon on 23 September 2004 (644). On 30 November 2005, Ryanair announced that it had reached agreement with Shannon to open a substantial base at Shannon Airport (645). EasyJet started flying the Gatwick-Shannon route in January 2005 with 1 daily frequency (646). In May 2005, Ryanair also entered the Gatwick-Shannon route with 7 weekly frequencies while increasing the frequencies to 21 frequencies in Stansted. In comparison to Summer 2004, its average price went down from € [50-60]* to € [10-20]* in Summer 2005 and € [10-20]* in Winter 2005/6 (647). As can be seen from the graph, during the period when easyJet operated on this market, Ryanair also started flights to London Luton. Thus the overall number of Ryanair's frequencies increased significantly. In July 2006 easyJet announced that it would leave the route and stopped operations on 1 October 2006. In October 2006 Ryanair also stopped flying into Luton. Thus, following easyJet's exit, Ryanair reduced significantly its overall frequencies from Shannon to London. As in the case of Cork-Stansted, the increase of frequencies following entry and the reduction of frequencies after exit indicate strategic behaviour on Ryanair's side (648). In its Response to the Statement of Objections, Ryanair notes that compared to the situation before easyJet's entry, Ryanair's total number of frequencies on that market was still higher after the competitor exited (649). This, however, does not imply that its previous behaviour was not aimed at deterring easyJet's entry. It may also be the result of a general increase in demand on that market over that period of time.

Graph 4: Entry effects Shannon - London

Gatwick-Knock (LGW-NOC)

647. EasyJet announced that it would start flying on the Gatwick-Knock route on 23 September 2004 and began operations on 28 January 2005. easyJet operated one daily frequency on that route until the end of Summer 2006. Ryanair announced, according to the Irish Department of Transport (hereinafter "DOT"), in November 2004 that it would enter the route with 1 daily frequency in January 2005, in addition to its existing daily Stansted- Knock operation. Ryanair's service started on 19 January 2005, 10 days before easyJet's operations on the route commenced.

648. Following Ryanair's response, in July 2006 easyJet announced that it would withdraw from this route and it stopped operations on 1 October 2005 (650). In July 2006 also Ryanair announced that it would withdraw its flights on the Gatwick-Knock route. It switched its capacity to Knock-Luton.

649. Ryanair points out that on this route post-exit its frequency did not fall while fares were around [35-45]*% lower compared to the pre-entry period. However, as pointed out for the other routes above, this does not put into question that Ryanair's behaviour was a strategic move to force its competitor to leave the route (651). As regards prices, one has to take into account the seasonality effect, as explained with regard to Graph 3 above, since the postexit period falls in the Winter term where on these routes demand is lower, and as a result prices tend to be lower as well.

7.8.5.2.The example of MyTravelLite's entry attempt

Dublin - Birmingham (DUB-BHX)

650. MyTravelLite started operating on the Dublin-Birmingham route on 30 September 2003 with 2 daily frequencies (652). Prior to this new entry, Ryanair operated, according to the DOT, approximately 40-45 weekly frequencies. MyTravelLite announced that it would be operating fares starting at €1.99 (excluding taxes). According to the DOT, Ryanair reduced its fare to down to €0.99 and increased its daily frequencies from three to five daily flights corresponding to 70 monthly frequencies. This strong capacity increase subsequent entry, which was accompanied by the sharp drop in average prices, again confirms a strategic behaviour by Ryanair.

651. However, Ryanair points out in its response to the Statement of Objections (p.45) further developments were then determined by a dispute between Ryanair and Birmingham airport. Following a strong price increase by the latter, in April 2004 Ryanair switched its capacity from Birmingham to East Midland's airport. The reduction in capacity in Birmingham was therefore not due to Ryanair abandoning its prior strategy with regard to MyTravelLite but driven by external events (the dispute with the airport). However, given the proximity between Birmingham and East Midland airports, the competitive constraint exerted on MyTravelLite remained. On 19 July 2004 MyTravelLite announced its exit from the route as of 31 October 2004.

652. Since MyTravelLite's exit, there have been no further attempts of entry on this route (653).

Graph 5: Entry effects Dublin - Birmingham

7.8.5.3.The example of GOFly's entry attempt

Dublin-Glasgow (DUB-GLA, DUB-PIK)

653. Ryanair entered the Dublin-Glasgow route in May 1994 to compete with Aer Lingus. According to easyJet, which took over GO Fly on 1st August 2002 (654), Go Fly started offering flights on this route on 19 September 2001, operating 38 flights per week. GoFly's starting fare on the route was £35 return (including taxes and charges) with maximum fare set at £140 return (including taxes and charges). As can be seen on the graph below, prior to entry, Ryanair's prices had gone up sharply. Subsequent to entry, Ryanair then increased the number of flights on this route (655) which led to a sharp decrease in its average price. At that time, the CEO of Ryanair announced its intention to cut prices to £1.50 per seat and extend the offer into the winter period if GoFly lowered its prices further (656).

654. On 21 March 2002 GoFly stopped its services between Dublin and Glasgow. The DOT quotes from a speech of the former Chief Executive of GoFly Cassani, who pointed out that "the result is that no one else will come onto the Dublin route, and Irish consumers will have no one but Ryanair for the rest of their lives [...]. Why go up against the biggest and nastiest elephant in the china shop? [...] If anyone starts an airline out of Dublin, all I would say to them is: make sure you have a lot of money. It would be commercial suicide [to go against Ryanair] (657)"

655. Indeed, Ryanair's increase in capacity, resulting in a price decrease, subsequent to GoFly's entry in this example appears, in the absence f any other convincing explanation, to follow the same pattern of a "retaliation" strategy as in the examples above.

Graph 6: Entry effects Dublin - Glasgow

Dublin-Edinburgh (DUB-EDI)

656. For several years Ryanair had been (unsuccessfully) seeking discounts from BAA Edinburgh Airport (658). GoFly announced a new route from Edinburgh to Dublin on 4 July 2001 and started operations on 19 September 2001 on 4 July 2001. Following an agreement with Edinburgh Airport eight days later on 12 July 2001, Ryanair also announced to operate on the same route with four daily flights from 30 August 2001 (659). Go Fly's offered fares on this route (including taxes) ranging from £17,50 to £80. For the first 29 days of September 2001, Ryanair's return fare on the route was £29.99 for 70% of flights and a maximum of £99.99 for the remaining flights on this route (including taxes) (660). According to the DOT (661), Go Fly then lowered its prices to £10 (including tax) for a single flight and Ryanair, for a certain period of time in 2001, responded by offering single flights for £5 (including tax) on the Dublin-Edinburgh route.

657. Go Fly stopped the services between Dublin and Edinburgh on 21 March 2002. Since Go Fly's exit, there have been no new entrants on this route. Ryanair maintained its 4 daily frequencies through 2002 and 2003.Ryanair reduced its capacity on that market in April 2004. According to Ryanair, this was in reaction to a price increase by Edinburgh Airport. The Commission, however, notes that the ongoing dispute on prices with Edinburgh Airport did not prevent Ryanair to respond to GoFly's entry on the Dublin - Edinburgh route, which led to a failed entry attempt on a route to/from Ireland. It is therefore unlikely that the issue of the airport charges were the only reasons for Ryanair's withdrawal after the competitor had exited.

7.8.5.4. Conclusion

658. The analysis of previous examples of entry supports the view expressed by competitors that Ryanair systematically reduces prices and increases frequencies when competitors enter the Irish market.

659. While price reductions are an expected reaction to entry, a capacity increase, if not justified by other events, can only be explained by the aim to oust the competitor. Several examples mentioned above show that Ryanair has applied such strategic behaviour against entry in the past. This has been accompanied by public statements of Ryanair according to which Ryanair would deter competitors. Building up a reputation of deterring entry creates a factual barrier to enter for new competitors as set out in the Horizontal Merger Guidelines.

660. It would be significantly easier for an even stronger Ryanair/Aer Lingus entity to react aggressively on entry after the proposed merger, since the merged entity would be the dominant operator on literally all routes from and to Ireland. Already today, Ryanair and Aer Lingus offer a higher number of frequencies than their competitors on all of their 35 overlap routes except for one (662). The fact that the merged entity would operate only large 190-seat aircraft and have, a large number of seats "to play with (663)" (i.e. to offer very low fares), and that the merged entity will have bases not only in Ireland, but at more than 15 other destinations would give the merged entity a high degree of flexibility to react swiftly to any attempt of entry on the Irish market.

7.8.6. Competitors consider other markets more attractive than the small Irish market

661. As set out above, the likelihood of entry is closely linked to the potential competitor's expectations with respect to the profitability of the route which he might open (664).

662. Any decision for an airline to enter a specific route means that an aircraft needs to be redeployed from another route or that an additional aircraft is used which could otherwise be used for another route. Hence, any decision to open a route from or to Ireland entails opportunity costs as discussed above (665) which the potential entrants must weigh against the expected benefits of an entry to a route from or to Ireland.

663. The Commission's market investigation has shown that the Irish market is not considered by many competitors as a particularly attractive market and that many potential competitors the Commission interviewed indicated that they would rather seek to open new routes to other destinations than routes to Ireland (666).

664. First, entering this market would require significantly higher marketing costs than entry on other markets where there are no two strong low-cost carriers with a strong base (667).

665. Second, the Irish market is regarded as a relatively small market, with only four million inhabitants, which is not even half of the population of the greater London area (668). The size of the market and the turnover to be achieved on this island with routes from and to Dublin and the other two potential airports is therefore limited.

666. Thirdly, when considering different potential markets for expansion plans, the small Irish market is in general not expected to be the fastest growing and most profitable market in the future by many competitors at least not on the main routes between Ireland and the UK. Many competitors have expressed a preference to start operations in regions where they are able to reach more potential customers with their flights and which are growing faster than the Irish market (e.g. Eastern Europe or Scandinavia (669)).

667. Fourth, the specific geographic situation of Ireland does not favour new entry but rather discourages it. Indeed, Ireland is not located centrally in Europe, but at the far North-West border of the continent, therefore not ideally placed for many central European airlines. More importantly, while bases in larger and more centrally located agglomerations allow connecting more traffic to other shorter and longer routes in any direction from this base, Dublin is not a "pass-by" destination (670) but more remote and allows only for short flights to the UK (671). Further, many important routes out of Ireland such as the London routes, can be considered mature.

668. Fifth, when it comes to serving Dublin, it should be noted that the incentives of a network carrier operating a route between Dublin and its hub are different from those of a point-topoint carrier. The network carriers use these operations mainly to feed their hub and the constraints that they exercise on point-to-point carriers may as a result be more limited. The Commission's market investigation has also shown (672) that a number of such carriers do not see expansion or entry into individual routes to/from Ireland as part of other commercial strategy.

669. Finally, in particular for low-frills operators Ireland has the disadvantage that is has no "secondary" airport in the Dublin area but that all traffic to the most important agglomeration of the country must go through the main "primary" airport in Dublin. As recognised by Ryanair, "there is no secondary airport to Dublin. Dublin is the gateway to Ireland..." (673). This does not only result in relatively higher airport charges. It leads also to the effect that any potential new entrant would have to compete directly with Ryanair and Aer Lingus at their home airport where both companies have based an important number of large aircraft. Airport congestion constitutes an important additional barrier to entry.

7.8.7. Airport congestion is an additional barrier to entry

670. Another barrier for potential entrants on routes from or to Dublin identified by the Commission concerns the airport capacity available at both ends of the routes. In order to start operations on a given route, potential entrants must be able to obtain sufficient terminal capacity (departure desks, queuing areas etc.), stand capacity (ideally with socalled "contact stands" which are directly connected to the terminals (674)) and, most importantly in practice, runway capacity. Airlines can request so-called "slots (675)", from the airport which allow the airline to use the runway for a specific "movement", i.e. a departure or arrival of their aircraft (landing and take-off slots). On many airports, the demand for such slots exceeds their availability, since the number of possible "movements" (arrivals or departures) at an airport is limited, notably because of the limited capacity of the runways and the requirement of security margins between the individual movements. The extent of slot scarcity depends on the congestion level of an airport. The attribution of slots is regulated by Regulation (EEC) No 95/93 of 18 January 1993 on common rules for the allocation of slots at Community airports (676) (the "Slot Regulation"). According to the Slot Regulation, different levels of congestion can be distinguished which require different rules for slot allocation. At airports which suffer from limited actual or potential congestion at some periods of the day, slots are distributed according to the rules for "schedules facilitated airports", which provide for a system of voluntary cooperation between air carriers in which a "schedules facilitator" is appointed in order to facilitate the operations of air carriers, i.e. to allocate slots. For airports suffering from a high level of congestion, the Slot Regulation provides the status of a so-called "coordinated airport". According to the rules of the Slot Regulation, slots are then not distributed on a voluntary basis with the help of a scheduled facilitator, but by a slot co-ordinator.

671. The Commission has found in previous cases that the limited availability of slots can be an important barrier to enter for potential competitors (677). Without access to a sufficient number of slots at commercially attractive times throughout the day, a new entrant is often unable to operate a regular service on a sufficient scale, and as a result, would not be able to provide an effective competitive constraint on the merged entity. Landing and take off slots are therefore critical inputs for any entrant wishing to operate a new service or expanding an existing service to or from Ireland. The Commission has analysed whether and to what extent airport congestion plays a role as an entry barrier in the present case.

7.8.7.1. Capacity constraints at Dublin Airport

672. The Commission's market investigation has shown that with respect to Dublin Airport, capacity issues were raised by potential competitors (678) and other third parties (679) mainly with regard to the availability of runway capacity, i.e. the availability of sufficient slots, of (contact-)stands and with regard to terminal throughput capacity.

673. Ryanair concedes that there may be some congestion problems for short periods in the morning, but takes the view that Dublin Airport can in general not be regarded as congested. In its view, at Dublin Airport there is still sufficient room for additional frequencies and hence for entry. As concerns runway capacity, Ryanair argues that congestion problems would only occur for a very short period of time (15 minutes (680)) in the morning. The existing problems were mainly the result of bad management of the airport. Not only do the relevant regulators, notably the Dublin Airport Authority (DAA) use too conservative standards (681); the existing runways at Dublin Airport could also be used more effectively, notably by using an existing smaller runway (682) for general aviation and for all smaller aircraft (683).

674. As concerns congestion at the level of the terminal, Ryanair stressed that the congestion problems in the arrival area which led to security problems in particular last summer, cannot be expected to reoccur, because customers increasingly use the electronic check-in, and the 24 new desks in the newly created "Area 14" would absorb a large number of customers.

675. Even if Dublin Airport were congested, Ryanair submits that this would not constitute a relevant entry barrier, because most of the overlap routes concern holiday/leisure routes, and customers on these routes had in general no preference for a specific departure time. Other, more time-sensitive passengers could book the two legs of their journey separately with different airlines, should one airline not be able to offer the preferred departure/arrival times.

676. The Commission's market investigation has, however, confirmed that potential entrants are deterred by congestion problems at Dublin Airport which are not likely to disappear in the near future.

The decisions by the CAR of February 2007

677. It should be noted that the competent Regulator for the Irish Airport, the Irish Commission for Aviation Regulation (CAR) has commissioned a detailed study into the current and future capacity at Dublin Airport (the "Jacobs Report") which was carried out by an independent consultant firm. The purpose of the Jacobs Report was to inform the Commission for Aviation Regulation on the appropriate scheduling status at Dublin Airport, in accordance with Article 3(3) of the Slot Regulation. Taking into account the results of the study and the comments of the main affected third parties, the CAR has, by way of two decisions of 12 and 13 February 2007 (684), designated Dublin Airport as a "coordinated" airport for the summer and winter 2007 scheduling seasons in accordance with Article 3 of the Slot Regulation (685).

678. In the decision, the CAR explicitly stated the following:

"The Commission has accepted the conclusion of the Report that peak demand at Dublin Airport in Summer [and Winter] 2007 will be greater than the airport's capacity, that consequently the airport will suffer "significant delays" and that such problems cannot be resolved in the short term (686)."

679. As concerns Ryanair's doubts as to the appropriateness of this decision and of the way the Dublin Airport is organised, it should be stressed that it is not for the Commission to take a decision on how to organise best the traffic at Dublin Airport. The task of the Commission in the context of this merger procedure is to assess whether the conditions as found by competitors are likely to lead to new entry.

680. The Commission notes that the CAR's view that the Irish Airport is indeed congested is shared by the current schedule facilitator ACL as well as by the DAA (687) and the Irish Government (688), i.e. the competent authorities for the management of Dublin Airport. The above cited decisions, which are based on detailed studies and the views of all main stakeholders at Dublin Airport, contain therefore strong evidence for the existence of congestion problems in Dublin.

681. The fact that a previous decision of the CAR is still subject to court proceedings has no direct impact on the present decision, which is legally valid and reflects the view of the public regulator.

682. The CAR decisions are in line with the findings of the Commission's investigation with regard to airport congestion at Dublin Airport.

Runway congestion

683. As concerns the problem of runway congestion, the CAR has analysed in detail alternative suggestions of to increase the existing capacity (e.g. reducing the aircraft separation rules or the construction of a new taxiway). These proposals were found to be inappropriate to solve the congestions problems for the different airport users, partly for safety reasons (e.g. changing separation rules), partly because they would not provide a timely and effective solution (new taxiways (689)). No third other party than Ryanair has invoked that a changed use of the small runway 11/29 (putting all small aircraft traffic on this runway) would be an effective means against the congestion problem. This is in line with the Commission's findings. Indeed the DAA has informed the Commission that the small runway was already regularly used by smaller aircraft, which was confirmed also by affected competitors (690).

684. The CAR found that congestion problems were particularly to be expected for the hours between 05:00h and 08:00, without having to decide on specific congestion times. The Jacobs Report, one of the main sources of the CAR's decision on runway capacity (691), come to the conclusion that delays are to be expected also between 21:00 and 22:00 (692). The CAR decision concludes that it can be expected that during the (morning) peak hours "out of every 10 flights during this peak period, one will experience delays upwards of 32-45 minutes (...)".

685. These findings of the competent regulator are in itself evidence to show that Dublin Airport can be considered as congested during peak hours. However, also the results of the replies from airports and competitors to the Commission's market investigation show clearly that competitors are well aware of congestion problems at Dublin airport and regard that congestion problem as an entry barrier (693).

686. Although it may be expected that a new runway in Dublin, which is planned to open in 2011 or 2012, will increase the limited runway capacity situation at peak times in Dublin, this will not be sufficient to lead to timely entry of potential competitors which would be necessary to remedy the identified competition problems (even presuming that these constraints were the only barrier to entry).

687. The maximum number of departing runway movements (take offs) at Dublin Airport is currently 25 per hour during the morning peak period. This has increased to a maximum of 31 take offs per hour from the start of the Summer 2007 IATA scheduling season. At certain peak times however, notably during the morning, ACL indicates that there remains runway congestion. Moreover, comparing it with its equivalent at the London Gatwick and Stansted airports, ACL indicates that a maximum of 31 take off movements is a relatively high number. Therefore, no other than marginal increases in the maximum number of runway movements can be expected.

Insufficient stand capacity

688. As concerns the problem of limited availability of stands, the CAR concludes (citing the Jacobs Report): "the number of stands available in Winter 2007 will not meet the peak demand requirements". In addition, the decision states that "between 40 and 50% of stands are likely to be remote". The Jacobs Report refers also to another study on stand availability which concluded that there would be a shortfall in stand availability between 2007 and 2010 (694).

689. This is also in line with the Commission's findings, since in particular smaller competitors have mentioned that they would also face significant difficulties to obtain appropriate (contact-)stand which is linked to the airport building and that not being connected to the terminal is a significant disadvantage vis-à-vis Ryanair and Aer Lingus (695).

Congestion at terminal level

690. As concerns the existence of congestion problems at the terminal level, the CAR decisions state (again based on the Jacobs Report) that "the departures concourse at Dublin Airport is significantly congested. In the Summer 2006 peak morning period of 3 a.m. to 8 a.m., (...), the capacity of the departures concourse was reached for 82% of the peak period." The Jacobs Report expects that also in 2007 and 2008 congestion problems at security and immigration in Summer 2007. However, since the Jacobs report and the CAR expect that the capacity enhancements proposed by the DAA will mitigate these problems, and since competitors mainly refer to runway and stand capacity problems, the Commission concludes that congestion problems at the terminal can be disregarded as potential entry barriers at Dublin for the purpose of this decision.

Conclusion

691. In the light of the foregoing the Commission therefore considers that the congestion at Dublin Airport at certain peak times is likely to deter entry on routes from or to Dublin.

7.8.7.2. Constraints at other airports

692. As regards the capacity situation at other airports on the overlap routes, the following airports are "coordinated airports" within the meaning of the Slot Regulation: London- Heathrow, London-Gatwick, London-Stansted, Manchester, Paris CDG, Lyon St Exupéry, Frankfurt, Amsterdam, Milan-Linate, Milan-Malpensa, Rome-Fiumicino, Rome-Ciampino, Madrid, Barcelona, Vienna and Brussels-Zaventem. As will be set out in detail below (696), the congestion problems at many of these airports are even more serious than in Ireland, since some of these airports (such as London Heathrow /Gatwick or Frankfurt) are congested throughout the whole day. Others only have congestion issues parts of the day.

693. Secondary airports are generally not congested. However, the fact that customers, in particular low-frills customers, are in most cases ready to substitute primary airports by secondary airports, does not, contrary to what Ryanair claims, imply that airlines could easily substitute congested primary airports by secondary airports (697). Although many of these co-ordinated airports are in principle substitutable from a demand-side point of view with secondary airports, the possibility to fly to less congested airports is not an alternative for many airlines, in particular for those airlines whose service or operating model does not allow secondary airports to be used.

694. The level of congestion at the end of each analysed route will be discussed individually on a case by case basis in the context of the analysis of the competitive situation on the respective routes.

7.8.7.3. Airport congestion plays a role as a deterrent factor for potential entrants in the present case

695. As concerns Ryanair's argument that airport congestion play a more limited role than in other cases, since many of the routes relate to holiday/leisure routes, the Commission agrees with Ryanair that leisure customers are indeed more likely to accept departure/arrival times outside the peak-hours than time-sensitive business passengers.

696. The Commission also shares Ryanair's view that Dublin Airport is not congested throughout the whole day (698) and that there are usually no congestion problems outside the peak-hours.

697. However, the fact that it is possible for potential competitors to enter routes outside the peak hours does not exclude that operating outside the peak-hours might give an economic disadvantage to potential entrants, which might still deter them from entry to and preventing them from competing head-to-head with Aer Lingus and Ryanair.

698. This is particularly true for those routes, on which Ryanair and Aer Lingus operate already with a high number of frequencies, such as many routes between Ireland and the UK. The Commission's market investigation has shown that offering a wide range of slots, including early morning and late evening slots, allows Ryanair and Aer Lingus to attract a large number of time-sensitive passengers/business customers on these flights (699). A competitor who would not be able to offer flights during peak hours, could not meet the demand of these customers and would therefore automatically attract less passengers, to the detriment of his profitability. Since many of the potential competitors are very much focused on profitable time-sensitive passengers for which they offer specific ticket classes, these carriers cannot be expected to compete successfully against Ryanair and Aer Lingus without having certainty to obtain sufficient early-morning and late-evening slots at both sides of the routes. The fact that potential competitors, in particular those operating smaller aircraft, are aware that they will not even obtain a contact stand but will be forced to transfer their passengers via buses to remote stands (while the large Ryanair and Aer Lingus aircraft have better chances of obtaining a contact stand), is a further disincentive for all potential competitors on such routes.

699. On "leisure" routes, time-sensitive passengers are less relevant for competitors who want to enter a new route. Passengers flying on "leisure" routes are, in general, more willing to accept flight times outside the peak-hours. At the same time, the ability to offer early morning and evening slots also for these flights is not irrelevant, since it allows passengers to use the whole day of their holiday or short visit. The ability to offer early morning slots can therefore increase the attractiveness of a competitor's offer.

700. The risk of not being able to obtain slots during peak times is also a general problem that applies to all routes, regardless of whether UK-Ireland or continental "leisure" routes. This is because airlines (in particular low-frills airlines) aim at in general 3 to 4 turnarounds for an aircraft per day and try to start their operation as early as possible (i.e. during the peakhours) in order to maximise their aircraft utilisation. As set out above, an entrant who wanted to successfully compete with Aer Lingus and Ryanair would ideally operate from a base in Dublin. Such an entrant, however, would expect to be able to use a large number of early-morning and late evening slots in order to maximise his aircraft utilisation. The congestion at Dublin discourages therefore airlines from establishing a base at Dublin.

7.8.8. The merged entity's strong position at Dublin Airport might hinder further expansion by competitors

701. The merger would also have an effect on the use of Dublin Airport, since it would combine the two airlines which are by far the largest users of the airport. Ryanair's acquisition of Aer Lingus would create a customer who would account for the largest part of the traffic (around 70% of all passengers), while the second largest user (Aer Arann) would not carry more than 5% of all passengers at Dublin and other competitors would be even more fragmented. In line with the Horizontal Merger Guidelines, the Commission has therefore assessed whether the merged entity could us its potential market power at Dublin Airport to hinder entry or expansion by competitors (700).

702. A number of third parties have raised concerns that the merged Ryanair/Aer Lingus' paramount position as a customer in Dublin could be used to further increase the existing barriers to entry at Dublin (701). Some competitors expressed concerns that the merged entity could try to exercise its influence as an important customer to obtain preferential airport charges, notably at Dublin Airport (702). Indeed, Ryanair has been able to negotiate such preferential conditions at other airports (703). The internal documents of Ryanair confirm that [...]* (704) Further, the internal documents of Ryanair also show that [...]* (705). It should be noted that the airport charges are regulated as a price cap and thus the price regulation does not exclude possible discounts. In this context, a unique position of the merged entity could thus play an important role in the negotiations with Dublin Airport.

703. Competitors were also complaining about an increased influence of Ryanair on the allocation of the existing airport facilities, notably the existing contact stands, which are particularly important for business customer oriented airlines (706). Entry could be deterred e.g. should the merged entity use its position to get easier access to preferential contact stands close to the terminal, since this would relegate smaller operators to distant spots on the tarmac, accessible only by bus, whereas Ryanair could park their aircraft next to the terminal building and use contact stands or air bridges (707). In this regard it is noted that it is important for low-frills airlines to be close to the gates to facilitate quick turn-around times.

704. Many third parties also claimed that the merged entity's increased position might also allow them to block further expansion plans at Dublin Airport, thereby maintaining the existing congestion problems to the detriment of competitors who wish to enter on routes from and to Dublin (708). In particular, the merged entity could use its increased buyer power to pressure the airport authority not to proceed with the development of a new terminal and a new runway.

705. Ryanair claims not to have any power to influence the airport charges or expansion plans of the Dublin Airport Authority. It notably referred to the example of Stansted airport, where Ryanair was not able to influence the expansion plans of the airport, although Ryanair strongly opposed such plans (709). It has also provided evidence of the expansion plans of Dublin Airport.

706. The Commission's market investigation has indicated that the merged entity's strong position as a customer at Dublin could be used to exercise influence on decisions concerning the airport infrastructure which could further deter entry or expansion by competitors. Although the merger will not give Ryanair new legal rights to directly veto decisions on charges, facilities or expansion, it is noted that the regulatory framework obliges the CAR and the DAA to take the view of its then by far largest customer into account (710). After the merger, the merged entity will be more than 15 times larger than its next competitor in Dublin. This will give the new company not only an increased factual influence in the decision-making process. It will also have more legal possibilities to influence the decision making process in its own interest. For instance, the merged entity's weight in the Dublin Airport Coordination Committee (DACC) will increase significantly after the merger. According to the constitution of the DACC, a carrier operating more than 56% of so-called "air transport movements (711)" would hold the majority of votes. This will be the case for the merged entity post merger. The merger would therefore lead to a situation in which one company would hold the majority of votes in the DACC, with only a small and fragmented number of other airlines present in the DACC. While it is true that the DACC is not itself the airport regulator, it has nevertheless an important advisory function in the process with regard to defining the capacity and service requirements at Dublin Airport (712).

707. The fear of Ryanair's competitors (713) that the merged entity could use its majority within the DACC and its unique strong position at Dublin Airport to influence the regulator to shape the airport according to Ryanair's requirements may therefore be not unfounded. Indeed, while Ryanair prefers only very basic airport services for its low-frills service model, most potential entrants follow a different business model, for which they prefer a higher service level (e.g. lounges, contact stands, modern buildings etc.). This conflict can be illustrated by the example of the current legal dispute over the expansion plans of the DAA at Dublin (concerning the new "Terminal 2" and a new runway). While airlines representing the majority of customers at Dublin are in favour of the DAA's expansion plans today, the merged entity (Ryanair) is the company strongly opposing to the expansion. Ryanair's opposition to date will gain further weight post merger when it will represent more than two thirds of the customers at Dublin. The chances of influencing the expansion plans in order to make them more tailored to the Ryanair business model will therefore increase through the merger.

708. It can thus be concluded that the merger also allows the merged entity to increase its weight in the consultation process for airport charges, airport facility allocation or expansion plans. This may make entry/expansion of its competitors more difficult.

7.8.9. Analysis of Ryanair's and Aer Lingus' actual and potential competitors

709. Ryanair argues that a number of airlines would be able to enter the Irish routes or expand their current operations if Ryanair/Aer Lingus were to raise prices. These actual or potential competing airlines includes inter alia: Aer Arann, Air Berlin, Bmi/bmibaby, Lufthansa, British Airways, Clickair, easyJet, Flybe, and SkyEurope which are described in the following paragraphs (714). This section describes those competitors which are considered by the Commission as belonging to the most important competitors on the overlap routes in order to provide a general overview of their abilities and incentives to enter the markets affected by the proposed transaction. The fact that some competitors are not mentioned in this Section does not mean that they are disregarded by the Commission, since the competitive constraint exercised by individual competitors will in any event be discussed in Section 7.9 (route-by-route analyses.

710. The Commission notes that Ryanair also mentions US airlines who are not yet present in the short-haul markets in Europe as their "main competitors". Ryanair also refers to the recently concluded "Open Skies" Agreement, which would lead to more activities of US airlines in Europe. While the Commission does not contest that the "Open Skies" Agreement will probably give rise to an increasing level of activities of US airlines in Europe, it can be expected that the activities of US airlines will, at least for a starting period and until they will be established in Europe, focus on the long-haul market. Indeed, the Commission has not received any indications from its market investigation that US carriers are regarded as credible competitors on the European short-haul markets in a short or even mid-term perspective. The Commission can therefore not rely on the potential entry of non-European airlines and will not further analyse this in this Section.

711. In the following paragraphs, the competing carriers based at the relevant Irish airports are discussed first (CityJet and Aer Arann), followed by the most important other competitors active in Europe.

7.8.9.1.CityJet (Air France)

712. CityJet is an airline registered and incorporated in Ireland, although it has been a 100% subsidiary of Air France since 1999. As is clear from its ownership and the radial structure of its current network to and from Paris CDG, CityJet is primarily a feeder of traffic for the hub and spoke operations at Paris CDG of its mother company Air France and the latter's partners in the Skyteam airline alliance.

713. CityJet operates a fleet of 20 BAe 146 aircraft capable of seating 93 passengers, i.e. more or less half of the seating capacity of Boeing 737-800s as configured and operated by Ryanair. BAe 146s are particularly suitable for short runway operations due to their four jet engines. It should be stressed that the unit operating costs of these aircraft are significantly higher than those of the aircraft used by Ryanair and Aer Lingus. Although CityJet has in principle access to larger Aircraft operated by Air France, it has no intention using large aircraft on Irish routes (715).

714. In Ireland, CityJet is based in Dublin with in total 3 based aircraft operating two routes: Dublin - London City and Dublin - Paris CDG with high frequency flights. Therefore, compared to the number of aircraft based in Dublin and routes operated by the merged entity, the scope of its activities in Dublin is much more limited.

715. In terms of service differentiation CityJet is clearly a full service carrier. It describes itself as being "a scheduled airline offering a network carrier service from a Low Cost Base [...]" and "a Network and Regional Business Carrier"716. Further, CityJet confirms (717) that its main focus is first on the connecting passengers (in particular for routes from/to Paris CDG) and second on point-to-point business passengers which represent a substantial part of the point-to-point operations of CityJet. Given the significantly lower capacity of its planes, CityJet must achieve a higher average yield than Ryanair/Aer Lingus to be profitable - this is secured by focusing on connecting passengers and business passengers at key times of the day.

716. London City airport, to which CityJet operates, is subject to operational restrictions due to its short runway. This means that neither Ryanair nor Aer Lingus could, with the aircraft currently in their fleet, fly between Dublin and London City Airport. This allows CityJet to differentiate strongly its services from those of Ryanair or Aer Lingus. This differentiation on the basis of the airport that is served is reinforced by the differentiation in terms of service level to the extent that CityJet offers a "full service" type of product according to two service classes. The more "upmarket" positioning of CityJet is corresponds with the fact that at CityJet's destination airport, London City Airport, primarily business passengers are served (718).

717. The Commission observes in this regard that CityJet has recently announced the launch, starting 26 March, of 6 new routes to/from London City (Belfast, Geneva, Milan Linate, Madrid, Nice and Zurich). CityJet indicated that due to congestion and difficult access to infrastructure such as contact stands, such expansion would not be possible from Dublin (719).

718. In a telephone interview with the Commission on 21 February, CityJet indicated that it could be interested in the future in entering specific routes out of Dublin which fit best with its business oriented model. CityJet did not specify the routes but indicated that entry into these routes is not possible at present for CityJet as it would not be able to get access to the necessary peak slots and other infrastructure (e.g. contact stands) at Dublin and at the destination airports needed for its high frequency business/connecting passengers services. CityJet insisted that in particular lack of access to the contact stands in Dublin was particularly harmful to it in so far as remote/bus airside operations were viewed as an inconvenience by its high-yield business passengers. CityJet indicated that potential entry to any other of the overlap routes to/from Dublin (in particular Continental European routes) would "be further down the list of priorities for CityJet" (720). It should also be noted that entry on a number of the "non-hub" or even leisure routes to Continental Europe would not be in line with CityJet's business model.

719. The Commission also notes that its regression analysis did indicate that CityJet does not exert a competitive constraint on Aer Lingus .In view of CityJet's business model, this will also be true with respect to Ryanair. For more details see Annex IV.

720. The Commission concludes that entry of CityJet on overlap routes that it currently does not serve is unlikely. Finally the Commission observes that, if entry by CityJet occurred or if CityJet expanded on the London-Dublin route, the resulting competitive constraints on the merged entity would be limited, given (i) the lower seating capacity of the aircraft operated by CityJet and resulting cost disadvantage and (ii) the fact that CityJet, on the one hand, and Ryanair or Aer Lingus, on the other, do not appear to be close competitors due to extensive differentiation in service level and business model.

7.8.9.2.Aer Arann

721. Aer Arann defines itself as a regional carrier. It operates 35 routes across the United Kingdom and Ireland, and has only two summer season destinations in Western France.

722. Aer Arann, although marketing itself as a "low cost" airline, is nevertheless more similar to carriers such as CityJet as concerns its business model, since is carries a large proportion of business and connecting passengers and uses small aircraft which does in general not allow the same low fares to be offered as airlines operating larger aircraft. The closeness to business oriented network carriers is further evidenced by its partnerships with full service carriers such as British Airways, Air France or SAS, and, more importantly by the fact that it sells unrestricted (fully flexible) tickets.

723. Aer Arann operates a fleet of ATR 42 and 72 turbopropeller aircraft Compared to Boeing 737 or Airbus 320 aircraft, ATR 42/72 aircraft are designed for (very) short haul operations, and have a limited operational range and cruising speed. Their seating capacity is much lower with 50 seats for the ATR 42 and 66 seats for the ATR 72. Due to the type of aircraft operated, it is practically excluded that Aer Arann would be able to enter any route beyond Ireland, the UK and the nearest Continental Europe destinations.

724. Aer Arann has 4 aircraft (2 ATR42 and 2 ATR72) based in Dublin and 3 aircraft (1 ATR42 and 2 ATR72) in Cork. Aer Arann also bases 4 aircraft in Galway, 1 aircraft in Waterford and 1 aircraft in the Isle of Man. Therefore, compared to the total number of aircraft based in Dublin and routes operated by the merged entity, the scope of its activities in Dublin is much more limited. The ability of Aer Arann to enter Cork - London in view of its base at Cork is discussed further also in the relevant analysis of the Cork - London route below.

725. As a result of the difference in the type of aircraft that they operate, Aer Arann explains that their operational costs per seat are higher than those of Ryanair (721). The Commission observes that, should Aer Arann enter a route in competition with Ryanair, it would have to operate at a frequency three times higher than Ryanair with a view to offering the same capacity on the route. This is due to the lower seating capacity of Aer Arann's aircraft. Aer Arann explains that they would not enter any route where Ryanair is already present (722). Ryanair itself admits with respect to Air Southwest's exit from the Dublin - Bristol route that it was likely that operating a 50-seat turbo prop aircraft on this route "was not economically sustainable" (723). The Commission observes that in this is confirmed by the behaviour of Aer Arann in the recent years as there was no instance (at least in the last 5 years) of entry by Aer Arann on a route already operated by Ryanair (724).

726. The Commission also notes that the regression analysis conducted by the Commission indicated that Aer Arann would not exert any competitive constraint on Aer Lingus and it can be assumed that in view of the Aer Arann's business model it would be the case also with respect to Ryanair. For more details see the description of the regression analysis in Annex IV.

727. In the light of the above, the Commission considers that the possible entry of Aer Arann on one or more of the relevant routes (markets) for the present purposes is very unlikely. Given this unlikelihood moreover, the Commission takes the view that the threat of possible entry of Aer Arann onto one or more of the relevant routes in this case would not result in an effective competitive constraint on the merged entity. It should be stressed that in view of its cost disadvantage due to its fleet and differing business model, Aer Arann would not be able to offer a comparable substitute to Aer Lingus' services even on the Ireland - UK routes.

7.8.9.3.easyJet

728. easyJet is Ryanair's historical rival in its capacity as one of the first, fast growing low frills carriers in the European Union since the nineties. Starting from its original base in London Luton airport, easyJet has developed a European wide network of 289 point to point routes across 74 routes. It has expanded across the UK and towards the West across continental Europe, and more recently to Morocco.

729. easyJet is a low-frills airline as evidenced by the company's mission Statement (725) and the principles driving its business (726): use of the Internet to reduce distribution costs, maximising the utilisation of substantial assets (in particular aircraft), ticketless travel, no free lunches, efficient use of airports, paperless operations.

730. In terms of the differentiation which air carriers seek to achieve between their respective services, easyJet positions its service slightly "upmarket" compared to Ryanair, as evidenced inter alia by its relative preference for primary airports and the fact that it does not charge passengers for the first piece of checked in luggage. In this regard, the respective services of easyJet and Aer Lingus and Ryanair can be regarded as close substitutes.

731. EasyJet operates a mixed fleet of Airbus 319 and Boeing 737-300 and 731-700 aircraft. Gradually, the latter are being phased out in exchange for the former. Except that they offer a lower seating capacity, the aircraft operated by easyJet are very similar to the Boeing 737-800 operated by Ryanair or the Airbus 320-321 operated by Aer Lingus.

732. Ryanair regards easyJet as its "main competitor (727)". EasyJet is, however, not an actual competitor on any of the 35 overlap routes, since easyJet does not operate flights from the Republic of Ireland. In view of the many points in common between easyJet, Ryanair and Aer Lingus, easyJet may be regarded as a potential competitor to the merged entity on several of the relevant routes in this case. However, the Commission observes that although Ryanair and easyJet have concurrently and at a fast pace expanded throughout Europe since the nineties, easyJet is not active on any Irish routes. Further the internal documents of Ryanair acknowledge that "[...]* (728)". Furthermore, easyJet made an attempt to enter the routes between London, on the one hand, and Shannon and Cork in the republic of Ireland, on the other, in January 2005 and was confronted with a vigorous competitive reaction from Ryanair until it exited the routes a year later. As explained in Section 7.8.5 above, the nature and intensity of this reaction as well as internal documents of Ryanair strongly suggest that Ryanair intended to deter easyJet from entering the Irish market.

733. Ryanair argues (729) that the unsuccessful entry to the Irish regional airports did not deter easyJet from competing with Ryanair as demonstrated by the fact that easyJet recently opened two bases (Milan in March 2006 and Madrid in February 2007) in cities where Ryanair was already based. However, the position of Ryanair in Milan (4 based aircraft and around 15 routes served) and Madrid (4 based aircraft and 24 destinations served) differs significantly from its much stronger position and reputation in Dublin and Ireland. Further, only in case of Madrid did easyJet set up a base at the same airport, as in Milan it chose Malpensa, while Ryanair is based at Bergamo (even though it should be acknowledged that these airports are substitutable from low-frills customers' point of view for short-haul routes as described above in the Market Definition section (730)).

734. Moreover, easyJet has no concrete plans to enter the Irish market on a small or large scale to replace Aer Lingus in case of a merger with Ryanair. Any entry into Dublin, easyJet explained in an interview, would have to be of a sufficient scale and require "important investments (731)". Starting operations to Dublin would according to easyJet need significant marketing efforts with significant costs to compete against such two well-established brands in Ireland such as Aer Lingus and Ryanair. EasyJet confirms that it would still not be profitable to start a route only on the basis of the UK passengers knowing easyJet (even though on the Ireland/UK routes the passengers are divided around 50/50 between British and Irish passengers). An important share of the Irish originating customers would be necessary as well. Any potential entry would thus have to be with a number of routes, not only on one or two routes. In this respect it should be noted that entry with a larger number of routes to/from Dublin (and in particular in case of the thick Ireland-UK routes) would also require significant capacities at the Dublin Airport including peak-time slots, which would due to the capacity constraints at Dublin Airport make entry even more difficult and unlikely.

735. easyJet indicated that there are a number of more profitable and less risky routes in Europe where its aircraft can be used and the opportunity costs of entering Ireland are simply too high. Moreover, easyJet indicated that the Irish market is relatively small, has only three (major) airports, it is not a pass-by destination, but located at the very border of the European market and there are already two strongly established brands. Due to these various aspects, entry would according to easyJet be more difficult than elsewhere in Europe.

736. In the light of the above, the Commission does not consider that the establishment of a base or the entry of easyJet on one or more of the relevant routes in this case would be likely in the short to medium term. This is in particular because it is unclear whether such entry would be sufficiently profitable, taking into consideration the price effects of injecting additional capacity and the potential responses of a combined Ryanair-Aer Lingus. It follows that easyJet is unlikely to exert a significant competitive constraint on the merged entity in the relevant markets.

7.8.9.4.Sky Europe

737. SkyEurope is a low frills carrier incorporated and registered in Slovakia. It is a typical low frills airline, operating "point to point" to primary as well as to secondary airports on a one-way basis (no return tickets) and offering just basic on board services.

738. It operates 12 aircraft (14 aircraft from 25 March 2007 onwards) out of 5 bases in Central and Eastern Europe to 99 destinations in 19 countries. Its fleet is entirely composed of Boeing 737, of which it has 12 on order for delivery in 2007 and 2008.

739. Ryanair argues that SkyEurope is a potential entrant and close competitor on routes to/from Dublin, in particular for destinations in Central or Eastern Europe.

740. SkyEurope recognises the demand for air transport services between points in Ireland and point in Central or Eastern Europe is growing. This seems to result from to the large migrant workforce from the new Member States of the EU in Ireland. In its reply to the Market Test of Ryanair's Final Commitments, SkyEurope has clearly stated that it does not consider opening a base in Dublin. SkyEurope had already in previous submissions pointed out entry barriers to the Irish markets (732).

741. It can therefore not be expected that SkyEurope would enter on any of the overlap routes in the Irish market after a possible merger between Ryanair. Furthermore, given that even SkyEurope's whole actual fleet (12 aircraft) would not suffice to cover the 35 overlap routes according SkyEurope (which is in line with the results of the Market Test (733)), the Commission cannot rely on the mere possibility of entry by SkyEurope in this case.

742. In addition, the internal documents of Ryanair confirmed the pressure put on SkyEurope [...]* (734). Further, as regards the financial situation of SkyEurope, Ryanair in its internal documents states that [...]* (735).

743. In the light of the above, the Commission considers that entry or expansion of SkyEurope against the strengthened merged entity on one or more of the relevant routes in this case is not likely in the short to medium term. This is in particular because it is unclear whether such entry would be sufficiently profitable, taking into consideration the price effects of injecting additional capacity and the potential responses of a combined Ryanair-Aer Lingus entity.

7.8.9.5.British Airways

744. British Airways is the United Kingdom's largest international scheduled airline, flying to over 550 destinations worldwide (it should be however noted that despite this, Ryanair carries significantly more passengers on its European routes than British Airways in their whole worldwide network - according to Ryanair's internal documents [...]* (736)). British Airways is a full service airline, with a hub and spoke operation at London Heathrow airport. It is a member of the One World airline alliance.

745. The fleet of British Airways consists of regional, short to medium haul as well as long haul aircraft. On European routes in particular, it consists of Boeing 737s, Airbus 319/320/321 aircrafts as well as regional aircraft with a lower seating capacity and operational range.

746. British Airways competes with Ryanair on the Dublin to London route, which British Airways serve from London Gatwick airport. Ryanair argues that British Airways could, in its capacity as one of the main carriers in Europe with its main hub and important operations in the United Kingdom, enter routes between points in Great Britain, on the one hand, and Dublin or other points located in the Republic of Ireland, on the other. Actual or potential entry by British Airways would exercise a competitive constraint on the merged entity on routes between Ireland and the United Kingdom in particular. Ryanair names British Airways as its second most important competitor (737).

747. However the Commission observes that British Airways, despite its size and financial strength, is only active on one single route in competition with the Merging Parties, where it holds a relatively limited market share of below 10% on the Dublin-London route.

748. In the market investigation, British Airways has indicated not to have any concrete intention to enter or expand any other route than the Dublin-London route (which might be an attractive route for British Airways in order to channel passengers to its Hub in London Heathrow) (738). Moreover, British Airways, similar in this respect to other major network carriers in the European Union, tends to disengage from regional or short haul point to point scheduled air transport services. On 3 November 2006, British Airways announced (739) that it has reached agreement in principle to sell the regional operation of its subsidiary airline BA Connect to Flybe. On this occasion, Willie Walsh, British Airways chief executive is reported to have said: "Point to point regional operations are not a strategic part of our business and we believe that such activities are better undertaken by a regional low cost airline. Despite the best efforts of the entire team at BA Connect, we do not see any prospect of profitability in its current form. London City services complement our mainline business at Heathrow. For this reason they are not included in the proposed sale." (740)

749. In the light of the above, the Commission does not consider that entry/expansion of British Airways on one or more of the relevant routes between Ireland and Great Britain except the Dublin-London route would be sufficiently likely in the short to medium term to remove the Commission's competition concerns on the overlap routes. This is in particular because such entry would deviate from British Airways' current strategy, as is apparent from the divestment of its participation in BA Connect.

7.8.9.6.Air Berlin

750. Air Berlin defines itself as a German charter and scheduled low frills as well as full service Carrier (741). It operates from several base airports in Germany, as well as a base in Palma de Mallorca and London Stansted.

751. Air Berlin serves popular holiday destinations along the Mediterranean coast, the Canary Islands, Northern Africa as well major European cities. Palma de Mallorca is the prime destination of Air Berlin's route network, where they operate a "shuttle" with daily flights from 12 German airports. The total number of weekly flights to Majorca should exceed 360 in the Summer 2007. In addition to its original focus on popular holiday destinations, Air Berlin launched in October 2002 the City Shuttle, whereby it connects German airports with major European cities like Amsterdam, Barcelona, Bournemouth, Budapest, Copenhagen, Helsinki, London-Stansted, Madrid, Milan-Bergamo, Manchester, Paris, Rome, Vienna and Zurich (742).

752. Air Berlin is positioned at the quality end of the low frills spectrum as appears from the various quality awards obtained by this carrier. Unlike other low frills carriers, its operation is not entirely point to point in so far as it operates three of its base airports as a hub: Nuremberg, Palma de Mallorca and London Stansted. It has also entered into the first alliance agreement between low frills carriers with Niki with a view to developing the Austria and Eastern European markets.

753. Air Berlin operate a mixed fleet of Airbus 320 (15 aircraft), Boeing 737 (40 aircraft) and, to a much lesser extent Fokker 100 (3 aircraft). Air Berlin is committed to a purchase order of 60 Airbus 320 aircraft, with purchase options for a further 40. It is therefore strongly geared towards growing.

754. Air Berlin explain that they are "not interested in opening direct routes from Dublin to continental destinations (743).". This would not fit in Air Berlin's plan for growth. Moreover, other European routes such as Scandinavia or Russia would have better yield and profit prospects (744). The Market Test of the Final Commitments has confirmed that Air Berlin does not even "consider" entry to any route from/to Dublin (745).

755. Further, from the internal documents of Ryanair it seems that even Ryanair does not see Air Berlin as a credible competitive threat as they state [...]* (746).

756. In the light of the above, the Commission considers that entry of Air Berlin on one or more of the relevant routes (markets) for the present purposes is unlikely. Given this unlikelihood moreover, the Commission takes the view that the threat of possible entry of Air Berlin into one or more Irish airports from airports in Germany or from London Stansted would not result in an effective competitive constraint on the merged entity.

7.8.9.7.bmi (British Midland Airways)

757. British Midland Airways (bmi) defines itself as a scheduled full service network airline (747) (bmi has a separate subsidiary, bmibaby, in the low frills segment of the market - see below, Section 7.8.9.8). It operates from London Heathrow as a hub, where it has long as well as short and medium haul services. In addition, bmi operates point to point regional air transport services under the "bmi regional" brand from Aberdeen, Edinburgh, Glasgow, Leeds Bradford, Manchester and East Midlands. bmi is a member of the Star Alliance. Lufthansa and SAS own respectively a 30% minus one share and a 20% shareholding in bmi.

758. As is apparent from the nature of its services, the radial structure (around London Heathrow) of its services (748) other than regional, or its ownership structure, bmi is positioned as a full service network carrier focusing on operating to and from Europe's busiest airport, London Heathrow. This is also confirmed by the submission of bmi which indicates Heathrow as the airline's principal base of operations for its core mainline services (749). Further, corporate contracts are an important component in bmi's overall sales activities contributing to a significant proportion of the airline's annual turnover as well as attracting a stronger level of yield compared to leisure sales (750).

759. bmi operates a fleet of Airbus 330 (long haul), Airbus 319, 320 and 321 (short to medium haul) and Embraer 135 and 145 (regional) aircraft.

760. bmi competes with Ryanair and Aer Lingus on the route between Dublin and London, where bmi operates a service between Dublin and London Heathrow (751). The Commission cannot exclude the possibility that bmi may be interested in expanding the capacity and/or the frequencies it operates between Dublin and London Heathrow. Similarly, the Commission cannot exclude that bmi may show some interest for launching a service between Shannon and London Heathrow or Cork and London Heathrow. All such options appear prima facie as compatible with bmi's hub operation in Heathrow or its position in the market. The Commission observes, however, that London Heathrow is one of the busiest airports in the world. It is heavily congested. Therefore, airlines with a slot portfolio at London Heathrow seek to maximise revenue opportunities from this portfolio. Depending on the demand for connecting services on long haul flights operated by bmi via Heathrow, serving London Heathrow from Shannon or Cork may therefore imply substantial opportunity costs. It appears, therefore, unlikely that bmi would consider replacing the competitive constraint exercised today by Aer Lingus on the relevant routes between London and Dublin, Cork or Shannon.

761. As regards the strength of the merged entity, bmi stated that they would expect the combination of Ryanair and Aer Lingus to exert additional competitive pressure on bmi's services, even though it is difficult to assess without knowing what the transaction will involve (752). As regards possible entry, bmi indicated that increase in competitors' fares on a route currently not served by bmi is not alone a credible reason that bmi (or any other airline) would consider entering such a route as there are further broad criteria which needs to be considered (e.g. consistency with bmi's network expansion strategy, slot availability, aircraft availability, financial resources, commercial viability and regulatory/bilateral issues (753). It should be noted that bmi has not in the last five years entered any new route between Ireland and the UK in direct competition with Ryanair or Aer Lingus (754).

762. As regards potential entry on the Ireland - UK routes, bmi as well as bmibaby state that they are confident that they can provide a competitive service to Ryanair/Aer Lingus but that "the ability to exercise a competitive constraint and to specifically enter or expand services on a particular Ireland-UK route may be restricted by high barriers to market entry at both ends..." (755).

763. Bmi also under "bmi regional" brand operates niche UK domestic and intra-European scheduled passenger, cargo and mail and ad hoc charter air transport services from UK regional airports (756). However, due to its full-service business model and a fleet of small aircraft (Embraer 145 with maximum 49 seats and Embraer 135 with maximum 37 seats), bmi regional cannot be considered as a potential entrant able to exert significant competitive constraints on the high volume services of the merged entity between Ireland and the UK.

764. Bmi has confirmed that neither bmi nor bmi regional can be expected to enter any routes in competition with the merged entity (757). bmi's activities concentrate on the London- Heathrow route and thus it would not enter any other routes than the currently served Dublin - London. They also confirmed that establishing a brand in Ireland able to compete effectively with Aer Lingus and Ryanair brands is costly. Further, entry/expansion to Ireland would involve opportunity costs as bmi would have to use aircraft operating now elsewhere. The expansion strategy of bmi is focused rather towards mid-haul routes to Continental Europe (e.g. Moscow or holiday destinations with their recent acquisition of BMed). (758)

765. In the light of the above, the Commission considers that entry by bmi or bmi regional on one or more of the relevant routes (markets) for present purposes is unlikely.

7.8.9.8.bmibaby

766. Bmibaby is the low frills subsidiary of British Midland Airways (See above). It defines itself as a scheduled low frills airline (759). It operates from four bases in Great Britain (Manchester, East Midlands, Birmingham and Cardiff). It was formed in 2002 by British Midland Airways in response to the intense competition from low frills carriers in the United Kingdom including Ryanair. Bmibaby serves 24 destinations in 8 EU Member States. It does not fly to any of the London airports (760).

767. Bmibaby operates a fleet of 19 Boeing 737 aircraft.

768. Bmibaby serves Knock and Cork in the Republic of Ireland from Manchester and Birmingham airports. The services of bmibaby on Knock-Manchester and Cork- Birmingham compete with the services offered by Ryanair respectively between Cork and East Midlands (as a substitute for Birmingham) and between Knock and Liverpool (as a substitute for Manchester). On these routes therefore, competition between Ryanair and bmibaby takes place on the basis of a differentiated service as regards the airports served. It is noteworthy that bmibaby and Ryanair used to compete on an airport to airport basis during March 2004 and March 2005 following Ryanair's entry on the Dublin to East Midlands route. In March 2005, bmiBaby exited the route. Bmi itself acknowledges that the market entry of Ryanair onto the East Midlands - Dublin route prompted bmibaby to withdraw its services (761). Commenting on this exit, bmi indicated the following: "The decision to withdraw the service was a commercial decision because the route no longer produced a viable economic return in direct competition with Ryanair's services at the time. bmibaby's load factor and overall yield was no longer sustainable to continue scheduled operations." (762) Further, since bmibaby commenced scheduled operations in March 2002, it has entered only one route in direct competition with Ryanair (London Gatwick - Cork commenced in 2004) but withdrew this route in 2005 "under similar circumstances to bmibaby's East Midlands - Dublin" (763). In the same period, bmibaby entered only one route in direct competition with Aer Lingus (Birmingham - Cork commenced in 2005) (764). It should also be noted that bmibaby does not operate any route to/from Dublin and that the overlaps between the Merging Parties on routes to/from the UK regional cities served by bmibaby are limited to Dublin only.

769. As regards potential entry to the Ireland - UK routes, bmi as well as bmibaby state that they are confident that they can provide a competitive service to Ryanair/Aer Lingus but that "the ability to exercise a competitive constraint and to specifically enter or expand services on a particular Ireland-UK route may be restricted by high barriers to market entry at both ends..." (765).

770. The phone conference with bmi/bmibaby indicated that even though bmibaby may be more likely to enter any of the overlap routes than bmi or bmi regional, there are no plans for such entry (766). Bmibaby confirmed that it had been difficult in the past for bmibaby to compete with Ryanair. Further, any entry would need to be with larger number of frequencies covering both morning and evening to serve also business customers. In that respect, congestion at the relevant airports represents a barrier to entry. Bmibaby indicated that Birmingham is particularly congested and thus capacity would be a crucial factor for the question of entering new routes. (767) Further, the interview with bmi/bmibaby indicated that bmibaby targets expansion to European leisure destinations (beach/city break) (768).

771. In the light of the above, the Commission considers that the possible entry of bmibaby on one or more of the relevant routes (markets) for the present purposes is unlikely.

7.8.9.9.Flybe/BA Connect

772. Flybe considers itself as a low fares regional airlines mainly with flights in the UK with bases in a number of regional cities in the UK. However, in comparison to Ryanair and Aer Lingus, Flybe is more focused on serving business passengers as they also offer separate more flexible type of tickets (Economy Plus), executive lounges at a number of UK airports, a frequent flyer program, corporate deals or dedicated check-in desks for Economy Plus passengers. The focus of Flybe/BA Connect on business passengers is also confirmed by the recent UK Office of Fair Trading decision concerning their merger which analysed in particular the impact of the merger on business passengers (769). Currently, Flybe is in the process of acquiring BA Connect, a UK regional airline belonging to British Airways with similar business focus as Flybe.

773. In Ireland, Flybe currently operates three routes to/from Dublin (Exeter, Southampton and Norwich) and three routes to Galway (Southampton, Belfast and Birmingham). However, Flybe does not compete on any route directly with Ryanair (only indirectly on Dublin - Southampton where Ryanair operates Dublin - Bournemouth).

774. Flybe operates three types of aircraft: Q400 with 78 seats, Embraer 195 with 118 seats and BAe 146-300 with 112 seats. BA Connect operates a number of different types of aircraft which are in general even smaller than Flybe's aircraft (around 50 seats with only a few larger aircraft). Therefore, the aircraft of Flybe/BA Connect are significantly smaller than the aircraft used by Ryanair and Aer Lingus and less able to compete efficiently with the low-frills operations of the Merging Parties. Flybe indicated that the small 50-seat aircraft of BA Connect will be gradually replaced by larger and more efficient aircraft. They also indicated that in view of this replacement it is unlikely that any expansion to new routes would be possible at least in the next two years (770).

775. The strategy of Flybe (including BA Connect) is to focus on regional routes to/from UK provincial cities. In general, Flybe is trying to avoid capital cities and prefers flying to regional destinations (771). Flybe has confirmed that it has no plans for further expansion of its activities to Ireland, even if prices increase on routes from/to Dublin (772). It would rather expand its activities in Northern Ireland (Belfast). Even after the replacement of its fleet, other destinations are "more interesting" than Irish destinations for Flybe (773).

776. In the light of the above, the Commission considers that the possible entry of Flybe/BA Connect on one or more of the relevant routes (markets) for the present purposes is unlikely.

7.8.9.10.Clickair

777. Clickair is a newly established airline which started its operations in October 2006. The founding partners, each with 20% of the total shares, are Iberia Líneas Aéreas airlines, Cobra (a Service Company belonging to Grupo ACS), Iberostar tour operators, Nefinsa, and the venture capital fund Quercus Equity (Grupo Agrolimen).

778. Clickair is currently based in Barcelona (BCN) airport but plans to set up also other bases in other Spanish cities. By the end of 2007, it plans to base 2-3 aircraft also in Seville and Valencia. In total, they plan to have 24 Airbus 320 aircraft by the end of 2007 (of which 21-22 will be based in Barcelona) (774). According to the web-page of Clickair, the creation of sub-bases will depend on the potential development of point-to-point routes that have a point of origin and place of destination other than Barcelona (775).

779. Clickair can be regarded as mid-frills carrier combining the low-frills model with some traditional features of network carriers. On its website, Clickair describes its mission as follows: "Our objective is to create a new, flexible and convenient product which combines the advantages of low cost airlines with the flexibility and value added services found at conventional companies, with the aim of becoming one of the top options for both leisure and business travellers." (776). Clickair also has a code-sharing agreement with Iberia for all its flights and participates in Iberia's frequent flyer programme (777). This code-sharing agreement allows Clickair to have its services available through GDS-powered points of sale. According to Clickair, the share of tickets sold under the code-share agreement with Iberia will indeed be important and in any event "larger than the usual airline-to-airline marketing code-share agreement, which may contribute a few points of percentage share in individual markets" (778).

780. As concerns possible entry plans to Dublin, Clickair has stated that it "is in principle interested in expanding its routes and in further growth across all the non-domestic markets (...). This does in principle include the Irish market. However, Dublin would not be a "core market" for Clickair. In particular after a merger between Ryanair and Aer Lingus, Ireland would become a less attractive alternative. It would not be on the "top 5" or even "top 10" list of possible potential future aircraft bases for Clickair (779)." In its reply to the Market Test of the Final Commitments, Clickair indicated that it would maybe consider to enter on two of the 35 overlap routes (Malaga, Barcelona (780)), but excluded that it would open a base in Dublin ("to consider competing with Ryanair at its own base would be financially irresponsible for just about any company") (781). Clickair therefore has no concrete plans at present as to when and which route they would be willing to enter. On the contrary, Clickair notes that it would have to face the strong "home carrier" against which it would be difficult to compete. According to Clickair, the home carrier would have the advantage of scale, which may provide the home carrier with the ability to significantly undercut the new entrant's proposed fares. Clickair also argued that it would require a critical mass of at least 5 aircraft based in Dublin to attract sufficient customers to have a chance to compete with the merged entity and that entry with only 2 or 3 aircraft would not be enough (782). Further, Clickair started their operations only recently and thus does not have such established brands as Ryanair and Aer Lingus and it is not clear at present to what extent it will be able to successfully enter the short-haul air transport markets. Even though the Iberia brand is established and can assist to Clickair in view of their code- sharing agreements, Iberia is not seen as a low-frills low-fares brand. Further, in the routes between Dublin and Spain, most of the passengers originate in Ireland and it is thus doubtful whether Clickair would be able to expand significantly its mid-frills operations against the merged entity's strong brands in Ireland, since it would have to win the large majority of its customers in Ireland, where it enjoys very little brand-awareness. Clickair explained that marketing costs to win customers on a new route from/to Dublin would not only be very high for the entry, but any new company would have to sustain its marketing investments, since the merged entity would fight constantly against the new competitor (783).

781. Clickair replaces Iberia on the Dublin - Barcelona route as of the Summer 2007 period. As regards possible entry into other overlap routes, it is less likely due to their centre of operations at Barcelona. If a base at Seville would be set up, the route Dublin - Seville could be potentially served by Clickair but it is doubtful whether Clickair would be able to exercise any significant constraints on the merged entity (for more details see bellow in Section 7.9.3 - Dublin - Seville). Clickair further claims that in case of a non-base entry (e.g. W-flights), the new entrant would face a significant disadvantage, because the costs of such a competitor would be higher (784).

782. In the light of the above, the Commission considers that the possible new entry of Clickair on one or more of the relevant routes is not likely enough. It is uncertain whether the relatively remote threat of a possible entry of Clickair onto one or more of the relevant routes in this case would result in an effective competitive constraint on the merged entity.

7.8.9.11.Other competitors

783. The Commission has not received an indication from any other airline that it would enter routes to\from Ireland. The Commission is therefore not able to conclude with the necessary degree of likelihood that actual or potential entry of another airline with a sufficient scale would compensate for the substantial loss of competition between the Merging Parties.

7.8.10. Conclusion on entry

784. The Commission has found that there are a number of important entry barriers to operating flights from or to Dublin in competition with Ryanair/Aer Lingus. These barriers go well beyond the problem of the partly congested airport and are in particular linked to Ryanair's and Aer Lingus' well established position at their home base. The investigation showed that, as a result of these barriers, entry is not likely, but even unlikely on almost all overlap routes. In the absence of potential entrants on the vast majority of overlap routes and given that competitors have unanimously indicated that they would not even consider entering in direct competition with the merged entity on a significant scale (in particular by opening a base at Dublin), the Commission concludes that entry is not likely, timely and sufficient to constitute a sufficient competitive constraint on the merged entity and defeat the likely anti-competitive effects of the proposed merger.

7.9. Route-by-route analysis

7.9.1. Overview

785. In general, two main groups of overlap routes with distinct characteristics can be distinguished - routes between Ireland and the United Kingdom and routes between Ireland and Continental Europe. Even though these main groups can be further subdivided according to a number of criteria, they have some common features and are thus discussed separately below.

7.9.2. Ireland - United Kingdom routes

786. The routes between Ireland and the United Kingdom are among the most important routes as regards the number of passengers. In particular the Dublin - London route is a "thick" route - with more than 300 rotations operated by all carriers each week between Dublin and London. However, other United Kingdom routes, in particular Manchester, Birmingham, Glasgow or Edinburgh are also among the most important routes to/from Ireland.

787. Further, most of the Ireland - United Kingdom routes involve a significant amount of business traffic. As it is shown in greater detail for each route below, both Ryanair and Aer Lingus carry an important share of business customers and it appears that it is vital for the Ireland - United Kingdom routes to provide services which also attract business customers. As indicated in the section on Market Definition, these customers are not secured by Ryanair and Aer Lingus by special offerings (for example, business class service, lounges) (785), but rather through high frequency services covering important peak times of the day. For this reason, any new entry or expansion would also need to offer high frequency services at convenient times in order to attract some higher-yield business or more "time-sensitive" passengers to ensure profitability of the operations (786).

788. As indicated in Section 6.8., the percentage of passengers booking late (for example, 7 days before departure) can be considered as one of the indications of the importance of more time sensitive passengers on the route. Both Aer Lingus and Ryanair submitted their calculations concerning the share of late booking passengers on their flights for individual routes indicated in the route-by-route analysis. Ryanair in fact submitted two sets of late bookings data: one based on average share of the passengers booking in the last seven days for flights on one particular day (787) and a second one based on comparison of load factors at the end of the month and 10 days before the end of the month for the last 60 months (788). The second set of data provides in general lower shares of late booking passengers than the first one. In its Response to the Statement of Objections, Ryanair used the second set of late bookings data for the analysis of individual routes and argued among others that where this share is below 10%, the route is a primarily leisure route (789). However, the second set of late booking data provided by Ryanair does not seem to be a good proxy for the actual level of the late bookings as it was rather intended to show the trend of late booking over the recent months. Indeed, when providing this second set of data, Ryanair itself indicated that it "provides a clear proxy of the trend of time-sensitive passengers using Ryanair who book ten days in advance of a Ryanair flight over a 60 month period and on an overlap route basis" (790). Further, as it provides the percentage of late bookers as a percentage of the total capacity (that is to say, all seats on the plane), it is by definition lower than the percentage of late bookers as a percentage of the total number of passengers (that is to say, only seats sold on the plane). As the average load factor (that is to say, percentage of seats sold) on Ryanair's planes is slightly above 80% on average over all routes (791), the actual percentage of passengers booking late as a percentage of total passengers would be higher than indicated in the second set of data (792). Therefore, while providing a good proxy for the trend, the second late bookings data underestimate the actual level of the late bookings. In addition, as these data are, unlike the first set, based on bookings 10 days before departure, they would be difficult to compare with Aer Lingus' data based on 7 days before departure (793). For all these reasons, the following analysis of individual routes uses the first set of late booking data as it provides a better (albeit not perfect) indication of the actual level of late booking passengers on individual routes.

789. The routes between Ireland and the United Kingdom are rather mature routes. According to Dublin Airport Authority, total traffic between Dublin and the United Kingdom grew in 2006 by 4% in comparison to the previous year, with just over 8.6 million passengers carried in 2006 (794). Further, both Ryanair and Aer Lingus are established carriers on these routes, having already offered their services for many years on most of the overlap routes, and thus having a very good brand recognition both in Ireland and in the United Kingdom.

790. In general, the merged entity would become very strong on the Ireland - United Kingdom routes. According to the information contained in the Aer Lingus IPO Prospectus, the Merging Parties represented in July 2006 in total around 78% of total capacity between Ireland and the United Kingdom (Ryanair - 54%, Aer Lingus - 24%) with the next most important competitor BMI with 5%, easyJet with 3% (who has however exited all routes to Ireland in October 2006) and British Airways with 2%. The fact that the Merging Parties are the leading carriers on the Ireland - United Kingdom routes is confirmed by Ryanair itself which states in the Form "20F" filed with the US Securities and Exchange Commission for the year 2006 that "... Ryanair had the leading market share (in terms of passenger volume) on most of its scheduled routes between Ireland and provincial cities in the U.K. and carried approximately 45% of all scheduled passenger traffic between Dublin and London, a share favourably comparable to the 33% share of Aer Lingus plc ("Aer Lingus"), its primary competitor on its U.K./Ireland routes." (795)

Dublin - London

791. The total number of passengers carried by scheduled airlines on the Dublin - London route in 2006 amounted to [4 250 000 - 4 300 000]* and, currently, a total of around 320 weekly rotations are operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 decreased by [0- 10%]* and was also without major changes in the years before. Therefore the route can be regarded as mature. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (796), as also acknowledged by Ryanair itself (797), and thus it can be excluded that charter operations would exert any meaningful competitive constraint on the merged entity (see also section 6.7.). Further, contrary to what Ryanair argues (798), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular yearround operations of the merged entity on this route.

792. On this route, Ryanair and Aer Lingus overlap on a city-to-city basis. In winter 2006/2007 season, Ryanair operated around 64 weekly rotations to Stansted airport (STN), 42 weekly rotations to Gatwick airport (LGW) and 23 weekly rotations to Luton airport (LTN) while Aer Lingus operated around 88 weekly rotations to Heathrow airport (LHR). These airports are considered as substitutable from the demand side (799). In the summer 2007 season, Ryanair increases rotations to Stansted to around 68 per week and keeps the rotations at Gatwick and Luton while Aer Lingus increases weekly rotations to around 90 at Heathrow. Apart from the Merging Parties, BMI operates a service between Dublin and Heathrow airport with around 47 weekly rotations, British Airways operates to Gatwick with around 27 and 26 weekly rotations in the winter 2006/2007 and summer 2007 seasons respectively and CityJet operates around 28 weekly rotations to London - City airport (LCY) both in the winter 2006/2007 and summer 2007 seasons (800).

793. The market shares of the Merging Parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in summer 2006 season are summarised in the following table. It must, however, be noted that both BMI and British Airways carry a share of passengers connecting at Heathrow and Gatwick respectively to their medium- and long-haul services. In contrast, Ryanair does not carry any connecting passengers while the share of Aer Lingus connecting passengers on this route (despite being important) (801) is still much lower than the share of point-to-point passengers. Therefore, in line with the definition of the relevant market as well as to better reflect the competitive situation in the market for flights between Dublin and London (and to distinguish it from, for example, flights between Dublin and New York, via London), it is appropriate in the case of this route to calculate the market share excluding connecting passengers as indicated in the last column of the following table.

<emplacement tableau>

794. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - London route from 5 to 4, with three much smaller remaining competitors. It would give the merged entity a very high combined market share, which is in itself evidence of a dominant position (802). Moreover, the transaction would combine Ryanair and Aer Lingus, the two closest competitors active on the Dublin - London Route (803).

795. The main competitor to the merged entity on this route is BMI with a market share of [10- 20%]*. Even though Ryanair operates to different airports in London from Aer Lingus and BMI operates to the same airport as Aer Lingus, there are substantial differences between the services offered by Aer Lingus and BMI and hence the constraint they put on Ryanair. BMI is a full-service network carrier which offers both economy and business class on the Dublin - London route. Further, BMI operates a network of medium and long-haul services out of Heathrow airport and uses the short-haul routes including Dublin - London to feed its hub at Heathrow, even though the majority of BMI passengers on this route are not connecting passengers. In contrast, Aer Lingus is a point-to-point low frills carrier operating a single class cabin on this route. Even though Aer Lingus also carries a share of connecting passengers between Dublin and Heathrow on the basis of its code-sharing agreement with British Airways, a large majority of its customers on this route are also point-to-point passengers804. Further, Aer Lingus operates almost twice as many rotations per week to London as BMI and carries twice as many passengers. Moreover, it is the only airline on the route which comes close to the total number of around 130 weekly rotations to London airports of Ryanair.

796. As regards British Airways, the same considerations apply as to BMI with the additional fact that British Airways fly to Gatwick airport where Ryanair also flies and that the frequencies offered by British Airways are even lower. The services offered by CityJet are further distinct from the other competitors as CityJet flies with smaller aircraft (805) to London - City airport. London City airport, with its location close to the city centre (806) is ideal in particular for business travellers (807). It has a short runway and is therefore unable to accommodate the large aircraft used by Ryanair and Aer Lingus. CityJet serves predominantly business customers, offering a full range of services. While both Ryanair (808) and Aer Lingus (809) serve an important number of business customers (to whom they are able to provide the highest frequencies of the existing carriers), they both have a majority of non-business, less time-sensitive customers (for which airport substitutability is also, in general, higher and which are thus prepared to switch between Ryanair and Aer Lingus services to different London airports).

797. Ryanair further argued that the independent MIDT/OAG data contained in the Form CO demonstrates the marked difference in aggregate fares between Ryanair and Aer Lingus on overlap routes, including Dublin - London. They claim that "even though those data sets were inaccurate in some respect as regards internet bookings, they would not be so inaccurate as to tell a different story" (810). Ryanair's fares data for Aer Lingus on overlap routes are mostly based on sources like MIDT, OAG and others (see paragraphs 107 and following of the Form CO) which mainly rely on data from the Global Booking System and are therefore, as acknowledged by Ryanair, inaccurate with respect to fares sold over Internet. In view of the fact that Aer Lingus sells more than 70% of tickets directly via Internet, it is doubtful whether these data can be relied on (not only in respect of this route, but also with respect to these fares data for all other overlap routes). This is also confirmed by the fact that on the basis of these data, an average fare for all the routes for which Ryanair provided these fare data (weighted by number of Aer Lingus passengers on each of these routes) would amount to around EUR 139 without taxes for the Summer 2005 season and to around EUR 157 without taxes for the Winter 2005/2006 season. In view of the fact that the routes covered represented around 70% of all Aer Lingus' passengers and that the total average short-haul fare of Aer Lingus without taxes in 2005 amounted to only EUR 66.81 (811), it is clear that these fare data grossly overstate the real average fare of Aer Lingus. Further, this is also confirmed by the comparison of these fare data with the actual average fare data provided by Aer Lingus for the purposes the Commission's quantitative analysis.

798. Therefore, in view of these inaccuracies in the data concerning the fares of Aer Lingus for individual routes provided by Ryanair in the Form CO, these data cannot be considered as reliable either for this route or for other overlap routes analysed in this Section. As regards this route, the inadequacy of these data is illustrated by the fact that according to them the average fare of Aer Lingus increased from EUR 130.56 in the Winter 2004/2005 season and EUR 108.84 in the Summer 2005 season to EUR 185.44 in the Winter 2005/2006 season (that is to say, by around 70%) without any evident justification (812). This is also not in line with the evaluation of the situation on this route during the Winter 2005/2006 season in Ryanair's internal documents (813).

799. Further, Ryanair argues that Aer Lingus targets a high share (over 60%) of time sensitive passengers on this route. This is based on a CAA 2005 survey according to which Aer Lingus carries 4.2% of business connecting passengers, 22.6% of leisure connecting passengers, 34.1% of business point-to-point passengers and 39.1% of leisure point-topoint passengers. As Ryanair considers all connecting as well as point-to-point business passengers as time sensitive, it claims that, overall, 60.9% of Aer Lingus' passengers on this route are time sensitive, compared to Ryanair's 20.1% time sensitive passengers at Gatwick, 43.6% at Stansted and 26.0 at Luton. (814).

800. However, as explained in Section 6, for the purpose of the assessment of the current transaction the relevant market concerns only scheduled passenger point-to-point air transport services and, thus, not connecting passengers, despite the possible importance from the point of view of the commercial strategy and focus of the carrier on the route. Further, it is not appropriate for the purpose of this case, to distinguish between time/nontime sensitive passengers for the reasons set out in Section 6.8. However, even if the distinction between business and leisure passengers on the basis of the data submitted by Ryanair in its Response to the Statement of Objections is taken into account, the share of business passengers of Aer Lingus as a percentage of their total point-to-point passengers would be 46.6%. Therefore, more than 50% of Aer Lingus' passengers on this route can be regarded as leisure passengers. The share of Ryanair's business passengers would amount to 30.8% at Stansted, 17.2% at Luton and 13.9% at Gatwick. The share of business passengers is thus also significant for Ryanair, amounting to around 24.4% on this route on average (weighted by passenger numbers to Gatwick, Stansted and Luton). On the other hand, the share of business passengers would be higher both for BMI (49.1%) and British Airways (55.1%). Therefore, even though Aer Lingus carries a higher share of business passengers, it still carries an even higher share of leisure passengers. This is also in line with the repositioning of Aer Lingus towards more low-frills model described in more details in sections 6.1. and 7.3.

801. Ryanair also argues that there is an "enormous number of passengers which EI [Aer Lingus] carries to Heathrow to connect with the rest of the world" (815) which makes Aer Lingus' focus very different from Ryanair's operations on the London route. However, the data submitted by Ryanair (816) show that in the Summer 2005 season Ryanair itself carried almost 15% "connecting (817)" passengers, that is to say around [160 000 - 180 000]* "connecting" passengers carried by Ryanair in the Summer 2005 season compared to around 250 000 connecting passengers of Aer Lingus (818). Despite this difference, it is not appropriate to consider that while Aer Lingus carries enormous number of connecting passengers on this route, the number of "connecting" passengers of Ryanair is small. In any event, the relevant market analysed is limited to point-to-point passengers and both Ryanair and Aer Lingus compete with each other with respect to a significant number of point-to-point passengers on the Dublin - London route.

802. Therefore, Ryanair and Aer Lingus are the only low-frills, predominantly point-to-point airlines active on the Dublin - London route. The fact that Ryanair and Aer Lingus are perceived by consumers as the closest competitors on this route was also strongly supported by the results of the Customer Survey commissioned by the Commission (819). Further, internal Ryanair documents submitted to the Commission by Ryanair also confirm that Aer Lingus is seen as the closest competitor on the Dublin - London route. [...]* (820), or: [...]* (821). This seems to be in clear contradiction with Ryanair's claim that "EI [Aer Lingus] services at Heathrow are unlikely to impact on FR [Ryanair] services at Luton, Stansted or Gatwick" (822) or that it is unlikely that Ryanair's services to/from Luton, Stansted or Gatwick constrain Aer Lingus services to/from Heathrow (823).

803. Therefore, Ryanair would, by the proposed transaction, acquire its closest rival on the Dublin - London route in terms of operating model, passenger focus as well as market share and capacity offered. Despite some remaining differentiation between Ryanair and Aer Lingus concerning, in particular, the airports served and in view of the continuing change of Aer Lingus towards the low-frills model operated by Ryanair (824), the proposed merger would lead to the elimination of the competitive rivalry between the two Merging Parties, which has been an important source of competition on this route to date and would be likely to even further intensify in the future.

804. The route is subject to the general barriers to entry described in Section 7.8. The Dublin - London route is an extremely busy route both as regards passenger numbers and the total number of frequencies operated by the Merging Parties. Therefore, any new entry would have to be of a substantial scale in terms of the operated capacity and frequencies to exert any competitive constraint on the merged entity. Ryanair, in its Response to the Statement of Objections argues that having a large number of frequencies is not a substantial competitive advantage due to the fact that fares are set on a one-way basis and that its passenger base is non-time sensitive point-to-point passengers. However, although slots and number of frequencies may not be important on all routes relevant for the purpose of the notified operation, on routes such as the London route they are. The merged entity would have more than 220 weekly rotations carrying more than 3 million passengers/year serving four London airports. Further, even though prices are set on a one way basis, around [60-80]*% of Ryanair's customers still book a return flight, that is to say both outbound and inbound flight together (825), and therefore the number of offered frequencies is of material importance on this route as well as on other thick and more business oriented routes analysed in this Section where any new entrant would need to provide sufficient competitive constraints on the merged entity by offering a comparable choice of departure and arrival times.

805. It is important for its competitiveness that any airline (including Ryanair and Aer Lingus) active on such routes attracts a certain proportion of business or more time-sensitive passengers (for the importance of business passengers on this route see paragraph 799 and following) by offering flights at all peak times of the day. As such customers in general book their tickets closer to departure they normally pay higher prices (826), thus increasing the average yield on the flight and are important for maintaining competitive operations. Further, a higher share of business customers would then, in general, also enable the carrier to offer lower fares for the passengers booking sufficiently far in advance (while keeping the overall average yield and load factor) and thus also compete more effectively with respect to the relatively more price-sensitive passengers and achieve sufficient load factor. On the routes where the share of the business passengers is significant (such as Dublin - London), it is thus more important for carriers to attract a significant share of these higher yield passengers to be competitive vis-à-vis other carriers than on routes where business traffic share is low. For this reason, the limited capacity of Dublin Airport during peak hours as described in more detail in Section 7.8., as well as of the fully coordinated airports of Heathrow (heavily congested throughout the day), Gatwick (congested throughout the day) and Stansted (congested at peak times) constitutes a significant barrier to entry to this route (827). Even though the Luton and City airports are not congested, the availability of capacity at these airports does not eliminate the additional barriers to entry to this route. As described in greater detail in paragraph 796, London City airport serves predominantly business passengers flying in smaller aircraft and cannot thus be used for a high volume entry covering the whole spectrum of passengers served by Ryanair and Aer Lingus. Further, both London City and Luton are, because of supply side considerations, not an alternative to the congested London airports (and in particular Heathrow) for a number of potential entrants, in particular due to their existing bases/hubs and network activities (828).

806. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) would be able to easily enter this route if the merged entity were to increase prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at the London airports: British Airways with 87 short-haul aircraft based in Heathrow, 33 short-haul aircraft based in Gatwick and 10 regional jets based in London City; easyJet with 18 aircraft based in Gatwick, 13 in Stansted and 17 in Luton; BMI with 14 aircraft based in Heathrow; Globespan with 1 aircraft based in Stansted. Further, even though Air Berlin does not have any aircraft based in London, it uses Stansted as a hub for its flights to some United Kingdom destinations and a number of holiday routes.

807. However, on the basis of the Phase II investigation none of these carriers appear to have sufficient abilities and/or incentives to enter or significantly expand their activities on this route, in the event of a price increase (829) by the merged entity. Globespan is a low-frills, as well as a charter airline operating mainly between the UK cities and holiday destinations in the Mediterranean, Switzerland and the Czech Republic together with some long-haul services (830). In view of its focus on holiday and long-haul traffic, Globespan is not a likely entrant into the Dublin - London route. Some network carriers with operations in Heathrow might, in theory, be able and potentially willing to enter or expand their activities between Dublin and London, but are not able to do so, in particular due to the severe capacity constraints at Heathrow as well as the peak time capacity scarcity at Dublin. Further, it is doubtful that traditional network carriers would constitute a significant competitive constraint to the low-frills, pointto- point operations of the combined entity, since they have a significantly different business model focusing on feeding their hub, mainly targeting the higher yield customer segment. Also their cost structure may not allow them to offer an alterative for classic "low-frills" customers comparable to Ryanair or Aer Lingus to any significant extent (831). As regards easyJet, it is also (apart from the peak time congestion at Dublin) partially constrained by the capacity situation in Gatwick and at peak times at Stansted (even though their third base at Luton is not constrained, the overall situation in London lowers the opportunities for an efficient entry with sufficient number of frequencies). Further, in view of the unsuccessful entry by easyJet to the London routes to regional Irish airports and in particular the reaction of Ryanair to this entry (for more details see in Section 7.8.), it is doubtful whether easyJet or any other new carrier would be willing and able to challenge the (even stronger) merged entity on the London - Ireland routes in the short to mid-term. In this regard, see Section 7.8.9 for more details about the abilities and incentives of easyJet to enter Irish routes. As regards Air Berlin, it is doubtful whether it can successfully apply its current "W"-model (832) used via London Stansted for some UK destinations for entering the Dublin - London route (833). This model would not allow Air Berlin to enter with a sufficient number of frequencies to constitute a credible competitor and would not make it possible to operate the important early morning and later evening frequencies. Further, neither easyJet nor Air Berlin has any existing operations in Ireland and entering the Dublin market would require substantial investments into brand awareness and marketing. As around 50% of passengers on this route originate in Ireland (834), it is still important for a successful new entry to attract a significant share of Irish originating customers. It should also be noted that with a total of 40 Ryanair aircraft based in Stansted and Luton, the merged entity would have the advantage of having bases at both sides of this route, with resulting greater flexibility and cost benefits. Further, Dublin - London is a mature route, making entry by a new carrier with additional capacities even more demanding. Finally, there are no examples of (even only attempted) new entry to this route in the recent years by any other scheduled carrier.

808. As explained in detail in Sections 7.3. and 7.8., the remaining competing carrier based in Dublin Airport (Aer Arann) cannot be considered a likely potential entrant to this route which would be able to exert sufficient competitive constraints on the merged entity (835). Further, the market investigation did not identify any likely potential new entrant currently not based in either Dublin or London for this route.

809. Therefore, the results of the market investigation demonstrate that the possibility of entry or expansion of capacity by the existing competitors is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

810. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - London route.

Dublin - Birmingham

811. The total number of passengers carried by scheduled airlines on the Dublin - Birmingham route in 2006 amounted to [800 000 - 850 000]*, and, in total, around 65 weekly rotations are currently operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006, compared to the situation in the year 2005, increased by [10- 20%]*. However there is no clear growth trend as the total number of passengers between 2004/2005 decreased by [10-20%]*. Overall, the route can be considered as mature. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (836), as also acknowledged by Ryanair itself (837), and thus it can be excluded that charter operations would exert any meaningful competitive constraint on the merged entity (see also section 6.7.). Further, contrary to what Ryanair argues (838), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

812. On this route, Ryanair and Aer Lingus overlap on both an airport-to-airport and a city-tocity basis. In the winter 2006/2007 season Ryanair operated around 17 weekly rotations to Birmingham International airport (BHX) as well as around 21 weekly rotations to East Midlands airport (EMA) and Aer Lingus operated around 27 weekly rotations to Birmingham International airport. These airports are considered as substitutable from the demand side (839). The planned rotations for the summer 2007 season remain the same for Ryanair and decrease to around 21 rotations per week for Aer Lingus. Apart from the parties, there are no other competitors active on this route. As regards the past examples of entry on this route, MyTravelLite entered the route Dublin - Birmingham route in September 2003 but exited after only one year of operation in October 2004. Further, BMIBaby exited the Dublin - Birmingham route (serving East Midlands airport) from March 2005 (around one year after Ryanair started its Dublin - East Midlands flights (840)) after having operated on this route from 2002 when it replaced the traditional operations by BMI on this route. BMI itself acknowledges that the market entry of Ryanair onto the Birmingham - Dublin route prompted BMIBaby to withdraw its services (841).

813. The market shares of the parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in summer 2006 season are summarised in the following table (842).

<emplacement tableau>

814. Therefore, the proposed transaction would lead to a reduction in the number of carriers active in the Dublin - Birmingham route from 2 to 1 and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (843).

815. The route is subject to the general barriers to entry described in Section 7.8.. As described in Section 7.8.5., there is a past example of unsuccessful entry by MyTravelLite in 2003/2004 against Ryanair and Aer Lingus on this route and there has been no attempted entry since then. Further, the Dublin - Birmingham route is a busy route both as regards the passenger numbers and the total number of frequencies operated by the Merging Parties (together, they would have around 9 rotations per day). Therefore, any new entry would have to be of a substantial scale in terms of both the operated capacity and frequencies to exert a competitive constraint on the merged entity (844). It is important for its competitiveness that any airline active on this route attracts a certain proportion of business or more time-sensitive passengers (845) by offering flights at all the peak times of the day. For this reason, the limited capacity of Dublin Airport in the peak times, described in more detail in Section 7.8., constitutes an additional significant barrier to new entry. Further, the Dublin - Birmingham route is a mature route making entry by a new carrier with additional capacities even more demanding.

816. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport would easily be able to enter this route if the merged entity were to increases price and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Birmingham and East Midlands airports: BA Connect with 15 based aircraft (4 aircraft with 110 seats and 11 aircraft with 49 seats), easyJet with 3 based aircraft, Flybe with 5 based aircraft, BMI/BMIBaby with 5-8 aircraft, Monarch with 2 based aircraft and Thomsonfly with 3 based aircraft. However, on the basis of the Phase II investigation none of these carriers appear to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in the event of a price increase by the merged entity. None of these carriers confirmed that they would be ready to enter the Dublin - Birmingham route in the near future. Further, as described in more detail in Section 7.8, most of these carriers have a different strategic focus of their activities and consider other routes as more profitable and less risky alternatives for utilisation of their aircraft. Flybe and BA Connect are regional UK airlines operating usually smaller aircraft which would not represent effective competition to Ryanair and Aer Lingus' higher volume operations on this route. Monarch and Thomsonfly are low-frills as well as charter airlines operating mainly between the UK cities and holiday destinations predominantly in the Mediterranean. As explained in more detail in Section 7.8, easyJet and BMI/BMIBaby can also not be considered as credible potential entrants on this route in the short to medium term. Further, many of these carriers (with the exception of bmi on the Dublin - London route as well as four routes to regional Irish airports operated by bmibaby and Flybe's 6 routes to Dublin and Galway - see section 7.8) have no existing operations in Ireland and entering the Dublin market would require substantial investments into brand awareness and marketing. As between 30% and 40% of passengers on this route originate in Ireland (846), it is less important for a successful new entry to attract Irish originating customers. However, as Ryanair has a strong presence (and indeed even a base) at both ends of this route, any new entrant not yet present in Ireland would potentially still be in a more disadvantageous position in comparison to Ryanair. It should also be noted that with 2 Ryanair aircraft based in East Midlands, the merged entity would have the advantage of having bases at both sides of this route, with resulting greater flexibility and cost benefits. Finally, there are no examples of a sustained entry to this route in recent years apart from Ryanair's expansion to East Midlands in April 2004. To the contrary, the examples of an unsuccessful entry by MyTravelLite and exit by BMIBaby prompted by Ryanair's expansion on this route demonstrate that entry against Ryanair and Aer Lingus on this route is not easy.

817. It could be argued that CityJet would potentially be able to enter the Dublin - Birmingham route as it has aircraft based in Dublin and also overnights aircraft at Birmingham and this route, with significant business traffic, could potentially be in line with its business oriented model (847). However, the potential focus of CityJet on this route would be business passengers served by its 93 seat aircraft (848). For this reason, CityJet would need to operate a high frequency service with all the peak times covered to be able to compete with Ryanair/Aer Lingus. Due to the slot constraints at peak times at Dublin Airport, the entry of CityJet into the Dublin - Birmingham route cannot be considered likely. In order to enter such a route with focus on business customers, CityJet would need the peak slots as well as access to other infrastructure such as contact stands (849). As explained in detail in Sections 7.3. and 7.8, the remaining competing carrier based in Dublin Airport (Aer Arann) is not a likely potential entrant on this route. Further, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Birmingham/East Midlands. Ryanair argues that airlines such as Air Malta, Hapag Lloyd Express, KLM and Lufthansa, which operate to Birmingham from other destinations and also already operate in Dublin (with sales in Ireland), could expand or open new routes (850). However, it is unlikely that they would be able to provide any meaningful competitive constraint to the merged entity on the Dublin - Birmingham route as they do not base aircraft either in Dublin or in Birmingham and as entry to a route without a base at either side is in general less efficient and as such rather uncommon (see also Section 7.8.). Further, these carriers are rather full-service network carriers with business models in most cases substantially different to Ryanair and Aer Lingus which would be unlikely to significantly constrain the merged entity.

818. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

819. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Birmingham route.

Dublin - Edinburgh

820. The total number of passengers carried by scheduled airlines on the Dublin - Edinburgh route in 2006 amounted to [400 000 - 450 000]* and in total around 36 weekly frequencies are currently operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006, compared to the situation in the year 2005, increased by [10- 20%]*. However, there is no clear growth trend as the growth rate for 2004/2005 was only around [0-10%]* and the route thus can be considered as mature. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (851), as also acknowledged by Ryanair itself (852), and thus it can be excluded that charter operations would exert any meaningful competitive constraint on the merged entity (see also section 6.7.). Further, contrary to what Ryanair argues (853), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

821. On this route, Ryanair and Aer Lingus overlap on an airport-to-airport basis. In the winter 2006/2007 season Ryanair operated around 23 weekly rotations to Edinburgh airport (EDI) and Aer Lingus operated around 13 weekly rotations to Edinburgh airport. The planned rotations for the summer 2007 season are around 21 per week for Ryanair and 7 per week for Aer Lingus. Apart from the parties, there are no other competitors active on this route. As regards the past examples of entry in this route, Go Fly entered the route in September 2001 but exited already in March 2002 after only around 6 months of operation (854).

822. The market shares of the parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (855).

<emplacement tableau>

823. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Edinburgh route from 2 to 1 and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (856).

824. The route is subject to the general barriers to entry described in Section 7.8.. As described in Section 7.8.5., there is a past example of unsuccessful entry by Go Fly in 2001/2002 against Ryanair and Aer Lingus on this route and there has been no attempted entry since then. Further, the Dublin - Edinburgh route is a rather busy route both as regards the passenger numbers and the total number of frequencies operated by the Merging Parties (together, they would have around 5 rotations per day). Therefore, any new entry would have to be of a substantial scale in terms of the operated capacity and frequencies to exert a competitive constraint on the merged entity. It is important for its competitiveness that any airline active on this route attracts a certain proportion of business or more time-sensitive passengers (857) by offering flights at all peak times of the day. For this reason, the limited capacity of Dublin Airport in peak times, described in more detail in Section 7.8., constitutes an additional significant barrier to new entry. Further, Dublin - Edinburgh route is a mature route making entry by a new carrier with additional capacities even more demanding.

825. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport would easily be able to enter this route if the merged entity were to increase prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Edinburgh airport: BA Connect with 3 based aircraft, easyJet with 3 based aircraft, Flybe with 3 based aircraft, Loganair with 1-2 based aircraft and Globespan with 4 based aircraft (858). On the basis of the Phase II investigation none of these carriers appear to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in the event of a price increase by the merged entity. Further, as described in more detail in Section 7.8, most of these carriers have a different strategic focus of their activities and consider other routes as more profitable and less risky alternatives for utilisation of their aircraft. Flybe, BA Connect and Loganair are regional UK airlines operating usually smaller aircraft which would not represent an effective competition to Ryanair and Aer Lingus' higher volume operations on this route. Flybe/BA Connect also indicated that they have no plans to expand their operations to/from Dublin and that their focus on expansion is on the United Kingdom to continental Europe routes (859). Globespan is a low-frills as well as charter airline operating mainly between the UK cities and holiday destinations in the Mediterranean, Switzerland and the Czech Republic together with some long-haul services. Even though it offers one long-haul service from Dublin to Toronto, it has no short-haul operations in Dublin and its reply indicates that the strength of the merged entity would constitute a barrier preventing it from providing a competitive service on the affected routes (860). As explained in more detail in Section 7.8, easyJet can also not be considered as a credible potential entrant on this route in the short to medium term. Further, none of these carriers (861) have any existing operations in Ireland and entering the Dublin market would require substantial investments into brand awareness and marketing. As almost 50% of passengers on this route originate in Ireland (862), it is essential for a successful new entry to attract a significant share of Irish originating customers. Ryanair argues that the combined share of Ryanair and Aer Lingus of the capacity at Edinburgh only amounts to 4% (863). However, this only indicates that there are other significant carriers operating any flights to/from Edinburgh but has no direct relevance for the assessment of the Dublin - Edinburgh route to the extent that the market investigation did not suggest that entry of any of these other carriers would be sufficiently likely and timely (864).

826. It could be argued that CityJet would potentially be able to enter the Dublin - Edinburgh route as it has aircraft based in Dublin and also overnights aircraft at Edinburgh and this route, with significant business traffic, could potentially be in line with its business oriented model (865). However, the potential focus of CityJet on this route would be business passengers served by its 93-seats aircraft. For that reason, CityJet would need to operate a high frequency service with all peak times covered to be able to compete with Ryanair/Aer Lingus. Due to the slot constraints at peak times at Dublin Airport, the entry of CityJet into the Dublin - Edinburgh route cannot be considered likely. In order to enter such a route with a focus on business customers, CityJet would need the peak slots as well as access to other infrastructure such as contact stands (866). As explained in detail in Sections 7.3. and 7.8, the remaining competing carrier based in Dublin Airport (Aer Arann) is not a likely potential entrant on this route. Even though Ryanair argues that Aer Arann is one of several strong credible entrants (867), this was not confirmed by the Commission's investigation and Aer Arann clearly states that they "would not enter any route where Ryanair is already present, even if Ryanair's prices were to increase after a merger between Ryanair and Aer Lingus..." (868) Ryanair also argues that Aer Arann has a history of following Ryanair but provides no evidence to this effect. On the contrary, as indicated in Section 7.8, the Commission has found no instances of entry by Aer Arann onto a route already operated by Ryanair in recent years. Also the regression analysis conducted by the Commission indicates that Aer Arann is not likely to exert any significant competitive constraints on the pricing of Ryanair and Aer Lingus (869). Further, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Edinburgh.

827. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

828. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Edinburgh route.

Dublin - Glasgow

829. The total number of passengers carried by scheduled airlines on the Dublin - Glasgow route in 2006 amounted to [400 000 - 450 000]* and in total around 49 weekly rotations are currently operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006, compared to the situation in the year 2005, increased by [0- 10%]* and remained at around the same level in the previous years. The route can thus be considered as mature. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (870), as also acknowledged by Ryanair itself (871), and thus it can be excluded that charter operations would exert any meaningful competitive constraint on the merged entity (see also section 6.7.). Further, contrary to what Ryanair argues (872), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

830. On this route, Ryanair and Aer Lingus overlap on a city-to-city basis. In the winter 2006/2007 season Ryanair operated around 17 weekly rotations to Prestwick airport (PIK) and Aer Lingus operated around 18 weekly rotations to Glasgow International airport (GLA). These airports are considered as substitutable from the demand side (873). The planned rotations for the summer 2007 season are around 19 for Ryanair and around 13 for Aer Lingus. Apart from the parties, only Loganair operates around 14 weekly rotations between Dublin and Glasgow, however not directly but with a stopover in Londonderry (LDY) (874). As regards the past examples of entry in this route, Go Fly entered the route in September 2001 but exited already in March 2002 after only around 6 months of Operation (875).

831. The market shares of the parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (876). It should be noted, however, that due to the fact that the operations by Loganair are indirect and some passengers fly only to/from Londonderry, the market share on the basis of capacity overstates the importance of Loganair on the Dublin - Glasgow route. Thus the market shares on the basis of passengers actually carried reflects better the situation on this route.

<emplacement tableau>

832. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Glasgow route from 3 to 2 and would give the merged entity a [90- 100]*% combined market share, coming close to a monopoly on this route. The very high combined market share of Ryanair and Aer Lingus on this route is in itself evidence of a dominant market position (877). Moreover, the transaction would combine Ryanair and Aer Lingus, the two closest competitors active on the route (878). The indirect (879) small-volume services operated by Loganair cannot be considered comparable to the significant operations by Ryanair and Aer Lingus post merger. Therefore, the merger would also lead to the elimination of the competitive rivalry between the Merging Parties' which has been an important source of competition on this route.

833. The route is subject to the general barriers to entry described in Section 7.8.. As described in Section 7.8.5., there is a past example of unsuccessful entry by Go Fly in 2001/2002 against Ryanair and Aer Lingus on this route and there has been no attempted entry since then. Further, the Dublin - Glasgow route is a busy route both as regards passenger numbers and the total number of frequencies operated by the Merging Parties (together, they would have around 7 rotations per day). Therefore, any new entry would have to be of a substantial scale in terms of the operated capacity and frequencies to exert a competitive constraint on the merged entity. It is important for its competitiveness that any airline active on this route attracts a certain proportion of business or more time-sensitive passengers (880) by offering flights at all peak times of the day. For this reason, the limited capacity of Dublin Airport in peak times, described in more detail in Section 7.8., constitutes an additional significant barrier to new entry. Further, the Dublin - Glasgow route is a mature route making entry by a new carrier with additional capacities even more demanding.

834. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport would easily be able to enter this route if the merged entity were to increase prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Glasgow and Prestwick airports: BA Connect with up to 4 based aircraft, easyJet with 4 based aircraft, Flybe with up to 3 based aircraft, BMI/BMIBaby with 2 aircraft, Loganair with 3-7 based aircraft and Globespan with 4 based aircraft. On the basis of the Phase II investigation none of these carriers appear to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in the event of a price increase by the merged entity. In particular, none of these carriers confirmed that they would be ready to enter the Dublin - Glasgow route in the near future even if the combined entity were to increase prices. Further, as described in more detail in Section 7.8, most of these carriers have a different strategic focus of their activities and consider other routes as more profitable and less risky alternatives for utilisation of their aircraft. Flybe, BA Connect and Loganair are regional UK airlines operating usually smaller aircraft which would not represent any effective competition to the merged entity's higher volume operations on this route. Globespan is a low-frills as well as charter airline operating mainly between the UK cities and holiday destinations in the Mediterranean, Switzerland and the Czech Republic together with some long-haul services. In view of its focus on holiday and long-haul traffic, Globespan is not a likely entrant into the Dublin - Glasgow route. As explained in more detail in Section 7.8, easyJet (881) and BMI/BMIBaby can also not be considered as credible potential entrants on this route in the short to medium term. Further, none of these carriers based in Glasgow or Prestwick and not yet active on the route (with the exception of BMI on the Dublin - London route, which is however specific as discussed in the relevant section) have any existing operations in Dublin and entering the Dublin market would require substantial investments into brand awareness and marketing. As around 50% of passengers on this route originate in Ireland (882), it is essential for a successful new entry to attract a significant share of Irish originating customers. It should also be noted that with 5 Ryanair aircraft based in Prestwick, the merged entity would have the advantage of having bases at both sides of this route, with resulting greater flexibility and cost benefits. As explained in detail in Section 7.8, the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Further, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Glasgow/Prestwick.

835. Even though Loganair already has operations between Dublin and Glasgow, it is not likely that it would have the ability and incentive to expand these operations in such a way that it would provide sufficient competitive constraint to the merged entity. First, Loganair operations are not direct and are served by substantially smaller planes (with 34 seats) than those operated by Ryanair (189 seats) and Aer Lingus (174 seats). Further, the Dublin - Londonderry route is operated by Loganair under a Public Service Obligation (PSO) contract tendered by the Irish Government for routes which cannot be operated commercially but which are important for public interest. The government subsidises these operations and also decides the schedule and maximum fare. Further, Loganair's primary business is connecting the regional Scottish airports with the network of its franchise partner British Airways and therefore Dublin operations are not part of its core business.

836. Therefore, the results of the market investigation demonstrate that the possibility of entry or expansion of capacity by the existing competitors is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

837. Taking into account all the above, the Commission concludes that the transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Glasgow route.

Dublin - Manchester

838. The total number of passengers carried by scheduled airlines on the Dublin - Manchester route in 2006 amounted to [1 250 000 - 1 300 000]* and in total around 95 weekly rotations are currently operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006, compared to the situation in the year 2005, increased by [0-10%]* with similar or lower growth in the preceding years. Overall, the route can be considered as mature. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (883), as also acknowledged by Ryanair itself (884), and thus it can be excluded that charter operations would exert any meaningful competitive constraint on the merged entity (see also section 6.7.). Further, contrary to what Ryanair argues (885), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

839. On this route, Ryanair and Aer Lingus overlap on both an airport-to-airport and a city-tocity basis. In the winter 2006/2007 season Ryanair operated around 24 weekly rotations to Manchester airport (MAN), 18 weekly rotations to Liverpool airport (LPL) as well as 19 weekly rotations to Leeds Bradford airport (LBA) and Aer Lingus operated around 27 weekly rotations to Manchester airport. These airports are considered as substitutable from the demand side (886). The planned rotations for the summer 2007 season for Ryanair are around 26, 21 and 19 weekly rotations for Manchester, Liverpool and Leeds Bradford respectively. Aer Lingus plans around 24 weekly rotations in the same period. Apart from the parties, only Luxair operates limited services between Manchester and Dublin with up to 7 weekly rotations (as part of the flights from Luxembourg to Dublin via Manchester and back). As of October 2004, Aer Lingus also operated on this route between Dublin and Liverpool airport but exited Liverpool airport in June 2006 and currently only operates to Manchester airport. There are no examples of other new entry to this route in recent years.

840. The market shares of the parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (887). It should be noted, however, that due to the fact that the operations by Luxair are stop-over flights from Luxembourg via Manchester and some passengers fly between Dublin and Luxembourg, the market share on the basis of capacity overstates the importance of Luxair on the Dublin - Manchester route. Thus the market shares on the basis of passengers actually carried reflects better the situation on this route.

<emplacement tableau>

841. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Manchester route from 3 to 2 and would give the merged entity almost 100% combined market share. The very high combined market share of Ryanair and Aer Lingus on this route is in itself evidence of a dominant market position (888). Moreover, the transaction would combine Ryanair and Aer Lingus, the two closest competitors active on the route (889) as the small-volume services operated by Luxair are not comparable to the significant operations by Ryanair and Aer Lingus. Further, Luxair's services on this route are focused on business customers (890) and are provided by significantly smaller aircraft (891). Therefore, the merger would also lead to the elimination of the competitive rivalry between the Merging Parties, which has been an important source of competition on this route (892).

842. The route is subject to the general barriers to entry described in Section 7.8. Further, the Dublin - Manchester route is a busy route both as regards passenger numbers and the total number of frequencies operated by the Merging Parties (together, they would have around 12 rotations per day). Therefore, any new entry would have to be of a substantial scale in terms of the operated capacity and frequencies to exert a competitive constraint on the merged entity. It is important for its competitiveness that any airline active on this route attracts a certain proportion of business or more time-sensitive passengers (893) by offering flights at all peak times of the day (894). For this reason, the limited capacity of Dublin Airport in peak times, described in more detail in Section 7.8., as well as the peak time congestion at the fully coordinated Manchester airport (895), constitutes additional barriers to new entry. Ryanair argues that the merged entity would be unlikely to retain all the early morning slots it currently operates and thus that a new entrant from Manchester would be able to exploit that gap in the market (896). Even presuming that the parties may give up some slots at this airport (the Commission has not been provided with any concrete plans to this effect), Manchester is a slot co-ordinated airport and it is difficult to predict which airline would be granted those particular slots if given up by Ryanair/Aer Lingus. Further, it is not clear why these early morning slots would be used by another carrier specifically for this route and no other route to/from Manchester (897). In addition, the Dublin - Manchester route can be considered a mature route making entry by a new carrier with additional capacities even more demanding.

843. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport would easily be able to enter this route if the merged entity were to increase prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Manchester, Liverpool and Leeds Bradford airports: BA Connect with 16 (898) based aircraft, easyJet with 8 based aircraft, Jet2 with 16 based aircraft, BMI/BMIBaby with 8 based aircraft and Monarch with 3 based aircraft. However, on the basis of the Phase II investigation none of these carriers appear to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in the event of a price increase by the merged entity. None of these carriers confirmed that they would be ready to enter the Dublin - Manchester route in the near future even if the combined entity were to increase prices. Further, as described in more detail in Section 7.8, most of these carriers have a different strategic focus of their activities and consider other routes as more profitable and less risky alternatives for utilisation of their aircraft. BA Connect is a regional UK airline operating smaller aircraft which would not represent effective competition to Ryanair and Aer Lingus' higher volume operations on this route. Monarch is a low-frills as well as charter airline operating mainly between the UK cities and holiday destinations predominantly in the Mediterranean. Similarly, Jet2 is a low-frills airline operating mainly between the UK cities and holiday destinations. Even though it operates some services to major European cities and also to Cork (from Newcastle (899)), it is focused principally on UK customers and would not be willing to expand in the Irish market where both Ryanair and Aer Lingus are well known established companies offering a large number of destinations (900). As explained in more detail in Section 7.8, easyJet and BMI/BMIBaby can also not be considered as credible potential entrants on this route in the short to medium term. Further, none of these carriers (with the exception of BMI on the Dublin - London route) have any existing operations in Dublin and entering the Dublin market would require substantial investments into brand awareness and marketing. As close to 50% of passengers on this route originate in Ireland (901), it is still important for a successful new entry to attract a significant share of Irish originating customers. It should also be noted that with 7 Ryanair aircraft based in Liverpool, the merged entity would have the advantage of having bases at both sides of this route, with resulting greater flexibility and cost benefits. Finally, there are no examples of a successful and sustained new entry to this route in recent years by any other scheduled carrier.

844. It could be argued that CityJet would potentially be able to enter the Dublin - Manchester route as it has aircraft based in Dublin and is a 100% subsidiary of Air France with substantial operations to/from Manchester and this route, with significant business traffic, could potentially be in line with their business oriented model (902). However, the potential focus of CityJet on this route would be business passengers served by its 93 seat aircraft. For that reason, CityJet would need to operate a high frequency service with all peak times covered to be able to compete with Ryanair/Aer Lingus. Due to the slot constraints at peak times both in Dublin and in Manchester, the entry of CityJet into the Dublin - Manchester route cannot be considered likely. In order to enter such a route with a focus on business customers, CityJet would need the peak slots as well as access to other infrastructure such as contact stands (903). As explained in detail in Sections 7.3. and 7.8, the remaining competing carrier based in Dublin Airport (Aer Arann) is not a likely potential entrant on this route. Even though Ryanair argues that Aer Arann already has flights between Manchester and Galway, Kerry and Waterford, these are not operated in competition with any of the parties and the market investigation did not indicate that Aer Arann would be likely to enter the Dublin - Manchester route and represent any significant competitive constraint on the merged entity (see Section 7.8). Also the regression analysis conducted by the Commission indicates that Aer Arann is not likely to exert any significant competitive constraints on the pricing of Ryanair and Aer Lingus (904). The market investigation did not identify any other likely new entrant for this route not based in either Dublin or Manchester/Liverpool/Leeds Bradford.

845. Therefore, the results of the market investigation demonstrate that the possibility of entry or expansion of capacity by the existing competitors is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

846. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Manchester route.

Dublin - Newcastle

847. The total number of passengers carried by scheduled airlines on the Dublin - Newcastle route in 2006 amounted to [250 000 - 300 000]* and in total around 26 weekly rotations are currently operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 increased by [20- 30%]*, with a relatively stable passenger numbers in the previous years. The high increase in 2006 could be possibly attributed to entry of Aer Lingus to the route in October 2006 which generated some additional demand. However, overall the route can be considered as rather mature. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (905), as also acknowledged by Ryanair itself (906), and thus it can be excluded that charter operations would exert any meaningful competitive constraint on the merged entity (see also section 6.7.). Further, contrary to what Ryanair argues (907), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular yearround operations of the merged entity on this route.

848. On this route, Ryanair and Aer Lingus overlap on both an airport-to-airport and a city-tocity basis. In the winter 2006/2007 season Ryanair operated around 14 weekly rotations to Newcastle airport (NCL) as well as around 5 weekly rotations to Durham Tees Valley airport (MME) and Aer Lingus operated around 7 weekly rotations to Newcastle airport (908). These airports are considered as substitutable from the demand side (909). The planned rotations for the summer 2007 season are the same for both Ryanair and Aer Lingus. Apart from the parties, there are no other competitors active on this route. As regards entry in recent years, Aer Lingus re-entered the route as of October 2006 after having exited the route in March 2001. Further, in March 2001 Brymon Airways (owned by British Airways) launched Dublin - Newcastle flights in competition with, at that time, only Durham Tees Valley flights of Ryanair. These flights were operated under Brymon Airways and subsequently under BA City Express (currently BA Connect) names until January 2003 when they exited the route. It is worth noting that Ryanair started operations between Dublin and Newcastle airport also in January 2003. Further, the UK regional airline Gill Airways started operations between Dublin and Newcastle in January 2001 but ceased its flights and went bankrupt in September 2001 (910). There are no examples of other new entry to this route in the recent years.

849. The market shares of the parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (911). As Aer Lingus started operations on this route only in October 2006, Ryanair was the only carrier active on this route in the summer 2006 season.

<emplacement tableau>

850. Therefore, the proposed transaction would lead to a reduction in the number of carriers active in the Dublin - Newcastle route from 2 to 1 and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (912).

851. The route is subject to the general barriers to entry described in Section 7.8. Further, the Dublin - Newcastle route is a busy route both as regards passenger numbers and the total number of frequencies operated by the Merging Parties (together, they would have almost 4 rotations per day). Therefore, any new entry would have to be of a substantial scale in terms of the operated capacity and frequencies to exert a competitive constraint on the merged entity. It is important for its competitiveness that any airline active on this route attracts a certain proportion of business or more time-sensitive passengers (913) by offering flights at all peak times of the day. For this reason, the limited capacity of Dublin Airport in peak times, described in more detail in Section 7.8., constitutes an additional significant barrier to new entry. Further, Dublin - Newcastle route is a rather mature route with a new entrant only as of October 2006 (Aer Lingus), making entry by a further new carrier with additional capacities even more demanding.

852. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport would easily be able to enter this route if the merged entity were to increase prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Newcastle and Durham Tees Valley airports: easyJet with 6 based aircraft and Jet2 with 3 based aircraft. On the basis of the Phase II investigation none of these carriers appear to have sufficient abilities and/or incentives to enter this route and provide a sufficient competitive constraint on the merged entity in case of a price increase. In particular, none of these carriers confirmed that they would be ready to enter the Dublin - Newcastle route in the near future even in the event that the combined entity were to increase prices. Further, as described in more detail in Section 7.8, most of these carriers have a different strategic focus of their activities and consider other routes as more profitable and less risky alternatives for utilisation of their aircraft. Jet2 is a low-frills airline operating mainly between the UK cities and holiday destinations. Even though it operates some services to major European cities and also to Cork (from Newcastle (914)), it is focused principally on the UK customers and would not be willing to expand in the Irish market where both Ryanair and Aer Lingus are well known established companies offering a large number of destinations (915). As explained in more detail in Section 7.8., easyJet can also not be considered as a credible potential entrant on this route in the short to medium term. Further, none of these carriers have any existing operations in Dublin (only Jet2 has some limited operations on one regional route to Ireland) and entering the Dublin market would require substantial investments into brand awareness and marketing. As between 40-50% of passengers on this route originate in Ireland (916), it is less important for any successful new entry to attract a significant share of Irish originating customers. Ryanair argues that as less than half the passengers on this route originate in Ireland and as Jet2 and easyJet both enjoy significant brand recognition at the other end of the route, they could easily compete with Ryanair on that basis. However, this fails to take into account the fact that Ryanair has a wellestablished brand not only in Ireland but also in the United Kingdom (917). Even though Ryanair argues that Aer Arann could enter this route (918), as explained in detail in Section 7.8, neither Aer Arann (nor the other competing carrier based in Dublin Airport- CityJet) are likely potential entrants to this route who would be able to provide significant competitive constraints to the merged entity. In particular, even if they were to enter, their current business model (smaller regional aircraft and for CityJet focus on high yield business customers) does not allow them to provide any significant constraint on the merged entity. Also the regression analysis conducted by the Commission indicates that Aer Arann and CityJet are not likely to exert any significant competitive constraints on the pricing of Ryanair and Aer Lingus (919). Finally, there are no examples of a successful and sustained new entry to this route in recent years by any other scheduled carrier.

853. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

854. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Newcastle route.

Shannon - London

855. The total number of passengers carried by scheduled airlines on the Shannon - London route in 2006 amounted to [850 000 - 900 000]* and in total around 50 weekly rotations are currently operated on this route by scheduled carriers. The total number of passengers carried on this route decreased slightly (by less than [0-10%]*) in 2006, compared to the situation in the year 2005. There was a significant increase of the number of passengers in 2005, which could be attributed to the entry of easyJet and the subsequent price war described in more detail in Section 7.8.5. and paragraph 857. In the years before, the total number of passengers was stable. Therefore the route can be regarded as mature. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (920), as also acknowledged by Ryanair itself (921), and thus it can be excluded that charter operations would exert any meaningful competitive constraint on the merged entity (see also section 6.7.). Further, contrary to what Ryanair argues (922), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

856. On this route, Ryanair and Aer Lingus overlap on a city-to-city basis. In the winter 2006/2007 season, Ryanair operated around 20 weekly rotations to Stansted airport (STN) and around 7 weekly rotations to Gatwick airport (LGW) while Aer Lingus operated around 21 weekly rotations to Heathrow airport (LHR). These airports are considered as substitutable from the demand side (923). In the summer 2007 season, Ryanair's number of rotations remains the same (924) while Aer Lingus increases weekly rotations to around 27 at Heathrow. Apart from the parties, there are no other competitors active on this route (925). Ryanair argues that Ryanair and Aer Lingus offer a differentiated product on this route where Aer Lingus' fare "is consistently in the region of double that of FR's [Ryanair's]", referring to a graph plotting Aer Lingus' fares based on MIDT/OAG data. However, as already indicated above (see analysis of the Dublin - London route, paragraph 796 and following), these fare data do not properly account for Internet sales and, in general, grossly overstate the actual Aer Lingus fares. Further, according to this data the fares on this route of not only Aer Lingus but also easyJet would for example in the winter 2005/2006 season be more than eight times higher than Ryanair's actual average fares (EUR 141.25 for Aer Lingus, EUR 130.31 for easyJet compared to EUR [10-20]* for Ryanair to Stansted and EUR [10-20]* for Ryanair to Gatwick), which seems unlikely in view of the business models of these carriers, their cost base (see Section 7.3.3.) and the "fare wars" (926) on this route following easyJet's entry. Further, as regards other differentiation of services between Ryanair and Aer Lingus, similar arguments as indicated in the analysis of the Dublin - London routes apply with the difference that this is a monopoly route and Ryanair and Aer Lingus are the only carriers on the route and thus no other actual competitors could be even considered as closer competitors for Ryanair or Aer Lingus.

857. As regards entry of competitors in the past, easyJet entered the route from Gatwick airport in January 2005 (together with also Cork - London and Knock - London routes). It should be stressed that Ryanair clearly reacted very aggressively to the entry of easyJet into this route. In response to easyJet's entry (927), Ryanair also started operations between Shannon and Gatwick as well as Shannon and Luton (928) in May 2005 (after having established Shannon as a base at the end of November 2004 (929)) and reacted by lowering fares (930). The intention of Ryanair to [...]* is also clear from the following Statement in the Ryanair internal documents referring to the Shannon, Cork and Knock routes to London: "[...]*" (931) Following Ryanair's reaction, easyJet reduced by half the number of frequencies between Shannon and London in August 2005 (932) and ultimately left all the Irish regional routes in October 2006. Following easyJet's exit, Ryanair discontinued its flights between Shannon and Luton and thus reduced its total number of weekly rotations between Shannon and London by around 20% from 34 to 27. Apart from easyJet's attempt, there has been no other entry by a competing carrier against Ryanair on the Shannon - London route in the recent years. Even though Virgin Express as well as City Flier Express were present on the route before Ryanair entered in April 2000, both exited the route after Ryanair's entry (Virgin Express in March 2001 and City Flier Express in October 2001) and only Aer Lingus remained on this route in competition with Ryanair.

858. The market shares of the parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (933).

<emplacement tableau>

859. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Shannon - London route from 2 to1 and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly that would eliminate all existing competition on this route (934).

860. The route is subject to general barriers to entry described in Section 7.8. The Shannon - London route is a busy route both as regards passenger numbers and the total number of frequencies operated by the Merging Parties. Therefore, any new entry would have to be of a substantial scale in terms of the operated capacity and frequencies to exert a competitive constraint on the merged entity. It is important for its competitiveness that any airline active on this route attracts a certain proportion of business or more time-sensitive passengers (935) by offering flights at all peak times of the day. For this reason, even though there appears to be sufficient capacity at Shannon airport, the limited capacity of the fully coordinated airports of Heathrow (heavily congested throughout the day), Gatwick (congested throughout the day) and Stansted (congested at peak times) constitutes a significant barrier to entry to this route (936). Even though the Luton and City airport are not congested, the availability of capacity at these airports does not eliminate the barriers to entry to this route. As described in greater detail in the Dublin - London route analysis, London City airport serves predominantly business passengers flying in smaller aircraft and cannot thus be used for a high volume entry covering the whole spectrum of passengers served by Ryanair and Aer Lingus. Further both London City and Luton are, because of supply side considerations, not an alternative to the congested London airports (and in particular Heathrow) for a number of potential entrants, in particular due to their existing bases/hubs and network activities.

861. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) would easily be able to enter this route if the merged entity were to increase prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at the London airports: British Airways with 87 short-haul aircraft based in Heathrow, 33 short-haul aircraft based in Gatwick and 10 regional jets based in London City; easyJet with 18 aircraft based in Gatwick, 13 in Stansted and 17 in Luton; BMI with 14 aircraft based in Heathrow; Globespan with 1 aircraft based in Stansted. Further, even though Air Berlin does not have any aircraft based in London, it uses Stansted as a hub for its flights to some UK destinations and a number of holiday routes.

862. However, on the basis of the Phase II investigation none of these carriers appear to have sufficient abilities and/or incentives to enter this route in the event of a price increase by the merged entity. The arguments related to individual carriers described in the Dublin - London route analysis above apply by analogy also to the Shannon - London route. Further, in view of the unsuccessful entry by easyJet to the Shannon - London route in 2005/2006 and in particular the reaction of Ryanair to this entry (for more details see in Section 7.8.), it is doubtful whether easyJet or any other carrier would be willing and able to challenge the further strengthened Ryanair in the Shannon - London route in the short to mid-term. As around 40-50% of passengers on this route originate in Ireland (937), it is less important for any successful new entry to attract a significant share of Irish originating customers. However, as Ryanair has a strong presence (and indeed even a base) at both ends of this route, any new entrant not yet present in Ireland would potentially still be in a more disadvantageous position in comparison to Ryanair. It should be noted that with 4 Ryanair aircraft and 2 Aer Lingus aircraft based in Shannon as well as in total 40 Ryanair aircraft based in Stansted and Luton, the merged entity would have the advantage of having bases at both sides of this route, with resulting greater flexibility and cost benefits. Further, Shannon - London is a mature route making entry by a new carrier with additional capacities even more demanding, as has been demonstrated by the entry of easyJet which made the increase of total passengers carried on this route possible only at the expense of very low yields (938). Further, there are no examples of (even only attempted) new entry to this route in recent years by any other scheduled carrier. Finally, apart from Ryanair and Aer Lingus, there are no other scheduled airlines based at Shannon airport. The market investigation did not identify any likely new entrant for this route not based in either Shannon or London. Ryanair argued that Centralwings could readily operate a W-route with aircraft based on one of its bases in Poland, for example Warsaw-Shannon-Stansted-Shannon-Warsaw. However, as discussed in Section

7.8.2., a new entry on this basis is less likely and less efficient. Further, in view of the length of the suggested W-route, it is also doubtful whether it could be operated without having to change the crew in view of the crew utilisation limitations (939). Further, Centralwings would thus be able to offer only one daily rotation on this route compared to almost 8 daily rotations of Ryanair and Aer Lingus combined. Therefore, it is not likely that, even if Centralwings would be able and willing to enter this route, it would be able to provide a significant competitive constraint on the merged entity.

863. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

864. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Shannon - London route.

Cork - London

865. The total number of passengers carried by scheduled airlines on the Cork - London route in 2006 amounted to [1 000 000 - 1 050 000]* and in total around 62 weekly rotations are currently operated on this route by scheduled carriers. The total number of passengers carried on this route decreased slightly (by less than [0-10%]*) in 2006, compared to the situation in the year 2005. There was a significant increase of the number of passengers in 2005, which could be attributed to the entry of easyJet and the subsequent price war described in more detail in Section 7.8.5. and in paragraph 867. In the years before, the total number of passengers was stable. Therefore the route can be regarded as mature. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible940, as also acknowledged by Ryanair itself (941), and thus it can be excluded that charter operations would exert any meaningful competitive constraint on the merged entity (see also section 6.7.). Further, contrary to what Ryanair argues (942), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

866. On this route, Ryanair and Aer Lingus overlap on a city-to-city basis. In the winter 2006/2007 season, Ryanair operated around 20 weekly rotations to Stansted airport (STN) and around 7 weekly rotations to Gatwick airport (LGW) (943) while Aer Lingus operated around 28 weekly rotations to Heathrow airport (LHR). These airports are considered as substitutable from the demand side944. In the summer 2007 season, Ryanair's number of rotations remains the same while Aer Lingus increases weekly rotations to around 34 at Heathrow. Apart from the parties, there are no other competitors active on this route (945).

867. As regards entry of competitors in the past, easyJet entered the route from Gatwick airport in January 2005 (together with Shannon - London and Knock - London routes). It should be stressed that Ryanair clearly reacted very actively to the entry of easyJet into this route. In response to easyJet's entry (946), Ryanair also started operations between Cork and Gatwick in November 2005 (after having established Cork as a base in September 2005) (947) and reacted by lowering fares (948). The intention of Ryanair to [...]* is also clear from the following Statement in the Ryanair internal documents referring to the Shannon, Cork and Knock routes to London: "[...]*" (949) Following Ryanair's reaction, easyJet ultimately left all the Irish regional routes in October 2006. Following easyJet's exit, Ryanair reduced the number of weekly rotations between Cork and Gatwick from 14 to 7 and thus decreased its total capacity operated on this route by around 20%. Apart from easyJet's attempt, BMIBaby entered the Cork - London (Gatwick) route in March 2004 but left in January 2005, after less than one year of operation. Further in the past, City Flyer Express operated Cork - Gatwick flights for a number of years but exited in October 2001. Therefore, despite the attempts of BMI Baby and easyJet in recent years, only Aer Lingus remains on this route in competition with Ryanair.

868. The market shares of the parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (950).

<emplacement tableau>

869. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Cork - London route from 2 to 1 and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (951).

870. The route is subject to the general barriers to entry described in Section 7.8. The Cork - London route is a busy route both as regards passenger numbers and the total number of frequencies operated by the Merging Parties. Therefore, any new entry would have to be of a substantial scale in terms of the operated capacity and frequencies to exert a competitive constraint on the merged entity. It is important for its competitiveness that any airline active on this route attracts a certain proportion of business or more time-sensitive passengers (952) by offering flights at all peak times of the day. For this reason, even though there appears to be sufficient capacity at Cork airport, the limited capacity of the fully coordinated airports of Heathrow (heavily congested throughout the day), Gatwick (congested throughout the day) and Stansted (congested at peak times) constitutes a significant barrier to entry to this route (953). Even though the Luton and City airports are not congested, the availability of capacity at these airports does not eliminate the other barriers to entry to this route. As described in greater detail in the Dublin - London route analysis, London City airport serves predominantly business passengers flying in smaller aircraft and cannot thus be used for a high volume entry covering the whole spectrum of passengers served by Ryanair and Aer Lingus. Further both London City and Luton are for supply side considerations not an alternative to the congested London airports (and in particular Heathrow) for a number of potential entrants, in particular due to their existing bases/hubs and network activities.

871. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) would easily be able to enter this route if the merged entity were to increase prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at the London airports: British Airways with 87 short-haul aircraft based in Heathrow, 33 short-haul aircraft based in Gatwick and 10 regional jets based in London City; easyJet with 18 aircraft based in Gatwick, 13 in Stansted and 17 in Luton; BMI with 14 aircraft based in Heathrow; Globespan with 1 aircraft based in Stansted. Further, even though Air Berlin does not have any aircraft based in London, it uses Stansted as a hub for its flights to some UK destinations and a number of holiday routes.

872. However, on the basis of the Phase II investigation none of these carriers appear to have sufficient abilities and/or incentives to enter this route in the event of a price increase by the merged entity. The arguments related to individual carriers described in the Dublin - London route analysis in paragraphs 794 to 806 apply by analogy also to Cork - London route. Further, in view of the unsuccessful entry to the Cork - London route by easyJet in 2005/2006 (as well as BMIBaby in 2004) and in particular the reaction of Ryanair to easyJet's entry (for more details see in Section 7.8.), it is doubtful whether easyJet or any other carrier would be willing and able to challenge the further strengthened Ryanair in the Cork - London route in the short to mid-term. As only slightly more than 40% of passengers on this route originate in Ireland (954), it is less important for any successful new entry to attract Irish originating customers. However, as Ryanair has a strong presence (and indeed even a base) at both ends of this route, any new entrant not yet present in Ireland would potentially still be in a more disadvantageous position in comparison to Ryanair. It should also be noted that with 1 Ryanair aircraft and 4 Aer Lingus aircraft based in Cork as well as in total 40 Ryanair aircraft based in Stansted and Luton, the Merged Entity would have the advantage of having bases at both sides of this route, with resulting greater flexibility and cost benefits. Further, Cork - London is a mature route making entry by a new carrier with additional capacities even more demanding, as has been demonstrated by the entry of easyJet which made the increase of total passengers carried on this route possible only at the expense of very low yields (955).

873. Ryanair also argued that several low-cost carriers (Wizzair, BMIBaby, Centralwings, Jet2, Malev and Air Southwest) not based at either side of the route but operating to Cork from other airports can be considered as likely entrants (956). It is not clear why Malev is considered by Ryanair as a low-cost airline as it is a Hungarian flag carrier belonging to the oneworld alliance and offering full-service passenger, cargo and charter services (957). It operates one single flight to Cork from its hub at Budapest and even if it would be willing and able to enter the Cork - London route it is difficult to see how it would be able to provide any significant competitive constraints on the merged entity. Ryanair also argues that Jet2 poses a significant competitive threat to the parties as it has been praised by the Chief Executive of DAA as being responsible for the increase of passenger volumes on Cork to provincial UK routes (958). However, Jet2 operates only one single route to Cork airport (from its base at Newcastle) with around 5 weekly rotations. It is not clear how this could provide a basis for Jet2 to enter the Cork - London route where the parties operate together in total more than 60 weekly rotations. Further, Ryanair argues that BMIBaby bases 20 aircraft at London airports, which is however incorrect as BMIBaby has no aircraft based at London airports (959) and the aircraft Ryanair refers to are probably BMI's aircraft (see Section 7.8.9.). Further, Ryanair argues that Centralwings could readily operate a W-route as it operates flights both to Cork and Stansted or Gatwick from its bases in Poland. However, as discussed in Section 7.8.2., a new entry on this basis is less likely and less efficient. Further, in view of the length of the suggested W-route, it is also doubtful whether it could be operated without having to change the crew in view of the crew utilisation limitations (960). Further, Centralwings would thus be able to offer only one daily rotation on this route compared to around 8 daily rotations of Ryanair and Aer Lingus combined. As regards Wizzair and Air Southwest, they offer only two routes (Gdansk, Katowice) and one (Newquay) route to Cork respectively, all with less than daily rotation. It is therefore not clear how this could provide the carriers with a basis for entering the Cork - London route where the parties operate together in total more than 60 weekly rotations. Therefore, even presuming that these carriers can be considered as likely entrants to the Cork - London route, which the Commission questions, the evidence does not suggest that they would have the ability and incentive to provide a significant competitive constraint on the merged entity.

874. Apart from Ryanair and Aer Lingus, Aer Arann is also based at Cork airport with 3 aircraft. However, these aircraft are 2 ATR72 planes (maximum 72 seats) and 1 ATR42 plane (maximum 48 seats) which are not able to provide effective constraints to Ryanair and Aer Lingus on a high volume route such as Cork - London. Ryanair argues that Aer Arann has recently initiated an aggressive advertising campaign against Ryanair (allegedly demonstrating that Ryanair's reputation does not deter aggressive competition), competes head to head with Ryanair on the domestic Cork-Dublin route and operates several other routes from Cork (961). However, Aer Arann stated that it is not willing to enter any route where Ryanair is active even in the event of a price increase by Ryanair (962). Also the regression analysis conducted by the Commission indicates that Aer Arann is not likely to exert any significant competitive constraints on the pricing of Ryanair and Aer Lingus (963) (see further in Section 7.8. for a more detailed discussion of Aer Arann's activities and its abilities and incentives to provide a competitive constraint to the merged entity on the overlap routes). In addition, the market investigation did not identify any likely new entrant for this route not based in either Cork or London.

875. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

876. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Cork - London route.

Cork - Manchester

877. The total number of passengers carried by scheduled airlines on the Cork - Manchester route in 2006 amounted to [150 000 - 200 000]* and in total around 22 weekly rotations are currently operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 decreased by [10- 20%]* but grew in the preceding years. The average annual growth of traffic from 2002 till 2006 was around [0-10%]* and the route can be considered as mature. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (964) and thus it can be excluded that charter operations would exert any meaningful competitive constraint on the merged entity (see also section 6.7.). Further, contrary to what Ryanair argues (965), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

878. On this route, Ryanair and Aer Lingus overlap on a city-to-city basis. In the winter 2006/2007 season Ryanair operated around 5 weekly rotations to Liverpool airport (LPL). In the summer 2007 season, Ryanair plans to operate around 7 weekly rotations on this route. Aer Lingus started its operations on this route as of 25 March 2007 and operates around 3 weekly rotations to Manchester airport (MAN) (966). These airports, together with Leeds Bradford (LBA), are considered as substitutable from the demand side (967). Apart from the parties, bmibaby operates around 7 weekly rotations to Manchester airport in summer 2007 and Aer Arann operates around 5 weekly rotations to Leeds Bradford airport. As of May 2002, bmi also operated on this route between Dublin and Leeds Bradford airport but exited in June 2006, around a year after Ryanair started its operations on this route in April 2005. Further, BA CityExpress operated on the route from Manchester airport at least as of March 2002 but exited in October 2005, a few months after Ryanair commenced its services on this route.

879. The market shares of the parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (968). It should be noted, however, that Aer Lingus commenced its operations only in March 2007, bmi exited in June 2006 and Aer Arann started its operations only in October 2006. Therefore, the market shares for summer 2006 and winter 2006/2007 do not reflect the current situation on the route and the market shares for summer 2007 should be regarded as the main indicator of the competitive position of the Merging Parties and their competitors post merger.

<emplacement tableau>

880. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Cork - Manchester route from 4 to 3 and would give the merged entity a very high combined market share (969). The very high combined market share of Ryanair and Aer Lingus on this route is in itself evidence of a dominant market position (970). Moreover, the transaction would combine Ryanair and Aer Lingus, the two closest competitors active on the route (971). As explained in more detail in Section 7.8, Aer Arann is a regional full-service carrier operating significantly smaller aircraft. Also in view of its minor presence on this route, Aer Arann cannot be considered as a significant competitive constraint on the Merging Parties. It should be acknowledged that bmibaby's business model is closer to the model of the Merging Parties, however it is not clear whether bmibaby would constitute a significant competitive constraint on the merged entity. As after the proposed merger, the merged entity would be competing with bmibaby directly on an airport to airport basis at Manchester airport (and not, as at present, when Ryanair operates to Liverpool which gives bmibaby some possibilities for differentiation), bmibaby would be faced with competition from both airports post merger. In this context it is relevant to note that bmibaby already exited from Dublin - East Midlands and Cork - London (Gatwick) in 2005 after entry of Ryanair as these routes "no longer produced a viable economic return in direct competition with Ryanair's services". (972) Further, the interview with bmi/bmibaby indicated that bmibaby targets expansion to European leisure destinations (beach/city break) (973). It is therefore at least doubtful, whether bmibaby would be able and willing to stay on the route even after entry of Aer Lingus and after the proposed merger and provide a significant competitive constraint on the merged entity. Therefore, it is likely that the merger would lead to the elimination of the competitive rivalry between the Merging Parties, which is an important source of competition on this route.

881. The route is subject to the general barriers to entry described in Section 7.8. It should however be acknowledged that the share of business passengers on this route would most likely be smaller than in the case of the Dublin - Manchester route discussed in paragraph

838 and following. This is also documented by the lower number of offered frequencies as none of the carriers offer more than one rotation per day. For this reason, the peak time congestion at the fully coordinated Manchester airport is probably a less significant barrier to new entry. There are no capacity problems at Cork airport. On the other hand, the Cork - Manchester route is a mature route, making entry by a new carrier with additional capacities even more demanding, as also documented by the exits of several carriers in recent years.

882. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport would easily be able to enter this route if the merged entity were to increase prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Manchester, Liverpool and Leeds Bradford airports: BA Connect with 16 (974) based aircraft, easyJet with 8 based aircraft, Jet2 with 16 based aircraft, bmi/bmibaby with 8 based aircraft and Monarch with 3 based aircraft. bmibaby is already present on the route and is discussed in paragraph 880. As regards the remaining potential entrants, on the basis of the Phase II investigation none of these carriers appear to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in case of a price increase by the merged entity. None of these carriers confirmed that they would be ready in the near future to enter the Cork - Manchester route even in case the combined entity increases prices. Further, as described in more detail in Section 7.8, most of these carriers have a different strategic focus of their activities and consider other routes as more profitable and less risky alternatives for utilisation of their aircraft. BA Connect is a regional UK airline operating smaller aircraft which would not represent effective competition to Ryanair and Aer Lingus' higher volume operations on this route. Monarch is a low-frills as well as charter airline operating mainly between the UK cities and holiday destinations predominantly in the Mediterranean. Similarly, Jet2 is a low-frills airline operating mainly between the UK cities and holiday destinations. Even though it operates some services to major European cities and also to Cork (from Newcastle (975)), it is focused principally on the UK customers and would not be willing to expand in the Irish market where both Ryanair and Aer Lingus are well known established companies offering a large number of destinations (976). As explained in more detail in Section 7.8, easyJet can also not be considered as a credible potential entrant on this route in the short to medium term, also in view of its recent exit from, among others, the Cork - London route after an aggressive reaction by Ryanair. Further, these carriers have no or only very minor existing operations in Ireland and entering the Irish market would require substantial investments into brand awareness and marketing. As between 40-50% of passengers on this route originate in Ireland (977), it is less important for a successful new entry to attract a significant share of Irish originating customers. However, as Ryanair has a strong presence (and indeed even a base) at both ends of this route, any new entrant not yet present in Ireland would potentially still be in a more disadvantageous position in comparison to Ryanair. With 7 Ryanair aircraft based in Liverpool, the merged entity would have the advantage of having bases at both sides of this route, with resulting greater flexibility and cost benefits. Finally, there are several examples of exits by carriers active on this route which followed a short time after Ryanair commenced its operations.

883. As explained above, the remaining competing carrier based in Cork airport (Aer Arann) is not a significant competitive constraint on the parties on this route. The market investigation did not identify any other likely new entrant for this route not based in either Cork or Manchester/Liverpool/Leeds Bradford.

884. Therefore, the results of the market investigation demonstrate that the possibility of entry or expansion of capacity by the existing competitors is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

885. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Cork - Manchester route.

7.9.3. Ireland - Continental European routes (+ Tenerife)

886. The overlap routes between Ireland and Continental Europe are a more varied group of routes which can be in general distinguished into the following three groups (even though some routes may fall into more of these categories with different characteristics): (1) routes to major European cities (such as Paris, Frankfurt, Rome, Madrid); (2) routes to holiday destinations (which can be further subdivided into summer (sun and sea) holiday destinations such as Faro, Tenerife, Alicante, Venice and winter (ski) holiday destinations such as Salzburg); and (3) routes to Central and Eastern European countries (such as Warsaw, Krakow, Poznan, Riga).

887. In general, the routes between Ireland and Continental Europe are rather dynamic and growing. According to Dublin Airport Authority, the total traffic between Dublin and the Continental Europe grew in 2006 by 25% in comparison to the previous year, with over 10 million passengers carried in 2006 (978). This growth can, in principle, be attributed in particular to the growing number of holiday destinations (predominantly in the Mediterranean) served by the scheduled operators. Further, the traffic to/from Central and Eastern European countries grew significantly also due to the fact that Ireland has already opened its labour market to the Central and Eastern European EU countries (979). On the other hand, the routes to the major European cities are more mature.

888. In connection with the growth of the Continental Europe routes in the recent years, a number of the overlap routes were started by Aer Lingus and, in particular, Ryanair (which was, in the past, more focused on the Ireland - United Kingdom market and entered the routes to Continental Europe destinations from Ireland only in the recent years (980)) only recently. Out of the in total 26 overlap routes to Continental Europe, on 17 routes the overlap was created by the entry of Ryanair or Aer Lingus (or both of them), since January 2006. This fact also illustrates the dynamic element of competition between Ryanair and Aer Lingus in the passengers air transport to/from Dublin.

Dublin - Alicante

889. The total number of passengers carried by scheduled airlines on the Dublin - Alicante route in 2006 amounted to [200 000 - 250 000]* and currently in total between around 8 weekly rotations in winter, and around 18 weekly rotations in summer are operated on this route by the Merging Parties. The total number of passengers carried on this route in 2006, compared to the situation in the year 2005, increased by around [30-40%]* and there was also significant growth in previous years. Even though there are significant traffic volumes throughout the year, there is a certain seasonal pattern with high summer traffic (as also illustrated by the significantly different number of frequencies in summer and winter). In addition, in view of the presence of charter carriers on this route (981), the possible competitive constraints on the merged entity on this route exercised by charter traffic are analysed in more detail in paragraph 894.

890. On the Dublin - Alicante route, Ryanair and Aer Lingus overlap on a city-to-city basis and as of May 2007 also on an airport-to-airport basis. In the winter 2006/2007 season, Ryanair operated around 4 weekly rotations to Murcia airport (MJV) and Aer Lingus operated around 4 weekly rotations to Alicante airport (ALC). These airports are considered as substitutable from the demand side (982). The planned weekly rotations for the summer 2007 season are around 7 to Murcia and, as of May 2007, also around 5 to Alicante for Ryanair and around 11 for Aer Lingus. Apart from the Merging Parties, there are no other competitors active on this route (983) and there are no examples of new entry to this route in recent years.

891. The market shares of the Merging Parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 seasons, as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (984).

<emplacement tableau>

892. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin Alicante route from two to one and would give the merged entity a 100% combined market share. The merger is thus a merger to monopoly which would eliminate all existing competition on this route (985).

893. Ryanair argues that Ryanair and Aer Lingus do not compete in the same space as there is ample evidence that there are substantial differences in pricing between Dublin - Alicante and Dublin - Murcia (986). In the Form CO, Ryanair claims that its average fare without taxes on this route in the Winter 2005/2006 season amounted to EUR [50-60]* while Aer Lingus' average fare without taxes according to the Ryanair's data amounted to EUR 144.83. However, as already indicated in the analysis of the Dublin - London route, paragraphs 797 and 798, these fare data do not properly account for Internet sales and are in general grossly overstating the actual Aer Lingus fares. This is also shown by the alleged Aer Lingus fares based on MIDT/OAG for the Dublin to Alicante route which are, according to the Form CO, significantly higher in winter (EUR 144.83 in Winter 2005/2006) than in summer (EUR 74.39 in Summer 2005). As this route is predominantly a summer leisure route with higher demand in summer, these fares seems illogical. This is also confirmed by Ryanair's own fares which show exactly the opposite pattern (EUR [50- 60]* in Winter 2005/2006 and EUR [100-110]* in Summer 2005).

894. In general, charter airlines are not considered to constitute a significant competitive constraint on Ryanair and Aer Lingus on the overlap routes (987). However, as charter traffic represents an important share of passengers carried on this route, in particular in the summer season, and taking into account the arguments to the contrary raised by Ryanair with respect to this particular route, the Commission has explored whether charter airlines could exercise a competitive constraint on the Merging Parties on this route. In particular, the following charter airlines operate on this route: Air Europa, Air Futura and Spanair (988). The charter airlines are present on this route practically only during the Summer season and thus do not have any influence on the activities of the Merging Parties during the rest of the year. The market investigation indicated that a large majority of seats of charter airlines in Ireland were not sold separately, but rather as a part of a package holiday, sold to final customers by tour operators and not by the charter airlines themselves (989). Even though Ryanair refers, for example to FirstChoice's submission and claims that it offers "dry seats" on the Irish market (990), it should be added that FirstChoice sells only 3% of its tickets in Ireland as dry seats and that 97% of its tickets are sold as part of package holidays (991). There are no indications that this would be different for charter airlines active on this particular route. Therefore, only a very small part of the [0-10%]* share of the charter traffic on this route were dry seats. Furthermore, none of these charter airlines offered a possibility to book flights only for Dublin - Alicante via their website, even though it is possible to do so at the webpages of some tour operators selling their tickets in Ireland (992). Most of the charter airlines sell their tickets to tour operators who then sell them either as part of the package holidays (which also means that it is, in general, the tour operator and not the charter airline who is deciding on which routes to operate and when (993)) or to a much smaller extent as dry seats. It should also be noted that the operations of the charter companies on this route during the Summer 2006 season were concentrated to weekends only (994). On the other hand, the merged entity would offer a significantly more flexible timetable with more than two rotations each day of the week during the summer season. Furthermore, in view of the decreasing charter market in Ireland in general as well as on this particular route (995), the likelihood of any significant entry by new charter carriers with a more flexible product offering which would provide any significant constraint on the merged entity is low (996). Therefore, as only a small part of the [0-10%]* share of charter traffic was constituted by dry seats and taking into account the limited period of the charter operations on this route, the difference in the product sold to the final customers and in the used distribution channels (tour operators) and the limited flexibility of the charter flights, charter airlines cannot be considered to represent any significant competitive constraint for the merged entity on this route.

895. The route is subject to the general barriers to entry described in Section 7.8.. It should be, however, acknowledged that airport congestion does not constitute a significant barrier to entry to this route as the predominantly holiday flights on this route are not dependent on peak time slots and high frequencies. On the other hand, it should be noted that also carriers focused on leisure routes prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier. (997). On the other hand, an absolute majority of the passengers on this route originate in Ireland (998) and any new entrant would therefore need to secure sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland, as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. Finally, there are no examples of any new entry to this route in the recent years.

896. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. According to Ryanair, Iberia is based in Alicante with two aircraft. However, Iberia in its reply did not confirm this fact (999). Nevertheless, even if this was the case, it is highly unlikely that Iberia would enter this route in view of its business model based on fullservice network operations (Iberia is a member of the "oneworld" alliance). Iberia is more focused on feeding its hubs at Madrid and Barcelona and its entry into point-to-point operations between Dublin and Alicante against the merged entity is highly unlikely, which also confirmed the reply of Iberia in the market test (1000). Furthermore, as a network carrier Iberia is also likely to have higher unit costs than Ryanair and Aer Lingus (see Section 7.3.). Ryanair further argued that Iberia has the scope to operate a no-frills service with its Clickair operation (1001). However, Clickair does not have any aircraft based in Alicante and it cannot be regarded as a likely entrant also for other reasons explained in detail above (1002).

897. As explained in detail in Sections 7.3. and 7.8, the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to the Dublin - Alicante route which would provide significant constraints on the merged entity. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Alicante airports. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

898. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Alicante route.

Dublin - Barcelona

899. The total number of passengers carried by scheduled airlines on the Dublin - Barcelona route in 2006 amounted to [400 000 - 450 000]* and currently in total between around 26 weekly rotations in winter and around 32 weekly rotations in summer are operated on this route by the scheduled carriers. The total number of passengers carried on this route in 2006, compared to the situation in the year 2005, increased by around [20-30%]* and also grew steadily in previous years. Even though there are significant traffic volumes throughout the year, there is a certain seasonal pattern with high summer traffic (as also illustrated by the different number of frequencies in summer and winter). Furthermore, in view of the presence of charter carriers on this route (1003), the potential competitive constraints on the merged entity on this route exercised by charter traffic are analysed in more detail in paragraph 903.

900. On the Dublin - Barcelona route, Ryanair and Aer Lingus overlap on a city-to-city basis. In the winter 2006/2007 season, Ryanair operated around 5 weekly rotations to Reus airport (REU) and around 7 weekly rotations to Girona airport (GRO) while Aer Lingus operated around 7 weekly rotations to Barcelona airport (BCN). These airports are considered as substitutable from the demand side (1004). The planned weekly rotations for the summer 2007 season are around 7 to Reus and 7 to Girona for Ryanair and around 11 to Barcelona for Aer Lingus. Apart from the Merging Parties, Iberia has traditionally operated on this route from Barcelona airport, but it is being replaced by its low-frills subsidiary Clickair as of March 2007. Iberia/Clickair operates around 7 weekly rotations on this route both in Winter 2006/2007 and in Summer 2007. Furthermore, Spanair operated scheduled services between Dublin and Barcelona in the Summer 2006 season but exited the route in October 2006 and does not have any scheduled services planned for Summer 2007.

901. The market shares of the Merging Parties and their competitors on the basis of the operated seat capacity in the Winter 2006/2007 and Summer 2007 seasons, as well as on the basis of the passengers carried in the Summer 2006 season are summarised in the following table. It must, however, be noted that Iberia carried a significant proportion of connecting passengers on this route. On the other hand, both Ryanair and Aer Lingus carry almost only (1005) point-to-point passengers. Therefore, in line with the definition of the relevant market and in order to better reflect the competitive situation on this market, it is appropriate to calculate the market share excluding connecting passengers as indicated in the last column of the table.

<emplacement tableau>

902. As Spanair has left the route, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Barcelona route from three to two and would give the merged entity a very high combined market share. The very high combined market share of Ryanair and Aer Lingus on this route is, in itself, evidence of a dominant market position, in particular as only point-to-point passengers should be taken into account (1006). Moreover, the transaction would combine Ryanair and Aer Lingus, the two closest competitors active on the route (1007). Even though Iberia serves the same airport in Barcelona as Aer Lingus, its business model based on full-service network operations is clearly different from Aer Lingus and Ryanair. It should be acknowledged that Clickair is closer to the model of Ryanair and Aer Lingus and it can be regarded as mid-frills carrier combining the low-frills model with some traditional features of network carriers (1008). Thus, Clickair, for example, has a code-sharing agreement with Iberia or would participate in Iberia's frequent flyer programme (1009). This code-sharing agreement allows Clickair to have its services available through GDS-powered points of sale and, according to Clickair, the share of tickets sold under the code-share agreement with Iberia is indeed important and in any event "larger than the usual airline-to-airline marketing code-share agreement, which may contribute a few points of percentage share in individual markets" (1010). It should be noted that Clickair considers possible expansion of operations to Dublin to be in line with their expansion plans even though they are not able to indicate any concrete plans at present. On the contrary, Clickair does not regard Dublin as an attractive market for expansion in view of the strength of the Merging Parties in their home market in Dublin and as that would require substantial investment into marketing (1011). Furthermore, Clickair started their operations only in October 2006 and thus does not have such established brands as Ryanair and Aer Lingus and it is not clear at present to what extent it would be able to successfully enter the short-haul air transport markets. Even though the Iberia brand is established and can assist Clickair in view of their code-sharing agreements, Iberia is not seen as a low-frills low-fares brand. Finally, in view of the large majority of passengers originating in Ireland on this route (1012), it is doubtful whether Clickair would be able to expand significantly its operations against the merged entity's strong brands in Ireland. In any event, the proposed merger would lead to elimination of the existing competitive rivalry between the Merging Parties' which has been an important source of competition on this route, in particular in the case of point-to-point passengers between Dublin and Barcelona on which both Ryanair and Aer Lingus focus.

903. In general, charter airlines are not considered to constitute a significant competitive constraint on Ryanair and Aer Lingus on the overlap routes (1013). However, as charter traffic represents a significant share of passengers carried on the Dublin - Barcelona route, in particular in the summer season, and taking into account the arguments to the contrary raised by Ryanair with respect to this particular route, the Commission has explored whether charter airlines could exercise a competitive constraint on the Merging Parties on this route. In particular, the following charter airlines operated on the Dublin - Barcelona route in recent years: Air Futura, Eirjet (which went bankrupt in October 2006), FirstChoice Airways, Monarch Airlines and Czech Airlines (1014). The charter airlines are present on this route mainly between May and September and their operations during the rest of the year are more limited. The market investigation indicated that a large majority of seats of charter airlines in Ireland were not sold separately but rather as a part of a package holiday, sold to final customers by tour operators and not by the charter airlines themselves (1015). Even though Ryanair refers, for example, to FirstChoice's submission and claims that it offers "dry seats" on the Irish market (1016), it should be added that FirstChoice sells only 3% of its tickets in Ireland as dry seats and that 97% of its tickets are sold as part of package holidays (1017). There are no indications that this would be different for the Dublin - Barcelona route. Therefore, only a very small part of the [10-20%]* share of the charter traffic on this route were dry seats. Furthermore, none of these charter airlines offered the possibility to book flights only for Dublin - Barcelona via their website, even though it is possible to do so at the webpages of some tour operators selling their tickets in Ireland (1018). However, it must be acknowledged that (even though practically not present on the Dublin - Barcelona route in the recent years), there is, according to the slot allocation lists provided by Ryanair, one charter airline present in Summer 2007 which does directly offer online flights between Dublin and Reus - XL Airways. However, XL Airways offers only one weekly rotation on Wednesday (1019). It should be stressed that also XL Airways sell only part of their tickets as dry seats and that they do offer package holidays as well (1020). Most of the other charter airlines sell their tickets to tour operators who then sell them either as part of the package holidays (which also means that it is in general the tour operator and not the charter airline who is deciding on which routes to operate and when (1021)) or to a much smaller extent as dry seats. However, even in case the small share of dry seats sold by XL Airways and other charter airlines is taken into account, the market share of the charter operators' combined dry seat activities would be unlikely to exceed 10% of the total traffic. Furthermore, also in view of their limited number of rotations restricted to some days of the week only (1022), charter airlines cannot be regarded as constituting a significant competitive constraint on the merged entity with its full year operations and up to around 25 weekly rotations in the summer season. The merged entity would thus offer a significantly more flexible timetable with more than three rotations each day of the week. Furthermore, in view of the decreasing charter market in Ireland in general and rather stable charter traffic in this particular route (1023), the likelihood of any significant entry by new charter carriers with a more flexible product offering is low (1024). Therefore, taking into account the limited period and extent of the charter operations on the Dublin - Barcelona route, the difference in the product sold to the final customers and in the used distribution channels (tour operators) and the limited flexibility of the charter flights, charter airlines cannot be considered to represent any significant competitive constraint for the merged entity on this route.

904. The Dublin - Barcelona route is subject to the general barriers to entry described in Section 7.8. As regards, in particular, access to airport infrastructure issues, it is noteworthy that Barcelona is a fully coordinated airport under the Slot Regulation (1025). In combination with the limited capacity of Dublin Airport during peak times, described in more detail in Section 7.8., this would constitute an additional barrier to an entry with sufficient number of attractive frequencies. Girona and Reus airports, which are not capacity constraint, may be substitutable to Barcelona airport from the demand side but less so from the supply side. It should be, however, acknowledged that airport congestion does not constitute a significant barrier to entry to this route as the more leisure oriented flights on this route are not dependent on peak time slots and high frequencies. On the other hand, it should be noted that carriers focused on leisure routes also prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier (1026).

905. Ryanair argues that, in particular, competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Barcelona airports: Iberia with more than 20 aircraft, Clickair with more than 20 aircraft (by the end of 2007), Spanair with 6 aircraft, Vueling with 9 aircraft. Ryanair argues that Hola Airlines also has a base in Barcelona. However, Hola Airlines are solely a charter airline without scheduled operations and cannot thus be regarded as a potential entrant which would provide significant competitive constraints to the merged entity's scheduled flights on this route (1027).On the basis of the Phase II investigation, the carriers based in Barcelona do not appear to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in case of a price increase by the merged entity. As regards Iberia/Clickair, this point has already been discussed in paragraph 902. In view of its recent exit from the Dublin - Barcelona route (as well as from the Dublin - Madrid and Dublin - Tenerife routes), it is also unlikely that Spanair would constitute a credible potential entrant to the Dublin - Barcelona route in the short to mid-term. In addition, Spanair is a full-frills scheduled as well as charter airline and part to the Star Alliance. Even though they plan a major expansion in the coming years, their aim is "to become the leading airline for business travel" (1028). Even though this expansion plan is focused on Barcelona, their recent exit from the Dublin - Barcelona route confirms that it cannot be regarded as a business route. As regards Vueling, it considers itself as a low-fares airline and it is based in Barcelona (and with an additional base in Madrid) serving in total 18 destinations, mainly in the Mediterranean (plus Paris, Brussels and Amsterdam). Vueling does not operate (and has never operated) any flights to/from Ireland and does not plan to enter Ireland "any time soon" (1029). Ryanair argued in its Response to the Statement of Objections that Vueling in its response to the Commission (however not specifying which response) stated that it would consider Dublin as a possible destination. However, none of the responses of Vueling (1030) include any such Statement but instead contain indications rather to the contrary, as indicated in the previous sentence. Vueling also indicates that the disproportionate marketing costs of the entry and a threat of aggressive pricing behaviour by the large merged entity would constitute further important barriers to entry (1031). Furthermore, as indicated in paragraph 902, a large majority of the passengers on the Dublin - Barcelona route originate in Ireland and any new entrant would therefore need to secure sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland (such as Clickair and Vueling, however, including also Spanair with only marginal presence in Ireland), as it would require significant investment in brand awareness and marketing taking into account the current strength of the Ryanair and Aer Lingus brands in Ireland. It should also be noted that with five Ryanair aircraft based in Girona, the merged entity would have the advantage of having bases at both sides of this route, with resulting greater flexibility and cost benefits. Finally, apart from the Spanair's short-lived entry for the Summer 2006 season only, there are no examples of sustained new entry to this route in the recent years (1032).

906. As explained in detail in Sections 7.3. and 7.8, the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to the Dublin - Barcelona route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Barcelona airports. Therefore, the results of the market investigation demonstrate that the possibility of entry or expansion of capacity by the existing competitors is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

907. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Barcelona route.

Dublin - Berlin

908. The total number of passengers carried by scheduled airlines on the Dublin - Berlin route in 2006 amounted to [150 000 - 200 000]* and currently a total of around 14 weekly rotations are operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006, compared to the situation in the year 2005, increased significantly most probably due to the entry of Ryanair to this route in June 2006. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (1033), as also acknowledged by Ryanair itself (1034), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1035), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

909. On the Dublin - Berlin route, Ryanair and Aer Lingus overlap on an airport-to-airport basis. In the Winter 2006/2007 season, Ryanair operated around seven weekly rotations to Berlin Schönefeld (SXF) and Aer Lingus operated around seven weekly rotations to that airport. The planned weekly rotations for the Summer 2007 season remain the same for both Merging Parties. Apart from the Merging Parties, there are no other competitors active on this route. In the past, Lufthansa operated the Dublin - Berlin route but exited in October 2000. Since that time, there have been no new entrants to this route apart from Aer Lingus (in April 2004) and Ryanair (June 2006).

910. The market shares of the Merging Parties and their competitors on the basis of the operated seat capacity in the Winter 2006/2007 and Summer 2007 season, as well as on the basis of the passengers carried in summer 2006 season are summarised in the following table (1036).

<emplacement tableau>

911. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Berlin route from two to one and would give the merged entity a 100% combined market share. The merger is thus a merger to monopoly which would eliminate all existing competition on this route (1037).

912. The route is subject to the general barriers to entry described in Section 7.8., even though, in view of the limited number of frequencies, and the smaller share of more time-sensitive passengers, the peak time congestion at Dublin Airport may not be as significant a barrier to entry. On the other hand, both Merging Parties operate early morning flights from Dublin (1038) and new entrants wishing to compete for the early morning passengers from Dublin, with the merged entity, may find it difficult to obtain the early morning peak time slots at Dublin Airport.

913. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to easily enter the Dublin - Berlin route in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Berlin Schönefeld airport or other Berlin airports (1039): Lufthansa with up to 10 based aircraft, Air Berlin/DBA with 16 based aircraft, Germanwings with 3 based aircraft, easyJet with 8 based aircraft. On the basis of the Phase II investigation, none of those carriers appears to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in the case of a price increase by the merged entity. Lufthansa is a traditional network carrier operating hub and spoke system different to the point-to-point low-frills models of Ryanair and Aer Lingus. Furthermore, in view of the exit by Lufthansa from the route in 2000 (when there was no other competitor on the direct Dublin - Berlin routes) it is doubtful whether it would now be willing to enter against a strong merged entity and whether its cost structure would allow it to offer an alternative for those customers who are interested in low cost flights. Even though Air Berlin and Germanwings are closer to the low-frills model of the Merging Parties, there are no indications that they would enter the Dublin - Berlin route in the case of a price increase by the merged entity (1040). Furthermore, from the internal documents of Ryanair it seems that even Ryanair does not consider Air Berlin as a credible competitive threat (1041). As explain in more detail in Section 7.8, easyJet also cannot be considered as a credible potential entrant on this route in the short to medium term. Furthermore, more than 50% of the passengers on this route originate in Ireland (1042) and any new entrant would therefore need to secure a sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland (such as Air Berlin or easyJet), as it would require significant investment in brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. Finally, apart from the Merging Parties, there are no examples of any new entry to this route in recent years.

914. As explained in detail in Sections 7.3. and 7.8, the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to the Dublin - Berlin route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Berlin. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

915. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Berlin route.

Dublin - Bilbao/Vitoria

916. The total number of passengers carried by scheduled airlines on the Dublin - Bilbao/Vitoria route in 2006 amounted to [0 - 50 000]*. However, it should be noted that Ryanair only entered the route in January 2007 and the passengers carried in 2006 are thus solely Aer Lingus passengers. Currently, a total of around four weekly rotations are operated on this route by scheduled carriers in the Winter season and around seven in the Summer season. The total number of passengers carried on this route decreased between 2005 and 2006 by around [10-20%]*. However, the route is relatively new as Aer Lingus started their operations only in April 2004 and, also in view of the recent Ryanair's entry, it thus cannot be regarded as a mature or declining route. The route is relatively thin as regards the total number of passengers and, considering the recent Ryanair's entry, it could be questioned whether there would be scope for any additional new entry. The route has no significant seasonal pattern. The presence of charter carriers on this route is only marginal (1043), as also acknowledged by Ryanair itself (1044), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also Section 6.7.). Furthermore, contrary to claims by Ryanair (1045), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining, to any significant extent, the regular year-round operations of the merged entity on this route.

917. On the Dublin - Berlin route, Ryanair and Aer Lingus overlap on a city-to-city basis and only in the Summer season. In the Winter 2006/2007 season, Ryanair operated around four weekly rotations to Vitoria airport (VIT), while Aer Lingus is not active on the route in the Winter season. The planned weekly rotations for the Summer 2007 season remain the same for Ryanair, while Aer Lingus plans to operate around three weekly rotations to Bilbao Sondica airport (BIO). Vitoria and Bilbao Sondica airports are considered as substitutable from the demand side (1046). Apart from the Merging Parties, there are currently no other competitors active on this route and there have been no other new entrants on this route in the recent years. Ryanair indicated in their Response to the Statement of Objections that they would exit this route in September 2007 (1047). However, as at the time of the Commission decision there is an actual overlap between the activities of Ryanair and Aer Lingus on this route, and as there is no legal certainty that Ryanair will indeed not be active on the route in the near future, it has to be regarded as an overlap route where the transaction would remove actual competition between Ryanair and Aer Lingus.

918. The market shares of the Merging Parties and their competitors on the basis of the operated seat capacity in the Winter 2006/2007 and Summer 2007 seasons, as well as on the basis of the passengers carried in summer 2006 season are summarised in the following table (1048).

<emplacement tableau>

919. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Bilbao/Vitoria route from two to one and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (1049).

920. The route is subject to the general barriers to entry described in Section 7.8. However, it should be acknowledged that airport congestion does not constitute a significant barrier to entry to this route as the predominantly holiday flights on this route are not dependent on peak time slots and high frequencies (1050). On the other hand, it should be noted that carriers focused on leisure routes also prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier. (1051) Furthermore, the majority of the passengers on this route originate in Ireland (1052) and any new entrant would, therefore, need to secure a sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland, as it would require significant investment in brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland.

921. Ryanair argues that, in particular, competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. According to Ryanair, there are no competing scheduled airlines based in Bilbao/Vitoria at present which would be able to enter the route. The market investigation indicated that Spanair bases aircraft in Bilbao (1053). Spanair is a full-frills scheduled as well as charter airline belonging to the Star Alliance (SAS). Even though they plan a major expansion in the coming years, their aim is "to become the leading airline for business travel" and their expansion plan is mainly focused on Barcelona (1054). Therefore, it seems unlikely that Spanair would constitute a credible potential entrant against the merged entity to this route, also in view of the fact that Spanair recently abandoned three other Dublin routes where they competed with Ryanair/Aer Lingus (1055). As explained in detail in Sections 7.3. and 7.8, the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Bilbao/Vitoria. There are no other examples of a sustained new entry to this route which would provide efficient competitive constraints to Ryanair and Aer Lingus. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

922. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position also on the Dublin - Bilbao/Vitoria route.

Dublin - Bologna

923. The total number of passengers carried by scheduled airlines on the Dublin - Bologna route in 2006 amounted to [0 - 50 000]*. However, it should be noted that Ryanair only entered the route in February 2007 and the passengers carried in 2006 are solely Aer Lingus passengers. Currently, in total around four weekly rotations are operated on this route by scheduled carriers in the Winter and around seven in the Summer season. The total number of passengers carried on this route in 2006, compared to the situation in the year 2005, decreased by around [10-20%]*, probably due to the fact that Aer Lingus operated the route from January to October in 2005 but only from March to October in 2006. The route is relatively thin as regards the total number of passengers and in view of the recent Ryanair's entry, it could be questioned whether there would be scope for any additional new entry. Currently, Aer Lingus operates the route only in the Summer season from March to October, while Ryanair started whole year services. Despite certain seasonality on the route, the presence of charter airlines on this route is marginal (1056), which Ryanair itself acknowledges (1057), and therefore does not need to be discussed separately as it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also Section 6.7). Furthermore, contrary to claims by Ryanair (1058), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

924. On the Dublin - Bologna route, Ryanair and Aer Lingus overlap on a city-to-city basis and only during the Summer season. In the Winter 2006/2007 season, Ryanair operated around four weekly rotations to Forlì airport (FRL). The planned weekly rotations for Summer 2007 season remain the same for Ryanair and will amount to around three weekly rotations for Aer Lingus flying to Bologna airport (BLQ). These airports are considered as substitutable from the demand side (1059). Apart from the Merging Parties, there are no other competitors active on this route. Before the entry of Aer Lingus in March 2003, there were no scheduled carriers operating on this route (apart from Ryanair's operations in the Summer 1997 season). Except for Ryanair's entry in February 2007, there have been no other new entrants on this route in the recent years.

925. The market shares of the Merging Parties and their competitors on the basis of the operated seat capacity in the Winter 2006/2007 and Summer 2007 season as well as on the basis of the passengers carried in Summer 2006 season are summarised in the following table (1060).

<emplacement tableau>

926. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Bologna route from two to one and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (1061).

927. The Dublin - Bologna route is subject to the general barriers to entry described in Section 7.8. However, it should be acknowledged that airport congestion at Dublin Airport does not constitute the most significant barrier to entry to this route as the predominantly leisure flights on this route are not dependent on peak time slots and high frequencies (1062). On the other hand, it should be noted that carriers focused on leisure routes also prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier (1063).

928. Ryanair argues, in particular, that competing carriers with bases or hubs at the destination airport(s) are able to easily enter the Dublin - Bologna route in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. Ryanair argues (1064) that Alitalia and Meridiana operate a base from Bologna (1065). However, in view of Alitalia's recent exit from the Dublin - Rome route, as well as from the Dublin - Milan route and its current financial situation (1066), it is unlikely that Alitalia would constitute a credible potential entrant to the Dublin - Bologna route in the short to medium-term. Furthermore, Alitalia is a full-frills network carrier which would, in any event not be a close competitor to the Merging Parties, having a different cost structure and hence a more limited ability to offer low fares. As regards Meridiana, it is an Italian low-frills airline serving in particular domestic destinations (1067). It is operating slightly smaller aircraft than Ryanair or Aer Lingus (1068). Meridiana is not active at all to Ireland, and as regards the United Kingdom, it only serves London Gatwick from Florence. Considering its regional focus, it seems rather unlikely that Meridiana would have the ability and incentive to enter the Dublin - Bologna route and provide significant competitive constraints to the merged entity. Furthermore, the majority of passengers on this route originate in Ireland (1069) and any new entrant would therefore need to secure a sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland, as it would require significant investment in brand awareness and marketing taking into account the current strength of the Ryanair and Aer Lingus brands in Ireland. Finally, apart from the Merging Parties, there are no examples of any new entry to this route in the recent years.

929. As explained in detail in Sections 7.3. and 7.8, the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants on the Dublin - Bologna route. Furthermore, the market investigation did not identify any other likely new entrant for this route that is not based in either Dublin or Bologna. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

930. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Bologna route.

Dublin - Brussels

931. The total number of passengers carried by scheduled airlines on the Dublin - Brussels route in 2006 amounted to [350 000 - 400 000]* and currently in total 24 weekly rotations are operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006, compared to the situation in the year 2005, increased by around [0- 10%]* and was rather stable in the previous years. Therefore, the route can be regarded as mature. The route has no significant seasonal pattern. The presence of charter carriers on this route is only marginal (1070), as Ryanair itself acknowledged (1071), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also Section 6.7.). Furthermore, contrary to claims by Ryanair (1072), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

932. On the Dublin - Brussels route, Ryanair and Aer Lingus overlap on a city-to-city basis. In the Winter 2006/2007 season, Ryanair operated around 12 weekly rotations to Charleroi Brussels South airport (CRL) and Aer Lingus operated around 12 weekly rotations to Brussels airport (BRU). These airports are considered as substitutable from the demand Side (1073). The planned weekly rotations for Summer 2007 season remain the same for Ryanair and increase to around 13 for Aer Lingus. Apart from the Merging Parties, there are no other competitors active on this route. Traditionally, this route was served by Sabena flying between Dublin and Brussels (BRU) airport. However, Sabena (now Brussels Airlines) left the route in November 2001 and since that time no other entry against Ryanair and Aer Lingus has taken place on this route.

933. The market shares of the Merging Parties on the basis of the operated seat capacity in the Winter 2006/2007 and Summer 2007, season as well as on the basis of the passengers carried in summer 2006 season are summarised in the following table (1074).

<emplacement tableau>

934. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Brussels route from two to one and would give the merged entity a 100% combined market share (1075). The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (1076).

935. The Dublin - Brussels route is subject to the general barriers to entry described in Section 7.8. Furthermore, the Dublin - Brussels route is a rather thick route both as regards the passenger numbers and the total number of frequencies operated by the Merging Parties. Therefore, any new entry would have to be of a substantial scale in terms of the operated capacity and frequencies to exert a competitive constraint on the merged entity. It is important for its competitiveness that any airline active on this route attracts a certain proportion of business or more time-sensitive passengers (1077) by offering flights at all the peak times of the day. For this reason, the limited capacity in the peak times of Dublin Airport, described in more detail in Section 7.8., as well as of the fully coordinated Brussels airport constitute an additional significant barrier to entry (even though the capacity constraints at Brussels airport seem to be more limited (1078)). Charleroi Brussels South airport, which is not capacity constraint, may be substitute to Brussels airport albeit subject to supply side considerations as it may not be acceptable in particular to airlines with hubs/bases at Brussels. Furthermore, Dublin - Brussels route is a mature route making entry by a new carrier with additional capacities even more demanding.

936. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Brussels airports: Brussels Airlines with more than 38 based aircraft (both long- and short-haul), KLM with 3 based aircraft, VLM with 3 based aircraft. Ryanair further included Transavia and Airlinair with bases in Eindhoven airport (1079), however the market investigation did not confirm that Eindhoven airport would be a meaningful substitute airport for Brussels or Charleroi (1080). Further, even if Transavia and Airlinair would be considered as potential entrants, they would not represent an efficient competitive constraint on the Dublin - Brussels route both due to the fact that they would fly to Eindhoven as well as in view of the business model and current operations of both Airliner (1081) and Transavia (1082). In any event, on the basis of the Phase II investigation none of these carriers appears to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in case of a price increase by the merged entity. Brussels Airlines (1083) and KLM are primarily full-frills, network carriers with a business model different from Ryanair and Aer Lingus. Furthermore, none of them is currently present on the Irish market. VLM is focused on serving business customers using smaller 50 seats aircraft between major cities in Benelux and several UK airports, with a focus on London - City (1084), which thus cannot be considered as a significant potential entrant on Dublin - Brussels route. Further, up to 50% of the passengers on this route originate in Ireland (1085) and any new entrant would therefore need to secure a sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland (such as Brussels Airlines, KLM or VLM) as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. It should also be noted that with 4 Ryanair aircraft based in Charleroi Brussels South, the merged entity would have an advantage of having bases at both sides of this route, with the resulting greater flexibility and cost benefits. Finally, apart from the Merging Parties, there are no examples of new entry to this route in the recent years.

937. As explained in detail in Sections 7.3. and 7.8., the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for the Dublin - Brussels route not based in either Dublin or Brussels/Charleroi. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

938. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Brussels route.

Dublin - Faro

939. The total number of passengers carried by scheduled airlines on the Dublin - Faro route in 2006 amounted to [200 000 - 250 000]* and currently in total around 7 weekly rotations in winter and 18 weekly rotations in summer are operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 increased by [70-80%]* and therefore the route can be regarded as growing. The route is seasonal with a significantly higher number of passengers during the summer Season (1086). It is noted that charter traffic represents a significant proportion of the total (charter and scheduled) passengers carried on this route in the summer season amounting to [40-50%]* in the summer 2006 (1087). For this reason, the possible competitive constraints on the merged entity on this route exercised by charter traffic are analysed in more detail in paragraph 944.

940. On the Dublin - Faro route, Ryanair and Aer Lingus overlap on an airport-to-airport basis. In winter 2006/2007 season, Ryanair operated around 3 weekly rotations to Faro airport (FAO) and Aer Lingus operated around 4 weekly rotations to the same airport. In summer 2007 season, Ryanair plans to operate around 7 weekly rotations and Aer Lingus around 11 weekly rotations on this route. Apart from the Merging Parties, there are no other competitors active on this route. As of June 2006, TAP Portugal was operating a service between Dublin and Faro with a stopover in Lisbon (LIS), however it exited the route as of February 2006 (1088) after less than 9 months of operation. Previously, Air Luxor was operating on this route as of February 2003 but exited in October 2004. Apart from that, there were no other new entrants on this route in recent years.

941. The market shares of the Merging Parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 seasons as well as on the basis of the passengers carried in summer 2006 season are summarised in the following table (1089).

<emplacement tableau>

942. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Faro route from two to one and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (1090).

943. In general, charter airlines are not considered to constitute a significant competitive constraint on Ryanair and Aer Lingus on the overlap routes in this case (1091). However, as charter traffic represents an important share of passengers carried on this particular route in the summer season and taking into account the arguments to the contrary raised by Ryanair with respect to this particular route, the Commission has explored whether charter airlines could exercise a competitive constraint on the Merging Parties on the Dublin - Faro route.

944. The following charter airlines in particular operated on the Dublin - Faro route in the recent years: Air Futura, Eirjet (which however went bankrupt in October 2006), Flightline, MyTravel Airways, FirstChoice Airways and Spanair (1092). However, their operations were in 2006 limited to the period between April and October and thus have no effects on the operations of the merged entity for the rest of the year. Further, the market investigation indicated that a large majority of seats of charter airlines were not sold separately but rather as a part of the package holiday, sold to the final customers by the tour operators and not by the charter airlines themselves (1093). Even though Ryanair refers for example to FirstChoice's submission and claims that it offers "dry seats" on the Irish market (1094), it should be added that FirstChoice sells only 3% of its tickets in Ireland as dry seats and that 97% of its tickets are sold as part of package holidays (1095). There are no indications that this would be different for this particular route. Therefore, only a very small part of the [40-50%]* share of the charter traffic on this route were dry seats. None of these charter airlines offered a possibility to book flights only for Dublin - Faro via their website as they do in the UK, even though it is possible to do so at the webpages of some tour operators selling their tickets in Ireland (1096). However, it must be acknowledged that (even though practically not present on the route in the recent years), there will be according to the slot allocation lists provided by Ryanair one charter airline present in Summer 2007 which does offer online flights directly between Dublin and Faro - XL Airways. However, XL Airways offers only 2 weekly rotations on Saturday and Sunday and no flights during the week (1097). It should be stressed that also XL Airways sell only part of their tickets as dry seats and that they do offer package holidays as well (1098). The other charter airlines normally sell their tickets to tour operators who then sell them either as part of the package holidays (which also means that it is in general the tour operator and not the charter airline who is deciding on which routes to operate and when (1099)) or to a much smaller extent as dry seats. However, even in case the small share of dry seats sold by XL Airways and other charter airlines is taken into account, the market share of the charter operators would be unlikely to exceed 10% of the total traffic. Further, also in view of their limited number of rotations restricted mainly to weekends only and operated solely during the summer season, charter airlines cannot be regarded as constituting a significant competitive constraint on the merged entity with full year operations and up to around 18 weekly rotations in the summer season. It should be noted that the operations of the charter companies on this route in summer 2006 were concentrated on Thursdays, Saturdays and Sundays only (1100). On the other hand, Ryanair and Aer Lingus offered a much more flexible timetable covering the whole week and in the peak season even twice daily services on some days. Furthermore, in view of the shrinking charter market in Ireland in general as well as in this particular route (1101), the likelihood of a significant entry by new charter carriers with more flexible product offering is low (1102). Therefore, taking into account the low share of dry seats sold in Ireland, the limited period of the charter operations on this route, the difference in the product sold to the final customers and in the used distribution channel and the limited flexibility of the charter flights, one may conclude that charter airlines would not represent any significant competitive constraint on the merged entity on the Dublin - Faro route.

945. The Dublin - Faro route is subject to the general barriers to entry described in Section 7.8. However, it should be acknowledged that airport congestion does not constitute a significant barrier to entry to this route as the predominantly holiday flights on this route are not dependent on peak time slots and high frequencies. On the other hand, it should be noted that carriers focused on leisure routes also prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier. (1103) Further, an absolute majority of the passengers on this route originate in Ireland (1104) and any new entrant would therefore need to secure sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland, as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland.

946. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to easily enter the Dublin - Faro route in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. However, there are no competing scheduled airlines based in Faro at present which would be able to enter the route. As explained in detail in Sections 7.3. and 7.8., the competing carriers based at Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Faro. Even though TAP Portugal entered the market with a stopover in Lisbon, its exit after less than 9 months of operations demonstrates that such indirect flights are not a viable alternative to the direct flights offered by Ryanair and Aer Lingus. There are no other examples of a sustained new entry to this route which would provide efficient competitive constraints to Ryanair and Aer Lingus. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

947. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Faro route.

Dublin - Frankfurt

948. The total number of passengers carried by scheduled airlines on the Dublin - Frankfurt route in 2006 amounted to [500 000 - 550 000]* and currently a total of around 46 weekly rotations are operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 increased by [10- 20%]*, which may be attributed to the fact that Ryanair entered the route in May 2005. However, before Ryanair's entry the number of passengers was rather stable and the route can in general be regarded as relatively mature. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (1105), as Ryanair itself acknowledged (1106), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1107), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

949. On the Dublin - Frankfurt route, Ryanair and Aer Lingus overlap on a city-to-city basis. In the winter 2006/2007 season Ryanair operated around 12 weekly rotations to Frankfurt- Hahn airport (HHN) and Aer Lingus operated around 13 weekly rotations to Frankfurt airport (FRA). These airports are considered as substitutable from the demand side (1108). The planned rotations for the summer 2007 season remain the same for Ryanair and increase to around 14 for Aer Lingus. Apart from the Merging Parties, Lufthansa operates a service between Dublin and Frankfurt airport with around 21 weekly rotations. Apart from Ryanair's entry in May 2005, there are no examples of a new entry to this route by other carrier in the recent years and the route has been traditionally operated by Aer Lingus and Lufthansa only.

950. The market shares of the Merging Parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in summer 2006 season are summarised in the following table. However, it must be noted that Lufthansa carries a major proportion of passengers connecting at Frankfurt airport to its medium- and long-haul services. On the other hand, Ryanair does not carry any connecting passengers and the share of Aer Lingus connecting passengers (amounting to around [0-15%]*) was significantly lower than in case of Lufthansa. Therefore, in line with the definition of the relevant market and to better reflect the competitive situation in the market for lights between Dublin and Frankfurt (and to distinguish it from e.g. flights between Dublin and New York, via Frankfurt), it is appropriate in case of this route to calculate the market share excluding connecting passengers as indicated in the last column of the table.

<emplacement tableau>

951. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Frankfurt route from three to two and would give the merged entity a very high combined market share. The very high combined market share of Ryanair and Aer Lingus on this route is in itself evidence of a dominant market position, in particular as only point-to-point passengers should be taken into account (1109). Moreover, the transaction would combine Ryanair and Aer Lingus as the closest competitors active on the route (1110). Even though Lufthansa serves the same airport in Frankfurt as Aer Lingus, its business model is clearly different from Aer Lingus and Ryanair. Lufthansa is primarily focused at feeding its medium and long-haul services at Frankfurt and the point-to-point passengers carried on the route form clearly only a minor part of its activities. For this purpose, Lufthansa operates a high frequency service with schedules accommodated according to the departure/arrival waves of at Frankfurt airport. Further, Lufthansa concentrates more on higher-yield business customers that it is the case for Aer Lingus (1111). Finally, according to the Customer Survey commissioned by the Commission, more Aer Lingus customers considered for flying Dublin - Frankfurt also Ryanair (33.3% of all Aer Lingus customers) than Lufthansa (20.4% of all Aer Lingus customers), even though Ryanair is flying to a different airport while Lufthansa to the same one. Further, 66.7% of Ryanair customers considered also Aer Lingus for flying Dublin - Frankfurt while only 9.3% of Ryanair passengers considered Lufthansa as an alternative. Therefore, the proposed merger would lead to the elimination of the competitive rivalry existing between the two Merging Parties, which has been an important source of competition on this route, in particular for point-topoint passengers between Dublin and Frankfurt. Something which both Ryanair and Aer Lingus are focusing.

952. The Dublin - Frankfurt route is subject to the general barriers to entry described in Section 7.8. Further, the Dublin - Frankfurt route is a very busy route both as regards the passenger numbers and the total number of frequencies operated by the Merging Parties. Therefore, any new entry would have to be of a substantial scale in terms of the operated capacity and frequencies to exert a competitive constraint on the merged entity. It is important for its competitiveness that any airline active on this route attracts a certain proportion of business or more time-sensitive passengers (1112) by offering flights at all the peak times of the day (1113). For this reason, the limited capacity in peak times at Dublin Airport, described in more detail in Section 7.8., as well as of the fully co-ordinated Frankfurt airport (1114) constitutes an additional significant barrier to entry. Frankfurt-Hahn airport, which is not capacity constraint, may be substitutable to the Frankfurt airport from a demand side but not necessarily from the supply side. In particular airlines with an existing hub or base at Frankfurt airport (such as Lufthansa) would in general not commence operations from Frankfurt-Hahn airport for a single route. Further, the Dublin - Frankfurt route is a rather mature route making entry by a new carrier with additional capacities even more demanding.

953. Ryanair argues that in particular competing carriers with bases or hubs at the destination airports are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Frankfurt airports: Lufthansa with more than 55 based aircraft, Condor Airlines, Air Berlin with 4 based aircraft (all of them at Frankfurt airport). However, on the basis of the Phase II investigation none of these carriers appears to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in case of a price increase by the merged entity. As already indicated, the current Lufthansa operations are focused primarily on feeding its hub. Even if they would be able to capture some additional higher fares passengers of the merged entity, it is unlikely that they would be able to provide significant additional competitive constraint to the merged entity with respect to the majority of their point-to-point customers and that they could offer low fares comparable to Ryanair and Aer Lingus with their cost structure. Even though Air Berlin is closer to the low-frills model of the Merging Parties, there are no indications that they would enter the Dublin - Frankfurt route even in case of a price increase by the merged entity (1115). Further, from the internal documents of Ryanair it seems that [...]* (1116). Condor Airlines, a charter airline that connects in particular Germany with a number of holiday destinations in the Mediterranean as well as elsewhere around the world (1117) cannot be regarded as a sufficiently likely entrant on the Dublin - Frankfurt route which would exert significant competitive constraints on the merged entity with a view to its different business model. Further, an important part of the passengers on this route originate in Ireland (1118) and any new entrant would therefore need to secure sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland (in particular Condor or Air Berlin), as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. It should also be noted that with 9 Ryanair aircraft based in Frankfurt - Hahn, the merged entity would have an advantage of having bases at both sides of this route, with resulting greater flexibility and cost benefits. Finally, apart from the two Merging Parties, there are no examples of new entry to the Dublin - Frankfurt route in the recent years.

954. As explained in detail in Sections 7.3. and 7.8., the remaining competing carriers based in Dublin Airport (CityJet and Aer Arann) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any other likely new entrant for this route not based in either Dublin or one of the Frankfurt airports. Therefore, the results of the market investigation demonstrate that the possibility of entry or expansion of capacity by the existing competitors is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on this route.

955. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Frankfurt route.

Dublin - Hamburg/Lübeck

956. The total number of passengers carried by scheduled airlines on the Dublin - Hamburg/Lübeck route in 2006 amounted to [100 000 - 150 000]* and currently a total of around 11 weekly rotations are operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 increased significantly (by around [30-40%]*) most probably due to entry of Ryanair to this route in June 2006. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (1119), as Ryanair itself acknowledged (1120), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1121), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

957. On the Dublin - Hamburg/Lübeck route, Ryanair and Aer Lingus overlap on a city-to-city basis. In the winter 2006/2007 season, Ryanair operated around 7 weekly rotations to Lübeck airport (LBC) and Aer Lingus operated around 4 weekly rotations to Hamburg airport (HAM). These airports are considered as substitutable from the demand side (1122). The planned weekly rotations for the summer 2007 season remain the same for both Merging Parties. Apart from the Merging Parties, there are no other competitors active on this route. In the past, Hapag-Lloyd Executive operated on the route (from Hamburg) as of April 2004 but exited in January 2006 (1123). Lufthansa operated the Dublin - Hamburg/Lübeck route in the past but exited in October 2000.

958. The market shares of the Merging Parties on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (1124).

<emplacement tableau>

959. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Hamburg/Lübeck route from two to one and would give the merged entity a 100% combined market share. The merger is thus a merger to monopoly which would eliminate all existing competition on this route (1125).

960. The Dublin - Hamburg/Lübeck route is subject to the general barriers to entry described in Section 7.8., even though in view of the limited number of frequencies and less important share of more time-sensitive passengers (1126), the peak time congestion at Dublin Airport may not be as significant a barrier to entry.

961. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity.

962. The following competing carriers have bases or hubs at Hamburg/Lübeck: Lufthansa with more than 10 based aircraft, Germanwings with 2 based aircraft and Air Berlin with 7 aircraft. On the basis of the Phase II investigation none of these carriers appears to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in case of a price increase by the merged entity. Lufthansa is a traditional network carrier operating hub and spoke system different to the point-to-point low-frills models of Ryanair and Aer Lingus. Further, Lufthansa already exited this route in 2000 when there was no other competitor on the direct Dublin - Hamburg/Lübeck route. Even in case Lufthansa would be willing to enter this route, it would not be able to provide a low frills alternative for customers and, in view of its business model, would be unlikely to represent a significant competitive constraint for the merged entity. Even though Germanwings and Air Berlin are closer to the low-frills model of the Merging Parties, there are no indications that they would enter the Dublin - Hamburg/Lübeck route even in case of a price increase by the merged entity - on the Contrary (1127). Further, from the internal documents of Ryanair it seems that [...]* (1128). It should be acknowledged that the share of passengers originating in Ireland on this route is significantly lower (1129) and thus the strength of the merged entity in Ireland represents a less significant barrier to entry compared to other routes. However, apart from the Merging Parties, there are no examples of sustained new entry of any comparable carrier to the Dublin - Hamburg/Lübeck route in recent years.

963. As explained in detail in Sections 7.3. and 7.8., the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Hamburg/Lübeck. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Hamburg/Lübeck route.

964. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Hamburg/Lübeck route.

Dublin - Krakow

965. The total number of passengers carried by scheduled airlines on the Dublin - Krakow route in 2006 amounted to [100 000 - 150 000]* and currently in total around 12 weekly rotations are operated on this route by scheduled carriers. The scheduled operations on this route commenced only at the end of the year 2005, when SkyEurope and Aer Lingus (both in October/November 2005) commenced their services. Ryanair entered the market in April 2006. The total numbers of passengers on this route are since then steadily growing and the route can be regarded as rather dynamic, in particular in view of the growing traffic between Poland and Ireland. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (1130), as Ryanair itself acknowledged (1131), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1132), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

966. On the Dublin - Krakow route, Ryanair and Aer Lingus overlap on an airport-to-airport basis. In the winter 2006/2007 season, Ryanair operated around 4 weekly rotations to the Krakow airport (KRK) and Aer Lingus operated around 5 weekly rotations to the same airport. Apart from the Merging Parties, SkyEurope operates a service between Dublin and Krakow airport with around 3 weekly rotations. The planned rotations for the summer 2007 season remain the same for all three carriers.

967. The market shares of the Merging Parties and their competitor on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (1133).

<emplacement tableau>

968. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Krakow route from three to two and would give the merged entity a very high combined market share. The very high combined market share of Ryanair and Aer Lingus on this route is in itself evidence of a dominant market position (1134). Moreover, the transaction would combine Ryanair and Aer Lingus, possibly the two closest competitors active on the route (1135). It is noted that SkyEurope is also a low-frills airline offering point-to-point services. However, SkyEurope has indicated that the merger would give the combined entity a market dominating position which would prevent SkyEurope from expanding even on routes which would be consistent with their business plan, including Dublin - Krakow (1136). Further, Ryanair itself in its internal documents states that SkyEurope [...]* (1137). It is thus doubtful, to what extend SkyEurope represents a credible competitive constraint to the combined entity in the short to mid-term, in particular in view of the fact that the combined entity would operate on the Dublin - Krakow route each day of the week while SkyEurope provides only three rotations per week. Even though a larger part of the passengers on this route originate in Ireland (1138), it should be acknowledged that part of this Irish originating traffic could be still generated by migrant Polish workers living in Ireland (1139) which could enhance the ability of SkyEurope to compete with the merged entity on this route. However, as explained in detail in Section 7.8., SkyEurope would consider expanding its operation only under certain conditions. The Commission therefore have serious doubts that SkyEurope would be able and willing to expand and compete effectively against the merged entity on this route on the basis of the notified operation. Therefore, the merger would lead to elimination of the competitive rivalry between the Merging Parties, which has been an important source of competition on this route and it is doubtful whether this loss could be compensated by the only remaining competitor SkyEurope.

969. The Dublin - Krakow route is subject to the general barriers to entry described in Section 7.8., even though in view of the limited number of frequencies and less important share of more time-sensitive passengers, the peak time congestion at Dublin Airport may not be a significant barrier to entry. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. However, except for SkyEurope with 3 based aircraft, which is already present on the route, there is no other competing carrier with a base in Krakow. Ryanair argues that Centralwings and Wizzair are expanding airlines which could enter the Dublin - Krakow route despite not having a base there. They argue that Centralwings already operate Shannon - Krakow (1140). It should be pointed out that Centralwings do not operate Shannon - Krakow but Cork - Krakow (1141). Further, even though Centralwings currently compete on three Irish routes (Dublin-Wroclaw, Dublin-Gdansk and Cork-Warsaw) with Ryanair, it was always Ryanair who entered the route against existing operations of Centralwings. It is therefore questionable whether Centralwings would be willing, without having a base at either end, to enter this route against the strong merged entity and be able to exercise significant competitive constraints. Wizzair is a low-frills carrier based mainly in Hungary and Poland and flying to a number of European destinations. However, apart from the recently started two flights to Cork from Katowice and Gdansk (where neither of the Merging Parties is active), they do not operate any flights to Dublin. It is therefore doubtful to what extent they would be able to exert efficient competitive constraints on the merged entity on the Dublin - Krakow route, without having a base at either end of the route. The fact that Wizzair "already works closely with FR [Ryanair] in the European Low Fares Airline Association and operates on numerous airports alongside FR [Ryanair]" (1142) cannot be regarded as mitigating factor against the identified barriers to entry in this case, in particular as Wizzair does not actually compete with Ryanair or Aer Lingus on any specific route to/from Ireland. Further, it should be noted that so far none of these airlines have entered a route between Poland and Ireland already served by Ryanair and that none so far has entered (at least between April 2003 and October 2006) any other intra-European route where Ryanair was already operating on the relevant airport-pair or city-pair (1143).

970. As explained in detail in Sections 7.3. and 7.8., the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Krakow. Therefore, the results of the market investigation demonstrate that the possibility of entry or expansion of capacity by the existing competitors is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Krakow route.

971. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Krakow route.

Dublin - Lyon

972. The total number of passengers carried by scheduled airlines on the Dublin - Lyon route in 2006 was [50 000 - 100 000]* and currently a total of around 6 weekly rotations are operated on this route by Ryanair and Aer Lingus. The route is relatively new from their perspective as Aer Lingus started its operations in April 2004 and Ryanair in December 2006. Even though the route has been quickly growing in the last two years, it is questionable whether there would be scope for further new entry as it is relatively thin with respect to the total number of passengers carried. The route has no significant seasonal pattern. Ryanair argues that charter carriers have significant operations on this route. As there is important charter traffic in the winter season on this route (1144), the possible competitive constraints on the merged entity on this route exercised by charter traffic are therefore analysed in more detail in paragraph 976.

973. On the Dublin - Lyon route, Ryanair and Aer Lingus overlap on a city-to-city basis. In the winter 2006/2007 season, Ryanair operated around 3 weekly rotations to Grenoble airport (GNB) while Aer Lingus operated around 3 weekly rotations to Lyon St Exupéry airport (LYS). These airports are considered as substitutable from the demand side (1145). The planned number of weekly rotations for the Summer 2007 season is around 2 for Ryanair and around 4 for Aer Lingus. Apart from the Merging Parties, there are no other scheduled carriers active on this route and there were no such carriers present at the time of Aer Lingus' entry in 2004. Apart from the Merging Parties, there have been no other new entrants on this route in the recent years.

974. The market shares of the Merging Parties on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (1146).

<emplacement tableau>

975. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Lyon route from two to one and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (1147).

976. In general, charter airlines are not considered to constitute a significant competitive constraint on Ryanair and Aer Lingus on the overlap routes (1148). However, as charter traffic carried a significant number of passengers on the Dublin - Lyon route in winter 2005/2006 and taking into account the arguments to the contrary raised by Ryanair with respect to this particular route, the Commission has explored whether charter airlines could exercise a competitive constraint on the Merging Parties on this route. It should be however noted that Ryanair entered the route only in December 2006 and thus the share of the charter traffic is likely to be lower than it was in 2006. In case the added capacity of Ryanair is taken into account (and even presuming hypothetically that entry of Ryanair would not take any passengers from the charter airlines), the share of total charter traffic would be less than 15% even in the winter season. In general, the charter operations are limited to the period between December and March (1149) and thus have no effects on the operations of the merged entity for the rest of the year. Apart from some ad-hoc one-off operations by other charter carriers, in particular Monarch Airlines operates on the Dublin - Lyon route (1150). This is also confirmed by the slot allocation for winter 2006/2007 provided by Ryanair where only Monarch Airlines have slots on more than only one or two days. As Monarch Airlines operate their flights solely on Saturdays while the merged entity would offer flights 5 to 6 days of the week it would thus provide significantly more flexibility and choice. Furthermore, the market investigation indicated that a large majority of seats of charter airlines (including Monarch Airlines (1151)) is not sold separately but rather as a part of package holidays sold to the final customers by the tour operators (and not by the charter airlines themselves) (1152). Even though Ryanair refers for example to FirstChoice's submission and claims that it offers "dry seats" on the Irish market (1153), it should be added that FirstChoice sells only 3% of its tickets in Ireland as dry seats and that 97% of its tickets are sold as part of package holidays (1154).There are no indications that this would be different for this particular route. Therefore, only a very small part of the less than 15% share of the charter traffic on this route were dry seats. In this regard the charter airlines (including Monarch Airlines) do not offer a possibility to book flights only for Dublin - Lyon via their website, even though it is possible to do so at the webpages of some tour operators selling their tickets in Ireland (1155). Most of the charter airlines sell their tickets to tour operators who then sell them either as part of the package holidays (which also means that it is in general the tour operator and not the charter airline who is deciding on which routes to operate and when (1156)) or to a much smaller extent as dry seats. Furthermore, in view of the decreasing charter market in Ireland in general and a decreasing number of passengers in this particular route (1157), the likelihood of a significant entry by new charter carriers with more flexible product offering is low (1158). Therefore, as only a small part of the less than 15% share of total charter traffic in the winter season would be constituted by dry seats and taking into account the limited period of charter operations, the difference in the product sold to the final customers and in the used distribution channels (tour operators) and the limited flexibility of charter flights, it cannot be considered that charter airlines would represent any significant competitive constraint on the merged entity on the Dublin - Lyon route.

977. The route is subject to the general barriers to entry described in Section 7.8. It should be, however, acknowledged that airport congestion at Dublin does not constitute a significant barrier to entry to this route as it is predominantly leisure flights on this route and they are not dependent on peak time slots and high frequencies. On the other hand, it should be noted that carriers focused on leisure routes also prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier. (1159)

978. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. According to Ryanair, Air France has a base in Toulouse Blagnac airport with 10 based aircraft. However, Air France has in Lyon only a regional hub primarily for its domestic flights. Therefore, also in view of the different business model of this full-frills network airline, Air France cannot be regarded as a credible potential entrant to the Dublin - Lyon route, which would provide significant competitive constraints to the merged entity, able to offer an alternative in particular for customers interested in low fares on this predominantly leisure route. Further, more than 50% of the passengers on this route originate in Ireland (1160) and any new entrant would therefore need to secure sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland, as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. Finally, apart from the Merging Parties, there are no examples of new entry to this route in the recent years.

979. There are also no indications that CityJet with its aircraft based in Dublin would be a credible entrant to the Dublin - Lyon route. As this route is primarily leisure oriented with a low share of business/more time-sensitive passengers (1161), it would not be in line with the business model of CityJet described in more detail in Section 7.8. Further, with its smaller planes CityJet would have difficulties to compete with the merged entity on this low-frequency route. For similar reasons, and as explained in detail in Sections 7.3. and 7.8., also the other competing carrier based in Dublin Airport (Aer Arann) is not a likely potential entrant to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Lyon. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Lyon route.

980. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Lyon route.

Dublin - Madrid

981. The total number of passengers carried by scheduled airlines on the Dublin - Madrid route in 2006 amounted to [200 000 - 250 000]* and currently in total between around 25 weekly rotations in winter and around 29 weekly rotations in summer are operated on this route by the scheduled carriers. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 increased by around [0-10%]* and was steadily growing also in the previous years, even though the growth rate is in general lower than on a majority of more dynamic overlap routes and the market is thus rather mature (also in view of the Ryanair's entry in December 2006 which may make any further new entry more difficult). The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (1162), as Ryanair itself acknowledged (1163), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1164), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular yearround operations of the merged entity on this route.

982. On the Dublin - Madrid route, Ryanair and Aer Lingus overlap on an airport-to-airport basis. In the winter 2006/2007 season, Ryanair operated around 7 weekly rotations to Madrid airport (MAD) and Aer Lingus operated around 11 weekly rotations to the same airport. The planned weekly rotations for the summer 2007 season remain the same for both Merging Parties. Apart from the Merging Parties, Iberia has traditionally operated on this route and it offers around 7 weekly rotations on this route in Winter 2006/2007 and around 11 weekly rotations in Summer 2007. Further, Spanair operated scheduled services between Dublin and Madrid in the summer seasons between of 2004-2006 but exited the route in October 2006 and does not have any scheduled services planned for Summer 2007. With the exception of Ryanair's entry in December 2006 and Spanair's purely seasonal operations, there have been no new entrants on this route in the recent years.

983. The market shares of the Merging Parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table. It must however be noted that Iberia carried a significant proportion of connecting passengers on this route. On the other hand, both Ryanair and Aer Lingus carry practically only point-to-point passengers (1165). Therefore, in line with the definition of the relevant market and to better reflect the competitive situation on this market, it is appropriate to calculate the market share excluding connecting passengers as indicated in the sixth column of the table. Further, in order to show the combined position of the Merging Parties in the point-to-point market after entry of Ryanair in December 2006, the last column of the table shows the market shares on the basis of capacity for the Summer 2007 season where the capacity corresponding to connecting passengers of Iberia is excluded on the presumption that the share of connecting passengers will stay as in Summer 2006.

<emplacement tableau>

984. As Spanair has left the route, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Madrid route from three to two and would give the merged entity a very high combined market share. The very high combined market share of Ryanair and Aer Lingus on this route is in itself evidence of a dominant market position, in particular as only point-to-point passengers should be taken into account (1166). Moreover, the transaction would combine Ryanair and Aer Lingus, the two closest competitors active on the route (1167). Iberia cannot be considered as a close competitor to the Merging Parties due to its business model based on full-service network operations, including significant proportion of connecting passengers on this route (see also the Section 7.4.). Further, as Clickair has currently a base only in Barcelona, it is not likely entrant on this route in the short to medium term. And even in case Clickair establishes in future a base in Madrid as well (1168), it is unlikely, for the same reasons as in case of the Dublin - Barcelona route (see paragraph 902), that it would provide a sufficient competitive constraints on the merged entity. Ryanair argues that "Iberia's low-cost subsidiary Clickair has recently commenced operating the Madrid route" (1169) and also that "the fact that Clickair now operates the Madrid-Dublin route is proof that full-service carriers can and do compete with FR [Ryanair] whether through their low cost subsidiaries or in their own right." (1170). However, these statements are factually incorrect as Clickair is not operating on the Dublin - Madrid route and even does not operate any scheduled flights to/from Madrid as its main base is Barcelona (1171). However, with respect to the entry of Clickair instead of Iberia on the Dublin - Barcelona route, Iberia indicated that the traffic is extremely sensitive to price and that in order to be profitable on such a route, a lower cost base is needed (1172). It also indicated that that route is rather seasonal and thus that a larger flexibility is necessary to adapt the supply to the demand. This seems to confirm that Iberia itself was less able to compete on its own right with the existing carriers on that route (Ryanair and Aer Lingus). Ryanair further argues that "Iberia is more likely EI's [Aer Lingus'] closest competitor on the Madrid - Dublin route given that the services offered by each airport is far more similar to that of FR [Ryanair].". This argument is however not clear as all the carriers on this route use the same airports (DUB and MAD). Therefore, the merger would lead to elimination of the existing competitive rivalry between the Merging Parties which has been an important source of competition on this route, in particular in case of point-to-point passengers between Dublin and Madrid on which both Ryanair and Aer Lingus focus.

985. The Dublin - Madrid route is subject to the general barriers to entry described in Section 7.8. As regards in particular access to airport infrastructure issues, it is noteworthy that Madrid is a fully coordinated airport under the Slot Regulation (1173) while Dublin Airport's capacity is limited during peak times, as described in more detail in Section 7.8. It should be however acknowledged that the Dublin - Madrid route is not the type of high frequency route and peak time congestion may not be that significant barrier to entry. On the other hand, it should be noted that carriers focused on leisure routes also prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier. (1174)

986. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity.

987. The following competing carriers have bases or hubs at Madrid airport: Iberia with more than 30 aircraft, Spanair with more than 5 aircraft, Vueling with more than 3 aircraft and easyJet with more than 3 aircraft. However, on the basis of the Phase II investigation none of these carriers appears to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in case of a price increase by the merged entity. Iberia has already been discussed in paragraph 984. In view of its recent exit from the Dublin - Madrid (as well as from the Dublin - Barcelona and Dublin - Tenerife routes), it is unlikely that Spanair would constitute a credible potential entrant to the Dublin - Madrid route in the short to mid-term. In addition, Spanair is a full-frills scheduled as well as charter airline and part to the Star Alliance. Even though it plans a major expansion in the next years, its aim is "to become the leading airline for business travel" (1175). Further, this expansion plan is focused on Barcelona and not Madrid. As regards Vueling, they consider itself as a low-fares airline and are based in mainly in Barcelona (with additional base in Madrid) serving in total 18 destinations, mainly in the Mediterranean (plus Paris, Brussels and Amsterdam). Vueling does not operate (and never operated) any flights to/from Ireland and does not plan to enter Ireland "any time soon" (1176). As explained in more detail in Section 7.8., also easyJet cannot be considered as a credible potential entrant on this route in the short to medium term. It should be also noted that when easyJet set up a base in Madrid in February 2007 and launched a number of new routes, it actually newly entered only one route in competition with Ryanair (Madrid - Rome) (1177). Ryanair argues that easyJet already overlaps with Ryanair on the Madrid-Paris and Madrid-London routes and claims that there is no reason it could not enter the Dublin-Madrid route (1178). However, while easyJet has a well established activities with a base at these destinations (Paris and London), it is not active in Ireland at all. Further, as explained in detail in Section 7.8., easyJet indicated that any entry to Dublin would have to be on a sufficient scale and require important investments. Further, around 50% of the passengers on this route originate in Ireland (1179) and any new entrant would therefore need to secure sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland (such as Clickair, easyJet and Vueling, however including also Spanair with only marginal presence in Ireland), as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. It should also be noted that with 4 Ryanair aircraft based in Madrid, the merged entity would have an advantage of having bases at both sides of this route, with resulting greater flexibility and cost benefits. Finally, there are no examples of sustained new entry to the Dublin - Madrid route in the recent years.

988. As explained in detail in Sections 7.3. and 7.8., the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Further, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Madrid airports. Therefore, the results of the market investigation demonstrate that the possibility of entry or expansion of capacity by the existing competitors is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Madrid route.

989. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Madrid route.

Dublin - Malaga

990. The total number of passengers carried by scheduled airlines on the Dublin - Malaga route in 2006 amounted to [350 000 - 400 000]* and currently a total of around 14 weekly rotations are operated on this route by scheduled carriers in the winter season and around 26 in the summer season. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 increased by around [30-40%]* and there was also significant growth in previous years. As evidenced also by the almost twice as many rotations operated by these carriers during the summer compared to the winter season, the route has strong seasonal characteristics with substantially higher traffic volumes during the summer season. It is also noted that charter traffic represented in 2006 around [10- 20%]* of the total (charter and scheduled) passengers carried on this route (and even [20- 30%]* in the summer season) (1180). For this reason, the possible competitive constraints on the merged entity on this route exercised by charter traffic are analysed in more detail in paragraph 994.

991. On the Dublin - Malaga route, Ryanair and Aer Lingus overlap on an airport-to-airport basis. In the winter 2006/2007 season, Ryanair operated around 5 weekly rotations to Malaga airport (AGP) and Aer Lingus operated around 9 weekly rotations to the same airport. In the summer 2007 season, Ryanair plans to operate around 10 weekly rotations and Aer Lingus around 14 weekly rotations on this route. Apart from the Merging Parties, Spanair operates as of February 2005 a service between Dublin and Malaga during summer only, with around 2 weekly rotations planned for the Sumer 2007 season. Further, CityJet operated traditionally on the route, but exited in November 2005. Except for the seasonal service of Spanair, there have been no other new entrants on this route in the recent years. Even though Ryanair argues that Jet2 is an actual competitor offering scheduled flights between Malaga and Belfast (Northern Ireland) (1181), the Commission has found that Belfast cannot be regarded as substitutable with Dublin for the purpose of the notified operation (1182) and thus Jet2 cannot be viewed as a competitor on this route providing a competitive constraint to the merged entity.

992. The market shares of the Merging Parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (1183).

<emplacement tableau>

993. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Frankfurt route from three to two and would give the merged entity a very high combined market share. The very high combined market share of Ryanair and Aer Lingus on this route is in itself evidence of a dominant market position (1184). Moreover, the transaction would combine Ryanair and Aer Lingus, the two closest competitors active on the route (1185). While the Merging Parties operate on the route throughout the year, Spanair is active only during the summer season. Further, Spanair's operations on the route are limited to weekend frequencies similar to charter airlines (1186) while the merged entity would during the summer offer around 3 rotations each day of the week. Finally, Spanair is a full-frills scheduled as well as charter airline and part to the Star Alliance. Even though they plan a major expansion in the next years, their aim is "to become the leading airline for business travel" and their expansion plan is mainly focused on Barcelona (1187). Therefore, in view of its limited operations and different type of services compared to the Merging Parties, Spanair cannot be considered as a significant constraint to the merged entity on this route. The merger would thus lead to the elimination of the competitive rivalry existing between the Merging Parties, which has been an important source of competition on this route.

994. In general, charter airlines are not considered to constitute a significant competitive constraint on Ryanair and Aer Lingus on the overlap routes (1188). However, as charter traffic represents an important share of passengers carried on the Dublin - Malaga route in the summer season and taking into account the arguments to the contrary raised by Ryanair with respect to this particular route, the Commission has explored whether charter airlines could exercise a competitive constraint on the Merging Parties on this route. In particular the following charter airlines operate on this route: Air Futura, Air Europa, Eirjet (which however went bankrupt in October 2006), FirstChoice Airways, Flightline and Spanair (1189). It should be noted that despite a peak in the summer season, the charter airlines are present on this route with important capacities throughout the year. However, the market investigation indicated that a large majority of seats of charter airlines in Ireland were not sold separately but rather as a part of a package holiday, sold to final customers by tour operators and not by the charter airlines themselves (1190). Even though Ryanair refers for example to FirstChoice's submission and claims that it offers "dry seats" on the Irish market (1191), it should be added that FirstChoice sells only 3% of its tickets in Ireland as dry seats and that 97% of its tickets are sold as part of package holidays (1192). There are no indications that this would be different for this particular route. Therefore, only a very small part of the [20-30%]* share of the charter traffic on this route were dry seats. Further, none of these charter airlines offered a possibility to book flights only for Dublin - Malaga via their website, even though it is possible to do so at the webpages of some tour operators selling their tickets in Ireland (1193). Most of the charter airlines sell their tickets to tour operators who then sell them either as part of the package holidays (which also means that it is in general the tour operator and not the charter airline who is deciding on which routes to operate and when (1194)) or to a much smaller extent as dry seats. It should also be noted that the operations of the charter companies on this route in summer 2006 were concentrated on Saturdays and Sundays only (1195). The merged entity would offer a significantly more flexible timetable with up to 3 rotations each day of the week. Furthermore, in view of the decreasing charter market in Ireland in general as well as in this particular route (1196), the likelihood of any significant entry by new charter carriers with a more flexible product offering is low (1197). Therefore, as only a small part of the [20- 30%]* share of charter traffic was constituted by dry seats and taking into account the difference in the product sold to the final customers and in the used distribution channels (tour operators) and the limited flexibility of the charter flights, charter airlines cannot be considered to represent any significant competitive constraint for the merged entity on the Dublin - Malaga route.

995. The Dublin - Malaga route is subject to the general barriers to entry described in Section 7.8. However, it should be acknowledged that airport congestion does not constitute a significant barrier to entry to this route as the predominantly holiday flights on the route are not dependent on peak time slots and high frequencies. On the other hand, it should be noted that carriers focused on leisure routes also prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier. (1198) Further, an absolute majority of the passengers on this route originate in Ireland (1199) and any new entrant would therefore need to secure sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland, as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland.

996. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. However, only Air Europa has up to 2 aircraft based at Malaga airport. Air Europa is a full service scheduled as well as charter airline operating domestic routes in Spain as well as some international short-haul and long-haul routes (1200). It is already operating charter flights to Ireland (including on Dublin - Malaga route), which are however sold exclusively by tour operators and it has no scheduled flights to Ireland. Therefore, Air Europa cannot be regarded as a credible new entrant to the Dublin - Malaga route for scheduled air transport services and which would be able to provide a significant competitive constraints to the merged entity. As explained in detail in Sections 7.3. and 7.8., the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Malaga. Therefore, the results of the market investigation demonstrate that the possibility of entry or expansion of capacity by the existing competitors is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anticompetitive effects of the merger on the Dublin - Malaga route.

997. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Malaga route.

Dublin - Marseille

998. The total number of passengers carried by scheduled airlines on the Dublin - Marseille route in 2006 was [50 000 - 100 000]* and currently a total of around 4 weekly rotations in the winter season and around 7 weekly rotations in the summer season are operated on this route by Ryanair and Aer Lingus. The route is relatively new for these carriers as Aer Lingus started its operations in March 2005 and Ryanair in April 2006. Even though the route has been growing rapidly in the last two years, it could be questioned whether there is scope for any further substantial entry as it is a relatively thin route with respect to the total number of passengers carried. Currently, Ryanair operates the route throughout the whole year whilst Aer Lingus offers flights only in the summer season. The presence of charter carriers on this route is negligible (1201), as Ryanair itself acknowledged (1202), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1203), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

999. On the Dublin - Marseille route, Ryanair and Aer Lingus overlap on an airport-to-airport basis and only during the summer season. In the winter 2006/2007 season, Ryanair operated around 4 weekly rotations to Marseille airport (MRS) while Aer Lingus is not active. The planned number of weekly rotations for the Summer 2007 season for Ryanair remains the same while Aer Lingus plans to operate around 3 weekly rotations also to Marseille airport. Apart from the Merging Parties, there are no other scheduled carriers active on this route and there were no such carriers present at the time of Aer Lingus' entry in 2005. Apart from the Merging Parties, there have been no other new entrants on this route in the recent years.

1000. The market shares of the Merging Parties on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (1204).

<emplacement tableau>

1001. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Marseille route from two to one and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (1205).

1002. The Dublin - Marseille route is subject to the general barriers to entry described in Section 7.8. However, it should be acknowledged that airport congestion at Dublin does not constitute the most significant barrier to entry to this route as the predominantly leisure flights on this route are not dependent on peak time slots and high frequencies. On the other hand, it should be noted that carriers focused on leisure routes also prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier (1206).

1003. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. According to Ryanair, Air France has a base at Marseille airport with 1 aircraft based there. However, Air France has its main hub at Paris CDG, its main domestic hub at Paris Orly and regional hubs at Lyon, Clermont-Fernand and Bordeaux (1207). To the extent they have any aircraft based in Marseille, these are primarily used for domestic flights and feeding the hubs. Therefore, also in view of the different business model of this full-frills network airline, Air France cannot be regarded as a credible potential entrant to the Dublin - Marseille route, which would provide significant competitive constraints to the merged entity, able to offer a comparable low-fare alternative to Ryanair/Aer Lingus. Further, more than 50% of the passengers on this route originate in Ireland (1208) and any new entrant would therefore need to secure a sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland, as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. It should also be noted that with 1 Ryanair aircraft based in Marseille, the merged entity would have an advantage of having bases at both sides of this route, with resulting greater flexibility and cost benefits. Finally, apart from the Merging Parties, there are no examples of new entry to this route in the recent years.

1004. There are also no indications that CityJet with its aircraft based in Dublin would be a credible entrant to the Dublin - Marseille route. As this route is primarily leisure oriented with a low share of business/more time-sensitive passengers (1209), it would not be in line with the business model of CityJet described in more detail in Section 7.8.. Further, with its smaller planes CityJet would have difficulties to compete with the merged entity on this low-frequency route. For the similar reasons, and as explained in detail in Sections 7.3. and 7.8., also the other competing carrier based in Dublin Airport (Aer Arann) is not a likely potential entrant to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Marseille. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Marseille route.

1005. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Marseille route.

Dublin - Milan

1006. The total number of passengers carried by scheduled airlines on the Dublin - Milan route in 2006 amounted to [250 000 - 300 000]* and currently a total of up to 21 weekly rotations are operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 increased by [30- 40%]*, probably due to Ryanair's entry on this route in April 2006. Also the year before the passenger number grew significantly again probably due to the fact that Aer Lingus opened a route to a second Milan airport (Bergamo) in March 2005. However, despite these dynamic developments in recent years, it could be questioned whether there would be scope for further significant entry. Ryanair argues that the Dublin - Milan route has strong seasonal characteristics. Even though there are some seasonal peaks, it cannot be regarded as a typically seasonal route as there is important traffic throughout the year. For illustration, in 2006 around 60% of passengers were carried in the six "summer months" (May to October) and around 40% in the six remaining "winter months". Ryanair also argues that charter airlines are active on this route. However, according to the available passenger data, their presence on this route is marginal (1210) and therefore does not need to be discussed separately as it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair1211, ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

1007. On the Dublin - Milan route, Ryanair and Aer Lingus overlap on a city-to-city basis. In winter 2006/2007 season, Ryanair operated around 7 weekly rotations to Milan - Bergamo airport (BGY) while Aer Lingus operated around 7 weekly rotations to Milan - Linate airport (LIN) and around 3 weekly rotations to Milan - Malpensa airport (MXP). These airports are considered as substitutable from the demand side (1212). The planned weekly rotations for the summer 2007 season remain the same for Ryanair and for Aer Lingus at Milan - Linate and increase to around 7 weekly rotations for Aer Lingus to Milan - Malpensa. Apart from the Merging Parties, there are no other competitors active on this route. Traditionally, this route was served by Alitalia flying between Dublin and Milan - Malpensa (and Milan - Linate before 1999). However, Alitalia left the route as of October 2006, only a few months after Ryanair entered the route. Apart from the Ryanair's entry to Milan - Bergamo in April 2006 and Aer Lingus additional flights to Milan - Bergamo as of March 2005 in addition to its existing Milan - Linate flights (and subsequently switched to Milan - Malpensa as of October 2006), there were no other entrants to this route in recent years.

1008. The market shares of the Merging Parties, on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (1213).

<emplacement tableau>

1009. As Alitalia has left the route, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Milan route from two to one and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (1214).

1010. The Dublin - Milan route is subject to the general barriers to entry described in Section 7.8. As regards in particular access to airport infrastructure issues, it is noteworthy that both Milan - Linate (congested throughout the day) and Milan - Malpensa (congested at peak times) are co-ordinated airports under the Slot Regulation (1215). In combination with the limited capacity of Dublin Airport in the peak times, described in more detail in Section 7.8., this would constitute an additional barrier to any entry during attractive peak hours. Even though Milan - Bergamo is not congested, it would not be any real alternative for some airlines (such as Alitalia, easyJet or AirOne) with hubs/bases at Linate or Malpensa.

1011. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Milan airports: Alitalia with more than 15 based aircraft, Air One with up to 4 based aircraft, easyJet with 3 based aircraft and MyAir with 5 based aircraft. However, on the basis of the Phase II investigation none of these carriers appears to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in case of a price increase by the merged entity. In view of its recent exit from the Dublin - Milan (as well as from the Dublin - Rome route - see below) and of its current financial situation (1216), it is unlikely that Alitalia would constitute a credible potential entrant to the Dublin - Milan route in the short to mid-term. Further, Alitalia is a full-frills network carrier which would anyway not be a close competitor to the Merging Parties. As regards Air One, its activities are in particular focused on domestic Italian routes and routes from Italy to Mediterranean holiday destinations. Even though it operates also flights to some European cities (such as to London-City, Paris or Copenhagen), its focus is clearly Italy and Southern Europe. Further, Air One is a full service carrier offering business class on its international routes and code-sharing with Lufthansa and other network carriers (1217). Therefore, Air One cannot be regarded as a likely potential entrant on the Dublin - Milan route which could provide sufficient competitive constraints to the merged entity. MyAir is an Italian low-frills airline operating a network of domestic routes as well as "a tailored and carefully planned international network" (1218). Its international destinations are mainly in the Mediterranean, in Eastern Europe and in France. They have no flights to Ireland (or the UK). Therefore, in view of MyAir's regional focus it does not seem likely that they would be prepared to enter the Dublin - Milan route and provide sufficient competitive constraints to the merged entity. As explained in more detail in Section 7.8., also easyJet cannot be considered as a credible potential entrant on this route in the short to medium term. In addition, easyJet is based at the fully coordinated Milan - Malpensa airport with peak time congestion. It should be also noted that when easyJet set up a base in Milan in March 2006 and launched a number of new routes, it actually did not enter any new route in competition with Ryanair (1219). Further, considering the recent failed attempt by easyJet to enter the Irish market it is unlikely that it would enter this route where (similarly as in case of Shannon- London and Cork-London routes) Ryanair has a base at both ends of the route. Further, up to 50% of the passengers on this route originate in Ireland (1220) and any new entrant would therefore need to secure sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland (such as Alitalia, easyJet and in particular the outside Italy less known Air One and MyAir), as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. As already mentioned, with 4 Ryanair aircraft based in Milan-Bergamo the merged entity would have an advantage of having bases at both sides of this route, with resulting greater flexibility and cost benefits. Finally, apart from the parties, there are no examples of new entry to the Dublin - Milan route in the recent years.

1012. As explained in detail in Sections 7.3. and 7.8., the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Milan airports. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Milan route.

1013. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Milan route.

Dublin - Paris

1014. The total number of passengers carried by scheduled airlines on the Dublin - Paris route in 2006 amounted to [800 000 - 850 000]* and currently a total of around 83 weekly rotations are operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006 compared to 2005 increased by [0-10%]* and was rather stable in the previous years. Therefore the route can be regarded as relatively mature. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (1221), as Ryanair itself acknowledged (1222), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1223), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

1015. On the Dublin - Paris route, Ryanair and Aer Lingus overlap on a city-to-city basis. In the winter 2006/2007 season Ryanair operated around 16 weekly rotations to Beauvais airport (BVA) and Aer Lingus operated around 21 weekly rotations to Paris Charles de Gaulle airport (CDG). These airports are considered as substitutable from the demand side, together with Paris Orly airport (ORY)1224. The planned rotations for the summer 2007 season remain the same for both Ryanair and Aer Lingus. Apart from the Merging Parties, CityJet operates a service between Dublin and Paris CDG airport with around 46 weekly rotations. There are no examples of an attempted new entry to this route by other carrier in the recent years.

1016. The market shares of the Merging Parties and City Jet on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table. It must however be noted that CityJet carries a significant proportion of passengers connecting at Paris CDG to medium- and long-haul services of Air France. On the other hand, both Ryanair and Aer Lingus carry almost only (1225) point-to-point passengers. Therefore, in line with the definition of the relevant market and to better reflect the competitive situation on this market (and to distinguish it from e.g. flights between Dublin and New York, via Paris), it is appropriate also to calculate the market share excluding connecting passengers as indicated in the last column of the table.

<emplacement tableau>

1017. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Paris route from three to two and would give the merged entity a very high combined market share. The very high combined market share of Ryanair and Aer Lingus on this route is in itself evidence of a dominant market position, in particular as only point-to-point passengers should be taken into account (1226). Moreover, the transaction would combine Ryanair and Aer Lingus, the two closest competitors active on the route (1227). Even though CityJet serves the same airport in Paris as Aer Lingus, its business model is clearly different from Aer Lingus and Ryanair. CityJet, which is a 100% subsidiary of Air France, is primarily focused at feeding the medium and long-haul services of Air France at Paris CDG. For this purpose, CityJet operates a high frequency service with schedules accommodated according to the departure/arrival waves of Air France at Paris CDG. As regards point-to-point passengers, CityJet concentrates on higheryield business customers. CityJet operates substantially smaller aircraft than Ryanair/Aer Lingus (1228) and thus would not be able to compete with them efficiently for the lower fares segments of customers. Further, the fact that the Merging Parties are the closest competitors on this route is confirmed by the Customer Survey conducted by the Commission. According to the survey 37.2% of Ryanair's customers considered Aer Lingus as an alternative (even though Ryanair flies to a different airport) and only 4.3% considered any other airline, in particular CityJet. And vice versa, 25.9% of Aer Lingus' passengers considered Ryanair as an alternative and only 11.1% considered any other airline (1229). Therefore, the merger would lead to elimination of the existing competitive rivalry between the Merging Parties' which has been an important source of competition on this route, in particular in case of point-to-point passengers between Dublin and Paris on which both Ryanair and Aer Lingus focus.

1018. The Dublin - Paris route is subject to the general barriers to entry described in Section 7.8.. Further, the Dublin - Paris route is a very busy route both as regards the passenger numbers and the total number of frequencies operated by the Merging Parties. Therefore, any new entry would have to be of a substantial scale in terms of the operated capacity and frequencies to exert a competitive constraint on the merged entity. It is important for its competitiveness that any airline active on this route attracts a certain proportion of business or more time-sensitive passengers (1230) by offering flights at all the peak times of the day. For this reason, the limited capacity during peak times at Dublin Airport, described in more detail in Section 7.8., as well as of the slot constraints at the fully coordinated Paris CDG and Paris Orly airports constitutes an additional significant barrier to entry (1231). Beauvais airport, which is not capacity constraint, may be substitutable to the two Paris airports from the demand side but less so from the supply side. Therefore, in particular airlines feeding their hubs (such as Air France/CityJet) or having bases (such as easyJet) at the Paris CDG or Paris Orly airport would in general find it more difficult to start operations from Beauvais airport. Further, Dublin - Paris route is a mature route making entry by a new carrier with additional capacities even more demanding.

1019. Ryanair argues that in particular competing carriers with bases or hubs at the destination airports are able to easily enter the Dublin - Paris route in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Paris airports: Air France/CityJet with more than 70 based aircraft at Paris CDG and Paris Orly, easyJet with 5 based aircraft at Paris Orly. However, on the basis of the Phase II investigation none of these carriers appears to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in case of a price increase by the merged entity. As already indicated, the current Air France/CityJet operations are focused on feeding Air France services and on business point-to-point passengers. Even if they would be able to capture some additional higher yield passengers from the merged entity, it is unlikely that they would be able to provide any significant additional competitive constraint to the merged entity with respect to the majority of the merged entity's customers. Further, taking into account the smaller aircraft operated by CityJet and the capacity constraints at Dublin as well as Paris CDG and Paris Orly, it is not likely that any expansion of Air France/CityJet would provide significant constraints to the merged entity and would be able to offer a low-fares alternative for Ryanair's/Aer Lingus' customers. As explained in more detail in Section 7.8., also easyJet cannot be considered as a credible potential entrant on this route in the short to medium term. In addition, easyJet is based at the coordinated Paris Orly airport and would have to sacrifice a substantial number of slots used for other routes in order to enter the Dublin - Paris market with sufficient number of frequencies. Further, the majority of the passengers on this route originate in Ireland (1232) and any new entrant would therefore need to secure a sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland (in particular easyJet), as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. Finally, apart from the Merging Parties, there are no examples of new entry to this route in the recent years.

1020. As explained in detail in Sections 7.3. and 7.8., the remaining competing carrier based in Dublin Airport (Aer Arann) is not a likely potential entrant on this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or one of the Paris airports. Therefore, the results of the market investigation demonstrate that the possibility of entry or expansion of capacity by the existing competitors is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Paris route.

1021. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Paris route.

Dublin - Poznan

1022. The total number of passengers carried by scheduled airlines on the Dublin - Poznan route in 2006 amounted to [50 000 - 100 000]* and currently a total of around 5 weekly rotations are operated on this route by the Merging Parties. The scheduled operations on this route commenced only in April 2006, when both Ryanair and Aer Lingus commenced their services between Dublin and Poznan. The route can be regarded as rather new and dynamic, in particular in view of the growing immigrant based traffic between Poland and Ireland. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (1233), as Ryanair itself acknowledged (1234), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1235), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular yearround operations of the merged entity on this route.

1023. On the Dublin - Poznan route, Ryanair and Aer Lingus overlap on an airport-to-airport basis. In both winter 2006/2007 and summer 2007 season, Ryanair operates around 3 weekly rotations to the Poznan airport (POZ) and Aer Lingus operates 2 weekly rotations to the same airport. Apart from the Merging Parties, there are no competing airlines on this route and there were no new entrants to this route in the recent years.

1024. The market shares of the Merging Parties on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (1236).

<emplacement tableau>

1025. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Poznan route from two to one and would give the merged entity a 100% combined market share. Therefore, the merger would eliminate all existing competition on this route (1237).

1026. The Dublin - Poznan route is subject to the general barriers to entry described in Section 7.8., even though in view of the limited number of frequencies and less important share of more time-sensitive passengers, the peak time congestion at Dublin Airport may not be in itself a significant barrier to entry. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. However, there are no competing airlines based at Poznan airport. Even though Ryanair argues that the service could be operated on a W-basis (1238), a new entry on this basis is less likely and efficient as discussed in Section 7.8.2. Further, locating an aircraft to Poznan only to serve Dublin - Poznan route does not seem plausible and thus would also depend on other possible routes to be served by that aircraft out of Poznan. As explained in detail in Sections 7.3. and 7.8., the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route.

1027. Ryanair argues that Wizzair is well-established at Poznan airport (1239). Wizzair is a low-frills carrier based mainly in Hungary and Poland and flying to a number of European destinations. However, apart from the recently started two flights to Cork from Katowice and Gdansk (where neither of the Merging Parties is active), they do not operate any flights to Dublin and there are no examples of entry by this carrier on any intra-European city-pair or airport-pair served by Ryanair in the recent years (1240). It is therefore doubtful to what extent they would be able and willing to enter and exert efficient competitive constraints on the merged entity on this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Poznan. Even though some operators already active on the routes between Poland and Ireland, in particular SkyEurope, could possibly enter the route, the market investigation indicated (1241) that such entry against the merged entity would not appear likely, also in view of the merged entity's strong existing position on the route and Ryanair's reputation with regard to reaction to new entrants.

1028. Ryanair further argues that the route is "already marketed and advertised extensively by both EI [Aer Lingus] and FR [Ryanair] (the latter on Europe's most popular travel website) means that any new entrant would have no additional expense in promoting the route or making the public aware of it" (1242). However, this also means that any new entrant would have to at least match this extensive marketing and advertising to attract sufficient numbers of passengers to buy its tickets at its website or other distribution channels and not on the merged entity's website. Therefore, even though the route is growing rather dynamically, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anticompetitive effects of the merger on the Dublin - Poznan route.

1029. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Poznan route.

Dublin - Riga

1030. The total number of passengers carried by scheduled airlines on the Dublin - Riga route in 2006 amounted to [150 000 - 200 000]* and currently a total of up to 15 weekly rotations are operated on this route by scheduled carriers. The scheduled operations on this route commenced only in April 2004, when Air Baltic started operating between Riga and Dublin. Subsequently, Aer Lingus commenced services on this route in November 2005 and Ryanair in January 2006. In this period, the total numbers of passengers on the route was growing significantly and the route can be regarded as rather dynamic, even though it could be questionable whether there would be scope for further new entry. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (1243), as Ryanair itself acknowledged (1244), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1245), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

1031. On the Dublin - Riga route, Ryanair and Aer Lingus overlap on an airport-to-airport basis. In winter 2006/2007 season, Ryanair operated around 5 weekly rotations to Riga airport (RIX) and Aer Lingus operated around 3 weekly rotations to the same airport. Apart from the Merging Parties, Air Baltic operated a service between Dublin and Riga airport with 4 to 7 weekly rotations. The planned rotations for the summer 2007 are around 7 for Ryanair, around 5 for Aer Lingus and again between 4 to 7 for Air Baltic. Apart from the three existing carriers, there was no new entry on this route in recent years.

1032. The market shares of the Merging Parties and Air Baltic on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (1246).

<emplacement tableau>

1033. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Riga route from three to two and would give the merged entity a very high combined market share. The very high combined market share of Ryanair and Aer Lingus on this route is in itself evidence of a dominant market position (1247). Moreover, the transaction would combine Ryanair and Aer Lingus, the two closest competitors active on the route (1248). Air Baltic is a full-frills airline with two class cabin operating a hub in Riga and an additional base in Vilnius (1249). The results of the market investigation suggest that Air Baltic is neither capable nor willing to expand its operations on the route against the combined entity (1250). It is thus doubtful, to what extend Air Baltic represents a credible competitive constraint to the combined entity in the short to mid-term, in particular in view of the fact that the combined entity would operate on the Dublin - Riga route in summer 2007 almost twice as many rotations as Air Baltic. Further, a larger part of the passengers on this route originate in Ireland (1251) and in order to expand, Air Baltic would therefore need to secure a sufficient number of Irish originating passengers. Further, the fact that the Merging Parties are the closest competitors on this route is confirmed by the Customer Survey conducted by the Commission. According to the survey 36.4% of Ryanair's customers considered Aer Lingus as an alternative and only 18.2% considered any other airline, in particular Air Baltic. And vice versa, 50.0% of Aer Lingus' passengers considered Ryanair as an alternative and only 14.8% considered any other airline (1252). Ryanair argues that the services of Ryanair and Aer Lingus are sufficiently differentiated in terms of the product being offered that the fares are different. However, in view of the findings set out in Section 7.4. and the fact that both carriers fly to the same airports on this route, it is not clear what this sufficient product differentiation refers to. As regards the fare difference, it should be stressed again that the fare data for Aer Lingus provided by Ryanair in the Form CO on the basis of MIDT/OAG data grossly overstate the real average fare of Aer Lingus and cannot be regarded as reliable (1253). Therefore, the merger would lead to the elimination of the competitive rivalry between the Merging Parties which has been an important source of competition on this route and it is doubtful whether this loss could be compensated by the only remaining competitor Air Baltic.

1034. The Dublin - Riga route is subject to the general barriers to entry described in Section 7.8., even though in view of the limited number of frequencies and less important share of more time-sensitive passengers, the peak time congestion at Dublin Airport may not be the most significant barrier to entry. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. However, except Air Baltic with around 8 based aircraft which is already present on the route and discussed in paragraph 10331033, there is no other competing carrier with a base in Riga. As explained in detail in Sections 7.3. and 7.8., the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Riga. Therefore, even though the route is growing rather dynamically, the results of the market investigation demonstrate that the possibility of entry or expansion of capacity by the existing competitors is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anticompetitive effects of the merger on the Dublin - Riga route.

1035. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Riga route.

Dublin - Rome

1036. The total number of passengers carried by scheduled airlines on the Dublin - Rome route in 2006 amounted to [250 000 - 300 000]* and currently a total of 20 weekly rotations are operated on this route by the Merging Parties. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 increased by less than [0- 10%]* and therefore the route can be regarded as mature. The route has no significant seasonal pattern. The presence of charter carriers on this route is only marginal (1254), as Ryanair itself acknowledged (1255), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1256), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

1037. On the Dublin - Rome route, Ryanair and Aer Lingus overlap on a city-to-city basis. In the winter 2006/2007 season, Ryanair operated around 10 weekly rotations to Rome - Ciampino airport (CIA) and Aer Lingus operated around 10 weekly rotations to Rome - Fiumicino airport (FCO). These airports are considered as substitutable from the demand Side (1257). The planned weekly rotations for the summer 2007 season are around 8 for Ryanair and around 11 for Aer Lingus. Apart from the Merging Parties, there are no other competitors active on this route. Traditionally, this route was served by Alitalia flying between Dublin and Rome Fiumicino. However, Alitalia left the route as of October 2005 (1258) and offers currently only indirect flights via Paris CDG on the basis of codesharing with Air France. As specified in the Market Definition part (Section 6.5.), indirect flights cannot be considered to constitute part of the relevant market in case of short haul services (1259).

1038. The market shares of the Merging Parties and Alitalia on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (1260).

<emplacement tableau>

1039. As Alitalia has left the route, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Rome route from two to one and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (1261).

1040. The Dublin - Rome route is subject to general barriers to entry described in Section 7.8. As regards in particular access to airport infrastructure issues, it is noteworthy that both Rome - Fiumicino and Rome - Ciampino are coordinated airports under the Slot Regulation even though neither of them seems to be significantly capacity constraint (1262). However, the limited capacity of Dublin Airport during peak times, described in more detail in Section 7.8., would constitute an additional barrier to an entry with sufficient number of frequencies. Further, Dublin - Rome route is a mature route making entry by a new carrier with additional capacities even more demanding.

1041. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to easily enter the Dublin - Rome route in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Rome-Fiumicino and Rome-Ciampino airports: Alitalia with more than 18 based aircraft, Air One with around 10 based aircraft. However, on the basis of the Phase II investigation none of these carriers appears to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in case of a price increase by the merged entity. In view of its recent exit from the Dublin - Rome (as well as from the Dublin - Milan route - see paragraph 1006 and following) and of its current financial situation (1263), it is unlikely that Alitalia would constitute a credible potential entrant to the Dublin - Rome route in the short to mid-term which could compensate for the loss of competition between Ryanair and Aer Lingus. Further, Alitalia is a full-service network carrier which would anyway not be a close competitor to the Merging Parties, having a different cost structure and hence a more limited ability to offer low fares. As regards Air One, its activities are in particular focused on domestic Italian routes and routes from Italy to Mediterranean holiday destinations. Even though it operates also flights to some European cities (such as to London-City, Paris or Copenhagen, its focus is clearly Italy and Southern Europe. Further, Air One is a full service carrier offering business class on its international routes and code-sharing with Lufthansa and other network carriers. Therefore, Air One cannot be regarded as a potential entrant on the Dublin - Rome route which could provide sufficient competitive constraints to the merged entity. Ryanair also argues that Centralwings serves Ciampino from its base at Krakow airport and that there is nothing to prevent it operating a W-route between Krakow-Rome-Dublin-Rome-Krakow if it became profitable to do so (1264). However, as discussed in Section 7.8.2., a new entry on this basis is less likely and less efficient. Further, in view of the length of the suggested W-route, it is also doubtful whether it could be operated without a necessity to change the crew in view of the crew utilisation limitations (1265). Further, the majority of the passengers on this route originate in Ireland (1266) and any new entrant would therefore need to secure sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland (such as Alitalia, and in particular the outside Italy less known Air One), as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. It should also be noted that with 5 Ryanair aircraft based in Rome-Ciampino, the merged entity would have an advantage of having bases at both sides of this route, with resulting greater flexibility and cost benefits. Finally, apart from the Ryanair's entry in March 2003, there are no examples of (even only attempted) new entry to this route in the recent years.

1042. As explained in detail in Sections 7.3. and 7.8., the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Rome-Fiumicino/Rome-Ciampino. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Rome route.

1043. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Rome route.

Dublin - Salzburg

1044. The total number of passengers carried by scheduled airlines on the Dublin - Salzburg route in 2006 was [0 - 50 000]*. As Aer Lingus started operations on this route in December 2005 and Ryanair in April 2004 and there were no scheduled carriers active before, it is a rather new route. Currently a total of around 6 weekly rotations are operated on this route by the Merging Parties during the winter season and around 3 during summer. The route is relatively thin as regards the total number of passengers and in view of the recent Aer Lingus' and Ryanair's entry, it is questionable whether there would be much scope for any significant new entry. Currently, Ryanair operates the route the whole year while Aer Lingus offers flights only in the winter season. Ryanair argues that during the winter season, the charter airlines have significant presence on the route (1267). For this reason and also in view of the seasonality of the services by Aer Lingus, the possible competitive constraints on the merged entity on this route exercised by charter traffic are analysed in more detail in paragraph 1048.

1045. On the Dublin - Salzburg route, Ryanair and Aer Lingus overlap on an airport-to-airport basis and only during the winter season. In winter 2006/2007 season, Ryanair operated around 3 weekly rotations to Salzburg airport (SZG) and Aer Lingus operated around 3 weekly rotations to the same airport. While the planned weekly rotations for the Summer 2007 season remain the same for Ryanair, Aer Lingus have no planned scheduled flights in the summer 2007 season. Apart from the Merging Parties, there are no other scheduled carriers active on this route and there were no such carriers present at the time of Aer Lingus' entry in December 2005. Apart from that, there have been no other new entrants on this route in the recent years (1268).

1046. The market shares of the Merging Parties on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in summer 2006 season are summarised in the following table (1269).

<emplacement tableau>

1047. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Salzburg route from two to one and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (1270).

1048. In general, charter airlines are not considered to constitute a significant competitive constraint on Ryanair and Aer Lingus on the overlap routes (1271). However, as charter traffic carried a significant number of passengers on the Dublin - Salzburg route in winter 2005/2006 and taking into account the arguments to the contrary raised by Ryanair with respect to this particular route, the Commission has explored whether charter airlines could exercise a competitive constraint on the Merging Parties on the Dublin - Salzburg route. It should be however noted that Ryanair entered the route only in April 2006 and thus the share of the charter traffic is likely to be lower than it was in the Winter 2005/2006 season. In case the added capacity of Ryanair is taken into account (and even presuming hypothetically that entry of Ryanair would not take any passengers away from the charter airlines), the share of total charter traffic would be around 35% in the winter season. In general, the majority of charter operations is limited to the period between December and March, even though in the past years there was less significant charter traffic also in summer, mainly between June and September. As also Aer Lingus' activities are limited to the winter season, the charter airlines are present most of the time when also the activities of Ryanair and Aer Lingus overlap. In particular the following charter airlines operate on this route: Monarch, Austrian Airlines, Thomas Cook Airline, Transavia and Flightline (1272). However, according to the slot allocation for winter 2006/2007 provided by Ryanair, all these charter carriers operate their flights solely on Saturdays, whereas both Ryanair and Aer Lingus offer flights on Tuesdays, Thursdays and Saturdays and thus provide significantly more flexibility. Further, the market investigation indicated that a large majority of seats of charter airlines were not sold separately but rather as a part of the package holiday, sold to the final customers by the tour operators and not by the charter airlines themselves (1273). Even though Ryanair refers for example to FirstChoice's submission and claims that it offers "dry seats" on the Irish market (1274), it should be added that FirstChoice sells only 3% of its tickets in Ireland as dry seats and that 97% of its tickets are sold as part of package holidays (1275).There are no indications that this would be different for this particular route. Therefore, only a very limited part of the around 35% share of total charter traffic on this route in the winter season were dry seats. None of these charter airlines offered a possibility to book flight only for Dublin - Salzburg via their website, even though it is possible to do so at the webpages of some tour operators selling their tickets in Ireland (1276). Most of the charter airlines sell their tickets to tour operators who then sell them either as part of the package holidays (which also means that it is in general the tour operator and not the charter airline who is deciding on which routes to operate and when (1277)) or to a much smaller extent as dry seats. Furthermore, in view of the decreasing charter market in Ireland in general and a stagnating number of passengers on this particular route (1278), the likelihood of significant entry by new charter carriers with more flexible product offering is low (1279). Therefore, as only a small part of the around 35% share of total charter traffic in the winter season would be constituted by dry seats and taking into account the still limited volume of charter operations, the difference in the product sold to the final customers and in the used distribution channel (tour operators) and the limited flexibility of the charter flights, the charter airlines do not represent any significant competitive constraint for the merged entity on the Dublin - Salzburg route.

1049. The Dublin - Salzburg route is subject to the general barriers to entry described in Section 7.8. It should, however, be acknowledged that airport congestion at Dublin does not constitute the most significant barrier to entry to this route as the predominantly leisure flights on this route are not dependent on peak time slots and high frequencies. On the other hand, it should be noted that carriers focused on leisure routes also prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier. (1280)

1050. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. Ryanair argues that Austrian Airlines maintain a base at Salzburg with a number of Aircraft (1281). However, Austrian Airlines operate from Salzburg only four scheduled routes to Austrian or German destinations (1282) and do not operate any scheduled routes to Ireland. In view of these facts and the focus of Austrian Airlines to full-service network operations, it cannot be regarded as a likely entrant on this route which would have the ability and incentive to provide significant competitive constraints on the merged entity. Further, a significant part of the passengers on this route originate in Ireland (1283) and any new entrant would therefore need to secure a sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland, as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. Finally, apart from the Merging Parties, there are no examples of new entry to this route in the recent years.

1051. Given that the route is seasonal, it is questionable whether any other scheduled airline would find it profitable to enter the route from Dublin against the merged entity. As explained in detail in Sections 7.3. and 7.8., the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Salzburg. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Salzburg route.

1052. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Salzburg route.

Dublin - Seville

1053. The total number of passengers carried by scheduled airlines on the Dublin - Seville route in 2006 amounted to [0 - 50 000]*. It should be noted however that Ryanair entered the route only in January 2007 and the passengers carried in 2006 are thus solely Aer Lingus passengers. Currently a total of around 5 weekly rotations are operated on this route by the Merging Parties in the winter and around 6 in the summer season. As Aer Lingus entered the route in March 2005, it is a rather new and still growing route. However, the route is relatively "thin" as regards the total number of passengers and in view of the recent Ryanair entry, it could be questioned whether there would be scope for any further significant new entry. Aer Lingus operated the route originally only in the summer season in 2005 and 2006 but as of October 2006 it operates the route all year round. The route has no significant seasonal pattern. The presence of charter airlines on this route is marginal (1284), which is also acknowledged by Ryanair itself (1285), and therefore does not need to be discussed separately as it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1286), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

1054. On the Dublin - Seville route, Ryanair and Aer Lingus overlap on an airport-to-airport basis. In the winter 2006/2007 season, Ryanair operated around 3 weekly rotations to Seville airport (SVQ) and Aer Lingus operated around 2 weekly rotations to the same airport. The planned weekly rotations for the summer 2007 season remain the same for Ryanair and will amount to around 3 weekly rotations for Aer Lingus. Apart from the Merging Parties, there are no other competitors active on this route. Before the entry of Aer Lingus in March 2005, there were no scheduled carriers operating on this route. Except for Ryanair's entry in January 2007, there have been no other new entrants on this route in the recent years.

1055. The market shares of the Merging Parties on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (1287). <emplacement tableau>

1056. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Seville route from two to one and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (1288).

1057. The Dublin - Seville route is subject to the general barriers to entry described in Section 7.8. However, it should be acknowledged that airport congestion at Dublin does not constitute a significant barrier to entry to this route as the predominantly leisure flights on this route are not dependent on peak time slots and high frequencies (1289). On the other hand, it should be noted that carriers focused on leisure routes also prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier. (1290)

1058. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport are able to easily enter the Dublin - Seville route in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. However, as also acknowledged by Ryanair, there are no competing carriers based in Seville. It should be noted that Clickair indicated that they plan to establish a base in Seville with 2-3 aircraft by the end of 2007 (1291). However, even if this intention is materialised, there are further doubts as to whether Clickair would be able to provide sufficient competitive constraints to the merged entity as described in detail in paragraph 902 in the description of the Dublin - Barcelona route. In particular, Clickair does not regard Dublin as an attractive market for expansion in view of the strength of the Merging Parties in their home market in Dublin and as that would require substantial investment into marketing (1292). Further, the majority of the passengers on this route originate in Ireland (1293) and any new entrant would therefore need to secure a sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland, as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. Finally, apart from the Merging Parties, there are no examples of new entry to this route in the recent years.

1059. As explained in detail in Sections 7.3. and 7.8., the competing carriers based at Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this predominantly leisure route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Seville. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Seville route.

1060. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Seville route.

Dublin - Tenerife

1061. The total number of passengers carried by scheduled airlines on the Dublin - Tenerife route in 2006 amounted to [50 000 - 100 000]* and currently a total of around 7 weekly rotations are operated on this route by the Merging Parties. The total number of passengers carried on this route was growing rapidly during the last years also in view of Spanair's entry in June 2006 and recent Ryanair's entry in December 2006. Ryanair argues that the route has strong seasonal characteristics. However, even though there are some seasonal peaks, it cannot be regarded as a typically seasonal route as there is important traffic throughout the year, which is also documented by the fact that charter carriers are present practically throughout the year. It is noted that charter traffic represented in 2006 around [50-60%]* of the total (charter and scheduled) passengers carried on this route (1294). For this reason, the possible competitive constraints on the merged entity on this route exercised by charter traffic are analysed in more detail in paragraph 1065.

1062. On the Dublin - Tenerife route, Ryanair and Aer Lingus overlap on a city-to-city basis. Both in the winter 2006/2007 season and in the Summer 2007 season, Ryanair operates around 4 weekly rotations to Tenerife North airport (TFN) and Aer Lingus operates 3 weekly rotations to Tenerife South airport (TFS). These airports are considered as substitutable from the demand side (1295). Apart from the Merging Parties, there are currently no other competitors active on this route. As of June 2006, Spanair was operating a service between Dublin and Tenerife South airport, however it exited the route as of November 2006 after only 5 months of operation. Apart from that, there have been no other new entrants on this route in the recent years. The Commission's market investigation also did not bring any indication that indirect flights would be used by customers in any significant extent and thus exert competitive constraints on the Merging Parties on this route (1296), which can be regarded as mid-haul (1297). 1063. The market shares of the Merging Parties and Spanair on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in summer 2006 season are summarised in the following table (1298).

<emplacement tableau>

1064. As Spanair has left the route, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Tenerife route from two to one and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (1299).

1065. In general, charter airlines are not considered to constitute a significant competitive constraint on Ryanair and Aer Lingus on the overlap routes (1300). However, as charter traffic represents an important share of passengers carried on the Dublin - Tenerife route and taking into account the arguments to the contrary raised by Ryanair with respect to this particular route, the Commission has explored whether charter airlines could exercise a competitive constraint on the Merging Parties on the Dublin - Tenerife route. It should be noted that Ryanair entered the route only in December 2006 and thus the share of the charter traffic is likely to be lower than it was in 2006. In case the added capacity of Ryanair is taken into account (and even presuming hypothetically that entry of Ryanair would not take any passengers from charter airlines), the share of total charter traffic would be less than 40%. In particular, the following charter airlines operate on this route: Air Futura, FirstChoice Airways, Monarch Airlines and Iberworld (1301). It should be noted that despite a peak in the summer season, the charter airlines are present on this route with important capacities throughout the year. However, the market investigation indicated that a large majority of seats of charter airlines were not sold separately but rather as a part of the package holiday, and sold to the final customers by the tour operators and not by the charter airlines themselves (1302). Even though Ryanair refers for example to FirstChoice's submission and claims that it offers "dry seats" on the Irish market (1303), it should be added that FirstChoice sells only 3% of its tickets in Ireland as dry seats and that 97% of its tickets are sold as part of package holidays (1304). There are no indications that this would be different for this particular route. Therefore, only a very small part of the les than 40% share of the charter traffic on this route were dry seats. None of these charter airlines offered a possibility to book flight only for Dublin - Tenerife via their website, even though it is possible to do so at the webpages of some tour operators selling their tickets in Ireland (1305). Most of the charter airlines sell their tickets to tour operators who then sell them either as part of the package holidays (which also means that it is in general the tour operator and not the charter airline who is deciding on which routes to operate and when (1306)) or to a much smaller extent as dry seats. It should also be noted that the operations of the charter companies on this route in summer 2006 were concentrated on Fridays, Saturdays and Sundays only (1307). On the other hand, the combined entity would offer a much more flexible timetable covering the whole week. Furthermore, in view of the decreasing charter market in Ireland in general as well as on this particular route (1308), the likelihood of a significant entry by new charter carriers with more flexible product offering is low (1309). Therefore, as only a small part of the less than 40% share of total charter traffic was constituted by dry seats and taking into account the difference in the product sold to the final customers and in the used distribution channel (tour operators) and the limited flexibility of the charter flights, the charter airlines can not be considered to represent any significant competitive constraint for the merged entity on the Dublin - Tenerife route.

1066. The Dublin - Tenerife route is subject to the general barriers to entry described in Section 7.8. However, it should be acknowledged that airport congestion does not constitute a significant barrier to entry to this route as the predominantly holiday flights on this route are not necessarily dependent on peak time slots and high frequencies. On the other hand, it should be noted that carriers focused on leisure routes also prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier. (1310) On the other hand, an absolute majority of the passengers on this route originate in Ireland (1311) and any new entrant would therefore need to secure sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland, as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland.

1067. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. According to Ryanair, there are no competing scheduled airlines based in Tenerife at present which would be able to enter the route. The market investigation indicated that Spanair bases aircraft in Tenerife (1312). Spanair is a full-frills scheduled as well as charter airline belonging to Star Alliance. Even though they plan a major expansion in the next years, their aim is "to become the leading airline for business travel" and their expansion plan is mainly focused on Barcelona (1313). Further, Spanair entered the Dublin - Tenerife route in June 2006 and exited (together also with Dublin - Barcelona and Dublin - Madrid) after only 5 months of operations. Therefore, it seems unlikely that Spanair would constitute a credible potential entrant against the merged entity to this route. As explained in detail in Sections 7.3. and 7.8., the competing carriers based at Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Tenerife. There are no other examples of a sustained new entry to this route which would provide efficient competitive constraints to Ryanair and Aer Lingus. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Tenerife route.

1068. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Tenerife route.

Dublin - Toulouse/Carcassonne

1069. The total number of passengers carried by scheduled airlines on the Dublin - Toulouse/Carcassonne route in 2006 was [100 000 - 150 000]* and currently a total of around 4 weekly rotations in the winter season and around 10 weekly rotations in the summer season are operated on this route by the Merging Parties. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 increased by [40-50%]*, probably also due to the fact that Ryanair started its services only in April 2005. Even though the route is relatively new and was rapidly growing in the last years, it could be questioned whether there would be scope for further significant new entry. Currently, Ryanair operates the route the whole year whilst Aer Lingus offers flights only in the summer season. Ryanair argues that during the winter season, the charter airlines have significant presence on the route. However, as the overlap between the Merging Parties on this route is limited to the summer season and the presence of charter airlines in summer season is marginal (1314), the possible competitive constraints by the charter airlines specific for this route do not have to be analysed separately (1315) as it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1316), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular yearround operations of the merged entity on this route.

1070. On the Dublin - Toulouse/Carcassonne route, Ryanair and Aer Lingus overlap on a cityto- city basis and only during the summer season. In winter 2006/2007 season, Ryanair operated around 4 weekly rotations to Carcassonne airport (CCF) while Aer Lingus is not active. The planned weekly rotations for the Summer 2007 season for Ryanair are around 6 weekly rotations while Aer Lingus plans to operate around 4 weekly rotations to Toulouse Blagnac airport (TLS). These airports are considered as substitutable from the demand Side (1317). Apart from the Merging Parties, there are no other scheduled carriers active on this route and there were no such carriers present at the time of Aer Lingus' entry in October 2003. Apart from the Merging Parties, there have been no other new entrants on this route in the recent years.

1071. The market shares of the Merging Parties on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (1318).

<emplacement tableau>

1072. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Toulouse/Carcassonne route from two to one and would give the merged entity a 100% combined market share. The proposed merger is thus a merger to monopoly which would eliminate all existing competition on this route (1319).

1073. The Dublin - Toulouse/Carcassonne route is subject to the general barriers to entry described in Section 7.8. However, it should be acknowledged that airport congestion at Dublin does appear to constitute a significant barrier to entry to this route as the predominantly leisure flights on this route are not dependent on peak time slots and high frequencies. On the other hand, it should be noted that carriers focused on leisure routes also prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier. (1320)

1074. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to easily enter the Dublin - Toulouse/Carcassonne route in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. According to Ryanair, Air France has a base in Toulouse Blagnac airport with 6 based aircraft. However, Air France has its main hub at Paris CDG, its main domestic hub at Paris Orly and regional hubs at Lyon, Clermont-Fernand and Bordeaux (1321). To the extent they have any aircraft based in Toulouse, these are primarily used for domestic flights. Therefore, also in view of the different business model of this full-frills network airline, Air France cannot be regarded as a credible potential entrant to the Dublin - Toulouse/Carcassonne route which would provide significant competitive constraints to the merged entity, able to offer a low fare alternative to Ryanair's/Aer Lingus' customers. Further, a significant part of the passengers on this route originate in Ireland (1322) and any new entrant would therefore need to secure sufficient number of Irish originating passengers. That would constitute a further obstacle to entry for carriers not yet present in Ireland, as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. Finally, apart from the Merging Parties, there are no examples of new entry to this route in the recent years.

1075. There are also no indications that CityJet with its aircraft based in Dublin would be a credible entrant to this route. As this route is primarily leisure oriented with a low share of business/more time-sensitive passengers (1323), it would not be in line with the business model of CityJet described in more detail in Section 7.8. Ryanair argues that CityJet "entered the route in 2006" (1324) and also commenced flights from London City to "a number of French destinations" (1325). However, CityJet has no scheduled flights to Toulouse/Carcassonne either from Dublin or from London City. According to the DAA data received by the Commission, CityJet operated in April 2006 one single rotation between Dublin and Toulouse. However, except this one-off operation they did not start any scheduled flights on the route. CityJet also indicated that operating services similar to those launched recently from the London City airport would not be possible from Dublin due to the capacity situation and insufficient access to facilities at Dublin Airport (1326). Further, with its smaller planes CityJet would have difficulties to compete with the merged entity on this low-frequency route. For the similar reasons, and as explained in detail in Sections 7.3. and 7.8., also the other competing carrier based in Dublin Airport (Aer Arann) is not a likely potential entrant to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Toulouse/Carcassonne. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Toulouse/Carcassonne route.

1076. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Toulouse/Carcassonne route.

Dublin - Venice

1077. The total number of passengers carried by scheduled airlines on the Dublin - Venice route in 2006 amounted to [100 000 - 150 000]* and currently a total of 9 weekly rotations are operated on this route by the Merging Parties. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 increased by almost [60- 70%]*, following Ryanair's entry in April 2006. Further, as Aer Lingus entered the route only in March 2004, it can be considered as rather new and dynamic route, even though it could be questionable whether there would be scope for further entry. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (1327), as Ryanair itself acknowledged (1328), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1329), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

1078. On the Dublin - Venice route, Ryanair and Aer Lingus overlap on a city-to-city basis. In the winter 2006/2007 season, Ryanair operated 4 weekly rotations to Treviso airport (TSF) and Aer Lingus operated 5 weekly rotations to Venice airport (VCE). These airports are considered as substitutable from the demand side (1330). The planned weekly rotations for the summer 2007 season remain the same for both Ryanair and Aer Lingus. Apart from the Merging Parties, there are no other competitors active on this route. Before the entry of Aer Lingus in 2004, there were no scheduled carriers operating on this route. Apart from the Merging Parties, there have been no other new entrants on this route in the recent years.

1079. The market shares of the Merging Parties on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (1331).

<emplacement tableau>

1080. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Venice route from two to one and would give the merged entity a 100% combined market share. Therefore, the merger would eliminate all existing competition on this route (1332).

1081. The Dublin - Venice route is subject to the general barriers to entry described in Section 7.8. However, it should be acknowledged that airport congestion at Dublin does not constitute the most important barrier to entry to this route as the predominantly leisure flights on this route are not necessarily dependent on peak time slots and high frequencies. On the other hand, it should be noted that carriers focused on leisure routes also prefer early morning slots and that they consider the existing slot congestion at Dublin as a relevant entry barrier. (1333)

1082. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Venice airports: Alpi Eagles with 4 based aircraft and Alitalia. However, on the basis of the Phase II investigation none of these carriers appears to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in case of a price increase by the merged entity. In view of its recent exit from the Dublin - Rome as well as from the Dublin - Milan routes and of its current financial situation (1334), it is unlikely that Alitalia would constitute a credible potential entrant to the Dublin - Venice route in the short to mid-term. Notwithstanding that, Alitalia is a full-frills network carrier which would anyway not be a close competitor to the Merging Parties. Further, Alitalia's activities at the Venice airport focus rather on domestic regional routes than international flights. Alpi Eagles is an Italian regional airline operating mainly domestic and some international routes (to Mediterranean and Central and Eastern Europe) with smaller aircraft (1335). In view of its activities (1336) and aircraft used, it cannot be regarded as a credible entrant to the Dublin - Venice route able to exert sufficient competitive constraints to the merged entity. Further, the majority of the passengers on this route originate in Ireland (1337) and any new entrant would therefore need to secure sufficient number of Irish originating passengers. That would constitute an additional obstacle to entry for carriers not yet present in Ireland (such as Alitalia, and in particular the outside Italy less known Alpi Eagles), as it would require significant investment into brand awareness and marketing considering the current strength of the Ryanair and Aer Lingus brands in Ireland. Finally, apart from the Merging Parties, there are no examples of new entry to this route in the recent years.

1083. As explained in detail in Sections 7.3. and 7.8, the competing carriers based at Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Venice. Therefore, the results of the market investigation demonstrate that the possibility of entry is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Venice route.

1084. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Venice route.

Dublin - Vienna/Bratislava

1085. The total number of passengers carried by scheduled airlines on the Dublin - Vienna/Bratislava route in 2006 amounted to [150 000 - 200 000]* and currently in total around 14 weekly rotations are operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 increased by around [70-80%]* (probably also due to entry by Ryanair in April 2006) and therefore the route can be regarded as rather dynamic. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (1338), as Ryanair itself acknowledged (1339), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1340), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

1086. On the Dublin - Vienna/Bratislava route, Ryanair and Aer Lingus overlap on a city-to-city basis. In the winter 2006/2007 season, Ryanair operated around 3 weekly rotations to the Bratislava airport (BTS) and Aer Lingus operated around 7 weekly rotations to the Vienna Schwechat International airport (VIE). These airports are considered as substitutable from the demand side (1341). Apart from the Merging Parties, SkyEurope operates a service between Dublin and Bratislava airport with around 4 weekly rotations. The planned rotations for the summer 2007 season remain the same for all three carriers. Ryanair argues that also Austrian Airlines operate at least during summer season on this route from Vienna Schwechat airport. However, even though Austrian Airlines traditionally operated this route and had some flights also in the Summer 2006 season, they were limited to 1 or 2 rotations per week (on Wednesdays and Sundays) and had more a character of charter flights (in view of its seasonality and limited number of rotations). Further, neither the website of the Dublin Airport (1342) nor the website of Austrian Airlines (1343) indicates that there would be any scheduled flights planned for the Summer 2007 season. In any way, due to their very limited and seasonal presence on this route, the inclusion of Austrian Airlines does not change the competitive situation on the market, as can be seen from the share of passengers carried in Summer 2006 indicated in the table below. Apart from these carriers, Lauda Air operated on this route (from Vienna Schwechat International airport) in the summer seasons in 2002 and 2003 but exited from the route afterwards. Further, Tyrolean Airways operated on the route (from Vienna Schwechat International airport) between March 1998 and October 2002 but are not present there since.

1087. The market shares of the Merging Parties and their competitor on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table (1344). <emplacement tableau>

1088. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Vienna/Bratislava route from three to two and would give the merged entity a very high combined market share. The very high combined market share of Ryanair and Aer Lingus on this route is in itself evidence of a dominant market position (1345). Moreover, the transaction would combine Ryanair and Aer Lingus, the two closest competitors active on the route (1346). It is noted that SkyEurope is also a low-frills airline offering point-to-point services and operates on this route to Bratislava airport together with Ryanair. However, SkyEurope has indicated that the merger would give the combined entity a market dominating position which would prevent SkyEurope from expanding even on routes which would be consistent with their business plan, including Dublin - Vienna/Bratislava (1347). Further, Ryanair itself in its internal documents states that SkyEurope [...]* (1348). It is thus doubtful, to what extend SkyEurope represents a credible competitive constraint to the combined entity in the short to mid-term, in particular in view of the fact that the combined entity would operate on the Dublin - Vienna/Bratislava route each day of the week while SkyEurope provides only three rotations per week. Further, a larger part of the passengers on this route originate in Ireland (1349) and in order to expand, SkyEurope would therefore need to secure sufficient number of Irish originating passengers. As regards Austrian Airlines, due to their limited presence on this route and focus on fullservice network operations, it cannot be regarded as close competitor to the Merging Parties exerting any significant competitive constraints. Further, the fact that at least Ryanair is the closest competitor to Aer Lingus on this route is confirmed by the Customer Survey conducted by the Commission. According to the survey 28.4% of Ryanair's customers considered Aer Lingus as an alternative and 37.0% considered any other airline, in particular SkyEurope (this is probably largely due to the fact that they fly to the same airport). And vice versa, 29.3% of Aer Lingus' passengers considered Ryanair as an alternative and only 16.2% considered any other airline (1350). Therefore, the merger would lead to elimination of the competitive rivalry between the Merging Parties which has been an important source of competition on this route and it is doubtful whether this loss could be compensated by the only remaining competitor SkyEurope.

1089. The Dublin - Vienna/Bratislava route is subject to the general barriers to entry described in Section 7.8. above, even though in view of the limited number of frequencies and less important share of more time-sensitive passengers, the peak time congestion at Dublin Airport and the fact that Vienna Schwechat International airport is fully coordinated (1351) may not be a significant barrier to entry. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity.

1090. The following carriers are based in Vienna Bratislava: Austrian Airlines with more than 20 aircraft at VIE, Lauda Air with 2 aircraft at VIE, SkyEurope with 7 aircraft at BTS, Air Slovakia with 3 aircraft at BTS and Slovak Airlines with 3 aircraft at BTS. On the basis of the Phase II investigation none of these carriers appears to have sufficient abilities and/or incentives to enter this route and provide sufficient competitive constraints in case of a price increase by the merged entity. As regards possible expansion of SkyEurope and Austrian Airlines, see in paragraph 1088 and with respect to SkyEurope, see also Section 7.8. Lauda Air operated on the route in the past, however it has been acquired by Austrian Airlines since and now operates full-service and charter flights to holiday destinations (1352). Slovak Airlines are a full-service scheduled as well as charter carrier based in Bratislava with in total 3-4 aircraft of which two are used for charter flights and the remaining one is used for the current only two scheduled flights operated by Slovak Airlines (Bratislava - Brussels and Bratislava - Moscow). It does not operate to Ireland and has no plans to enter this market (1353). Air Slovakia is a full-service scheduled and charter airline operating with its in total 3 aircraft scheduled flights to the Middle East and charter flights to holiday destinations in the Mediterranean (1354). Therefore, neither Slovak Airlines nor Air Slovakia can be regarded as potential entrants to the Dublin - Vienna/Bratislava route able to exert any significant competitive constraint on the merged entity. As explained in detail in Sections 7.3. and 7.8., the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Vienna/Bratislava. Therefore, even though the route is growing rather dynamically, the results of the market investigation demonstrate that the possibility of entry or expansion of capacity by the existing competitors is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anticompetitive effects of the merger on the Dublin - Vienna/Bratislava route.

1091. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Vienna/Bratislava route.

Dublin - Warsaw

1092. The total number of passengers carried by scheduled airlines on the Dublin - Warsaw route in 2006 amounted to [150 000 - 200 000]* and currently in total around 28 weekly rotations are operated on this route by scheduled carriers. The total number of passengers carried on this route in 2006 compared to the situation in the year 2005 increased by around [30-40%]* and the route can be regarded as rather dynamically growing. The route has no significant seasonal pattern. The presence of charter carriers on this route is negligible (1355), as Ryanair itself acknowledged (1356), and thus it can be excluded that charter operations would exert any appreciable competitive constraint on the merged entity (see also section 6.7.). Furthermore, contrary to claims by Ryanair (1357), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route. Furthermore, contrary to claims by Ryanair (1358), ad hoc operations of charter airlines (for example, in the case of large sports or cultural events or holidays) cannot be considered as constraining to any significant extent the regular year-round operations of the merged entity on this route.

1093. On the Dublin - Warsaw route, Ryanair and Aer Lingus overlap on an airport-to-airport basis. In the winter 2006/2007 season, Ryanair operated (as of December 2006) around 7 weekly rotations to the Warsaw airport (WAW) and Aer Lingus operated around 7 weekly rotations to the same airport. Apart from the Merging Parties, LOT operates around 10-11 weekly rotations and, as of September 2006, Norwegian Air Shuttle operates around 3 weekly rotations on this route. The planned rotations for the summer 2007 season remain the same for all four carriers.

1094. The market shares of the Merging Parties and their competitors on the basis of the operated seat capacity in the winter 2006/2007 and summer 2007 season as well as on the basis of the passengers carried in the summer 2006 season are summarised in the following table. It must however be noted that LOT carries a significant proportion of connecting passengers on this route. On the other hand, both Ryanair and Aer Lingus carry practically only pointto- point passengers. Therefore, in line with the definition of the relevant market and to better reflect the competitive situation on this market, it is appropriate to calculate the market share excluding connecting passengers as indicated in the sixth column of the table. Further, in order to show the combined position of the Merging Parties in the point-topoint market after entry of Ryanair in December 2006, the last column of the table shows the market shares on the basis of capacity for the Summer 2007 season where the capacity corresponding to connecting passengers of LOT is excluded on the presumption that the share of connecting passengers will stay as in Summer 2006.

<emplacement tableau>

1095. Therefore, the proposed transaction would lead to a reduction in the number of carriers active on the Dublin - Warsaw route from four to three and would give the merged entity a very high combined market share, leaving customers with only two significantly smaller competitors. The very high combined market share of Ryanair and Aer Lingus on this route is in itself evidence of a dominant market position, in particular as only point-topoint passengers should be taken into account (1359). Moreover, the transaction would combine Ryanair and Aer Lingus, the two closest competitors active on the route (1360). LOT is the Polish national flag carrier. It is a member of Star Alliance. Its business model based on full-service operations with significant proportion of connecting passengers is significantly different from both Ryanair and Aer Lingus. Norwegian Air Shuttle is a lowcost point-to-point airline with a main base in Oslo which entered the Polish market (including the Dublin - Warsaw route) only recently (1361). Apart from this route, they do not operate any other route to/from Ireland. Therefore, even though their business model is similar to Ryanair and Aer Lingus, it is doubtful whether they would be able to provide significant competitive constraint in view of their new entry into Poland and only limited capacities operated currently on this route (the merged entity would operate almost five times more frequencies than Norwegian Air Shuttle). Therefore, the merger would lead to elimination of the competitive rivalry between the Merging Parties which has been an important source of competition on this route and it is doubtful whether this loss could be compensated by either LOT or Norwegian Air Shuttle.

1096. The Dublin - Warsaw route is subject to the general barriers to entry described in Section 7.8., even though in view of the limited number of frequencies and the less important share of more time-sensitive passengers, the peak time congestion at Dublin Airport may not be the most important barrier to entry. Ryanair argues that in particular competing carriers with bases or hubs at the destination airport(s) are able to enter this route easily, in case the merged entity increases prices and would thus provide sufficient competitive constraints on the behaviour of the merged entity. The following competing carriers have bases or hubs at Warsaw airport: LOT with more than 20 aircraft, Norwegian Air Shuttle with 2 aircraft, Centralwings with 3 aircraft, SkyEurope with 1 aircraft and Wizzair with 4 aircraft. As discussed in paragraph 10951095, expansion of the current competitors on the route (LOT and Norwegian Air Shuttle) is unlikely to provide significant competitive constraint on the merged entity. It is noted that SkyEurope is also a low-frills airline offering point-to-point services968. However, SkyEurope has indicated that the merger would give the combined entity a market dominating position which would prevent SkyEurope from expanding even on routes which would be consistent with their business plan (1362). In addition, Ryanair itself in its internal documents states that SkyEurope [...]* (1363). Centralwings is a low-frills subsidiary of LOT based in Poland and operating a number of flights between Poland and the rest of Europe, including several flights to Ireland (1364). Even though Centralwings currently compete on three of these routes (Dublin-Wroclaw, Dublin-Gdansk and Cork- Warsaw) with Ryanair, it was always Ryanair who entered the route against existing operations of Centralwings. It is therefore questionable whether Centralwings would be willing to enter this route against the strong merged entity and be able to exercise significant competitive constraints. Wizzair is a low-frills carrier based mainly in Hungary and Poland and flying to a number of European destinations. However, apart from the recently started two flights to Cork from Katowice and Gdansk (where neither of the Merging Parties is active), it does not operate any flights to Dublin (1365). It is therefore doubtful to what extent it would be able to exert efficient competitive constraints on the merged entity. Further, a larger part of the passengers on this route originate in Ireland (1366) and in order to expand, any new entrant would therefore need to secure sufficient number of Irish originating passengers. It should be acknowledged that part of this Irish originating traffic could be generated by migrant Polish workers living in Ireland (1367) which could enhance the ability of the low-frills airlines with significant presence in Poland (in particular Centralwings, SkyEurope or Wizzair) to compete with the merged entity on this route. However, it should be noted that so far none of these airlines have ever entered a route between Poland and Ireland already served by Ryanair and even none of them entered (at least between April 2003 and October 2006) any other intra-European route where Ryanair was already operating on the relevant airport-pair or city-pair (1368). It is therefore doubtful whether these carriers would be able and willing to enter and compete effectively against the merged entity on the Dublin - Warsaw route.

1097. As explained in detail in Sections 7.3. and 7.8., the competing carriers based in Dublin Airport (Aer Arann and CityJet) are not likely potential entrants to this route. Furthermore, the market investigation did not identify any likely new entrant for this route not based in either Dublin or Warsaw. Therefore, even though the route is growing rather dynamically, the results of the market investigation demonstrate that the possibility of entry or expansion of capacity by the existing competitors is not likely, timely and sufficient to constitute a competitive constraint on the merged entity and defeat the anti-competitive effects of the merger on the Dublin - Warsaw route.

1098. Taking into account all the above, the Commission concludes that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the Dublin - Warsaw route.

7.10. Possible efficiencies are not likely to outweigh the competitive harm

1099. In the Notification (1369) and various follow-up submissions in reply to comments and specific questions made by the Commission (1370), Ryanair emphasised the potential for substantial efficiencies brought about by the proposed transaction, which would allegedly benefit all customers.

7.10.1. The principles

1100. According to recital 29 in the preamble to the Merger Regulation (1371) and the Horizontal Merger Guidelines (1372), it is possible that efficiencies brought about by a merger counteract the effects on competition and in particular the potential harm to consumers that it might otherwise have. Parties to a concentration may thus detail the efficiency gains generated by the concentration that are likely to enhance the ability and the incentive of the merged entity to act pro-competitively for the benefit of consumers. Typical examples of such efficiencies include cost savings, new product introduction and service or product improvement. Efficiency claims need to be verifiable (namely reasoned, quantified and supported by internal studies and documents if necessary). To counteract the anticompetitive effects of a merger, such efficiencies must be likely to benefit consumers (and in particular, to benefit directly customers in the relevant markets where competition concerns have been identified) and could not have been achieved to a similar extent by means that are less anticompetitive than the proposed concentration (the so called "merger specificity"). The three conditions - verifiability, merger specificity and consumer benefit - are cumulative.

1101. According to the Horizontal Merger Guidelines, to declare compatible with the common market a transaction for which competition concerns have been identified, the Commission should be in a position to conclude, on the basis of sufficient evidence, that: "the efficiencies generated by the merger are likely to enhance the ability and the incentive of the merger entity to act pro-competitively for the benefit of consumers, thereby counteracting the adverse effects on competition which the merger might otherwise have".

1102. Moreover, the Horizontal Merger Guidelines state that "the incentive on the part of the merged entity to pass efficiency gains on to consumers is often related to the existence of competitive pressure from the remaining firms in the market and from potential entry". It is further indicated that: "It is highly unlikely that a merger leading to a market position approaching that of a monopoly, or leading to a similar level of market power, can be declared compatible with the common market on the ground that efficiency gains would be sufficient to counteract its potential anti-competitive effects" (1373).

1103. In the light of these principles and of the anti-competitive effects identified above and in view of the very high market shares of Ryanair/ Aer Lingus on the relevant markets, any efficiencies, even if they were found to be verifiable, merger specific, and likely to benefit consumers, would have to be particularly substantial to prevent the significant impediment to effective competition set out above.

7.10.2. Ryanair's claims

1104. In the Notification, Ryanair argued that due to low barriers to entry, expansion and a high degree of differentiation between Aer Lingus and Ryanair, the merger should not raise any competition concerns. However, even if that were not the case, Ryanair submits that the merger would not give rise to unilateral effects due to efficiency gains and its particular business model.

1105. Ryanair indicated that the efficiencies brought about by the proposed transaction would derive mainly from operational cost savings, as a result of (i) larger scale and (ii) rationalisation within Aer Lingus once Ryanair's business model (and related expertise in generating lower costs and greater efficiencies) would be applied to it (including via the introduction of better and more innovative management). These savings would concern several fields: staff costs, aircraft ownership costs, maintenance costs, airport charges and ground operational costs, ancillary sales and finally, distribution efficiencies.

1106. Ryanair provided some detailed calculations as well as complementary information upon the Commission's request. (1374) Initial post-merger efficiencies are estimated by Ryanair as summarised in the following table.

<emplacement tableau>

1107. On staff costs, Ryanair argues that it could "feasibly reduce Aer Lingus' total staffing by approximately 1000-1500, or roughly [...]* of Aer Lingus' current staff numbers" (1376), including both clerical and cabin staff. Ryanair also argued that this reduction in cabin staff would not be detrimental to the service level because it would be coupled with better work practices. For instance, the contribution of cabin crew to completing the boarding of the aircraft and the cleaning function, the outsourcing for catering, etc.

1108. On aircraft ownership costs, Ryanair claims that the savings will mainly derive from the better bargaining position of the combined group coupled with Ryanair's experience in negotiating discounts on large aircraft orders. It also points out that it could make available at no cost some of the 67 Boeing aircraft for which Ryanair has an option until 2012 (that is to say, price has already been negotiated but there is no "firm" order, yet) to replace the 14 Airbus aircraft that Aer Lingus currently leases (and the contracts for which will expire between 2009 and 2012), generating a considerable cost reduction, according to Ryanair.

1109. On maintenance costs (representing together with staff costs around [60-70%]* of the claimed savings), Ryanair estimates that its scale (Ryanair's fleet is of around 120 Boeing aircraft) would allow it to negotiate more favourable maintenance contracts for the much smaller fleet currently available to Aer Lingus (around 35 Airbus aircraft). These costs would be even lower if Aer Lingus were to switch to a Boeing fleet, namely by reducing the need for duplicate spare parts or dual rated engineers.

1110. On airport charges and ground operational costs, Ryanair argues that the main potential for savings derives from scale in all airports where both airlines currently operate. In fact, many of Ryanair's discount deals with airports are conditional on delivering a certain minimum number of passengers per year. By adding Aer Lingus' passengers to Ryanair's ones better discounts could be achieved (or at least thresholds would be more likely attained).

1111. On distribution (and advertising) costs, all savings claimed by Ryanair would stem from shifting Aer Lingus' sales model to a near 100% online sales model based on Ryanair's online platform. In addition, the "internet exposure" created by ryanair.com would reduce the need for Aer Lingus to use alternative advertising and marketing means.

1112. Ryanair submits that these efficiencies that would be brought about by the proposed transaction cannot be obtained by any alternative transaction and would not be achieved by the companies individually in the absence of the merger, for instance through on-going (or planned) reorganisation of Aer Lingus and/or internal growth. In support of this Statement, Ryanair argues that it substantially outperforms Aer Lingus on a wide range of efficiency measures, such as operational punctuality, fewer lost bags and a lower flight cancellation level. Therefore, each of the efficiencies achieved as a result of the merger would be timely, likely and concentration-specific. Also in terms of merger-specificity, Ryanair claims that Aer Lingus shows no signs of implementing changes on its own initiative and even when such changes happened in the past (namely a reduction of its staff by up to 50%) there was little or no change in Aer Lingus's service levels.

1113. Ryanair argues that these efficiency gains will be passed on to consumers in terms of reduced fares, higher frequencies and more routes for passengers, and better quality products and services. Cost savings realised within Aer Lingus would enable it to be more competitive and to enhance its product offering, while Ryanair would benefit from Aer Lingus's brand and image on routes where it is not currently present.

1114. Ryanair argues that the claimed cost savings would in no way affect Aer Lingus' quality of service. The service levels would either be maintained or, more likely, improved. According to Ryanair, Aer Lingus' mid frills service would be maintained, which involves the current brand, flights to major capital city and regional city airports, allocated seating and a business class service on long haul flights.

1115. Although most of the claimed cost savings concern (short-run) fixed costs, Ryanair claims that "the established track record Ryanair enjoys of converting low costs to low fares throughout its network"1377 would ensure that efficiencies are passed on to consumers.

7.10.3. Aer Lingus' position

1116. In its submissions and replies to the Commission's questionnaires (1378), Aer Lingus argues that Ryanair's claims "appear to be based on incorrect facts and/or adopt a counterfactual that ignores the improvements in operational efficiency achieved by Aer Lingus over the past few years as well as its plans for future cost-cutting measures, (1379)"

1117. On the counterfactual, Aer Lingus argues the following. ???Some of the savings put forward considered by Ryanair derive from not comparing like-with-like, namely not considering that part of Aer Lingus' business (and therefore fleet and staff) is allocated to long-haul flights, while Ryanair is only active on short-haul flights. A comparison limited to Aer Lingus short-haul business would result in much lower or no savings (for instance in staff costs or airport charges). ???Ryanair seems to assume that it could simply translate its business model (and related costs) to Aer Lingus and that Aer Lingus would not be able to realise any savings in the absence of the merger. This is strongly opposed by Aer Lingus, by putting forward past and planned savings as well as the fact that some costs are simply non-compressible (such as airport charges in primary airports) or not dependent on scale (such as fuel costs, which depend more on an appropriate hedging). ???Cost reductions will not automatically translate into lower fares and are more likely to result in lower quality of the service offered by Aer Lingus in respect of aspects that are relevant for its current customers (such as seat allocation). ???Claimed reduction of aircraft costs would arise only under the assumption of realising the book value from the sale of the current Aer Lingus' fleet and its replacement with an asset (namely Ryanair's options on new Boeing aircraft), which is currently valued in Ryanair's accounts "around 32% below the current Aer Lingus (and market) aircraft cost despite the need to meet substantial transaction costs. The implication is that Ryanair's Boeing option price is well below the current market value of a Boeing 737-800, and the options are extremely valuable. Seen in this light, the Ryanair proposal has no efficiency effect at all, and in fact it has no real effect on the merged entity. It simply involves Ryanair transferring existing options for 28 Boeing aircraft with a value of approximately €42 million into the accounts of its Aer Lingus subsidiary. The analysis above is correct on the assumption that the scenario only involves a reshuffling of existing assets". (1380) ???Most of the efficiencies identified by Ryanair are the result of its buyer power. According to Aer Lingus a number of these savings do not represent efficiencies as they would not raise total welfare and primarily represent a transfer of rents from one party (for example, shareholders of Airbus or Boeing in the case of aircraft costs, airports in the case of airports charges) to another (the new entity). Aer Lingus states that this critique applies to all of the savings (apart from staff costs) claimed by Ryanair.

1118. As regards the incorrect nature of the facts on which Ryanair's claims are based, Aer Lingus argues the following. ???Some savings are not netted out of the reduced revenues that they would generate (such as those deriving from ending the cargo service) or do not consider the costs attached (such as the costs of outsourcing some tasks). ???Aer Lingus also opposes the view that cabin crew can be reduced and that turnaround time is compressible to 25 minutes.

(i) As regards cabin crew, Aer Lingus argues that Ryanair's figures are wrong and that Aer Lingus is already running at the legal minimum of 4 cabin crew on A320 and 5 on A321.

(ii) As regards turnaround, Aer Lingus argues that 40 minutes is the standard turnaround (not 55, apart from when a meal break is provided to the crew) and that 25 minutes are possible only in secondary airports where Ryanair operates but not in primary, congested airports.

1119. Based on the above, Aer Lingus opposes the estimation provided for many of the savings claimed by Ryanair, with an impact on verifiability, merger specificity and pass-on to consumers.

7.10.3.1. Verifiability

1120. Focusing in particular on the largest savings claimed by Ryanair, that is to say, staff costs, aircraft costs, aircraft utilisation and maintenance costs, representing all together around [80-90]*% of the efficiency gains claimed by Ryanair (EUR [150-200m]*of the total EUR [200-250m]*, see table 4 below), some of the erroneous assumptions and inaccuracies invoked by Aer Lingus support its claim that these savings would not be attainable or are, in any case excessive. Aer Lingus also points to various cost savings it has already achieved in the areas of airport costs, fuel management, maintenance and distribution through re-negotiating agreements with the relevant suppliers. Aer Lingus provides its own estimates of the savings that it considers more reasonable and attainable. (1381)

1121. Overall, Aer Lingus' estimate of possible cost savings is significantly below Ryanair's projections, as shown by table 4 below:

<emplacement tableau>

Sources: Ryanair's reply of 25 January 2007 (folio no. 1795) to Commission's questionnaire of 11 January 2007 and Aer Lingus' submission of 14 February (folio no. 5704) as well as replies of 2 and 6 March 2007 (folio no. 5704 and 5024) to Commission's questionnaire of 22 February 2007 as well as submission of 9 March 2007 (folio no. 5704).

7.10.3.2. Merger specificity

1122. Aer Lingus also contests the merger-specificity of the claimed efficiencies. It argues that it can achieve those savings on its own, based on past and on-going internal policies. Aer Lingus claims to have focussed heavily on cost reduction and productivity since 2001. It claims also that it has reduced unit costs across the organisation by 47% since that year. This focus on costs is set to continue independently of the proposed concentration. At the time of the Aer Lingus IPO, Aer Lingus told investors that it would "focus relentlessly on cost." (1382)

1123. Based on this, Aer Lingus argues that even the savings claimed by Ryanair that are verifiable (that is to say, which are acknowledged by Aer Lingus in substance and size) would not be merger-specific.

7.10.3.3. Pass-on to consumers

1124. Aer Lingus also argues, in particular via a paper prepared by its economic consultants, that the "virtuous circle" argument (that is to say, Ryanair would have the incentive to pass lower costs on to consumers in terms of lower fares because those lower fares will create higher volumes and lead to even lower costs, and so on) is flawed because it assumes that Ryanair would be immune from competition. On the contrary, the fact that it would be immune from competition on some routes and not on others, would result in the normal profit-maximising scenario of lowering fares on the routes where there is competition and simply increasing margins where Ryanair is immune from competition.

1125. Finally, Aer Lingus claims that Ryanair fails to "shed any light on the extent to which this particular merger would reduce marginal (incremental) costs [... in relation] to output expanding efficiencies". (1383)

1126. Aer Lingus concludes that even if Ryanair could demonstrate some verifiable and merger-specific efficiency gains, and even if these might be passed on to consumers on some routes, the pass-on would be limited (or inexistent) on the Dublin routes where the transaction would eliminate effective competition.

7.10.4. Commission's assessment

1127. The Commission has reviewed the material submitted by Ryanair and Aer Lingus. It welcomes the efforts by the Merging Parties to assess whether the claimed efficiencies meet the relevant criteria of verifiability, merger specificity and benefit consumers as well as to provide indications of their magnitude. For the reasons set out below, the Commission has reached the conclusion that none of the claimed efficiencies meets all the necessary criteria to be taken into consideration. Moreover, even if these criteria were met, the claimed efficiency gains would in all likelihood be insufficient in magnitude to reverse the anti-competitive effects identified above in Sections 7.1-7.9.

7.10.4.1. General principles

1128. A basic principle underlying the efficiency section of the Horizontal Merger Guidelines is the assumption that companies generally pursue profit maximisation strategies. The potential for efficiencies to offset any anticompetitive merger effects is assessed under the same assumption.

1129. Ryanair's submission on merger efficiencies (1384) and its response to the Statement of Objections rely in part on the notion that "Ryanair's business model is radically different from the various models used by any other airline in Europe" (1385). Ryanair argued that it is not in competition with other airlines but with other uses of disposable income. Further, Ryanair claims that "[g]iven that Ryanair's model is growth dependant and that growth relies entirely on fare reductions to generate it, there is no realistic prospect whatsoever of Ryanair increasing its fares." (1386) The company reacts indignantly to the suggestion, that it might not pass on efficiencies and that "it will simply use lower costs as an excuse to raise margins" (1387).

1130. However, as shown in detail above (see e.g. Section 7.4), Ryanair is clearly in competition with other airlines and adjusts its fares both upwards and downwards in response to competitive conditions and costs (1388). Other elements also show that Ryanair is able to increase prices in order to maximise profits. For example, in a report on its latest financial results, Ryanair indicates that inter alia competitors' price increases (through fuel surcharges) led it to raise its own average fare (yield) by 7%. (1389) In 2005, Ryanair increased its average fare by 2%. Conversely, in Ryanair's 2004 annual report, increased capacity and "a profusion of lower fares offered by flag carriers and start up airlines" is blamed for a 14% decline in yields. Also, Ryanair's industry-leading 22% operating margin (in FY 2006) would appear to be consistent with normal profit-maximising conduct.7

1131. Hence, despite Ryanair's arguments, there are no indications that any increase in Ryanair's market power would automatically benefit consumers.

7.10.4.2. Verifiability

1132. Ryanair's efficiency claims are based on documents produced for the purposes of the merger procedure. The basic line of argument is that Ryanair will apply its management skills and important elements of its business model to Aer Lingus, which will thereby lower its costs towards Ryanair's existing levels. Some claimed efficiencies also relate to the increased scale of the combined business.

1133. The Commission considers that several of Ryanair's efficiency claims rely on very strong assumptions which cannot be independently verified. In particular, the proposition that Ryanair will be able to fully transfer its business model, and in particular the related cost levels to Aer Lingus without offsetting downgrades in product characteristics and revenue appears highly optimistic. Ryanair presents no objective or convincing evidence in this respect other than a general confidence in "Ryanair's more ruthless management style" (1390) compared to Aer Lingus. There appear not to exist business documents, dated pre-merger, which objectively and independently assess the scope for efficiency gains from acquiring Aer Lingus. The Commission considers that such documents are critical to show, first, that Ryanair's business model is different, non-replicable and superior to that of Aer Lingus and, secondly, that its cost structure can be successfully replicated in Aer Lingus' post merger.

1134. Ryanair points to its integration of a previous acquisition, the KLM subsidiary "Buzz", as an example of its successful track record in implementing efficiencies. (1391) However, that airline was entirely absorbed into Ryanair. That is, contrary to the premises of the present efficiency claim, the Buzz brand disappeared, the entire Buzz fleet was withdrawn (returned to lessors) within 18 months of the acquisition and services to most primary airports were discontinued. (1392) Rather than supporting Ryanair's efficiency claim, the Buzz acquisition, thus, sheds further doubt on the assumptions underlying Ryanair's cost projections.

1135. Therefore, the Commission has not found any verifiable evidence that the claimed efficiencies can be realised when measured against the three most relevant dimensions of interest for consumers, namely fares, frequencies and service quality (for each route). It has not been demonstrated that Ryanair can reduce Aer Lingus costs without offsetting reductions in service quality (different airports, different use of the crew, etc).

1136. The same concern applies to the claims made about possible reductions in advertising costs. The Commission notes that if Ryanair plans to maintain the Aer Lingus brand and quality differentiation, it will continue to be necessary for Aer Lingus to follow its own, differentiated marketing strategy. Indeed, it may be necessary to maintain advertising expenditures in order to ensure that consumers do not identify Aer Lingus with Ryanair.

1137. With respect to aircraft ownership costs, it seems likely that Ryanair acquired the options for new aircraft for the purposes of developing its own business. Transferring the options to Aer Lingus would mean that Ryanair forgoes the opportunity of expansion planned in the absence of the merger. Even if Ryanair had no alternative use for the new aircraft in its own (as opposed to Aer Lingus') fleet, there would still be opportunity costs, because it could alternatively sell the new aircraft to another airline for an immediate profit. This opportunity cost exists as long as Ryanair's negotiated delivery price remains below the "spot market" price. In other words, the opportunity cost of the aircraft's alternative use is exactly equal to the claimed efficiency (and thereby makes it irrelevant). Transferring options from Ryanair to Aer Lingus at below market prices would amount to no more than a rent transfer.

1138. In addition, assuming that the efficiency did exist (that is to say, that the two foregoing caveats did not apply), the assumptions under which of aircraft ownership costs are quantified seem particularly optimistic, notably in terms of the possibility to realise the book value of Aer Lingus' fleet.

1139. As regards airport charges and ground operations costs, the Commission notes that these are often regulated and thus fixed, particularly at the primary airports used by Aer Lingus. This may also be one explanation for Ryanair's relatively higher success in negotiating lower airport charges. However, cost reductions from merely switching Aer Lingus services to secondary airports would in themselves not constitute efficiencies.

1140. Furthermore, the hostile nature of the takeover may complicate integration of the two companies (especially if Ryanair obtained less than 100% of Aer Lingus' shares) and thus reduce the likelihood that claimed efficiencies can be achieved. Significant uncertainties remain about Ryanair's ability to reduce Aer Lingus costs, in particular, staff costs, without reducing the quality of service which currently allows Aer Lingus to somewhat differentiate its product in relation to Ryanair.

1141. In conclusion, the Commission finds that it cannot verify the efficiencies claimed by Ryanair so as to be reasonably certain that these efficiencies are likely to materialise (1393). Ryanair's claims are often based on mere assertions such as that Aer Lingus' unit costs can be reduced to its own levels, without explaining to what extent this would be at the expense of other elements beneficial to consumers, such as quality of service (1394) or airport location. In addition, it appears from the absence of any supporting documents pre-dating the bid that Ryanair itself did not systematically assess the scope for efficiencies when preparing its bid for Aer Lingus. The supporting documents submitted by Ryanair were prepared specifically for the purposes of the Commission's merger procedure.

1142. Ryanair's efficiency claim, therefore, fails to meet the criteria of verifiability. Nevertheless, for the sake of completeness, the following sections provide an assessment of the merger specificity of Ryanair's efficiency claims and of the potential consumer benefits.

7.10.4.3. Merger specificity

1143. In addition, a number of efficiencies claimed by Ryanair are likely to be not merger specific. In particular, the claimed reduction in staff costs, improved aircraft utilisation, and reductions in fuel and distribution costs could also be achieved by Aer Lingus independently of the proposed merger. Indeed, since 2001 Aer Lingus has built a significant track record of reducing unit costs, as outlined in Section 7.3.2. The fact that the airline has placed more emphasis on vertical product differentiation than Ryanair cannot by itself be taken as an indication of inefficiency. Its ability to compete successfully against Ryanair on most of its short-haul network, to the contrary, would seem to indicate that Aer Lingus has chosen an efficient price/quality combination. Likewise, Ryanair's professed low esteem of Air Lingus' new management team (1395) does not constitute evidence that the airlines will not continue to implement possible efficiencies. In the absence of the merger, Aer Lingus will continue to be exposed to competition from Ryanair, which in itself provides incentives to remain an efficient operator.

1144. With respect to the efficiencies related to reductions in maintenance charges and aircraft costs, it seems, as argued by Aer Lingus, that these (and other) savings, to the extent they can be achieved, would largely represent a re-distribution of rents in intermediate markets, rather than real efficiencies. Such a transfer of rents from one party (for instance the airports) to another (the new entity) would not lead to any change in welfare. In addition, as explained above (Section 7.10.4.2), the transfer of Ryanair aircraft delivery options to Aer Lingus at below market prices would merely constitute a rent transfer between the Merging Parties. Similar arguments apply to fuel hedging contracts and other rents originating in financial market transactions.

1145. In any case, even if cost reductions through rent transfers from third Merging Parties (such as aircraft manufacturers) were considered as merger-specific efficiencies, they affect essentially fixed (1396), rather than marginal, costs. While the Horizontal Merger Guidelines do not explicitly limit the notion of efficiencies to reductions in marginal costs, a pass-on of cost reductions within a reasonable timeframe is less likely for fixed cost reductions because they do not immediately affect a firm's profit maximising price/output decisions.

7.10.4.4. Benefits to consumers

1146. Ryanair argues that reductions in aircraft operating costs affect the airline's "marginal" decision whether or not to operate a flight on a certain route. (1397) "Such cost reductions have an effect similar to the reduction in the 'hurdle rate' of return that must be earned in order to make any given flight profitable, which in turn will increase the number of potential flights that can be profitably operated." (1398)

1147. The Commission accepts that the argument that lower fixed entry costs into a route lower the yield levels at which entry becomes profitable, and that such entry tends to increase competition for a given route, is valid in principle. It is essentially an argument about long-run competitive equilibrium. This raises two important questions:

(i) the degree of certainty and timeliness of any pass-on, and

(ii) whether pass-on will benefit the consumers affected by the merger.

1148. With regard to certainty and timeliness, the claimed fixed cost efficiencies would not affect Ryanair's price setting decisions on existing flights. Any consumer benefit would materialise only when and if Ryanair opted to increase frequencies on existing routes (which would ceteribus paribus lower the average fare it can achieve on those routes) or if and when it opened a new route that was not viable before the fixed-cost reduction but became so after the merger. Even if the claimed efficiencies were realised, any consumer benefit would therefore not be immediate but conditional on a chain of events and thus considerably less certain than the price effect of a marginal cost reduction (which would create immediate incentives for price reductions).

1149. The benefits of fixed cost efficiencies would in theory be most felt on "thin" routes that are at present not viable (1399). However, since Ryanair only argues that Aer Lingus' costs can be brought down towards its own levels, there would be no change compared to the pre-merger situation. Ryanair's "hurdle yield" and, thus, the scope of potential profitable routes would not change, even if all claimed efficiencies were realised.

1150. By contrast, on existing overlap routes, the immediate effect of the merger lies in the internalisation of Ryanair's and Aer Lingus' pricing and output decisions. The combined airline (assuming profit maximising conduct) would have an incentive to raise prices on these routes because, due to the very high combined market shares, any diversion of passengers between the Ryanair and Aer Lingus brands would largely (and entirely on the routes moving from duopoly to monopoly) remain within the merged group. Theoretically, this effect could be counteracted if and when fixed-cost efficiencies made it profitable for Ryanair to add extra frequencies to the affected routes, which would in turn put downward pressure on prices. Given the extremely high combined market shares, often approaching monopoly levels, and the fact that the claimed efficiencies extend largely to the Aer Lingus brand, it is highly unlikely that the price-reducing effect of such efficiencies would be sufficient to reverse the price increase from the horizontal overlap and consequent loss of competition.

7.10.5. Conclusion on efficiencies

1151. In conclusion, the efficiencies claimed by Ryanair are not verifiable and are not merger specific. Even if they were, these efficiencies would affect Aer Lingus' fixed (aircraft operating) costs, which makes it uncertain that they would benefit consumers. Finally, as set out in the Horizontal Merger Guidelines "[i]t is highly unlikely that a merger leading to a market position approaching that of a monopoly, or leading to a similar level of market power, can be declared compatible with the common market on the ground that efficiency gains would be sufficient to counteract its potential anti-competitive effects" (1400).

1152. In the light of these principles and of the anti-competitive effects identified in Sections 7.1 to 7.9, the Commission concludes that the merger does not give rise to efficiencies that would counteract the significant impediment to effective competition resulting from the notified transaction.

8. COMMITMENTS PROPOSED BY RYANAIR

1153. Ryanair first submitted commitments in Phase I of the investigation on 29 November 2006 ("Initial Phase I Commitments"), which were subsequently replaced by a revised set of commitments on 14 December 2006 (1401) ("Modified Phase I Commitments"). On 17 April 2007, Ryanair submitted, together with its reply to the Statement of Objections, commitments in order to remove the competition concerns identified in the Statement of Objections ("Initial Phase II Commitments"). After a State-of-Play meeting with the Commission which was held on 26 April 2007 and in which Ryanair was informed about the Commission's preliminary assessment of the initial Phase II Commitments, Ryanair submitted a modified set of commitments (hereinafter "Final Commitments") on 3 May

2007. On 29 May 2007, a State-of-Play meeting was held in the form of a conference call in which Ryanair was informed about the results of the market test of those commitments and the Commission's preliminary assessment of the Final Commitments.

8.1. Description of the proposed commitments

1154. The Final Commitments proposed by Ryanair contain, essentially, a modification of Ryanair's commitments offered in Phase I.

8.1.1. The Commitments offered in Phase I

1155. The Initial Phase I Commitments submitted by Ryanair consisted of two main elements, a slot-related remedy and a behavioural remedy related to the fares of Aer Lingus. Ryanair notably proposed a [2-4]*-year commitment to accommodate entry by allowing a new entrant that is unable to obtain the necessary slots on [5-10]* specified airport-pair routes to obtain them from Ryanair or Aer Lingus, provided that no more slots are surrendered than those required to offer a frequency equal to the frequencies of Ryanair or Aer Lingus (1402). Ryanair further proposed not to increase capacity for three years (beyond the level prior to entry) after the facilitated new entry occurs and to reduce Aer Lingus' average short haul fares by a minimum of 2.5 per cent per annum for four years.

1156. In its Modified Phase I Commitments, Ryanair offered also additional slots at Dublin Airport, including morning peak time slots necessary, according to Ryanair, to base [2-6]* aircraft at Dublin aircraft. Further, it increased the number of routes for which slots would be made available. The remedy related to Aer Lingus fares was eliminated in this version.

1157. The Commission in its decision to initiate proceedings pursuant to Article 6(1)(c) of the Merger Regulation took the view that the proposed commitments were not sufficient to clearly remove all identified competition concerns. First, the commitments did not cover all the routes on which the Commission had identified competition concerns in its Article 6(1)(c) decision. The slots offered were also regarded as insufficient to allow a competitor to establish itself with a sufficient size at Dublin, notably by establishing a base. The Commission also noted that slots made available at non-congested airports were not likely to attract new entry. The Commission's market test of the remedies also identified a number of other concerns related to the effectiveness and practicability of the proposed remedy. More importantly, during the market test, no airline expressed a concrete interest in establishing itself with a significant presence or even a base at Dublin Airport and thus in ensuring that the competitive rivalry of carriers at Dublin Airport would be maintained post-merger. Although the Modified Phase I Commitments created a theoretical possibility for an airline to obtain slots which would be sufficient to base up to [2-6]* aircraft at Dublin, there was thus no indication that such entry would actually occur, nor that any entrant would provide a sufficient competitive constraint to the merged entity (together having more than 40 aircraft in Dublin) (1403).

8.1.2. The Initial Phase II Commitments

1158. Ryanair submitted revised commitments in Phase II. Like the commitments offered in Phase I, the Initial Phase II Commitments submitted on 17 April 2007 consisted of a "behavioural" commitment related to fares and fuel surcharges (referred to in the commitments text as "Commitment No. 1") and a part related to the release of slots (referred to in the commitments text as "Commitment No. 2"), which was modified as compared with the previous proposal as described in the following two paragraphs.

1159. In the Initial Phase II Commitments, Ryanair committed to reduce Aer Lingus' short-haul fares by at least 10% immediately, to eliminate the fuel surcharges Aer Lingus applies on its long-haul flights immediately, to retain Aer Lingus' brand and to continue to operate Ryanair and Aer Lingus separately.

1160. Ryanair also offered to make available "slots" on the existing overlap routes. These slots included (1) a certain number of slots for flights between Dublin and London/Heathrow and (2) slots for other routes from/to Dublin, sufficient, according to Ryanair, to operate with up to [4-8]* based in Dublin.

1161. Having analysed the Initial Phase II Commitments, the Directorate-General for Competition ("DG Competition") informed Ryanair in a State-of-Play meeting that the Commitments were not sufficient to eliminate the identified competition concerns and that they would not be sent to third parties in order to market test them (1404).

8.1.3. The Final Commitments

1162. On 3 May 2007, Ryanair sent a "set of commitments (1405)" by e-mail and fax which consisted of several documents, namely ?

??a "Commitments Letter" which contains (i) a description of the commitments proposed by Ryanair and (ii) general comments on the Commission's investigation and on the appropriateness of commitments in the present case;

??an Annex (with the title "Annex 1") which is aligned to the commitments format of the Commission's model texts and describes "the details of the proposed mechanics of Commitment no 2 (1406)";

1163. Among the document sent to the Commission was also a letter signed by the Chief Operating Officer and Deputy Chief Executive of Ryanair which again contains a description of the commitments as well as general comments on the Commission's investigation and on the appropriateness of commitments, including an annex (with the title "Appendix 1") which sets outs [...]*. Ryanair, however, declared later in an e-mail of 15 May 2007 that this letter did not form part of the commitments (1407).

1164. The Final Commitments pick up certain proposals of the Initial Phase I Commitments (Heathrow slots, slots for other Dublin routes, fare-/brand-related commitments) and add several new elements (slots for Shannon/Cork routes, partial up-front buyer proposal, and frequency freeze). Since the exact content of the Final Commitments was not entirely clear, the Commission has tried to obtain clarifications from Ryanair. (1408) Their main content can be summarised as follows (see in detail Section 8.2.1 below):

Slot related commitments

(1) Heathrow slots: Ryanair committed to make available, for an unlimited period (1409), slots for the London-Heathrow route under a so-called "leasing arrangement" (1410). According to the text of the Final Commitments, these slots are exclusively reserved for British Airways and Air France (allowing them to each offer [2-6]* daily flights in each direction) (1411).

(2) Slots for other routes from/to Dublin: Ryanair also committed, if necessary (1412), to make available slots for other overlap routes from and to Dublin, allowing airlines, according to Ryanair, to operate with up to [4-8]* aircraft based in Dublin (1413). Ryanair further offers to make available an equivalent number of slots at specific destination airports on the overlap routes, if necessary (1414).

(3) Slots for routes from/to Shannon/Cork: Ryanair also committed in its Final Commitments to make available slots for overlap routes starting at Cork and Shannon if necessary (1415) ([6-10]* daily slots in Cork and [6-10]* daily slots in Shannon and an equivalent number of slots at London/Stansted for flights to London/Stansted plus [0-4]* arrival and [0-4]* departure [slot/slots] in Cork and Liverpool in order to facilitate entry on the Cork-Manchester/Liverpool route (1416)).

(4) "Up-front buyer": In the "Commitments Letter", Ryanair also offers "not to complete the acquisition of Aer Lingus (1417)" before it has found a "buyer" that has committed to taking up the slots for the [4-8]* based aircraft operation at Dublin (1418).

Non slot-related commitments

(5) Fare/brand-related commitments: Ryanair repeats its offer to reduce Aer Lingus' short-haul fares by at least 10% immediately, to eliminate the fuel surcharges Aer Lingus applies on its long-haul flights immediately, to retain Aer Lingus' brand and to continue to operate Ryanair and Aer Lingus separately.

(6) "Frequency freeze": Ryanair commits not to increase the number of frequencies on any of the claimed overlap routes "in the event of a new entrant to the route", in excess of the frequencies jointly operated by Ryanair and Aer Lingus on each route for a period of six IATA seasons after completion of the merger (1419). It also commits not to reduce the frequencies on these routes "unless a route is or becomes unprofitable".

1165. The Commission sent a non-confidential version of the Final Commitments together with a detailed questionnaire to third parties on 8 May 2007 in order to allow them to express their views on the Commitments and to clarify certain factual issues (for example, sufficient scope of the remedy, likelihood of entry triggered by the Final Commitments). However, the Commission made clear to Ryanair (1420) and the consulted third parties (1421) that the fact that a market test was conducted did not prejudice the Commission's final assessment as to whether or not the Final Commitments remove the competition concerns raised by the concentration.

8.2. Assessment of the proposed commitments

1166. The Commission has assessed the Final Commitments and concludes that they cannot remove the identified competition concerns brought about by this merger. Not only are they formulated in an unclear and often contradictory manner (see Section 8.2.1), their substance is also not sufficient to resolve the identified competition concerns (see Section 8.2.2). The Commission's market test of the Final Commitments has confirmed this view.

8.2.1. Formal shortcomings: The Commitments are not sufficiently clear to be implemented and enforced

1167. Under the Merger Regulation, the Commission can only accept such remedies which are sufficiently clear so that it can be expected that they can be implemented and enforced by the Commission without further problems, notably without creating a manifest risk of diverging interpretations of main elements of the commitments (1422). Since any delay in the implementation of the Commitments which may be caused by problems of interpretation or even by litigation can have severe consequences for the affected customers, it is crucial for commitments to be accepted that they provide a sufficient degree of legal certainty as concerns their content and that they are "fully workable (1423)".

1168. The Final Commitments do not meet these minimum standards. They contain numerous contradictions and vague or ambiguous formulations which put into question the viability of the commitments as such, since it is doubtful whether the commitments would be at all workable and enforceable.

1169. The Commission has further identified a number of shortcomings which concern core parts of the commitments (1424).

8.2.1.1. The format of the commitments

1170. The original "set of commitments" consisted of a number of different letters and appendices. Ryanair later explained that only the "Commitments Letter" and one annex ("Annex 1") should be formally regarded as the submitted commitments (1425). The fact that Ryanair did not use the Commission's recommended model text (1426) as a basis for its commitments proposal but spread the legal content of the commitments in an incoherent manner between two documents and mixed it with statements and opinions throughout the text, makes it difficult and sometimes even impossible to identify the actual legal content of the commitments. This approach also led to a number of contradictions in the text, as set out in the next paragraphs (1427). Indeed, the only part of the text which contains "pure" legal text and which is aligned to the Commission's recommended commitments format, the submitted "Annex 1", covers only some parts of the offered commitments, while other important elements of the commitments (e.g. the "up-front buyer", the number of slots to be released at Cork and Shannon, Ryanair's willingness to offer slots at destination airports, but also the "frequency freeze" commitment and the entire "Commitment No 1") are only described in general in the attached letters.

8.2.1.2. The commitment relating to the "up-front buyer"

1171. In particular Ryanair's commitments relating to the offered "up-front solution" are not sufficiently clear. The commitments mention in various paragraphs Ryanair's willingness to find an up-front buyer before Ryanair acquires control of Aer Lingus (1428). However, Ryanair does not submit a text that sets out how such an "up-front solution" would work in detail. In particular Annex 1, which contains the relevant provisions on the divestiture of slots, remains silent in this respect and does not provide any description of the functioning of the up-front buyer solution.

1172. A commitment that aims at an up-front buyer solution must not only set out clearly the criteria according to which the Commission can accept or reject the respective purchaser ("Purchaser Requirements"); it must notably contain a clause that makes clear that the final sale and purchase agreement (or the final bid) is conditional on the Commission's approval of the purchaser. It is this clause that distinguishes an ordinary divestiture from an up-front buyer solution. Such a clause is entirely missing in the text of the Final Commitments which sets out the rules for the slot divestiture (Annex 1) (1429).

1173. Ryanair further refers in its "Commitments Letter" to a "time limit" for the implementation of the up-front buyer. On pages 2 and 5 it states: "Ryanair reserves all of its rights, including the right to request a review of this condition should it become impossible to secure an upfront buyer for these slots within the time limit set out above.". However, the Commission notes that such a "time limit" is not defined anywhere in the text. At the State-of-Play meeting, Ryanair indicated that there are deadlines set under the Irish Takeover Rules which would not allow for an undue delay of the bid. They claim that the bid would, if any conditions of the bid including those relating to the commitments are not fulfilled, lapse at the latest 81 days after Ryanair's offer document is posted, unless the Irish Takeover Panel grants an extension. However, the Final Commitments contained no detailed explanation of the relevant deadlines. Further, the definite deadline for finding an up-front buyer would depend on the date when the new bid is launched by Ryanair and thus the overall timeframe of the commitments is unclear.

1174. In this context it should also be noted that Clause 3 of Annex 1 ("Duration of Commitment No. 2") is in contradiction with the upfront buyer solution referred to in the "Commitments Letter". The former refers to a situation in which slots are not "taken up on or before the sixth anniversary of the date of the Consummation of the Concentration (...)". In an "upfront" scenario, however, the consummation of the concentration depends on whether all slots are actually taken up, hence, there cannot be a situation in which the slots are not taken up after the consummation.

1175. It is further not entirely clear from the text whether the upfront buyer solution refers only to "slots for the [4-8]* based aircraft operation at Dublin Airport" as set out on pages 2 and 5 of the "Commitments Letter", or whether Ryanair might have intended to include also the slots offered at Shannon and Cork Airports (see the respective indirect Statement on page 9 of this Letter: "(...) to require Ryanair to find an up-front buyer to base aircraft at Dublin, and take up slots at Cork and Shannon Airports (...)").

1176. Another ambiguity concerning the upfront buyer is the question whether slots can only be divested to different airlines or to one single purchaser (see Clauses 2.5-2.7. of Annex 1). Indeed, key provisions of the commitments use the term "up-front buyer" and "Prospective New Entrant" in the singular form, which seems to imply that the offered slots should go to one single purchaser (see for example, pages 2 and 5 of the "Commitments Letter" and Clause 3 of Annex 1). However, on page 4, the commitments refer to "airline (or airlines)" (1430).

8.2.1.3. The mechanics of the slots divestiture

1177. The Commission notes that the description of the offered slot divestiture in the "Commitments Letter" and in "Annex 1" is often not coherent and lacks a sufficiently detailed description of the intended mechanism of the slot divestiture and the number of slots made available. As the Commission's experience with previous slot remedies has shown, an unclear description of the detailed mechanism of the slot divestiture can cause significant delays and problems related to the slot divestiture. It is therefore important that the details of the slot divestiture be adapted to the requirements of the specific case, to the identified competition concerns, that the details be discussed with the Commission and can be sent for market testing in order to exclude such problems in the implementation phase.

1178. By way of example, the definition for "Heathrow Slots" in Annex 1 reads as follows: "Such slots will be allocated from time to time so as to comply with the Articles of Association of Aer Lingus (as amended from time to time)." The vague formulation in this definition does not meet the necessary requirements of an enforceable commitment text. Also the content of the term "binding leasing arrangement" referred to on page 4 of the "Commitments Letter" remains unclear. As regards the "early morning peak departure slot" for the Dublin-London route, the commitments do not define whether the offer refers to a slot at Dublin or Heathrow.

1179. The Commission further notes that Ryanair's willingness to make available slots at destination airports is not clearly expressed in the "Commitments Letter" and that the "Relevant Airport Pairs" as defined in "Annex 1" do not, as claimed in the submitted text, correspond to Table 2 of the Statement of Objections. For example, Annex 1 of the submitted text does not contain any indication with regard to the number of slots to be released at Cork and Shannon.

1180. The final Commitments are also ambiguous with regard to the question whether new entrant(s) could use the offered slots (except the London/Heathrow slots (1431)) for any route or whether the use is restricted to the overlap routes. While Ryanair supports the first interpretation (1432), which may find some support in the text of the Commitments Letter (1433), the relevant text of Annex 1 stipulates that slots are provided on the "Identified Relevant Airport Pairs" which are defined in Point 1 as being the overlap routes. Moreover, Point 2.15 of Annex 1 provides that slots that are not used on the Identified Relevant Airport Pair(s) have to be returned to the Merged Entity immediately. Also other details of the slot release mechanism are incomplete or not adapted to the present case (1434).

8.2.1.4. The description of Commitment No. 1

1181. As regards "Commitment No.1" the Commission notes that the current text only describes Ryanair's intentions in general, without defining in detail the scope of what is offered. The text is also void of all details concerning the implementation and monitoring of the proposed commitment (for example, reporting obligations on Ryanair, nomination of a Trustee to monitor compliance with these commitments, etc.).

8.2.1.5. Conclusion on the formal shortcomings

1182. The Commission's market test confirmed that the inconsistency and lack of clarity of the Final Commitments is also perceived by some competitors as an obstacle to the viability of the offered commitment (1435). The Commission therefore concludes that the Final Commitments fail to meet the requirements in terms of legal clarity and coherence in order to be acceptable to dismiss the identified competition concerns.

8.2.2. Substance: The commitments cannot eliminate the identified competition problems

1183. The Final Commitments do not consist in a business divestiture, but mainly in a so-called "access remedy (1436)". Following the model of previous airline cases, Ryanair's commitments propose to address the problem of airport congestion and intend to remove existing entry barriers for other airlines by making slots available to potential entrants. According to Ryanair, this would allow airlines to compete with Ryanair/Aer Lingus on the actual 35 overlap routes as well as other potential overlap routes and eliminate any impediment to competition (1437).

1184. As set out above (Section 8.1.3), Ryanair's distinguishes in its commitments between slots for the Dublin/Heathrow route and slots for other overlapping routes from/to Dublin, Cork and Shannon. The Commission's investigation showed that these "access remedies" will, independently of their formal shortcomings, not be sufficient to dismiss the identified competition concerns on the identified routes (see Sections 8.2.2.1/8.2.2.2). The additional behavioural commitments offered can also not compensate for the loss of competition (see Section 8.2.2.3).

8.2.2.1.The slots package for the overlapping routes from/to Dublin, Cork and Shannon (except London/Heathrow)

1185. The Commission's market investigation and the market test of the Final Commitments showed that there are no indications of sufficient entry on the vast majority of the identified routes and that the offer of slots for non-Heathrow routes (1438), will not restore competition on the more than 30 affected overlap routes. Slot remedies are not an appropriate remedy in this case, since they are not likely to trigger any substantial entry on the overlap routes

1186. As set out in Section 8.1.3, the offered Final Commitments do not consist in a business divestiture, but in "slot"-remedies which are intended to guarantee that actual or potential entry would sufficiently constrain the Merging Parties post-merger (1439). The Commission is of the view, as the market investigation suggests, that access to airport infrastructure (slots) would not be sufficient to restore competition in the present case.

1187. Indeed, the investigation showed that Dublin Airport is only partly congested, and that there are many other reasons which prevent competitors from entering into direct competition with Ryanair/Aer Lingus on the overlap routes in Ireland (see: Section 7.8, for example, Ryanair's and Aer Lingus' strong brands, their large bases at Dublin, Ryanair's extreme low cost model etc.).

1188. The Commission considers that access remedies may be acceptable in circumstances where it is sufficiently clear that there will be actual entry of new competitors that would eliminate any significant impediment to effective competition. However, in a situation in which it cannot be concluded that the lowering of the entry barriers by the proposed commitments will lead with the sufficient degree of certainty to the entry of sufficient new competitors in the market, the Commission has to reject such a remedies package (1440). This holds in particular true in situations in which the divested slot is not the only and not even the main entry barrier, and where accordingly the competitive constraint by the mere "threat" of potential entry by a competitor seems to be even more remote to effectively restore competition (1441).

1189. The Commission therefore had to verify whether it can be concluded with the required degree of certainty that the Final Commitments would actually trigger entry on the relevant overlap routes or not.

1190. A clear majority of the respondents to the market test of the Final Commitments does not regard slot remedies as an appropriate measure to trigger entry and to restore competition in the present case (1442). The market test has also confirmed the Commission's analysis of the main competitors' abilities to enter and the fact that the expectation of entry, if any, is limited to a very few routes (1443). It also showed that the commitments have not changed the general reluctance of almost all competitors to enter into competition with Ryanair/Aer Lingus on any of the overlap routes except the Dublin-London/Heathrow route.

1191. Indeed, none of the competitors answering to the market test has indicated that they would actually enter any of the non-Heathrow overlap routes, even in view of the Final Commitments (1444). Only three of all answering airlines indicated that they would consider entering on some routes (1445). Only two of the three competitors (1446) specified a route on which they would consider entry in the market test (Dublin-Hamburg; -Barcelona, - Malaga). The competitor considering entry on the Hamburg-Dublin route, however, indicated that most of the other overlap routes did not fit into its business model and that in any event further clarifications on the process of slot transfer would be necessary. It should also be noted that this competitor would only operate with small 70-100 seat aircraft on the considered route, which is not likely to provide effective competition on this route due to higher operation costs (1447). Clickair, while considering entry on two overlap routes, expressly excluded that it would enter on a larger scale by opening a base in Dublin ("to consider competing with Ryanair at its own base would be financially irresponsible for just about any company") (1448). The remaining potential competitor on an overlap route, Air France/CityJet, wants the Commission to consider its answer on potential entry plans on other specific routes than London/Heathrow confidential. The Commission, however, notes that CityJet also operates with small aircraft with less than 130 seats. It should also be noted that Air France/CityJet had already previously explained that there are important "barriers which would need to be removed before CityJet could actually successfully enter new routes to/from Dublin" and refers notably to insufficient access to airport facilities (i.e. non slot-related entry barriers) (1449).

1192. The Commission further notes that easyJet, which has a similar business model as the Merging Parties, indicated in relation to the Market Test that it would not enter any of the overlap routes. In this context, easyJet also stated that it believed it had not provided any indication that it was willing to enter/base aircraft at Dublin airport to Ryanair or the Commission. This contradicts the Statement made by Ryanair according to which easyJet had confirmed its willingness to enter/base aircraft at Dublin. easyJet also confirmed that it has no concrete plans at the present time to open a base at Dublin or to fly on routes to Dublin (1450). This is in line with easyJet's previous submissions in this case (1451).

1193. With respect to the possibility of establishing a base at Dublin Airport, the market test clearly indicated that no single competitor would actually open a new base at Dublin (1452) or consider opening such a base in view of the Final Commitments. As regards Ryanair's claim that in previous submissions several carriers (notably British Airways, easyJet and Air France/CityJet) had confirmed their "willingness to base aircraft at Dublin Airport", the Commission notes that neither British Airways nor easyJet have expressed such an intention to the Commission. On the contrary, British Airways (BA) indicated clearly in a previous questionnaire that "irrespective of the merger, BA would not set up a base in Dublin (1453)" and that "BA would not consider opening a base as it is not in line with our Strategic Direction (1454)". As set out in the previous paragraph, also easyJet has no such intention.

1194. The Commission therefore concludes, on the basis of the market investigation and the Market Test of the Final Commitments, that the lowering of the entry barriers intended with the proposed commitments would not lead with a sufficient degree of certainty to the entry of new competitors in the affected market.

1195. It should be noted that this result is in line with the Commission's assessment that the entry barriers which discourage competitors from entering into direct competition with Ryanair/Aer Lingus are not only related to airport congestion. Indeed, the Commission has identified numerous other important entry barriers in the present case which result, notably, from the fact that the proposed merger would combine two low frills/point-topoint carriers with a strong base at the same airport. The Final Commitments aimed, however, at removing the entry barriers related to airport capacity and did not provide a solution to address the other identified barriers to entry (such as access to sufficient infrastructure at Dublin Airport, difficulties to build up brand recognition, etc. (1455)).

1196. It is noted that the market test of the Final Commitments confirmed that slot-release remedies were not a sufficient means to eliminate the significant impediment to effective competition in this case (1456). Indeed, most respondents indicated that the ability to obtain slots for a given route would not be enough to stimulate entry and to ensure that one or more entrants would actually compete effectively against the merged entity. The "up-front" proposal cannot remedy the missing perspective of entry

1197. Commitments proposed by the parties will meet the condition that they are capable of rendering the notified transaction compatible with the common market only in so far as the Commission is able to conclude, with sufficient certainty, that it will be possible to implement them and that the new commercial structures resulting from them will be sufficiently workable and lasting (1457). Ryanair has offered not to complete the proposed acquisition of Aer Lingus (1458) before it has found an entrant that has committed to taking up the slots for the [4-8]* based aircraft operation at Dublin. The Commission has in previous cases accepted so-called "up-front buyer" solutions where the viability of the proposed remedies package to a large extent depends on the identity of a new entrant (or purchaser). In such a scenario, the Commission issues a clearance decision which makes the completion of the acquisition conditional on the approval of a suitable purchaser by the Commission (1459).

1198. A clearance decision which is combined with an "up-front" condition can be accepted in cases in which there is a sufficient degree of likelihood that (i) an entrant would eliminate all identified competition concerns and (ii) that the remedy with the up-front solution would be workable and implementable, in particular that the design of the remedy would allow for a suitable entrant to be found within a reasonable period of time (1460). Both conditions are, as set out in previous paragraphs, not fulfilled in this case. The proposed up-front solution is limited in scope and would lead to fragmented entry. Thus, it would not have sufficient scope to eliminate the impediment to competition on all identified overlap markets (1461). As set out in Section 8.1, the provisions on the "up-front" solution in the Final Commitments are in any event not sufficiently clear to be implemented.

The scope of the proposed commitments is insufficient

1199. The Commission believes, based on the results of the market investigation, that airport congestion is not the main barrier to entering into direct competition with Ryanair/Aer Lingus (1462). Regardless, however, of the appropriateness of the instrument of "slot remedies" in the present case (1463), the scope of the offered remedies also cannot be regarded as sufficient to stimulate entry of a scale that would allow the entrant(s) to replace the loss of competition caused by the proposed merger. Independently of the question whether slot remedies can be regarded as an appropriate measure at all in the context of the present transaction and assuming that despite the identified entry barriers, the Final Commitments would stimulate competitors to enter in competition with Ryanair/Aer Lingus and to take up all offered slots, this entry would still not be sufficient to effectively constrain the merged entity.

1200. As regards routes from/to Dublin, Ryanair offers to make available [4-8]* early morning peak hour departure slots per day at Dublin Airport for any airline(s) wishing to base up to [4-8]* aircraft (1464). In addition, Ryanair commits to make available to any of these carriers further slots at Dublin Airport "necessary to maintain a schedule (1465)" throughout the day for these (up to [4-8]*) aircraft. It is understood that the maximum number of slots offered at Dublin Airport amounts, according to Ryanair's own estimates, to around 336 weekly slots (1466), which would allow an entrant to operate with around 170 frequencies per week (provided that the entrant did not face any entry barriers at destination airports (1467)). 1201. By contrast, Aer Lingus uses around 540 slots (270 frequencies) on the overlap routes from/to Dublin for the Summer 2007 season, and around 700 (350 frequencies) when taking into account flights to London. Even if an entrant were to be found for all available slots (which is far from certain given the existing barriers and the absence of committed entrants), it would thus by far not reach the number of frequencies Aer Lingus can currently offer on the overlap routes.

1202. As regards aircraft available at Dublin Airport, Aer Lingus and Ryanair today operate with 23 and 20 aircraft respectively. Although Aer Lingus does not only serve the overlap routes with all 23 aircraft, the investigation confirmed that [4-8]* aircraft (1468) would not suffice to serve all overlap routes from/to Dublin (except Heathrow) on a scale that would constrain the merged entity.

1203. Aer Lingus itself, while keeping the exact number of aircraft used to serve the overlapping routes confidential, indicated that even for a hypothetical "minimum frequency (1469)" on each route (which would fall significantly behind Aer Lingus' actual offer), a competing airline would need around 10.5 aircraft (1470). In order to replicate also the dynamic competitive pressure that Aer Lingus currently exercises on Ryanair and have the necessary flexibility in its overall route portfolio, a minimum of two further aircraft would be necessary according to Aer Lingus (1471).

1204. This estimate is in line with the replies of other competitors. No respondents to the market test indicated that they would be able to operate with around 350 frequencies on the 34 overlap routes from/to Dublin with only [4-8]* or [6-10]* aircraft (1472). The estimates of these third parties varied between 12 and 20 aircraft, with an average of around 15 to 16 aircraft (1473). This is clearly above the [4-8]* aircraft (plus two aircraft for the Dublin - Heathrow route) offered by Ryanair to different competitors.

1205. Some respondents pointed to the fact that in view of the geographic location of Ireland, the many routes from/to Dublin are longer (in particular with respect to the destinations in Continental Europe) and thus the possible number of daily rotations of an aircraft based at Dublin is actually lower than four rotations (1474). The respondents also indicated that an additional spare aircraft may potentially be necessary to ensure reliability of the schedules and that maintenance operations also need to be taken into account.

1206. It can thus be concluded that, even if the slot remedy would stimulate entrants to use all offered slots, the entrant would still not reach the critical size which would allow it to restore the required degree of competition in Dublin on all affected overlap routes, without even considering the additional routes where the potential competition would be lost due to the transaction.

Slots at important destination airports are missing

1207. Although the text is ambiguous in this respect (1475), Ryanair appears to offer in "Annex 1" of the Final Commitments to make slots available if necessary at a certain number of destination airports. The Commission notes that this commitment does not reflect the Commission's analysis of the affected markets as "city-pairs" but offers slots only for airport-pairs. Furthermore, the offer to make slots available is often limited to secondary airports (for example, Paris Beauvais but not Paris CDG, Frankfurt Hahn but not Frankfurt Main, Charleroi but not Brussels, Milan Bergamo but not Milan Linate or Milan Malpensa etc.). Most of these airports are, however, usually not congested. Therefore, offering slots at secondary airports is not likely to remove any important entry barrier and to attract any additional entrant.

1208. The fact that Ryanair does not offer slots for all the affected overlap route (including primary airports) but limits its offer to one airport, which is in most cases a secondary airport, limits the likelihood of entry significantly. Indeed, such a remedy would facilitate entry only for the few airlines which accept to fly to secondary airports. As set out above (1476), even though these airports are substitutable from the demand side, this is not necessarily true for the supply-side. Indeed, most potential entrants identified by Ryanair itself operate from main airports. It is, therefore, unlikely that carriers such as Air France/CityJet or easyJet (with substantial operations at fully coordinated Paris CDG and Paris Orly airports) would find it viable to enter the Dublin - Paris route using the Paris Beauvais airport (which is not currently congested and where no "slot" divestiture is thus needed to enter). This limitation on the destination slots made available makes entry by a credible new competitor on the overlap routes substantially less likely.

1209. A large majority of the respondents to the market test of the Final Commitments have confirmed that the fact that some of the most congested destination airports such as Frankfurt Main, Paris CDG, Milan Linate or Milan Malpensa are not covered by the proposed commitments, would effectively prevent airlines from entering on these routes (1477). This further diminishes the likelihood that effective competition would be restored on these routes by the offered slots. The Commitments could only provide for "fragmented" entry which is less likely to restore effective competition

1210. As set out in detail above (1478), the Commission's market investigation showed that airlines operating not only with a few routes but from a large base enjoy significant advantages, for example, in terms of scale efficiencies and increased flexibility (route adjustment). The Final Commitments do not, however, provide for a solution that guarantees that a sufficiently strong competitor will establish itself with a large base at Dublin.

1211. On the contrary, the Final Commitments expressly exclude the possibility that one single entrant would take over all offered slots in order to establish a large base and restore competition, since the London/Heathrow slots are exclusively reserved to two different airlines. These slots can therefore not be acquired by any other airline that would wish to operate on the remaining overlap routes. Further, neither British Airways nor Air France are likely to take up also the other Dublin slots to base up to [4-8]* additional aircraft. This leads to an in-built fragmentation of any hypothetical entry at Dublin, which is likely to reduce the entrants' ability to compete head-to-head with the merged entity. The separation between the Heathrow slots and the other slots deprives this entrant of the possibility to combine this economically highly important route to London with other overlap routes. A majority of third parties have confirmed that the exclusion of any new entrant except British Airways and Air France from using the London/Heathrow slots would negatively affect its ability to compete effectively with Ryanair/Aer Lingus on the overlap routes. Competitors explained that the Dublin-London/Heathrow route is a "core business market", a "necessary backbone" and an "absolute must" for any new entrant who would like to establish operations of a certain scale at Dublin (1479).

1212. The market test of the Final Commitments also confirmed the Commission's findings in the previous market investigation that fragmented entry (that is to say entry by single competitors operating just on one or two single routes) is less likely to restore effective competition in Dublin than entry of a competitor with a large base. Indeed, a majority of the responses to the market test confirmed that a fragmented entry by more than one carrier would not be sufficient to eliminate competition concerns, since it would not allow the entrants to reach a "critical mass" of operations in Dublin. This may make the entrant more vulnerable to aggressive reaction and retaliation measures by the merged entity (1480). The Commitments disregard the operating model of the entrant

1213. As set out in the assessment part of the decision (1481), the ability to replace the competitive constraint Aer Lingus and Ryanair currently exert on each other depends also on the operating model of the hypothetical entrant. According to its operation and service model, as well as its cost structure, the competitor may be more or less able to offer a comparable alternative to the affected customers. In this context, it is important to note that the decisive criterion for the effectiveness of the offered commitments is not only whether a third airline could operate profitably on a given route. The Commission also has to assess the commitments from the perspective of the customers which would be affected by the disappearance of Aer Lingus as an independent airline and the closest competitor to Ryanair. Only if a new entrant would be able to offer an alternative which would be regarded as credible and sufficient substitute by the affected customers currently flying with Aer Lingus or Ryanair could the commitments be considered to address the significant impediment to effective competition identified above.

1214. The Final Commitments do not provide any clear requirements as to the type of operating model, service model or cost structure of the hypothetical entrant (1482). The market investigation has shown that the ability of an airline to offer services which are competitive from a customers' point of view can, for example, depend on the type of aircraft used on a given route. Due to the higher operating costs of smaller aircraft of less than 130 seats, carriers operating with such a type of aircraft are usually not able to offer as low prices as competitors operating with larger jets such as Ryanair or Aer Lingus (1483). These competitors therefore mainly focus on business customers and connecting passengers. While this strategy may allow an airline to operate profitably, it does not offer an alternative for the typical "low frills"/"point to point" customers of Ryanair and Aer Lingus.

1215. The commitments do not, therefore, provide a sufficiently clear-cut solution for the Commission to reject new entrants which are not likely to restore effective competition on the affected routes and also do not define workable criteria that would enable the Commission to decide with the required degree of confidence on approval of the new Entrant (1484).

8.2.2.2.The slot package for the London/Heathrow-Dublin route

1216. As concerns the slot package for the London/Heathrow-Dublin route (slots for eight daily rotations, to be equally split between Aer France/CityJet and British Airways), the Market Test has confirmed the Commission's analysis that British Airways and Air France (as well as at least one other competitor) are interested in obtaining these slots and in expanding their existing operations on the Dublin-London market. Although, due to the exclusivity clause for the two existing competitors, the offered slots would not lead to new entry on the Dublin-London route, the Commission acknowledges that the slot offer, should they actually be divested (1485), would most likely lead to expansion by competitors on this route which could constrain Ryanair/Aer Lingus on the Dublin-London route to a certain extent.

1217. However, the sufficiency and effectiveness of the commitment regarding the Heathrow slots is questionable for a number of reasons. It is not clear whether both airlines, being focused on business- and connecting passengers, could offer "low-fares" services in competition to Ryanair that would substitute the disappearing competitive pressure from Aer Lingus to a sufficient extent (1486). Indeed, both British Airways and Air France are "hub-carriers", focussed on feeding their respective hubs and not operating according to the point-to-point model (1487). British Airways and Air France/CityJet are thus less close substitutes to Ryanair than Aer Lingus. As discussed above, CityJet has a significantly lower capacity in its smaller planes and must therefore achieve a higher average yield than Ryanair/Aer Lingus to be profitable (1488). British Airways and Air France have significantly higher unit costs than either of Ryanair and Aer Lingus. Although both may be able to offer "cheap flights" to point-to-point customers to fill their capacity, it cannot be reasonably expected that their cost structure will allow them to offer tickets at a price which is comparable to Ryanair or Aer Lingus (1489).

1218. It is also uncertain whether the offered number of [2-6]* frequencies, offered exclusively to two different competitors (Air France and British Airways), is sufficient to impose effective competitive constraints on the merged entity. The Commission notes in this context that there is no good reason why the slots for London/Heathrow should be split between two "exclusive" acquirers instead of giving them to one single acquirer. Excluding other purchasers might prevent a solution in which a more appropriate buyer could take over the slots (for example, a low frills/point to point carrier which is closer to the Merging Parties' business model than Air France/British Airways (1490)). There is also no good reason why the slots should be "split", which weakens the entrants position vis-à-vis Ryanair/Aer Lingus compared to a situation in which all slots would go to one single competitor on this route. Indeed, the market share of Ryanair and Aer Lingus on the Dublin-London route amounts to [70-80%]*, while both Air France and British Airways hold less than 10% of the market. The fact that both competitors will only operate with [20-30]* weekly rotations (compared to 130 rotations offered by Ryanair) and that Ryanair offers only [6-10]* of Aer Lingus' 13 slot pairs at Heathrow, shows that the commitment, if it can even be implemented at all (1491), is very unlikely to effectively constrain the merged entity (1492).

1219. Taking also into account the lack of clarity of the Final Commitments with regards to important details of the slot release (1493) and the legal uncertainty whether the Heathrow slots can be divested at all (1494), the Commission concludes that the offered slots for London Heathrow cannot lead with the required degree of certainty to the entry of new competitors or expansion of existing competitors in the market which would eliminate the significant impediment to effective competition on the Dublin-London market.

8.2.2.3.The additional commitments

1220. Ryanair commits to immediately reducing Aer Lingus' short-haul fare by at least 10%, to eliminate Aer Lingus' fuel surcharges, to retain the Aer Lingus brand and to continue to operate the two airlines "separately", and commits for a period of six IATA seasons after completion of the merger not to increase the number of frequencies jointly operated by Ryanair and Aer Lingus on any of the claimed overlap routes in the event of new entry and not to reduce the frequencies on these routes unless a route is or becomes "unprofitable".

1221. These additional commitments are not structural in nature, but concern elements which are related to Ryanair's future behaviour. As set out in Section 8.1., these behavioural remedies are worded in very general terms described in the "Commitments Letter", without providing any further detailed rules for the functioning of the remedies (for example, definitions of key terms, detailed description of the functioning and the mechanism for monitoring these remedies). The absence of workability of the proposed remedies does, therefore, not allow the Commission to consider them as a solution for the identified competition problems. Moreover, also the content of these remedies cannot eliminate the identified competition concerns.

1222. In particular, Ryanair's proposed commitment to immediately reduce Aer Lingus' fares by 10% does not provide any certainty as to its content and duration. Any reduction could, theoretically, be withdrawn immediately, since Ryanair does not indicate how long such a reduction should be in place. Further, the commitment seems to refer to average fares and does not guarantee that fares on any overlap route would be reduced. It does not prevent Ryanair from making selective price increases on particular overlap routes. There is also no guarantee that the quality of Aer Lingus' service would not be negatively affected by the price reduction, neither would Ryanair be prevented from increasing Aer Lingus' other price components declared as "taxes" or "charges". In addition, the 10% price reduction does not exclude the possibility that existing Aer Lingus services on overlap routes are closed down. The Commission also notes that the commitments only relates to Aer Lingus prices and does not exclude possible Ryanair price increases. More generally, it is impossible for the Commission to anticipate what would constitute a "competitive" price level at a future point in time in a market that has seen significant fluctuations in average fares and in input costs, such as fuel, from year to year (sometimes in excess of 10%). A key function of competitive markets is precisely to discover efficient outcomes in terms of prices, output, quality and similar parameters. A static price commitment would not restore this important market function.

1223. With respect to Ryanair's commitment to eliminate Aer Lingus' fuel surcharges, the Commission notes that Aer Lingus applies fuel surcharges only to its long-haul flights. Hence, this commitment is not directly related to the short-haul overlap markets in which the Commission has identified competition concerns and is not likely to have any impact on the short-haul routes affected by the concentration (1495). The Commission agrees with third parties which have also stated in the market test of the Final Commitments that the monitoring of such a type of commitment would be complex and hardly workable (1496). Hence, this commitment does not alter the competitive impact of the proposed transaction.

1224. Ryanair further commits to retaining the Aer Lingus brand and to continuing to operate the two airlines separately. This remedy does not directly address any identified competition concern. Moreover, the content and the effect on competition of the offer to operate the two airlines separately is entirely unclear, and there is no explanation on how to monitor this commitment. It may also be noted that based on any plausible economic theory, it is not clear how the proposed commitment would even partially remove the significant impediment to effective competition created by the notified transaction. Assuming profit-maximising behaviour, common ownership of Ryanair and Aer Lingus is sufficient to lead to the anticompetitive effects set out above (1497). If anything, a dual branding strategy may enable Ryanair to exploit its market power more effectively and, thus, extract additional profits as compared to a single-branding strategy from its customers. Dual branding can make it easier for a firm to charge higher prices to customers with a higher willingness to pay and lower prices to more price-elastic customers, increasing the firm's profits.

1225. Ryanair finally proposes a "frequency freeze" remedy, by which it commits for a period of six IATA seasons after completion of the merger not to increase the number of frequencies jointly operated by Ryanair and Aer Lingus on any of the claimed overlap routes in the event of new entry and not to reduce the frequencies on these routes, unless a route is or becomes "unprofitable". Again, the precise content of this commitment (for example, the content of the concept of an "unprofitable route") is not defined in Annex 1 or anywhere else in the text of the offered commitments. It is also not clear what the intended effects on competition of the proposed remedies would be and how they would address the identified competition problems on the respective routes (1498).

1226. Therefore, the Commission is of the view that the additional remedies proposed by Ryanair cannot eliminate the significant impediment to effective competition, neither on a stand-alone basis, or as part of a package of remedies.

8.2.2.4.It is unclear whether the proposed commitments can be implemented without the consent of other Aer Lingus shareholders

1227. While it is already clear from the previous paragraphs that the Final Commitments cannot be accepted, the Commission notes that there are also doubts as to whether Ryanair's commitments could be implemented in a successful, full and timely manner. (1499)

1228. Aer Lingus claims that under Irish company law and Aer Lingus' Articles of Association, the offered commitments in general, and the offer to surrender slots at London Heathrow in particular, cannot be lawfully delivered by Ryanair. Aer Lingus made several arguments to substantiate its claims (1500) and Ryanair stated its views in a memorandum of 4 December 2006 and a submission of 30 May 2007. The Irish Department of Transport also submitted that the provision in the Articles of Association of Aer Lingus in relation to the Heathrow slots introduces a degree of uncertainty concerning the implementation of this commitment, such that it should not be accepted. (1501)

1229. Without it being necessary to assess all these issues in full, the Commission considers that there are, in particular, a number of serious doubts as to whether Ryanair's commitment to make available, for an indefinite period of time, [6-10]* of Aer Lingus' 13 pairs of Heathrow slots, is likely to be implemented in a successful, full and timely manner.

1230. Article 10 of the Articles of Association of Aer Lingus contains a specific procedure for the disposal of the London Heathrow slots involving the shareholders of Aer Lingus. According to Aer Lingus, the Irish government and the Aer Lingus Employee Share Option Trust ("ESOT"), both shareholders in Aer Lingus, can together block any disposal of Aer Lingus' slots at London Heathrow. Ryanair seems not to dispute this but puts forward that these slots could be made available by way of short term leases, which are exempted from the rules on disposal of Heathrow slots that require the involvement of Aer Lingus' shareholders. (1502)

1231. Although the text of that provision may be open to different interpretations, it seems nonetheless reasonably clear that, even on the narrow and technical reading of Article 10 advocated by Ryanair, Aer Lingus would most likely not be able to lease the eight pairs of London Heathrow slots for a period of more than 36 months by way of short term leases. Any subsequent lease for the same slots (whether in favour of the same or different parties) would no longer qualify as a short term lease but as a disposal transaction. Consequently, Aer Lingus would not be able to enter into a second set of short term leases for these eight pairs of London Heathrow slots.

1232. Aer Lingus has argued that its Articles of Association would only allow the company to enter into a lease of one slot for up to 36 months. While the Commission cannot, on the basis of the information available, determine whether the competent Irish court would construe Article 10 in that sense or in the meaning advocated by Ryanair, Aer Lingus' position seems prima facie not implausible. In any event, in view of all legal arguments put forward by the parties, there is clearly scope for legal dispute, which in itself is sufficient to cast serious doubts on Ryanair's capability of delivering this remedy in time. (1503)

1233. Consequently, without taking a position on the other arguments raised by Aer Lingus, the Commission concludes that it is not sufficiently likely that the commitment of Ryanair to make available eight pairs of Aer Lingus' London Heathrow Slots could be implemented in a successful, full and timely manner.

8.2.2.5. Conclusion on the substance of the commitments

1234. In view of the above and irrespective of the formal shortcomings identified in Section 8.2.1., the Final Commitments would not be sufficient to remove all the identified competition concerns.

8.3. Conclusion

1235. On the basis of the examination of the Final Commitments, the Commission concludes that those commitments are not sufficient to remedy the identified significant impediment to effective competition and, thus, cannot render the proposed concentration compatible with the common market.

8.4. Draft commitments submitted on 1 June 2007

1236. On 1 June 2007, following the State-of-Play meeting on 29 May 2007 (1504) and subsequent additional discussions, Ryanair submitted, in draft form, revised commitments (Draft Modified Final Commitments) (1505). According to Ryanair, the proposed modifications aimed at addressing the identified shortcomings of the Final Commitments of 3 May 2007. Having analysed the Draft Modified Final Commitments, the Commissioner for Competition informed Ryanair by letter of 4 June 2007 that she did not consider these draft commitments sufficient to resolve the competition concerns raised by the transaction and that she would be obliged to draw the necessary conclusions when proposing a final decision for adoption by the College of the Commission (1506).

1237. It must be underlined that following the Final Commitments of 3 May 2007, Ryanair did not formally submit any new or modified commitments. The Draft Modified Final Commitments were provided explicitly in draft form, without signature and without complying with the formal requirements necessary pursuant to Article 20 of the Commission Regulation (EC) No 802/2004 of 7 April 2004 implementing Council Regulation (EC) No 139-2004 of 20 January 2004 on the control of concentrations between undertakings (1507) ("the Implementing Regulation"). Since Ryanair has not formally submitted new commitments, the Commission is not obliged to assess them in the present decision. In particular, the Commission could not make an authorisation conditional on compliance with such draft commitments. In any event, even if the Draft Modified Final Commitments had been formally submitted, they would still not have been sufficient to allow the Commission to conclude, on the basis of information already available, that they fully and unambiguously resolve the competition concerns identified. It must be noted that the last day for submitting commitments, in line with the Implementing Regulation, was 3 May 2007. Therefore, the Draft Modified Final Commitments were provided 17 working days after this deadline and thus very late in the procedure. Further, the Implementing Regulation provides in Article 19(2) that, in exceptional circumstances, the Commission may accept commitments offered after the expiry of the time limit, provided that the procedure provided for in Article 19(5) of the Merger Regulation is complied with. Furthermore, paragraph 43 of the Remedies Notice indicates that where the parties subsequently modify the proposed commitments, the Commission may only accept them where: (i) it can clearly determine - on the basis of its assessment of information already received in the course of the investigation, including the results of prior market testing, and without the need for any other market test - that such commitments, once implemented, resolve the competition problems identified, and (ii) there is sufficient time for proper consultation of Member States. (1508)

1238. The Draft Final Modified Commitments clearly would not have been sufficient to address all the above described shortcomings of the Final Commitments. In particular, the Draft Final Modified Commitments are still based primarily on slot transfer (enabling access to airport infrastructure) and do not provide any new elements which would address the other identified barriers to entry and enable thus the Commission to re-evaluate the negative results of the market test as to likelihood of an actual entry (see Section 8.2.2.1, first subheading). Further, the scope of the guaranteed new entry is still insufficient as the commitments provide for a new up-front entrant with only [4-8]* (1509) based in Dublin plus two aircraft for the Dublin - London Heathrow route (which is according to the results of the market test clearly insufficient - see Section 8.2.2.1, third subheading). The London Heathrow slots are still reserved only for some competitors (1510) and thus one single upfront buyer is not guaranteed, leading to a possibility of fragmented entry (see Section 8.2.2.1, fifth sub-heading). Further, there remains the legal uncertainty with respect to the London Heathrow slots (see Section 8.2.2.4). The Draft Modified Final Commitments also do not provide for the transfer of slots at all relevant destination airports, in particular not for slots at congested airports (see Section 8.2.2.1, fourth subheading) (1511). As regards the up-front buyer provisions, while some of the ambiguities and uncertainties appear to be addressed, the criteria for the up-front buyer would still not be sufficiently specific to ensure the timely entry of a competitor which would provide sufficient competitive constraints to the Merged Entity on all identified overlap routes (see Section 8.2.2.1, second subheading). Further, the draft still disregards the operational (business) model of a potential entrant (see Section 8.2.2.1, sixth subheading). In addition, a number of ambiguities and uncertainties would remain with respect to the implementation and workability of the commitments, for example with respect to slot transfer mechanism, dispute resolution, monitoring of the additional commitments or implementation of the up-front provisions (see Section 8.2.1 and following).

1239. Therefore, even if the Draft Modified Final Commitments had been formally submitted, they would clearly not have allowed the Commission to determine with the required degree of certainty (and without the need for another market test) that such commitments, once implemented, would sufficiently resolve the identified competition concerns caused by the proposed concentration.

9. CONCLUSION

1240. For the reasons mentioned above, the Commission concludes that the notified concentration would significantly impede effective competition in the common market or a substantial part thereof within the meaning of Article 2(3) of the Merger Regulation, in particular as a result of the creation of a dominant position of Ryanair and Aer Lingus on 35 routes from and to Dublin, Shannon and Cork, and the creation or strengthening of a dominant position on 15 other routes from and to Dublin and Cork. The concentration must, therefore, be declared incompatible with the common market and the EEA Agreement pursuant to Article 8(3) of the Merger Regulation and Article 57 of the EEA Agreement.

Has adopted this decision:

Article 1

The concentration whereby Ryanair acquires sole control of Aer Lingus within the meaning of Article 3(1)(b) of Regulation (EC) No 139-2004 is hereby declared incompatible with the common market and the EEA Agreement.

Article 2

This decision is addressed to:

Ryanair Holdings plc

Ryanair Corporate Headquarters,

Dublin Airport

Ireland - County Dublin

Notes

1 OJ L 24, 29.1.2004, p. 1.

2 OJ C ...,...200. , p....

3 OJ C ...,...200. , p....

4 See: http://www.ryanair.com/site/EN/about.php?page=About&sec=fleet.

5 Interlining agreements allow airlines to combine their own flights with flights from other airlines in order to offer enlarge their portfolio of destinations. See in detail below.

6 See e-mail of Ryanair (A&L Goodbody) of 19.12.2006, folio no. 9861, and paragraph 866 of Ryanair's response to the Statement of Objections.

7 Turnover calculated in accordance with Article 5(1) of the Merger Regulation and Commission Notice on calculation of turnover under Council Regulation (EEC) No 4064-89 on the control of concentrations between undertakings (OJ C66, 2.3.1998, p. 25). Hereinafter referred to as "the Notice on the calculation of turnover".

8 Ryanair EUR 1,692 million, Aer Lingus EUR 883 million. [...]*

9 Ryanair EUR [...]*, Aer Lingus above [...]* EUR.

10 See paragraph 5 of the Notice on the calculation of turnover; CFI judgment in the case T-417-05 - Endesa v. Commission, of 14 July 2006.

11 See paragraph 7 of the Notice on the calculation of turnover.

12 See the Commission decisions in cases M.130 - Delta Airlines/PanAm, M.157 - Air France/Sabena, M.259 British Airways/TAT, M.616 - Swissair/Sabena (II), M.857 - British Airways/Air Liberté, M.1354 - SAirGroup/LTU, M.1494 - Sair Group/AOM.

13 E.g. the Commission decision in case M.130 - Delta Airlines/PanAm.

14 According to Aer Lingus, [70-80%]* of all its intra EC tickets in the period November 2005 - October 2006 was sold over the Internet.

15 Draft Commission Consolidated Jurisdictional Notice under Council Regulation (EC) No 139-2004 on the control of concentrations between undertakings, published on 28.9.2006, available at the Commission website: http://ec.europa.eu/comm/competition/mergers/legislation/jn.pdf.

16 Turnover of Aer Lingus allocated according to this methodology would exceed EUR 25 million only in Ireland (EUR [...]*) and the United Kingdom (EUR [...]*). The third largest national turnover within the Community amounting to EUR [...]* would be generated in [...]*.

17 Ryanair currently sells around [90-100]*% of tickets via Internet while Aer Lingus around [70-80%]*.

18 Case T-417-05 Endesa v Commission judgment of 14 July 2006.

19 The Draft Consolidated Jurisdictional Notice published on 28 September 2006 states in paragraph 170: "Air transport cases fall outside the categories set out above as the service consists in enabling the customer to travel. The turnover generated by air transport is to be attributed to the location of the customer at the moment at which provision of the service commences, i.e. the place of departure. This is normally the country where the ticket was bought."

20 See the third subsidiary criterion mentioned in paragraph 46 of the Notice on calculation of turnover. This observation relates solely to the identification, for a whole category of cases and for the sole purposes of determining the geographical allocation of turnover, of the place where a given notifying party could be thought likely to face competition to provide services to customers departing from a given airport. It is without prejudice to the definition of markets for the purposes of the competitive assessment, which depends on a concrete examination of the circumstances of each individual case.

21 Aer Lingus cannot rely on the Draft Consolidated Jurisdictional Notice, which has not been formally adopted by the Commission yet.

22 It should be noted that this does not prejudge the conclusion on whether it would be necessary to split return tickets also in case of more traditional network carriers selling the traditional return tickets under more advantageous conditions than two one-way tickets.

23 Ireland (Ryanair EUR [...]*, Aer Lingus around EUR [...]*), the United Kingdom (Ryanair EUR [...]*, Aer Lingus EUR [...]*), Spain (Ryanair EUR [...]*, Aer Lingus EUR [...]*).

24 Ireland (Ryanair EUR [...]*, Aer Lingus EUR [...]*), the United Kingdom (Ryanair EUR [...]*, Aer Lingus EUR [...]*), Spain (Ryanair EUR [...]*, Aer Lingus EUR [...]*).

25 As opposed to corporate customers which can be contacted via the respective corporations. The Merging Parties sell a large majority of their tickets over the internet and the share of corporate customers in total sales is limited.

26 See also Annex I, answers to question 9 and 3 of the Customer Survey.

27 See for a more detailed discussion of the criticism Section 7.3.5 and Annex I.

28 See e.g. Ryanair's quotes of single third party replies to the Commission's market investigation or the market test of the remedies in Ryanair's response to the Statement of Objections.

29 It should be noted that the number of customers who have been contacted and whose answer has been analysed in the Customer Survey carried out on behalf of the Commission is largest number of third parties the Commission has ever contacted in a merger investigation.

30 See page 36 of the Ryanair's Response to the Statement of Objections: "Detailed examination of the responses makes clear that respondents often did not understand the questions with many contradictory answers".

31 The Commission analyses in particular answers by competitors very carefully, since they may have an interest to make the transaction of their competitors more difficult, in particular in the framework of a contested bid. However, in the present case many airlines were not directly affected by the merger, and other airlines indicated even that they would welcome the merger since they were generally in favour of airline consolidation. Indeed, since a number of airlines have just undergone merger control procedures at the occasion of national or European mergers, carriers interested in future acquisition might be rather interested to avoid the view that airline consolidation could have negative aspects.

32 For example, the first questionnaires to competitors and corporate customers were made on the assumption that it may be relevant to define separate markets for so-called time-sensitive and non-time-sensitive passengers in this case, see Section 6.8 below.

33 In its Response to the Statement of Objections, Ryanair quotes predominantly from questionnaires in Phase I and largely ignores the more detailed "follow-up" questionnaires and minutes with the respective third parties.

34 In case of unclear or contradicting answers, the Commission has attached greater weight to the later detailed conversation than to the previous written answer.

35 Also referred to as "carriers" in this decision.

36 Without a hub, a direct route from every origin to every destination would be necessary.

37 See CFI, T-177-04 EasyJet v Commission, of 4 July 2006, ECR [2006], II-1913, paragraph 118.

38 See below, Section 7.3.4.

39 It should be noted that most network carriers are members of international airline alliances.

40 Since smaller regional airlines with small aircraft need more frequencies than carriers with larger aircraft to transport the same passengers and have higher operation costs, they are usually only a limited competitive constraint to larger point-to-point carriers.

41 Typically, such carriers would be the former small to medium sized "flag carriers" in Europe. The term "flag carriers" refers to national airlines which were (or still are) state-owned and considered as the countries' only or leading airline (such as Air France for France, Lufthansa for Germany, British Airways for the UK etc.). Most of these airlines operate according to the network model. However, some smaller "flag carriers" (such as Aer Lingus) have changed their operating model into a low-frills model, while retaining some long-haul operations as a "second limb" of their operations. Such small flag carriers continue to operate a limited number of long haul services from their main airport. These operations and/or the portfolio of destinations that are otherwise available on a direct service from their main airport are not sufficient for these carriers to operate according to the hub and spoke model (see paragraph 42 above). To a varying degree however, these carriers seek to maintain some connectivity at this airport to feed their long haul services and are willing to adapt their operation and product offering to this effect.

42 See in greater detail in Section 7.9.

43 See the Notification, paragraph 201.

44 See below Section 6.8.

45 Aer Lingus changed its former business model dramatically after 2001 from a "traditional" full-service flag carrier by significantly reducing the service offered. Today, Aer Lingus' European operation is perceived by most customers rather as a "low-frills" than a "full-service" business model , even though Aer Lingus still offers a somewhat more elaborate and "upmarket" product than Ryanair. Aer Lingus fulfils many criteria that qualify them as a low-frills carrier (very high percentage - more than 70% - of direct distribution over the Internet, only one-way restricted fares, baggage fees, single economy-type cabin class service, no complimentary meals on board, etc.). The market investigation has indeed confirmed that Aer Lingus has changed significantly in the recent years (see in particular replies to question 30 of the questionnaire sent to competitors on 6 November 2006 and question 8 and 15 of the questionnaire sent to customers on 31 October 2006). Further, the presentation "Aer Lingus Brand & Advertising Study" prepared for Aer Lingus by Research Solutions in January 2006 confirms the increased perception of Aer Lingus as a low fares airline (see also e-mail from Aer Lingus (Simone Warwick) of 27 March 2007) , folio no. 6316.

46 See the above footnote. For more details, see also Section 7.3 below.

47 CityJet is a 100% a subsidiary of Air France. On the services it operates to and from Paris CDG airport, CityJet acts as a provider of feeder traffic into Air France's hub airport.

48 For more details on individual competing airlines, see Section 7.8.9.

49 See in particular Ryanair's submission of 21 November 2006, folio no. 22736.

50 See in detail below in particular in Section 7.4.

51 Notification, paragraphs 170 and 183.

52 See e.g. cases M.3940 - Lufthansa/Eurowings, paragraph 10; M.3770 - Lufthansa/Swiss, paragraph 12; M.3280 - Air France/KLM, paragraphs 9-18 (confirmed by CFI, case T-177-04 easyJet v Commission, of 4 July 2006 ECR [2006], II-1913, at paragraphs 54-61; M.1855 - Singapore Airlines/Virgin Atlantic, paragraph 16; M.1494 - Sair Group/AOM, paragraph 14; M.857 - British Airways/Air Liberté, paragraph 14; M.278 - British Airways/DanAir, paragraph 10 (confirmed by CFI case T-2-93 Air France v Commission , of 19 May 1994 ECR ), ECR (1994), 320, at paragraphs 84 and 85); M.157 - AirFrance/Sabena, paragraph 25; see also ECJ, case 66-86 of Ahmed Saeed Flugreisen, of 11 April 1989 ECR (1989), 803, at paragraphs 39- 41.

53 See ECJ, case 66-86 Ahmed Saeed Flugreisen of 11 April 1989 ECR (1989), 803, at paragraphs 39-41; See CFI case T-2/93 Air France v Commission , of 19 May 1994 ECR 320, at paragraphs 84 and 85; See CFI, case T-177-04 easyJet v Commission, of 4 July 2006 ECR [2006], II-1913, at paragraphs 54-61.

54 See case M.3280 - Air France/KLM.

55 See e.g. Notification, paragraph 73; see also interview with easyJet of 15.2.2007, folio no. 6170.

56 For more details about the position of Ryanair and Aer Lingus in Ireland and in particular Dublin, see below Sections 7.2.

57 See for a more detailed analysis of the advantages of operating from airport bases in Section 7.3. below.

58 See in detail Section 7.8.3 below.

59 Other third parties mentioned that the Irish market is also different with respect to the absence of "pass by" customers which fly on to other destinations from Ireland see interview with easyJet of 15.2.2007, folio no. 6170. See further below in Section 7.8.6.

60 E.g. the common base in Dublin, the advantage of brand recognition for Irish operators, the possibility to shift flights between different destinations out of Dublin etc.

61 See e.g. submission of the DOT to the Commission of 13.11.2007, p 49, folio no. 6444.

62 See in particular paragraph 13 of the Commission Notice on the definition of the relevant market, OJ C 372, 09/12/1997, p. 03.

63 See e.g. case M.3770 - Lufthansa / Swiss, paragraph 12

64 See more in detail below in Section 7.8.3.

65 idem.

66 idem.

67 See further Commission Notice on the definition of the Relevant Market for the purposes of Community competition law, OJ 97-C 373-03), para. 20 ff.

68 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 19; it should be noted that even some network carriers, who in previous cases involving network carriers voted for a wider definition in order to cover the "network effects" (e.g. FFPs), supported the O&D approach in the present case involving two point-to-point carriers.

69 The Commission does, however, recognise that the single "O&D" markets are not entirely independent from each other and will thus take account of the commonalities between different routes and of supply-side substitutability and other forms of potential competition considerations whenever appropriate.

70 See below Section 7.9.

71 See already M.3940 - Lufthansa/Eurowings, paragraph 10; M.3770 - Lufthansa/Swiss, paragraph 11; M.3280 - Air France/KLM, paragraph 12 and, CFI, case T-177-04 easyJet v Commission, of 4 July 2006 ECR (2006), II-1913, at paragraph 56; M.1855 - Singapore Airlines/Virgin Atlantic, paragraph 16 with further references.

72 See the Commission decisions in cases COMP.37.730 Lufthansa/Australian Airlines, M.3280 - Air France/KLM and COMP/38.712 British Midland/Lufthansa/SAS.

73 These routes are Dublin - Manchester, Dublin - Birmingham, Dublin - Newcastle and Dublin - Alicante.

74 The Commission has notably sought the view of the affected airports, competitors and the Civil Aviation Authorities of the respective Member States and conducted a price correlation analysis.

75 The Commission has also verified to what extent the airport of Dublin is potentially substitutable with the airport in Belfast.

76 See replies to the Airport Questionnaire (substitutability) send on 3 January 2007, in particular questions 3 and 4.

77 See in detail Section 6.8.

78 Notification, paragraph 81, and Aer Lingus IPO Prospectus, page 67.

79 See e.g. minutes of the interview with Aer Arann of 13 February 2007, folio no. 6170, and reply of British Airways to the Questionnaire to Competitors sent on 6 November 2006, folio no. 22168, question 22 .

80 This is in line with evidence submitted by the DOT, which provided a table for the airports relevant for this case indicating the time difference, accounting for these factors, as a percentage of total travel time that it takes to reach the destination city. The table shows that, on average, passengers must spend just 15% more time travelling if they choose to travel to an adjacent airport compared to travelling to a city airport; See: A response to RBB's paper "Comments on the LECG report for the DOT", LECG, 29 November 2006, folio no. 6145.

81 See e.g. replies to the questionnaire to airports ("Airport Questionnaire (substitutability)") sent on 3 January 2007, in particular questions 10 and 11.

82 The Commission notes that studies conducted by the UK Civil Aviation Authority suggest that the catchment area of airports in the United Kingdom extends up to 2 hours driving time. See e.g. Airport price control review - Initial proposals for Heathrow, Gatwick, Stansted - December 2006 available at http://www.caa.co.uk/default.aspx?categoryid=5&pagetype=90&pageid=7162.

83 See in this respect also Ryanair's Response to the Statement of Objections, paragraph 462 ("the reality is much more complex").

84 Section 6.8.

85 Ryanair has in particular submitted data on the distribution of passengers using London airports by County outside the Greater London and by Metropolitan District within the Greater London in its reply to the Commission's Art 6(1)(c) decision in order to show that the "distance or time to city centre"-criterion referred to by the Commission is "too simplistic".

86 See press release of Ryanair "RYANAIR ANNOUNCES BIGGEST EVER EXPANSION AT DUBLIN 12 NEW EUROPEAN ROUTES START FROM DECEMBER" of 09/08/2006. Similarly in its press release "RYANAIR OPENS 2 NEW ROUTES TO WARSAW AND STOCKHOLM & EXTENDS PRIORITY BOARDING OPTION TO ALL PASSENGERS" of 18/10/2006, Ryanair compares its fares with the fares of Aer Lingus on a number of routes where Ryanair and Aer Lingus serve different airports, including Tenerife, Bilbao/Vitoria, Bologna or Lyon//Grenoble.

87 See e.g. "Hahn", "Girona" or even "Stansted"

88 See e.g. answers to question 22 of the Commission's Questionnaire to Competitors sent on 6 November 2006.

89 The Commission conducted a price correlation analysis for seventeen city-pair routes out of Dublin for which sufficient data was available. The scope and methodology for this empirical analysis is explained in Annex III to this Decision.

90 See the results of the Customer Survey as described in Annex I.

91 The Customer Survey only includes a subset of the routes where the Merging Parties serve different airports. The Customer Survey is therefore informative only for those routes.

92 In its responses to the Statement of Objections, Ryanair alleges that the Commission ignores the results of its own investigation and that it disregards the views of major corporate customers such as CRH, Unilever, IBM EMEA, Siemens or Microsoft. The Commission contests these allegations. As set out in Section 5, in cannot base its conclusions on single answers to the market test but has to weigh the validity of each answer and compare it with the other evidence in the case file. Furthermore, Ryanair seems even to "cherry pick" those responses from corporate customers which fit the argument they want to make. For example, Ryanair underlines the view of Siemens that Manchester, Liverpool and Leeds Bradford are not substitutable whereas they make no mention of the views of Siemens that Birmingham and East Midlands airports are substitutable.

93 OJ L 240, 24.8.1992, p. 8.

94 The importance of these elements for the purposes of influencing the purchasing behaviour of passengers should not be underestimated as evidenced from Ryanair's internal documents: [...]* (Quote from Board Paper 4 (Board Meeting [...]*) : PAGE 4.2, folio no. 629)

95 It should be noted that the Commission was only able to conduct such analysis on 17 routes as for the other routes there was not sufficient data available to the Commission to conduct a price correlation. The Commission was therefore not able to provide such a test on all relevant airport/city pairs. The scope and methodology for this empirical analysis is explained in Annex III to this Decision.

96 Notification page 104. The argument was made in relation to the competitive position of Dublin Airport.

97 See Ryanair's submission of 26 January 2007, first paragraph of the cover letter.

98 Approximately one and a half hour away by car according to Ryanair, Notification, para. 301.

99 See e.g. minutes of the interview with easyJet of 15 February 2007, folio no. 6170, as well as replies to question 23 of the Questionnaire to Competitors sent on 6 November 2006.

100 Questionnaire of 9.02.2007 on substitutability between Dublin and Belfast airports, folio no. 3870.

101 See reply of Belfast International of 02/03/2007 to the Questionnaire on substitutability between Dublin and Belfast airports, folio no. 6128, in particular annexed CAA 2006 Survey - Key Passenger Profile Findings, January 2007.

102 See also in Annexes I and II, the complete result and presentation of the Customer Survey conducted in Dublin Airport by the Commission.

103 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 380.

104 See the Notification, paragraph 92.

105 While Ryanair reacted very aggressively to easyJet's entry to the routes between Shannon/Cork/Knock and London in January 2005 [...]* (for more details see below Section 7.8.5), there was no such reaction to the growth of the easyJet's base at Belfast International airport (e.g. when easyJet announced in September 2006 a new route Belfast - Krakow and further expansion of its Belfast base - see press release of easyJet "easyJet announces Belfast expansion" of 21 September 2006, http://www.easyjet.com/en/News/belfast_to_krakow.html). ???Parts of this text have been edited to ensure that confidential information is not disclosed; those parts are enclosed in square brackets and marked with an asterisk. 106 See e.g. Board Paper 3 for the Board Meeting of Ryanair Holdings plc on [...]*, page 3.15. and others (folio no. 629).

107 Source: viamichelin.com and rac.com

108 Source: market investigation: responses to airport substitutability questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)]

109 Source: market investigation: responses to airport substitutability questionnaire

110 Source: ryanair.com

111 Average journey time by coach (National Express) is 65 minutes; market test suggests 135 minutes of journey time by bus.

112 Average journey time by coach (National Express) is 90 minutes; market test suggests 145 minutes of journey time by bus.

113 See the Commission decision 92/213 in case IV/33.544 British Midlands/Aer Lingus, of 26 February 1992.

114 In its Response to the Statement of Objections, Ryanair refers to various pieces of evidence with a view to showing the non-substitutability of London airports. These pieces of evidence point to the special position of London Heathrow airport as an important connecting point, in its capacity as one of Europe's major hub airports. Ryanair refers in this respect to special provisions in the articles of association of Aer Lingus in respect of the latter's slots at London Heathrow airport. The purpose of these provisions appears to be to safeguard the connectivity of Ireland with the rest of the world, in connection with services that are available from Heathrow. As explained in Section 6.3.1 above however, it follows from the Commission's application of the "Origin and Destination" (O&D) approach that only point to point air transport services on the relevant markets are considered in the present decision. Services on O&D pairs that are not relevant are disregarded, even if such services imply that, as part of the overall journey on this O&D pair, one or more legs is flown on an O&D pair which is relevant for the present decision. It follows that Ryanair's objections concerning the specificity of London Heathrow and in particular Aer Lingus' slots at this airport do not change the present assessment.

115 See on the limited "precedent value" of previous merger cases and the Commission's obligation to analyse markets also case T-210/01 - General Electric v Commission [2005] ECR II-5575, paragraphs 118-119

116 See Commission decision in the case M.967 - KLM/Air UK of 22/09/1997, paragraph 24, Commission decision in the case COMP/D2/38.479: - British Airways/Iberia/GB Airways of 10/12/2003, paragraphs 21- 24 of the summary of the Commission's assessment published on 10/01/2004.

117 Figures according to the population census of 2001, available from http://www.statistics.gov.uk/census2001/pyramids/pages/h.asp

118 On the basis of mere distance considerations, it can for example not be excluded that passengers starting their journey in Croydon would consider services to Dublin that are available from London City as well as from Gatwick Airport and not exclusively services available from Gatwick Airport as suggested by the study of York Aviation.

119 See in particular the following documents of the UK Civil Aviation Authority: CAA Passenger Survey Report 2005 available at http://www.caa.co.uk/default.aspx?categoryid=81&pagetype=90&pageid=6554 and Airport price control review - Initial proposals for Heathrow, Gatwick, Stansted - December 2006, available at http://www.caa.co.uk/default.aspx?categoryid=5&pagetype=90&pageid=7162..

120 See replies to question 22 of the Questionnaire to Competitors of 06.11.2006 and in particular replies of Aer Lingus (folio no. 4122), BMI (folio no. 22283), CityJet (folio no. 1990) and British Airways (folio no. 22168).

121 See replies to the Questionnaire to Airports sent on 9 November 2006 and the Airport Questionnaire (substitutability) send on 3 January 2007.

122 http://journeyplanner.tfl.gov.uk

123 See in particular page 211 of the response, Ryanair's comments on paragraph 78 of the Statement of objections.

124 Source: www.ryanair.com

125 See in particular, page 3.2 of Board paper 3 (Board meeting of [...]*), page 3.2 of Board paper 3 (Board Meeting [...]*), pages 2.2 and 2.3 of Board paper 4 (Board Meeting [...]*) , page 3.2 of Board paper 3 (Board Meeting [...]*), page 3.2 of Board paper 3 (Board Meeting [...]*), Page 3.5 of Board paper 3 (Board Meeting [...]*), page 3.3 of Board paper 3 (Board Meeting [...]*), Pages 1.3 of Board paper 1 and 3.4 of Board paper 3 (Board Meeting [...]*), Page 3.2 of Board Paper 3 (Board Meeting [...]*). All under folio no. 629.

126 See Minutes of the Board Meeting of Ryanair Holdings plc on [...]*, page 1.3., which were presented as Board Paper 1 for a Board Meeting of Ryanair Holdings plc on [...]* (folio no. 629).

127 See for example the forecasted business plan dated [...]* prepared by Ryanair for the [...]* period (see folio no. 629, Annex 2.1) with a view to demonstrating [...]*.

128 It is important to note with respect to all the analysed city pairs that the Commission has taken the economic evidence on price correlation into account as additional elements that support its view. Therefore, even the absence of this confirmation by the price correlation analysis would not imply that the services in question belong to separate markets. On the one hand, a positive price correlation over time serves as a meaningful confirmation that two or more products or services belong to the same relevant market. On the other hand, a negligible or negative correlation of prices over time does not necessarily imply that the products/services belong to separate markets. This is because prices movements may be influenced by other factors which can lead to low price correlation. In the particular case of the Aer Lingus service between Dublin and London Heathrow, substantial variations in the number of connecting passengers can be observed. A different seasonality pattern for these inter-connecting passengers or a different pattern of demand may also lead to a low correlation coefficient and would not reflect the underlying structural relationship between the two price series. See Annex III for further details.

129 See Annex I.

130 Ryanair, in its Response to the Statement of Objections compares the 18.4% of Aer Lingus passengers considering Ryanair with the 16.7% of passenger considering Belfast as an alternative airport. It argues that the differences are low and thus it is inconsistent to regard Heathrow as an alternative to other London airports but not Belfast to Dublin. However, as explained in the previous section the two questions are very different. Firstly the question on Belfast asks whether the respondent would ever consider Belfast and does not cover for which route and which airline (as noted neither Aer Lingus nor Ryanair flies from Belfast). It is thus a significant bias towards a positive answer. In contrast the 18.4% refers to actual past behaviour, that is to say whether the respondent flying with a certain carrier actually considered a specific other carrier on the London route for the particular flight in question.

131 Ryanair's customers were questioned on a weekday for this route (see also Annexes I and II).

132 See Ryanair response to the Statement of Objections (paragraph 413).

133 Source: viamichelin.com and rac.com

134 Source: market investigation: responses to airport substitutability questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

135 Source: market investigation: responses to airport substitutability questionnaire.

136 Source: www.ryanair.com.

137 Source: www.manchesterairport.co.uk

138 See response by the CAA to the Questionnaire to Member States' Civil Aviation Authorities, Question 5, folio no. 1435.

139 See replies (in particular Luxair, folio no. 23152) to the Questionnaire to Competitors sent on 6 November 2006, question 22.

140 Decision of the UK Office of Fair Trading in the case Flybe/BA Connect published 15 February 2007 http://www.oft.gov.uk/advice_and_resources/resource_base/Mergers_home/decisions/2007/Flybe.

141 See paragraphs 12, 27 and 52 of the OFT decision .

142 See paragraph 27 of the OFT decision.

143 UK airports: Report on the market study and proposed decision to make a market investigation reference OFT 882 (December 2006), see in particular paragraph 5.87.

144 See Board Paper 3 for the Board Meeting of Ryanair Holdings plc [...]*, page 3.4 (folio no. 629).

145 Source: viamichelin.com and rac.com.

146 Source: market investigation: responses to airport substitutability questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

147 Source: market investigation: responses to airport substitutability questionnaire.

148 Source: ryanair.com.

149 See response by the CAA to the Questionnaire to Member States' Civil Aviation Authorities, folio no. 1435, Question 5.

150 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

151 See Board Paper 3 for the Board Meeting of Ryanair Holdings plc on [...]*, page 3.4 (folio no. 629).

152 Source: viamichelin.com and rac.com.

153 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

154 Source: market investigation: responses to Airport Substitutability Questionnaire.

155 Source: ryanair.com.

156 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

157 Source: viamichelin.com and rac.com.

158 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU).].

159 Source: market investigation: responses to Airport Substitutability Questionnaire.

160 Source: www.ryanair.com.

161 Source: http://www.firstgroup.com/scotrail/index.php

162 See in particular page 211 of the response, Ryanair's comments on paragraph 78 of the Statement of objections.

163 Source: www.ryanair.com

164 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

165 In its response to the Statement of Objections Ryanair argues that a number of airlines have considered that the two airports are not substitutable for time-sensitive passengers. The Commission notes however that although it does not consider it appropriate to distinguish between time/non-time sensitive passengers for the purpose of this case, there is no evidence in the file suggesting that the two airports should not be substitutable for non-time sensitive/leisure passengers. On the contrary, a clear majority of competitors indicated that these airports are substitutable for non-time sensitive passengers, see replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

166 A positive price correlation over time serves as a meaningful confirmation that two or more products or services belong to the same relevant market. Conversely, a negligible or negative correlation of prices over time does not necessarily imply the products/services belong to separate markets. See Annex III for further details.

167 Source: viamichelin.com and rac.com.

168 Source: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

169 Source: market investigation: responses to Airport Substitutability Questionnaire.

170 Source: ryanair.com.

171 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22, in particular CityJet (folio 5351).

172 Source: viamichelin.com and rac.com.

173 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

174 Source: market investigation: responses to Airport Substitutability Questionnaire.

175 Source: ryanair.com.

176 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

177 Source: viamichelin.com and rac.com.

178 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

179 Source: market investigation: responses to Airport Substitutability Questionnaire.

180 Source: www.ryanair.com.

181 Source: www.toulouse.aeroport.fr

182 The difference made by Toulouse Blagnac between competition between airports, on the one hand, and substitutability from the passenger's point of view on the other illustrates the distinction to be made between demand and supply side substitutability (See Section 6.4 below).

183 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

184 No customer survey was conducted for this route.

185 Source: viamichelin.com and rac.com.

186 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

187 Source: market investigation: responses to Airport Substitutability Questionnaire.

188 Source: www.ryanair.com.

189 Source: www.rennes.aeroport.fr

190 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

191 See in this respect the difference compared to other airports such as example of Frankfurt/Hahn and Frankfurt am Main and others.

192 Source: viamichelin.com and rac.com.

193 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

194 Source: market investigation: responses to Airport Substitutability Questionnaire.

195 Source: www.ryanair.com.

196 Questionnaire to Member States' Civil Aviation Authorities, Question 5.

197 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

198 See above the part describing the substitution of London airports and Annex III.

199 In particular, in so far as 38.4% of Ryanair passengers to Charleroi have considered flying Aer Lingus to Brussels National Zaventem whilst 60.6% of Aer Lingus passengers have considered flying Ryanair to Brussels South Charleroi, the two airports may be considered substitutable for a sufficiently large number of passengers.

200 See e.g. Board Paper No 2 for the Board Meeting of Ryanair Holdings plc to be held on [...]*, Section 3.1 and others (folio no. 629).

201 Commission Decision (2004-393-EC) of 12 February 2004 concerning advantages granted by the Walloon Region and Brussels South Charleroi Airport to the airline Ryanair in connection with its establishment at Charleroi, OJ L 137, 30/04/2004, p. 1 - 62, ("the Charleroi Decision").

202 See paragraphs 299 and 300 of the Charleroi Decision.

203 See paragraphs 248 and 249 of the Charleroi Decision.

204 Source: viamichelin.com and rac.com.

205 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

206 Source: market investigation: responses to Airport Substitutability Questionnaire.

207 Source: ryanair.com.

208 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

209 See Commission decision in the case M.967 - KLM/Air UK of 22/09/1997, paragraph 24.

210 Source: viamichelin.com and rac.com.

211 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

212 Source: market investigation: responses to Airport Substitutability Questionnaire.

213 Source: ryanair.com.

214 Hanseatic Higher Regional Court (Hanseatisches OLG Hamburg), judgement of 30.09.2004 (case no. 5 U 176/03) and preliminary injunction of 19.12.2002 (case no. 5 U 137/02).

215 See Board Paper 3 for the Board Meeting of Ryanair Holdings plc on [...]* (folio no. 629), page 3.5. as well as [...]*. Similarly see also Board Paper 3 for the Board Meeting of Ryanair Holdings plc on [...]* (folio no. 629), page 3.4. ("[...]*"). Further also Board Paper 3 for the Board Meeting of Ryanair Holdings plc on [...]* (folio no. 629), page 3.3.; Board Paper 3 for the Board Meeting of Ryanair Holdings plc on [...]* (folio no. 629), page 3.5. and others. [...]*.

216 The Commission's finding is without prejudice to the fact that the two airports are not substitutable from the carriers' point of view (the supply-side substitutability). Further, this finding is without prejudice to any finding by the Commission in relation to Council Regulation No 2408-92.

217 According to the Customer Survey, Frankfurt/Hahn airport does not seem acceptable for more than 90% of Lufthansa's passengers, many of which are connecting passengers transiting in Frankfurt/Main to another (usually long-haul) flight.

218 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22..

219 Ryanair entered the Dublin - Frankfurt route only in May 2005.

220 See above the part describing the substitution of London airports. In the particular case of the Aer Lingus service between Dublin and Frankfurt am Main, substantial variations in the number of connecting passengers can be observed. A different seasonality pattern for these inter-connecting passengers or a different pattern of random demand shocks would lead to a low correlation coefficient and would not reflect the underlying structural relationship between the two price series. See Annex III for further details.

221 Source: viamichelin.com and rac.com.

222 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

223 Source: market investigation: responses to Airport Substitutability Questionnaire.

224 Source: ryanair.com.

225 Source: www.ham.airport.de and www.jasper.de

226 Source: www.bahn.de and www.flughafen-luebeck.de .

227 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

228 See notably the comments of the responsible German Ministry for Transport, stressing the willingness of low-cost customers to accept significantly longer travel distances (response to Airport Substitutability Questionnaire of 13/01/2007, folio no. 1312, e.g. answers to questions 2, 5, 8).

229 Source: viamichelin.com and rac.com.

230 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

231 Source: market investigation: responses to Airport Substitutability Questionnaire.

232 Source: www.ryanair.com.

233 Source: www.viennaairport.com

234 Source: www.eurolines.at

235 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

236 See the published version of the decision in case BWB/Z-22 - Flughafen Wien Aktiengesellschaft; PENTA INVESTMENTS LIMITED; Airport Bratislava, available at the webpages of the Austrian authorities: http://www.bwb.gv.at/BWB/Veroeffentlichungen/Zusammenschluesse/Zusammenschluesse_2006/z_0022.ht m.

237 See the press release of the Slovak national competition authority (the decision is currently under appeal and is thus not yet published) http://www.antimon.gov.sk/eng/article.aspx?c=387&a=2362.

238 See e.g. the press release of the Slovak authorities, which states among others as follows: "Regarding the fact that the airports VIE and BTS serve nearly same catchment area, which does not overlap with catchment areas of other airports, which would represent current or potential competitors of the airports VIE and BTS, the Office on a basis of complex competition assessment of many factors has found out that the airports VIE and BTS are mutually substitutable and competition for airlines and passengers proceeds between them in given region in the relevant market of providing infrastructure to regular regional flights.".

239 In particular, in so far as 38.4% of Ryanair passengers to Charleroi have considered flying Aer Lingus to Brussels National Zaventem whilst 60.6% of Aer Lingus passengers have considered flying Ryanair to Brussels South Charleroi, the two airports may be considered substitutable for a sufficiently large number of passengers.

240 Source: viamichelin.com and rac.com.

241 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

242 Source: market investigation: responses to Airport Substitutability Questionnaire.

243 Source: www.ryanair.com.

244 Source: www.barcelona-airport.com

245 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

246 Source: viamichelin.com and rac.com.

247 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

248 Source: market investigation: responses to Airport Substitutability Questionnaire.

249 Source: www.ryanair.com.

250 Source: www.spanish-airport-guide.com

251 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

252 Data submitted by Ryanair.

253 Source: viamichelin.com and rac.com.

254 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

255 Source: market investigation: responses to Airport Substitutability Questionnaire.

256 Source: ryanair.com.

257 The Commission did not have sufficient data for a price correlation for this route. Further, no customer survey was conducted for this route.

258 Source: viamichelin.com and rac.com

259 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)]

260 Source: information not available from market test questionnaires or the Internet.

261 Source: ryanair.com

262 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

263 In particular, in so far as 38.4% of Ryanair passengers to Charleroi have considered flying Aer Lingus to Brussels National Zaventem whilst 60.6% of Aer Lingus passengers have considered flying Ryanair to Brussels South Charleroi, the two airports may be considered substitutable for a sufficiently large number of passengers.

264 Source: viamichelin.com and rac.com.

265 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

266 Source: market investigation: responses to Airport Substitutability Questionnaire.

267 Source: www.ryanair.com.

268 Source: www.adr.it

269 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

270 In particular, in so far as 38.4% of Ryanair passengers to Charleroi have considered flying Aer Lingus to Brussels National Zaventem whilst 60.6% of Aer Lingus passengers have considered flying Ryanair to Brussels South Charleroi, the two airports may be considered substitutable for a sufficiently large number of passengers.

271 Source: viamichelin.com and rac.com.

272 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

273 Source: market investigation: responses to Airport Substitutability Questionnaire.

274 Source: www.ryanair.com.

275 Source: www.comunedimilano.it

276 Source: www.comunedimilano.it

277 Source: http://coachtv.ibooking.com via www.ryanair.com

278 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

279 Source: viamichelin.com and rac.com.

280 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)].

281 Source: market investigation: responses to Airport Substitutability Questionnaire.

282 Source: www.ryanair.com.

283 Source: www.alilaguna.it

284 Source: www.atvo.it

285 Source: http://www.atvo.it/orari/pdf_airport/Ryanair.pdf

286 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

287 Source: viamichelin.com and rac.com

288 Source: market investigation: responses to Airport Substitutability Questionnaire [and rac.com (UK) and viamichelin.com (rest of EU)]

289 Source: market investigation: responses to Airport Substitutability Questionnaire

290 Source: ryanair.com

291 Source: www.forliairport.com

292 See replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

293 See Section 6.1.1 above.

294 See also below Section 7.3.4.

295 See Section 6.1.

296 Airport Substitutability Questionnaires of 9 November 2006, 3 January 2007 and 30 January 2007.

297 See replies to question 22 of the Questionnaire to Competitors of 06.11.2006.

298 For example, due to its business model or existing hub/base, a particular carrier may be only willing to enter a specific route if it obtains slots to operate from a primary airport.

299 See e.g. cases M.3940 - Lufthansa/Eurowings, paragraph 11; M.3770 -Lufthansa/Swiss, paragraph 16; M.3280 - Air France/KLM, paragraph 20; British Airways/DanAir, paragraph 10;M.2672 - SAS/Spanair.

300 In exceptional circumstances, customers might consider indirect flights on longer continental routes ("mid-haul" routes, see case M.3940 - Lufthansa/Eurowings, paragraph 11). This might be the case when the direct flight does not allow for a typical one day return trip. Given that the mid-haul routes in the present case (in particular Dublin - Tenerife which is the only overlap routes with travelling time significantly above 3 hours and thus comparable to the flights between Spain and Scandinavia identified as mid-haul in the case M.2672 - SAS/Spanair, paragraph 14) are primarily non-business destinations, the need for a "same-day" return trip does not play a significant role in the customers' decision-making process here (see also see also Lufthansa/Swiss, paragraph 17; M.3280 - Air France/KLM, paragraph 20).

301 Notification, paragraph 236.

302 See e.g. cases M.3940 - Lufthansa/Eurowings, paragraphs 11, 51; M.3770 -Lufthansa/Swiss, paragraphs 12, 56; M.3280 - Air France/KLM, paragraphs 9 and 71; see also ECJ, case 66/86 Ahmed Saeed Flugreisen, of 11 April 1989 [1989] ECR 803, at paragraph 40.

303 See e.g. cases M.3940 - Lufthansa/Eurowings, paragraphs 11, 51; M.3770 - Lufthansa/Swiss, paragraphs 12, 56; M.3280 - Air France/KLM, paragraphs 9 and 71.

304 See www.bahn.de.

305 Although Ryanair mentions in the Notification that it may be "possible" to include ferries and buses "in certain contexts" in an O&D pair (see paragraph 173and 230/231 of the Notification), the Commission has not found any evidence in support of such an assumption. It should also be noted that Ryanair submits in the Notification that indirect flights on short-haul routes should be defined as separate markets because of the longer flight time of an indirect flight. The same reasons (significant time difference), however, apply to comparison of flights and bus/ferry or train/ferry trips.

306 See e.g. interview with CityJet (Air France) of 21.2.2007, folio no. 6170, in which CityJet particularly refers to the specific "island" situation of Dublin, the "poor ferry connections" and the absence of a train link.

307 The "seats" can also be sold to end customers by tour operators which bought them from the charter airline, see e.g. ThomasCook for FlyMonarch.

308 In previous decisions the Commission has also used the term "seat-only" for sales of charter airlines of seats without a holiday package to end customers, see e.g. case M.1494 - Sair Group/AOM, paragraph 11.

309 It should be noted that also scheduled airlines sell, to a limited extent, seat packages to tour operators. However, this is not the case for Ryanair. The merger is therefore not likely to have an impact on the market for sales of seats to tour operators.

310 See case M.1524 - Airtours/First Choice, paragraphs 34-50 (insofar not affected by the CFI's annulment decision); M.157 - AirFrance/Sabena, paragraph 25; see also cases M.1494 - Sair Group/AOM, paragraph 11 and M.3770 - Lufthansa/Swiss, paragraph 20.

311 See e.g. paragraph 171 of the Notification, where Ryanair refers to the example of Irish residents owning a holiday home (e.g. in Spain), who could turn to charter airlines to buy "dry seats".

312 As concerns dry seat sales, the distinction between scheduled and charter flights is to a certain extent misleading. Indeed, although a flight might be regarded as a "charter flight" since it serves mainly package holiday customers, from the point of view of a dry seat customer, the same flight will be regarded as a "scheduled" flight which he booked individually for a specific date.

313 See e.g. Ryanair's Response to the Commission's 6(1)(c)-decision, pages 34 and 35; Ryanair's submission "Additional Information on Charter Airlines" of 31.1.2007 and 7.2.2007; RBB Position Paper on Barriers to Entry, 20.2.2007, folio no. 4135, pages 30-32.

314 See e.g. Response to the Statement of Objections, page 196/197.

315 See Ryanair's submission of 31.1.2007.

316 In particular, the Commission notes that none of the 36 references to Ryanair Board minutes and papers or filings with the US SEC contain a reference to an O&D pair that is affected by the present transaction. None of the 7 US SEC filings pointed to the Commission's attention by Ryanair identifies one or more charter carriers as competitors.

317 It should be noted, however, that the fact that the "market" share of one product drops if it is put together with another seasonal product (such as charter flights) during the time in which the seasonal product is sold (see e.g. RBB paper of 20.2.2007, folio no. 4135, page 31) does not necessarily show that both products are in one market. The dropping market share may just be a result of putting them in one market. It might also well be possible that charter flights just capture a seasonal demand for package holidays (including the related services) without any direct impact on scheduled flights. Customers might just prefer to buy the product "package holiday" in summer, without considering a simple flight.

318 FirstChoice, for instance, estimates that only 3% of all its seats on flights to/from Ireland are sold as "dry seats" to end customers, see interview with FirstChoice of 8.3.2007, folio no. 6170.

319 The limited competitive impact of these charter airlines on the parties on the routes where the share of charter traffic (mainly "non-dry seats") is higher will be discussed further below within the framework of the competitive analysis of these individual routes.

320 See questionnaire of 21, 23 and 28 November 2006 sent to charter airlines, answers to questions 13(b)/7, as well as additional questions to charter airlines during the Phase II investigation It should also be noted that a majority of airlines have answered that charter airlines would not exert any competitive pressure on Ryanair/Aer Lingus (See questionnaire of 6 November 2006 sent to scheduled airlines, answers to question 53).

321 See minutes from the meeting with Ryanair in Dublin of [...]* (folio no. 6170); the internal documents provided by Ryanair contain presentations on Ryanair's "main competitors" [...]*. See the documents named "Board Paper 3 for the Board Meeting of Reynar Holdings plc" for the meetings held on [...]*, [...]*, [...]*, [...]*, [...]* etc. (folio no. 629).

322 See e.g. questionnaire of 21, 23 and 28 November 2006 sent to charter airlines, answers to questions 15-18 as well as additional questions to charter airlines during the Phase II investigation; A number of respondents to the Commission's market investigation indicated that former package holiday customers at least partly switched from booking through a charter/tour operator to booking their holiday packages on a "do-ityourself"- basis, combined of an independently booked accommodation and a scheduled flight with Ryanair or Aer Lingus.

323 CAA study "Low-Frills Carriers: Revolution or Evolution?", 15 November 2006, CAP 770, 2.33 ff.

324 See e.g. cases M.3940 - Lufthansa/Eurowings, paragraph 12; M.3770 - Lufthansa/Swiss, paragraph 15; M.3280 - Air France/KLM, paragraph 19; the question was left open in case COMP/JV 19 - KLM/Alitalia, paragraph 21.

325 See COMP/JV 19 - KLM/Alitalia, paragraph 21.

326 On its European network, Aer Lingus issues exclusively one way restricted tickets, with the exception of tickets issued in accordance with corporate contracts. As explained above however, such contracts represent a negligible proportion of Aer Lingus' turnover.

327 The "Saturday Night Rule" is a typical restriction imposed on roundtrip tickets: these tickets are valid only if there is one Saturday night between the outbound and the inbound legs on the ticket. This rule is used to discriminate passengers travelling on business from leisure passengers. The latter are supposedly (i) more willing than business passengers to spend the Saturday Night away from home and (ii) more price sensitive. Therefore, roundtrip tickets subject to the Saturday night rule are normally cheaper.

328 See for an example that summarizes the Commission's findings e.g. the interview with Aer Arann, folio no. 6170: "...the catchment areas of airlines can vary significantly. In particular, the catchment area of Low Cost Carriers can be much larger than those of network carriers as customers are prepared to travel further to an airport to fly on their low cost flight." In the view of competitors, non time-sensitive passengers (which can be deemed to represent the majority of Aer Lingus and Ryanair's passengers, see Annex I) are much more inclined to substitute primary with secondary airports than time-sensitive customers, see replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

329 See e.g. cases M.3280 - Air France/KLM, paragraph 19; M.3770 - Lufthansa/Swiss, paragraph 15. This trend was also confirmed by the market investigation in the current case - see replies to Questionnaire to Competitors of 6 November 2006, questions 20 and 21.

330 It is noteworthy that the possibility to price discriminate between different customer groups was the main reason in previous decisions to define separate markets for time-sensitive and non-time-sensitive passengers (See in particular case M.3770 - Lufthansa/Swiss, paragraph 15).

331 It should be noted that Aer Lingus serves, but only to a very minor extent, also so-called "corporate customers", i.e. customers who can book flexible tickets under a framework contract with major corporate customers. These sales, however, apply only to customers under a corporate contract and account only for less than 5% of its total turnover (see Aer Lingus' e-mail of 27 March 2007, folio no. 6350) and are therefore no proper means to capture all time-sensitive passengers of Aer Lingus.

332 The Commission has, however, taken into account the differences between both customer groups where these differences were considered to be relevant for the competitive analysis.

333 The rationale for this "time to departure" approach is that low frills carriers like Ryanair or Aer Lingus tend to raise ticket prices for any given flight as the plane fills up with passengers. Late bookers therefore tend to pay a higher price, which could have constituted a meaningful form of price discrimination for the present purposes.

334 See minutes from the meeting with Ryanair in Dublin of 6 February 2007, folio no. 6170.

335 For more details on yield management systems of Ryanair and Aer Lingus, see the Section 7.4.1.

336 See for example the interview with Aer Arann, folio no. 6170, which summarizes the Commission's findings in this regard: "...the catchment areas of airlines can vary significantly. In particular, the catchment area of Low Cost Carriers can be much larger than those of network carriers as customers are prepared to travel further to an airport to fly on their low cost flight." In the view of competitors, "non time-sensitive" passengers (which can be deemed to represent the majority of Aer Lingus and Ryanair's passengers, see e.g. Annex I) are much more inclined to substitute primary with secondary airports than "time-sensitive" customers See e.g. replies to the Questionnaire to Competitors sent on 6 November 2006, question 22.

337 See Annex I, answers to question 5 of the Customer Survey.

338 See Section 6.5.

339 See Section 6.3.

340 For example, the merger of a network carrier operating according to a hub and spoke model with a charter/tour operator (see cases M.2218 - British Airways/Thomas Cook; M.1354 - Sair (Swissair)/LTU; M.1128 - KLM/Martinair; M.2093 - Airtours/FTI; M.1524 - Airtours/First Choice), the merger of a network carrier and a smaller regional airline (see Cases M.157 - Air France/Sabena; M.278 - British Airways /DanAir; M.857 - British Airways /Air Liberté; M.806 - BA/TAT; M.1855 - Singapore /Virgin; M.1494 - Sair (Swissair)/AOM; M.3940 - Lufthansa Eurowings) or the merger of two network carriers (a smaller and a larger (see cases IV/JV.19 - KLM /Alitalia; COMP M.3770 - Lufthansa/Swiss; M.562 - Swissair/Sabena).

341 See above Section 6.1.

342 See Association of European Airlines, Yearbook 2006, page 10.

343 See cases M.2218 - British Airways/Thomas Cook; M.1354 - Sair (Swissair)/LTU; M.1128 - KLM/Martinair;M.2093 - Airtours/FTI; M.1524 - Airtours/First Choice.

344 Cases M.157 - Air France/Sabena; M.278 - British Airways /DanAir, M.857 - British Airways /Air Liberté; M.806 - BA/TAT; M.1855 - Singapore /Virgin; M.1494 - Sair (Swissair)/AOM; M.3940 - Lufthansa Eurowings.

345 See cases IV/JV.19 - KLM /Alitalia; COMP M.3770 - Lufthansa/Swiss; M.562 - Swissair/Sabena.

346 See Section 7.8.

347 Both airlines constantly seek to maximise the use of their fleet and crew as these are fixed costs. As a result, they will constantly adjust the number of routes they operate (and the frequencies on these routes) from a given hub or base. This is exemplified by the fact that the actual number of the overlap routes has changed significant in the past and would even have been different, had the merger been notified a few months earlier or later (and actually changed during the course of the Commission's investigation Since the Article 6(1)(c) decision, the overlap routes no longer include anymore the Dublin - Bristol route as Aer Lingus ceased its operations on this route at the end of March 2007. See in detail Section 7.6 on potential competition.

348 It should be noted that Ryanair's and Aer Lingus' point-to-point operations are regularly adjusted and modified.

349 The market shares in the table are, except for Dublin-Paris and Dublin-Frankfurt, calculated on the basis of the capacity operated on the route (seats/week). The last passenger data for the parties and their competitors are available to the Commission only for the Summer 2006 season. Since then, there have been a number of changes in the structure of the markets (in particular competitors exiting or one of the mergin parties entering the route). The actual capacity data therefore give the most up-to-date "picture" of the current structure of the individual markets. Further, these market shares to a large extent correspond to the market shares calculated on the basis of the actual point-to-point passenger numbers (for the routes where these numbers are already available) with the exception of in particular two routes to major hub airports - Frankfurt and Paris - where existing network carriers have a significant proportion of connecting passengers and where thus the combined market share of the Merging Parties on the basis of point-to-point passenger traffic is significantly higher than on the basis of capacity data. For this reason, the table provides for these two routes market shares based on actual point-to-point passengers. For all other routes, the differences between market shares on the basis of capacity and actual point-to-point passenger numbers are only minor, if any (see for more details Section 7.9).

350 For confidentiality reasons, the market shares as well as some other figures from third parties are provided in this Decision as ranges in square brackets.

351 The question whether dry seat sales of charter airlines or tour operators should be included in the affected markets was left open (see Section 6.7) and has, as the table shows, no significant effect on the assessment of the transaction. See more in detail in the detailed analysis of the respective routes in Section 7.9.

352 As indicated above in footnote 349, these figures take account of the significant number of connecting passengers of the network carrier Lufthansa which are disregarded (see Section 6.3.1). They are based on actual point-to-point passengers and not overall capacity.

353 As indicated above in footnote 349, these figures take account of the significant number of connecting passengers of the carrier CityJet which are disregarded (see Section 6.3.1). They are based on actual point to-point passengers and not overall capacity

354 Unlike the Article 6(1)(c) decision, the overlap routes do not include anymore the routes Dublin-Bristol, Dublin-Fuerteventura and Dublin-Turin, as Aer Lingus ceased its operations on this route as of end of March 2007 (Bristol) and June (Fuerteventura, Turin). Further, the list of overlap routes does not include Dublin - Amsterdam/Eindhoven and Dublin - Nantes/Rennes routes where the in-depth investigation showed that the relevant destination airports cannot be regarded as substitutable (see above in the Section 6.3.). On the other hand, the Merging Parties have already announced to open new routes after the date of the sending of the Statement of Objections, some of which might create new overlap routes (e.g. Dublin - Copenhagen, to be opened by Aer Lingus as of Winter 2007/2008, very likely in competition with Ryanair's existing operation marketed as Malmö/Copenhagen, or Shannon - Dublin, to be opened by Ryanair as of Winter 2007/2008 in competition with the existing operations of Aer Lingus).

355 Even if dry seat sales from charter airlines were included, the market share would still come close to a monopoly due to the very limited number of places the offer of the parties.

356 These are: Dublin-Alicante; Dublin - Berlin, Dublin - Birmingham, Dublin - Edinburgh, Dublin - Faro, Dublin - Krakow, Dublin - Malaga, Dublin - Manchester, Dublin - Marseille, Dublin - Poznan, Dublin - Riga, Dublin - Salzburg, Dublin - Madrid, Dublin - Newcastle, Dublin - Seville, Dublin - Warsaw,.

357 It should be noted that in case of three routes (Dublin - Manchester, Dublin - Birmingham, Dublin - Alicante and Dublin - Newcastle) Ryanair flies both to the same airport as Aer Lingus (Manchester, Birmingham and Newcastle respectively) and to a secondary/adjacent airport(s) (Liverpool and Leeds, East Midlands and Durham respectively). Therefore, the airport substitutability was analysed in the market definition section also with respect to these routes, in addition to the 16 only city-to-city overlapping routes.

358 Even if dry seat sales from charter airlines were included, the market share would still come close to a monopoly due to the very limited number of places the offer of the parties.

359 The remaining "non-overlap routes" of the Merging Parties therefore represent only about 10% of intra- European customers at Dublin. The calculation is based on passenger data provided by DAA.

360 The total passenger number was calculated by the Commission on the basis of the (confidential) passenger numbers submitted by the Dublin Airport Authority ("DAA") on 1 February 2007 for individual routes.

361 See for a detailed analysis of the passenger number and the frequencies offered on each route the route-byroute analysis in Section 7.9.

362 Examples of such seasonal routes are: Aer Lingus operates only in summer Dublin - Bilbao, Dublin - Bologna, Dublin - Marseille and Dublin - Toulouse and only in winter Dublin - Salzburg. The majority of routes are however permanent.

363 Connecting or transit passengers accounted, according to Aer Lingus, only for [5-10%]* on the 34 overlap routes from/to Dublin as identified by the Commission in the Statement of Objections.

364 See Horizontal Merger Guidelines, OJ C 31, 5.2.2004, pages 5-6, paragraph 14.

365 See Horizontal Merger Guidelines, OJ C 31, 5.2.2004, page 5, paragraph 17.

366 See Case T-210-01 General Electric v Commission [2005] ECR II-5575, paragraph 115.

367 Case T-177-04 easyJet v Commission [2006] ECR II-1913, paragraph 174.

368 See Horizontal Merger Guidelines, paragraphs 28-30.

369 In particular in markets where the two merging firms are by far the largest competitors, there is a strong presumption that they are each others' closest competitors.

370 See Horizontal Merger Guidelines, paragraph 28: "(...) The merging firms' incentive to raise prices is more likely to be constrained when rival firms produce close substitutes"

371 See in more detail below, Section 7.9.

372 See Section 7.3.4.

373 See above, Section 7.2.1, Table 2.

374 At least on two of these three routes (Dublin-Paris and Dublin-London), the large proportion of connecting passengers of cityJet and Lufthansa should be deducted in order to compare the market shares, see above Section 7.2.1.

375 See more in detail below, Section 7.3.4

376 See Section 6.1.

377 See replies to question 13, 14 and 15 of the Questionnaire to Customers of 31.10.2006 and replies to questions 28,29 and 30 of the Questionnaire to Competitors of 06.11.2006.

378 See in this respect e.g. Ryanair's comment in the Response to the Statement of Objections, Page 70; see replies to question of the questionnaire to customers of 31.10.2006 (questions 8, 9, 16, 17 and 18). Although a number of corporate customers indeed emphasises the differences between both airlines for business customers, these differences are less relevant in the present case, which mainly affects leisure customers; see also some replies of the questionnaire to competitors of 06.11.2006 (questions 31, 32, 45). It should however be noted that even business customers do not at all disregard Ryanair but fly to a significant extent with Ryanair, see the results of the Commission's Customer Survey in Annexes I and II, the questionnaire to customers of 31.10.2006 (questions 15, 17 and 18)., the CAA-study "No-Frills Carriers: Revolution or Evolution" (2006), chapter 4 page 3 et seq.

379 The changed model of Aer Lingus does not relate to its long-haul operations. However, as the overlaps between the parties are limited to intra-European flights and these flights are no affected by the long-haul operations as Aer Lingus does not feed these flights through its short-haul network as traditional hub-andspoke carriers, the long-haul operations of Aer Lingus can be disregarded for the purpose of the competitive assessment of the proposed concentration.

380 See for example Aer Lingus' slogan: "Low Fares -Way Better".

381 It should however be noted that Aer Lingus' customers usually have to pay less for the transfer to the airport and can reach the primary airports served by Aer Lingus easier.

382 On many routes, the parties face network carriers such as Lufthansa on the Dublin- Frankfurt route, Iberia on flights to Madrid, Air France/CityJet to Paris CDG, Austrian to Vienna, British Airways and bmi to London. These carriers feed traffic into their respective hub airport. As described above in Section 6.1, Ryanair's and Aer Lingus' "point-to-point" operating model differs from "feeder-oriented" services of network carriers. While network carriers mainly cater for transfer passengers and primarily feed their hub airport. Ryanair does not compete for transfer passengers at all, and also Aer Lingus, while carrying to a certain extent connecting passengers for third airlines, clearly focuses its strategy on point-to-point (and not feeder) services. Network carriers also provide a different service model (price/quality) mix than Aer Lingus and Ryanair, and therefore they are less close competitors. They operate into main airports (their "hubs") and mostly provide full-service. Their presence on a particular route is mainly driven by the number of transfer passengers they can feed into their network. Also many of the remaining competitors on the other routes cannot be regarded as closer to Ryanair than Aer Lingus in terms of their business model (e.g. Spanair operates on Dublin-Malaga only in the summer season around 2 weekly rotations limited to weekends and is thus rather closer to charter carrier operations than to the Merging Parties daily services throughout the year). The remaining "low frills" competitors Ryanair and Aer Lingus face (e.g. Sky Europe (on two routes) bmibaby, Clickair and Norwegian (on one route respectively), have no comparable base at Dublin.. Aer Arann's service model may also be characterised as a "low-frills" model, but Aer Arann operates with 70-seat aircraft and does thus not have a cost structure that would allow it to be considered as a closes competitor to Ryanair (see below, Section 7.3.3).

383 Referred to by the market also as "low cost carriers" ("LCCs", see above, paragraph. See as to the terminology Section 6.1.

384 See as to Flybe's characterisation as a "regional" low-frills carrier, flying with smaller aircraft to regional airports the minutes of the interview with FlyBe of 12.3.2007, folio no. 6170.

385 This does, however, not mean that Ryanair's prices are "always" lower than Aer Lingus' prices as claimed by Ryanair in its Response to the Statement of Objections (page 230). According to the booking status, it is well possible that a customer can buy a flight at a cheaper price from Aer Lingus than from Ryanair. The Commission's case team has observed examples of such instances at the occasion of the visit of Ryanair's and Aer Lingus' headquarters in Dublin (see in this respect also the minutes of the meeting with Ryanair of 6.2.2007 and with Aer Lingus of 5.2.2007, folio no.6170).

386 European average short-haul prices 2006 as indicated by Ryanair: € 41 (it is, however, unclear which taxes and charges such as credit card fees, airport charges, baggage fees, etc. are included); Aer Lingus: € 91 (€ [65-75]* before charges and taxes).

387 See e.g. http://business.guardian.co.uk/story/0,,1769707,00.html

388 See e.g. http://www.ryanair.com/site/about/invest/docs/present/quarter1_05.pdf, page 5.

389 At the same time, the airport charges included in Aer Lingus' prices are usually higher than the charges of secondary airports Ryanair flies to.

390 See Annex III.

391 See Ryanair's comment in its Response to the Statement of Objections (page 187). When stressing the differences between both airlines, Ryanair particularly refers to Aer Lingus' "...lack of robust interpretation of customer service rules and regulations".

392 See e.g. the descriptions of websites describing Aer Lingus: http://www.eurodestination.com/european-lowcost- airlines-guide.html: "Irish National Carrier that has now gone low cost"; http://www.cheapflights.co.uk/low-cost-about.html: "Aer Lingus has repositioned itself as a low-cost carrier in recent years." http://en.wikipedia.org/wiki/Aer_Lingus: "They are currently positioning themselves as competition to the European no-frills airlines". See also Aer Lingus' study on the Aer Lingus brand (see email from Aer Lingus (Simone Warwick) of 27 March 2007, folio no. 6316).

393 See e.g. Response to the Statement of Objections, page 223.

394 Stage-length adjusted operating cost per available seat kilometre. Sources for all carriers except Aer Lingus: IATA Economics Briefings n.5 ; http://www.iata.org/NR/rdonlyres/D1AFC924-8F16-4E9D-AEC1- FD18C1FE1336/51284/airline_cost_perfomance.pdf ; for Aer Lingus: "Circular in relation to Ryanair offer", page 8, www.flyaerlingus.com. Estimation of fuel costs for Aer Lingus based on IATA Economics Briefings n.5 (see above).

395 cf. RBB Economics: Ryanair/ Aer Lingus: Merger Efficiencies, folio no. 19873, p.6

396 Now part of Brussels Airlines

397 See e.g. Response to the Statement of Objections, page 223.

398 cf. Aer Lingus preliminary announcement of results for the year ended 31 December 2006, available on http://www.aerlingus.com/cgi-bin/obel01im1/Corporate/ir_finan_info_reports.jsp

399 cf. Response to the Statement of Objections, par.33. It appears from the Response that Ryanair derives its estimate from information about Aer Lingus average fares provided in its IPO prospectus.

400 €ct/ASK; source: Aer Lingus (http://www.aerlingus.com/cgi-bin/obel01im1/Corporate/ir_overview_about_ei.jsp) .

401 See Section 6.1.

402 The term "base" is used in this decision to characterise the advantages of concentrating significant operations at one single airport. As further explained in this section, airlines use different concepts of the term "base" (see Base Questionnaire, e.g. questions 3 and 11). It is therefore not possible to define an "exact" set of qualitative and quantitative criteria for a "base" airport. However, it can be observed that the term "base" is usually only used when more then one route is served by an airport. As set out above in Section 6.1., the concept of a "base" has to be distinguished from the concept of a "hub", since the term "hub" relates to the "huband- spoke" system and its system of connecting "feeding" traffic into a network which is not the case for the concept of a "base" (although a base can of course also be a hub).

403 See in particular answers 7, 8, 9 and 10 to the Questionnaire to Competitors (base competition) sent on 7 December 2007 (hereinafter "Base Questionnaire").

404 By way of example, Ryanair handles its luggage in dedicated luggage handling facilities at Dublin Airport.

405. In its response to the Statement of Objections (p. 54), Ryanair concludes therefore that a base would be irrelevant for such stand-by services. However, it should be noted that Ryanair's approach confirms that the cost of such stand-by-aircraft can be minimised with very large fleets, i.e.; there exist economies of scale when handling disruption cost. Second, the fact that Ryanair can spread such cost over a number of bases results from its particular business model (i.e. to operate with a base network). Third, given that Ryanair mainly connects its various bases, this in effect implies that - when not actually in used - in effect these [2- 6]* aircraft are most likely on stand-by at Ryanair's base airports, one of them being Dublin.

406 Although it is true that some airports offer incentive schemes to new entrants, carriers holding already a slot at an airport have an advantage since they can easily enter on a new route using a slot previously used for another route ("grandfather rights").

407 See answers to question 8 and 9 of the Base Questionnaire.

408 While these advantages may at least until today not play an important role at Dublin Airport, Ryanair has been able to negotiate significant rebates and other advantages at other airports, see also Section 7.8.8.

409 See in more detail e.g. submission by Aer Lingus of 9 February 2007, folio no. 4122, pp.2-3.

410 This fact was stressed by competitors in the interviews conducted by the Commission, see interview with easyJet of 15.2.2007 and with Jet2.com of 1.2.2007, folio no. 6170.

411 The incremental scale efficiencies become, however, less important with very large bases (e.g. the additional savings from operating with 70 instead of 60 aircraft on a base are relatively lower).

412 Ryanair response to the Article 6(1)(c) decision of 19 January 2007, p.13

413 Fixing technical problems is more complex and expensive in the case of an aircraft break-down on routes which are not connected to a base where technical support is located.

414 See Ryanair's Response to Preliminary comments/questions on the efficiencies paper submitted on 20 October 2006, folio no. 24006, section Staff Costs.

415 See e.g. the results of the so-called Base Questionnaire sent to competitors. In its submission of RBB of 20.2.2007, folio no. 4135 (pages 45-51), RBB has, based on Ryanair data, considered a one off expansion from 10 to 15 aircraft at Dublin Airport and finds that this does not give rise to a reduction in Ryanair's operating costs. It also compares the cost of flying in the Summer 2006 season between Gatwick (not a Ryanair base) and Cork (where one Ryanair aircraft is based) with the cost of flying between Gatwick and Dublin (where 15 aircraft are based). RBB finds that flying from Dublin does not give rise to a cost advantage. The validity of the study's results is questionable. It is neither clear how RBB controls for the differences in the routes operated out of the various airports nor which part of maintenance and operating costs were included in the cost calculation. Further, as certain routes in the benchmark group were (rightly) excluded from the analysis (as explained in footnote 42 of the report, routes have been excluded which link the benchmark group airports with a base airport at the other end) it is not clear how possible scale economies linked to the particular base have been allocated between these routes and those which have been taken into account. The report does not discuss whether, when comparing Dublin with the benchmark group of 5 other Ryanair base airports (out of 18), there exists a bias in the selection of these airports. As regards marketing costs, it can be assumed that the study only includes current expenses without taking into account historic expenses linked to establishing the Ryanair brand. These costs are not comparable to the costs of Ryanair's competitors, as set out further below in Section 7.8.

Aer Lingus has also provided its own calculation on cost savings from having a base. It compares a 1-fleet aircraft to a 5-fleet one to illustrate the benefits of arising from operating having a base. It would not, however, necessarily imply that the minimum efficient scale for a base is uniformly 5 aircraft. The calculation takes into account cost for disruption, staff, ground handling, sales and marketing, total block hours etc. and concludes that a cost saving of about 20% can be realised with 5 aircraft. This would only marginally increase to 21% in the case of 10 aircraft (see Aer Lingus' submission of 19.2.2007, folio no.4122).

416 Aer Lingus, for instance, only operates with one base.

417 See, with regard to the example of Charleroi, e.g. Commission Decision of 12 February 2004, Official Journal L 137-1 p.2, 30.4.2004.

418 On 24 of the 35 overlap routes, Aer Lingus or Ryanair changed their weekly rotations operated in Winter 2006/2007 in Summer 2007. See also the various changes of overlapping routes during the time of the Commission's investigation e.g. on the routes Dublin-Bristol, Dublin-Turin, Dublin-Fuerteventura and Manchester-Cork; see also Ryanair's announced shift of six routes at Frankfurt on 18.4.2007, see Ryanair's Response to the Statement of Objections, page 219.

419 With regard to Dublin, this risk of a "negative demand shock" was, however not mentioned by any competitor in the market test, see answers to question 8 and 9 of the Base Questionnaire.

420 See questionnaire to competitors of 6.11.2007, question 36.

421 This is particularly important in the present case as both Aer Lingus and Ryanair sell a large majority of their tickets via their website.

422 See replies to question 44 of the Commission's questionnaire sent to competitors on 6.11.2006; see also minutes of the interview with easyJet of 15.2.2007 and minutes of the interview with bmi of 9 March 2007, folio no. 6170.

423 Aer Lingus indicates at most six routes of all its routes ex-Dublin, see Aer Lingus' submission of 9 February 2007, folio no.4122, page 6. Further, with one exception (Luxair on the Dublin - Manchester route), the relevant carrier was Spanair, which partly operates according to a "charter airline" model (see further discussion of Spanair within the framework of the individual routes in Section 7.9.). Further, according to its website, Spanair currently offers scheduled flights from Dublin only to two of these destinations - Lanzarote and Gran Canaria.

424 According to Ryanair's response of 8 December 2006 these are Treviso-Beauvais, Beauvais-Porto, Tampere- Riga and Newcastle-Top.

425 See interview with easyJet of 15.2.2007, folio no. 6170.

426 See also Ryanair's internal documents : "[...]*". Board paper 3 (Board Meeting the [...]*. P.3.4, folio no. 629). Ryanair's internal documents : [...]* (Board Paper 3; Board Meeting the [...]*: P. 3.4, folio no. 629); "[...]*" (Board Paper 4; Board Meeting the [...]*: P. 1, folio no. 629, text highlighted by the Commission); "[...]*" (Board Paper 4; Board Meeting the [...]*: P. 2, folio no. 629); "[...]*" (Board Paper 3; Board Meeting the [...]* : P.3.2, folio no. 629; text highlighted by the Commission).

427 Ryanair reply of 8 December 2006; pages 8-9.

428 Ryanair press release of 06.02.06.

429 See answers to questions 8, 9 and 10 of the Base Questionnaire; see also answers to question 38 of the questionnaire to competitors sent on 6.11.2006. By way of example, discussing the advantages associated with having a base, a low-cost carrier has stated that it can "start eight new routes using three aircraft by opening up a single new airport". It is a way of increasing "the efficiency of our operations in those destination cities and helps to improve asset utilisation and reduce unit costs", see: easyJet press release of 13.02.2004; another carrier describes the disadvantages of not having a base airport as follows: "[...] this means that for maintenance requirements the aircraft will have to position empty to a maintenance base and a substitute aircraft will have to position empty to replace it. Also if there is any disruption to the operation there is no opportunity to get the operation back on track by substituting another aircraft as there is at a hub/ base. Flight crew must either be positioned to the airport and night stop at a hotel at a cost or be based at the airport. The required number of crew would be higher than the average required per aircraft at a base to allow for standby duties. There are also probable inefficiencies in handling services at the airports if there is only the one aircraft based there.", see Loganair's answer to question 38 in the Questionnaire to Competitors sent on 6.11.2006, folio no. 21059.

430 See answers to question 11 of the Base Questionnaire.

431 See answers to question 11 of the Base Questionnaire; for example British Airways has put forward that a base of 25 aircraft optimises the minimal fleet size with respect to maintenance and standby aircraft (see British Airways' answer of 7 February 2007 to the Commission's "follow-up questions" (No 14). Aer Lingus has provided a detailed account of cost advantages which it argues would arise from operating at a base airport. While it does not define what the minimum efficient scale for a base is, it carries out a comparison of a 1 aircraft fleet with a 5-aircraft fleet to illustrate economies of scale arising from a base. It calculated cost savings for the latter of about [...]*%, see Aer Lingus' submission of 9 February 2007, folio no. 4122, p.4.

432 Ryanair's response to the Article 6(1)c decision of 19 January 2007, p.13.

433 As regards the Ireland-London (Cork, Dublin and Shannon combined) the two companies serve around 93% of traffic on the basis of 2005 data. Charter airlines have formed less than 1% of all traffic on such routes.

434 See Ryanair's Reply to the Statement of Objections, page 57.

435 See above Section 7.3.4.1.

436 See replies to questions 3 and 15 of the Base Questionnaire. In the Reply to the Statement of Objections, even Ryanair does not argue that these carriers have a base at Dublin. Ryanair argues that the following 5 airlines have a base at Dublin: Ryanair, Aer Lingus, Aer Arann, CityJet and BMI (see page 57 of the Reply to the Statement of Objections). However, BMI itself has confirmed to the Commission that they do not consider their aircraft overnighted at Dublin as constituting a "base" (see e.g. BMI's reply to the Commission's "follow-up" questions of 26 February 2007, folio no.43348, answers to questions 5 and 15, and BMI's answer to the Commission's Market Test of the Final Commitments in Phase II, folio no. 9133).

437 This is also confirmed by the reply of DAA of 16 November 2006 to the Questionnaire to Airports, question 4.

438 Moreover, with a seating capacity of 50 seats for the ATR 42 and 66 seats for the ATR 72, it has a significantly higher cost base than Ryanair and Aer Lingus. To match the capacity of 1 Ryanair frequency, with these aircraft, it would have to put three frequencies on the route.

439 The fact that Air France has indicated in its very first reply to the Commission's questionnaire that its smaller aircraft are "perfectly appropriate for regional operations", the directly concerned company CityJet has later confirmed that CityJet must "achieve a higher average yield than Ryanair/Aer Lingus to be profitable - which is secured by focusing on connecting passengers and business passengers". It must therefore focus on quality and not on low-frills/low cost (see e.g. interview with CityJet (Air France) of 21.2.2007, folio no. 6170.

440 See also Section 7.8.

441 Cork is only served by Aer Arann.

442 (see Ryanair's map. "Competitor Bases - All Ryanair City Pairs and Aer Lingus City Pairs from Dublin).

443 Taking into account a "safety margin" and only counting those routes as "outbound" on which Irish passengers account for more than 60% according to the Commission's data.

444 Ryanair rightly mentioned that the DAA offers an "incentive scheme" which is intended to attract new entrants to Dublin. This scheme could, however, apparently not remove the existing entry barriers to a sufficient extent to attract a large number of new airlines competing with Ryanair.

445 The fact that some charter carriers offer some minor volumes (>5%) of dry-seats on some single routes does not alter this conclusion.

446 See e.g. Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 52; text in brackets inserted by the Commission.

447 See further Annex I on the details of the Customer Survey.

448 See Ryanair's Response to the Statement of Objections of 17 April 2007, for example, paragraph 69 or 370

449 Ryanair also argues that "it is not correct to say that passengers were "interviewed" and the use of this expression gives more weight to the results than is due to a self completion questionnaire" (See Ryanair's Response to the Statement of Objections of 17 April 2007, see paragraph 381). However, Ryanair itself uses this term repeatedly, for example four paragraphs above in 377, in 376, 374 etc.

450 477 passengers out of 1166 travelling with Ryanair (40.9%) said they had considered using Aer Lingus in the route they were going to travel on that day. Only 12.8% of Ryanair passengers considered any other carrier. The difference in percentages is statistically significant at less than 1% level. As regards Aer Lingus, 430 passengers out of 1042 travelling with Aer Lingus (41.3%) stated that they had considered using Ryanair on the route they were going to travel on that day. Only 10.4% of Aer Lingus passengers considered any other carrier. The difference in percentages is statistically significant at less than 1% level. For further details see Annex I, Section 3 and table A.4.5 in Annex II.

451 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 71

452 Similarly neither party expressed any concern with respect to the use of the word "route". Ryanair in its response to the Statement of Objections (paragraph 387, 388 and 389) now claims that it is imprecise as respondents may not be aware that there may be other airports at the destination city. This, however, would only bias the responses towards less substitutability, not more. In any event it would be inappropriate to regard this as a bias since what is of interest is precisely whether the passengers regard other carriers as substitutes. One reason they might not is that they are unaware that a rival is flying to an airport nearby to the destination city. In any event, the respondent had the possibility to ask the interviewer for clarification if needed.

453 See further table A.4.4 in Annex II.

454 Similar proportions apply in the case of Aer Lingus passengers. Of those indicating that they always travel with Aer Lingus 21% had considered Ryanair and 8% other airlines). In contrast 68.9% of passengers that do not always travel with Aer Lingus considered Ryanair (and 16% other airlines)

455 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 70, second indent.

456 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 374.

457 The raw data was gathered by the Commission and therefore was not confidential in any way. To generate the weights it is possible to use as an approximation the number of yearly passengers on a given route (this is the approach followed by the Commission as explained below).

458 Response of Aer Lingus of 28 November 2006 to Additional Questionnaire, Annex 4, folio no. 23422.

459 Notification, paragraph 50.

460 See also Section 7.4.3 and Annex IV.

461 See also the figure for 2005 in Aer Lingus' IPO Prospectus, page 40.

462 See in more detail in the next paragraph.

463 Cf. the "Revenue Management System"/"RMS" for Ryanair and the "PROS 5.2" system for Aer Lingus'.

464 See e.g. minutes of the Commission's site visit in Dublin (Ryanair) of 6.2.2007, folio no. 6170, and minutes of the Commission's site visit in Dublin (Aer Lingus) of 6.2.2007, folio no. 6170.

465 [...]*

466 See minutes of the Commission's site visit in Dublin (Ryanair) of 6.2.2007 as changed and agreed by Ryanair by e-mail of 14 February 2007, folio no. 6170,.

467 See e.g. minutes of the Commission's site visit in Dublin (Ryanair) of 6.2.2007, folio no. 6170; see also minutes of the Commission's site visit in Dublin (Aer Lingus) of 6.2.2007, folio no. 6170.

468 See minutes of the meeting with Aer Lingus at Dublin on 5 February 2007and internal Aer Lingus presentation "Aer Lingus Revenue Management - April & May 2006 Presentations Combined", slide 24, folio no. 6170.

469 See Section 7.4.1.

470 See Section 7.4.2.

471 See as to the constraint of Ryanair by other airlines including Aer Lingus Ryanair's internal documents, e.g. Board Paper 1 (Board Meeting of [...]*, folio no. 629): P 3.3: "[...]*"; Board Paper 4 (Board Meeting the [...]*, folio no. 629): P. 2.2.: "[...]*" and Board Paper 3 (Board Meeting the [...]*, folio no. 629): P. 3.3.:"[...]*".

472 See Section 7.4.3.

473 See in this respect also Ryanair's "lowest fares" guarantee to customers, see http://www.ryanair.com/site/EN/notices.php?notice=070508_EN_doubledifference&pos=CDAY.

474 See for example press releases of Ryanair of 8 October 2006, "Ryanair opens 2 new routes to Warsaw and Stockholm"; of 9 August 2006, "Ryanair announces biggest ever expansion at Dublin (...)", in which Ryanair states: " "On every route where there they have a choice, millions more passengers prefer Ryanair to Aer Lingus. Today's twelve new routes will end Aer Lingus' high fare monopoly to destinations such as Madrid, Bologna, Almeria and Seville. Irish passengers who previously had no alternative to Aer Lingus' high fares and frequent delays on these routes will now have a low fare, on-time option with Ryanair. Irish passengers will also enjoy Ryanair's low fares to new European destinations such as Norway (Oslo Torp), Denmark (Billund), Finland (Tampere) and Eastern Poland (Rzeszow) which are not currently served by Aer Lingus." 10 May 2006, "Ryanair announces new route from Dublin to Berlin ", 21 December 2007, "Ryanair launches 18 routes from Dublin to Europe", in which Ryanair states: "Aer Lingus can't compete with Ryanair's prices, they can't match our punctuality, and now they can't match our route network from Dublin to the UK or Europe". See also Aer Lingus' submission of 28 November 2007, Annex 5, folio no. 4122.

475 See e.g. Aer Lingus' submission of 22 February 2007 Annex 9, in which Aer Lingus directly reacts to a recent Ryanair campaign; see also the examples of advertisements directly comparing Ryanair's and Aer Lingus' average fares in Aer Lingus' submission of 19 February ("Further Comments on the Relevance of Base Competition"), folio no. 4122.

476 See Ryanair's internal documents: "[...]*" Board Paper 4 (Board Meeting the [...]*, folio no. 629): P. 2.2. and "[...]*" Board Paper 3 (Board Meeting the [...]*, folio no. 629): P. 3.3.

477 See Annex 6 to Aer Lingus' Response of 28/11/2006 to Additional Questionnaire dated 22 November 2006, folio no. 4122. According to Aer Lingus, a Aer Lingus' promotion (in October 2001) covered flights to several cities, including destinations also served by Ryanair (e.g. Barcelona. Lyon, Milan, Berlin and Rome). In this instance, Ryanair lowered, according to Aer Lingus, its price in response to Aer Lingus' promotional prices both in the case of airport-pair and city-pair overlap routes, and extended the duration of its promotion in Ireland to match Aer Lingus' promotion. Similarly, during the time of Aer Lingus promotional prices to Lyon St Exupéry (LYS), Ryanair had, according to Aer Lingus, massively lowered its prices but increased them once Aer Lingus promotion came to an end. The same happened on the route to Milan (Aer Lingus flying to Linate (LIN) and Ryanair to Bergamo (BGY)). After Aer Lingus promotion ended, Aer Lingus' minimum fare rose, according to Aer Lingus.

478 Sources of route heterogeneity possibly affecting fares include: characteristics of the destination city, number of alternative airports at destination, characteristics of these airports, the popularity of the route according to purpose of travel, customer awareness and expectations, route distance, duration of travel, air traffic regulations at country of destination, population levels and population density and population income at destination, cultural affinity between origin and destination countries. Not all these variables can be measured or indeed observed.

479 RBB "Bargepole: Ryanair cross-sectional regression results", 10 December 2006.

480 The term "specification" refers to the regression equation.

481 In technical terms the estimated coefficients will be subject to bias when important variables affecting price in different routes cannot be included in the regression (for example because they cannot be observed or measures) and are correlated with one or more of the explanatory variables in the regression. This problem is often referred to as omitted variable bias.

482 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 142.

483 RBB "ANNEX: Is Aer Lingus a competitive constraint on Ryanair?", 08 February 2007

484 RBB "ANNEX: Is Aer Lingus a competitive constraint on Ryanair?", 08 February 2007, page 3.

485 RBB "Comparing prices on "monopoly" and non-monopoly routes out of Frankfurt Hahn" included as Annex 1 in RBB "Position Paper on Barriers to Entry" dated 20th February 2007, folio no. 4135.

486 In order to capture more events of Aer Lingus entering, the extensive data set has enabled the Commission to consider a longer time period, starting from Aril 1997. While in fact Aer Lingus' presence has a significant negative effect on Ryanair's prices in that regression, for a number of reasons - as set out in Annex IV - the Commission does not give weight to this result.

487 See on the mutual competition between Ryanair and Aer Lingus also Section 7.4.2 (reaction to promotions) and on competition with other carriers e.g. Section 7.8.5 (retaliation in reaction to entry).

488 It may in this context be noted that the Commission's evidence has shown that not only Aer Lingus reacts to Ryanair, but that also Ryanair reacts to Aer Lingus competitive behaviour. However, for a serious competition concern it would not even be necessary that both Aer Lingus and Ryanair constrain each other in equal measure. It is well possible (and in this case even likely) that one firm exercises on the other a stronger competition concern than conversely. If customers respond by switching to a change in relative prices in one direction (say from Aer Lingus to Ryanair) at least this customer group can be expected to switch back if the change of relative prices is reversed.

489 Horizontal Merger Guidelines, paragraph 8. Although not explicitly mentioned here, the merger will also reduce the choice for Irish customers who can currently choose between two differentiated airlines on most of the affected routes. The Commission also notes that the merger may well have an impact on the competition on quality between the two Merging Parties, to the detriment of the customers.

490 In its response to the Statement-of-Objections, Ryanair argues that its expansion was no reaction to the expansion of Aer Lingus, but that it was hindered by conflicts with Dublin Airport to expand. However, the Commission notes that only started expanding ex-Ireland subsequent to Aer Lingus' restructuring, possibly as a result of a need for Ryanair to re-orientate itself to the Irish market. Ryanair also argues that, in spite of opening new routes, Aer Lingus has not expanded capacity. However, while it is correct that Ryanair's has expanded more in terms of aircraft capacity and overall passenger figures on a European-wide level than Aer Lingus (and any other airline in Europe), the situation is more balanced if one looks at the development in Ireland. By way of example, Aer Lingus operates today more routes from/to Dublin than Ryanair). Further, by replacing its fleet with larger aircraft Aer Lingus has indeed added capacity on the market and increased its passenger numbers from 6.6 million in 2001 to 8,6 million in 2006.

491 See in detail Section 7.6. on potential competition.

492 Ryanair argues that Ryanair's "pattern of opening routes at Dublin is a result from Ryanair's practice to "join the dots", i.e. to add routes between airports that are already in the [Ryanair] network", see response to the Statement of Objections, page 46. However, as the total number of destinations served by Ryanair was steadily growing in recent years (according to the Ryanair's annual Form 20-F Statements for the US SEC (source: www.ryanair.com), their network of locations served evolved as follows: Sept 2003 - 83 locations served, Sept 2004 - 88 locations served, Sept 2005 - 107 locations served, Sept 2006 - 115 locations served) and in view of the overall expansion plans of Ryanair, it cannot be presumed that there will be no other "dots" added to the Ryanair's network in the near future.

493 Airlines opening new routes face typically some uncertainty about the size and the development of the market and demand often grows only over time. The "first-mover", once established in the market, is therefore well positioned to develop the market by adding more frequencies as demand grows.

494 The intensity of the competition to open new routes is also linked to the fact that both Ryanair and Air Lingus have a large base in Dublin. As discussed above, the base allows a more efficient use of aircraft and also provides the ability to shift capacity and enter new routes higher speed. To maximise aircraft utilisation, 3-4 round trips per aircraft have to be carried out during one day. Such an efficient use of the fleet requires optimising schedules across routes. Only by operating on a bundle of routes out of the same airport an airline can fly on "thinner" routes which on their own would not generate sufficient traffic to dedicate one large aircraft (with 3-4 daily round-trips). Further, a large slot portfolio at a base airport allows for greater operational flexibility to react quickly to changes in the market.

495 The term "increased fares" in this context should be understood as encompassing also lower decrease of fares than absent the merger.

496 See e.g. the Commission decisions in cases M.3280 - Air France/KLM, M.3770 - Lufthansa/Swiss, M.3940 - Lufthansa/Eurowings; see also the discussion of potential competition by the CFI, case T-177/04 easyJet v Commission, of 4 July 2006 ECR (2006) II-1913, at paragraphs 63 et seq.

497 See as regards the advantage of increased flexibility Section 7.3.4.1.

498 The previous overlap routes Dublin-Bristol, Dublin-Turin and Dublin-Fuerteventura identified in the Article 6(1)(c)-decision and the Statement of Objections "disappeared" during the time of the investigation because one airline ceased operations on these routes, while at least one other overlap route (Manchester-Cork) was created during the investigation (it may also be noted that the Commission has evidence that further overlap routes will be created in the near future, e.g. through the announced opening of a route from Dublin to Copenhagen by Aer Lingus in Winter 2007/2008).

499 See paragraphs 58-60 of the Horizontal Merger Guidelines.

500 See paragraph 59 of the Horizontal Merger Guidelines.

501 Aer Lingus's and Ryanair's large operations in Dublin, their access to airport facilities and in particular their easy access to Irish (and for Ryanair even non-Irish) customers make shifting routes and opening new routes from/to Irish airports significantly easier and cheaper than it would be for their competitors, who would face considerably higher sunk costs. See in detail below, Section 7.8 on entry barriers (notably 7.8.3 and 7.8.4).

502 See paragraph 59 of the Commission's Notice on the Assessment of Horizontal Mergers: "A merger with a potential competitor can generate horizontal anti-competitive effects, whether coordinated or noncoordinated, if the potential competitor significantly constrains the behaviour of the firms active in the market. This is the case if the potential competitor possesses assets that could easily be used to enter the market without incurring significant sunk costs. Anticompetitive effects may also occur where the merging partner is very likely to incur the necessary sunk costs to enter the market in a relatively short period of time after which this company would constrain the behaviour of the firms currently active in the market."

503 See Aer Lingus press release "Preliminary Announcement of Results for the Year Ended 31 December 2006" of 13 March 2007, available at: www.airlingus.com.

504 Table in paragraph 21 of Ryanair's Response to DoT Briefing Paper

505 See Ryanair's Response to the Statement of Objections of 17 April 2007, page 244.

506 See Aer Lingus Response to Commission Request of 19 March 2007 submitted on 21 March 2007, folio no. 5933.

507 See Notification, paragraph 21.

508 See Notification, paragraph 23.

509 See the public Statement of 01/08/2006 "RYANAIR ANNOUNCE RECORD Q.1 RESULTS - NET PROFIT RISES 80% TO €116m - TRAFFIC GROWS 25% TO 10.7m" available at Ryanair's webpage: http://www.ryanair.com/site/EN/about.php?page=Invest&sec=download&ref=2007

510 See Board Paper 3 for the Board Meeting of Ryanair Holdings plc on [...]* (folio no. 629), page 3.6., section 3.4.1. - Aer Lingus stating [...]*.

511 See e.g. press release of 09/01/2007, "2 New Routes from Dublin to Bydgoszcz and Gdansk" stating among others as follows: "Ryanair, Europe's largest low fares airline today (Tuesday, 9th January 2007) announced that from May it will launch 2 new routes from Dublin to Bydgoszcz and Gdansk in Poland and increased frequencies from Dublin to Bratislava, Biarritz, Carcassonne, Kaunas, Krakow, Malaga, Murcia, Riga and Rome. Ryanair will launch a total of 22 new European destinations from Dublin for Summer 2007."

512 It should be however noted, that there is at least one additional destination served by Aer Lingus from Dublin which is already part of the Ryanair's network, namely Santiago de Compostela (served by Ryanair currently from Liverpool, London, Frankfurt and Rome - see www.ryanair.com).

513 According to the Ryanair's annual "Form 20-F Statements" for the US SEC (source: www.ryanair.com), their network of locations served evolved as follows: Sept 2003 - 83 locations served, Sept 2004 - 88 locations served, Sept 2005 - 107 locations served, Sept 2006 - 115 locations served.

514 See e.g. Ryanair's internal documents: "[...]*". Board Paper 1 (Board Meeting the [...]*, folio no. 629) : P. 1.2

515 It is noted that Aer Lingus already had a strong presence in Ireland as the national "flag carrier".

516 As mentioned above, there will be no actual competition between the parties anymore on the overlap routes "Dublin-Turin" and "Dublin-Fuerteventura" as of Winter 2007/2008.

517 See Ryanair's internal documents: "[...]*". Board Paper 3 (Board Meeting the [...]*, folio no. 629) : P. 3.4.

518 See also the recently announced entry by Aer Lingus to the Dublin - Copenhagen route as described in footnote 357 above.

519 Brussels Airlines were formed through the merger of SN Brussels and Virgin Express.

520 See: Barbara Cassani: Go - an airline adventure, p. 260; quoted from DoT briefing paper of 13 November 2006, folio no. 6444.

521 The Commission notes that the number of "potential" overlap routes is much higher than the number of routes identified as "most likely" entry routes in the following paragraphs and that the fact that other routes are excluded does not mean that there is no threat of entry at all.

522 It is argued that it is not the current traffic on the route (i.e. how many frequencies are currently operated and thus how thick or thin the route is) which is relevant for possibilities of entry, but rather the potential for further growth. This is also evidenced by some current overlap routes as Ryanair for examples entered in January 2007 the route Dublin - Seville even though Aer Lingus operated only 2 weekly rotations on this route.

523 This criterion shows that the Commission's approach is rather prudent, since the remaining competitors may be small or following a different operating/service model and therefore be a very limited threat to expansion.

524 Based on the information available on the websites of Aer Lingus (www.airlingus.com), Ryanair (www.ryanair.com) and a Dublin Airport (www.dublinairport.com).

525 One rotation is defined as an outbound and inbound flight between the two ends of the O&D pair, i.e. two one-way flights in the opposite directions.

526 The route between Dublin and Santiago de Compostela is being launched by Aer Lingus as of 27 March 2007.

527 See Section 7.9.3.

528 According to the Ryanair's annual Form 20-F Statements for the US SEC (source: www.ryanair.com), their network of locations served evolved as follows: Sept 2003 - 83 locations served, Sept 2004 - 88 locations served, Sept 2005 - 107 locations served, Sept 2006 - 115 locations served.

529 See paragraph 60 of the Horizontal Merger Guidelines.

530 As indicated above, all these five destinations are predominantly holiday destinations with majority of Irish passengers and potential for further growth. Further, there are no competing carriers based at the destination airports (DBV, NAP, NCE, SCQ) or these carriers would be less strong competitors compared to Ryanair (PMI). Finally, these airports are not capacity constraint and cannot be considered as primary airports.

531 Based on the information available on the websites of Aer Lingus (www.airlingus.com), Ryanair (www.ryanair.com) and of Dublin Airport (www.dublinairport.com).

532 See also paragraph 569 below.

533 See Section 7.9.3.

534 Based on the information available on the websites of Aer Lingus (www.airlingus.com), Ryanair (www.ryanair.com) and of Dublin Airport (www.dublinairport.com).

535 See Section 7.9.3.

536 See Ryanair's Preliminary Response to the Article 6(1)(c) Decision of 19 January 2007.

537 For instance, Ryanair has indicated that it would be prepared to base more aircraft at Cork, if it can reach an agreement with the airport.

538 See paragraph 60 of the Horizontal Merger Guidelines.

539 Virgin Express closed its Shannon base in 2001. EUjet established a base in Shannon in May 2003 but ceased operations in July 2005. In 2004 Skynet withdrew, in 2005 Thomsonfly, Hapag-Lloyd, BA Cityexpress and Flybe and in 2006 easyJet.

540 Based on the information available on the websites of Aer Lingus (www.airlingus.com), Ryanair (www.ryanair.com) and of Dublin Airport (www.dublinairport.com).

541 See also paragraph 569 below.

542 See Section 7.9.3.

543 Horizontal Merger Guidelines, paragraph 31.

544 The term "prices increase" in this context should be understood as encompassing also lower decrease of prices than absent the merger, as well as reduction of output, choice or quality or lower innovation (see paragraph 8 and footnote 7 of the Horizontal Merger Guidelines).

545 Horizontal Merger Guidelines, paragraph 31.

546 Horizontal Merger Guidelines, paragraph 64, in particular the definition of countervailing buyer power: "Countervailing buyer power in this context should be understood as the bargaining strength that the buyer has vis-à-vis the seller in commercial negotiations due to its size, its commercial significance to the seller and its ability to switch to alternative suppliers."

547 Horizontal Merger Guidelines, paragraph 68.

548 See e.g. Notification, paragraph 210; Ryanair's answer to the Commission's Article 6(1)(c) -decision, pages 20-27; RBB paper of 20.2.2007 (Position Paper on Entry Barriers).

549 See Horizontal Merger Guidelines, paragraph 70. It is important to note that in order to constitute a barrier to entry, it is therefore not necessary to entirely exclude competitors from the market.

550 Case T-282-02 Cementbouw Handel & Industrie v Commission [2006] ECR II-319, paragraph 219.

551 In order to verify whether there exist significant entry barriers or not, the Commission has not only sent questionnaires to the actual and potential competitors and to customers of airlines. It has also taken into account the results of the Customer Survey at Dublin Airport that was carried out on behalf of the Commission and the data submitted by Aer Lingus and Ryanair on passenger numbers and market shares on the different routes. The Commission has also conducted specific interviews with the main potential entrants in order to learn more about the factual elements that might or might not prevent them from entering the Irish market.

552 It should be noted that this is in line with the Commission's case practice in previous cases, as confirmed by the ECJ, see: CFI, case T-177-04 easyJet v Commission, of 4 July 2006 ECR (2006), II-1913, at paragraphs 197 et seq.

553 See Notification, paragraph 210.

554 See for intra-European flights insofar already the Commission's decision in case M.3770 - Lufthansa/Swiss.

555 Although it might be argued that non-EU-based airlines still face significant regulatory entry barriers, entry of non-EU based airlines on the short-haul market is in any event very unlikely.

556 See replies to questions 35 and 42 of the questionnaire to competitors of 06.11.2006. Only one competitor mentioned regulatory barriers at all.

557 See Commission Decision of 4 July 2005 in the case M.3770 - Lufthansa/Swiss, paragraph 43. Similarly, see also paragraph 73 of the Commission Decision of 11 February 2004 in the case M.3280 - Air France/KLM.

558 Aer Arann is also in Cork. However, its operations there are limited.

559 See Section 7.3.4. It should be noted that the most significant benefits from operating with a large number of based aircraft can be achieved at Dublin. As regards the other Irish airports, the benefits are smaller. Three airlines have declared to have a base at Cork. Aer Lingus has based 4 A320 aircraft, operating 17 destinations, Ryanair 4 destinations with one B 737-800 aircraft and Aer Arann 9 destinations with 3 aircraft. At Shannon, Ryanair operates 26 destinations with 4 aircraft, Aer Lingus long-haul services and two shorthaul ones to Dublin and London. For the latter operations the aircraft are not based in Shannon. Thus, on these airports in comparison to newcomers the "base advantage" of the parties is more limited.

560 See interview with Aer Arann of 13.2.2007 and with CityJet (Air France) dated 21 February 2007, folio no. 6170.

561 See interview with CityJet (Air France) dated 21 February 2007, folio no. 6170.

562 See minutes with Aer Arann of 13.2.2007, folio no. 6170. It should be noted that the "base disadvantage" at Cork for Aer Arann is, if any, significantly smaller, given that it has already 3 aircraft at that airport. However, given its fleet of turboprop aircraft, it cannot be regarded as an effective competitor

563 See minutes of the interviews with Ryanair's potential competitors, folio no. 6170 and 8091; see also replies to the Questionnaire to Competitors sent on 06/11/2006 and to the follow-up questions to individual competitors.

564 See e.g. replies of competitors to the Market Testing of Proposed Remedies - Competitors, sent on 30 November 2006, in particular question 10 and minutes of interviews with competitors, , folio no. 6170 and 8091.

565 See e.g. Ryanair's response to the Statement of Objections, page 238 et seq.

566 See in detail below in Section 7.9.6.

567 It should be noted that the other entry barriers discussed below militate not only against entry or expansion as such, but notably against opening a base at Dublin.

568 Jetmagic, a low-cost airline, established its base in Cork with 3 Embraer 135/145 in 2003 but withdrew its operations in January 2004. As explained in detail below, easyJet withdrew from these airports as a result of Ryanair's strategic behaviour. In the case of Shannon, Virgin Express closed its base in 2001. Skynet withdrew in 2004, and Thomsonfly, Hapag-Lloyd, BA Cityexpress and Flybe in 2005. EUjet established its base in May 2003 with Fokker F100 aircraft but ceased its operations in July 2005..

569 See in more detail below, Section 7.8.4.

570 There are 14 overlap routes where the merged entity would have a base on both sides and they represent almost 75% of total traffic on all overlap routes. These routes are as follows: Dublin - London, Dublin - Glasgow, Dublin - Manchester, Dublin - Birmingham, Dublin - Brussels, Dublin - Frankfurt, Dublin - Madrid, Dublin - Barcelona, Dublin - Milan, Dublin - Rome, Dublin - Marseille, Cork - London, Shannon - London and Cork - Manchester.

571 Calculation based on information from Ryanair (map: competitor bases - all city pairs from Dublin).

572 Aircraft may also be flown in empty to the starting point of the route from a base in the morning and return empty in the evening, however, with additional costs; aircraft may also be parked over night at destinations where a carrier has no base (this however involves additional costs such as hotel expenses for the crew, technical support at non-base airport and others).

573 For instance, CityJet offers such a triangular service when flying Dublin-London, London-Paris and Paris- Dublin. See for other examples (often referring to the past or to charter carriers) RBB Paper of 20.2.2007, pages 17-19.

574 See answers to question 38 of the Questionnaire to Competitors of 6.11.2007.

575 idem.

576 See minutes from interview with easyJet, folio no. 6170.

577 See answers to question 38 of the Questionnaire to Competitors sent on 6 November 2006.

578 From all 437 routes Ryanair operates, only 4 are, according to Ryanair, not base-connected. From all 275 routes easyJet operates, only 2 are not base-connected, see submission of Ryanair/RBB Paper of 20.2.2007. See also minutes from interview with easyJet, folio no. 6170.

579 See e.g. Ryanair's reply to the Commission's Article 6(1)(c)-decision, pages 25 and 26and RBB report of 20 February 2007,p.24.

580 Horizontal Merger Guidelines, paragraph 71 (c).

581 See Commission Decision of 4 July 2005 in the case M.3770 - Lufthansa/Swiss, paragraph 44.

582 Idem, paragraph 45.

583 Response of Aer Lingus of 28 November 2006 to Additional Questionnaire, Annex 4, folio no. 4122.

584 See e.g. Notification, paragraph 219.

585 See answers to Questions 35 and 42 of the Questionnaire to Competitors sent on 6 November 2006 and interview with competitors, folio no. 6170 and 8091.

586 See interview with easyJet of 15.2.2007, folio no. 6170.

587 See Ryanair's answer to the Commission's Article 6(1)(c)-decision, page 24.

588 See Ryanair Paper of 20.2.2007, page 4. For example [...]*.

589 See further on this point the route-by-route analysis.

590 The Commission's Customer Survey and the data submitted by the DOT (see DOT submission of 8 February 2007, page 9, folio no. 6230) seem to suggest that the share of Irish-based passengers might be higher than 50% even on some of the Ireland-UK routes.

591 These are London Stansted, London Luton, Liverpool, East Midlands and Glasgow.

592 Interview with easyJet of 13.2.2007, folio no. 6170.

593 See e.g. minutes of interview with easyJet of 15.2.2007, folio no. 6170.

594 It should be noted that even a relatively well-known low-frills carriers such as easyJet indicated that it would be difficult to compete against the two well-established brands Aer Lingus and Ryanair in Dublin. Even on routes where about 50% of the customers originate in the UK, "it would still not be profitable to start a route only on the basis of the UK customers knowing easyjet. An important share of Irish customers is necessary as well", see interview with easyJet of 15.2.2007, folio no. 6170.

595 See in this respect e.g. the Statement of Jet2.com, interview of 1.2.2007, folio no. 6170: "...many Irish flights are outbound flights which makes entry for foreign companies more difficult"; interview with Flybe of 12.3.2007, folio no. 6170: "[Flybe] would rather expand its activities in Northern Ireland (Belfast), where the "Flybe" brand is already established and well-known. Establishing a brand at a new destination, according to [Flybe], require[s] "a major investment"; interview with BMI of 9.3.2007, folio no. 6170: "Entering new markets requires establishing a brand, which is costly."; interview with AerArann of 13.2.2007, folio no. 6170: "Opening a new route would require significant investment due to the costs necessary to establish a brand on the Irish market which would be able to compete with two the wellestablished brands Aer Lingus and Ryanair"; interview with Air Berlin of 13.2.2007, folio no. 6170: "An entry into the Irish market would, according to Mr Rautenberg, in any event require "a lot of money", in particular to establish a brand in Ireland against the now well-known Ryanair and Aer Lingus brands."

596 See e.g. notably minutes of interviews with Flybe of 12.3.2007, BMI of 9.3.2007; AerArann of 13.2.2007; Jet2.com of 1.2.2007, folio no. 6170.

597 See: http://www.alexa.com/site/ds/top_sites?cc=IE&ts_mode=country&lang=none

598 http://www.alexa.com/site/ds/top_500?qterm=

599 Even taking into account that no other competitor was explicitly mentioned, this result can still deemed to be relevant for the assessment; see further Annex I.

600 See Notification.

601 See e.g. case M.3770 - Lufthansa/Swiss, paragraph 44.

602 It should be noted that two of the four competitors Ryanair refers to (e.g. CityJet (AirFrance) or easyJet) have very well highlighted the existence of significant entry barriers, inter alia in terms of marketing costs, see e.g. CityJet (Air France), interview of 21.2.2007, folio no. 6170: "...setting up operations in a new airport/country is a complex operation ("a massive operation"), requiring new ground services, customer care, technical operations, sales and marketing."; interview with easyJet of 15.2.2007, folio no. 6170: "entry to Ireland would be a 2significant decision" with significant investment (...); starting operations would need significant marketing efforts with significant costs. (...) Therefore, to get customers on any Ireland-UK route, easyJet would have to make significant investments into building a brand in Ireland."

603 See e.g. Notification, paragraph 500.

604 See Ryanair's reply to the Commission's Article 6(1)(c)-decision, page 25; RBB Paper of 20.2.2007, page 24; Ryanair's response to the Statement of Objections, see e.g. page 241.

605 See in this respect e.g. the Statement of Jet2.com, interview of 1.2.2007, folio no. 6170: "It would be a "huge problem" to try to compete in Dublin, where Ryanair is so well-known. In particular, it would be hard to establish a brand which is sufficiently recognised to be eligible for customers. Aer Lingus and Ryanair, being present in Ireland for 15 years, have a strong presence in Ireland and enjoy an important advantage over their competitors.";

Interview with Flybe of 12.3.2007, folio no. 6170: "[Flybe] would rather expand its activities in Northern Ireland (Belfast), where the "Flybe" brand is already established and well-known. Establishing a brand at a new destination, according to [Flybe], require[s] "a major investment";

Interview with BMI of 9.3.2007, folio no. 6170: "Entering new markets requires establishing a brand, which is costly.";

Interview with AerArann of 13.2.2007, folio no. 6170: "Opening a new route would require significant investment due to the costs necessary to establish a brand on the Irish market which would be able to compete with two the well-established brands Aer Lingus and Ryanair";

Interview with Air Berlin of 13.2.2007, folio no. 6170: "An entry into the Irish market would, according to Mr Rautenberg, in any event require "a lot of money", in particular to establish a brand in Ireland against the now well-known Ryanair and Aer Lingus brands."

Interview with Clickair (Iberia) of 26.3.2007, folio no. 8091: "According to [Clickair], "not anybody will step in" in case of a merger between Ryanair and Aer Lingus. This is because it would require a critical mass of aircraft (more than 5) positioned at a base at Dublin. (...) Entry would therefore not be a realistic option, since a sufficient number of Irish customers could only be won with a base at Dublin and a very significant sustained marketing budget."

606 See the Commission's "Charleroi" decision concerning state aids, Commission Decision of 12 February 2004, Official Journal L 137-1 p.3, 30.4.2004.

607 See interview of 12.3.2007 with flybe, folio no. 6170; see also interview of 15.2.2007 with easyJet, folio no. 6170: "important investment"; interview with Clickair of 26.3.2007, folio no. 8091: "very significant, sustained marketing budget"; interview with Aer Arann of 13.2.2007, folio no. 6170: "significant investment"; interview with Air Berlin of 13.2.2007, folio no. 6170: "[would require] a lot of money".

608 See in this respect also answers of competitors to the Market Test of Ryanair's Final Commitments, answers to questions 3 and 4.

609 Horizontal Merger Guidelines, paragraph 69.

610 See e.g. interview with Jet2.com of 1.2.2007, folio no. 6170: € 4 million, "mainly for advertising".

611 See also Commission Decision of 4 July 2005 in the case M.3770 - Lufthansa/Swiss, paragraph 44.

612 Depending on the level of service the airline offers, airlines would also have to build or pay for lounges etc.; see in this respect also CityJet (Air France), interview of 21.2.2007, folio no. 6170: "...setting up operations in a new airport/country is a complex operation ("a massive operation"), requiring new ground services, customer care, technical operations, sales and marketing."

613 Some airports provide some of these services themselves (such as Dublin Airport for security and luggage handling) and include the costs in the airport charges.

614 See in this respect e.g. the interview with CityJet (Air France) of 21.2.2007, folio no. 6170.

615 See interview with Aer Arann of 13.2.2007, folio no. 6170 (referring to the examples of the failed entry by GO and easyJet); Aer Arann considers that Ryanair can afford to employ this strategy due their "deep pockets", i.e. it is financially much stronger than most of its competitors and Aer Arann in particular. See also interview with Jet2com of 1.2.2007, folio no. 6170. Jet2.com pointed out that the example of easyJet, having recently tried to enter the Irish market without success, showed very clearly that Ryanair reacts "very aggressively, offering flights "for nothing", using its size and its market power in such a situation". See also interview with easyJet of 15.2.2007, folio no. 6170;

616 See Ryanair's answer to the Commission's Article 6(1)(c)-decision, page 25; RBB Paper of 20.2.2007, pages 26 et seq.

617 See agreed minutes of the Commission's visit of Ryanair's headquarters of 6.2.2007, folio no. 6170: "[...]*".

618 See e.g. the description of Ryanair's aggressive reaction by Jet2.com (interview of 1.2.2007, folio no. 6170); Aer Arann (interview of 1.2.2007, folio no. 6170); easyJet (interview of 15.2.2007, folio no. 6170); Clickair (interview of 26.3.2007, folio no. 8091).

619 In its response to the Statement of Objections Ryanair points out that "Ryanair's entire business model revolves around systematically undercutting competitor's prices by a substantial degree on every market in which it operates". This would therefore not be a special reaction to entry. However, as discussed below, in the case of entry, this has been used strategically, i.e. with the intention to make the competitor leave the market.

620 Ryanair has argued that easyJet has set up a base in Madrid only after Ryanair had done so as well. As both would compete on Madrid-Paris and Madrid-London, it would be clear that easyJet is not deterred from entry against Ryanair. However, also this example refers to markets which are outside the scope of this investigation and therefore is not well suited to judge the likelihood of new entry on routes to/from Ireland. In comparison to such routes, on the Madrid-Paris and Madrid-London, easyJet seems to be in a particular strong position as it has a base airport at both ends.

621 See on the relevance of past experience with entry also the Horizontal Merger Guidelines, paragraph 70.

622 (in total 63 examples of such entry have been identified by Ryanair in the period April 2003 - October 2006), see Ryanair's submission "Ryanair/Aer Lingus: Position Paper on Barriers to Entry" of 20 February 2007, folio no. 4135.

623 See on the fact that Ireland on the specific difficulties to enter the Irish market above, Section 7..8.6.

624 On the routes Dublin - Liverpool and Dublin - Bristol.

625 This concerned the following routes: Cork - London (bmibaby), Glasgow - Shannon (Flybe), Dublin - Bristol (Air Southwest), Dublin - Glasgow (British Airways), Knock - London (easyJet) and Shannon - London (easyJet).

626 It points out: "It is no surprise that entrants cannot sustain competition against Ryanair, when Ryanair has significantly lower costs and lower fares than all other airlines" (p. 243).

627 Public statements by Ryanair itself could be understood of building up such a reputation. Responding to a question what Ryanair would do with a surplus of EUR 2.4 bn, in an interview for "Wirtschaftswoche" on 12 March 2007, M. O'Leary declared that Ryanair would need most of the money to deter competitors in the case of a price war. http://www.wiwo.de/pswiwo/fn/ww2/sfn/bm_artikel/bmpara/1567bmpara/41525420

628 See e.g. the description by Jet2.com (interview of 1.2.2007, folio no. 6170); Aer Arann (interview of 1.2.2007, folio no. 6170) and easyJet itself (interview of 15.2.2007, folio no. 6170).

629 Interview with easyJet of 15.2.2007, folio no. 6170.

630 Ryanair's CEO Mr. O'Leary reacted publicly to easyJet's entry on the Irish Sea routes: "We can't wait to welcome the sixth candidate from the UK that has come to liberate the people of the west of Ireland. They will follow in a long line of others who have left" (emphasis added). See: Article in the Irish Times, 24 September 2004.

631 See submission of the DOT of 24.11.2006, folio no. 6233, and of 7.2.2007, folio no. 6233.

632 See for a confirmation of the dates related to easyJet's Irish operations e-mail sent by easyJet on 16 May 2007, folio no. 9859.

633 See Ryanair's Response to the Statement of Objections, paragraph 194.

634 See interview with Ryanair (visit of Ryanair's headquarters) of 6.2.2007, folio no. 6170. Further, this is also confirmed by a number of Ryanair's internal documents: the Board Paper No 8 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629), page 8.2: "[...]*"; Board Paper No 5 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629): "[...]*"; Board Paper No 3 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629): [...]*; Board Paper No 3 for the Board Meeting of Ryanair Holding plc on [...]*, page 3.3 (folio no. 629), stating with regard to routes from Cork/Shannon/Knock to London as follows: "[...]*."

635 See easyJet's e-mail of 16 May 2007.

636 The Commission has based this and the following graph on frequency and fare data which have been submitted by DAA and Ryanair, respectively.

637 In its response to the Statement of Objections, Ryanair points out that just before easyJet's entry prices were already below €50 (p.244). However, as can be seen, this is due to the seasonal effect. Moreover, presumably easyJet had already taken its entry decision a few months before.

638 The frequency increase was accompanied by various promotions. For example, Ryanair announced in October 2005 that the price of seats on all its routes would start from 29 pence (excluding taxes and other charges) for the period 3 November 2005 to 9 February 2006.

639 Response to the Statement of Objections, p. 243.

640 On the assumption that there are no other changes on that particular route (as demand shifts etc).

641 See Ryanair's Response to the Statement of Objections, paragraph 191.

642 As can be seen from Graph 3, Ryanair's post-exit frequency and price level do not show an overall improvement, as they only correspond to its frequency and price levels in the period 2002-2003.

643 The fare and capacity figures in this and following paragraphs are based on a reply from Ryanair of 8 December 2006 to the Commission questions of 6 December 2006. 644 See for a confirmation of the dates related to easyJet's Irish operations e-mail sent by easyJet on 16 May 2007.

645 In its Response to the Statement of Objections (p. 244) Ryanair claims that this announcement was made before easyJet announced that it would start flying from Gatwick to Shannon.

646 easyJet then operated 2 daily frequency in Summer 2005 and again 1 in Winter 2005/6 and Summer 2006.

647 Ryanair introduced various promotions, including an offer of 20 000 'free' seats (excluding taxes) on SNNUK routes, including SNN-LGW, LTN and STN, in conjunction with announcing SNN as its 12th European base. http://www.ryanair.com/site/EN/news.php?yr=05&month=apr&story=pro-en-.

648 This is also confirmed by a number of Ryanair's internal documents: Board Paper No 3 for the Board Meeting of Ryanair Holding plc on [...]*, page 3.3 (folio no. 629), stating with regard to routes from Cork/Shannon/Knock to London: "[...]*"; Board Paper No 8 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629), page 8.2: "[...]*"; Board Paper No 5 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629): "[...]*"; Board Paper No 3 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629): [...]*.

649 Ryanair also mentions that Aer Lingus' fare reductions were equally "aggressive" on the Shannon-Heathrow route in response to easyJet's entrance. While this could not be verified, a similar behaviour by Aer Lingus would only strengthen the concern that the merged entity's reputation would constitute an entry barrier.

650 Submission of the DoT of 7 February 2007, p. 29, folio no. 6230. In its Response to the Statement of Objections, Ryanair claims that easyJet made its announcement to withdraw from this route only in September 2006.

651 This is also confirmed by a number of Ryanair's internal documents: Board Paper No 3 for the Board Meeting of Ryanair Holding plc on [...]*, page 3.3 (folio no. 629), stating with regard to routes from Cork/Shannon/Knock to London: "[...]*"; Board Paper No 8 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629), page 8.2: "[...]*"; Board Paper No 5 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629): "[...]*"; Board Paper No 3 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629): [...]*.

652 See for a confirmation of the dates related to MyTravelLite's Irish operations e-mail sent by MyTravel on 15 May 2007, folio no. 9283.

653 See also DOT, submission of 24.11.2006, "New entry to compete with Ryanair on any route post-merger is unlikely", folio no. 2333, page 2.

654 See for a confirmation of the dates related to GOFly's Irish operations e-mail sent by easyJet on 16 May 2007, folio no. 9859.

655 According to the DOT, by adding extra flights on the weekend. DOT, submission of 7.2.2007, folio no. 6230,page 10.

656 See http://archives.tcm.ie/irishexaminer/2001/07/14/story7930.asp.

657 Quote taken from the Barbara Cassani's monography "Go - an airline adventure", London 2003, p. 260; In its Response to the Statement of Objections, Ryanair points out that Ms Cassani had also acknowledged that Ryanair's cost were substantially lower than GoFly's. As regards this episode see also the bulletin from the Strategic Aviation Special Interest Group7/2001, page 5: "Ryanair has announced that it is to add more capacity on its routes from Dublin to Edinburgh and to Glasgow to rival its competitor GO, a bid which will undoubtedly fuel a price war with GO. Ryanair will be adding 20% capacity on the Dublin-Edinburgh route and will be charging 37p each way on the route excluding taxes during its period of special fare promotions", see http://www.sasig.org/pdfs/bullet2001/July01.pdf..

658 See Ryanair's Response to the Statement of Objections, p. 259.

659 See Ryanair's Response to the Statement of Objections, p. 259.

660 http://rte.ie/business/2001/0712/go.html

661 See DOT, submission of 24.11.2007, "New entry to compete with Ryanair on any route post-merger is unlikely", folio no. 2333, page 4.

662 On the Dublin-Paris route, CityJet (flying with small aircraft) has more frequencies, but still only 20% of the passengers, while the Merging Parties hold 80%. See in more detail below.

663 See interview with Aer Arann of 13.2.2007, folio no. 6170.

664 Horizontal Merger Guidelines, paragraph 69: "For entry to be likely, it must be sufficiently profitable taking into account the price effects of injecting additional output into the market and the potential responses of the incumbents."

665 See e.g. interview with easyJet of 15.2.2007, folio no. 6170.

666 See minutes of interviews with competitors, folio no. 6170 and 8091.

667 See Section 7.8.4.

668 See interview with CityJet of 21.2.2007, folio no. 6170: "[Ireland is] "a drop in the Ocean"; interview with easyJet of 15.2.2007

669 See e.g. interview with Air Berlin of 13.2.2007, folio no. 6170: "There s currently a number of better possible routes in Europe for Air Berlin to enter with better yield and profit prospects than Irish routes, for example routes to Russia or Scandinavia"

670 See e.g. interview with easyJet of 15.2.2007, folio no. 6170.

671 See e.g. interview with Air Berlin of 13.2.2007, folio no. 6170: "[Air Berlin has a] strategy to focus its operations to regions with a large catchment area, which allows to connect traffic from other points to this destination."

672 See replies to Question 38 of the Questionnaire to Competitors of 06.11.2006.

673 Notification, page 104.

674 Passengers can also be transported to their aircraft by bus; however this causes an additional delay and is not appreciated in particular by airlines focussing on business passengers, see e.g. interview with AerArann of 13.2.2007, folio no. 6170, and interview with CityJet (Air France)of 21.2.2007, folio no. 6170.

675 Also the term "Air Transport Movements" ("ATMs") is used, see e.g. Constitution of the Dublin Airport Coordination Committee.

676 OJ L 14, 22.1.1993, p. 1-6, as last amended by Regulation (EC) No 793-2004 of the European Parliament and of the Council of 21 April 2004 amending Council Regulation (EEC) No 95/93 on common rules for the allocation of slots at Community airports (OJ L 138, 30.4.2004, p. 50-60).

677 See e.g. cases M.3940 - Lufthansa/Eurowings, M.3770 -Lufthansa/Swiss; M.3280 - Air France/KLM.

678 See e.g. replies to questions 35 and 42 of the questionnaire to competitors of 6.11.2007.

679 See replies to Question 7, Questionnaire to airports, 09.11.2006; see also the statements by third parties summarised in the report "Dublin Airport Capacity Review" of 6.12.2006, prepared by Jacobs Consultancy for the Irish Commission for Aviation Regulation (the "Jacobs Report").

680 See RBB Paper of 20.2.2007, page 13.

681 E.g. a criterion for the definition of a "delay" of only 8 minutes, see RBB Paper of 20.2.2007, page 12.

682 Runway 11/29 is too short for Boeing or Airbus aircraft. It can only accommodate regional aircraft or planes designed specifically for short runway operations such as the BAe146/Avro RJ 85- 100.

683 See minutes of the meeting with Ryanair at Dublin of 6.2.2007; RBB Paper of 20.2.2007, page 13.

684 It should be noted that the CAR had had already designated Dublin Airport as a coordinated airport in April 2005 with effect from summer 2006. This decision, however, was the subject of a judicial review by Ryanair and was declared invalid in first instance; the procedure is still pending.

685 See http://www.aviationreg.ie/images/ContentBuilder/2007-02-13%20cp3%202007%20final.pdf.

686 See press Statement of the CAR of 12.2.2007: http://www.aviationreg.ie/images/ContentBuilder/2007-02- 12%20press%20release%20summer%2007%20designation%20of%20dublin%20airport.pdf.

687 See Jacob's Report.

688 See submission of Irish DOT of 2.2.2007 ("Impact of Ryanair's dominance at Dublin Airport"), folio no. 6444.

689 CAR decision of 12.2.2007, pages 21 et seq.; Jacobs Report, page 62.

690 See interview with Aer Arann of 13.2.2007, folio no. 6170.

691 The CAR refers also to a study carried out by the air traffic consultant NATS for the DAA.

692 Jacobs Report, page 58.

693 6 of 25 competitors responding to the Commission's market investigation in relation to this issue did not mention "congestion problems" or "missing airport capacity at Dublin" when they were asked to describe potential barriers to entry. Of these, only two of these 6 operators actually fly to Dublin. See also interview with CityJet, folio no. 6170: "Dublin Airport is "completely congested" according to CityJet between 6am- 9am and between 4pm- 7pm. CityJet does not consider the second smaller runway to be an alternative. Also due to the current capacity situation and lack of fair regulation at the Dublin Airport, CityJet recently announced its expansion out of London City airport to other cities in Europe (such as Madrid, Milan Linate, Geneva, Zurich, Nice, Belfast City). Operating these services from DUB would not be possible due to capacity situation and insufficient access to facilities such as contact stands."

694 Jacobs report, page 55.

695 See minutes from interviews with Aer Arann and CityJet, folio no. 6170.

696 See Section 7.9.

697 See Ryanair's Response to the Statement of Objections, page 249.

698 See e.g. replies to Question 5 of the Questionnaire to Airports of 09.11.2006.

699 The estimated percentage of business/time-sensitive passengers is between 20% and 30% on these routes.

700 See recital 36 of the Horizontal Merger Guidelines [Merged entity able to hinder expansion by competitors] "Some proposed mergers would, if allowed to proceed, significantly impede effective competition by leaving the merged firm in a position where it would have the ability and incentive to make the expansion of smaller firms and potential competitors more difficult or otherwise restrict the ability of rival firms to compete. [...]"

701 See e.g. answers to question 49 of the questionnaire to competitors sent on 6.11.2006; see also interview with Aer Arann of 13.2.2007, folio no. 6170; interview with CityJet (Air France) of 21.2.2007, folio no. 6170; interview with Jet2.com of 1.2.2007, folio no. 6170; interview with Air Berlin of 13.2.2007, folio no. 6170; see also the Irish DoT's submission on "Impact of Ryanair's Dominance at Dublin Airport on the Development of the Airport and on Competition" of 2.2.2007, folio no. 6444.

702 Answers to question 49 of the questionnaire to competitors sent on 6.11.2006

703 See in this context also interview with Aer Arann of 13.2.2007, folio no. 6170.

704 See Board Paper 1 for the Board Meeting of Ryanair Holdings plc on [...]*, page 1.4-1.5 (folio no. 629).

705 See Board Paper 1 for the Board Meeting of Ryanair Holdings plc on [...]* (folio no. 629), page 1.5. stating as follows: "[...]*" See also Board Paper 1 for the Board Meeting of Ryanair Holdings plc on [...]*, page 1.7 (folio no. 629).

706 See e.g. interviews with Aer Arann of 13.2.2007, folio no. 6170, or with CityJet of 21.2.2007, folio no. 6170. CityJet stated in this respect: "Taking into account the focus of CityJet on high-yield business passengers who expect for the higher fares a certain level of services and comfort, this inconvenience for the passengers is particularly harmful for CityJet."

707 See recital 71 of the Horizontal Merger Guidelines, "Barriers to entry can take various forms:

(a) [...]

(b) The incumbents may also enjoy technical advantages, such as preferential access to essential facilities, [...] which make it difficult for any firm to compete successfully."

708 See answers to question 49 of the questionnaire to competitors sent on 6.11.2006; the Irish DoT's submission of 2.2.2007, folio no. 6444; see also interviews with Jet2.com of 1.2.2007, folio no. 6170: "...the merged entity's strong position on the Dublin Airport could also give rise to concerns, given the influence on flight times, landing rights and influence on decisions on the airport in general."; interview with Aer Arann of 13.2.2007, folio no. 6170: "Ryanair/Aer Lingus could use its dominance as an airport customer at the Irish airports to make entry for competitors even more difficult (...)"; interview with Air Berlin, folio no. 6170: "[Air Berlin] expressed concerns that the merged entity could use its strength as a customer at Irish airports to cause these airports to discriminate other competitors as concerns airport charges, i.e. more favourable fees than its competitors."; see also interview with CityJet (Aer France) of 21.2.2007. . interview with CityJet (Air France)of 21 February 2007, folio no. 6170: "...allowing them even more to dominate Dublin Airport policy".

709 It should, however, be noted that Ryanair's reference to the situation at Stansted is not able to mitigate the Commission's concerns in this regard.

710 See section 33 of the Irish Aviation Regulation Act 2001 as amended by section 22 of the Irish State Airports Act 2004:

"(1) In making a determination the objectives of the Commission are as follows-

(a) to facilitate the efficient and economic development and operation of Dublin Airport which meet the requirements of current and prospective users of Dublin Airport,

(b) to protect the reasonable interests of current and prospective users of Dublin Airport in relation to Dublin Airport (...)" (text highlighted by the Commission).

711 Air Transport Movements ("ATMs") are defined as "the scheduled time of arrival or departure available or allocated to an aircraft movement on a specific date at Dublin", see section 11 of the Constitution.

712 See notably 3.1.1. and 3.1.2 of the Constitution.

713 See e.g. answers to question 49 of the questionnaire to competitors sent on 6.11.2006; see also interview with Aer Arann of 13.2.2007, folio no. 6170; interview with CityJet (Air France) of 21.2.2007, folio no. 6170; interview with Jet2.com of 1.2.2007, folio no. 6170; interview with Air Berlin of 13.2.2007, folio no. 6170.

714 Section 7.8.9.1 -7.8.9.11.

715 See minutes with CityJet (Air France) of 21.2.2007 ("CityJet must achieve a higher average yield than Ryanair/Aer Lingus"); according to the presentation of Morten, Beyer and Agnew, Inc. made at the Morgan Stanley Conference - Regional Jet Market Update in March 2004, slide 18, the operating unit costs (in terms of USD per ASM (air seat mile) of BAe 146 are around twice as those of the aircraft used by Ryanair and Aer Lingus. Even though according to other slides of the presentation the total operating costs per departure (slide 15) and block hour costs (slide 16) are lower for BAe 146, it is the unit cost per seat and mile which is most relevant from the point of view of the final fare paid by the customer and thus for the competitiveness of the airline. The same presentation (see slides 19-25) also acknowledges that regional jet carriers face a threat of the low-cost carriers and that as a result of that many of the regional jet operations may become unprofitable. See: http://www.mba.aero/presentations/040301_regional_jet_market_update.pdf, folio no. 13020.

716 Source: CityJet responses to questions 2 and 3, Competitor questionnaire of 15.11.2006, folio no. 1990.

717 See interview with CityJet (Air France)of 21 February 2007, folio no. 6170.

718 Source: London City Airport response to question 16, airport substitutability questionnaire.

719 The Commission does not contest that CityJet (Air France) has indicated to be interested to expand its London-Dublin route if attractive slots were made available; however, CityJet indicated that "entry on other routes, notably continental routes, would be further down the list of priorities for CityJet.". This is because "Ireland is not a very attractive market to start operations [since] average yields are quite low." Furthermore, CityJet points inter alia at the difficulty of obtaining contact stands as a further entry deterrent for CityJet, see minutes of the interview with CityJet (Air France) of 21 February 2007, folio no. 6170.

720 Ibid.

721 Source: Telephone interview with Aer Arann, 13 February 2007, folio no. 6170.

722 Source: idem

723 See RBB document "Ryanair/Aer Lingus: Position Paper on Barriers to Entry", of 20 February 2007, folio no. 4135, page 29.

724 Even though both Aer Arann and Ryanair operate currently between Cork and Shannon, it was Ryanair who entered in November 2005 in competition with existing operations of Aer Arann.

725 "To provide our customers with safe, good value, point-to-point air transport services. [...]", Source: http://www.easyjet.com/EN/About/index.html, 02 March 2007.

726 Source www.easyJet.com/EN/About/Information/index.html, 02 March 2007.

727 Response to the Statement of objections, page 251.

728 See Board Paper 3 for the Board Meeting of Ryanair Holdings plc on [...]*, page 3.6. (folio no. 629).

729 See RBB document "Ryanair/Aer Lingus: Position Paper on Barriers to Entry", of 20 February 2007, folio no. 4135, page 29.

730 See Section 6.3.4.21.

731 Source: telephone interview with easyJet, 15 February 2007, folio no. 6170.

732 See reply of SkyEurope of 6 December 2006 to the Questionnaire on proposed commitments, folio no. 2596, questions 2 and 12.

733 See in detail Section 8.2.2.1.

734 See Board Paper 3 for the Board Meeting of Ryanair Holdings plc on [...]* (folio no. 629), page 3.4. which states as follows: [...]*.

735 See Board Paper No 3 for the Board Meeting of Ryanair Holdings plc on [...]*, page 3.7 (folio no. 629).

736 See Board Paper 3 for the Board Meeting of Ryanair Holdings plc on [...]*, page 3.6. (folio no. 629).

737 See response to the Statement of Objections, page 251.

738 See e.g. response to questions 3 and 4 of the Market Test of the Final Commitments.

739 Source: http://www.britishairways.com/travel/bapress/public/en_gb. On 5 March 2007, British Airways announced in the form of a press release that it had completed the sale of BA Connect to FlyBe.

740 See press release of British Airways "Proposed Sale of BA Connect to Flybe" of 03/11/2006, Ref 108/SR/06, available at the website: http://www.britishairways.com/travel/bapress/public/en_gb.

741 Source: Air Berlin responses to questions 2 and 3, Competitor questionnaire of 15.11.2006

742 Source: http://www.air-berlin.com

743 Interview with Air Berlin, 13 February 2007, folio no. 6170.

744 Interview with Air Berlin, 13 February 2007, folio no. 6170.

745 See Market Test, response to questions 3 and 4.

746 See Board paper No 3 for the Board Meeting of Ryanair Holdings plc on the [...]*, page 3.5 (folio no. 629).

747 Source: bmi responses to questions 2 and 3, Competitor questionnaire of 15.11.2006, folio no. 22283.

748 It is noteworthy that, in addition to the services it operates with its own aircraft, bmi also offers several services to/from London Heathrow airport through code share agreements. This is in particular the case on routes to/from points located in Germany or Scandinavia, the home market of its two airline shareholders.

749 Source: bmi responses to question 2, Competitor questionnaire of 15.11.2006, folio no. 22283.

750 Source: bmi responses to question 16, Competitor questionnaire of 15.11.2006, folio no. 22283.

751 bmi also offers connecting flights (via London Heathrow) between Dublin and some provincial cities in Great Britain which are served by Ryanair on the basis of a direct service from Dublin. As explained in section "5.5. Joint market for direct flights and indirect flights?" above however, indirect services are considered to be in a distinct market for the present purposes.

752 Source: bmi response to question 50, Competitor questionnaire of 15.11.2006, folio no. 22283.

753 Source: bmi response to question 33, Competitor questionnaire of 15.11.2006, folio no. 22283.

754 See bmi/bmi regional/bmibaby reply of 26 February 2007 to follow-up questions, folio no. 4348, question 3.

755 See bmi/bmi regional/bmibaby reply of 26 February 2007 to follow-up questions, folio no. 4348, question 8. See also bmi's reply to the Market Test, question 3.

756 See bmi response to question 3, Competitor questionnaire of 15.11.2006, folio no. 22283.

757 See minutes of the interview with bmi/bmibaby of 9 March 2007, folio no. 6170, and BMI's response to the Commission's Market Test of the Final Commitments.

758 See minutes of the interview with bmi/bmibaby of 9 March 2007, folio no. 6170.

759 Source: bmibaby responses to questions 2 and 3, Competitor questionnaire of 15.11.2006, folio no. 22283.

760 Source: http://www.bmibaby.com

761 See reply of BMI of 17 November 2006 to the Questionnaire to competitors, folio no. 22283, question 34

762 See bmi/bmi regional/bmibaby reply of 26 February 2007 to follow-up questions, folio no. 4348, question 4.

763 See bmi/bmi regional/bmibaby reply of 26 February 2007 to follow-up questions, folio no. 4348, question 3.

764 See bmi/bmi regional/bmibaby reply of 26 February 2007 to follow-up questions, folio no. 4348, question 3.

765 See bmi/bmi regional/bmibaby reply of 26 February 2007 to follow-up questions, folio no. 4348, question 8.

766 See minutes of the interview with bmi/bmibaby of 9 March 2007, folio no. 6170, and BMI's response to the Commission's Market Test of the Final Commitments.

767 See minutes of the interview with bmi/bmibaby of 9 March 2007, folio no. 6170.

768 Ibid.

769 Decision of the UK Office of Fair Trading in the case Flybe/BA Connect published 15 February 2007 http://www.oft.gov.uk/advice_and_resources/resource_base/Mergers_home/decisions/2007/Flybe.

770 See minutes from the interview with Flybe of 12 March 2007, folio no. 6170.

771 See minutes from the interview with Flybe of 12 March 2007, folio no. 6170.

772 See minutes of the interview with Flybe of 12 March 2007, folio no. 6170, and Flybe's response to the Commission's Market Test of the Final Commitments.

773 See reply of Flybe of 11 December 2006 to Questionnaire to Competitors, question 33 and minutes from the interview with Flybe of 12 March 2007, folio no. 6170.

774 See minutes of the interview with Clickair of 26 March 2007, folio no. 8091.

775 See http://www.clickair.com/view/default.aspx?lang=2&menu=81&pagina=0, folio no. 13021.

776 See: www.clickair.com, folio no. 13021.

777 See: http://grupo.iberia.es/portal/site/grupoiberia/menuitem.8d4d9fb9661bc259f54c0f10d21061ca/.

778 See minutes of the interview with Clickair of 26 March 2007, folio no. 8091.

779 See minutes of the interview with Clickair of 26 March 2007, folio no. 8091.

780 Clickair indicated that it would also consider opening a route to "North of Spain".

781 See reply to the Market Test, answers to questions 3 b) and d) and to question 6 a).

782 See minutes of the interview with Clickair of 26 March 2007, folio no. 8091.

783 Idem.

784 Idem.

785 Even though Aer Lingus still have some corporate contracts, they represent less than 5% of the total turnover of Aer Lingus - see Aer Lingus' e-mail of 27 March 2007, folio no. 6350.

786 See in particular replies to Market Testing of Proposed Remedies - Competitors sent on 30 November 2006, questions 5, 6 and 7.

787 See Ryanair's route analysis spreadsheet sent by e-mail on 02/02/2007, last column, folio no. 2620.

788 See Ryanair's late bookings data sent by e-mail on 12/02/2007, folio no 10928.

789 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 468, table 19 and data provided in the analysis of individual routes.

790 See Ryanair's late bookings data sent by e-mail on 12/02/2007, folio no 10928.

791 According to the Ryanair's Annual Report for the year ended 31 March 2006, the average load factor was 83% (see Form CO, Annex 16, page 10).

792 In its e-mail on 12/02/2007, folio no 10928, Ryanair gives an example that "a move from 70% to 74% registered for a full month on the basis of bookings taken in the last 10 days of the month represents 12% (4% x 3) of the applicable flights on which the bookings were taken". If it is presumed that the final load factor of 74% is spread more or less evenly throughout the month, then the share of passengers booking in the last 10 days of the month on the basis of these data would be around 16% of all actual passengers on these flights, and not 12% which relates to all seats (even though around 26% of them were not occupied).

793 It should be noted that the Commission specifically asked Ryanair for the data on bookings 7 days or less before departure (see e-mail "M.4439 - Ryanair/Aer Lingus - follow-up issues" sent to Ryanair on 30.01.2007, folio no. 5232).

794 See press release of DAA "More than 2.7m extra passengers at Dublin Airport in 2006" of 24 January 2007, available at: http://www.dublinairport.com/about-us/media-centre/press-releases/2006_pax_figures.html, folio no. 13020.

795 See Form 20F for the US Securities and Exchange Commission filed on 27 September 2006, page 19 (see reply by Ryanair to the Commissions' request for information of 22 December 2006, Annex 2.7, folio no. 629). Similar statements are also in the Form 20F submissions to the US SEC in the previous years (see folio no. 629). Ryanair argues that the Form 20F is not an anti-trust law document and cannot substitute for an antitrust analysis made in the Notification. However, the Statement in the Form 20F clearly demonstrates that Ryanair regards Aer Lingus as its main competitor on the Ireland - UK routes and systematically compares its performance (market share) with that of Aer Lingus.

796 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

797 See Ryanair's route analysis spreadsheet sent by e-mail on 02/02/2007, folio no. 2620.

798 See Ryanair's Response to the Statement of Objections of 17 April 2007.

799 See Section 6.3.

800 Even though Aer Lingus has a code-share agreement for this route with British Airways and American Airlines, it is only limited to passengers connecting at Heathrow (see reply of Aer Lingus of 20/11/2007 to Additional Questionnaire, folio no. 4122, question 15). Therefore, even though British Airways and American Airlines sell seats on Aer Lingus flights to Heathrow as part of their (usually long-haul) routes (e.g. Dublin - New York via Heathrow), they do not sell any point-to-point tickets between Dublin and Heathrow on Aer Lingus flights and thus cannot be regarded as competitors on the point-to-point market on the Dublin - London route (apart from British Airways' own flights to London Gatwick mentioned above)

801 Ryanair carried no connecting passengers while the share of connecting passengers on Aer Lingus flights to Heathrow in summer 2006 was around [20-30%]* of total carried passengers, see response of Aer Lingus of 20/11/2006 to Questionnaire to Competitors, folio no. 4122.

802 See CFI, case T-210-01 case - General Electric v Commission, of 14 December 2005, ECR II-5575, paragraph 115: "Furthermore, although the importance of market shares may vary from one market to another, very large shares [of over 50%] are in themselves, and save in exceptional circumstances, evidence of the existence of a dominant position." (text in brackets added by the Commission). See also the Horizontal Merger Guidelines, OJ C 31, 5.2.2004, page 5, paragraph 17

803 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see e.g. Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 ff). However, as discussed in more detail above, the two carriers do compete on the same market (see Section 6.) and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Further, the fares of Aer Lingus for individual routes based on MIDT/OAG data provided by Ryanair in the Form CO cannot be considered as reliable estimates of the actual fares (see paragraph 796 and following below).

804 Aer Lingus carried on this route around [20-30%]* connecting passengers, see response of Aer Lingus of 20/11/2006 to Questionnaire to Competitors, folio no. 4122.

805 CityJet uses 93-seats BAe 146 aircraft. Ryanair argues in its Response to the Statement of Objections of 17 April 2007 that these aircraft are perfectly appropriate for the current regional operations of Ryanair and Aer Lingus. However, CityJet itself indicated that "Given the significantly lower capacity of its planes, CityJet must achieve a higher average yield than Ryanair/Aer Lingus to be profitable - this is secured by focusing on connecting passengers and business passengers at key times of the day (emphasis on schedule and product)." (see minutes of the interview with CityJet of 21/02/2007, folio no. 6170). This clearly indicates that CityJet is not able with its smaller aircraft to compete for the same passengers as Ryanair and Aer Lingus in particular as regards the more non-time sensitive point-to-point passengers (predominantly leisure) which provide for more than half of the point-to-point passengers of both Ryanair and Aer Lingus (see paragraph 799 below).

806 This was also confirmed by the Customer Survey commissioned by the Commission, according to which 72.4% of passengers who fly to London City airport indicated as one of the reasons for choosing that particular flight the fact that City airport is closest to their final destination. In the case Heathrow passengers, this figure was only 31.8%.

807 According to the Customer Survey commissioned by the Commission, the share of business travellers to London City was around 60%.

808 According to the CAA Departing Passenger Survey for 2005 available at http://www.caa.co.uk, folio no. 9782, 21% of scheduled passengers between Stansted and Ireland, 18% between Luton and Ireland and 19% between Gatwick and Ireland were business passengers. Even though in case of Gatwick, these figures also include the passengers of British Airways and easyJet (which at that time operated between Irish regional airports and London), it can be assumed on the basis of the figures for Stansted and Luton served only by Ryanair that around 20% of Ryanair passengers are business. It is also confirmed by the fact that that around 20% of bookings on Ryanair flights from/to Stansted and Luton and 15% of bookings from/to Gatwick are made within the seven last days before the flight. These figures were largely confirmed by the Customer Survey commissioned by the Commission, even though it indicated lower share of business passengers for Stansted (14%) but on the contrary higher for Luton (28%). Even though Ryanair point in their Response to the Statement of Objections of 17 April 2007 (paragraph 704) to the Statement of Manchester Airport that the CAA surveys are based on limited passenger sampling and could for individual destinations be less reliable, it can still be considered as a useful independent source of information providing, in the absence of any other clear data and if interpreted in light of other available information from the parties and other sources, at least a proxy for the actual situation.

809 According to the CAA Departing Passenger Survey for 2005 available at http://www.caa.co.uk, folio no. 9782, 36% of scheduled passengers between Heathrow and Ireland were business passengers. As only around [15-30%]* of bookings on Aer Lingus flights from/to Heathrow are made within the last seven days before the flight (based on data by Aer Lingus of 02/02/2007, folio no. 2935), which is comparable to the Ryanair figure for Stansted and Luton, the share of business passengers of Aer Lingus may actually be lower than 36% and that the balance of business passengers for Heathrow is accounted by BMI. This would also be in line with the business models of these two carriers described above.

810 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 653.

811 See Aer Lingus Annual Report 2005, Form CO, Annex 17.

812 According to the data submitted by Ryanair in the Form CO, the frequencies offered by Aer Lingus as well as its competitors remained largely stable in this period and the fares charged by Ryanair and other competitors (which are full-service network carriers for which MIDT/OAG data may be more reliable) in the same period even decreased or did not increase in any such magnitude (as can be seen from the graph in Table 21 on page 151 of the Ryanair's Response to the Statement of Objections of 17 April 2007).

813 See in particular Board Paper No3 for the Board Meeting of Ryanair Holdings plc on [...]* (folio no. 629), page 3.2. in section 3.3.1. - Dublin to London: "[...]*". Similarly, see Board Paper No3 for the Board Meeting of Ryanair Holdings plc on [...]* (folio no. 629), page 3.4.

814 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraphs 669 ff and table 22.

815 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 675.

816 See Table 22 on page 152 of Ryanair's Response to the Statement of Objections of 17 April 2007.

817 It should be noted that Ryanair, unlike Aer Lingus, offers exclusively point-to-point connections. When it refers to "connecting" passengers, these passengers purchase two tickets separately. For the purpose of this decision, these passengers are regarded as "point-to-point" passengers, since connecting passengers pursuant to the wording of this decision do not buy two tickets for the different sectors of their route but just one "connecting" ticket, see Section 6.3.

818 This figure is approximate and based solely on the data provided by Ryanair in the Form CO according to which Aer Lingus offered in Summer 2005 a weekly one-way capacity of 17.327 seats. With 31 weeks of the Summer 2005 season (from 27/03/2005 to 30/10/2005) this would give around 1.074.274 seats per season two-ways. Considering the share of connecting passengers of 26.8% (see Table 22 on page 152 of Ryanair's Response to the Statement of Objections of 17 April 2007) and presuming purely hypothetically a load factor of Aer Lingus of 85% (Ryanair was not able to provide Aer Lingus' load factor but it is presumed that it would not be significantly higher than Ryanair's load factor ranging between 80 and 86%), this would give around 245.000 connecting passengers of Aer Lingus in the Summer 2005 season.

819 Of all passengers using Aer Lingus, 18% considered also Ryanair as an alternative while only 9% considered any other of all the remaining competitors. In case of Ryanair, 45% of passengers travelling to Gatwick, 30% of passengers to Stansted and 25% of passengers to Luton considered also Aer Lingus as an alternative, while only 26%, 12% and 2% respectively consider any other of the remaining carriers as an alternative. Ryanair argues that the Customer Survey's results are "significantly skewed" in case of the flights of Aer Lingus to London Heathrow where the survey was conducted on weekends (see Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 70, first indent). However, as indicated above it is unlikely that the bias (if any) would be "significant" (see Section 6.3.4. - London airports). Further, this does not affect at all the high shares of Ryanair's customers considering Aer Lingus as an alternative. See also Annexes I and II.

820 See Minutes of the Board Meeting of Ryanair Holdings plc held on [...]* (presented as Board Paper No1 for the Board Meeting of Ryanair Holdings plc on [...]*), page 1.3 (folio no. 629)

821 See Minutes of the Board meeting of Ryanair Holdings plc held on [...]* (presented as Board Paper No 1 for the Board Meeting on [...]*), page 1.3. (folio no. 629)

822 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 677, first indent.

823 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 677, second indent.

824 See e.g. the introduction of the check-in luggage charge at the beginning of 2007.

825 See Ryanair's e-mail of 08/11/2006.

826 Although neither Aer Lingus nor Ryanair discriminate between different customer groups they price higher closer to departure. For a more detailed explanation of the parties' yield management see Section 7.4.

827 See reply of BAA of 17/11/2006, folio no. 22229, and of Airport Coordination Ltd of 17/11/2006, folio no. 22175, to the Questionnaire to Airports.

828 See Section 6.4.

829 The term prices increase for the purpose of this section on route-by-route analysis should be understood (in line with paragraph 8 and footnote 7 of the Horizontal Merger Guidelines, OJ C 31, 5.2.2004, page 5) as encompassing also reduction of output, choice or quality, lower innovation or lower decrease of prices than it would otherwise be the case without the merger.

830 In fact the only scheduled routes of Globespan from Stansted are Stansted - Tenerife and Stansted -Lamaca (Cyprus), see www.flyglobespan.com, folio no. 13020.

831 For more details, see Section 7.8.9. concerning abilities and incentives of competing airlines to enter any of the overlap routes.

832 See Section 7.8.3.4.

833 Air Berlin uses Germany based aircraft to serve some London - UK destinations (Manchester, Glasgow, Cardiff) by flying in the morning from German bases to Stansted, then to the UK destinations and back and in the evening back to Germany.

834 According to the CAA Departing Passenger Survey for 2005 available at http://www.caa.co.uk, folio no. 9782, around 64% of passengers flying between Heathrow and Ireland, 50% of passengers between Gatwick and Ireland, 56% of passengers between Luton and Ireland and 46% of passengers between Stansted and Ireland were non-UK passengers. The data provided by the parties confirm an important proportion of Irish originating passengers (Ryanair [30-40]*% for Stansted, [40-50]*% for Gatwick and [30-40]*% for Luton, Aer Lingus [45-60%]* for Heathrow - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965). According the Irish DOT, [30-40%]* of passengers on this route originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

835 In view of Aer Arann's operated smaller aircraft and business model, it would anyway not be a close competitor to Ryanair and Aer Lingus for the similar reasons as CityJet.

836 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

837 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

838 See Ryanair's Response to the Statement of Objections of 17 April 2007.

839 See Sections 6.3.

840 In its reply of 26 February 2007 to the follow-up questions, folio no. 4348, question 4, BMI noted with respect to the withdrawal from East Midlands - Dublin route the following: "The decision to withdraw the service was a commercial decision because the route no longer produced a viable economic return in direct competition with Ryanair's services at the time. bmibaby's load factor and overall yields was no longer sustainable to continue scheduled operations."

841 See reply of BMI of 17 November 2006 to the Questionnaire to Competitors, question 34, folio no. 22283.

842 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

843 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see e.g. Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 ff). However, as discussed in more detail above, the two carriers do compete on the same market (see Section 6.) and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Further, the fares of Aer Lingus for individual routes based on MIDT/OAG data provided by Ryanair in the Form CO cannot be considered as reliable estimates of the actual fares (see paragraph 796 and following).

844 Ryanair in its Response to the Statement of Objections of 17 April 2007 argues that thickness of the route is not important as all that is required is that entry replaces lost capacity. However, this assumption is incorrect as in order to consider a potential entry as a mitigating factor against significant impediment to competition, the new entrant would need to provide sufficient competitive constraint on the merged entity. In case the merged entity operates a large number of frequencies on a thick route like Dublin - Birmingham or other thick routes analysed in this section, a small-scale entry is unlikely to provide such sufficient competitive constraint. Further, this presumption would also mean that in case there is no lost capacity after the merger (Ryanair and Aer Lingus continue to operate the same capacities), there would be no need for any entry to eliminate competition concerns which is clearly incorrect.

845 According to the CAA Departing Passenger Survey for 2003 (neither Birmingham nor East Midlands were included in the CAA surveys for 2004 or 2005) available at http://www.caa.co.uk, folio no. 9782, 23% of scheduled passengers between Birmingham and Ireland and 27.5% between East Midlands and Ireland were business passengers. It is also noted that Ryanair itself serves a significant number of more time-sensitive customers on this route, as around [15-30]*% of bookings on Ryanair flights on this route are made within the seven last days before the flight. Further, [15-30%]* of Aer Lingus passengers book within the last seven days on this route (based on data by Aer Lingus of 02/02/2007, folio no. 2935).

846 According to the CAA Departing Passenger Survey for 2003 available at http://www.caa.co.uk, folio no. 9782, around 30% of passengers flying between Birmingham/East Midlands and Ireland were non-UK passengers. The data provided by the parties indicate currently even higher proportion of Irish originating passengers (Ryanair [40-50]*% for Birmingham and [30-40]*% for East Midlands, Aer Lingus [30-45%]* based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965). According the Irish DOT, [50-60%]* of passengers to BHX and [60-70%]* of passengers to EMA originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

847 See reply of CityJet to the Questionnaire to Competitors sent on 6 November 2006, folio no. 1990 and minutes of the interview with CityJet of 21 February 2007, folio no. 6170.

848 See minutes of the interview with CityJet of 21 February 2007, folio no. 6170.

849 See minutes of the interview with CityJet of 21 February 2007, folio no. 6170.

850 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 520.

851 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

852 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

853 See Ryanair's Response to the Statement of Objections of 17 April 2007.

854 For more details see section 7.8.5.

855 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

856 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see e.g. Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 ff). However, as discussed in more detail above, the two carriers do compete on the same market (see Section 6.) and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Further, the fares of Aer Lingus for individual routes based on MIDT/OAG data provided by Ryanair in the Form CO cannot be considered as reliable estimates of the actual fares (see paragraph 796 and following).

857 According to the CAA Departing Passenger Survey for 2005 available at http://www.caa.co.uk, folio no. 9782, 27.8% of scheduled passengers between Edinburgh and Ireland were business passengers. It is also noted that Ryanair itself serves a significant number of more time-sensitive customers on this route, as around [10-20]*% of bookings on Ryanair flights on this route are made within the seven last days before the flight. Further, [15-30%]* of Aer Lingus passengers book within the last seven days on this route (based on data by Aer Lingus of 02/02/2007, folio no. 2935).

858 Ryanair argued that also BMI has 4 aircraft based in Edinburgh, however this was not confirmed by the information obtained from BMI (see reply of bmi of 26/01/2007 to Questionnaire to Competitors (Base Competition), folio no. 1933, Question 3).

859 See e.g. reply of Flybe of 11/12/2007 to the Questionnaire to Competitors, folio no. 3533, question 33 and minutes of the interview with Flybe of 13 March 2007, folio no. 6170.

860 See reply of Globespan Airways of 15/11/2006 to the Questionnaire to Competitors, folio no. 21876, in particular questions 42, 43 and 52.

861 With the exception of a small scale operation of the Scottish regional carrier Loganair which flies under British Airways franchise between Glasgow and Dublin indirectly via Londonderry with small 34-seat aircraft Saab 340. The Dublin - Londonderry route is operated by Loganair under a Public Service Obligation (PSO) contract tendered by the Irish Government for routes which cannot be operated commercially but which are important for public interest. Further, Loganair's primary business is connecting the regional Scottish airports with the network of its franchise partner British Airways (see reply of Loganair of 08/11/2006 to the Questionnaire to Competitors, folio no. 21059, and www.loganair.co.uk, folio no. 13020).

862 According to the CAA Departing Passenger Survey for 2005 available at http://www.caa.co.uk, folio no. 9782, 48.3% passengers flying between Edinburgh and Ireland were non-UK passengers. This corresponds also to the figures submitted by the parties (Ryanair [40-50]*%, Aer Lingus [45-60%]* based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965). According the Irish DOT, [25-35%]* of passengers on this route originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

863 See Response of Ryanair to the Statement of Objections of 17 April 2007, paragraph 549.

864 As there are in general no indications that the situation in this respect would be different for any other of the overlap routes, a similar reasoning applies also to all the other routes analysed in this section for which Ryanair argues that the combined share of the merged entity at the destination airport would be small and that there are numerous other airlines operating to/from the destination airport (but not on the affected route). However, in case there are specific arguments in relation to a specific carrier or carriers, these are analysed in more details in the relevant route's analysis.

865 See reply of CityJet to the Questionnaire to Competitors sent on 6 November 2006, folio no. 1990, and minutes of the interview with CityJet of 21 February 2007, folio no. 6170.

866 See minutes of the interview with CityJet of 21 February 2007, folio no. 6170.

867 See reply of Ryanair to the Statement of Objections of 17 April 2007, paragraph 552.

868 See minutes of the interview with Aer Arann of 13 February 2007, folio number 6170, folio no. 6170.

869 For more details see Section 7.4.3. and Annex IV.

870 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

871 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

872 See Ryanair's Response to the Statement of Objections of 17 April 2007.

873 See Section 6.3.

874 Loganair flies between Glasgow and Dublin under British Airways franchise indirectly via Londonderry with small 34-seat aircraft Saab 340 (see reply of Loganair of 08/11/2006 to the Questionnaire to Competitors, folio no. 21059, and www.loganair.co.uk, folio no. 13020).

875 For more details see Section 7.8.5.

876 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

877 See CFI, case T-210-01 - General Electric/Commission, of 14 December 2005, ECR II-5575, paragraph 115 and Horizontal Merger Guidelines, OJ C 31, 5.2.2004, page 5, paragraph 17

878 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see e.g. Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 ff). However, as discussed in more detail above, the two carriers do compete on the same market (see Section 6.) and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Further, the fares of Aer Lingus for individual routes based on MIDT/OAG data provided by Ryanair in the Form CO cannot be considered as reliable estimates of the actual fares (see paragraph 796 and following).

879 The total duration of the indirect flight via Londonderry is 2 hours, 10 minutes compared to 55 and 50 minutes of the direct flights to Glasgow airport and Prestwick airport respectively (see on-line timetable on www.dublinairport.com, folio no. 13020).

880 According to the CAA Departing Passenger Survey for 2005 available at http://www.caa.co.uk, folio no. 9782, 37% of scheduled passengers between Glasgow and Ireland and 8% of scheduled passengers between Prestwick and Ireland were business passengers. It is also noted that Ryanair itself serves a significant number of more time-sensitive customers on this route even if it serves Prestwick, as around [10-20]*% of bookings on Ryanair flights Dublin - Prestwick are made within the seven last days before the flight. Therefore, despite the lower share of business purpose passengers at the Prestwick airport, there are indications that Ryanair still serves an important share of more time-sensitive customers. Further, [15-30%]* of Aer Lingus passengers book within the last seven days on this route (based on data by Aer Lingus of 02/02/2007, folio no. 2935).

881 Ryanair argues in its response to the Statement of Objections of 17 April 2007 (paragraph 608) that easyJet in their response to the remedies package of 29 November 2006, did not rule out entry on Dublin - Glasgow. However, without specifically referring to this route, easyJet only stated that "easyJet evaluates routes on an ongoing basis and potentially the aforementioned routes could be included in our network evaluation options." In light of the other submissions by easyJet including the minutes of the interview with them of 15/02/2007, folio no. 6170, easyJet however cannot be considered as sufficiently likely and timely entrant on this route. For more details see Section 7.8.

882 According to the CAA Departing Passenger Survey for 2005 available at http://www.caa.co.uk, folio no. 9782, around 56% of passengers in Glasgow and 60% of passengers in Prestwick flying to/from Ireland were non-UK passengers. The data provided by the parties confirm a large proportion of Irish originating passengers (Ryanair [40-50]*%, Aer Lingus [30-45%]* - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965). According the Irish DOT, [45-55%]* of passengers on this route originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

883 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

884 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

885 See Ryanair's Response to the Statement of Objections of 17 April 2007.

886 See Section 6.3.

887 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

888 See CFI, case T-210-01 - General Electric v Commission, judgment of 14 December 2005, ECR II-5575, paragraph 115 and Horizontal Merger Guidelines, OJ C 31, 5.2.2004, page 5, paragraph 17

889 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see e.g. Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 ff). However, as discussed in more detail above, the two carriers do compete on the same market (see Section 6.) and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Further, the fares of Aer Lingus for individual routes based on MIDT/OAG data provided by Ryanair in the Form CO cannot be considered as reliable estimates of the actual fares (see paragraph 796 and following).

890 See reply of Luxair of 24 November 2006 to the Questionnaire to Competitors, folio no. 23152, in particular question 3.

891 Luxair operates on this route Embraer RJ145 aircraft with 49 seats (and to a smaller extent Embraer RJ 135 aircraft with 37 seats) (see on-line timetable on www.dublinairport.com and www.luxair.lu, folio no. 13020).

892 The fact that there is indeed competition between Ryanair and Aer Lingus is also illustrated e.g. by the following Statement in the Board Paper No 3 for Board Meeting of Ryanair Holdings plc on [...]*, on page 3.4 (folio no. 629): "[...]*"

893 According to the CAA Departing Passenger Survey for 2005 available at http://www.caa.co.uk, folio no. 9782, 37.0% of scheduled passengers between Manchester and Ireland and 31.1% between Leeds Bradford and Ireland were business passengers. According to the same survey for 2003, the share of business passengers between Liverpool and Ireland amounted to 21.8%. It is also noted that Ryanair itself serves a significant number of more time-sensitive customers on this route, as between [10-20]*% and [15-25]*% of bookings on Ryanair flights on this route are made within the seven last days before the flight. As regards Aer Lingus, [15-30%]* of bookings were made within the last seven days (based on data by Aer Lingus of 02/02/2007, folio no. 2935).

894 Even though it is correct that a substantial proportion of passengers travelling on these routes could be considered as non-time sensitive (as pointed out in the Ryanair's Response to the Statement of Objections of 17 April 2007), the share of more "time sensitive passengers" is important for both Ryanair and Aer Lingus and cannot be disregarded (see footnote 893).

895 See reply of Airport Coordination Ltd of 17/11/2006 to the Questionnaire to Airports, folio no. 22175.

896 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 710.

897 In particular in view of the findings that many competitors considered other intra-European routes more attractive (see Section 7.8.6.).

898 BA Connect bases in Manchester 7 Dash 8 aircraft with 50 seats, 8 Embraer 145 aircraft with 49 seats and 1 BAE aircraft with 95 seats (see reply of British Airways of 7 February 2007 to follow-up questions, folio no. 2997, question 4).

899 Jet2 is the only airline serving Cork - Newcastle as neither Ryanair nor Aer Lingus are present there.

900 See minutes of the interview with Jet2 of 1 February 2007, folio no. 6170.

901 According to the CAA Departing Passenger Survey for 2005 available at http://www.caa.co.uk, folio no. 9782, around 50.2% of passengers flying between Manchester and Ireland and 43.4% between Leeds Bradford and Ireland were non-UK passengers. The corresponding figure for Liverpool in 2003 is 40.9%. This is largely confirmed by the data provided by the parties (Ryanair [40-50]*% for MAN, [40-50]*% for LPL and [30-40]*% for LBA, Aer Lingus [30-45%]* - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965). According the Irish DOT, [35-45%]* of passengers to MAN/LPL and [35- 45%]* of passengers to LBA originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

902 See reply of CityJet to the Questionnaire to Competitors sent on 6 November 2006, folio no. 1990, and minutes of the interview with CityJet of 21 February 2007, folio no. 6170.

903 See minutes of the interview with CityJet of 21 February 2007, folio no. 6170.

904 For more details see Section 7.4.3. and Annex IV.

905 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%. 906 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

907 See Ryanair's Response to the Statement of Objections of 17 April 2007.

908 Ryanair claims in their Response to the Statement of Objections of 17 April 2007 that "it may be that EI [Aer Lingus] will fly the route only between October and March: in the foreseeable future" (see paragraph 741). The Commission notes however that Aer Lingus is planning and have schedules for flying on this route in the Summer 2007 season.

909 See Section 6.3.

910 See the BBC News article "UK Regional airlines runs out of money" of 20 September 2001, available at

http://news.bbc.co.uk/1/hi/business/1554570.stm, folio no. 13022.

911 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

912 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see e.g. Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 ff). However, as discussed in more detail above, the two carriers do compete on the same market (see Section 6.) and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Further, the fares of Aer Lingus for individual routes based on MIDT/OAG data provided by Ryanair in the Form CO cannot be considered as reliable estimates of the actual fares (see paragraph 796 and following).

913 According to the CAA Departing Passenger Survey for 2005 available at http://www.caa.co.uk, folio no. 9782, 38.4% of scheduled passengers between Newcastle and Ireland and 18.4% between Durham Tees Valley and Ireland were business passengers. It is also noted that Ryanair itself serves a significant number of more time-sensitive customers on this route, as [10-20]*% and [10-20]*% of bookings on Ryanair flights on this route are made within the seven last days before the flight for Newcastle and Durham Tees Valley respectively. As regards Aer Lingus, [0-15%]* of bookings were made within the last seven days (based on data by Aer Lingus of 02/02/2007, folio no. 2935) - however this figure is based only on the few months of operations after October 2006 and may underestimate the importance of more time-sensitive passengers.

914 Jet2 is the only airline serving Cork - Newcastle as neither Ryanair nor Aer Lingus are present there.

915 See minutes of the interview with Jet2 of 1 February 2007, folio no. 6170.

916 According to the CAA Departing Passenger Survey for 2005 available at http://www.caa.co.uk, folio no. 9782, around 42.1% of passengers flying between Newcastle and Ireland and 21.1% between Durham Tees Valley and Ireland were non-UK passengers. This is largely confirmed by the data provided by the parties (Ryanair [25-40]*% for NCL and [20-35]*% for MME, Aer Lingus [45-60%]* - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965). According the Irish DOT, [65-75%]* of passengers on this route originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

917 See e.g. minutes of the interview with easyJet of 15 February 2007, folio no. 6170.

918 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 742.

919 For more details see Section 7.4.3. and Annex IV.

920 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

921 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

922 See Ryanair's Response to the Statement of Objections of 17 April 2007.

923 See Section 6.3.

924 However, as of the Winter 2007/2008 season, Ryanair launches also one daily rotation between Shannon and London Luton airport (LTN), see press release of Ryanair of 31/05/2007, available at the webpage: http://www.ryanair.com/site/EN/news.php?yr=07&month=may&story=rte-en-310507.

925 Even though Aer Lingus has a code-share agreement for this route with British Airways and American Airlines, it is only limited to passengers connecting at Heathrow (see reply of Aer Lingus of 20/11/2007 to Additional Questionnaire, folio no. 4122, question 15). Therefore, even though British Airways and American Airlines sell seats on Aer Lingus flights to Heathrow as part of their (usually long-haul) routes (e.g. Shannon - New York via Heathrow), they do not sell any point-to-point tickets between Shannon and Heathrow on Aer Lingus flights and thus cannot be regarded as competitors on the point-to-point market on the Shannon - London route.

926 See Board Paper 3 for the Board Meeting of Ryanair Holdings plc on [...]*, folio no. 629, page 3.6, first sentence of the last paragraph stating "[...]*".

927 Ryanair in its Board Paper No 5 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629) states the following: "[...]*" Further, the Board Paper No 8 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629) states on page 8.2 the following: "[...]*"

928 Until that time, Ryanair operated flights only between Shannon and Stansted.

929 See Ryanair's press release of 30 November 2004 "Ryanair invests $240 million in Shannon", available at http://www.ryanair.com/site/EN/news.php?yr=04&month=nov&story=rte-en-301104, folio no. 13020.

930 In the Board Paper No 3 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629) Ryanair states: [...]*.

931 See Board Paper No 3 for the Board Meeting of Ryanair Holding plc on [...]*, page 3.5 (folio no. 629). Similarly, see also Board Paper No 3 for the Board Meeting of Ryanair Holding plc on [...]*, page 3.3 (folio no. 629), stating with regard to routes from Cork/Shannon/Knock to London as follows: "[...]*".

932 See also Ryanair's press release from 9 August 2005 "Ryanair beats easyJet every time - CAA statistics prove it", available at http://www.ryanair.com/site/EN/news.php?yr=05&month=aug&story=pax-en-090805, folio no. 13020.

933 Even though Aer Lingus carries an important share of connecting passengers on this route (in Summer 2006 this share amounted to [20-30%]* of all passengers, see response of Aer Lingus of 20/11/2006 to Questionnaire to Competitors, folio no. 4122), it is still significantly lower than the share of point-to-point passengers and due to the monopoly position of the merged entity on the route, exclusion of connecting passengers would not make any material difference to the resulting combined market share.

934 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see e.g. Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 ff). However, as discussed in more detail above, the two carriers do compete on the same market (see Section 6.) and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4). Further, the fares of Aer Lingus for individual routes based on MIDT/OAG data provided by Ryanair in the Form CO cannot be considered as reliable estimates of the actual fares (see paragraph 796 and following).

935 According to Ryanair's data, [10-20]*% and [5-15]*% of bookings on Ryanair flights on this route to Stansted and Gatwick respectively are made within the seven last days before the flight.

936 See reply of BAA of 17/11/2006, folio no. 22229, and of Airport Coordination Ltd of 17/11/2006, folio no. 22175, to the Questionnaire to Airports.

937 According to the data provided by Ryanair, the share of Irish customers is [35-50]*% for Stansted and [35- 50]*% for Gatwick.

938 See in particular paragraph 857.

939 Even presuming a short 40 minutes turnaround time, it would take the aircraft and its crew in total more than 11 hours to come back to their base at Warsaw (the flight time between Shannon - Warsaw being around 190 minutes and Shannon - Stansted around 85 minutes - see www.shannonairport.com, folio no. 13020).

940 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

941 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

942 See Ryanair's Response to the Statement of Objections of 17 April 2007.

943 It should be noted that Ryanair also operates 7 weekly rotations between Kerry and Stansted. Given that Kerry is significantly more than 100 km and 1 hour driving time away from Cork (source: www.viamichelin.com, folio no. 13020) it does not seem to be substitutable for Cork. However, even in case Kerry is taken into account, the analysis of this route would not be changed as there are no other carriers operating flights between Kerry and London apart from Ryanair neither any additional based competing scheduled carriers.

944 See Section 6.3.

945 Even though Aer Lingus has a code-share agreement for this route with British Airways and American Airlines, it is only limited to passengers connecting at Heathrow (see reply of Aer Lingus of 20/11/2007 to Additional Questionnaire, folio no. 4122, question 15). Therefore, even though British Airways and American Airlines sell seats on Aer Lingus flights to Heathrow as part of their (usually long-haul) routes (e.g. Cork - New York via Heathrow), they do not sell any point-to-point tickets between Cork and Heathrow on Aer Lingus flights and thus cannot be regarded as competitors on the point-to-point market on the Cork - London route.

946 Ryanair in its Board Paper No 5 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629) states the following: "[...]*" Further, the Board Paper No 8 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629) states on page 8.2 the following: "[...]*"

947 Until that time, Ryanair operated flights only between Cork and Stansted.

948 In the Board Paper No 3 for the Board Meeting of Ryanair Holding plc on [...]* (folio no. 629) Ryanair states: [...]*.

949 See Board Paper No 3 for the Board Meeting of Ryanair Holding plc on [...]*, page 3.5. (folio no. 629). Similarly, see also Board Paper No 3 for the Board Meeting of Ryanair Holding plc on [...]*, page 3.3 (folio no. 629), stating with regard to routes from Cork/Shannon/Knock to London as follows: "[...]*".

950 Even though Aer Lingus carries an important share of connecting passengers on this route (in Summer 2006 this share amounted to [20-30%]* of all passengers, see response of Aer Lingus of 20/11/2006 to Questionnaire to Competitors, folio no. 4122), it is still significantly lower than the share of point-to-point passengers and due to the monopoly position of the merged entity on the route, exclusion of connecting passengers would not make any material difference to the resulting combined market share.

951 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see e.g. Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 ff). However, as discussed in more detail above, the two carriers do compete on the same market (see Section 6.) and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Further, the fares of Aer Lingus for individual routes based on MIDT/OAG data provided by Ryanair in the Form CO cannot be considered as reliable estimates of the actual fares (see paragraph 796 and following).

952 According to Ryanair's data, [10-20]*% and [5-15]*% of bookings on Ryanair flights on this route to Stansted and Gatwick respectively are made within the seven last days before the flight.

953 See reply of BAA of 17/11/2006, folio no. 22229, and of Airport Coordination Ltd of 17/11/2006, folio no. 22175, to the Questionnaire to Airports.

954 According to the data provided by Ryanair, the share of Irish customers is 41.5% for Stansted and 40.4% for Gatwick.

955 See in particular paragraph 867.

956 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraphs 839-840.

957 See: http://www.malev.hu/bp/eng/I%5FCORPORATE%5FENG/.

958 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraphs 840.

959 See reply of BMI/BMIBaby of 26/01/2007 to the Commission's Questionnaire to Competitors (Base Competition), folio no. 1933, question 3.

960 On a W-route Warsaw-Cork-Stansted-Cork-Warsaw, even presuming a short 40 minutes turnaround time, it would take the aircraft and its crew in total almost 11 hours to come back to their base at Warsaw (the flight time between Cork - Warsaw being around 180 minutes and Cork - Stansted around 80 minutes - see www.corkairport.com, folio no. 13020).

961 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraphs 840.

962 See minutes of the interview with Aer Arann of 13 February 2007, folio no. 6170.

963 For more details see Section 7.4.3. and Annex IV.

964 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

965 See Ryanair's Response to the Statement of Objections of 17 April 2007.

966 It should be noted that Aer Lingus currently does not offer any flights on this route between 2 June 2007 and 28 October 2007. However, this is only a temporary arrangement as Aer Lingus already now offers on its website flights on this route again from 28 October 2007 onwards and even increases the number of weekly rotations to 4 (see: www.aerlingus.com, folio no. 13020).

967 See Section 6.3.

968 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

969 Further, the combined market share of the merged entity would be even higher as of 28 October 2007 as Aer Lingus increases the number of offered rotations to 4 per week.

970 See CFI, case T-210/01 - General Electric v Commission, judgment of 14 December 2005, ECR II-5575, paragraph 115 and Horizontal Merger Guidelines, OJ C 31, 5.2.2004, page 5, paragraph 17

971 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see e.g. Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 ff). However, as discussed in more detail above, the two carriers do compete on the same market (see Section 6.) and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Further, the fares of Aer Lingus for individual routes based on MIDT/OAG data provided by Ryanair in the Form CO cannot be considered as reliable estimates of the actual fares (see paragraph 796 and following).

972 See bmi/bmi regional/bmibaby reply of 26 February 2007 to follow-up questions, folio no. 4348, question 4.

973 See minutes of the interview with bmi/bmibaby of 9 March 2007, folio no. 6170.

974 BA Connect bases in Manchester 7 Dash 8 aircraft with 50 seats, 8 Embraer 145 aircraft with 49 seats and 1 BAE aircraft with 95 seats (see reply of British Airways of 7 February 2007 to follow-up questions, folio no. 2997, question 4).

975 Jet2 is the only airline serving Cork - Newcastle as neither Ryanair nor Aer Lingus are present there.

976 See minutes of the interview with Jet2 of 1 February 2007, folio no. 6170.

977 According to the CAA Departing Passenger Survey for 2005 available at http://www.caa.co.uk, folio no. 9782, around 50.2% of passengers flying between Manchester and Ireland and 43.4% between Leeds Bradford and Ireland were non-UK passengers. The corresponding figure for Liverpool in 2003 is 40.9%.

978 See press release of DAA "More than 2.7m extra passengers at Dublin Airport in 2006" of 24 January 2007, available at: http://www.dublinairport.com/about-us/media-centre/press-releases/2006_pax_figures.html, folio no. 13020.

979 According to the DAA press release referred to in footnote 985, the total number of passengers from Poland increased from 143 000 in 2005 to over 580 000 in 2006 and the Baltic routes (Estonia, Latvia and Lithuania) collectively grew from just 174 000 passengers in 2005 to over 340 000 in 2006.

980 According to the DOT Briefing Paper of 13 November 2006, folio no. 6444, in 2002 Ryanair operated -from Dublin only two routes to Continental Europe - to Paris (Beauvais) and Brussels (Charleroi). Subsequently Ryanair opened a large number of new routes to Continental Europe, including a number of routes already served by Aer Lingus.

981 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is around [0-10%]*. It should be noted that this includes all passengers carried by charter airlines, namely both those who bought a package holiday as well as those who bought a dry seat. As discussed in more detail in paragraph 894, the share of dry seat sales is significantly lower. Furthermore, the share of charter traffic is likely to be even lower in the Summer 2007 season in view of the additional flights of Ryanair to Alicante which started in May 2007.

982 See Sections 6.3.

983 Even though Spanair operated weekly flights in the Summer 2005 and 2006 seasons, these were rather charter flights, limited to one day per week. According to their reply of 16 February 2007 to the follow-up questions, folio no. 3734, question 4, as well as information on its website www.spanair.com and the website of the Dublin airport www.dublinairport.com, Spanair does not have any scheduled services between Dublin and Alicante.

984 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

985 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see, for example, Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 217 and following). However, as discussed in more detail in Section 6., the two carriers do compete on the same market and exercise competitive constraints on each other (see, in particular, Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

986 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 491.

987 See Section 6.7.

988 According to the DAA data, these charter airlines represented more than 95% of charter passengers in 2006.

989 See replies to Questionnaire to Charter Airlines, questions 3, 6 and 8, as well as Questionnaire to Charter Airlines II, questions 2, 4 and 5.

990 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 479.

991 See reply of FirstChoice of5.3.2007 to the Charter Questionnaire, folio no. 4916, question 4.

992 Such as www.falconholidays..ie, www.panoramaholidays.ie, www.steintravel.ie, www.cassidytravel.ie or www.joewalshtour.ie.

993 See replies to Questionnaire to Charter Airlines as well as Questionnaire to Charter Airlines II.

994 According to the data from the Dublin Airport Authority, more than 90% of charter aircraft movements in 2006 fell into Saturday and Sunday and the slot allocation schedule for the summer 2007 season indicates that the situation will be similar also this year.

995 According to the data provided by the Dublin Airport Authority, the total number of passengers carried by charter airlines on the Dublin - Alicante/Murcia route decreased between 2002 and 2006 by almost [70- 80%]*.

996 See also minutes of the interview with FirstChoice of 8 March 2007, folio no. 6170, replies of the charter airlines to the Questionnaire to Charters of 23/11/2006 and to subsequent further questions addressed to charter airlines.

997 See, for example, minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

998 According to Ryanair, around [80-90]*% of its passengers on this route originate from Ireland, which corresponds also to figures for Aer Lingus ([75-90%]* of revenue from Irish originating customers- based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965). According the Irish DOT, [5-15%]* of passengers on this route originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

999 See reply of Iberia of 17 November 2006 to Questionnaire to Competitors, folio no. 22281, questions 3 and 39.

1000 See reply of Iberia of 17 November 2006 to Questionnaire to Competitors, folio no. 22281.

1001 See paragraph 485 of the Response of Ryanair to the Statement of Objections, 17 April 2007.

1002 See Section 7.8., as well as minutes of the interview with Clickair of 26 March 2007, folio no. 8091.

1003 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is around [10-20%]*. It should be noted that this includes all passengers carried by charter airlines, namely both those who bought a package holiday, as well as those who bought a dry seat. As discussed in more detail in paragraph 903, the share of dry seat sales is significantly lower.

1004 See Sections 6.3.

1005 Ryanair carried no connecting passengers while the share of connecting passengers on Aer Lingus flights to Barcelona was negligible.

1006 See Case T-210-01, General Electric v Commission, judgment of 14.12.2005 (ECR II-5575, paragraph 115), and the Horizontal Merger Guidelines (OJ C 31, 5.2.2004, p. 5, paragraph 17).

1007 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see, for example, Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 217 and following). However, as discussed in more detail in Section 6., the two carriers do compete on the same market and exercise competitive constraints on each other (see, in particular, Sections 7.3. and 7.4.. Furthermore, the fares charged by Aer Lingus for individual routes based on MIDT/OAG data provided by Ryanair in the Form CO cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1008 On its website, Clickair describes its mission as follows: "Our objective is to create a new, flexible and convenient product which combines the advantages of low cost airlines with the flexibility and value added services found at conventional companies, with the aim of becoming one of the top options for both leisure and business travellers.". See: www.clickair.com, folio no. 13021.

1009 See: http://grupo.iberia.es/portal/site/grupoiberia/menuitem.8d4d9fb9661bc259f54c0f10d21061ca/, folio no. 13021.

1010 See minutes of the interview with Clickair of 26 March 2007, folio no. 8091.

1011 See minutes of the interview with Clickair of 26 March 2007, folio no. 8091.

1012 According to Ryanair, around [60-70]*% and [80-90]*% of its passengers on this route originate from Ireland, for Reus and Girona respectively, while Aer Lingus derives [75-90%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965. According the Irish DOT, [10-20%]* of passengers on this route originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

1013 See Section 6.7.

1014 According to the DAA data, these charter airlines represented almost 99% of charter passengers in 2006.

1015 See replies to Questionnaire to Charter Airlines, questions 3, 6 and 8 as well as the Questionnaire to Charter Airlines II, questions 2, 4 and 5.

1016 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 495.

1017 See reply of FirstChoice of 05/03/2007 to the Charter Questionnaire, folio no. 4916, question 4.

1018 Such as www.falconholidays..ie, www.panoramaholidays.ie, www.steintravel.ie, www.cassidytravel.ie or www.joewalshtour.ie. However, in view of the recent decision of the UK Office of Fair Trading in the merger Flybe/BA Connect, it must be noted that the situation is different in the United Kingdom where a number of these charter airlines sell dry seats directly on their webpages (and not via tour operators) From the indicated charter carriers, at least First Choice Airways (http://www.firstchoice.co.uk) and Monarch Airlines (www.flymonarch.com) offer directly on their websites a possibility to book "flights only" from a number of the UK airports to various holiday destinations.

1019 See the slot allocation tables for Summer 2007 provided by Ryanair by e-mail of 31 January 2007 as well as information on www.xl.com, folio no. 13021.

1020 See: www.xl.com, folio no. 13021.

1021 See replies to Questionnaire to Charter Airlines as well as Questionnaire to Charter Airlines II.

1022 According to the data from the Dublin Airport Authority, almost 80% of charter aircraft movements in 2006 were on Monday, Friday and Sunday and the slot allocation schedule for the Summer 2007 season indicates that the situation will be similar also this year.

1023 According to the data provided by the Dublin Airport Authority, the total number of passengers carried by charter airlines on the Dublin - Barcelona route increased between 2002 and 2006 by only around [0-10%]*.

1024 See also minutes of the interview with FirstChoice of 8 March 2007, folio no. 6170, replies of the charter airlines to the Questionnaire to Charters of 23.11.2006 and to subsequent further questions addressed to charter airlines.

1025 See reply of AENA of 24/11/2006 to the Questionnaire to Airports, folio no. 23214.

1026 See, for example, minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

1027 See www.holaairlines.com and www.wikipedia.org, folio no. 13021.

1028 See http://www.spanair.com/web/en-gb/About-Spanair/Letter-from-our-President/, folio no. 13021.

1029 See reply of Vueling of 17 November 2006 to the Questionnaire to Competitors, folio number 22226.

1030 See reply of Vueling of 17 November 2006 to the Questionnaire to Competitors, folio number 22226 and reply of Vueling of 7 December 2006 to the Market Testing of Proposed Remedies, folio number 24193.

1031 See reply of Vueling of 7 December 2006 to the Market Testing of Proposed Remedies, folio number 24193, question 8.

1032 It should be noted that Clickair only replaces the previous operations of its parent company Iberia on this route with the same number of rotations and with a code-sharing agreement with Iberia. It cannot be therefore regarded as an entry of a new competitor.

1033 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

1034 See Ryanair's route analysis spreadsheet sent by an e-mail of 2.2.2007, folio no. 2620.

1035 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1036 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation, excluding connecting passengers, is not necessary.

1037 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see, for example, Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 217 and following). However, as discussed in more detail in Section 6., the two carriers do compete on the same market and exercise competitive constraints on each other (see, in particular, Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1038 Ryanair at 6:05 and Aer Lingus at 6:40/6:45 local time (source: www.dublinairport.com)

1039 It can be left open whether Berlin Schoenefeld is substitutable with other Berlin airports, namely Berlin Tegel and Berlin Tempelhof, as the competitive assessment would not change under any scenario.

1040 See reply of Germanwings of 15.11. 2006, folio no. 21802, and minutes of the interview with Air Berlin of 13.2. 2007, folio no. 6170.

1041 The Board paper No 3 for the Board Meeting of Ryanair Holdings plc on the 1.6.2006 (folio no. 629) states in a section devoted to Air Berlin on page 3.5, inter alia, the following: "... they were at best breakeven with practically no operating profit over the last 3 years. In reality they are a charter carrier (high cost at that), with an average cost per passenger of €90 but an average revenue per passenger of €83. It stacks very poorly to Ryanair on the financial side and their IPO also highlighted that 335% of their passengers or 3.5m passed through Palma De Majorka."

1042 According to Ryanair, around [45-55]*% of its passengers on this route originate from Ireland, while Aer Lingus derives [45-60%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26.1.2007, folio no. 1965. Furthermore, according to the Customer Survey commissioned by the Commission, 73.8% of passengers on the Dublin - Berlin route started their journey in Dublin.

1043 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 was below 5%.

1044 See Ryanair's route analysis spreadsheet sent by an e-mail of 2.2.2007, folio no. 2620.

1045 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1046 See Sections 6.3..

1047 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraphs 844.

1048 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1049 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see, for example, Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 217 and following). However, as discussed in more detail in Section 6., the two carriers do compete on the same market and exercise competitive constraints on each other (see, in particular, Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1050 For illustration, Aer Lingus' share of passengers booking in the last seven days was [0-15%]* (based on data by Aer Lingus of 2.2.2007, folio no. 2935).

1051 See, for example, minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

1052 As Ryanair only started operations on this route in January 2007, there are not yet available any reliable data on the origin of Ryanair's passengers. However, Aer Lingus derives [60-75%]* of revenue from Irish originating customers (based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965) and according the Irish DOT, only [15-25%]* of passengers on this route originated in Spain in 2005 - see DOT submission of 8.2. 2007, page 9, folio no. 6230.

1053 See reply of Spanair of 16.2. 2007 to the follow-up questions of the Commission, folio no. 3734, question 5.

1054 See http://www.spanair.com/web/en-gb/About-Spanair/Letter-from-our-President/, folio no. 13021.

1055 Dublin - Tenerife, Dublin - Madrid and Dublin - Barcelona (see www.spanair.com and www.dublinairport.com).

1056 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is around [0-10%]*.

1057 See Ryanair's route analysis spreadsheet sent by an e-mail of 2.2.2007, folio no. 2620.

1058 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1059 See Sections 6.3.

1060 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation, excluding connecting passengers, is not necessary.

1061 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see, for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6., the two carriers do compete on the same market and exercise competitive constraints on each other (see, in particular, Sections 7.3. and 7.4.). Furthermore, the fares charges Aer Lingus for individual routes, based on MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1062 For illustration, Aer Lingus's share of passengers booking in the last seven days was [0-15%]* (based on data by Aer Lingus of 02/02/2007, folio no. 2935).

1063 See, -for example, minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

1064 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 852. However, it should be noted that in their previous submission (route analysis spreadsheet sent by an e-mail of 2.2.2007, folio no. 2620) Ryanair did not indicate any competing carrier based in Bologna or Forlì airports.

1065 It should be noted that according to the reply of Forlì airport of 17.11. 2006 to the Questionnaire to Airports, WindJet operates a hub at Forlì even though it does not have a base there. WindJet is Italian low-frills airlines based in Catania, Sicily and operating mainly Italian domestic flights and only a few international destinations. From its hub at Forlì, WindJet operates flights to Bucharest, Moscow, Saint Petersburg and (as of 4 July 2007) Zakynthos but it does not operate any flights to destinations in Western Europe. See: www.volawindjet.it, folio no. 13021.

1066 See, for example, the press article of BBC News "Alitalia warns of widening losses" of 29.1.2007, available at: http://news.bbc.co.uk/2/hi/business/6308775.stm, folio no. 13021. The financial difficulties of Alitalia are also acknowledged by Ryanair internal documents, [...]*. See Board Paper No. 3 for the Board Meeting of Ryanair Holdings plc on [...]* (folio no. 629).

1067 From Bologna, Meridiana operates flights to Olbia, Cagliari, Palermo, Catania, Malta and Tenerife (source: www.meridiana.it, folio no. 13021).

1068 In particular Airbus A319 with 132 seats, MD 82 and MD 83 (up to 172 seats) (see: www.ch-avation.ch, folio no. 13021).

1069 As Ryanair only started operations on this route in February 2007, there are not yet any reliable data available on the origin of Ryanair's passengers. However, Aer Lingus derives [45-60%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26.1.2007, folio no. 1965.

1070 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

1071 See Ryanair's route analysis spreadsheet sent by an e-mail of 2.2.2007, folio no. 2620.

1072 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1073 See Sections 6.3.

1074 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1075 It should be stressed, that the Customer Survey commissioned by the Commission confirms that despite flying to different airports in Brussels, Aer Lingus and Ryanair are regarded by a large number of consumers as substitutable: on the Dublin - Brussels route, 60.6% of passengers flying with Aer Lingus considered also Ryanair while 38.4% of Ryanair's passengers considered also Aer Lingus for their flight between Dublin and Brussels.

1076 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see e.g. Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1077 It is noted that Ryanair itself serves a significant number of more time-sensitive customers on this route, as [10-25]*% of bookings on Ryanair flights on this route are made within the seven last days before the flight. The share of Aer Lingus' passengers booking within the last seven days is similar, amounting to [15-30%]* (based on data by Aer Lingus of 02/02/2007, folio no. 2935).

1078 See reply of Brussels Airport of 15/11/2006 to the Questionnaire to Airports, folio no. 21927.

1079 Transavia does not operate any scheduled flights to/from Brussels airport and does not have a base there, even though it may operate some charter flights to/from Brussels (see reply of Transavia of 17 November 2006 to the Questionnaire to Competitors, folio no. 22319).

1080 According to www.viamichelin.com (folio no. 13021), the distance between Brussels centre and Eindhoven airport is 139 km and the travelling time 1 hour 31 minutes. These figures are clearly above the indicative thresholds of 100 km and 1 hour driving time. Further, in view of the existence of both primary and secondary airport (Brussels-Zaventem and Charleroi), it is doubtful that flights from Eindhoven to Dublin would constitute any meaningful competitive constraints on flights from Brussels/Charleroi - Dublin.

1081 Airlinair is a regional French airline operating mainly domestic regional flights (with exception of two routes: Toulouse - Amsterdam and Paris - Eindhoven) using a fleet of smaller ATR42 and ATR72 aircraft. Therefore, it is not a credible entrant to the Dublin - Brussels route which would be able to exercise sufficient competitive constraints on the merged entity (see www.airlinair.com, folio no. 13021).

1082 Transavia, part of the Air France/KLM group, is a low-frills as well as charter airline operating mainly holiday destinations from Amsterdam, Rotterdam and to a much smaller extent Eindhoven (only routes from Eindhoven are Malaga, Faro, Alicante, Las Palmas and Herakion (Crete)). Transavia does not operate any route to Ireland (it exited the route Dublin - Rotterdam in March 2006 where it was the only carrier active on this route) and it is not credible that it would start operations from Eindhoven to compete with the merged entity on Dublin - Brussels route. See the reply of Transavia of 17 November 2006 to the Questionnaire to Competitors, folio number 22319, as well as www.transavia.com, folio no. 13021.

1083 See reply of Brussels Airlines of 07/02/2007 to the Questionnaire to Competitors, folio no. 2912, questions 2 and 3.

1084 Source: www.flyvlm.com, folio no. 13021.

1085 According to Ryanair, around [35-50]*% of its passengers on this route originate from Ireland, while Aer Lingus derives [45-60%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965. According to the data provided by the Irish DOT, the share of passengers originating from Belgium on this route in 2005 amounted to [35-45%]* - see DOT submission of 8 February 2007, page 9, folio no. 6230. Further, according to the Customer Survey commissioned by the Commission, 69.4% of passengers on the Dublin - Brussels route started their journey in Dublin

1086 For illustration purposes, the total number of passengers carried by scheduled airlines on this route in July 2006 was according to DAA data four times higher than in December 2006.

1087 It should be noted that this includes all passengers carried by charter airlines, namely both those who bought a package holiday as well as those who bought a dry seat. As discussed in more detail below, the share of dry seat sales is significantly lower.

1088 See e-mail by TAP Portugal of 15/02/2007, folio number 3937.

1089 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1090 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1091 See Section 6.7.

1092 Even though Ryanair refers also to some additional charter airlines on this route (see Response to the Statement of Objections), the charter carriers listed are the most important carriers which together accounted according to the DAA data for almost 90% of charter passengers in 2006.

1093 See replies to Questionnaire to Charter Airlines, questions 3, 6 and 8 as well as Questionnaire to Charter Airlines II, questions 2, 4 and 5.

1094 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 564.

1095 See reply of FirstChoice of 05/03/2007 to the Charter Questionnaire, folio no. 4916, question 4.

1096 Such as www.falconholidays..ie, www.panoramaholidays.ie, www.steintravel.ie, www.cassidytravel.ie or www.joewalshtour.ie. However, in view of the recent decision of the UK Office of Fair Trading in the merger Flybe/BA Connect, it must be noted that the situation is different in the UK where a number of these charter airlines sell dry seats directly on their webpages (and not via tour operators). At least MyTravel Airways (http://www.mytravel.com) and First Choice Airways (http://www.firstchoice.co.uk) offer in the UK directly on their websites a possibility to book flights only from a number of the UK airports to various holiday destinations.

1097 See the slot allocation tables for Summer 2007 provided by Ryanair by e-mail of 31 January 2007 as well as information on www.xl.com, folio no. 13021.

1098 See: www.xl.com, folio no. 13021.

1099 See replies to Questionnaire to Charter Airlines as well as Questionnaire to Charter Airlines II.

1100 According to the data from the Dublin Airport Authority, almost 95% of charter aircraft movements in 2006 fell into these three days.

1101 According to the data provided by the Dublin Airport Authority, the total number of passengers carried by charter airlines on the Dublin - Faro route fell between 2003 and 2006 by in total [30-40%]*.

1102 See also minutes of the interview with FirstChoice of 8 March 2007, folio no. 6170, replies of the charter airlines to the Questionnaire to Charters of 23/11/2006 and to subsequent further questions addressed to charter airlines.

1103 See for example, minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

1104 According to Ryanair, around [80-90]*% of its passengers on this route originate from Ireland, which corresponds also to figures for Aer Lingus ([75-90%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965). According the Irish DOT, [0-10%]* of passengers on this route originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

1105 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

1106 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

1107 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1108 See Sections 6.3.

1109 See Case T-210/01, General Electric v Commission, judgment of 14.12.2005 (ECR II-5575, paragraph 115), and the Horizontal Merger Guidelines (OJ C 31, 5.2.2004, p. 5, paragraph 17).

1110 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6., the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4. ). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1111 A significant share, about [...]*, of the revenue of Lufthansa on this route is generated by corporate deals.

1112 It is noted that Ryanair itself serves a number of more time-sensitive customers on this route, as [10-15]*% of bookings on Ryanair flights on this route are made within the seven last days before the flight. In case of Aer Lingus, [0-15%]* of passengers book within the last seven days on this route (based on data by Aer Lingus of 02/02/2007, folio no. 2935).

1113 It should be noted that all the three carriers active on the route offer daily early morning flight from Dublin (between 6:45 and 7:00 local time).

1114 See reply of Fraport of 16/11/2006 to the Questionnaire to Airports, folio no. 22060.

1115 See reply of Germanwings of 15 November 2006, folio no. 21802, and minutes of the interview with Air Berlin of 13 February 2007, folio no. 6170.

1116 The Board paper No 3 for the Board Meeting of Ryanair Holdings plc on [...]* (folio no. 629) states [...]*.

1117 See www.condor.com, folio no. 13021.

1118 According to Ryanair, only around [25-35]*% of its passengers on this route originate from Ireland. On the other hand, Aer Lingus indicates that [45-60%]* of its revenue comes from Irish originating customers (based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965) and the Irish DOT indicates that the share of Germany-originating passengers on this route in 2005 amounted to only [25-35%]* - see DOT submission of 8 February 2007, page 9, folio no. 6230. Further, according to the Customer Survey commissioned by the Commission, 66.5% of the passengers on the Dublin - Frankfurt route started their journey in Dublin while only 27.9% were returning to Frankfurt (the remaining 5.6% of passengers did not reply to this question).

1119 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%

1120 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

1121 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1122 See Sections 6.3.

1123 Further, Hapag-Lloyd Executive operates a fleet of very small aircraft (with maximum 11 seats) exclusively for business passengers and thus cannot be regarded as any competitive constraint for the parties (see: http://www.hl-executive.de/, folio no. 13021).

1124 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1125 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4. ). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1126 The share of Ryanair passengers booking in the last seven days on this route amounts to [0-10]*%. In case of Aer Lingus, [0-15%]* of passengers book within the last seven days on this route (based on data by Aer Lingus of 02/02/2007, folio no. 2935).

1127 See reply of Germanwings of 15 November 2006, folio no. 21802, and minutes of the interview with Air Berlin of 13 February 2007, folio no. 6170.

1128 The Board paper No 3 for the Board Meeting of Ryanair Holdings plc on [...]* (folio no. 629) states [...]*.

1129 According to Ryanair, around [20-35]*% of its passengers on this route originate from Ireland, while Aer Lingus derives [30-45%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965. According the Irish DOT, [60-70%]* of passengers on this route originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

1130 According to the data from the Dublin Airport Authority, there were no charter flights on this route in 2006.

1131 See Ryanair's route analysis spreadsheet sent by e-mail on 02/02/2007, folio no. 2620.

1132 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1133 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1134 See Case T-210-01, General Electric v Commission, judgment of 14.12.2005 (ECR II-5575, paragraph 115), and the Horizontal Merger Guidelines (OJ C 31, 5.2.2004, p. 5, paragraph 17).

1135 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see e.g. Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1136 See submission of SkyEurope of 06/12/2006, folio no. 2596, in particular replies to questions 2 and 12.

1137 See Board Paper No 3 for the Board Meeting of Ryanair Holdings plc on [...]*, page 3.7 (folio no. 629).

1138 According to Ryanair, around [45-60]*% of its passengers on this route originate from Ireland, while for Aer Lingus is [60-75%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965. While Ryanair states in the Response to the Statement of Objections of 17 April 2007 (paragraph 628) that a substantial proportion of passengers would originate rather in Poland, it does not provide any additional evidence or arguments to this effect.

1139 See Ryanair's Response to the Statement of Objections of 17 April 2007, for example, paragraphs 628, 761 or 828.

1140 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 630.

1141 See www.centralwings.com as well as www.corkairport.com and www.shannonairport.com, folio no. 13021.

1142 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 633.

1143 See Ryanair's submission of 20 February 2007, "Ryanair/Aer Lingus: Position Paper on Barriers to Entry", folio no. 4135, Annex III. There is no example of Centralwings, SkyEurope or Wizzair entering between April 2003 and October 2006 a city-pair or airport-pair already served by Ryanair. In all instances where Ryanair competes with one of these carriers on the routes between Ireland and Poland (SkyEurope on Dublin - Krakow and Centralwings on Dublin - Wroclaw and Dublin - Gdansk), it was always Ryanair who entered against existing operations of that carrier.

1144 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 amounts to [10-20]*%, while the share in the Winter 2005/2006 season amounted according to DAA data to around [20-30%]*. However, as Ryanair entered the route only in December 2006 it can be presumed that the share of charter passengers in the Winter 2006/2007 season was significantly lower as the offer of scheduled seats more than doubled with the entry of Ryanair. Further, it should be noted that this includes all passengers carried by charter airlines, namely both those who bought a package holiday as well as those who bought a dry seat. As discussed in more detail below, the share of dry seat sales is significantly lower.

1145 See Sections 6.3.

1146 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1147 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1148 See Section 6.7.

1149 According to the data from the DAA, there were no charter flights outside these winter months in the year 2006.

1150 According to the DAA data, this charter airline represented around 99% of charter passengers in 2006.

1151 See reply of Monarch of 15/02/2007 to the Questionnaire to Charters, folio no. 3664, question 4.

1152 See replies to Questionnaire to Charter Airlines, questions 3, 6 and 8 as well as Questionnaire to Charter Airlines II, questions 2, 4 and 5.

1153 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 685.

1154 See reply of FirstChoice of 05/03/2007 to the Charter Questionnaire, folio no. 4916, question 4.

1155 Such as www.falconholidays..ie, www.panoramaholidays.ie, www.steintravel.ie, www.cassidytravel.ie or www.joewalshtour.ie. However, in view of the recent decision of the UK Office of Fair Trading in the merger Flybe/BA Connect, it must be noted that the situation is different in the UK where a number of these charter airlines sell dry seats directly on their webpages (and not via tour operators). A number of charter airlines, including Monarch Airlines (www.flymonarch.com), offer on their websites a possibility to book "flights only" from a number of the UK airports to various holiday destinations.

1156 See replies to Questionnaire to Charter Airlines as well as Questionnaire to Charter Airlines II.

1157 According to the data provided by the Dublin Airport Authority, the total number of passengers carried by charter airlines on the Dublin - Lyon route decreased between 2005 and 2006 by more than [50-60%]*.

1158 See also minutes of the interview with FirstChoice of 8 March 2007, folio no. 6170, replies of the charter airlines to the Questionnaire to Charters of 23/11/2006 and subsequent further questions addressed to charter airlines.

1159 See for example minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

1160 As Ryanair started operations on this route only in December 2006, there are not yet available any reliable data on origin of Ryanair's passengers around. However, Aer Lingus derives [45-60%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965.

1161 According to Aer Lingus, only [0-15%]* of passengers on this route booked their flight in the last seven days before the flight (based on data by Aer Lingus of 02/02/2007, folio no. 2935).

1162 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 2%.

1163 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

1164 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1165 Ryanair carried no connecting passengers while the share of connecting passengers on Aer Lingus flights to Madrid was negligible (in summer 2006 it was only around [0-10%]* of total carried passengers, see response of Aer Lingus of 20/11/2006 to Questionnaire to Competitors, folio no. 4122).

1166 See CFI, , case T-210-01 - General Electric v Commission, of 14 December 2005, ECR II-5575, paragraph 115 and Horizontal Merger Guidelines, OJ C 31, 5.2.2004, page 5, paragraph 17

1167 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1168 It should be noted that Clickair states at its website that "In a near future -not before 2007- sub-bases will be opened in other major Spanish cities." See: www.clickair.com, folio no. 13021.

1169 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 857.

1170 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 858.

1171 See www.dublinairport.com and www.clickair.com, folio no. 13021. The only route Clickair operates to Dublin is from Barcelona (see the analysis of the Dublin - Barcelona route).

1172 See reply of Iberia of 17/11/2006 to the Questionnaire to Competitors, folio no. 22281, question 40.

1173 See reply of AENA of 24/11/2006 to the Questionnaire to Airports, folio no. 23214.

1174 See, for example minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

1175 See http://www.spanair.com/web/en-gb/About-Spanair/Letter-from-our-President/, folio no. 13021.

1176 See reply of Vueling of 17 November 2006 to the Questionnaire to Competitors, folio number 22226.

1177 For more details see Section 7.8 on easyJet.

1178 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 856.

1179 As Ryanair started its operations only in December 2006, there are no reliable figures as to origin of customers, however Aer Lingus derives [45-60%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965. According the Irish DOT, [40-50%]* of passengers on this route originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

1180 It should be noted that this includes all passengers carried by charter airlines, namely both those who bought a package holiday as well as those who bought a dry seat. As discussed in more detail below, the share of dry seat sales is significantly lower.

1181 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 695.

1182 See Section 6.3.

1183 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1184 See Horizontal Merger Guidelines, OJ C 31, 5.2.2004, page 5, paragraph 17

1185 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1186 According to the schedule for the Summer 2007 season, Spanair will operate two frequencies both on Saturday and Sunday (source: www.dublinairport.com, folio no. 13021).

1187 See http://www.spanair.com/web/en-gb/About-Spanair/Letter-from-our-President/, folio no. 13021.

1188 See Section 6.7.

1189 According to the DAA data, these charter airlines represented more than 95% of charter passengers in 2006.

1190 See replies to Questionnaire to Charter Airlines, questions 3, 6 and 8 as well as Questionnaire to Charter Airlines II, questions 2, 4 and 5.

1191 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 696.

1192 See reply of FirstChoice of 05/03/2007 to the Charter Questionnaire, folio no. 4916, question 4.

1193 Such as www.falconholidays..ie, www.panoramaholidays.ie, www.steintravel.ie, www.cassidytravel.ie or www.joewalshtour.ie. However, in view of the recent decision of the UK Office of Fair Trading in the merger Flybe/BA Connect, it must be noted that the situation is different in the UK where a number of these charter airlines sell dry seats directly on their webpages (and not via tour operators). From the indicated charter carriers, at least First Choice Airways (http://www.firstchoice.co.uk) and Monarch Airlines (www.flymonarch.com) offer on their websites a possibility to book "flights only" from a number of the UK airports to various holiday destinations.

1194 See replies to Questionnaire to Charter Airlines as well as Questionnaire to Charter Airlines II.

1195 According to the data from the Dublin Airport Authority, around 97% of charter aircraft movements in 2006 fell into these two days and the slot allocation schedule for the summer 2007 season indicates that the situation will be similar also this year.

1196 According to the data provided by the Dublin Airport Authority, the total number of passengers carried by charter airlines on the Dublin - Malaga route fell between 2005 and 2006 by around [30-40%]*.

1197 See also minutes of the interview with FirstChoice of 8 March 2007, folio no. 6170, replies of the charter airlines to the Questionnaire to Charters of 23/11/2006 and to subsequent further questions addressed to charter airlines.

1198 See, for example, minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

1199 According to Ryanair, around [75-90]*% of its passengers on this route originate from Ireland, while Aer Lingus generates [75-90%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965. According the Irish DOT, [0-10%]* of passengers on this route originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

1200 See reply of Air Europa of 15.11.2006 to the Questionnaire to Competitors, folio no. 21820, question 2, as well as reply of Air Europa of 16.02.2007 to the Questionnaire to Charter Airlines II, folio no. 3730, and www.air-europa.com, folio no. 13021.

1201 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

1202 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

1203 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1204 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1205 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1206 See, for example, minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

1207 See reply of Air France of 16 November 2006 to the Questionnaire to Competitors, folio no. 2507, question 39.

1208 According to Ryanair, around [45-60]*% of its passengers on this route originate from Ireland, while Aer Lingus derives [45-60%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965.

1209 According to Ryanair, only [0-10]*% passengers on this route booked their flight in the last seven days before the flight while the corresponding share for Aer Lingus is [0-15%]* (based on data by Aer Lingus of 02/02/2007, folio no. 2935).

1210 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is around [0-10%]*.

1211 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1212 See Sections 6.3.

1213 The share of connecting passengers carried by Ryanair and Aer Lingus on this route is negligible. To the extent that Alitalia carried connecting passengers in the Summer 2006 season, the separate market share calculation excluding connecting passengers is not necessary in view of its exit from the route (and anyway limited market share in Summer 2006 even when all passengers are taken into account).

1214 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1215 See reply of Societa Esercizi Aeroportuali p.A. - S.E.A. of 17/11/2006 to the Questionnaire to Airports, folio no. 22228.

1216 See, for example, the press article at BBC News "Alitalia warns of widening losses" of 29/01/2007, available at: http://news.bbc.co.uk/2/hi/business/6308775.stm, folio no. 13021. The financial difficulties of Alitalia are also acknowledged by Ryanair internal documents commenting that "Al Italia have been bankrupt for pretty much most of the last 20 years". See Board Paper No. 3 for the Board Meeting of Ryanair Holdings plc on 14th December 2006 (folio no. 629).

1217 See: www.flyairone.it, folio no. 13021.

1218 See: www.myair.com, folio no. 13022.

1219 For more details see Section 7.8.

1220 According to Ryanair, around [35-50]*% of its passengers on this route originate from Ireland, while Aer Lingus derives around [45-60% ]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965. Further, according to the data provided by the Irish DOT, the share of passengers originating at destination on this route in 2005 amounted to only [25-35%]* - see DOT submission of 8 February 2007, page 9, folio no. 6230.

1221 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

1222 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

1223 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1224 See Sections 6.3.

1225 Ryanair carried no connecting passengers while the share of connecting passengers on Aer Lingus flights to Paris CDG was negligible (in summer 2006 it was only around [0-10%]* of total carried passengers, see response of Aer Lingus of 20/11/2006 to Questionnaire to Competitors, folio no. 4122)

1226 See CFI, , case T-210-01 - General Electric v Commission, of 14 December 2005, ECR II-5575, paragraph 115 and Horizontal Merger Guidelines, OJ C 31, 5.2.2004, page 5, paragraph 17

1227 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4. ). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1228 CityJet operates on this route 93-seats BAe 146 aircraft, compared to 174-seats Aer Lingus and 189-seats Ryanair aircraft.

1229 For more details on consumer study, please see Annexes I and II.

1230 It is noted that Ryanair itself serves a significant number of more time-sensitive customers on this route, as [0-15]*% of bookings on Ryanair flights on this route are made within the seven last days before the flight for a price which is in general higher than for earlier bookings. In case of Aer Lingus, [0-15%]* of passengers book within the last seven days on this route (based on data by Aer Lingus of 02/02/2007, folio no. 2935).

1231 See reply of Aeroports de Paris of 16/11/2006 to the Questionnaire to Airports, folio no. 22149, as well as www.cohor.org.

1232 According to Ryanair, around [50-65]*% of its passengers on this route originate from Ireland, which corresponds also to figures for Aer Lingus ([60-75%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965) and for the route as a whole (according to the data provided by the Irish DOT, the share of France-originating passengers on this route in 2005 amounted to only [20-30%]* - see DOT submission of 8 February 2007, page 9, folio no. 6230.

1233 According to the data from the Dublin Airport Authority, there were no charter flights on this route in 2006.

1234 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

1235 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1236 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1237 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1238 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 760.

1239 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 755.

1240 See Ryanair's submission of 20 February 2007, "Ryanair/Aer Lingus: Position Paper on Barriers to Entry", folio no. 4135, Annex III. There is no example of Wizzair entering between April 2003 and October 2006 a city-pair or airport-pair already served by Ryanair.

1241 See in particular response of SkyEurope of 6 December 2006 to the Market Testing of Proposed Remedies questionnaire, folio no. 2596, in particular questions 2, 3, 12 and response of SkyEurope of 7 February 2007 to the follow-up questions by the Commission, folio no. 2878.

1242 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 762.

1243 According to the data from the Dublin Airport Authority, there were no charter flights on this route in 2006.

1244 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

1245 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1246 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1247 See CFI, , case T-210-01 - General Electric v Commission, of 14 December 2005, ECR II-5575, paragraph 115 and Horizontal Merger Guidelines, OJ C 31, 5.2.2004, page 5, paragraph 17

1248 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1249 See reply of Air Baltic of 15.11.2006 to the Questionnaire to Competitors, folio no. 21855, questions 2, 3, 6 and 7.

1250 See submission of Air Baltic of 15 November 2006, folio no. 21855.

1251 According to Ryanair, around [50-65]*% of its passengers on this route originate from Ireland, while for Aer Lingus [60-75%]* of revenue comes from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965.

1252 For more details on consumer study, please see Annexes I and II.

1253 See paragraphs 797 and 798 in the Dublin - London route assessment.

1254 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is around [0-10%]*.

1255 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

1256 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1257 See Sections 6.3.

1258 It should be noted that in the summer 2006 season Alitalia offered again some direct flights between Dublin and Rome, however these were limited to Saturday and Sunday operations only and represented only around [0-10%]* of total passengers in summer 2006. However, for summer 2007 there are no direct scheduled flights between Dublin and Rome offered by Alitalia (see: www.dublinairport.com, folio no. 13022).

1259 In this particular case, while direct flights of Ryanair and Aer Lingus take around 3 hours or even less, the indirect flights of Alitalia take usually between 5 and 8 hours, depending on the connection in Paris CDG (see schedule information on the website of Dublin Airport www.dublinairport.com).

1260 The share of connecting passengers carried by Ryanair and Aer Lingus on this route is negligible. Even though Alitalia carried connecting passengers in Summer 2006 season, the separate market share calculation excluding connecting passengers is not necessary in view of its exit from the route (and anyway limited market share in Summer 2006 even when all passengers are taken into account).

1261 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1262 See reply of ADR of 22/11/2006 to the Questionnaire to Airports, folio no. 22732.

1263 See, for example, the press article at BBC News "Alitalia warns of widening losses" of 29/01/2007, available at: http://news.bbc.co.uk/2/hi/business/6308775.stm, folio no. 13021. The financial difficulties of Alitalia are also acknowledged by Ryanair internal documents commenting that "Al Italia have been bankrupt for pretty much most of the last 20 years". See Board Paper No. 3 for the Board Meeting of Ryanair Holdings plc on 14th December 2006 (folio no. 629).

1264 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 776.

1265 Even presuming a short 40 minutes turnaround time, it would take the aircraft and its crew in total around 12 hours to come back to their base at Krakow (the flight time between Dublin - Rome being around 3 hours and Rome - Krakow around 2 hours - see www.dublinairport.com and www.adr.it, folio no. 13022).

1266 According to Ryanair, around [60-75]*% of its passengers on this route originate from Ireland, which corresponds also to figures for Aer Lingus ([60-75%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965) and for the route as a whole (according to the data provided by the Irish DOT, the share of passengers originating from Italy on this route in 2005 amounted to only [20-30%]* - see DOT submission of 8 February 2007, page 9, folio no. 6230.

1267 This is confirmed also by the DAA data according to which the share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 was [30-40%]* and even higher in the winter 2005/2006 season. It should be however noted that Ryanair entered this route only in April 2006 and that this includes all passengers carried by charter airlines, namely both those who bought a package holiday as well as those who bought a dry seat. As discussed in more detail below, the share of dry seat sales is significantly lower.

1268 It should be noted that according to the data from DAA, Aero Lloyd operated some at least semi-scheduled services with several interruptions between 1999 and 2002. Aero Lloyd was primarily a charter airline which ceased its operations in October 2003 (see: http://en.wikipedia.org/wiki/Aero_Lloyd, folio no. 13022).

1269 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1270 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1271 See Section 6.7.

1272 According to the DAA data, these charter airlines represented more than 95% of charter passengers in 2006.

1273 See replies to Questionnaire to Charter Airlines, questions 3, 6 and 8 as well as Questionnaire to Charter Airlines II, questions 2, 4 and 5.

1274 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 784.

1275 See reply of FirstChoice of 05/03/2007 to the Charter Questionnaire, folio no. 4916, question 4.

1276 Such as www.falconholidays..ie, www.panoramaholidays.ie, www.steintravel.ie, www.cassidytravel.ie or www.joewalshtour.ie. However, in view of the recent decision of the UK Office of Fair Trading in the merger Flybe/BA Connect, it must be noted that the situation is different in the UK where a number of these charter airlines sell dry seats directly on their webpages (and not via tour operators). For example MyTravel Airways (http://www.mytravel.com), First Choice Airways (http://www.firstchoice.co.uk) or Thomas Cook Airlines (www.flythomascook.com) offer on their websites a possibility to book "flights only" from a number of the UK airports to various holiday destinations. It should be stressed that not even Austrian Airlines, which has an on-line booking system for its scheduled services does not offer direct Dublin - Salzburg flights on its website.

1277 See replies to Questionnaire to Charter Airlines as well as Questionnaire to Charter Airlines II.

1278 According to the data provided by the Dublin Airport Authority, the total number of passengers carried by charter airlines on the Dublin - Salzburg route decreased slightly between 2005 and 2006 by [0-10%]*.

1279 See also minutes of the interview with FirstChoice of 8 March 2007, folio no. 6170, replies of the charter airlines to the Questionnaire to Charters of 23/11/2006 and to subsequent further questions addressed to charter airlines.

1280 See, for example, minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

1281 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 782.

1282 Vienna, Linz, Frankfurt and Düsseldorf (see: www.austrian.com, folio no. 13022).

1283 According to Ryanair, around [45-60]*% of its passengers on this route originate from Ireland, while Aer Lingus derives [75-90%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965.

1287 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1288 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1289 For illustration, the share of passengers booking in the last seven days for Aer Lingus was [0-15%]* (based on data by Aer Lingus of 02/02/2007, folio no. 2935).

1290 See, for example, minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

1291 See minutes of the interview with Clickair of 26 March 2007, folio no. 8091.

1292 See minutes of the interview with Clickair of 26 March 2007, folio no. 8091.

1293 As Ryanair started operations on this route only in February 2007, there are not yet available any reliable data on origin of Ryanair's passengers. However, Aer Lingus derives [75-90%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965.

1294 It should be noted that Ryanair entered this route only in December 2006 and that this figure includes all passengers carried by charter airlines, namely both those who bought a package holiday as well as those who bought a dry seat. As discussed in more detail in paragraph 1065, the share of dry seat sales is significantly lower.

1295 See Sections 6.3.

1296 See replies to the Questionnaire to Competitors send on 6 November 2006, question 27.

1297 The total duration of the flight is around 4 hours 20 minutes. On the other hand, the duration of the indirect flights is still significantly longer, ranging from 6 to 10 hours.

1298 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1299 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1300 See Section 6.7.

1301 According to the DAA data, these charter airlines represented almost 99% of charter passengers in 2006.

1302 See replies to Questionnaire to Charter Airlines, questions 3, 6 and 8 as well as Questionnaire to Charter Airlines II, questions 2, 4 and 5.

1303 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 696.

1304 See reply of FirstChoice of 05/03/2007 to the Charter Questionnaire, folio no. 4916, question 4.

1305 Such as www.falconholidays..ie, www.panoramaholidays.ie, www.steintravel.ie, www.cassidytravel.ie or www.joewalshtour.ie. However, in view of the recent decision of the UK Office of Fair Trading in the merger Flybe/BA Connect, it must be noted that the situation appears different in the UK where a number of these charter airlines sell dry seats directly on their webpages (and not via tour operators). From the indicated charter carriers, at least MyTravel Airways (http://www.mytravel.com), First Choice Airways (http://www.firstchoice.co.uk) and Monarch Airlines (www.flymonarch.com) offer on their websites a possibility to book "flights only" from a number of the UK airports to various holiday destinations.

1306 See replies to Questionnaire to Charter Airlines as well as Questionnaire to Charter Airlines II.

1307 According to the data from the Dublin Airport Authority, around 98% of charter aircraft movements in 2006 fell into these three days. According to the slot allocation for winter 2006/2007 and summer 2007, the same pattern can be expected also in these seasons.

1308 According to the data provided by the Dublin Airport Authority, the total number of passengers carried by charter airlines on the Dublin - Tenerife route fell between 2005 and 2006 by around [10-20%]*.

1309 See also minutes of the interview with FirstChoice of 8 March 2007, folio no. 6170, replies of the charter airlines to the Questionnaire to Charters of 23/11/2006 and to subsequent further questions addressed to charter airlines.

1310 See, for example, minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

1311 According to Ryanair, around [50-65]*% of its passengers on this route originate from Ireland, while Aer Lingus generates [75-90%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965.

1312 See reply of Spanair of 16 February 2007 to the follow-up questions of the Commission, folio no. 3734, question 5.

1313 See http://www.spanair.com/web/en-gb/About-Spanair/Letter-from-our-President/, folio no. 13021.

1314 According to the DAA data, the share of charter passengers on total (both scheduled and charter) passengers carried on this route in Summer 2006 season was only [0-10%]*.

1315 Further, even on a yearly basis the presence of charters is rather limited, amounting to [0-10%]* of total traffic in 2006.

1316 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1317 See Sections 6.3.

1318 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1319 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1320 See, for example, minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

1321 See reply of Air France of 16 November 2006 to the Questionnaire to Competitors, folio no. 2507, question 39.

1322 According to Ryanair, around [60-75]*% of its passengers on this route originate from Ireland, while Aer Lingus derives [60-75%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965.

1323 According to Ryanair, only [0-5]*% passengers on this route booked their flight in the last seven days before the flight while the corresponding share for Aer Lingus is [0-15%]* (based on data by Aer Lingus of 02/02/2007, folio no. 2935).

1324 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 805.

1325 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 804. It should be however noted that CityJet commenced scheduled flights from London City to the following Continental Europe destinations: Geneva, Madrid, Milan, Nice and Zurich (see press release of CityJet of 15/02/2007, www.cityjet.com, folio no. 13022). Therefore, the only other French destination added to the already served Paris is Nice and the Statement the Statement of Ryanair in paragraph 804 that CityJet commenced routes to "a number of French destinations" is thus factually incorrect.

1326 See minutes of the interview with CityJet of 21/02/2007, folio no. 6170.

1327 According to the data from the Dublin Airport Authority, there were no charter flights on this route in 2006.

1328 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

1329 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1330 See Sections 6.3.

1331 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1332 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1333 See, for example, minutes of the interview with FirstChoice of 8.3.2007, folio no. 6170.

1334 See, for example, the press article at BBC News "Alitalia warns of widening losses" of 29/01/2007, available at: http://news.bbc.co.uk/2/hi/business/6308775.stm, folio no. 13021. The financial difficulties of Alitalia are also acknowledged by Ryanair internal documents commenting that "Al Italia have been bankrupt for pretty much most of the last 20 years". See Board Paper No. 3 for the Board Meeting of Ryanair Holdings plc on 14th December 2006 (folio no. 629).

1335 The fleet of Alpi Eagles consists of 8 Fokker 100 aircraft with 94 seats (see: www.alpieagles.com, folio no. 13022)

1336 The website of Alpi Eagles states the following: "Our mission is to increase a market niche of routes connecting the North-East with destinations to the South and Islands, becoming the chosen company for North -South routes. In recent years, we have developed international routes to Athens, Barcelona and Madrid." (see: www.alpieagles.com, folio no. 13022)

1337 According to Ryanair, around [50-65]*% of its passengers on this route originate from Ireland, while the corresponding figure for Aer Lingus is even higher ([60-75%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965).

1338 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

1339 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

1340 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1341 See Sections 6.3.

1342 See www.dublinairport.com, folio no. 13022.

1343 See www.austrian.com, folio no. 13022.

1344 As the share of connecting passengers carried by all airlines operating on this route is negligible, the separate market share calculation excluding connecting passengers is not necessary.

1345 See Case T-210/01, General Electric v Commission, judgment of 14.12.2005 (ECR II-5575, paragraph 115), and the Horizontal Merger Guidelines (OJ C 31, 5.2.2004, p. 5, paragraph 17).

1346 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4.). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1347 See submission of SkyEurope of 06/12/2006, folio no. 2596, in particular replies to questions 2 and 12.

1348 See Board Paper No 3 for the Board Meeting of Ryanair Holdings plc on [...]*, page 3.7 (folio no. 629).

1349 According to Ryanair, around [45-60]*% of its passengers on this route originate from Ireland, while for Aer Lingus is [45-60%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965. According the Irish DOT, [15-25%]* of passengers to/from Vienna originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

1350 For more details on consumer study, please see Annexes I and II.

1351 See reply of Flughafen Wien AG of 14/11/2006 to the Questionnaire to Airports.

1352 See www.laudaair.com, folio no. 13022.

1353 See reply of Slovak Airlines of 14 November 2006 to the Questionnaire to Competitors.

1354 See: www.airslovakia.sk and www.ch-aviation.ch, folio no. 13022).

1355 The share of charter passengers on total (both scheduled and charter) passengers carried on this route in 2006 is below 1%.

1356 See Ryanair's route analysis spreadsheet sent by an e-mail of 02/02/2007, folio no. 2620.

1357 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1358 See Ryanair's Response to the Statement of Objections of 17 April 2007.

1359 See Case T-210/01, General Electric v Commission, judgment of 14.12.2005 (ECR II-5575, paragraph 115), and the Horizontal Merger Guidelines (OJ C 31, 5.2.2004, p. 5, paragraph 17).

1360 Ryanair argues that the fares charged by Aer Lingus are substantially higher than those of Ryanair and their business model is substantially different and thus that there are only minor constraints between the two carriers (see , for example, Ryanair's Response to the Statement of Objections of 17 April 2007, para. 217 and following). However, as discussed in more detail in Section 6, the two carriers do compete on the same market and exercise competitive constraints on each other (see in particular Sections 7.3. and 7.4. ). Furthermore, the fares charged by Aer Lingus for individual routes, based on the MIDT/OAG data provided by Ryanair in the Form CO, cannot be considered as reliable estimates of the actual fares (see paragraphs 797 and 798).

1361 According to their Quarterly Report for Third Quarter 2006 available on their website, Norwegian Air Shuttle based 2 aircraft in Warsaw only from the fourth quarter of 2006 (see: http://www.norwegian.no/graphics/Investorrelations/Reports/2006/NAS3.Quarter2006.pdf, folio no. 13022) In this report they state that they plan a considerable strengthening and increase in the marketing activities in Poland. The base in Warsaw is not expected to be profitable in the first year of operation. However, they also acknowledge that "Previous experience shows that it takes time to incorporate new routes, and with the expansion in Poland the company is entering into a new market which will present the company with new challenges. ... The base in Warsaw is not expected to be profitable in the first year of operation."

1362 See submission of SkyEurope of 06/12/2006, folio no. 2596, in particular replies to questions 2 and 12.

1363 See Board Paper No 3 for the Board Meeting of Ryanair Holdings plc on [...]*, page 3.7 (folio no. 629).

1364 Centralwings fly to Dublin from Wroclaw, Gdansk, Szczecin and Katowice, to Cork from Wroclaw, Warsaw and Krakow and to Shannon from Gdansk, Warsaw and Katowice. See: www.centralwings.com

1365 See: www.wizzair.com, folio no. 13022.

1366 As Ryanair started its operations only in December 2006, there are no reliable figures as to origin of customers, however Aer Lingus derives [60-75%]* of revenue from Irish originating customers - based on data submitted by Aer Lingus on 26/01/2007, folio no. 1965. According the Irish DOT, [20-30%]* of passengers on this route originated at destination in 2005 - see DOT submission of 8 February 2007, page 9, folio no. 6230.

1367 See Ryanair's Response to the Statement of Objections of 17 April 2007, for example, paragraphs 628, 761 or 828.

1368 See Ryanair's submission of 20 February 2007, "Ryanair/Aer Lingus: Position Paper on Barriers to Entry", folio no. 4135, Annex III. There is no example of Centralwings, SkyEurope or Wizzair entering between April 2003 and October 2006 a city-pair or airport-pair already served by Ryanair. In all instances where Ryanair competes with one of these carriers on the routes between Ireland and Poland (SkyEurope on Dublin - Krakow and Centralwings on Dublin - Wroclaw and Dublin - Gdansk), it was always Ryanair who entered against existing operations of that carrier.

1369 Notably "Ryanair/Aer Lingus: Merger Efficiencies" of 20 October 2006 by RBB Economics (folio no. 23962 and also Annex 1 of the Notification).

1370 Ryanair submitted the paper "Ryanair/Aer Lingus: Merger Efficiencies" during the prenotification phase (folio no. 19873). The Commission commented on it on 24 October 2006 (folio no. 23962) and expressed its reservations as to the completeness and the merger specific character of the argumentation. The answers to those queries were provided on 5 December 2006 (folio no. 24006) and could not be considered as exhaustive and satisfactory in order to fully endorse Ryanair's efficiency claims and avoid a Phase II investigation. On 11 January 2007 (folio no. 466) and 22 February 2007 (folio no. 5130), the Commission sent an additional set of questions to Ryanair, to which Ryanair replied on 27 January 2007 (folio no. 4558) and 27 February (folio no. 4558), respectively.

1371 See Article 2(1)(b) and Recital 29.

1372 Guidelines on the assessment of horizontal mergers under the Council Regulation on the control of concentrations between undertakings, OJ C 31, 5.2.2004, p. 5.

1373 See § 84 of the "Guidelines on the assessment of horizontal mergers under the Council Regulation on the control of concentrations between undertakings".

1374 See footnote 1370.

1375 Ryanair allows for some leeway in this estimation, underlining that even if only half of those savings were achieved, this would amount to €[...]*.

1376 Ryanair's reply of 27 February 2007 (folio no. 4558) to Commission's questionnaire of 22 February 2007.

1377 Ryanair Response to the Statement of Objections, par. 247

1378 Notably, the replies to Commission's questionnaire of 22 February and the CRA memorandum of 2 March therein attached (folio no. 5704), commenting on the RBB paper of 20 October 2006.

1379 CRA memorandum of 2 March commenting on the RBB paper of 20 October 2006, folio no. 5704.

1380 CRA memorandum of 2 March commenting on the RBB paper of 20 October 2006, folio no. 5704.

1381 Aer Lingus letter of 14 February to Mrs Larsson-Haug, folio no. 5704.

1382 Aer Lingus letter of 14 February to Mrs Larsson-Haug, folio no. 5704.

1383 CRA memorandum of 2 March commenting on the RBB paper of 20 October 2006, folio no. 5704.

1384 cf. RBB Economics: "Ryanair/ Aer Lingus: Merger Efficiencies", folio no. 19873.

1385 ditto, Annex 1, p. 2.

1386 ditto, Annex 1, p. 8.

1387 Reply to the Statement of Objections, p.77

1388 See in detail Section 7.4.

1389 Part of the fare increase was imposed through checked baggage fees See Ryanair's report of the results of Quarter 3/2006: "The 7% rise in yields largely reflects the impact of competitors excessive and unjustified fuel surcharges as well as revenues from checked baggage charges which were introduced in March '06 and lack of a prior year comparative in this quarter's numbers." http://www.ryanair.com/site/about/invest/docs/2007/q3_2007_doc.pdf.

1390 Reply to the Statement of Objections, par. 253

1391 Cf. Response to the Statement of Objections, "Letter by Michael O'Leary to the European Commission"

1392 Cf. Ryanair's reply of 26 January 2007 to a Commission request for information (folio no. 1795)

1393 See in this respect paragraph 86 of the Guidelines on the assessment of horizontal mergers under the Council Regulation on the control of concentrations between undertakings (Horizontal Merger Guidelines).

1394 See e.g. Ryanair's announcement to reduce the number of crew members in Aer Lingus' aircraft.

1395 Cf. Response to the Statement of Objections, "Letter by Michael O'Leary to the European Commission".

1396 The costs in question are fixed in relation to the services that are sold in the relevant market, i.e. passenger seats on scheduled flights. By contrast, Ryanair argues that the claimed efficiencies relate to the marginal costs involved in allocating an additional aircraft to a route. This aspect is discussed below (section titled "Benefits to consumers".

1397 Cf. Response to the Statement of Objections, par 278-285

1398 Cf. Ryanair's reply of 26 January 2007 to a Commission request for information (folio no. 1795)

1399 Cf in this respect also Horizontal Merger Guidelines, paragraph 79.

1400 See § 84 of the Horizontal Merger Guidelines; see also paragraph 79.

1401 The revised remedies were submitted first on 13 December 2006. However, as the two versions of these remedies sent by e-mail and faxed differed and as further clarifications as to the actual number of slots offered was needed, the corrected and final version of the revised remedies was submitted on 14 December 2006 in the afternoon.

1402 Reference was made to the lower of the relevant Aer Lingus or Ryanair frequency. Under the Phase I Commitment, a prospective new entrant would be able to get slots at the Dublin Airport and the relevant other airport on the route needed to support in aggregate: up to [1-4]* roundtrips per day on the Dublin - London-Heathrow route, up to [1-4]* roundtrips per day on the Dublin - London-Stansted route, up to [1- 10]* roundtrips per week on the Dublin - Paris-Beauvais route, up to [1-2]* roundtrips per day on the Dublin - Frankfurt-Hahn route, up to [1-3]* roundtrips per week on the Dublin - Milan-Bergamo route, up to [1-2]* roundtrips per day on the Dublin - Rome-Fiumicino route, - up to [1-3]* roundtrips per week on the Dublin - Bratislava route, - up to [1-4]* roundtrips per week on the Dublin - Barcelona-Girona route. The conditions for the provision of slots included that slots released by the merged entity, and taken up by new entrants, should represent an equitable mixture of peak and non-peak slots at Heathrow, Stansted and Rome Fiumicino. The duration of the commitments relating to slots was limited to [2-4]* years. Further, the slots released by the Parties were to be within 45 minutes of the time requested by the new entrant. Moreover, so called "mega-carriers" such as British Airways/oneworld, Air France/KLM/CityJet/SkyTeam or Lufthansa/British Midlands/Star Alliance were excluded from the possibility to obtain the offered slots.

1403 The second remedy package did not address all shortcomings (confirmed by the participants to the market test) of the first remedies package related to e.g. the conditions under which the remedies would be provided (e.g. the window for provision of the slots remained 45 minutes, at least on some routes the remedies again related only to secondary airports, the "mega carriers" were still excluded for slots at their hub airports) and it also limited the possibility of carriers to take-up the remedies within [2-4]* years.

1404 The State-of-Play Meeting between the DG Competition and Ryanair took place on [...]*. Ryanair announced in this meeting that [...]*. Prior to submitting the Final Commitments Ryanair did not contact DG Competition in order to discuss the details of the draft Final Commitments and the necessary improvements in terms of form and substance with the DG Competition (as recommended by the Remedies Notice, paragraph 40: "The Commission is available to discuss suitable commitments prior to the end of the threemonth period. The parties are encouraged to submit draft proposals dealing with both substantive and procedural aspects which are necessary to ensure that the commitments are fully workable.") before it submitted its Final Commitments on 3 May 2007.

1405 See title and text of Ryanair's e-mail of 3 May 2007 (23.54 h).

1406 See page 5 of the Commitments Letter.

1407 See Ryanair's e-mail of 15 May, page 2.

1408 It may be noted that Ryanair has in various e-mails (see mails of 8, 11, 15 and 18 May 2007) proposed different "interpretations" of some elements of the Final Commitments (e.g. that the slots for London Heathrow are not exclusively offered to British Airways/Air France or that the "up-front solution" should also apply to the Heathrow slots, see Ryanair's e-mail of 8 May 2007). However, Ryanair has not provided further reasons for its interpretation, and the different e-mails were even contradictory as regards Ryanair's views (e.g., Ryanair wrote on 8 May 2007: "Therefore, to clarify, the Heathrow slots are not exclusive to BA and Air France." On 11 May 2007 Ryanair wrote: "These slots are exclusively reserved for British Airways and Air France (...)". See on other incoherent statements (e.g. as regards the scope of the up-front solution [...]*) the Commission's e-mail to Ryanair of 14 May 2007 (sent at 12:06). In the absence of any formal clarification or modification of the commitments text, the Commission has to base its assessment on the text as submitted for the purpose of this decision.

1409 See point 3 of Annex 1 - Mechanics of Commitment No. 2.

1410 See "Commitment No.2", (a) of the Initial Phase II Commitments. See also "Commitments Letter", page 3 of the Final Commitments.

1411 Ryanair offered to make available [6-10]* out of 13 Aer Lingus' daily slot pairs at London Heathrow specifically to British Airways and Air France (equally divided between them) to operate on the Dublin - London route. In order to accommodate early morning departure, Ryanair would also make available [1-2]* early morning peak departure [slot/slots] (between 6:00 and 8:00) for each of these two carriers. At the same time Ryanair committed to make available slots at Dublin Airport necessary for these two carriers to serve the Dublin - London route. Ryanair thus committed to make available in total [24-40]* Dublin - Heathrow slots per day ([12-20]* each to British Airways and Air France - [6-10]* arrival and [6-10]* departure slots).

1412 The slots would only be offered if the interested airline has demonstrated that "all reasonable efforts" to obtain slots for the respective overlap route have failed, see Clause 2.9. of Annex I of the Final Commitments.

1413 As set out below in Section 8.1, it is not entirely clear from the text of the Final Commitments whether the entrant can use the slots on other than the identified overlap routes.

1414 See in detail Clauses 2, 3, 4 and 5 of the Final Commitments. The airports covered by this commitment are as follows: Alicante, Barcelona (Girona and/or Reus), Bilbao and/or Vitoria, Berlin (Schönefeld), Birmingham, Brussels (Charleroi), Edinburgh, Faro, Frankfurt (Hahn), Fuerteventura, Glasgow Prestwick, Hamburg Lübeck, Krakow, London (Luton, Stansted and/or Gatwick), Lyon and/or Grenoble, Malaga, Madrid, Manchester, Marseille, Milan (Bergamo), Newcastle, Paris (Beauvais), Poznan, Riga, Rome (Ciampino and/or Fiumicino), Seville, Tenerife (North and/or South), Toulouse and/or Carcassonne, Turin, Venice (Treviso), Vienna and/or Bratislava, Warsaw.

1415 The slots would, again, only be offered if the interested airline has demonstrated that "all reasonable efforts" to obtain slots for the respective overlap route have failed, see e.g. Clause 2.9.

1416 See "Commitments Letter", page 5.

1417 Ryanair further explains that it means by "non-completion" that "the offer would not become unconditional". See the description on page 2 and 5 of the Commitments Letter: "Ryanair (..) shall not complete the acquisition (i.e. the new offer would not become unconditional) unless and until the identified upfront buyer has committed to taking up the slots for the [4-8]*based aircraft operation at Dublin Airport."

1418 In the Commission's practice, a clause according to which a merger cannot be implemented until a buyer of certain divested assets or rights has been approved by the Commission is usually referred to as "upfront buyer solution, see e.g. the Commission Notice on remedies acceptable under Council Regulation (EEC) No 4064/89 and under Commission Regulation (EC) No 447-98 ("Remedies Notice"), at paragraph 20.

1419 See e.g. page 5 of the "Commitments Letter".

1420 See e-mail of Mr Lücking to Mr Power of 8 May 2007, folio no. 11390.

1421 See cover letter of Annex I (questionnaire) of the Market Test.

1422 See e.g. paragraph 6 of the Remedies Notice: "(...) the parties are required to show clearly, to the Commission's satisfaction in accordance with its obligations under the Merger Regulation, that the remedy restores conditions of effective competition in the common market on a permanent basis." (text underlined by the Commission).

1423 See Remedies Notice, paragraph 40: "(...) to ensure that the commitments are fully workable." (text underlined by the Commission). See also CFI in Case T-210-01 General Electric v Commission of 14 December 2005, (2005), p. II-5575, at paragraph 555: "It must be noted that under Regulation No 4064-89 the Commission has power to accept only such commitments as are capable of rendering the notified transaction compatible with the common market (...). It must be held in that regard that structural commitments proposed by the parties will meet that condition only in so far as the Commission is able to conclude, with certainty, that it will be possible to implement them and that the new commercial structures resulting from them will be sufficiently workable and lasting to ensure that the creation or strengthening of a dominant position (...)."

1424 The elements mentioned in the following section are not meant to be exhaustive but to summarise the most salient formal/legal problems which directly affect the viability of the remedy.

1425 Originally, the "set of commitments" also comprised a letter from Ryanair's COO and a further annex ("Appendix 1").

1426 See e.g. the Best Practice Guidelines which explain the Commission's practice and the use of the proposed model texts, http://ec.europa.eu/comm/competition/mergers/legislation/note.pdf.

1427 See Sections 8.2.1.1 - 8.2.1.4.

1428 See page 2, 5, 9 of the Commitments Letter, page 2 of the letter of Ryanair's COO.

1429 There is only a general description of Ryanair's willingness to agree to an up-front solution in the Commitments Letter (see above, Section 8.1.3); in the text which sets out the legal details of the slot divestiture an up-front condition is missing. See for previous cases with commitments involving up-front buyer solution and the respective clauses used e.g. Case COMP/M.3796 - Omya/Huber PCC of 19 July 2006; Case COMP/M.2972 - DSM/Roche Vitamins of 23 July 2003; Case 3732 - Procter & Gamble/Gillette; Case COMP/M.2060 - Bosch/Rexroth of 13 December 2000; Case COMP/M.2337; Nestlé/Ralston Purina of 27 July 2001; COMP/M.2544 Masterfoods/Royal Canin of 15 February 2002; COMP/M.2947 Verbund/Energie Allianz of 11 June 2003.

1430 It should be also noted that Ryanair had also claimed in its e-mail of 8 May 2007 to have the "intention" that the up-front buyer provisions should also apply to the London/Heathrow slots. The Commission has informed Ryanair by e-mail of 10 May 2007 that this "intention" is not reflected in the submitted remedies text and that the remedies would need to be formally modified should Ryanair wish to enlarge the scope of the "up-front buyer" solution. In its e-mails of 11 May 2007, 15 May 2007 and 18 May 2007, Ryanair has, however, dropped its claim of 8 May 2007 and agree with the Commission's interpretation that the London/Heathrow slots are not part of the "upfront-buyer" solution.

1431 At least the "Commitments Letter" restricts the use explicitly to the Dublin/Heathrow route, see page 3.

1432 See Ryanair's letters of 11 May 2007, 15 May 2007 and 18 May 2007 (2 c) respectively).

1433 See page 4: "(...) to operate on any route (or routes) including any or all of the alleged overlap routes, without restriction".

1434 See e.g. Clause 2.6 of "Annex 1" (unclear provisions for the "discount" of frequencies of entrants that already operate on an overlap route), Clause 2.11 (further limitations for potential entrants; the justification and the content of the provision are unclear); the text also fails to make clear what happens in the event that the London/Heathrow slots are not used or "misused", see e.g. Clause 2.15 of "Annex 1".

1435 See replies to the Market test, notably to question 5.

1436 See e.g. Remedies Notice, paragraph 26-30.

1437 It should be noted that Ryanair itself seems to put into question the likelihood of successful entry as a result of the commitments. In the "Appendix 1" which was submitted to the Commission as part of the Final Commitments, Ryanair states on the necessity of an "up-front" solution: "The Commission is attempting to penalise Ryanair with a condition [the up-front buyer requirement] which it knows no other airline will accept because Ryanair's success in offering low costs and low fares to/from Dublin." (text highlighted/inserted in brackets by the Commission).

1438 See description of the Final Commitments, (2) and (3).

1439 See Remedies Notice, paragraph 28.

1440 See in this respect CFI, T-177-04 EasyJet v Commission, of 4 July 2006, ECR [2006] II-1913, paragraph 197 et seq; in particular as regards the requirement of a real possibility of new entry see paragraph 206 of the judgment: "The Commission was not required to identify a definite new entrant since various competitors had expressed an interest in entering the affected markets."

1441 As in the present case, see notable above, Section 7.8.

1442 See answers to question 4 a) of the Market Test. See e.g. the answer from Sky Europe (folio no. 2596): slots are "not the only driver in the market"; it should be noted that amongst the few airlines who considered slot remedies in principle as appropriate in this case were at least two airlines interested in obtaining London/Heathrow slots; the others made important reservations (LTU (folio no. 8934): "only if strong entrant"; SAS (folio no. 9013): "at least better" than today) or were not conclusive ([other competitor] (folio no. 8977): ..."On non-congested airports it should be a sufficient remedy").

1443 See notably Section 7.9.

1444 See answers to the Market Test, question 3 a), second tick box and the respective explanations.

1445 See answers to the Market Test, question 3 a), third tick box and the respective explanations.

1446 The identity of this competitor is kept confidential.

1447 See Section 7.3.5.

1448 See reply to the Market Test, answers to questions 3 b) and d) and to question 6 a).

1449 See e.g. interview with CityJet (Air France) of 21.2.2007, folio no. 6170.

1450 See email of 31.05.2007, folio number 10623.

1451 easyJet, as Ryanair has rightly pointed out in its Reply to the Statement of Objections, had indicated in an early response that it "evaluates routes on an ongoing basis and potentially the aforementioned routes could be included in our network evaluation options" (see easyJet's reply to the market test of the remedies phase 1, folio no. 24385), it has, in a later and more detailed interview with the Commission explained that it "does not have any concrete plans to enter the Irish market" and that "entry to Ireland would be a "significant decision" with significant necessary investment which would depend on many other factors [apart from slots]" (see agreed minutes from the interview with easyJet held on 15/02/2007, folio no. 6170). easyJet also pointed to the high investment costs and to the fact that other opportunities in Europe for new routes are "more profitable and less risky" than flying to Ireland.

1452 See answers to the Market Test, question 3 d) to f).

1453 Reply of British Airways of 11 December 2006 to the Questionnaire to Competitors (base competition), folio no. 1558, question 15 (submitted together with the reply to the Market Testing of Proposed Remedies questionnaire, i.e. with full knowledge of the proposed commitments).

1454 Reply of British Airways to question 3 f) of the Market Test (folio no. 9239).

1455 See in detail above, Section 7.8.

1456 See responses to question 4 of the Market Test.

1457 See e.g. CFI in Case T-210-01 General Electric v Commission of 14 December 2005, (2005), p. II-5575, at paragraph 555: "It must be noted that under Regulation No 4064-89 the Commission has power to accept only such commitments as are capable of rendering the notified transaction compatible with the common market (...). It must be held in that regard that structural commitments proposed by the parties will meet that condition only in so far as the Commission is able to conclude, with certainty, that it will be possible to implement them and that the new commercial structures resulting from them will be sufficiently workable and lasting to ensure that the creation or strengthening of a dominant position (...)."; see also paragraph 7 of the Remedies Notice.

1458 Ryanair further explain that that it means by "non-completion" that "the offer would not become unconditional". See the description on page 2 and 5 of the Commitments Letter: "Ryanair (..) shall not complete the acquisition (i.e. the new offer would not become unconditional) unless and until the identified upfront buyer has committed to taking up the slots for the [4-8]*based aircraft operation at Dublin Airport."

1459 See Remedies Notice, paragraph 20.

1460 See e.g. CFI in Case T-210-01 General Electric v Commission of 14 December 2005, (2005), p. II-5575, at paragraph 555: "It must be noted that under Regulation No 4064-89 the Commission has power to accept only such commitments as are capable of rendering the notified transaction compatible with the common market (...). It must be held in that regard that structural commitments proposed by the parties will meet that condition only in so far as the Commission is able to conclude, with certainty, that it will be possible to implement them and that the new commercial structures resulting from them will be sufficiently workable and lasting to ensure that the creation or strengthening of a dominant position (...).";see also paragraph 7 of the Remedies Notice.

1461 See in detail above, Section 8.2.2.1 and Section 8.1.

1462 See notably Section 7.8.

1463 See in detail below ("Offering slot remedies is not likely to trigger any substantial entry on the overlap routes").

1464 As set out above Section 8.2.1, it is not clear whether the acquirer of the slots would be allowed to operate on other than the overlap routes.

1465 See page 4 of the Commitments Letter.

1466 On page 3 of the Commitments Letter Ryanair assumes that an aircraft can fly approximately 4 frequencies (8 one-way flights) per day. This would amount, strictly calculated, to 336 weekly slots or 168 frequencies per week. If either BA or Air France would use all slots made available at other airports than London Heathrow (which they do not intend to do as confirmed by the market test), they could in theory use around 400 slots (London slots plus the other slots for routes from/to Dublin).

1467 See in this respect above, Section 7.8.

1468 Or seven aircraft, if British Airways or Air France were to take the non-Heathrow slots

1469 Aer Lingus speaks about the "bare minimum" of frequencies required in order to offer "at least a partial alternative", see Aer Lingus, reply to the Market Test of 15 May 2007, page 21, folio no. 9267.

1470 According to Aer Lingus, this calculation includes also 1 aircraft to serve the Dublin - London route, in addition to the two aircraft reserved under the commitments to British Airways and Air France and not included in this figure, as the new entrant would still be likely to want to offer services to London, given the profitability and size of the London route (see answer to question 2 b) of Aer Lingus' reply to the Market Test, page 21-22, folio no. 9267).

1471 See answer to question 2 g) of Aer Lingus' reply to the Market Test (page 28), folio no. 9267.

1472 This is based on the assumption that new airlines would be able to operate with 2 aircraft on the Heathrow336 Dublin route and with [4-8]*aircraft on the remaining Dublin routes. See below on the further problem whether "fragmented" entry by different players (i.e. two for the "reserved" Heathrow routes and even more for the remaining routes) would be as efficient as entry by a single player with a large base ("Fragmented entry is less likely to restore competition").

1473 See replies to question 2 a) of the Market Test. Even if only carriers with a low cost model and thus a high utilisation of aircraft are taken into account, the number of aircraft that would be necessary to operate at a comparable size as Aer Lingus still ranges between 12 and 15 aircraft to cover only the actual overlap routes. See also answers to question 2 b) and g) of the Market Test.

1474 The responses indicate that in average around 4 rotations per day are possible with one aircraft, however depending on a number of factors such as length of the individual routes, congestion at the destination airports, business model of the entrant etc. See e.g. answer of Air Berlin to the Market Test (folio no. 9116 and 9328).

1475 The Commitments letter does not explain that slots at destination airports are made available.

1476 See above Section 6.4.

1477 See answers to Question 2 c) of the Market Test.

1478 See notably Section 7.3.4.

1479 See answers to Question 2 c) of the Market Test.

1480 See answers to Question 2 d) of the Market Test.

1481 See in detail Section 7.8; see also the route-by-route analysis in Section 7.9.

1482 The only Clause which concerns criteria on the eligibility of a new entrant ("business justification") is in itself unclear and seems even to favour entrants which need to provide connecting services. This is certainly not a suitable criterion for the entrant in the present case, which should ideally follow the same point-to-point focussed model as Ryanair and Aer Lingus (see: Annex 1, Clause 2.11).

1483 See e.g. above, Section 7.3.5.

1484 See in this respect also the answers to question 2 d) of the Market Test. The Commission notes that some airlines seemed to answer from their own perspective (i.e. taking only into account whether they could provide a profitable service and not whether this service would replace Aer Lingus from a customers' point of view).

1485 See below Section 8.2.2.4.

1486 See e.g. above Sections 7.8 and 7.9.

1487 While Air France does not have a hub at London Heathrow, it would be able to feed traffic to the flights of its partners in SkyTeam alliance operating to Heathrow.

1488 See, for example, minutes of the interview with CityJet (Air France) of 21.2.2007, folio no. 6170.

1489 See above Section 7.4.

1490 It should be mentioned that the airline bmi has also expressed interest for Dublin-London/Heathrow slots in the Market Test (folio no. 9133).

1491 See with respect to the legal issues in connection with the divestiture of the London/Heathrow slots Section 8.2.2.4.

1492 See also the answers to question 3 g) of the Market Test. It should be noted that according to BA and Air France, they would rather switch their services to Heathrow than adding these services to the existing ones.

1493 See above Section 8.1. (e.g. definition of "slot misuse", consequences of a "misuse" of slots etc.). It is also noted that the slots are transferred by way of a so-called "leasing arrangement", which might require further monitoring.

1494 See below, Section 8.2.4.

1495 The Commission also notes that it makes little difference in substantive terms how different elements of the final price paid by consumers are labelled and that Ryanair itself states in its results announcement for the third quarter of 2006 that competitors' fuel surcharges enabled it to increase average fares by 7%.

1496 See Market Test, answers to Question 6.

1497 See Sections 7.1.-7.9.

1498 The Commission notes that the commitments are only related to frequencies and would not restrict Ryanair's pricing policy.

1499 Cf. paragraph 7 of the Commission's Remedies Notice.

1500 See Aer Lingus' submissions of 28 November 2006 (folio no. 4122), 18 April 2007 (folio no. 8035), Aer Lingus' reply to the Market Test of 15 May 2007 (folio no. 9267) and the submission of 30 May 2007 (folio no. 10216).

1501 See e.g. the DOT's response to the Market Test, folio no. 9072, page 7.

1502 See Commitments Letter, page 3 and Ryanair's submission of 30 May 2007.

1503 Let alone to conclude with certainty that it will be possible to implement them. See case T-210-01, General Electric v Commission, paragraph 555, quoted above. See also Remedies Notice, paragraph 32 "Where the parties submit proposed remedies that are so extensive and complex that it is not possible for the Commission to determine with the required degree of certainty that effective competition will be restored in the market, an authorisation decision cannot be granted."

1504 This meeting was held by way of a telephone conference.

1505 See letter of Ryanair of 1 June 2007.

1506 Letter from Commissioner Kroes to Ryanair of 4 June 2007.

1507 OJ L 133, 30.04.2004, p.1,.

1508 Cf. Case T-87-05 EDP v Commission [2005] ECR II-3745, paragraphs 161-163.

1509 Even though the Draft Modified Final Commitments provide for a possibility for further growth of this entrant up to 10 aircraft based in Dublin within the next three years, they provide no guarantee of such growth of the new entrant's activities as the up-front requirement is limited to an entrant with [4-8]* aircraft. In view of the results of the market test of the Final Commitments which did not identify any potential entrant which would be prepared to establish a strong presence in Dublin, a simple possibility of offering slots for 10 aircraft without any guarantee, e.g. in the form of a sufficiently clear up-front provision, cannot be considered as sufficient.

1510 British Airways, Air France, bmi/Star Alliance or any other Heathrow-based airline.

1511 The exceptions are Dublin-London (Heathrow slots), Cork-London (Stansted slots), Shannon-London (Stansted slots) and Cork-Manchester (Liverpool slots).

ANNEX I:

Customer Survey - General Overview M.4439 - Ryanair/Aer Lingus

1. INTRODUCTION TO THE SURVEY

1. The European Commission prepared a survey in the context of the Ryanair/AerLingus case in the month of February 2007. The purpose of this survey was to obtain a representative sample of responses from customers who departed from Dublin.

2. The Commission hired an experienced Contractor (TNS mrbi) to conduct a customer survey at Dublin Airport. The Commission designed the questionnaire, processed and analysed all the responses. Given the tight deadlines and the need to collect a sufficiently large and representative sample of responses the questionnaire was intentionally short and all questions were multiple choice (that is the questionnaire includes no open-ended questions requesting the respondent to explain or motivate his/her answer). It takes between 5 and 15 minutes to fully answer the questionnaire. Importantly, no questions were asked that would allow the Commission or any party with access to the responses to trace the identity of the respondent.

3. TNS mrbi is one of the most experienced service research companies in Ireland, being also one of the founders of AIMRO (1) which established the minimum standards for the market research fieldwork in Ireland. Furthermore, it was the firm that proposed to include in the survey the largest amount of routes in the very limited period of time given. The fieldwork was done between the 1st and the 10th February and the data was sent to the European Commission the 13th February.

2. SAMPLE

4. The Commission proposed a sample of routes of short haul flights from Dublin to EU destinations that were relevant for the case divided into different categories (2). Category A included the routes where both carriers operated into the same airport. Category B included routes in which AerLingus and Ryanair operated into different airports. Finally, category C involved routes in which also other carriers operated. From that list of different routes TNS mrbi chose four of each category. The company was allowed to choose the routes so that the Commission would not bias the results by choosing specific routes.

5. In total 2674 questionnaires were collected distributed by route as follows:

<emplacement tableau>

6. The break down by carriers is as follows:

<emplacement tableau>

7. Finally and for the sake of completion the graph below indicates the flight number by time of day and carrier.

<emplacement tableau>

2. DESCRIPTIVE ANALYSIS

2.1 Question 1

Did you buy a return ticket or a one-way ticket for today's trip?

8. The respondents were only allowed to give one answer out of the three possible options that they were given. Most of the people that were interviewed for this survey, the 82.3%, had a return ticket, while only a 15.9% had a one way ticket. There was a very small percentage of respondents that did not know whether they had a return or a one-way ticket or that just did not answer at all.

<emplacement tableau>

9. Passengers mostly buy return tickets

<emplacement tableau>

2.2 Question 2

How much was the total price of this ticket? (Including taxes and charges but excluding any travel insurance costs)

10. The respondents had multiple choices, i.e. different ranges of prices, but could only choose one of the options. They paid in a higher proportion prices between 51 and 150 euros for the selected routes: 29.8% at the range of price of 51 to 100 euros and 26.8% of them from 101 to 150 euros, but there was a much smaller proportion of respondents that bought their tickets at a higher price than 200 euros. The 10% of respondents did not know how much they had paid, which is almost the same proportion of respondents who paid up to 50 euros and from 151 to 200 euros.

<emplacement tableau>

11. Passengers spend mostly between 51 and 150 euros on the ticket

2.3 Question 3

What is the main purpose for making this trip today?

12. The respondents could only answer one of the multiple options that they were given, although they had one open-ended answer where they could fill in other reasons that were not mentioned in the other options given in the questionnaire. The majority of respondents, 41.8% of them, were flying for leisure while the proportion of business and visiting friend travellers is much smaller, 23.8% and 29% respectively. The amount of respondents that answered that the reason for travelling was different than business, leisure or visiting friends and the number of those who did not answer is very small.

<emplacement tableau>

13. Passengers travel mostly for leisure reasons

2.4 Question 4

Did you start your journey today or is this a return flight?

14. The respondents could only choose one of the two options given in the questionnaire. Most of the passengers in the survey were actually starting the journey that day from Dublin's airport, 68.3% of them, while only a 28.7% were actually returning.

<emplacement tableau>

15. Most of the passengers flying from Dublin's airport actually have it as the origin

2.5 Question 5

If started journey today at question 4: Are you using the same airline for your return flight to Dublin?

16. The respondents could only choose one of the three different answers given in the questionnaire. The majority of respondents, the 77.9%, flew back and forth with the same company (more than half of them), while the proportion of those who did not use the same airline, did not answer or were just not applicable was in comparison very small.

<emplacement tableau>

17. Passengers mostly return to Dublin with the same airline as they flew

2.6 Question 6

If chosen return flight at question four: Is the airline for this flight the same as the airline you used to travel to Dublin?

18. The respondents could only choose one of the two options given in the questionnaire. As in the previous question, most respondents who were flying back to the origin (65.7%) flew in both directions with the same airline.

<emplacement tableau>

19. Passengers mostly fly back to their departure airport from Dublin with the same airline

2.7 Question 7

Do you always use the same airline when travelling this route?

20. The respondents could only choose one of the two options given in the questionnaire. In this case there was not such a big difference (just a 4%) between the people that tend to use the same airline and those who did not when travelling on the specific route they were taking when they were interviewed. There was also a relatively big amount of passengers who thought it was not applicable.

<emplacement tableau>

21. Passengers travel always with the same airline nearly in the same proportion as those who do not

2.8 Question 8 Which other airlines, if any, did you consider using for this route?

22. The respondents were allowed to give more than one answer and they even had the possibility to write the name of the other airlines considered if they were not in the list included in the questionnaire. Most of them, 27.1%, did not consider any other airline as an alternative while 26.6% took into account AerLingus, 24.4% considered Ryanair and in a much smaller percentage they considered other airlines when flying from Dublin's airport.

<emplacement tableau>

23. Most passengers flying from Dublin's airport do not consider any airline as an alternative to their chosen airline, but if they do it is mainly either AerLingus or Ryanair

(a) If the ticket price offered by an alternative airline had been 10% lower than the price you paid, would you have purchased that ticket instead?

24. The respondents were only allowed to choose one option out of the several options given. More than half of the respondents would change airline if the tickets were a 10% lower in another airline, while the percentage of those who wouldn't or that have not considered it is much more reduced.

<emplacement tableau>

25. Passengers would change to another airline if its tickets were a 10% cheaper

2.9 Question 9 Which of the reasons below describe why you chose to fly with this airline today?

26. The respondents had multiple options but were allowed to choose more than one answer. Most of them chose the airline with which they were flying when they were interviewed because it had offered them the best price (33.2%), the best time (23.3%) and a the closest connection to the city they were travelling to (16.3%). The respondents took into account the other reasons in a much more reduced percentage.

<emplacement tableau>

27. The best price was one of the main factors why passengers chose a certain airline

2.10 Question 10

Would you ever consider flight to/from Belfast as and alternative to using Dublin airport?

28. The respondents were allowed to choose only one of the two options given in the questionnaire. Most of the respondents, the 65.5%, would never consider flying from/to Belfast as an alternative to Belfast. The proportion of passengers that would consider it, or that do not know, is quite similar but small compared to the amount of people who would not take it into account.

<emplacement tableau>

29. Passengers would not consider Belfast as an alternative to Dublin's airport

3. COMMISSION'S RESPONSE TO THE COMMENTS MADE BY RYANAIR IN THE REPLY TO THE STATEMENT OF OBJECTIONS

30. In the Response to the Statement of Objections Ryanair argues that customer survey is deficient because "it was undertaken on a self-completion basis" and "the questions posed are ambiguous and often leading "questions" requiring knowledge on the part of the respondent, particularly in the case of the Consumer (passenger) survey, which it was not reasonable to assume it existed" (3). The Commission notes that self-completion questionnaires are a standard technique to gather information from consumers in all sectors, including air travel. Second, all questions refer to the respondents own actions, perceptions and beliefs so it is quite reasonable to assume he/she can have sufficient knowledge. It is possible that in some cases the respondent may not have arranged the trip. For this reason in all critical questions the option "not applicable" or "not applicable/did not arrange flight" was given (4). Furthermore, in this case, the respondent is in fact in the process of experiencing the service it has purchased (5).

31. Ryanair also argues that "it is a manifest error that no interviews were carried out at the other ends of each route to ensure that the views of Irish and non-Irish passengers were fairly represented" (6). The Commission notes that it had to consider the the timing and economic constraints imposed by the regulatory framework it works under when it decided to conduct the survey, including how many routes and customers to cover. This applies also as to whether consider the views of customers at both ends of the routes. In any event responses to question 4 show that 68.3% of passengers started their journey at Dublin (7). This in itself suggests that a large majority of passengers travelling on the routes covered by the survey originate/started their journey in Dublin. Further, no significant error should result from conducting the survey in Dublin only given that 82.3% of passengers reported having purchased a return ticket. Indeed, the survey did not select only on-bound customers; it also asked the question to "returning" passengers. Therefore, all passengers irrespectively if they were commencing their journey in Dublin or were returning to their destination were covered. Thus, it makes no significant difference whether they are interviewed in Dublin or at the other end of the route.

32. Finally Ryanair points out that another source of bias may be that 22.7% of Aer Lingus passengers were on UK routes whilst 39.6% of Ryanair passengers were on UK routes. As UK routes may be expected to have more business passengers than some of the European routes this will distort comparisons between the two airlines (8). Ryanair also points out in its reply that Aer Lingus passengers travelling to Heathrow were only surveyed on a weekend which would have "a skewed results towards carrying a disproportionate high percentage of leisure passengers on the Heathrow route. This will impact significantly on the validity of any conclusions drawn about the substitutability of Aer Lingus and Ryanair on that route"(9).

33. The Commission acknowledges that conducting the survey on a weekend for the London route may have some impact on the percentage of business travellers on the routes. This may have induces some bias in the responses but it is unlikely that the bias should be "significant" overall for the UK routes or on the London route specifically. It is noted that a significant proportion of Ryanair customers have considered Aer Lingus although Ryanair fly into different airports. In fact, even if Ryanair does not impose a significant constraint on Aer Lingus on Heathrow it is clear that Aer Lingus does exert a constraint on Ryanair, because passengers, whether business or leisure is considering Aer Lingus and that Heathrow may be a more convenient location.

Results based on unweighted data

34. The results of the customer survey provide clear evidence that passengers flying on routes out of Dublin consider Aer Lingus and Ryanair to be the closest substitutes. In response to question 8, 477 passengers out of 1166 travelling with Ryanair (40.9%) (10) said they had considered using Aer Lingus in the route they were going to travel on that day (see table below). Only 12.8% of Ryanair passengers considered any other carrier. The difference in percentages is statistically significant at less than 1% level (11).

35. Conversely, 430 passengers out of 1042 travelling with Aer Lingus (41.3%) (12) said they had considered using Ryanair in the route they were going to travel on that day. Only 10.4% of Aer Lingus passengers considered any other carrier. The difference in percentages is statistically significant at less than 1% level (13).

<emplacement tableau>

36. In contrast, passengers travelling with a carrier other than Ryanair or Aer Lingus considered Ryanair much less frequently (25.8% in the case of Ryanair and 39.1% in the case of Aer Lingus).

37. The Commission also points out that there is certain symmetry in the responses to question 8 by passengers of Ryanair and Aer Lingus. In particular the difference in the proportion of passengers flying with FR and EI that considered the other (40.9% and 41.3%, respectively) is not statistically significant (14). This suggests that the competitive constraint that both carriers impose on each other is symmetric. This is relevant because, in contrast to the Commission's regression analysis that focuses on the impact of presence of one firm on the prices of the other, respondents to question 8(a) are not asked whether they considered the other carrier on the basis of a particular dimension (e.g. price).

Questionnaire design does not lead to inflated results

38. Ryanair argues in its reply to the Statement of Objections that these results are not conclusive proof of substitutability because the figures are inflated for two reasons. First "respondents were prompted to tick FR and EI (but not other carriers)" and second "many consumers may "consider" both FR and EI only to conclude that the two airlines are not substitutable" (15). The Commission believes that neither of these criticisms is valid to dismiss the results of the survey.

39. First, it should again be emphasised that the primary goal of the survey was to test (i.e. validate or refute) Ryanair's claim that Ryanair and Aer Lingus do not compete in routes out of Dublin from the perspective of consumers. In a series of submissions to the Commission, Ryanair continuously claims that, in case it were to raise prices its customers would choose not to fly rather than switch to Aer Lingus (that is Ryanair is constrained by consumers, not Aer Lingus or any other competitor). The Commission considers that asking passengers directly whether they considered Aer Lingus or Ryanair when flying in a particular route out of Dublin is the best way to avoid bias in the response to the critical question that serves to test Ryanair's claim that Merging Parties do not constraint each other.

40. A second general remark is that both Aer Lingus and Ryanair were consulted on the questionnaire design before using it. With respect to this particular question, neither party expressed any concern or insisted in the necessity to list all airlines or alternatively to leave the question open (16).

41. Ryanair suggests that the fact that "other carriers" were not mentioned by name and a box was provided would bias the respondents against ticking this box, or giving the name of the alternative carrier considered. The Commission disagrees. In fact by providing a box for "others" the respondent is given the possibility to tick the box even if he/she does not remember exactly the name of the carrier but recalls it had considered some carrier other than the one of choice. Indeed 52 out 307 (17%) of respondents did exactly this - ticked the "others" box without writing in the name of the alternative carrier. At the same time it appears that most respondents could easily recall the name of the other carrier they might have considered - up to 24 different airlines were mentioned by name.

42. In any event, excluding the London routes, in all other routes surveyed there is at most one competitor and in 6 out of 12 only the Merging Parties are present. Moreover, adding a long list of all carriers operating out of Dublin may have induced certain confusion in the respondent (17).

43. As to the use of the word "consider" the Commission believes that it is appropriate to test the hypothesis that customers perceive the Merging Parties to be substitutable. Indeed when the respondent has purchased a ticket with a particular airline it has made its choice and therefore it is clear that, at least for that flight, he/she did not believe that any other alternative was better. However, what is of interest here is whether the passenger, before making its choice, has indeed considered that some other airline might offer a competing alternative. Moreover the question is phrased in the past-tense so as to ensure that the respondent reports actual behaviour, and not hypothetical future behaviour.

44. Furthermore, the Commission also points out that the responses to question 8 are very likely to be biased downwards, precisely because as argued by Ryanair, if a passenger has in the past considered various airlines but then made up its mind in favour of one, it is likely that in future travel on the same route it will not consider the alternatives again and may directly book with its airline of choice the last time. Thus in answering to question 8 a passenger that travels often on a given route may respond that it did not considered any other alternative. The Commission was aware of this source of downward bias when designing the questionnaire. To partly resolve the problem question 8 makes reference to the "route", not the "flight". This, however, has not been sufficient to eliminate the downward bias in responding that another airline had been considered (i.e. an upwards bias in responding that "none" alternative airlines were considered). Indeed, of those Ryanair passengers that in response to question 7 indicated that they always use Ryanair to travel in the route, only 22% of them responded that they had considered Aer Lingus as an alternative (and only 5% considered any other airline) (18). In comparison up to 66.1% of Ryanair passengers that do not always use Ryanair to travel in the route had considered Aer Lingus as an alternative (and only 24.5% considered any other airline). This implies that for passengers that have not necessarily made up their mind to always travel with the same airline 2/3 of them had considered Aer Lingus (19).

Results based on unweighted data by route category

45. Routes sampled were intentionally selected on the basis of certain criteria to ensure the sample would be representative. In particular all London routes were included, routes were the Merging Parties were travelling to the same airport; routes were they served different airports and finally routes in which they faced competition from other carriers (20). Table A.4.1 in fact shows that despite some differences across type of route a significant number of passengers consider Ryanair as an alternative and vice-versa.

46. As may be expected, where both airlines operate into the same airport at destination, the percentage of passengers which have considered the other Irish airline is the highest. 54.2% of Ryanair customers have considered flying with Aer Lingus and 51.8% of Aer Lingus customers have considered Ryanair.

47. On routes where Aer Lingus and Ryanair compete with a third carrier, 40.4% of Ryanair's passengers consider Aer Lingus an alternative, while 17.4% consider (all) other competitors. Similarly, for Aer Lingus, 32.5% of its passengers consider Ryanair as an alternative while 15.7% consider (all) other competitors.

48. On the Dublin-London market, Aer Lingus flies into Heathrow in direct competition with bmi while Ryanair flies into Gatwick, Stansted and Luton and overlaps directly with BA in Gatwick. CityJet moreover flies into London City. In spite of the fact that the two parties fly into different airports at the other end and that each overlaps directly with another competitor flying into the same London airport, still their passengers consider the other Irish airline as the closest substitute. 33.9% of Ryanair's customers have considered Air Lingus and 14.6% have considered (all) other airlines. In the case of Aer Lingus, 18.4% of its customers have considered Ryanair and 9.2% have considered others airlines.

49. These results hold also when one distinguishes different customer groups (as business or leisure travellers) or between different reasons why passengers bought a ticket. A further indication is the number of Ryanair passengers that have selected Best price, Best time, Close Airport and Punctuality (the four most popular reasons) and whether they had considered Aer Lingus and other airlines (all aggregated). Of the 860 passengers that indicated they had chosen Ryanair because it offered the best price 44.4% considered Aer Lingus as an alternative, as opposed to 14.7% who considered carriers other than Aer Lingus. The same pattern is apparent whether the respondent's preference for Ryanair is due to the fact that it offered a good departure time, the destination airport was conveniently located or punctuality.

50. The results are similar when one looks at Aer Lingus customers. Interestingly, Aer Lingus passengers also seem to have a strong preference for Ryanair as an alternative again irrespective of the reason why they have selected Aer Lingus in the first place.

Results based on weighted data

51. Ryanair argues in its Response to the Statement of Objections (21) that data has to be weighted by route. Ryanair argues that "some of the analysis presented by the Commission aggregate results across the full 2674 interviews completed. This is not valid as different routes will have different characteristics and data should properly have been weighted by the number of passengers carried on each flight as a proportion of all passengers carried by the relevant carriers from Dublin over the period" (22). Notwithstanding this concern, Ryanair does not make any attempt to produce results based on weighted data (23).

52. The Commission does not disagree with Ryanair that in some instances the results based on weighted data may produce very different, if not opposite results than if the data is left unweighted. This, however, is not here the case.

53. A sample is, by definition, smaller than the corresponding population. The sample could be a disproportionate representation of the population if some groups are over- or underrepresented. Applying weights to samples seeks the general goal of making the sample more like the population at large. For example, as pointed out by Ryanair the customer survey includes 299 responses in the Vienna-Bratislava route but only 148 in the Heathrow route. However almost 10 times as many passengers travel in the Heathrow route in any given week. In other words, in stratified sampling, proportions in the overall sample may not coincide with proportions in the population. Correcting for this is straightforward with proportional weights, which are specific to every subdivision of the sample for which the sampling ratio is homogeneous (e.g. each stratum) (24).

54. To weight the sample the Commission has used as an estimate for the population size at route level the weekly average number of passengers on that route (from DAA data on yearly number of passengers). The table below reproduces the results reported above correcting for the over-sampling or under-sampling in each route.

<emplacement tableau>

55. The percentage of Aer Lingus passengers that considered Ryanair as an alternative is lower than in the case of the unweighted data (34.5% as opposed to 41.3%). However it can still be regarded as a significant that slightly more than 1/3 of Aer Lingus passengers had considered Ryanair as an alternative. It is also still above the 8.8% of Aer Lingus passengers that considered any other carrier. Furthermore when considering only Aer Lingus passengers that stated that they do not always travel with Aer Lingus the percentage that considered Ryanair increases to 62.3%.

56. All other percentages do not differ significantly from the results with unweighted data. The percentage of Ryanair passengers that considered Aer Lingus is 40.4% with weighted data and 40.9% with unweighted data. The percentage of passengers travelling with other airlines that considered Ryanair does not change at all (25.8% whether using weighted or unweighted data) and those that considered Aer Lingus increases from 39.1% to 51.3%. It can therefore be concluded that the results do not differ significantly when based on weighted and unweighted aggregate data.

57. The closeness of Aer Lingus and Ryanair is also confirmed by another survey, conducted for Aer Lingus. The majority of respondents, irrespective of gender, age, social class or region of residence cited either Ryanair or Aer Lingus as their preferred airline of choice in routes departing out of Ireland for both UK and European destinations. Moreover they were also the number one and number two airline in terms of overall spontaneous awareness and awareness of routes offered to UK and European destinations. Indeed the evidence led the survey analysts to compare both airlines with each other with respect to the most relevant factors that influence choice, without making such comparisons against any other airline (25).

Conclusion

58. On the basis of the above it can be concluded preliminarily that Aer Lingus and Ryanair are each others closest competitors.

59. Furthermore, absent the merger, it is likely that they remain the closest competitors on routes to/from Ireland. Given its base in Dublin and Cork as well as its presence in Shannon, Aer Lingus is a far more committed competitor to Ryanair than destination based carriers.

FINAL QUESTIONNAIRE

2ND FEBRUARY 2007

TNS mrbi have been commissioned to conduct research amongst air passengers on a number of specified routes departing from Dublin airport.

Please complete the questions below as fully as you can. If you have any queries, please speak to the person who gave you the questionnaire for clarification.

All completed questionnaires should be returned to a representative of TNS mrbi before boarding.

Your co-operation is greatly appreciated.

<emplacement tableau>

List of routes from which the sample was chosen

Dublin-London: must be covered

Category A: routes where both carriers operate into the same airport ?

??Dublin-Birmingham ?

??Dublin-Edinburgh ?

??Dublin-Newcastle ?

??Dublin-Berlin ?

??Dublin-Poznan ?

??Dublin-Turin

Category B: routes on which Aer Lingus and Ryanair operate into different airports

??Dublin-Brussels ?

??Dublin-Glasgow ?

??Dublin-Rome ?

??Dublin-Alicante ?

??Dublin-Hamburg/Luebeck ?

??Dublin-Toulouse/Carcassone ?

??Dublin-Venice ?

??Dublin-Lyon ?

??Dublin-Tenerife

Category C: routes on which also other carriers operate ?

??Dublin-Manchester ?

??Dublin-Paris

??Dublin-Frankfurt/Hahn ?

??Dublin-Barcelona ?

??Dublin-Warsaw ?

??Dublin-Malaga ?

??Dublin-Madrid ?

??Dublin-Faro ?

??Dublin-Vienna/Bratislava ?

??Dublin-Riga

ANNEX II:

Customer Survey - Main Results

M.4439 - Ryanair/Aer Lingus

1. Customer preferences and behaviour:

<emplacement tableau>

Table A.1.1.: % of customers buying return tickets (Question1)

Table A.1.1. summarises the responses to Q1: "Did you buy a return ticket or a one-way ticket for today's trip?" ?

??Ryanair: 81.6% of customers flying with Ryanair bought a return ticket. 16.4% bought a one-way ticket; ?

??Aer Lingus:: 84.4% bought a return and 14.5% a one-way ticket.

<emplacement tableau>

Table A.1.2. % of customers returning with the same airline to Dublin (Q5)

Table A.1.2. summarises the responses to Q5 (and Q6). ?

??Ryanair: 79.9% of passengers fly with the same airline both ways ?

??Aer Lingus: 78.6% of passengers fly with the same airline both ways

<emplacement tableau>

Table A.1.3. Purpose of the trip (Q3) ?

??Ryanair and Aer Lingus have about 20% and 25% business customers, respectively.

<emplacement tableau>

Table A.1.4.: % of passengers who would purchase cheaper ticket from other airline (Q8 (a))

??90.6% of Ryanair passengers would have bought a cheaper ticket from another airline; ?

??66.1% of Aer Lingus passengers would have bought a cheaper ticket from another airline; ?

??The passengers of the 3 network carriers are the least price sensitive ones.

2. Origin of passengers

<emplacement tableau>

Table: A.2.1. % of passengers who started journey in Dublin for each destination (Q4) ?

??On most routes more that 2/3 of passengers start their journey in Dublin. The notable exceptions are Dublin to Gatwick, Newcastle, Glasgow.

Customers by residence

<emplacement tableau>

Table A.2.2.: Carriers' passengers by residence ?

??For both Ryanair and Aer Lingus, about 60% of their passengers have their residence in Ireland.

<emplacement tableau>

Table A.2.3.: Carriers' passengers by residence and destination ?

??Example: 75% of Aer Lingus passengers flying to Heathrow have their residence in Ireland; 23% in London.

3. Belfast

<emplacement tableau>

Table A.3.1. % of passengers who would ever consider flying from Belfast (Q. 10)

Q 10 asked: "Would you ever consider flights to/from Belfast as an alternative to using Dublin airport". ?

??Only 16.6% of all passengers would ever think about using Belfast in the future. ?

??about 2/3 of all passengers respond that they would never consider Belfast as an alternative.

The low proportion of respondents willing to ever consider Belfast for departure confirms the Commission's assessment that routes out of Belfast do not provide a sufficient competitive constraint to routes out of Dublin. One should note that this question is drafted to capture all possible instances of substitution, whether actual or prospective: First, the question is forward looking since it covers the possibility that, for instance subsequent to the merger, somebody establishes a large base in Belfast and offers more and perhaps more attractive (in terms of price and quality) services than what is provided today at Belfast airport. Second the question is not related to any specific route but instead it directly assesses the scope for substitutability between airports. A respondent would answer positively even if he or she considered using Belfast as an alternative to travel to any destination.

Belfast - current routes

<emplacement tableau>

Table A.3.2.: % of carriers' passengers considering flying easyJet from Belfast (Q8)

Table A.3.2. lists all routes on which easyJet provides a service from Belfast to the point of destination. The last column provides the number and % of customers who would consider flying with easyJet. It does not cover other carriers which also fly from Belfast.

??On these routes between 0 and 4.7% of passengers flying from Dublin would also consider flying with easyJet from Belfast.

??Only Aer Lingus flies to Heathrow. About 26% of all Heathrow passengers would consider flying also with Ryanair from Dublin to another London airport. 0% would consider flying with easyJet from Belfast to London.

??On the other London routes, only Ryanair flies from Dublin. easyJet flies to these London airports while Aer Lingus does not. Still 24-46% of total passengers would consider flying Aer Lingus (to Heathrow) while only 0-4.7% consider flying with easyJet from Belfast.

4. Closeness of competitors

<emplacement tableau>

Table A.4.1.% of carrier's customers who have considered flying with another airline to the point of destination (Q8)

This table has bundled routes according to what airlines are flying:

(a) London: This is considered to be a distinct category. Ryanair and Aer Lingus are flying on this route, but into different airports at destination. Other carriers also fly - some of them fly into the same airport as either Ryanair or Aer Lingus.

(b) Duopoly routes-same airport: Only Aer Lingus and Ryanair are flying from Dublin. At destination, they fly into the same airport. This bundle includes the following routes: Birmingham, Newcastle, Berlin, Poznan.

(c) duopoly - different airport: Only Aer Lingus and Ryanair are flying on these routes from Dublin. At destination, they fly to different airports. This bundle includes: Brussels, Glasgow, Rome, Tenerife.

(d) oligopoly routes On these routes, Ryanair and Aer Lingus compete with another airline. This includes: Paris,

Frankfurt/Hahn, Vienna/Bratislava, Riga.

* Overlap routes:

- Where both airlines operate into the same airport at destination, the highest percentage of passengers has considered the other airline; 54.2% of Ryanair customers have considered Aer Lingus and 51.8% of Aer Lingus customers have considered Ryanair. ?

??On oligopoly routes:

- 40.4% of Ryanair passengers also consider Aer Lingus while 17.4% consider other competitors.

- 32.5% of Aer Lingus passengers have considered Ryanair compared to 15.7% who have considered other competitors. ?

* London: Aer Lingus and Ryanair each face a third airline which flies into the same London airport,

- 33.9% of Ryanair customers have considered Aer Lingus and 14.6% have considered others.

- 18.4% of Aer Lingus customers have considered Ryanair and 9.2% have considered others.

Closeness of competitors: by passenger group

<emplacement tableau>

Table A.4.2.: % of carrier's customers having considered a competitor - by customer group (Q8 and Q3) ?

??For each individual passenger group, for both Aer Lingus and Ryanair, the highest % of customers has considered the other party; ?

??Among Ryanair customers, those travelling for business purposes are most likely to consider Aer Lingus an alternative;

??Among Aer Lingus customers, business passengers are the most likely to consider a Ryanair.

<emplacement tableau>

Table A.4.3.: Ryanair customers' preferences and closeness to Aer Lingus: (Q8 and Q9) ?

??44.4% considered Aer Lingus as an alternative, as opposed to 14.7% who considered carriers other than Aer Lingus. ?

??The same pattern is apparent whether the respondent's preference for Ryanair is due to the fact that it offered a good departure time, the destination airport was conveniently located or punctuality.

Closeness of competitors: by loyalty

<emplacement tableau>

Table A.4.4.: of carrier's customers having considered a competitor - by loyalty (Q8)

Closeness of competitors: by carrier

<emplacement tableau>

Table A.4.5.: % of carrier's customers having considered a competitor (Q8)

5. Dublin-London

The survey covered the following flights: ?

??Two flights for Aer Lingus to Heathrow ?

* Two flights for Ryanair to each of its 3 London airports: Gatwick, Stansted, Luton ?

??Two flights at Heathrow by bmi ?

??Two flights of British Airways to Gatwick ???4 flights of CityJet/Air France to London City

Total No of replies: 548

<emplacement tableau>

Table A.5.1.: % of passengers who considered a competitor on routes to London (Q8) ?

??18% of Aer Lingus passengers (at Heathrow) have considered Ryanair as an alternative; ?

??45% of Ryanair passengers at Gatwick have considered Aer Lingus (at Heathrow) as an alternative; ?

??30% (25%) of Ryanair passengers at Stansted (Luton) have considered Aer Lingus as an alternative; ?

??46% of CityJet/Air France passengers have considered Aer Lingus and 17% have considered Ryanair as an alternative. ?

??49% of British Airways passengers (Gatwick) have considered Aer Lingus (Heathrow) as an alternative; ?

??33% of bmi passengers (Heathrow) considered Ryanair as an alternative

Ryanair: ?

??Gatwick: 45% consider Aer Lingus and 26% all other carriers as an alternative; ?

??Stansted and Luton: the % for Aer Lingus is significantly higher than for all other carriers together; ?

??More passengers of all other carriers consider Aer Lingus to be an alternative than Ryanair, in spite of the fact that Aer Lingus only flies into Heathrow and Ryanair into 3 different London airports.

Closeness of airport

<emplacement tableau>

Table A.5.2. Closeness of airport to final destination (Q9) ?

??31.8% of passengers who fly to Heathrow consider Heathrow airport to be the closest airport to their final destination. ?

??London City: it is 72.4%; ?

??73.7% of passengers flying to Luton consider this airport to be the closest to their final destination.

6. Dublin-Frankfurt

<emplacement tableau>

Table A.6.1. % of carrier's passengers flying to Frankfurt who have considered a competitor (Q8)

On that route, Ryanair flies into Frankfurt-Hahn and Aer Lingus/Lufthansa fly into Frankfurt main airport. The latter is Lufthansa 's main hub. Thus, Lufthansa can be expected to have a considerable number of transfer passengers on the Dublin-Frankfurt route. ?

??66.7% of Ryanair passengers have considered Aer Lingus; 33.3% of Aer Lingus passengers have considered Ryanair. ?

??47.2% of LH passengers have considered Aer Lingus and 5.7% have considered Ryanair. ?

??About 20% of Aer Lingus passengers have considered Lufthansa.

Dublin-Frankfurt

<emplacement tableau>

Table A.6.2. % of carrier's passengers flying to Frankfurt who have considered a competitor - by passenger category (Q8) ?

??90% of Ryanair business customers have considered flying with Aer Lingus and 9% flying with Lufthansa; ?

??10% of Lufthansa business customers have considered Ryanair (63% Aer Lingus), but no leisure customer has done so; ?

??Fewer Lufthansa leisure customers than Lufthansa business customers have considered flying Aer Lingus or Ryanair (perhaps the effect of transfer?)

Dublin-Brussels

<emplacement tableau>

Table A.6.3. % of carrier's passengers flying to Brussels who have considered a competitor (Q8) On Dublin-Brussels, Aer Lingus flies to Brussels-Zaventhem and Ryanair flies to Charleroi: ?

??38.4% of Ryanair passengers have considered Aer Lingus ; ?

??60.6% of Aer Lingus passengers have considered Ryanair

<emplacement tableau>

Table A.6.4. % of carrier's passengers flying to Brussels who have considered a competitor - by category(Q8)

Dublin-Glasgow

<emplacement tableau>

Table A.6.5. % of carrier's passengers flying to Glasgow who have considered a competitor (Q8)

??14.8% of Ryanair passengers consider Aer Lingus and 33.8% of Aer Lingus passengers consider Ryanair.

Dublin-Riga

<emplacement tableau>

Table A.6.6. % of carrier's passengers flying to Riga who have considered a competitor (Q8)

??36.4% of Ryanair passengers consider Aer Lingus and 50.0% of Aer Lingus passengers consider Ryanair.

??18.2% of Air Baltic passengers have considered Aer Lingus and 21.2% have considered Ryanair.

Dublin-Vienna/Bratislava

<emplacement tableau>

Table A.6.7. % of carrier's passengers flying to Vienna/Bratislava who have considered a competitor (Q8) ?

??28.4% of Ryanair passengers considered Aer Lingus and 37% considered other carriers ?

??29.3% of Aer Lingus passengers considered Ryanair and 16.2% considered other carriers. ?

??23.5% of SkyEurope passengers considered Aer Lingus and 42.9% have considered Ryanair.

Dublin-Paris

<emplacement tableau>

Table A.6.8. % of carrier's passengers flying to Paris who have considered a competitor (Q8)

??37.2% of Ryanair passengers considered Aer Lingus and 4.3% considered other carriers. ?

??25.9% of Aer Lingus passengers considered Ryanair and 11.1% considered other carriers. ??28.8% of Air France/CityJet passengers considered Aer Lingus and 10.2% have considered Ryanair.

Dublin-Tenerife

<emplacement tableau>

Table A.6.9. % of carrier's passengers flying to Tenerife who have considered a competitor (Q8)

??32.3% of Ryanair passengers considered Aer Lingus and 33.3% of Aer Lingus passengers considered Ryanair.

Dublin - Rome

<emplacement tableau>

Table A.6.10. % of carrier's passengers flying to Rome who have considered a competitor (Q8)

??50.5% of Ryanair passengers considered Aer Lingus and 55.6% of Aer Lingus passengers considered Ryanair.

Notes

1 Association of Irish Market Research Organisations.

2 See appendix.

3 See Ryanair's Response to the Statement of Objections of 17 April 2007, for example, paragraph 69 or 370

4 From a total number of 2674 responses 443, (that is a 16.6%) mentioned that they had not arranged the flight. Of these 443 only 20 (4%) also mentioned that they had considered another airline. This means that only in the case of 20 out of 2231 (0.9%) the respondent may not have had accurate information with respect to the alternative that were considered when making the booking.

5 Ryanair also argues that "it is not correct to say that passengers were "interviewed" and the use of this expression gives more weight to the results than is due to a self completion questionnaire" (See Ryanair's Response to the Statement of Objections, paragraph 381). However, Ryanair itself uses this term repeatedly, for example four paragraphs above in 377, in 376, 374 etc.

6 See Ryanair's Response to the Statement of Objections for example, paragraph 69 or 377

7 Ryanair claims that most of its passengers originate outside Ireland whereas most of Aer Lingus passenger originate in Ireland. The customer survey does not reveal, however, stark differences. As shown in the table below 66.2% of Ryanair passengers and 69.8% of Aer Lingus passengers originate in Dublin. If data is weighted as explained below the respective values are not very different (63.8% and 70.7% respectively).

8 See Ryanair's Response to the Statement of Objections, paragraph 376.

9 See Response to the Statement of Objections, paragraph 70, first sentence.

10 The confidence interval is [38.0% - 43.7%] at a 95% confidence level. The confidence interval is the plusor- minus figure usually reported in newspaper or television opinion poll results. The confidence level represents how often the true percentage of the population who would pick an answer lies within the confidence interval. In this case one can be 95% certain that the true percentage of Ryanair passengers that considered Aer Lingus as an alternative lies between [38.0% - 43.7%]. The Commission does not report confidence intervals for all percentages mentioned in this decision since given the high number of responses obtained the confidence intervals are sufficiently narrow in each case so as to not affect the given interpretation.

11 The Commission used for simplicity a standard test of significance of the difference between two independent proportions. The difference is also significant at the 1% level if responses are excluded that mention both Aer Lingus and another carrier as considered alternatives.

12 The confidence interval is [38.2% - 44.2%] at a 95% confidence level.

13 All differences in percentages reported in this section are significant at least at the 5% level.

14 Passengers travelling with airlines other than EI or FR considered more often EI than FR. This is consistent with the hypothesis that Ryanair is less constrained by airlines other than Aer Lingus than is Aer Lingus.

15 See Ryanair's Response to the Statement of Objections, paragraph 71.

16 Similarly neither party expressed any concern with respect to the use of the word "route". Ryanair in its response to the Statement of Objections (paragraph 387, 388 and 389) claims that it is imprecise as respondents may not be aware that there may be other airports at the destination city. This, however, would only bias the responses towards less substitutability, not more. In any event it would be inappropriate to regard this as a bias since what is of interest is precisely whether the passengers regard other carriers as substitutes. One reason they might not is that they are unaware that a rival is flying to an airport nearby to the destination city. In any event, the respondent had the possibility to ask the interviewer for clarification if needed.

17 For further reassurance the Commission has checked whether on the basis of the names given of alternative carriers it can be inferred that the respondents had a good knowledge of the presence of other competitors in the route. On the 7 routes where there are no competitors other than Ryanair and Aer Lingus less than 5% of respondents mentioned they had considered carriers that were not in fact operating a point-to-point to service to the destination city. In practically all cases the carrier mentioned was either the flag carrier in the country of destination, a charter airline serving the country of destination (but the destination city in that route) or an airline that had cancelled its service on that route only recently. In the case of the London routes except for very few respondents that mentioned easyjet (less than 4% in total) all respondents mentioned only carriers that are currently serving London from Dublin. Finally with respect to the other 5 routes in which there is one competitor to the Merging Parties the large majority of respondents indicated they had considered other airlines cited the one that is currently serving the route (except in the case of Glasgow were no respondent mentioned Loganair but 1 mentioned British Airways and 3 easyjet out of 125 respondents).

18 See table A.4.4 in Annex II.

19 Similar proportions apply in the case of Aer Lingus passengers. Of those indicating that they always travel with Aer Lingus 21% had considered Ryanair and 8% other airlines). In contrast 68.9% of passengers that do not always travel with Aer Lingus considered Ryanair (and 16% other airlines)

20 This is an example of a stratified sample. The population is exhaustively divided into two or more strata, and then a simple random sample is drawn from each stratum, not leaving any stratum aside. This makes more likely that cases are evenly selected from all the strata, thus reducing the possibility that the sample is disproportionately concentrated on one part of the population. Also, if the stratification is somehow related to the variables of interest, then the variance of those variables within each stratum will be lower than the overall variance of the population at large. This reduces overall sampling error.

21 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 70, second indent.

22 See Ryanair's Response to the Statement of Objections of 17 April 2007, paragraph 374

23 The raw data was gathered by the Commission and therefore was not confidential in any way. To generate the weights it is possible to use as an approximation the number of yearly passengers on a given route (this is the approach followed by the Commission as explained below).

24 Proportional weights and can be designated as Pk with the general form Pk =% of stratum in population/% of stratum in the sample: (formul) ??Where Nk is the size of the population strata k, nk is the size of the sample from population strata k, n is the sample size and N is the population size. The sampling ratio is fk=nk /Nk.

25 Response of Aer Lingus of 28 November 2006 to Additional Questionnaire, Annex IV.[CONFIDENTIAL TO AER LINGUS].

ANNEX III:

Price Correlation Analysis for Geographic Market Delineation M.4439 - Ryanair/Aer Lingus

1. Use and relevance of correlation analysis

1. The purpose of price correlation analysis is to identify the extent to which two price series moved together over time. If two products or geographic areas belong to the same market, their prices will not move indefinitely far from each other in the long run. The economic intuition is a simple arbitrage argument: if the products are very close substitutes, either on the demand or on the supply side, their prices cannot move too far apart, since either consumers or producers will shift between them in such a way as to eliminate the more expensive one from the market.

2. However, absolute price convergence is not necessary for products or services to belong to the same geographical market. Rather it is of importance for market definition to see whether the price behaviour of one product transmits to other in other areas. Two price series for separate geographical areas belong to the same market if the difference between them is stable over time. This notion is closer to the SSNIP methodology. It is the comovement of prices rather than the law of one price that is of importance in market definition.

3. Price correlation is widely recognised as a useful indicator of whether two products should be considered as forming part of the same market. The Commission Notice on market definition states that for defining product markets:

4. "There are a number of quantitative tests that have been specifically designed for the purpose of delineating markets. These tests consist of... tests based on similarity of price movements over time..."

5. "Generally the same quantitative tests used for product market definition might well be used in geographic market definition..."

6. Price correlation analysis is a technique that has been used in the past by the European Commission in merger cases to establish the extent of the relevant market (1).

7. The correlation between two variables A and B measures the degree of contemporary linear association between them. The correlation between the price of a good in two areas can be influenced by a number of factors, including variations in production costs, demand, the competitive environment, and the price of the good in the other area. It might be anticipated that the closer the movement of prices between two markets, the more these markets are integrated.

8. The extent of co-movements between two price series can be summarised in their correlation co-efficient. The correlation coefficient will range from minus-one to plusone. A correlation coefficient of minus one would indicate that the two price series moved in perfect opposition to each other such that whenever one price rose the other fell. A correlation coefficient of zero suggests that there was no relationship between the two price series under investigation. A correlation coefficient of one would indicate perfect co-movement between the two price series, such that whenever one price rose the other rose as well. It is rare to find perfect co-movement between two price series and a correlation coefficient of one is rarely observed. However, the higher the correlation coefficient between two price series the higher is the degree of co-movement between them.

9. A high and significant correlation coefficient close to unity may indicate that the two candidate markets belong to the same relevant market. A low or insignificant correlation coefficient may indicate that the two candidate markets do not belong to the same relevant market.

2. Role of price correlation analysis in the present case

10. In the context of this case the Commission must assess whether the geographic market definition of short-haul routes out of Dublin should include one or more destination airports in the proximity of each other. For every airport-pair and city-pair considered the Commission has gathered extensive qualitative evidence to make such assessment (see Market Definition section in the Statement of Objections). Price correlation analysis can complement this evidence. However, price correlation is not a determinative test for market definition. The result of a price correlation test is not credible on its own unless it fits into a consistent and intuitive assessment or can be confirmed by other studies or observations.

11. The interpretation of price correlation analysis in this case is as follows: If the price in airport-pair A is raised relative to the price in airport-pair B, can passengers shift from A to B or can airlines re-route planes from B to A? If there is a strong correlation between the prices in the airport-pairs, then this may indicate that the two are linked by arbitrage opportunities (when, for instance, passengers or airlines can move from one airport-pair to the other without incurring large transportation or transaction costs, respectively), in this case, following a price increase in airport-pair A, prices in A should fall back rapidly to their initial level. This could be an indication that the two areas should be part of the same relevant geographic market. If areas A and B are in different markets, then price changes in A should not affect prices in B. That is, the correlation between the prices in the two airport-pairs should be low. Thus, the likelihood that two airport-pairs are part of the same relevant geographic market increases as the correlation between the prices of the two airport-pairs increases.

12. The closer the correlation coefficient is to 1 the higher (contemporaneous) association exists between the two price series. There is no generally agreed level or threshold which defines whether series are moving sufficiently together for two airport-routes to belong to the same geographical market.

13. A partial, but generally accepted, response to this critique is the use of benchmarking techniques. As a benchmark against which to compare other correlations, one can use the correlation coefficient between the price of two series which one is willing to state on a priori grounds lie in the same relevant market. If the correlation coefficient between the two other products lies above the benchmark this can be interpreted as meaning that these two products lie in the same relevant market. In this case we use as a first benchmark the average correlation on the routes where both Aer Lingus and Ryanair fly to the same airport.

3. Limitations

14. The Commission acknowledges that price correlation analysis is subject to a number of limitations.

15. First a high correlation coefficient may suggest that markets should be widely defined when in fact the correlation is spurious. Spurious correlation occurs when two series seem to be correlated but in fact are not. The correlation in this case is a 'coincidence' and is not the product of any interrelation between the two products. There are several sources of spurious correlation:

(i) Price series may display serial correlation. This means that prices charged in the past could affect the level of prices today. If this is the case, then the correlation between two variables will be driven to some extent by each of the prices own serial correlation. In this case, the resulted correlation between the two will be affected by the relation of each the prices with their past values and lead, in this way, to erroneous conclusions. In certain cases, series of price are integrated of order 1, which means that the level of the price at certain point in time affects all the all future values reported by this variable. This problem can be corrected to a large extent by measuring correlations of the first or second difference in prices.

(ii) Spurious price correlation arises when two price series are subject to a common influence. For example, the price of fuel is an important determinant of the cost of a flight. If this common factor is not accounted for, then the correlation coefficients of air fares between two airport-pairs would be biased upward leading potentially to a finding of an overly broad market. Alternatively a high simple correlation can occur as a result of a common seasonal pattern in the series without the single series being directly related to each other.

(iii) Correlation techniques may become invalid if applied to variables that are not stationary. It is therefore important to determine whether an economic time series is stationary or not. A price series is said to be stationary if its mean, variances and co-variances remain constant during the time of reference (2). A price series is nonstationary if it fails to satisfy any part of this definition. Large sample theory breaks down whenever any of the explanatory variables in a regression equation is non-stationary.

16. Second a low correlation coefficient may lead to a narrow market definition when in fact the two price series are related but subject to significant random disturbances which break such correlation. This can arise because:

(i) Price correlation can only measure contemporary linear association between two variables. When prices respond to changes in market conditions with a time lag, the correlation between contemporaneous prices will not capture the response of the other goods' prices to the price change of any given good. Even though the markets are connected, one will observe little correlation. In this case, the use of price correlation tests would cause markets to be defined too narrowly.

(ii) If the 'noise to signal' ratio is high, one will also observe little correlation between the prices. But this result will be driven by random short lived shocks to the prices of the product and the apparent lack of correlation will not reflect the underlying structural relation between the products. For instance, suppose the inputs are really different for the two goods and input prices move around a lot. Then correlation will be small due to the high variance in the price series caused by shocks to input prices even though the two series may exhibit some co-movement. When the size the shocks is large relative to the movement of the price series over the period observed, this problem will be exacerbated since the 'noise to signal' ratio will be high.

17. Finally the Commission observes that no unique criterion exists to determine whether a correlation is large enough to place two areas in the same relevant geographic market. However, as mentioned above, in some cases it is possible to obtain non-arbitrary guidance by benchmarking correlations between candidate markets on correlations between markets are known to be unintegrated.

18. In any event, as argued by Bishop & Walker:

"...despite these weaknesses, price correlation analysis if used and interpreted correctly can provide useful information to aid the definition of the relevant market. In particular, given its relatively low information requirements and ease of use, it would be foolish to ignore a technique that can potentially provide useful information. Instead, practitioners must be fully aware of what the potential shortcomings of correlation analysis are and how these affect the interpretation of the results".

4. Methodology

19. The Commission relies on data on average monthly prices across a number of routes between Dublin (and Cork & Shannon) and other European destinations where there are potential horizontal overlaps, either because the merging parties fly to the same airport or because they fly to different airports in the same geographical area.

20. The relatively high level of data aggregation (i.e. monthly as opposed to weekly or daily) (3) implies that for a number of more recent routes of interest where Ryanair and Aer Lingus overlap we have insufficient observations to obtain any meaningful results. This is the case for Berlin, Bilbao, Bologna, Fuerteventura, Hamburg, Krakow, Lyon-Grenoble or Madrid.

21. Statistical tests have been proposes to correct for the possible existence of spurious correlation in the price series. The methodology followed in this analysis by the Commission implements several of these techniques in order to minimise the risk of defining markets too widely. As a result in a number of markets the results obtained are inconclusive. The Commission has not been able to test whether the existence of random disturbances may have systematically lowered an otherwise high correlation in prices. However, it believes the risk of defining markets too narrowly is mitigated by the use of benchmarking techniques.

4.1. Stationarity tests

22. In the analysis of price correlations it is important to discriminate between price processes being either stationary or non-stationary. A series of prices that are integrated of order lower than 1 are said to be stationary. Price series integrated of order one are nonstationary.

23. In order for price comparisons to make sense it is essential that the price series to be compared are of the same order to integration. If the price series under scrutiny are of different orders of integration it is unlikely that the underlying products or areas belong to the same geographical market.

24. Calculation of the price correlation coefficient raises no concerns when the underlying price levels are stationary. However, when the series are non-stationary this would result in spuriously high correlation coefficients even though the series may not be significantly correlated. Notwithstanding this, when price levels are integrated of order 1 it still makes sense (and is valid statistically) to calculate the correlation coefficients of the price changes, that is, correlations of the transformed series (formul). In this situation the economic interpretation of the correlations differ because it is the price changes rather than the price levels that potentially will correlate. In this case, it is relative convergence that is being tested. A "high" correlation of price changes across products or services for different geographical areas still indicates that these prices co-vary and hence suggests products belong to the same geographical market.

25. An alternative method is to statistically test whether non-stationarity might be a problem. To do so we compute an "Augmented Dickey-Fuller' (ADF) test for each price series to see whether each price series is non stationary (4). Table 1 reports the results of the ADF tests for the individual raw price series for each carrier and airport. Stationary series to a 5% level of significance are in bold. It is apparent that half the series are likely to be nonstationary.

26. We use the ADF test to see whether the relative prices are stationary. If the hypothesis that the two individual series are stationary is rejected but the relative price series is stationary then we can claim that the result is equivalent to getting correlation between stationary prices. An alternative is to apply the ADF test to the log of the ratio of the price series (5). If two products are in the same market one would expect the null hypothesis that the log of the ratio of the price series is non-stationary to be rejected. Table 2 reports the results for this analysis. In most cases the hypothesis of non-stationarity is rejected. It cannot be rejected in a number of airport pairs of interest in particular: Amsterdam- Eindhoven, Barcelona-Reus and Frankfurt-Hahn. For other pairs the number of observations is small both for the ADF test to be meaningful and to run price correlations.

4.2. Controlling for seasonality common input costs

27. A key concern prior to any test of co-movement is to purge price series for all correlation that is caused by common factors unrelated to competition, e.g. inflation, exogenous temporal variation or prices on key inputs common for the two candidate markets. If common factors are not purged, correlation may be spuriously high with a serious risk of drawing erroneous conclusions about the size of the relevant market.

28. For stationary time series, the proper measure for co-movement of prices is the partial correlation coefficient, not the ordinary correlation coefficient. The partial correlation coefficient differs from the ordinary correlation coefficients, as the former, but not the latter, has been purged from the influence of common factors. The partial correlation coefficient (PAC) is calculated by regressing all price series on time series of the price of fuel using OLS. The residuals from this regression are considered as a measure of prices purged for common factors and the partial correlation coefficient is the ordinary correlation coefficient between the residuals from two such regressions (6).

29. We also correct for seasonality by introducing quarterly dummies in the least squares regression. This is done in order to account for possible seasonal variation in the data. Several other factors could have been of relevance to include, but due to lack of such data an extended analysis is omitted in the present exposition. In general, the partial correlation coefficients should be preferred to the standard simple correlations.

4.3. Serial correlation

30. We found, using the Durbin-Watson test, that the residuals of the OLS regression were serially correlated - this implies past values of wholesale prices influence current values. This is usually an indication that there is some sort of misspecification in the model, e.g., that explanatory variables have been omitted. By including lagged values of wholesale prices in our estimations, we corrected for the problem of serial correlation.

4.4. Steps in our analysis

31. We follow two complementary approaches. Both approaches generate similar results and do not affect the conclusions.

I. Correlations on price levels

* Plot the candidate price series

* Compute the raw price correlation coefficient

* Regress the price series on the common influence variables (fuel costs and seasonal dummies).

* Regress the price series on the common influence variables (fuel costs and seasonal dummies) and on the lag of prices and

* Correlate the residuals from the above regressions

II. Correlations on first differences

* ADF test of stationarity on the individual price series (results reported in table 1)

* ADF test of stationarity on the log ratio

* First Strategy

- Take the first difference of the price series and the common costs

- Regress the first difference of the price series on the first difference of the common costs and seasonal dummies

- Correlate the residuals from the above regression (results in column FD-1 in table 2)

* Second Strategy

- Regress the price series on the common influence variables (fuel costs and seasonal dummies)

- Take the first difference of the residuals from the regression above

- Compute the correlation coefficient of the first differences of the residuals (results in column FD-2 in table 2)

32. The results are reported in Table 2. The first column indicates the main city close to the airports that are in the same area. The second and third columns refer to the airport that Aer Lingus (EI) serves. Columns 4-7 indicate the airport or airports served by Ryanair. Column 8 reports the number of observations (monthly average net prices). Column 10 reports the ADF test on the log price ratio. Column 11 indicates whether the airport pair identified in the same row corresponds to Aer Lingus and Ryanair flying to the same airport (S), to a different airport (D) or Ryanair is flying to two different airports (R). The next columns report the price correlations coefficients: the column labels RAW indicates the correlation between the raw price series; SC reports the partial correlation coefficient controlling for seasonality and common costs; SCL reports the partial correlation coefficient controlling for seasonality, common costs and including a lagged variable; RAWFD reports the raw correlation coefficient of the price first differences The column labelled FD-1 reports the correlation following the first strategy presented above. Finally the last column FD-2 indicates the partial correlation in price differences controlling for seasonality and common costs.

33. Our preferred approach is the second one given that in many routes the price series are likely to be non-stationary (on the basis of the ADF test). In particular, we report in detail the results from the second strategy described above (FD-2) but point out that although the correlations are generally lower in the case of the first strategy the qualitative results do not change. In the Market Definition section of the SO we only refer to the correlations in the last column (FD-2) in table 2.

5. Overview of results

34. The Commission has calculated correlation coefficients for a number of related airportpairs where sufficient time series data was available. Price correlations are of three types. A first category of price correlations concerns the situation where both Aer Lingus and Ryanair both serve the same airport. A second category concerns situations where Ryanair serves two different airports in close proximity to a given destination. Finally, the third category refers to the situation where Ryanair and Aer Lingus serve different airports close to a given destination. The results presented here refer only to the last correlation coefficient in table 2, namely the correlation in first differences controlling for common costs and seasonality.

35. We use as a first benchmark the average correlation on the routes where both Aer Lingus and Ryanair fly to the same airport. In our data we have 6 such routes. In five of these airports the correlation coefficients are high ranging from 0.82 in Malaga to 0.43 in Manchester. The average correlation is 0.69 (7). Given that the price series belong to Ryanair and Aer Lingus these high correlations in prices also provide a first indication that the merging parties are close competitors in these routes.

36. First we point out that, in general, the amount of correlation observed in the data cannot be explained by common variation in the costs. In some instances correcting for common costs leads to higher values of the correlation coefficient. This is often the case when common costs are observed to move in the opposite direction to the two price series.

37. As regards Manchester where both airlines also fly to the same airport the correlation is weaker at 0.43 but still statistically significant. This provides a within route benchmark for pairwise comparisons between Manchester airport and other adjacent airports. Ryanair's own fares to the two adjacent airports, Leeds and Liverpool, are very high around the 0.69 general benchmark and higher than the 0.43 within route benchmark. In theory there is no reason to expect that the correlation between Ryanair's own fares to different but adjacent airports should be always higher than fares from two different competitors to the same airport. On the one hand Ryanair's service to two different adjacent airports is likely to be a closer substitute than Aer Lingus. On the other hand if the two airports were not related one would not expect even Ryanair's prices to be positively correlated as even Ryanair would price in routes to one airport independently of routes to another airport. Nonetheless, it is more likely that certain (common) factors (besides costs) induce a higher correlation. In any event the highly significant ADF test certainly suggests that at least from Ryanair's perspective pricing in these three airports is highly interdependent (8).

38. In the case of routes where the merging parties fly to different airports and there are no other airports in the area the correlation coefficient is higher than the 0.69 benchmark in the case or Alicante, Barcelona, Rome, Paris and Toulouse. The ADF test on the log ratio is also highly significant. At the very least this result does not contradict the qualitative evidence that strongly suggests all the above routes constitute city-pair markets. With respect to Newcastle the correlation coefficient is also high and significant at 0.80 thus suggesting that Newcastle and Durham Tees airports are indeed part of the same market.

39. In contrast the correlation analysis is inconclusive with respect to Amsterdam and Eindhoven. The number of observations in small but the correlation coefficient (0.46) is only significant at the 10% level (p=0.056) and rather far from the benchmark. Also the ADF test cannot reject the hypothesis of non-stationarity.

40. As regards B'ham and East-Midlands the correlation between Aer Lingus and Ryanair fares is also below the benchmark at 0.45. However it is significant at the 5% level (p=0.010). Furthermore Ryanair's fares to these two airports are strongly correlated (0.80) which suggests that at least Ryanair regards these airports as substitutable.

41. In the case of Bristol the correlation coefficients suggest that Bristol and Cardiff are in the same market. In fact the correlation coefficient between Bristol and Cardiff at 0.73 is just higher than the correlation between the merging parties prices to Bristol airport. The correlation coefficient of Ryanair fares to Cardiff and Bournemouth and Bristol and Bournemouth is 0.50 and 0.51 respectively - and significant. However the correlation of Aer Lingus fares to Bristol and Bournemouth is insignificant and low at 0.21. This would suggest that Bournemouth is notpart of the same market as Cardiff and Bristol.

42. With respect to Frankfurt and Glasgow the price correlations are moderately high at 0.53 and 0.62 and significant at the 5% level. However these values are below the benchmark. In the Frankfurt case the ADF does not reject the hypothesis of non-stationarity, given that the test is sensitive to small samples and we only have 18 observations. For this reason we regard these results as only indicative of a city-pair market. A conclusion in this respect can only be made on the basis of complementary qualitative evidence.

43. The evidence is also inconclusive with respect to Heathrow and the other three London airports. Even though the correlation in price levels between Heathrow and the other airports is high, the non-stationarity hypothesis cannot be strongly rejected for Heathrow whereas the other three series are likely to be stationary. Indeed, the correlations in first differences are all low and insignificant. However one element that points in favour of Heathrow being part of the same market is the fact that the ADF test on the log of the ratio of the price series suggests the series are integrated. In any event it is important to emphasize that the correlations between the three airports served by Ryanair are all very high, above 0.90 and highly significant. This is strong evidence that Ryanair's pricing at these airports is highly interrelated and thus supports the view that Stansted, Gatwick and Luton are in the same market. As regards the question whether Heathrow is part of the same market more qualitative evidence is required to reach a conclusion.

44. The result for Brussels is odd. We obtain a negative correlation in prices between the two different Brussels airports. Such negative correlation is difficult to explain economically. It is also inconsistent with all other available evidence that suggests these two airports are part of the same market.

45. Finally, the Commission also compute the correlation in first differences of price (adjusted for common costs and seasonality) for Cork-London. Aer Lingus flies to Heathrow in this route whereas Ryanair flies to Stansted and, as of November 2005, also to Gatwick. Concerning the latter, the low number of observations makes the price correlation exercise inconclusive. Considering, indeed, Ryanair schedules on Stansted, there are 55 observations in the series from May 2002 to November 2006. The raw correlation coefficient was 0.74 and significant. However, after correcting for seasonality and common costs, and running the correlation on first differences the correlation coefficient is 0.35 (but remains statistically significant).

46. A very similar picture arises in the Shannon-London route. Here the raw correlation coefficient is 0.79 and statistically significant. However the partial correlation coefficient in first differences is also 0.35. This value is below the benchmark used for Dublin routes and therefore it does not confirm whether Heathrow and Stansted are in the same market in routes out of Cork. However the Commission notes that in both markets the correlation was very strong up to the end of 2004. At this point Ryanair drastically decreased its prices (by about 50%) in both routes. This is likely as a reaction to the entry into these routes of Easyjet operating from Gatwick airport which neither Ryanair (nor Aer Lingus) served from Cork or Shannon (as explained in section 7.10.4 of the SO). In addition to the lowering of fares, Ryanair expanded its capacity in this route by opening a service to Gatwick and Luton airport. Ryanair's prices in all these three airports are correlated (although the number of observations is small). Aer Lingus also decreased its prices at the same time but to a lesser extent. Prices have not varied significantly since (except for a spike in the summer of 2006) and has reduced decreases the scope for correlation.

6. RBB Comments regarding the Commission's price correlation analysis

47. Annex 3 in RBB's Dataroom submission provides comments on the Commission's price correlation analysis. RBB essentially argues that " While the Commission has rightly acknowledged the problems with price correlation analysis and attempted to strip out many common influences on price, we nonetheless find that the Commission's results are severely biased due to the failure to control for important spurious factors - notably common demand shocks - that affect all Dublin routes as well as all similar destinations". In particular RBB refers to a Dublin effect and a demand destination effect.

48. According to RBB the Dublin effect leads to high spurious correlation because "To the extent that the willingness to fly out of or into Dublin varies over time and affects all routes, then a positive price correlation is found between two routes even if they are not part of the same market. A factor that influences all routes out of Dublin over time would be Irish public holidays. An example of what affects air travel into Dublin may be sports events in Dublin".

49. RBB explains the destination demand effect as follows: "Suppose that two airports A and B are close to each other but not part of the same relevant geographic market. It is however still possible that passengers to A and B fly to that region for similar reasons. An intuitive example would be where demand would be higher in summer and lower in winter for all beach destinations. For winter destinations (e.g. ski resorts), on the other hand, the effect is the opposite with high demand in winter and low demand in summer. A more idiosyncratic example would be large sports events that cause an increase in demand for an entire region and thereby affect air fares on all routes to various airports in that region".

50. The Commission has carefully reviewed these arguments but has reached the conclusion that the so-called Dublin effect and demand destination effect are unlikely bias the results for a number of reasons:

51. First with respect to the "Dublin effect" there is no reason to expect that the willingness to fly out or into Dublin will "vary over time and affect all routes". Demand to travel on say the Dublin-London route is not likely to be systematically related over time to demand to travel on the Dublin-Rome route or the Dublin-Faro route, for example. There is simply no evidence that events in Dublin such as sport events are of such relevance that it would significantly affect demand and route prices in all routes to or from Dublin at the same time. Also note that the price correlation compares monthly prices, not daily prices. Average monthly prices are unlikely to be significantly affected by one-time events. Moreover for such events to bias the price correlations they would need to occur on a regular basis through the duration of the sample- there is no evidence of any such events, sporting or otherwise. The Commission points also out that there are no visible spikes in demand due to Irish public holidays (which is unsurprising since data is on monthly averages) but in any event, the seasonality correction would control for any related spurious correlation.

52. Second, the demand destination effect is essentially the same as the Dublin effect but the source of the alleged demand shock is at the other end. As a result all the above arguments would apply. For example note that if the prices in two routes are correlated because they are both winter destinations then the seasonality correction would control for the risk of spurious correlation. Nonetheless the Commission fails to see the relevance of the demand destination effect given that the only price correlations of interest are between Ryanair and Aer Lingus' prices on a route when flying to the same airport and/or to the same city.

53. In any event, and with respect to the more general argument that demand shocks may lead to spurious correlation the Commission makes two remarks: ???First, note that demand driven factors underlying any possible serial correlation are controlled for by first differencing. ???Second and most importantly, if demand shocks were to affect prices on all routes at the same time as RBB claims this would tend to increase all correlation coefficients. In that case the use of benchmarks ensures this would have little effect on the findings since also the benchmark would increase.

54. In conclusion the Commission does not consider RBB's criticisms convincing to dismiss the findings from the price correlation analysis reported above. In any case the Commission points out that it is well known that time-series analysis is not a fool-proof method of determining market and that it should be used in conjunction with more traditional analysis. The Commission' price correlation analysis is indeed only one element that has been taken into account to delineate the geographic market in this case.

Notes

1 See for example Nestle/Perrier (Case IV/M190 [1997] O.J. L356/1), Mannesmann/Vallourec/Ilva (Case IV/M315 [1994] O.J. L102),

2 A series is therefore stationary when a shock that occurs at a certain point in time has not a persistent effect in the following periods.

3 Monthly data, however, appears reasonable to mitigate the problem of delayed response identified above. At least for the case of no-frills airlines such as Ryanair and Aer Lingus price reactions to competitors in the same market are likely to be fast. This is because these airlines sell all or the majority of their tickets in the internet, where the transaction costs of changing prices are very low. Further price comparison software allows each company to monitor and react to the price variations of rivals. Therefore price series in the same market are likely to be related in the short-run, thus justifying our use of net price monthly averages as the reference variable.

4 We follow the common practice of trying the ADF for different number of lagged differences and a trend and conclude in favour of stationarity if the null hypothesis of non-stationarity is rejected.

5 If prices are co-stationary in first differences of the logs, stationarity of the log of the ratio is equivalent to cointegration with cointegrating vector [1 -1], see Forni (2002) "Using Stationarity tests in antitrust market definition" CEPR Discussion Paper DP 3236

6 If the correlation between two price series is simply spurious it is typically seen as the simple correlation being relatively large whereas the partial correlation controlling for common costs is relatively small. In general, however, the partial correlation coefficient can be either smaller or larger than the simple correlation coefficient and the coefficients may not even have the same sign even though one would expect this in most cases analyzing price data.

7 The average correlation in the case of the first strategy for correlations on first differences is 0.59.

8 For 20 months Aer Lingus also served the Liverpool airport. The correlation coefficient between Manchester and Liverpool for Aer Lingus own fares is 0.66 and significant at the 5% level. Interestingly the correlation coefficient between Ryanair and Aer Lingus at the same Liverpool airport is 0.47.

<emplacement tableau>

Correlation coefficients in bold significant at the 99% level of confidence.

Correlation coefficients in italics are significant at the 95% level of confidence.

Average correlations on first differences of the prices series adjusted for seasonality and common costs: (excluding non-significant observations)

RYANAIR AND AER LINGUS FLY TO SAME AIRPORT (6 observations): 0.69 (FD-2)

ANNEX IV:

Regression Analysis Technical report

M.4439 - Ryanair/Aer Lingus

1. Introduction

1. The Commission had access to quantitative data related to the markets affected by the proposed Ryanair/Aer Lingus merger. This annex reports on the regression analysis undertaken by the Commission. It also provides an assessment of the various econometric reports submitted separately by the merging parties during the market investigation.

2. The expert economists of Ryanair and Aer Lingus, RBB and CRA respectively, were given access for a period of three days each to the full data set and computer code used by the Commission in its investigation. Furthermore, each party granted access to its own econometric submission to the expert economists of the other party. Both RBB and CRA presented their "data room" findings in a report submitted as part of their client's respective response to the Statement of Objections. This annex also integrates the Commission's reply to these "data room" submissions.

3. Regression analysis is a statistical tool for understanding the relationship between two or more variables. Multiple regression analyses involves a variable to be explained -called the dependent variable- and additional explanatory variables that are thought to produce or be associated with changes in the dependent variable. The mathematical model of their relationship is the regression equation or regression specification. The dependent variable is modelled as a random variable because of uncertainty as to its value, given values of the independent variables. A regression equation contains estimates of one or more unknown regression coefficients which quantitatively link the dependent and independent variables. The coefficients are estimated from given data.

4. Regression analysis can be useful (1) in determining whether or not a particular effect is present; (2) in measuring the magnitude of a particular effect; and (3) in forecasting what a particular effect would be, but for an intervening event. In the context of the present case this corresponds to (1) assessing, for example, whether the presence of one merging party in a given route has an effect on the fares charged by the other merging party; (2) quantifying the magnitude of such effect; and (3) predicting whether prices would increase on average if the proposed merger is allowed to proceed.

5. In interpreting the results of a multiple regression analysis, it is important to distinguish between correlation and causality. Two variables are correlated when the events associated with the variables occur more frequently together than one would expect by chance. A correlation between two variables does not imply that one event causes the second to occur. Therefore, in making causal inferences, it is important to avoid spurious correlation. Spurious correlation arises when two variables are closely related but bear no causal relationship because they are both caused by a third, unexamined variable.

6. Causality cannot be inferred by data analysis alone - rather, one must infer that a causal relationship exists on the basis of an underlying causal theory that explains the relationship between the two variables. In the context of the present case the underlying theory of harm is that the merger between Ryanair and Aer Lingus may significantly impede effective competition in certain routes by removing important competitive constraints the merging parties exert on each other. The most direct effect of the merger will be the loss of competition between the merging firms, allowing the merged entity to exercise increased market power to the detriment of customers (1).

7. Even when an appropriate theory has been identified, causality can never be inferred directly - one must also look for empirical evidence that there is a causal relationship. Such evidence is reported in section 7 of this Decision, in particular sections 7.3, 7.4.1 and 7.4.2.

8. Thus, the presence of a non-zero correlation between two variables does not guarantee the existence of a relationship; it could be that the regression model does not reflect the correct interplay among the explanatory variables. Conversely, the absence of correlation does not guarantee that a causal relationship does not exist. Rather, lack of correlation could occur if (1) there are insufficient data; (2) the data are measured inaccurately; (3) the data do not allow multiple causal relationships to be sorted out; or (4) the model is specified wrongly. There is a tension between any attempt to reach conclusions with near certainty and the inherently probabilistic nature of multiple regression analysis. In general, statistical analysis involves the formal expression of uncertainty in terms of probabilities. The reality that statistical analysis generates probabilities that there are relationships should not be seen in itself as an argument against the use of statistical evidence (2).

9. It follows that in some cases, regression analysis will lead to inconclusive results. In other cases, the results may validate or refute one or more of the tested hypothesis. In any event, the Commission points out that econometric analysis cannot be determinative by itself. This is for at least two reasons. First, the analyst is often obliged to make technical assumptions about the nature of the studied relationships that may be difficult to validate. Second, the data can be inaccurate, incomplete or insufficient to make robust inferences. Therefore, under a strict cautionary principle the Commission can only attach limited weight to econometric results.

10. Furthermore the results from a regression analysis should never be interpreted in isolation. Econometric results are particularly relevant to the extent that they are consistent with other independent sources of evidence such as internal documents, customer surveys, industry reports, and submissions made by third parties, regulators or other market participants.

11. In interpreting econometric results it is necessary to distinguish between economic and statistical significance. Economic significance means that the magnitude of the effect being studied is not de minimis. Statistical significance is a formal way of assessing whether observed associations are likely to be explained by chance alone.

12. In the context of this case the Commission first formulates a number of hypotheses derived from the theory that the proposed merger eliminates the competitive constraint that Aer Lingus and Ryanair exert on each other. Regression analysis is then used to determine whether the associations of interest are economic and statistically significant. For example it is of interest to evaluate a (null) hypothesis that the presence of Ryanair on a route has no impact on Aer Lingus prices against the alternative hypothesis Aer Lingus will charge lower prices when Ryanair is present on the route. A null hypothesis is a hypothesis set up to be nullified or refuted in order to support an alternative hypothesis. When used, the null hypothesis is presumed true until statistical evidence in the form of a hypothesis test indicates otherwise.

13. In regression analysis the null hypothesis is often stated in terms of a particular regression coefficient being equal to 0. In terms of the above example this corresponds to the null hypothesis that the presence of Ryanair has "zero" impact on Aer Lingus prices. The alternative hypothesis is that the coefficient measuring the effect of Ryanair's presence on Aer Lingus is positive. A result is said to be statistically significant when it is possible to reject the null hypothesis. This is done by posing the question "If the null hypothesis is really true how likely it is to observe associations as large or larger than those observed in the current analysis simply by chance alone?" This probability is measured by the pvalue of the coefficient which is generated from the regression. The lower the p-value the lower the chance of falsely rejecting the null hypothesis. In terms of our example a very small p-value corresponds to a very low probability of rejecting the null hypothesis when it is true, that is incorrectly concluding that Ryanair's presence is negatively associated with Aer Lingus' prices (this is also known as a type I error).

14. The standard criterion for "statistical significance" is a p-value less than 0.05 on the pertinent statistical test. This means that if the null hypothesis were true we would observe a difference as large as or larger than the estimated coefficient in fewer than 5 out of 100 samples drawn from the same population (less than 5%).

15. It should be stressed that there is no specific percentage threshold above which a result is statistically significant. While the 5% criterion is typical, reporting of more stringent 1% significance tests or less stringent 10% tests can also provide useful information. This is because statistical significance is determined, in part, by the number of observations in the data set. The more observations used to calculate the regression coefficients the smaller the standard error of each coefficient. A smaller standard error reflects less random variability in the estimated coefficient (or estimate) (3). Other things being equal, the statistical significance of a regression coefficient increases as the sample size increases. If the data set is sufficiently large, results that are economically significant are often also statistically significant. However, when the sample size is small it is not uncommon to obtain results that are economically significant but statistically insignificant (4). The Commission follows the convention of reporting coefficients, p-values, standard errors and the size of the sample. In the description of the results the Commission emphasises statistically significant findings to the 5% level (i.e. p-value<0.05). However the Commission does not ignore findings significant at the 10% level, in particular where the coefficient of interest is economically significant and the sample size is small.

16. There is another important reason not to ignore findings when the p-value falls between 0.05 and 0.10. When evaluating the null hypothesis that a variable of interest has no association with a dependent variable against the alternative hypothesis that there is an association, a two-tailed test that allows for the effect to be either positive or negative is usually appropriate. A one-tailed test would usually be applied when it is believed, perhaps on the basis of other direct evidence, that the alternative hypothesis is either positive or negative, but not both. Because one-tailed tests produce p-values that are onehalf the size of the p-value using a two-tailed test, the choice of a one-tailed test makes it easier to reject a null hypothesis. Correspondingly, the choice of a two-tailed test makes null hypothesis rejection less likely. The Commission believes both economic theory and qualitative evidence suggest that in this case the relationship between the presence of one merging party and the prices of the other is negative or zero, but not positive. It would be therefore justifiable to adopt a one-tailed test.

17. Notwithstanding this, and since there is some arbitrariness involved in the choice of an alternative hypothesis, the Commission reports the p-value in all its regressions on the basis of a two-tailed test. As it turns out, using the Commission's preferred (5) regression technique the estimated coefficients of interest have p-values so small that they are statistically significant at the 5% level (6) whether using a one-tailed test (the relationship is different from zero but it is expected to be only negative) or the stricter two-tailed test (the relationship is different from zero but can be positive or negative).

18. The issue of robustness - whether regression results are sensitive to slight modifications in assumptions (e.g., that the data are measured accurately)-is also of serious importance. If the assumptions of the regression model are valid, standard statistical tests can be applied. However, when the assumptions of the model are imprecise, standard tests can overstate or understate the significance of the results. However, the Commission notes that the violation of an assumption does not necessarily invalidate a regression analysis. In some cases in which the assumptions of multiple regression analysis fail, there are more advanced statistical methods that are appropriate.

19. This report is structured as follows: The first section identifies the hypotheses of interest in this case which, a priori, can be tested with the use of regression analysis. Next, we describe the database that the Commission has compiled to conduct its own analysis. Section 4 explains the adequacy, strengths and weaknesses of two regression techniques used in this case: cross-section and fixed-effects. Section 5 and 6 report on the results from cross-section and fixed-effects regressions, respectively, and derive conclusions relevant for the competitive assessment. Section 7 discusses a number of technical issues in more detail in reaction to the comments made by both RBB (for Ryanair) and CRA (for Aer Lingus) in their response to the Statement of Objections. Section 8 concludes.

2. Hypotheses of interest

20. According to its own submission (7), to achieve a load factor target Ryanair constantly varies its price (i.e. the available fare). If a flight appears to be filling up too slowly to achieve the target load factor, Ryanair would react by lowering fares in order to increase the chance of meeting the target. Thus, if Ryanair were competitively constrained by the presence of Aer Lingus, it can be expected that Ryanair would have to offer lower fares on average when Aer Lingus is on the same airport or city pair (since if not, it would fail to meet the target). Conversely, if Ryanair imposes a competitive constraint on Aer Lingus we should expect that Aer Lingus fares are negatively affected by Ryanair's continued presence. This line of argument gives rise to the following testable hypothesis:

Hypothesis I

I-A Ryanair's presence is associated with a statistically (8) and economically significant reduction in Aer Lingus fares in the various short-haul routes where they overlap.

I-B Conversely, Aer Lingus presence is associated with a statistically and economically significant reduction in Ryanair's fares.

Hypothesis II

II-A Ryanair exerts a stronger competitive constraint on Aer Lingus' fares than any other actual or potential competitor does.

II-B Aer Lingus exerts a stronger competitive constraint on Ryanair's fares than any other actual or potential competitor does.

Hypothesis III

The existence of an actual or potential competitor operating from a base at the destination airport on a route originating in Dublin has a limited impact on the merging parties' prices.

Hypothesis IV

IV-A The stronger the presence of Ryanair in the route the more pronounced the effect on Aer Lingus fares.

IV-B The stronger the presence of Aer Lingus in the route the more pronounced the effect on Ryanair's fares

21. In testing these hypotheses, it must be stressed that two types of error are possible. First the hypothesis that Aer Lingus and Ryanair competitively constraint each other might be accepted even though it is false (leading to a "false conviction" - or type 1 error). The converse error is to reject this hypothesis even though it is true (leading to a "false acquittal" - or type 2 error). Both types of error typically result from omitting important variables in the regression specification. For example: ?* Even if Aer Lingus' fares appear to be lower on routes where Ryanair is present this relationship may be spurious. This may be the case if the presence of Ryanair is positively correlated with the presence of other competitors (for example because entry into these routes is easier). To some extent, this may be corrected by controlling in the regression analysis for the presence of other rivals (9) - for example, by introducing a variable indicating the number of rivals in the route or even their identity. ?* In contrast, a regression analysis may fail to detect any impact of Ryanair's presence on Aer Lingus' fares when it exists. In cross-section regressions, in particular, this may occur because the threat of entry by one of the merging parties may be almost as strong as actual presence. Indeed, in such cases a cross-section comparison of Aer Lingus' prices contingent on Ryanair's presence leads to a negative result even though Ryanair constrains Aer Lingus by the mere threat of entry even if it is not actually active in the route.

22. Econometric techniques can be used to mitigate the scope for either type of error, including the use of alternative specifications in order to assess the robustness of the results. However, the risk of making type 1 and type 2 error, no matter how much reduced, is an important reason why the Commission uses regression analysis as complementary to its overall qualitative and descriptive assessment of the facts of the case.

23. A further remark is here pertinent. Ryanair argues that it targets discretionary passengers whose alternative is not so much to fly with another airline but whether to fly the route at all. According to this view, Ryanair's closest constraint is the price sensitivity of these customers - not Aer Lingus. Thus, to the extent that competition between Ryanair and Aer Lingus exists, it is likely to be asymmetric - i.e. Ryanair constrains Aer Lingus to some degree but the reverse does not apply. In turn, this asymmetry of competition would indicate that Ryanair is unlikely to be in a position to sustain higher prices because of the merger.

24. Econometric analysis can provide some indication of the price effects of a merger on both of the merging parties. A possible conclusion is that the competitive constraint that one of the merging parties exerts on the other is more significant than the converse, possibly for any of the reasons advanced above.

25. However, the Commission first notes that for there to be a serious competition concern it is not necessary that both Aer Lingus and Ryanair constrain each other in equal measure. According to economic theory, it is possible that one firm exercises on the other a stronger competition concern than conversely (10). Such asymmetric effects may, inter alia, depend on: (i) the relative costs and degree of differentiation of the services provided by both parties; (ii) the availability and characteristics of other rivals in a given route; (iii) the distribution of customer preferences over the alternative offers; or (iv) exogenous factors affecting demand or supply, such as type of destination or airport charges.

26. In any event, the Commission emphasizes that the ultimate concern is with consumer welfare. Harm to consumers may primarily result from price increases but the Commission must also consider other sources of harm. A number of factors can affect the reciprocal influence of one firm on the other. For example, Ryanair may have a particularly strong effect on Aer Lingus' prices, due to its low-price strategy and its recent and aggressive entry into Aer Lingus routes out of Dublin. Conversely, Aer Lingus may have little impact on Ryanair's prices but a significant effect on its frequencies or load factors. Furthermore, Aer Lingus may also affect Ryanair's decisions regarding the expansion of its network out of Dublin. It may force Ryanair to increase advertising, and reduce prices of its ancillary services, which in certain routes may contribute to more than 70% of the overall profits on that route. Thus, it is possible that the constraints imposed on each other are asymmetric in nature but not in strength. Price regressions provide little insight into such effects. Other types of evidence, qualitative in nature, are therefore required to assess these competitive constraints.

3. Data description

27. The Commission's regression analysis is based on three sets of data submitted by Ryanair (FR), Aer Lingus (EI) and the Dublin Airport Authority (DAA). Early after the opening of the second phase, the Commission held a number of meetings with the merging parties and its economists. The purpose of these meetings was to assess the format and characteristics of the data that the parties could realistically make available to the Commission in a short time frame.

28. The Commission's Formal Request for Information of 23 January 2007 included a common template to facilitate the data collection and data management. Each of the merging parties was asked to complete three excel spreadsheets. In the first, each carrier had to provide data on its own fares and costs for each of the routes it operates out of Dublin (fares data). The second spreadsheet requests information on the competitive framework in the relevant routes (carriers data). Finally, the last spreadsheet refers to route specific characteristics (route data). Further, the Commission requested the DAA to provide information regarding the merging parties' competitors on all relevant routes out of Ireland.

29. At the outset it should be emphasized that the data obtained has been critical not only for the regression analysis described in this annex but also more generally to assess the current and past competitive dynamics in the routes affected by the notified merger. For all routes, the Commission has generated graphs and descriptive statistics - in particular of prices, market shares and number of passengers (normally for all carriers present in the route). This descriptive analysis has proved useful and has improved the Commission's interpretation of other evidentiary documents. It has provided an accurate picture of the competitive strength of the merging parties in all individual routes out of Dublin affected by the merger.

3.1. Dublin Airport Authority data

30. The DAA has provided in its first submission (1 February 2007) three datasets, containing, respectively, information on routes out of Dublin, Shannon and Cork (airports in Ireland where the merging parties are both present).

31. Each dataset contains information for each of these airports covering the period as of January 2000 until December 2006. It includes, for each carrier with scheduled flights and charter airlines (non-scheduled flights): ?* total monthly frequencies;

* total number of passengers, split between connecting and point-to-point passengers;

* total monthly frequencies per day of the week;

* total number of passengers per day of the week.

32. On 12 February, the DAA submitted an enlarged sample covering the period as of January 1996 until December 2000. On 14 February, the DAA further extended the data set to cover additional routes out of Dublin.

3.2. Aer Lingus data

33. On 17 January, CRA International submitted on behalf of Aer Lingus the three complete spreadsheets requested by the Commission.

34. The "Fares dataset" includes Aer Lingus data on short haul routes out of Dublin between 2004 and 2006. The included variables are:

* Number of passengers (business and economy);

* Average total fare, airport taxes, net fare;

* Volume and share of internet sales in terms on passengers;

* Revenues, split in fares revenues and revenues from other charges;

* Share of fare revenue from passengers originating from Dublin

* Expenses: fuel costs, maintenance charges, airport charges, marketing expenses;

* Number of seats, number of flights and load factor;

* Number of O & D passengers and number of connecting passengers.

35. The "Carrier dataset" contains data obtained from the DAA on scheduled departures from Dublin for short haul routes operated by Aer Lingus between January 2002 and November 2006. The spreadsheet reports for each carrier: ?* Carrier IATA name;

* Type of carrier (LCC, mid-frill, flag);

* Presence of a base at destination airport (a carrier is supposed to have a base at an airport if it offers flights to more than 10 different destinations from that airport);

* Average daily frequencies;

* Average weekly frequencies;

* Monthly frequencies (only from August 2003);

36. Finally, the "Route dataset" reports general information on the routes served by Aer Lingus. The information includes: ?* Distance;

* Average population of the destination city for the periods 1994-1998 and 1999-2003;

* Average monthly temperature in the destination city;

* Type of destination, i.e. leisure of business, based on the proportion of business class travel.

37. On 26 January, Aer Lingus submitted additional data including a revised version of the "Fares dataset" and of the "Carrier dataset". The new version of the "Fares dataset" covers a larger period of time, providing information on flights as of January 1996 in terms of:

* Number of passengers;

* Revenues (rev. from fares, rev. from other chargers);

* Number of single segment and multi-segment flights (11);

* Number of seats on single-segment flights.

38. The costs variables, provided only for single-segments flights, is enlarged respect to the previous submission to cover the period as of January 2002. Furthermore, information of "Direct Operating Costs" and "Maintenance Charges" was added.

39. On 20 February, CRA International has provided the EU Commission with additional data obtained by Schedule Reference Service (SRS) on airport use by airline. The information included refers to the number of outbound scheduled flights per week to destinations in the EU-25 (excluding Bulgaria and Romania), provided by airport and airline from January 2002 until November 2006. This submission has enabled the Commission to assess the degree of presence of each of the competitors at the destination airport of each route.

40. On 8 March, CRA informed the EU Commission about a limited number of errors on their "Fare dataset" submitted on January 26th. The errors affect the field "number of multisegment flights" for three routes: Dublin-Cork, Dublin-Kerry and Dublin-Manchester. The next day, CRA provided the EU Commission with a third amended version of the "Fares dataset". The Commission's regression analysis uses only this updated data-set.

3.3. Ryanair data

41. After a number of extensions of the initial deadline agreed with the Commission, Ryanair replied to the Commission's Formal Request for Information of 23 January 2007. Ryanair also submitted three separate spreadsheets in a format somewhat different from the template provided by the Commission. This complicated the necessary data management exercise undertaken by the Commission to combine all submitted data into a single data set.

42. The first file named "Apr Oct 03-06 Comp stats" includes data on weekly frequencies provided by carriers serving all Ryanair routes, not only out of Dublin, between 2003 and 2006. For each carrier, average weekly frequencies are reported with two observations per year, in April and in October.

43. The other two spreadsheets, info "request All 02-06" and "info request DUB 97-06", provide information on Ryanair business variables available as of April 2004 until December 2006 for all Ryanair routes and from April 2004 until the same date for routes out of Dublin, respectively.

44. The variables concerning Ryanair flights reported in both datasets are the following:

* Number of flown passengers;

* Total revenue from fares;

* Average total fare per passenger;

* Load factor.

45. Other variables, included only in the second dataset referring to routes out of Dublin and reported as of April 2004, are:

* Total revenues;

* Fuel price;

* Maintenance costs;

* Marketing and Distribution costs;

* Airport charges;

46. Each of the datasets reports also distance and population at the city of departure and destination.

47. On 19 February, the Commission informed Ryanair that it had detected several inconsistencies and errors in the cost variables reported. In reply to the subsequent request for clarifications, Ryanair submitted on 21 February 2007 an amended version of the dataset. The Commission's regression analysis uses only this updated data-set.

3.4. Data management

48. In order to perform its econometric analysis, the Commission has merged in one dataset the data provided by the DAA, Ryanair and Aer Lingus. This dataset contains information on all short haul routes out of Dublin on a monthly basis, by Ryanair or Aer Lingus in the period between January 1996 and December 2006 (12).

49. In order to have a measure of the parties' frequencies and capacity comparable with all other competitors in the routes, the Commission has relied on the DAA submission for these variables - after checking for the consistency of all three data sets. This procedure has notably enriched the variety of competitive situations envisaged by the data, allowing the analysis to cover periods in which none of the parties were operating some specific routes and, for this reason, no information were submitted by them.

50. The dataset has been constructed in accordance with the conclusions reached by the Commission with respect to the geographic market delineation. All the data available at the airport level has been aggregated for the relevant catchments area at the destination city. Frequencies and capacities provided have been summed up across airports to have a measure of the presence of the competitors in each city-pair route as precise as possible; for fares, revenues and costs the Commission has used an average measure weighted by the number of passengers flying to each airport for each carrier.

51. The final dataset contains information for 81 different airports for a total number of 5427 observations between January 1996 and December 2006. The data, however, is complete for most relevant variables only as of January 2002. Ryanair fare data goes back as far as April 1997. However, its cost data only starts in April 2004. Aer Lingus fare data (13) goes further back to January 1996 and its cost data to January 2002.

52. Variables relating to the competitive environment at the destination airport provided by Aer Lingus from SRS go back to January 2002. These variables are important in testing hypothesis III and they provide controls to test the robustness of the Commission's econometric specification. This information gives a measure of the potential competition exerted by carriers offering a high number of total frequencies (that is, the number of flights within a specific period of interest) at the destination airport. The Commission used this data to construct a dichotomous (or dummy) variable equal to 1 when there is at least one flag carrier with a strong presence at the destination airport and 0 otherwise. Likewise a dichotomous variable was created for non-flag carriers.

53. At this general level of analysis it was not possible to assess for each carrier at each destination airport, for each month in the sample, whether significant presence at that airport led the carrier to enjoy competitive advantages normally associated with setting a base. The Commission's analysis indicates that both Ryanair and Aer Lingus benefit from competitive advantages derived from their strong presence at Dublin where they currently base 23 and 20 planes, respectively. However in other airports, other carriers may enjoy similar advantages with less based aircraft. To allow for this possibility a carrier was assumed to have a strong presence at the destination airport where it operated more than 200 flights a week during the month in question out of that particular airport (or city) (14). This roughly corresponds to operating 10 planes with a turnover slightly below 3 return flights per day for every day of the week.

54. The Commission has chosen January 2002 as the starting point of analysis. This is for three reasons. First, as mentioned above, most relevant variables are available only as of this date. Second Aer Lingus' initiated its restructuring process in 2002. This restructuring has allowed Aer Lingus to reposition itself as a no-frills company competing directly with Ryanair on routes out of Dublin, as explained in the competitive assessment section in the SO. Third, in any event, five years prior to the proposed merger appears a reasonable period to assess the competitive effects of the merger. The further back into the past the analysis goes the more likely that the competitive conditions in the market have been affected by exogenous factors. These factors cannot be adequately controlled for in the regressions due to the lack of data.

55. The Commission's preferred econometric approach in this case (panel fixed-effects regressions) is only informative if there is sufficient variation in the explanatory variables of interest during the time frame considered and within each specific route. In the case of regressions to explain the influence of Aer Lingus on Ryanair fares, the explanatory variable of interest is Aer Lingus' presence in the route or the number flights it offers on a route in a given month (monthly frequencies). Unfortunately, there is very little variation in the explanatory variable of interest in the past five years, that is, since January 2002 (15).

56. Following the methodology pursued by CRA and RBB a similar dataset has been constructed on an airport-pair basis.

57. Where relevant the econometric analysis performed on the market definition dataset has been replicated in the airport-pairs dataset to assess the sensitivity of the results to the market definition adopted by the Commission in section 6 of the decision.

4. Econometric methodology

58. In order to test the four hypotheses set out above, the merging parties and the Commission have all considered a reduced-form specification. The basic idea is to regress some measure of airline fares on a vector of firm and route characteristics. In assessing the competitive effects of a merger, the key variable of interest is normally the effect of one of the merging parties on the competitive behaviour of the other. For example, RBB, CRA and the Commission all include specifications where the dependent variable is the price of one merging party and the explanatory variable of interest is whether the other is present on the route. Other variables are added to "control" for other possible systematic influences on fares, which refer to route characteristics that may affect demand or supply on that route.

59. The Commission's (merged) panel data tracks the prices set by both Ryanair and Aer Lingus in individual routes over time. The data includes observations on the average price charged by each airline by month since January 2002 to December 2006 both on routes to the same airport and on routes to the same city but a different airport. We have considered in our analysis the average monthly fares net of airport charges. Neither party appears to keep information on actual fares charged, so the fare submitted is in both cases the result of the allocation of the total net revenue on a passenger basis.

60. Data in this format makes it possible to pursue two different empirical strategies to assess the extent to which the merging firms exert a competitive constraint on each other (holding constant other factors such as competition from other airlines):

(i) Cross-section regression analysis: examines differences in prices across a number of affected routes at a point in time.

(ii) Fixed-effects regression analysis with panel data, which exploits the variation in market structure at individual routes over time.

61. Cross-section regressions use information on different market structures across routes, controlling for observed route specific factors that affect fares. The primary advantage of this approach arises where market structure varies substantially across routes and where there are a large number of routes in the data. Ryanair's expert economists (RBB) have proposed this approach.

62. The disadvantage of using a cross-section approach is that it may not be possible to control for important but unobserved or unmeasured influences on price that vary from route to route. When important variables affecting price in different routes cannot be observed and are correlated with the explanatory variables included in the regression, the estimated coefficients can be subject to bias. This problem is often referred to as omitted variable bias. For example, prices may be higher in monopoly routes not because there is no competition but because in this particular route, demand is relatively low or costs are relatively high (e.g. when high entry barriers are correlated with high operation costs). The opposite is also true, prices on monopoly routes may be relatively low because demand is weak (or very elastic) in such routes.

63. An alternative to making inferences about price effects from cross-sectional comparisons is to exploit the variation in market structure at individual routes over time. This approach, introducing appropriate assumptions on the distribution of the error term, uses information on changes in the market structure within a route over time. For example, the entry of Ryanair on a route dominated by Aer Lingus may affect the latter's price (after controlling for observable changes in other variables such as entry by other rivals). Effectively the method compares the level of Aer Lingus prices on a route after Ryanair entered, with the level before Ryanair entered. This before-and-after comparison is done systematically for all routes where Aer Lingus operates and thereby generates the average effect of Ryanair's presence on Aer Lingus fares.

64. Note that the fixed-effects procedure compares the incumbent's prices before-and-after entry of a rival within the same route. Such comparison can mitigate the omitted variable bias that affects cross-section regressions because it is more likely that unobservable or non-measurable cost or demand factors affecting fares and varying across routes are not likely to vary over time within a given route (such as the type of destination, the popularity of the route according to purpose of travel, customer awareness, destination airport characteristics, number of alternative airports at destination, safety considerations, total duration of travel, air traffic regulations at country of destination etc). Thus, the primary advantage of fixed-effects regressions comes where most unobservable or nonmeasurable factors affecting price are unlikely to vary much during the sample period.

65. Fixed-effects regressions are suitable if there is sufficient time series variation in the data to permit precise estimates of the relationship between price and presence of a rival. Unfortunately, in most panel data sets there is more variation across than over time. When the time series variation in the data is limited, fixed effect estimates of the price effects of market structure may be very imprecise. Fixed effect models also exacerbate the bias that may result from imprecisely measuring the explanatory variables. This is often referred to as "errors-in-variables" bias, which is the difficulty in detecting the influence of an explanatory variable when that variable is measured with error. Aer Lingus points out in its reply to the Statement of Objections that where errors in measuring the explanatory variable of interest are random this will attenuate the estimated magnitude of the coefficient estimate towards zero (that is any such errors would lead to an underestimate of the coefficient). This is correct. However, it is not often not possible to identify whether measurement error is random or systematic. In any event, the Commission has no reason to suspect that the market structure variables the Commission uses in its regression analysis- supplied by the DAA - are subject to any kind of measurement error.

5. Cross-sectional regression analysis

5.1. RBB's cross-sectional regressions on behalf of Ryanair

66. On 10 December 2006 Ryanair's expert economists (RBB) submitted a cross-sectional Analysis (16) based on a data set of 313 city pairs for the month of October 2006. RBB examines whether the presence of Aer Lingus in a Ryanair route has a statistically significant impact on lowering Ryanair's fares in city pairs. RBB also considers evidence on whether the absence of "base competition" leads to a statistically significant increase in Ryanair fares.

67. The dependent variable in the regression specification is the natural logarithm of Ryanair's fares for October 2006. The explanatory variables and results for each are:

* Logarithm of flown distance, which proxies for fuel costs, an important cost variable. This variable has a positive and significant coefficient.

* Base dummies for each airport in which Ryanair has a base. The sign on the Dublin base dummy ("BD1") is positive and statistically significant. According to RBB this may reflect relatively high underlying demand at Dublin or relatively high airport charges levied by the DAA.

* Base market power dummy for certain airports so as to check whether having a large share of scheduled flights at the base airport leads to higher fares. This hypothesis is rejected by the data in RBB regressions.

* An Aer Lingus dummy which turns out to be positive in most specifications and is never statistically significant. This finding would imply that Aer Lingus is "not an effective competitive constraint on Ryanair on existing overlap routes".

68. To estimate the regression coefficients RBB uses Ordinary Least Squares (OLS) augmented to correct for possible heteroskedasticity in the errors (17). To control for seasonal effects RBB proposes to focus on October 2006 fares as it is a "shoulder" month (18). To test whether the results hold for other months, RBB runs the same regressions for the months of April 2006-August 2006 and reports no significant changes.

69. RBB also controls for the number of competitors in the route. This approach makes sense to the extent that in the short-run the number of competitors is pre-determined. This appears to be a reasonable assumption in this case (19).

5.2. Commission's assessment of RBB cross-section analysis

70. RBB claims that the empirical evidence submitted on Ryanair's monopoly routes "demonstrates" that its business model is different from that of other airlines, and its pricing behaviour does not change with the number of competing airlines - so that the proposed merger should be expected to have no effects (see RBB's paper of 21 November, pages 3 and 6; and section 2 of RBB's paper of 20 February). In particular, RBB argues the results of its cross-section regressions show no evidence to suggest that (i) the presence of Aer Lingus or (ii) the presence of base competition led to a statistically significant reduction in Ryanair fares.

71. As a preliminary check, the Commission first reviewed RBB's analysis and confirmed that there are no technical errors. The Commission, however, identified a number of considerations that were omitted in the written submission yet are critical for understanding and interpreting the results. Most importantly, the data set is not constrained to routes out of Dublin (or even routes out of Ireland). Instead, it actually includes more than 300 of Ryanair routes across Europe. This implies that the reported results, even if valid, do not actually apply directly to the routes affected by the proposed merger. Indeed, RBB's cross-section analysis provide no indication of whether Aer Lingus' presence has a significant effect on Ryanair prices in overlap routes, relative to Ryanair's price in non-overlap routes out of Ireland (the main hypothesis of interest) (20).

72. The Commission first remarks that Ryanair's claim that it does not compete with other airlines is both counterintuitive, and at variance with considerable documentary and qualitative evidence. In any event, the Commission highlights the following difficulties with RBB's cross-section analysis.

(i) Cross-sectional regressions are not meaningful (i.e. they are likely biased) unless it is possible to control for factors that affect Ryanair prices and vary across routes (21). This is particularly critical given that RBB's regression includes a very large number of routes all across Europe likely to exhibit a significant degree of heterogeneity, both in terms of demand and competitive environment. If some influential variable is omitted (e.g. because it is unobserved or cannot be measured accurately) all coefficients, including the one indicating the presence of Aer Lingus will be biased. In any event, RBB has not provided any complementary analysis supporting the assumption that unobserved differences across routes are likely not to bias the results.

(ii) A further concern is that the variable "number of competitors" is probably a weak control for market structure and cannot fully capture the competitive constraints on Ryanair imposed by rivals other than Aer Lingus. The Commission believes that it is necessary at least to separate between Aer Lingus, other no-frills carriers and network carriers.

(iii) A related issue is that "presence" alone is only a rough measure of the competitive constraints a firm might exert on the other. One possible correction to this problem is to identify some variable, which measures the strength of such presence. The Commission's preferred variable in this respect is the number of monthly flights on the route (i.e frequencies) of a rival carrier. This may provide a more refined measure of the competitive constrain exerted on Ryanair by competitors, Aer Lingus in particular. However, in a cross-section regression the introduction of frequencies instead of the presence dummies would require an even more accurate set of controls to avoid the omitted variable bias.

(iv) With respect to the market power dummy, RBB claims that market power is adequately proxied by Ryanair's market share (above 80%) on an airport-pairs route (22). However, the selected airports are secondary airports. In at least four out of these five cases the geographic market also includes substitutable airports with significant operators with a base (Frankfurt - Lufthansa, Barcelona - Iberia, Vueling, Air Europa, Brussels - Virgin/SN, Glasgow - EasyJet, BA Connect, Loganair, Globespan). The market share of Ryanair in these city-pair routes is lower. This significantly reduces the information value of the market power dummy.

73. For all these reasons, in particular the concern with omitted variable bias, the Commission cannot attach any weight to RBB's cross-section regressions. A priori, the most reasonable solution to mitigate the risk of omitted variable bias is to use a fixed-effects procedure with panel data that exploits the variation in market structure at individual routes over time. Before turning to such analysis we report on the results from an alternative strategy which is to run a cross-section on routes out of Dublin only. This can be expected to partially reduce the risk of bias due to unobserved differences across routes by controlling for demand and supply characteristics at least at one end of the route. However this is at the expense of increased variability in the estimated coefficients due to the small sample size. This may lead to statistically insignificant coefficient estimates (see also paragraph 101 below).

5.3. The Commission's cross-section analysis

74. As explained in the previous section the Commission's view is that cross-sectional analysis is unlikely to provide reliable and convincing evidence of the possible competitive constraints that Ryanair and Aer Lingus mutually exert on each other. However, the Commission also recalls that it had to a more comprehensive data set than Ryanair or Aer Lingus' expert economists. Despite its severe limitations in this case, the Commission has decided to undertake also a cross-section analysis to satisfy the interest of the parties who have no access to the Commission's combined data set.

75. An additional motivation to conduct a cross-section analysis is to correct some of the above mentioned difficulties with RBB's own cross-section regression. In particular we focus on routes out of Dublin only and we introduce dummy variables for the presence of other competitors, besides Aer Lingus. We also test a specification that measures the strength of presence of each carrier on the route. Strength of presence is proxied by the number of frequencies. This, however, raises technical difficulties which we acknowledge cannot be fully resolved.

Baseline regressions

76. The Commission's cross-section regression analysis tests two different baseline specifications. The first specification (reported below for Ryanair) tests the effect of the presence of different competitors on the fares charged by each of the parties in its routes (this specification is referred to as presence specification throughout this report).

ln_pi = ß0 + ß1 ln_Ryan_freqi + ß2 Aer_Lingus_presencei + ß3 flag_presencei +

ß4non_flag_presencei+ ß4 ln_route_tot_paxi + ß5 ln_distancei + ei

77. For each route i operated by Ryanair (alternatively Aer Lingus), the (logarithm of) average fare has been regressed on measures of competition and a control for demand and costs. The competition variables included in this first specification are three dummy variables indicating (i) the presence of Aer Lingus, (ii) at least one flag (23) carrier and (iii) at least one non-flag carrier (24). It is common in applied work for the dependent variable to appear in logarithmic form, with one or more dummy variables appearing as independent variables. The coefficient on a dummy variable can then be interpreted as the percentage difference in the dependent variable. For example, in the above regression equation, if the coefficient on Aer Lingus presence were to be -0.05, this would imply that the presence of Aer Lingus is associated with Ryanair charging on average 5% less than in routes where Aer Lingus is not operating (25).

78. The log of total passengers in the route is introduced as a control for demand variation across routes and the log of distance as a control for cost variation across routes. The log of own frequencies in the route is added as a scaling factor. Furthermore, to test the effect of the presence of a rival airline with a base at the destination airport we run further regressions including two dummy variables indicating whether at least one flag carrier and at least one non-flag carriers with a strong presence at the destination airport affects the prices of the merging parties (see also paragraph 53).

79. A second, more refined specification has also been tested and is reported below for Ryanair (this specification is referred to as the frequencies specification throughout this report).

ln_pi = ß0+ ß1 ln_Ryan_freqi + ß2 ln_Aer_Lingus_freqi + ß3 ln_flag_freqi +

ß4 ln_non_flag_freqi+ ß5 absence_Aer_Lingusi+ ß6 absence_flagi+

ß7 absence_non_flagi + ß8 ln_route_tot_paxi + ei

80. The frequencies specification tests the strength or degree of presence of rivals in the route. In this specification the "presence" dummies have been replaced by the total frequencies offered that month in the given route by Aer Lingus, the total number of frequencies of all others competitors, aggregated into flag and non flag carriers, as well as Ryanair's own frequencies, all inserted in logarithmic form.

81. For each of the competitors a dummy of absence has been inserted in the specification. These dummies, takes the value of one if respectively (i) Aer Lingus, (ii) other flag26 or (iii) non-flag carriers are absent in the route. These dummies allow for a correct introduction of the logarithm in the frequencies. When the number of frequencies is equal to zero, because the carrier is not present in the route the observation would be dropped given that the log of zero does not exist. By adding the dummy indicating absence, we ensure that the observation is properly included in the regression as one where the carrier was not present.

82. The absence dummy should not be interpreted as the inverse equivalent to the "presence" dummies in the presence specification. This is because the "presence" dummies allow for true comparisons of fares when the rival is present and when it is absent. In contrast, the "absence" dummies as introduced in our second specification compare "absence" with the different degrees of presence of the carrier (measured in the number of monthly frequencies it offers). The coefficient of the absence dummy does not measure the price difference of one firm if the other is absent from the route (27)

83. In the "frequency" specification, the coefficient of interest is that of the (log) of frequencies of the other merging party. This coefficient can be interpreted as the elasticity of fares with respect to the number of monthly frequencies that a rival offers in the route. For example, if the coefficient is 0.2, this means that a 1% increase in a rival's monthly frequencies leads to a 0.2% increase in fares. It should be noted, however, this can not be compared to the estimated price effect given by the coefficient in the "presence" specifications. To estimate a comparable price effect from the "frequency" specification it is necessary to make assumptions regarding the competitive situation after the merger and make additional calculations. This is explained in detail in section 7.3.1.2 below. This is exactly the same approach that was also followed by the US Federal Trade Commission and the parties' economic expert in the Staples/Office Depot case where both the crosssection and fixed-effects approach were also used. In this case the regression specification was very similar to the Commission's proposed frequency specification (28).

84. This specification tests the strength or degree of presence of rivals in the route. In a routeto- route comparison such as the one proposed, costs are expected to be negatively correlated with the frequencies of each competitor. We face a difficult trade-off here. If one includes a control for costs, because of the high correlation with the explanatory variable of interest this would lead to high variance of the coefficient estimates, this problem is augmented by the small sample size in the cross-section regressions. However omitting a control for costs would likely lead to biased coefficients. We have tested both alternatives and conclude results are indeed sensitive to both problems. This further reinforces the view that cross-section regressions are unsuitable to test the hypothesis of interest in this case.

85. Both regressions have been performed on Ryanair and Aer Lingus average fares charged in the months of May and October 2006. These months have been chosen for two reasons. First, to allow for direct comparisons with RBB results, which also restrict the analysis to October 2006. Second, because as argued by RBB, the month of October and May are less sensitive to seasonal variations that, with so few observations, could obscure the underlying effects of interest as they would likely affect the routes to different degrees.

Extensions

86. The regression has been performed first in the data set constructed in accordance with the market definition, considering city pairs. The analysis has been replicated on an airport pair basis. Moreover, a variable indicating the presence of flag carriers and non-flag carriers with a base at the destination airport has also been included in a second set of regressions (or base at destination dummies).

Results from the presence specification

87. Table 1 reports the results of the presence specification applied to Aer Lingus fares based on the city-pairs database. Table 2 refers to the same regressions based on the airportpairs database.

Table 1

Presence specification/city pairs database (DB) / Dependent variable (DepVar): (ln) Aer Lingus Fares

<emplacement tableau>

88. Allowing for airport substitution in the city of destination, the city pair regression reports a significant negative effect of the presence of Ryanair on Aer Lingus fares in May and October (-14% and -27% respectively). All values are statistically significant at the 5% level. Non-flag presence also has a negative effect but is not statistically significant. Flag carriers also exert a competitive constraint on Aer Lingus but this is weaker than that of Ryanair.

89. Ryanair presence coefficients are robust to the inclusion of indicator variables for flag carrier and non-flag carrier with a base at the destination airport (flag_based and non_flag_based dummies, respectively). It appears that a flag carrier with a base at the destination city imposes a weakly statistically significant constraint on Aer Lingus, however this is the case only in May.

90. The effect of Ryanair's presence on Aer Lingus is less stark when the data set is restricted to airport pairs. A possible explanation for such a result could be the fact that on an airport basis the number of overlapping routes between Ryanair and Aer Lingus is reduced (29). In small samples, this effect could outweigh the one derived from the higher degree of demand-side substitutability between the two carriers if they fly to the same airport. However even in this case, Ryanair is the only competitor whose presence in the route is statistically significant, both in the extended specification (both months) and in the baseline specification (in October). The coefficients are also economically significant.

Table 2

Presence specification /aiport-pairs database (DB) / Dependent variable (DepVar): (ln) Aer Lingus Fares

<emplacement tableau>

91. Tables 3 and 4 report the equivalent results for Ryanair. The presence of flag carriers reports a negative and significant coefficient at the city level, which disappears at airport level. The Commission notes that in the routes where Ryanair faces competition from flag carriers Aer Lingus is always present. The presence of flag carriers and Aer Lingus is highly correlated across city-pairs. It is likely that the regression arbitrarily distributes the combined effect on Ryanair fares of their presence. It is therefore not surprising that the dummy for Aer Lingus presence is not statistically significant in the city-pairs regressions even if Aer Lingus may be, in fact, exerting a significant competitive constraint on Ryanair.

Table 3

Presence specification/city pairs database (DB)/ Dependent variable (DepVar): (ln) Ryanair Fares

<emplacement tableau>

92. This interpretation is confirmed by observing the results from the airport-pairs regression. Here the correlation between Aer Lingus and flag carrier presence is much more attenuated. Aer Lingus appears to have an independent effect on Ryanair. Indeed the coefficients became economically important and statistically significant in the baseline regression (23.9% and 18.9% for May and October respectively) (30).

Table 4. Presence specification - airport pairs DB- DepVar: Ryanair fares (ln)

<emplacement tableau>

Results from the frequencies specification

93. The first two columns of tables 5 and 6 report the results of the frequencies specification for the months of May 2006 and October 2006 applied on the city pair data set, respectively for Aer Lingus and Ryanair (31). The results for Aer Lingus indicate a high and significant effect of the number of frequencies provided by Ryanair on the fare charged by Aer Lingus on routes where it faces Ryanair competition. Other competitors, flag or nonflag, do not have a similar effect. For example in May and October, a 1% increase in Ryanair's frequencies would lead to a 0.17% and a 0.26% decrease in the fares of Aer Lingus (conditional on Ryanair being present in the route) (32). These coefficients are statistically significant.

94. Similar results, though slightly less statistically significant, are reported for the effect of Aer Lingus frequencies on Ryanair fares. Here the result is similar for the two months considered and again none of the other competitors has an effect.

95. The third and fourth column of each table reports the outcome of the introduction of the indicator variable for a flag or non-flag carrier with a strong presence at the destination airport (base at destination dummies). Interestingly, in the October regressions, Aer Lingus is constrained by the presence of a flag carrier (both on the route and at the destination airport) when the base at destination dummies are added but not otherwise. The parties' dummy variables in contrast appear robust to the inclusion of the base at destination dummies.

96. However, the Commission draws attention to the significant variation in the value of the coefficients attached to the explanatory variables of interest between the two months. Some very high coefficients attached to Ryanair or Aer Lingus' frequencies may depend upon non-observed route specific characteristics.

97. More generally, this lack of stability in the value of the regression coefficients as well as their statistical significance casts doubt on the reliability of cross-section regression to capture the competitive constraint that Ryanair or Aer Lingus exert on each other.

Table 5. Frequencies specification - city pairs - DepVar: (ln) Aer Lingus fares

<emplacement tableau>

Table 6. Frequencies specification - city pairs - DepVar: (ln) Ryanair fares <emplacement tableau>

98. Finally, the airport pair specification confirms the above results. Despite a lower number of overlap routes, the frequencies of each of the parties provide a significant competitive constraint on each other.

Table 7. Frequencies specification - airport pairs - DepVar: (ln) Aer Lingus fares

<emplacement tableau>

Table 8. Frequencies specification - airport pairs - DepVar: (ln) Ryanair fares

<emplacement tableau>

99. In spite of indications from the above cross-section regressions, that Ryanair and Aer Lingus do in fact constrain each other, the Commission does not regard these results as sufficiently reliable to reach any definite conclusion in this respect. The results from the frequency specification are particularly problematic because omitting a control for costs, such as route distance, likely biases the results as pointed out by RBB (33). Unfortunately when the frequency specification includes a control for costs (e.g. route distance) this would eliminate this source of bias it also results on much higher variance for the coefficients of interest. This is not surprising given the small number of available observations and the high correlation between frequencies provided by competitors and distance across routes (34).

5.4. RBB & CRA comments on the cross-sections

100. RBB raises two concerns regarding the Commission's cross-section analysis. First they point out that the natural logarithm of total passengers "is clearly endogenous by the Commission's own description of the yield management processes of Aer Lingus and Ryanair (see paragraph 380 of the SO), which explains how fares adjust rapidly depending on how quickly passengers fill up available seats on any given flight". The Commission notes that this variable was included to control for the thickness of the route, which potentially will have a high influence on prices. However, the Commission agrees with RBB that introducing this variable in the cross-section specification might lead to endogeneity bias. If that were the case RBB correctly remarks that "the reported results are therefore likely to be biased and inconsistent with invalid standard errors".

101. To partially correct for the risk of endogeneity we rerun all the regressions reported above replacing the natural logarithm of total passengers for the month in question with the same variable lagged six months. It can be argued that this approach removes the endogeneity concerns given that the fare in a given month is not simultaneously determined with the number of total passengers travelling in that route six-months before. This modification leads to results in line with those presented in table 5. While the value of Aer Lingus frequencies, lagged six months, is significant only for the October regressions, the log of Ryanair frequencies remains significant across all the specifications. This is an indication that the alleged endogeneity of frequencies is not a serious concern. In any case this result does not alter the Commission's view that these cross-sections are not sufficiently robust.

Table 5.b. Frequencies specification - airport pairs - DepVar: (ln) Aer Lingus fares

<emplacement tableau>

102. RBB also criticizes the fact that in the Statement of Objections the Commission did not include a table reporting the results from the frequency specification including "route distance" as an explanatory variable. RBB writes that "distance is not reported in the frequency specification. However, we learn that had the Commission included distance in the frequency regression [...]* In other words the Commission tables deliberately reports a functional form that was clearly incorrect". RBB complains of Commission bias in reporting results. The Commission cannot accept this accusation. First the Statement of Objections explicitly indicated the effects of including the "distance" variable in the frequency specification (Annex 4 of the SO, paragraph 81)" (as RBB itself points out). The effect of including distance in the regression is therefore not something RBB "discovered" in the data room. The Commission has not hidden any meaningful results. In fact, the Commission has reported the results from it cross-section analysis purely in the interest of transparency and despite the fact that it does not attach any weight in its competitive assessment to the cross-section regressions.

103. In any event, the Commission agrees with RBB that distance is a good proxy for the costs of fuel and omitting this variable can lead to bias. However as is explained in basic econometric textbooks (see for example Wooldridge 2003, page 100) (35) " the choice of whether or not to include a particular variable in a regression model can be made by analyzing the trade-off between bias and variance". As explained above (and in the Statement of Objections) the variables of interest, that is, the log of frequencies of competitors, is strongly and negatively correlated with "route distance". Likely, this is because in point-to-point routes the shorter the route distance the lower the flight time and the larger the number of rotations that can be done. High (but not perfect) correlation between two or more explanatory variables is called multicolliniarity (36). Multicolliniarity does not lead to bias but it often results in higher variance for the coefficient of interest. This problem is compounded if samples are small. In this case we have a very small sample (less than 50 observations in most regressions). As a result the high correlation between "distance" and "number of frequencies" would lead to very high variance. This, in turn, results in statistically insignificant coefficients, even though there might in fact be a relationship between number of frequencies and the fares.

104. In summary, the Commission notes that there is a trade-off between bias and variance in this case. Such trade-off would be irrelevant if despite omitting "distance" from the frequency specification the coefficients of interest were neither economically nor statistically significant. Distance is positively related with fares yet negatively related with number of frequencies; this likely leads to a negative bias in the frequencies coefficient. This implies that the absolute values of the frequency coefficient are higher than they should be. Unfortunately it is not possible to assess whether the bias is small or large. This is one reason why the Commission rejected the results from the frequency specification in the Statement of Objections (see paragraph 81 in the SO).

105. CRA points out37 that for Aer Lingus, all Commission cross-section regressions indicated that Ryanair constrained Aer Lingus' fares. For Ryanair fares, the simple presence specification for city-pair data indicates that Aer Lingus presence does constrain Ryanair. The fit is less good, and again there is considerable variation in coefficients across the various specifications. As a result, CRA argues that the Commission's crosssection results are consistent with both Ryanair being a constraint on Aer Lingus and Aer Lingus being a constraint on Ryanair. However CRA also remarks that "the small number of observations and the problems with omitted characteristics imply that these results are not conclusive"

106. In conclusion, the Commission considers that neither RBB's nor its own cross-section analysis is robust to the necessary standard. No definite conclusions can be derived from these results given the impossibility to control for unobserved factors that affect prices and differ across routes, the small number of observations, the sensitivity of the results to the month considered or the fact that the inclusion of statistically insignificant explanatory variables sometimes affects the coefficients of other variables.

6. RBB's two-step (fixed-effects) regressions on behalf of Ryanair

107. Ryanair expert economists (RBB) submitted two further reports that include econometric analysis based on a fixed-effects procedure.

6.1. RBB Report "Is Aer Lingus a competitive constraint on Ryanair? (8 Feb)

108. In the first report, RBB examines whether the presence of Aer Lingus has an effect on Ryanair prices on routes out of Dublin.

109. RBB's dataset consists on a panel of Ryanair one-way fares from Dublin (i.e. single fares from Dublin to other European destinations) over the period April 2002-December 2006, as well as other potential explanatory variables. RBB uses a panel regression of fares on fixed-effects (dummy variables to control for time, season and route effects) as well as fuel costs and a proxy for demand at the destination airport.

110. However, RBB does not use the standard fixed-effects procedure that can be found in econometric textbooks. Instead RBB introduces an ad-hoc two-step method which they claim avoids the problems encountered in the cross-section regressions yet exploits at the same time the panel data structure. The first step comprises a fixed-effects model whereby Ryanair's price is regressed on time and route fixed-effects as well as on other explanatory factors. The model takes the form (38):

(formule)

111. Where i indexes each route and t indexes the month and year of the observation, yt is a time and 8i a route fixed effect; pit measures Ryanair's average yield for a given route i in a given month t; xit represents a demand proxy and fuel and oil costs. In order to control for the shifts in demand over time RBB constructs the following measure: For all destinations X out of Dublin, they compute an average yield from another destination to X to control for the demand for the flight Dublin to X. For instance, the route London to Venice (Treviso) is used as a control for the route Dublin to Venice (Treviso). As control for costs, they use Ryanair's fuel costs per flight that Ryanair records on a monthly basis for each route. Finally, Eit is the error term that (presumably) satisfies the standard assumptions (see Annex IV.1 for details).

112. For the second step of the methodology RBB uses the estimated route fixed-effects, 8i ? and regress it on a measure of Aer Lingus' presence and the distance between departure and destination:

(formule)

113. There are two different measures RBB uses to capture the presence of Aer Lingus. The first is an indicator variable equal to 1 if Aer Lingus (EI) is active on a city pair which is also served by Ryanair (FR). The second is a measure of whether EI operates on the same airport pair as FR. The coefficient of interest is B2 which indicates whether the presence of Aer Lingus on a route explains any pricing differences across routes that persist over time. According to RBB if it were the case that Ryanair systematically charges lower prices on routes where Aer Lingus is present then the coefficient for EIi should be negative and statistically significant (39).

114. RBB runs several first stage regressions: (a) a time (month and year) and route fixedeffects model (excluding demand and cost variables); (b) a "full" model (including time and route effects as well as demand and cost variables); and (c) variants of the full model dropping either cost or demand variables. The first stage allows to generate a predicted "route fixed effect", i.e. the influence of route specific factors on fares having controlled for other features that impact persistently on fares. According to RBB, on overlap routes the route fixed effect includes any influence of Aer Lingus because Aer Lingus is almost always present on overlap routes throughout the whole of the time period.

115. In the second stage RBB uses a cross-sectional regression of the route fixed effect on the presence of Aer Lingus (whether on a city pair or an airport pair) and distance. This allows determining whether the presence of Aer Lingus adds any further explanatory power - in particular, whether Aer Lingus' presence leads to lower Ryanair fares.

116. RBB reports that for all the considered specifications there is no statistically significant impact of the presence of Aer Lingus either on a city pair or on an airport pair basis. According to RBB, this finding also supports Ryanair's claim that its business model is to target a passenger base that other (higher cost) airlines cannot profitably target. It follows that the strongest competitive constraint on Ryanair is not Aer Lingus but the price sensitivity of its customer base. Attempts by Ryanair to sustain higher fares would not be profitable - not because of switching to Aer Lingus but due to passengers switching to alternative activities.

117. RBB justifies this unconventional two-step methodology as follows: "Initially, we performed a fixed-effects panel regression of Ryanair's fare (the monthly outbound average yield from Dublin) on the presence of Aer Lingus and other controls such as costs, demand and time and route fixed-effects. We found that on routes where Aer Lingus is present prices are not systematically different from other routes. However, we realised that such a methodology may not be appropriate to address the question of interest as Aer Lingus' presence virtually does not vary over time. Only in three instances, there is a change in the Aer Lingus presence variable (two of which occurring on the same route). In other words, even though the standard fixed-effects approach would imply that Aer Lingus does not constrain Ryanair, we did not consider this evidence to be sufficiently robust due to the lack of variation in entry by Aer Lingus" (40).

118. The Commission agrees that a standard fixed-effects procedure is not informative in this situation. The competition effect is identified only from the observations where entry/exit occurred. Intuitively a "before-and-after" comparison is not possible if there are insufficient episodes of entry and exit happening. However the two-step method proposed by RBB as a solution is not adequate either, as explained below. It essentially corresponds to a cross-section analysis as explained in section 0 below. Hence, all the reservations and concerns expressed above with respect to the Commission's cross-section regressions apply to RBB's two-step regressions.

6.2. RBB's report "Comparing prices on "monopoly" and non-monopoly routes out of Frankfurt Hahn" (41) (20 Feb)

119. RBB submitted a complementary report to show that Ryanair does not set higher fares on "monopoly" routes (on either a city or airport pair basis). They argue this hypothesis is consistent with Ryanair's business model of targeting a price sensitive passenger base that other (higher cost) airlines could not profitably serve.

120. Essentially RBB follows the same two-stage fixed-effects approach described in their earlier submission. There are two important differences:

121. First they consider average monthly yield on all origin-destination pairs from Frankfurt Hahn. They identify the presence of competitors on city-pairs served by Ryanair from OAG data. Information regarding whether a competitor has a base at the other end of the route was collected by York Aviation for the end of 2006 (42).

122. Second, since Aer Lingus is not present on routes out of Frankfurt (except to Dublin) the variable of interest separates routes in which Ryanair is a monopolist and others where it faces actual or potential competition. RBB identifies a monopoly route where capacity constraints rule out potential competition at one end of the route and where there are no based carriers at the other end of the route. The tested hypothesis is whether Ryanair fares are higher in monopoly routes.

123. The results indicate that the presence of competing carriers is never statistically significant. Nonetheless, the impact is negative in all four reported specification.

6.3 The Commission's assessment of RBB's two-stage regressions taking CRA's "data-room submission" into account

124. The Commission acknowledges that in light of the reduced number of instances of Aer Lingus entry into Ryanair routes a standard fixed-effects regression may be uninformative as to the effect that Aer Lingus entry or exit might have on Ryanair's prices.

125. However, the Commission considers that the two-step methodology proposed by RBB cannot resolve this difficulty and is thus inadequate to assess the effect of Aer Lingus on Ryanair over time, for a number of reasons:

126. First and foremost, RBB's two-step methodology is effectively a cross-section analysis where each observation for the dependent variable 8i (the estimated route fixed effect) is in a sense a price average over time. Note that the estimated fixed-effects (8i) capture all time-invariant and unobservable influences on Ryanair's prices within a given route. If one or more of these factors are correlated with the presence of Aer Lingus, the coefficient of interest in the second stage regression will be biased (43). For example, if, as can be expected, Aer Lingus is more often present on routes where demand is less elastic than the average on Ryanair routes this would lead to a positive bias in the "EI presence" coefficient in the second stage regression. In other words, the advantages of the fixedeffects procedure, namely to control for all these unobservable (or non-measurable) route specific characteristics, disappears in RBB's two step approach.

127. Note also that RBB includes the variable "distance" in the second stage regressions. Nevertheless, this is a time-invariant variable and it would be unnecessary to control for distance when using the standard fixed-effects technique as the effects distance on prices would be controlled for by the fixed-effects, allowing the other explanatory variables to be accurately estimated. However, RBB includes "distance" in the second stage, which is effectively a cross-section regression. Unfortunately, as in the case of RBB's standard cross-section regressions (see section 5.1) RBB includes no other controls for time invariant factors affecting prices such as such as the type of destination, the popularity of the route according to purpose of travel, customer awareness, destination airport characteristics, number of alternative airports at destination, safety considerations, duration of travel, population at destination, etc. In fact, a second problem is that neither the first nor the second stage regressions actually control for the presence of rivals other than Aer Lingus. This can also lead to bias.

128. In summary, the Commission's concerns with respect to the value of cross-sectional analysis in this case also apply here. In particular, the small number of observations, which leads to high standard errors and the difficulty to control for unobserved factors that affect prices and differ across routes which results in omitted variable bias. It follows the results based on RBB's two step methodology are likely biased and inaccurately measured.

129. RBB does not agree with this argument. In their Data room report, RBB claims that omitted variable bias is more likely in the Commission's standard fixed-effects regressions than in their own two-step regression and also that the second stage regression is correctly specified. RBB argues that "the Commission implies that our two step approach is ultimately a cross-section approach. However, our methodology is clearly based on panel data regression techniques. Our model therefore satisfies all properties of a fixed-effects approach, in particular the ability to deal with unobserved variables that vary over time or regions" (44).

130. In contrast, CRA agrees with the Commission's analysis and conclusions. Aer Lingus in its response to the Statement of Objections (page 29) argues that RBB's two-step procedure "is simply the same cross-section approach applied to a panel of observations - since it does not exploit the variation in market structure within given routes over time. A paper cited by RBB indeed describes this fixed-effect approach as in fact similar to the cross-sectional model (45). To make this point clear CRA indicates that the two-step estimation procedure chosen by RBB is unnecessarily complex and in fact it is possible to estimate equations (1) and (2) above in a single regression by substituting (2) into (1). This would lead to:

(formule)

131. As CRA correctly states equation (3) indicates that RBB's two-step procedure is in fact not a route fixed-effects model at all but a cross-section. As a consequence it would suffer from all the shortcomings that affect cross-section regressions in this case.

132. The Commission also points out that the use of cross-sectional fixed-effects allows each explanatory variable to have a different mean in each cross-section. In other words, the average effect of each x is included in the fixed effect. This is why the consequences of policy changes can be identified only through time-series variation the data. It also means that RBB's Statement that "using a fixed-effects model stripping out demand and cost effects of the price as well as seasonality patterns" (46) does not make sense. Had their analysis been performed on residuals, demand and cost factors would have been removed. However, the estimated fixed-effects include the mean cost and demand for each crosssection. To see this consider the equation:

(formule)

where D is a demand and C is a cost variable. This equation is equivalent to

(formule)

where bars denote means

133. The estimated (sigle) is then (formule). This indicated that the average demand and cost effects have not been removed. Now suppose that entry occurs only in markets with low costs and high demand. When (formule) is regressed on EIi, the coefficient of EI that is estimated measures the effect not only of El but also of D and C (since those variables are correlated with EI). This means that, even if the true effect of EI by itself is negative, the estimated effect can have any sign and any significance, depending on whether D, C, or EI dominates.

134. Furthermore, even if the RBB analysis were correct, the demand that is relevant for entry is a route, not a city effect. To understand this distinction, suppose that Di measures the number of passengers that would like to fly from Dublin to city i at an average price p. There is no reason to believe that Di is closely approximated by the average number of passengers that would like to fly from Madrid to city i at (a distance corrected) average price. However, the price of a ticket from Madrid to i reflects the latter and not the former.

135. The RBB estimated fixed-effects (8i) have been purged of seasonal patterns of demand and cost (which are not time-invariant by definition). However, entry is a longterm decision (otherwise there would be no problem with a standard fixed-effects regression - we would observe lots of time-series variation in entry and exit). With longterm decisions, seasonal effects are not important. The bottom line is that the things that are correlated with entry/exit decisions, average levels of demand and cost, have not been removed at the second stage of the two-stage procedure, whereas the things that are irrelevant, time-series variation in those variables, have been purged. In conclusion the Commission believes that RBB's two stage analysis suffers from the same shortcomings as cross-section regressions.

136. As regards the "matched route" demand proxy the Commission further notes that even though it may work in some cases there are some reasons to expect that it may fail in others. Prices may be strongly influenced by demand factors that are not correlated between the route of interest and the proxy route. The selection of the proxy route appears critical in this respect (47).

137. With respect to the use of the "matched route" to control for unobserved route characteristics CRA also points out that: "In specifications without the matched route fare variable, the fixed effect measures the route effect for the route out of Dublin. By contrast, in specifications with the matched route fare variable, the estimated fixed effect is the difference between the route effects for the route out of Dublin and the matched route. This implies that fixed effect estimates from specifications that include the matched route fare will generally be much smaller than estimates from specifications without". This means that it is unlikely that any sort of relationship can be found, except by chance, between Ryanair's fares and Aer Lingus presence. This is because the fixed-effects estimated in the first stage do not in reality correspond to the average fares. Indeed the fixed-effects from RBB's Model 1 (the model, in which both the matched route fare and cost variables are included) are uniformly below the estimates from Model 2 from which the control variables are excluded. Estimation of an incorrect specification is unlikely to generate reliable estimates. According to CRA specifications that include the matched route should be ignored. The Commission agrees with CRA that this is strong evidence that RBB's second-stage regression is severely misspecified.

138. Note that as pointed out by RBB the year fixed-effects are negative and statistically significant. This is consistent with lower Ryanair prices over time but tells us nothing of the reason for such price decreases. Also the estimated coefficients on Aer Lingus' presence vary widely in the four models, changing both in magnitude and, most critically, in sign. Moreover, it could be expected that the average effect of distance on the fares to be similar across the four models; instead, the effect of distance on fares is almost three times as large in Model 4 than in Model 1. This is an indication of the lack of robustness of the estimation procedure.

139. CRA also identifies several other difficulties:

* CRA also identifies that there are 12 routes in RBB's dataset, for which there exist only one monthly observation (in 8 of them this is because Ryanair entered in December 2006, the rest cannot be explained). The estimates of the route fixed-effects in these cases are thus based on a single observation (that is 15% of the sample). Moreover, in several of these routes the presence of Aer Lingus is again not measured properly.

* An additional technical problem is that RBB estimates the first-stage regressions in levels and not in logs. However, the large variability in the dependent variable likely leads to all of the first-stage regressions to fail the Ramsey test (48), which indicates that the regressions are likely subject to omitted variable bias. The large range of values of the dependent variable reduces the precision of estimation of the route fixed-effects this renders the second stage also imprecise.

* Finally CRA correctly points out that RBB's preferred specification (model 1) fits very poorly and explains only 18 percent of the variance in average fares. This means that RBB derives its key result based on a model, in which more than 80 percent of the variation in Ryanair's fares is not explained. Such a model is likely to have omitted variable problems. Also the first stage regressions suffer from outliers. The outliers appear to have a significant impact on the estimated route fixed-effects.

140. In light of these arguments Aer Lingus concludes in its response to the Statement of Objections that "these serious problems with RBB's econometric analysis compound the methodological shortcomings of RBB's chosen cross-sectional technique (in particular the omitted variables problem) that we discuss above. We therefore fully concur with the Commission's assessment" (49).

141. CRA also identifies a number of problems with the quality of the data used by RBB in the estimation of their results. First Aer Lingus presence, the key explanatory variable, was measured with error (50). In a number of Aer Lingus seasonal routes, RBB's dataset indicates that Aer Lingus is present every month of the year. Also there are at least 12 airport routes on which information on Aer Lingus' airport presence is missing. There are also significant problems with the variable "fuel costs per flight" since it varies very widely within a route. In one example, fuel cost per seat varies by a factor of almost 200. Some of the largest variations occur within three months. The problem exists for a considerable number of routes. CRA concludes that this probably explains why this variable is not significant in RBB's regressions. Finally, the information on route distance is missing for a small number of routes, which results in the affected routes being dropped from the second-stage regression.

142. Following CRA's submission the Commission conducted its own analysis and confirmed the claims made by CRA regarding the quality of the data set used by RBB to generate its results. It should be mentioned that the Commission had already identified deficiencies in the data set submitted by Ryanair in response to the Commission's Formal Request of 23rd of January (see paragraph 47). This led to further delays.

143. Failure to provide correct data is a serious issue. At best it interferes with the ability of the Commission to use empirical analysis, at worst it may lead the Commission to accept misleading or erroneous conclusions derived from such analysis, whether conducted by the parties or by the Commission itself. It is apparent that the quality of the data set used by RBB in its regression analysis is seriously compromised. This in itself, methodological problems aside, implies that RBB's results are at best invalid if not misleading.

144. With respect to the second submission comparing prices on monopoly routes and nonmonopoly routes out of Frankfurt Hahn all the arguments mentioned above apply. In particular the use of the same two-step methodology implies that it still effectively follows a cross-sectional approach which cannot control for all the determinants of fares on routes. Furthermore the focus on Frankfurt-Hahn implies that this analysis does not, in itself, provide evidence on the competitive constraints that Ryanair faces on the affected routes - that is routes out of Ireland. Most importantly, it does not provide a useful test of any of the hypothesis identified by the Commission in section 2.

145. CRA agrees with the Commission's assessment on this point. They argue that "an analysis of routes out of Hahn is of obviously limited relevance to the markets affected by this transaction. Competition on city-pairs involving Frankfurt is likely to be quite unique, and clearly non-representative of competition on Irish routes, due to the peculiar features of both Hahn and Frankfurt Main. Hahn is a small regional airport where only Ryanair is present, and is clearly a very different base than Dublin, a primary airport where Ryanair faces strong competition. Moreover, competition from airlines flying into Frankfurt Main is also likely to be affected by the fact that a significant share of traffic into Main is accounted for by connecting passengers (given Frankfurt's importance has a European hub [...]*. Conversely, "real" monopoly routes out of Hahn are likely to be routes to secondary airports (where no other carrier is based), with relatively low and elastic overall demand. These features can both explain the fact that fares on monopolist routes are relatively low and Ryanair's position as a monopolist on some routes, and can therefore bias the econometric results". The Commission agrees with this argument and therefore concludes that RBB's second report on Frankfurt-Hahn can be given no weight in assessing the likely competitive effects in this case.

7. Fixed-effects regression analysis

146. The results from cross-section analysis (or RBB's two-step methodology) in this case are not sufficiently robust for at least two reasons. First, the number of independent observations is constrained to the number of routes operated by the merging parties and is therefore rather small. Second, even restricting the sample to routes out of Ireland there is no reason to think that possible omitted variable bias can be ignored. One possible solution to the omitted variable problem in cross-section regressions is to try to control for sources of route heterogeneity that likely affect prices such as such as the type of destination, the popularity of the route according to purpose of travel, customer awareness, destination airport characteristics, number of alternative airports at destination, safety considerations, duration of travel etc. The problem is that these variables may be unobserved or difficult to measure accurately. As a result the Commission has little confidence that it is possible to specify completely (or accurately measure) all the exogenous factors that may differ across routes yet influence the merging parties' fares.

147. An alternative way to control for differences across routes is to view the unobserved factors affecting fares as consisting of two types: those that are constant and those that vary over time. Consider the following basic specification: (formule) denotes for example the (log) price of Aer Lingus in route i in month t. Xit denotes the explanatory variable of interest, for example, a dummy variable indicating whether Ryanair is present on route i in month t. The variable ai captures all unobserved, time constant factors that affect Yit. The fact that ai has no time subscript tells us that it does not change over time. Generally ai is called an unobserved effect. It is also known in applied work as "fixed effect". A model that includes ai is often referred to as fixed effect model.

148. Both CRA and the Commission assume that the unobserved or fixed effect is a coefficient to be estimated for each route i. The way to estimate an intercept for each route i is to include a dummy variable for each route, along with the explanatory variables (51). Also dummy variables for each time month period are included. These are known as the time fixed-effects.

149. The main advantage of the fixed-effects approach is that the route fixed-effects capture all factors affecting prices that differ across routes but do not vary over time. Thus such factors can be omitted from the regression specification without any risk that this might lead to omitted variable bias. In contrast, in the case of cross-section regressions if an important variable, say flight duration, was omitted but this variable was correlated with any of the explanatory variables this would lead to biased estimates (52).

150. The Commission's preferred regression approach is thus to employ a fixed-effects procedure to estimate directly the effect of route and rivalry on prices. Incidentally, this is also the approach followed by Aer Lingus' expert economists. Ryanair's economists also acknowledge the merits and suitability of fixed-effects regression analysis for this case. RBB, ex-ante, first considered using a fixed-effects approach. However as mentioned in paragraph 117, the lack of variability in the data set with respect to the entry/exit pattern of Aer Lingus on Ryanair routes diminishes the value of this method when Ryanair's prices are the dependent variable.

151. More generally, fixed-effects is a suitable approach when (i) data contains many examples of entry and exit over time (ii) unobservable influences on prices are timeinvariant in a given route (iii) there is little reason to expect inaccuracies in measuring key explanatory variables. These conditions are all met in this case with respect to regression assessing the impact of Ryanair on Aer Lingus' prices:

* First, there has been a consistent pattern of recent entry by Ryanair on routes where Aer Lingus already operated. Aer Lingus has also entered (or exited) routes where Ryanair was present but to a much lesser extent. As a result the Commission agrees with RBB that the first condition is not met to the necessary degree when considering the impact of Aer Lingus on Ryanair's prices

* Second all explanatory variables of interest, for example presence in the route, number of monthly frequencies or capacity in the route does not raise severe measurement concerns.

* Third, the Commission has detected few inaccuracies when comparing data on the same variable submitted by the various parties.

7.1. CRA reports on behalf of Aer Lingus

152. On 23 November 2006 Aer Lingus' expert economics (CRA) submitted an econometrics report. CRA uses a fixed-effects procedure. This report was followed by two other submissions complementing and extending the initial results on 6 December 2006 and 9 February 2007.

7.1.1. CRA's report "Econometric Assessment of Competition between the two Airlines" (23 Nov)

153. The main objective of the analysis was to investigate whether Aer Lingus' average fares and load factor on a given route systematically differ whenever Ryanair also serves that particular route, and therefore whether there is any evidence that Ryanair imposes a competitive constraint on Aer Lingus.

154. Specifically, the analysis aims to test empirically whether Aer Lingus and Ryanair actually compete on routes where they fly to the same destination that is defined either at the airport-pair level or at the city-pair level (e.g. Paris CDG and Paris Beauvais are treated as the same destination).

155. CRA's analysis is based on two main sources of data:

* Monthly data on Aer Lingus' average one-way fare and load factor on all of its shorthaul routes out of Dublin from January 2003 to September 2006; and

* Monthly data from the Dublin Airport Authority (DAA) on the total number of scheduled departures and seats (i.e. capacity) by airline on all routes out of Dublin, during the period April 2004 to September 2006.

156. CRA merged both sources of data by route and month and constructed two alternative data sets. The first data set contains Aer Lingus' average fare, load factor and the total seat capacities of all the airlines (including Aer Lingus and Ryanair) for 55 routes at the airport-pair level, while the second data set contains this information for 54 routes at the city-pair level. The first data set thus aims to address the question of whether Ryanair poses a particularly close competitive constraint on Aer Lingus if both airlines fly to the same airport; the second extends this question to city pairs, i.e. those several routes where Ryanair does not fly to the same airport as Aer Lingus, but to the same city destinations.

157. CRA's regression analyses examines which factors drive Aer Lingus' average monthly prices and load factors, conditional on Aer Lingus serving a particular route in a given month. The tested hypothesis is that both variables are negatively affected by the "presence" or "strength of presence" of Ryanair in the route during the period of analysis. CRA considers three alternative measures of Ryanair's presence on an Aer Lingus: (i) share of total seat capacity, (ii) absolute monthly seat capacity and (iii) a dummy indicating presence of Ryanair in the route.

158. CRA employs a standard panel fixed-effects procedure. Panel data analysis examines the behaviour of fares on multiple routes over time. As explained in section 4, panel data regressions with fixed-effects exploit the fact that market structure (i.e. the presence of competitors) changes over time, and analyses the effect of such changes on fares. For example, when examining an Aer Lingus route where Ryanair entered halfway through a given period, the panel data regression compares the level of Aer Lingus fares on that route after Ryanair entered, with the level before Ryanair entered. CRA motivates the use of fixed-effects in that this methodology can mitigate the problem of omitted variables, as the unobservable cost and demand factors that are likely to affect fares across routes (such as the characteristics of the destination) are likely to be relatively constant over time.

159. CRA's base specification model the log of Aer Lingus' average monthly fare or the log of its average monthly load factor as a function of (53):

160. The log of Aer Lingus' total monthly capacity from Dublin to a particular destination;

* A set of monthly dummy variables accounting for aggregate time effects that impact Aer Lingus' average fares and load factor on all routes, such as for example changes in fuel prices over time, and seasonal effects;

* A set of route-specific fixed-effects to account for different characteristics and market dynamics that are peculiar to each route (these fixed-effects for example proxy for the distance between Dublin and the particular destination);

* A two-class dummy that proxies for Aer Lingus' abolition of premier class in November 2004 and March 2005, and the switch to an-economy only service

* Aer Lingus and Ryanair's share of seat capacity supply on a given route to capture the competition effect between the two airlines as well as the share of seat capacity supply of the airlines other than Aer Lingus and Ryanair.

161. In all the regressions, the coefficient on Ryanair's market share is negative and statistically significant at the 5% significance level with the exception of the load factor in the "city-pair" regression, where Ryanair's share coefficient is significant at 10% level. This suggests that whenever Ryanair is also operating on Aer Lingus' routes, Aer Lingus' fares and load factor are systematically lower on average than Aer Lingus' fares and load factor on routes that are not served by Ryanair. Aer Lingus' capacity variable is statistically significant, positively related to prices and negatively related to load factors.

162. A number of month- and route-specific effects are also statistically significant. This matters insofar as these dummy variables capture influences of other variables such as the distance of a given route, the total size of the market, or in the case of month dummies, time-varying effects. According to CRA this implies that the control variables serve as important corrections in the regression ensuring that the effects of the Ryanair on Aer Lingus is not over or underestimated.

163. In summary, CRA results indicate that on routes where both carriers are operating, Aer Lingus' fares and load factor are systematically lower on average than on routes which are not served by Ryanair. Moreover, it appears that the effect on Aer Lingus prices is greater when Ryanair is present than when other carriers are present on routes.

7.1.2. CRA's report "Ryanair/Aer Lingus: Further econometric analysis" (6 Dec)

164. CRA extended its earlier analysis on a separate note submitted on 6 December 2006. This analysis is based on the same data on capacity between city and airport pairs for short-haul routes flown out of Dublin by Aer Lingus, as well as average fares and load factors for Aer Lingus on these routes, as was used in the earlier memorandum. This information is used to calculate dummy variables that indicate: (i) Ryanair's presence on airport-to-airport routes flown by Aer Lingus out of Dublin; (ii) Ryanair's presence on routes out of Dublin serving another airport for the same cities; (iii) other airlines' presence on the same airport-to-airport route; and (iv) other airlines' presence on routes out of Dublin serving another airport for the same cities.

165. As before, they use a "fixed-effects" approach to control for both differences across routes (with route fixed-effects), and aggregate differences over time (with time fixedeffects). The explanatory variables of interest are dummy variables that measure the presence of Ryanair and other airlines at (a) the airport served by Aer Lingus and (b) at other airports serving that same city. This means that the results are based on the effects of changes over the sample period in Ryanair and other airlines' presence on the routes considered. As controls they include the log of Aer Lingus' total monthly capacity from Dublin to a particular destination; a zero-one variable that indicates if Aer Lingus was offering a two class or economy class only service and a set of time-specific fixed-effects that account for aggregate effects that vary over time and impact on Aer Lingus' average fares and load factors on all routes, such as variations in fuel prices.

166. CRA reports the following results:

* Aer Lingus' average fares are reduced by [5-10]* percent by Ryanair's presence at the same airport, by [0-5]* percent by Ryanair's presence at a different airport serving the same city, by 1.8 percent by other airlines at the same airport, and by 0.8 percent by other airlines at another airport;

* Aer Lingus' load factors are reduced [5-10]* percent by Ryanair's presence at the same airport and [0-5]* percent by Ryanair's presence at a different airport; however they are reduced by 0.7 percent by other airlines at the same airport, and there is no effect for other airlines at another airport.

167. According to CRA these results confirm that Ryanair's presence both at the same airport and at a different airport serving the same city has a substantial impact on Aer Lingus' fares, load factors and revenues. They also indicate that the impact of Ryanair is much greater than that of other airlines, confirming that Ryanair is Aer Lingus' closest competitor on these routes out of Dublin.

7.1.3. CRA's report "Ryanair/Aer Lingus: Econometric analysis" (9 Feb)

168. CRA submitted a third report extending further the results in the two earlier reports. First they replicate the analysis given in the 23 November report on a large dataset which covers the period as of January 2002 until November 2006. (54) With a longer panel data CRA obtains similar results as in previous submissions. The estimated impact of Ryanair on Aer Lingus is significant and marginally larger. The airport pair data indicates that Ryanair presence reduces Aer Lingus average total fares by [5-10]* percent and net fares by [5-10]* percent. The city pair data indicates that the Ryanair presence reduces Aer Lingus average net and total fares by [5-10]* and [5-10]* percent respectively (55).

169. The CRA study also uses data from Aer Lingus on revenues and passengers by sales region from 2004 to 2006 to examine the effect of competition from Ryanair on passengers originating in Dublin and passengers originating from elsewhere. This analysis indicates that the competitive effect of Ryanair on Dublin originating passengers is larger than for non-Dublin passengers. Competition from Ryanair reducing fares for Dublin origin passengers by up to [10-15]* percent, which is a very substantial price effect. These results are again consistent with a proposition that having a base at Dublin (and more generally a strong market presence in Ireland) leads to more intense competition for Irishoriginating passengers. According to CRA replicating this competitive pressure without a base at Dublin would be extremely difficult.

170. A third set of regressions includes the presence of competitors in the route. The results indicate that Ryanair's presence at the same airport reduces Aer Lingus' average net fares by [5-10]* percent; Ryanair's presence at another airport serving the same city reduces Aer Lingus' average net fare by [5-10]* percent; other airlines' presence at the same airport reduces Aer Lingus' average fare by 1.9 percent but is not statistically significant; and other airlines' presence at other airports reduces Aer Lingus' average fare by 3.1 percent (again not statistically significant). According to CRA this confirms that Ryanair is a much closer competitor for Aer Lingus than other airlines, which is consistent with the view that base competition at Dublin is an important source of competitive constraint.

7.1.4. The Commission's assessment of CRA's fixed-effects regressions taking RBB's taking "data-room submission" into account

171. The Commission considers the fixed-effects approach proposed by CRA appropriate to assess the mutual influence of the merging parties on each other on the affected routes. One obvious limitation of the CRA analysis, due to the lack of data, is that it only pertains to the effect of Ryanair on Aer Lingus. However, as RBB has pointed out and the Commission's own analysis suggests, the fixed-effects approach fails when analyzing the effect of Aer Lingus on Ryanair. This is due to insufficient variation in the presence of Aer Lingus when Ryanair is in the route. In contrast Ryanair's entry into a significant number of routes in the past few years where previously Aer Lingus was active provides an useful before-and-after test to assess the constraint that Ryanair's presence may exert on Aer Lingus', avoiding the problem of unobserved (or uncontrolled for) differences across routes.

172. RBB argues that CRA regressions (56) provide no meaningful indication of the impact of Ryanair on Aer Lingus (57). They criticize CRA's "capacity share" regressions, which according to CRA allow measuring the relative significance of Ryanair's presence on a route. RBB raise two concerns. First, RBB shares the Commission concern (see Annex 4 in the Statement of Objections, paragraph 167) that the inclusion of contemporaneous measures of concentration as explanatory variables can lead to biased results and the standard errors may not be informative as regards significance levels. Second, "CRA's original baseline regression (the "capacity share" approach) imposes strong restrictions on the estimated coefficients which are not consistent with theory". RBB points out that the explanatory variables employed by CRA in the regressions including "capacity shares" are linear combinations of each other. The Commission agrees with this argument. Essentially CRA's regressions that include capacity shares as explanatory variables implicitly assume that the coefficients of interest are related to each other in a specific way. To see this RBB notes the following ways of writing CRA's capacity share regression: (58)

(formule)

173. where p is the log of Aer Lingus' fare and EI and FR refer to Aer Lingus and Ryanair respectively. This can also be written:

<emplacement tableau>

174. RBB correctly remarks that "if [2] were estimated, the coefficient on EI share would be (ß1 - ß2 ), while if [3] were estimated the coefficient on FR share would be - (ß1 - ß2 ), i.e. the negative of the coefficient on EI share in [2]. Further, the impact of Others' share in [3] is the negative of FR share in [1]. Clearly the specification involves prior restrictions - further, the restrictions are not reasonable in terms of theory as (a) the impact of Others' share and FR share would not be expected to have equal and opposite effects on Aer Lingus' fares; and (b) the impact of EI share and FR share would not be expected to have equal and opposite effects on Aer Lingus' fares" (59).

175. The Commission agrees with this analysis. As a result the Commission does not attach any relevance to the results obtained by CRA based on the specifications that include "capacity shares" as explanatory variables (that is, tables 1 and 2 in the November 23 submission and table 2 and table 4 in the February 9 submission).

176. The Commission, however, notes that the results based on regressions that include a "presence" dummy (similar to the Commission's own fixed effect regressions) are not affected by this criticism. This refers to table 3 in the November 23 submission, tables 1 and 3 in the February 9 submission and also the whole of the results included in the December 6 report. All these regressions consistently indicate that Ryanair's presence is strongly associated with lower Aer Lingus prices in the order of [5-10]* percent.

177. RBB also raises two concerns with respect to CRA's "presence specification". First, RBB criticizes the use of use robust regression techniques without demonstrating the need for this technique. The Commission uses the same technique in its own fixed-effects regressions. "Robust regression" is a regression algorithm that differs from standard ordinary least squares (OLS) in that it assigns different weights when calculating the line of best fit. The motivation for this approach is that, to the extent that outliers are present in the data OLS, results can be very much driven by a few extreme observations. According to RBB, CRA uses this technique "due to an erroneous belief that this technique resolves omitted variable and heteroskedasticy problems" (RBB's Dataroom submission). This is simply false. CRA makes it clear that "all of the above models were estimated using robust regressions to eliminate outliers and improve the regression fit" see CRA's November 23 report, section 2.4. It is true that CRA does not motivate the use of "robust regression". However, the Commission regards the use of "robust regression" as entirely justified as explained further in section Use of robust regression estimation7.3.1.2.1 below.

178. Second, according to RBB, CRA's presence regressions are likely to suffer from omitted variable bias. They argue that "the most likely cause of this relates to the failure to control for demand appropriately" (60). However, CRA's regressions include Aer Lingus capacity, which CRA argues is an adequate control for demand in the route. Indeed, consistent with this assumption the sign of this variable in all of CRA's price regressions in positive and significant.

179. Notwithstanding the above, RBB argues that capacity is an endogenous variable simultaneously determined with prices and therefore one would expect the sign of the capacity coefficient to be negative. According to this interpretation a positive sign is "contrary to the law of demand that the relationship between price and quantity is negative".

180. The Commission believes that whether the "capacity" variable is an adequate control for demand in the route or instead, as RBB argues an endogenous variable simultaneously determined with prices (resulting in endogeneity bias) depends on the circumstances in which the merging parties make their capacity decision. To this effect, soon after starting the investigation, the Commission explicitly requested from CRA a clarification on this point.

181. In communications with the Commission's economists CRA has acknowledged that the endogeneity of the capacity variable can be a problem. However, CRA has argued that this potential problem is mitigated by certain features of the market and of the analysis.

182. Firstly, airlines generally determine their allocation of capacity twice yearly normally at least three months before the start of the period. Once the schedule is set, load management software over the selling period varies the price to maximize returns and hence both quantity and price are endogenous. The key point is that airlines have to determine the level of capacity well before they observe demand and hence the level of capacity is predetermined relative to demand.

183. Second, by including fixed-effects, the estimates of the parameters of interest are a function of the effect of changes in the variable for a particular route, not of the effect of variations in the variable across routes. The present study also concerns the same firms competing across multiple markets. Hence, standard stories about the effect of endogeneity (which concern a cross-section of industries with different firms) do not apply.

184. Third, the fixed-effects procedure allows capturing the impact of entry/increases in capacity by Ryanair on Aer Lingus. Such entry/expansion would likely lead to falls in average fares and load factors by Aer Lingus. However, increased Aer Lingus prices due to increased demand would be also likely linked to Ryanair entry/capacity expansion (not the reverse). This in fact implies that the endogeneity bias would create a positive relation between Ryanair capacity and price. This in turn implies that endogeneity causes the magnitude of the (negative) competitive effects to be underestimated.

185. The Commission agrees with the first two arguments advanced above. However with respect to the direction of the endogeneity bias the Commission points out that if the costs of operating a certain route (or many) increases over time this would likely increase Aer Lingus' prices but also reduce the scope of entry or capacity expansion by Ryanair (and vice versa). This would lead to an overestimate of the absence of Ryanair on Aer Lingus prices. It is not possible without using more sophisticated techniques such as instrumental variable estimation, to determine the direction of the bias - or for that matter, whether these two sources of bias cancel out (61).

186. The Commission acknowledges that it is difficult to find good instrumental variables that could resolve any potential endogeneity problem in the fixed-effects regressions (62). However, we also point out that fixed-effects regressions are less likely to be affected than cross-sectional regressions given that most of the variation observed in the affected routes is across routes, not within routes.

7.2. The Commission's fixed-effects analysis

187. As explained above, CRA's submissions on behalf of Aer Lingus indicate that Ryanair exerts a significance constraint on Aer Lingus' pricing. The Commission, however, believes it is necessary, before accepting any such results to conduct its own analysis intended to (i) test the robustness of these findings and (ii) improve and extend the analysis based on data set compiled by the Commission as explained in section 3 of this report.

188. The Commission's fixed-effects regressions extend the analysis submitted by both CRA and RBB. The most significant differences are the following:

* It relies on an extended data set that allows running each regression analysis both on Ryanair and Aer Lingus prices.

* It distinguishes between different types of competitors besides Aer Lingus and Ryanair

* It includes indicators of base presence at the destination airport of rival to the merging parties

* It includes additional control variables for within route variation in demand and costs

189. Furthermore the Commission tests two different type of specification as explained below:

7.2.1. Regression specifications

190. The Commission's empirical strategy focuses on the impact of Ryanair's "presence" and "strength of presence" on Aer Lingus' average net monthly fares and vice versa.

191. The baseline fixed-effects regression is as follows:

(formule)

Where:

* The dependent variable is the average net monthly fares of first Aer Lingus and then Ryanair.

* ai is the route fixed effect (time invariant dummy variables =1 for the route and 0 otherwise). The ai dummy accounts for systematic but unobserved or non-measurable differences in costs or demand within that route.

* f(.) is a function of competitor variables. These are the explanatory variables of interest.

* Dt is a dummy for each time period (a month). The month dummies allows for an identification of cost shocks that affected all routes during the same time period.

* Xit is a vector of cost and demand controls added in certain specifications

192. As explained in section 3 above, the Commission has constructed two data sets. The first is consistent with the city-pairs market definition adopted in the SO. The second assumes that each airport close to the destination city is a separate market. All specifications, if appropriate, are performed on both data sets and relevant differences are pointed out.

193. As in the case of the cross-sectional regressions, we run a number of alternative specifications that essentially differ in the competitor variables included - see section 5.3. We first test the presence specification where a dummy variable for the other merging firm and for other rivals present in the route is included. Then we test the frequencies specification where the frequencies of the other merging firm and that of other rivals is included in logarithmic form (in addition to absence dummies). This specification allows measuring the sensitivity of each firm's fares to the strength of its various rivals in the route.

194. We first start with the simplest possible specification (or baseline) and gradually include relevant controls, paying particular attention to the statistical robustness and economic significance of the explanatory variables of interest. All results are presented in tables below that allow for direct comparison of coefficients in alternative, slightly more specific specifications.

195. In all cases, we employ a "robust regression" technique that controls for the influence of outliers (essentially it drops some outliers and assigns less weight to others). We start by reporting the results from fixed-effects regressions on Aer Lingus price. We consistently find that Ryanair exerts a competitive constraint on Aer Lingus' prices. Next, we turn to describing our efforts in applying the fixed-effects procedure to test the influence of Aer Lingus on Ryanair prices, which has lead to very limited results.

196. We consistently used panel fixed effect estimation for all the regression presented. In the present case, the introduction of fixed effect can capture differences across routes affecting price that are not explicitly considered as regressors. An alternative approach known as "random-effects" imposes the assumption that the route effect is uncorrelated with each explanatory variable. This assumption is often not valid in practice. Fixedeffects models offer the advantage that the route effect also captures all time-invariant factors affecting price and likely to be correlated with the exogenous variables. Comparing the "fixed effects" and the "random effects" techniques can be a test of whether there is correlation between the ai and the explanatory variables assuming that the explanatory variables and the error term are uncorrelated across time periods. Hausman first suggested this test (63).

197. In order to test for the suitability of fixed-effects over random-effects in the present case, we use a Hausman test for all the regressions64. A large value of the Hausman test statistic leads to the rejection of the null hypothesis that the route fixed-effects are uncorrelated with included explanatory variables and to the conclusion that fixed-effects are present (that is routes differ significantly and these differences are correlated with the explanatory variables of interest - for example route distance might be correlated with the presence of Aer Lingus across routes). The replications of the test have confirmed the correctness of the fixed effect model in practically all cases (65).

7.2.2. Regressions with Aer Lingus' monthly average net fares as the dependent variable

7.2.2.1. Presence Specification

198. The presence specification includes dummy variables for the presence of (i) Ryanair, (ii) one or more flag carriers and (iii) one or more non-flag carriers. Furthermore Ryanair has argued that besides Ryanair and Aer Lingus, other carriers, in particular Aer Arann and Cityjet have a base at Dublin airport. Allegedly, these carriers can exert a significant competitive constraint on the merging parties. We have introduced dummy variables for these two carriers separately from that of other non-flag carriers. This allows us to compare directly the effect of their presence both in absolute terms and relative to that of Ryanair.

199. The baseline regression also includes Aer Lingus' log of capacity (seats) as a scale factor and a dummy to capture the impact of Aer Lingus' final stage in its restructuring, namely the move towards an internet based sales strategy that was implemented in September 2004. The date of this variable coincides, roughly with the middle point in our sample which starts in January 2002. The results of this regression are presented in the first column of table 9 below.

TABLE 9 -PRESENCE SPECIFICATION - BASELINE

<emplacement tableau>

200. It is instructive to review the coefficients of this regression because they will prove robust to many alternative specifications. The coefficient of most interest that of Ryanair's presence, is 0.077 and significant at the 1% level. It means that over the period considered and across all routes, on average, Aer Lingus prices are 7.7% lower when Ryanair is also present in the route - controlling for Aer Lingus' own capacity and the presence of other competitors. This value is in line with that obtained by CRA in a similar specification (without distinguishing between different rivals).

201. Equally important is to point out that no other rival has a negative and statistically significant effect on Aer Lingus' fares. Aer Arann's presence is associated with a positive increase in Aer Lingus prices. This, of course, is counterintuitive. One explanation is that Aer Arann is often present when Ryanair is not. Since Ryanair's presence is associated with lower Aer Lingus' prices, Aer Arann's presence would then be generally associated with higher Aer Lingus' prices. Indeed, during the period of analysis in around 85% of observations where Aer Arann is present, Ryanair is not. Hence, it is not surprise that when Aer Arann's presence is interacted with that of Ryanair this effect becomes statistically insignificant as shown in the regression reported in column 2.

202. The Sept'04 dummy takes a value of 1 for the period before the move to the internet and a value of 0 after September 2004. It is highly significant both statistically and economically. It indicates that prior to September 2004 Aer Lingus prices were about 45% higher than after this period. The dummy corrects for the risk that the regression would assign to Ryanair's presence the general reduction in prices that was associated with Aer Lingus' restructuring. CRA included in its regressions a similar dummy to control for the time when Aer Lingus was offering business class in each route. As shown in column 3 this dummy leads to practically identical results.

203. Column 4 in table 9 reports the results of adding two dummy variables to indicate the presence of at least one flag and one non-flag carrier with a base at the destination airport (66). The presence of a non-flag carrier with a base at destination has a negative and statistically significant effect on Aer Lingus' prices. However the impact is economically small (-2.75%), particularly in relation to the impact of Ryanair (-7.6%)

204. Finally we point out that Aer Lingus' capacity has a significant and positive relationship with average fares (i.e. on average, more capacity is planned on routes with higher expected demand, and therefore higher average fares). Incidentally, this is consistent with CRA's results.

205. In all our regressions, the r-squared is high (in this case 0.88), which implies that the model fits adequately. This is largely due to the route and month fixed-effects. The route fixed-effects capture all time-invariant differences across route, for example distance or thickness of the route that might affect prices. The monthly dummies capture unobserved sources of variation over time, including seasonal peaks in demand. The Hausman test in all regressions in table 9 is statistically significant implying that the random effects model is rejected in favour of the fixed effect model.

206. As an additional robustness check, we plotted the values of the monthly coefficient to assess whether the pattern of movement over time is consistent with the expected seasonal variation in demand. This indeed is the case as shown in the graph below:

<emplacement tableau>

207. It is evident that the month coefficients adequately control for the expected seasonal surges in demand in the summer period (with peaks at Sept 03, Aug 04, Aug 05, and Aug 06). In addition, note that the time fixed-effects pick up a downward sloping trend and therefore controls for reductions in price unrelated to the presence of Ryanair in the route. The time fixed effect therefore ensures that the competitive constraint exercised by Ryanair on Aer Lingus is not overestimated.

208. We also plotted the route fixed-effects against the distance to check our assumption that these dummies control for the increase in fares that result from the higher costs of operating longer distance routes. Again, we observe the expected positive association between distance and the route fixed-effects - as shown in the graph below. This strengthens our confidence on the robustness of the specification.

<emplacement tableau>

209. Table 10 reports the results from adding a number of additional controls to the baseline specification. In this set of regressions, we drop the capacity variable. We introduce first a direct measure of total costs in the route (ln_EI_costs) (67). We also introduce the log of total frequencies offered by all carriers at the destination airport (ln_dest_freq_total). This variable controls for the traffic at the destination airport and is therefore a good proxy for variations in demand (anticipated at least at the time that carriers set their frequencies). Both variables turn out to be highly significant in all specifications. As shown in the first column there is little variation in the presence variables. In particular, Ryanair's presence remains statistically highly significant and even increases slightly (7.9%).

210. The second regression (column 2) introduces the "base at destination" variables. Again, the presence of a flag carrier based at the destination has no effect on Aer Lingus prices whereas the presence of a non-flag carrier with a base has a small effect but 1% point larger than in the baseline regression. It should be noted that this variable aggregates the effect of all non-flag carriers with a base. It is also interesting to point that non-flag carriers actually present in the route do not seem to constrain Aer Lingus prices.

TABLE 10 -PRESENCE SPECIFICATION - ADDITIONAL CONTROLS

<emplacement tableau>

211. We have also explored whether the presence of Ryanair primarily affects Aer Lingus prices in the routes where they fly to the same airport. For this purpose we introduced two dummies for Ryanair. The first takes the value of 1 when it serves the same airport as Aer Lingus in the given route, and zero otherwise. The other takes the value of 1 when it serves a different airport, and zero otherwise. The fourth column includes the "base at destination" dummies whereas the regression in column 3 omits these variables. In both cases the presence of Ryanair remains significant and around 7% on average. In fact, Aer Lingus appears to be slightly more sensitive to Ryanair's presence on the route when the latter serves a different airport. The difference does not appear to be important but in any event these results rebut the argument that when Ryanair serves a different airport it does not compete with Aer Lingus in the relevant city-pair.

212. As a final remark, note that the coefficients of all the remaining variables are very similar to that of the baseline regression in table 9. As an additional test for the correctness of the specification, the Hausman test is statistically significant in regression in columns 1,2 and 4, implying that the fixed effect model is preferred in terms of efficiency to the random effect one in the present case. The Wooldridge f-test also confirms this result for regression in column 3.

213. Before introducing a more refined specification, we report the results from regressions, which include the route HHI (based on number of frequencies) (68). It is often highly controversial, and in any event not advisable to introduce a direct measure of concentration into any price regressions. This is because in a reduced form price specification as the one we use, the measure of concentration is likely to be endogenous, thus leading to biased estimates. With this caveat in mind the Commission believes it is its duty to report the results from the regressions including the HHI, in particular because including this variable leads to a small reduction in the influence of Ryanair's presence on Aer Lingus prices.

TABLE 11-PRESENCE SPECIFICATION - HHI's

<emplacement tableau>

214. As shown in the table the presence of the route frequency HHI reduces the impact of Ryanair's presence on Aer Lingus by around 3% percentage points to 4.5%. However, it remains negative and highly statistically significant. It also worth noting that the regression presence of non-flag carriers becomes significant but positive, that is, it leads to higher, not lower prices. This is contrary to what is theoretically expected and is an indication that the introduction of the HHIs in the regression may lead to endogeneity problems. On the other hand an alternative explanation for this result has been suggested by CRA - see paragraph 242 below.

Airport-pairs database

215. As already mentioned we tested the presence specification also in the airport-pairs database. In this case, the maintained assumption is that if Ryanair serves a different airport to a given city it does not compete at all with Aer Lingus. This assertion has been made by Ryanair. The results of these regressions are presented in table 12 below.

TABLE 12 -PRESENCE SPECIFICATION - AIRPORTS

<emplacement tableau>

216. Note that Ryanair continues to be the only carrier that constrains Aer Lingus. Its presence leads to approximately a 5% reduction in Aer Lingus' fares. It is also noteworthy that a non-flag carrier with a base at the destination also appears to have a relevant effect. If these two dummies are highly collinear, i. e. a non-flag carrier is based at the destination with higher frequency when Ryanair is in the route, it could be argued that the model is not able to disentangle these two separate effects. To solve this ambiguity, we added an interaction term (non_flag_b_ryan) between Ryanair presence the non-flag base dummy in the regression in column 4. This isolates the effect of the non-flag-base when Ryanair is not present. The interaction term is not significant suggesting that the full effect on Aer Lingus derives from Ryanair's presence on the route, not that of a flag carrier with a base at the destination.

Intermediate specification

217. As an intermediate step between the presence and frequencies specifications we tested a specification which includes the frequencies of Ryanair (plus the required absence dummy). However for the other carriers we only indicate their presence. Table 13 below presents the results of these set of regressions.

218. Column 1 reports the results from the baseline regression. The log of Ryanair's frequencies has a negative and highly significant effect on Aer Lingus' prices. In economic terms, the impact is even more striking than when only Ryanair's presence is considered (although both regressions are not directly comparable). The coefficient of 0.028 can be interpreted as follows: A 1% increase in Ryanair's frequencies (that is, a rather small increase in capacity in the route) is associated with a 0.028% reduction in Aer Lingus' prices.

TABLE 13 -INTERMEDIATE SPECIFICATON

<emplacement tableau>

219. Controlling for demand and adding the "base at destination" variable does not reduce but in fact slightly increases the influence of Ryanair. All other presence variables, including the effect on non-flag based rivals remain similar to the values in the presence specification.

220. Column 4 reports the result from distinguishing whether Ryanair operates in the same or in a different airport to Aer Lingus at the destination city. Once we measure the strength of presence, instead of the mere presence, it appears that Ryanair imposes a more significant constraint on Aer Lingus when it serves the same airport, as would be expected. However, note that the effect at a different airport is still significant and in fact larger than in the baseline regression from column 1.

221. In this, our last regression with presence dummies it is interesting to report that the route and month time effects appear to be working properly. In fact, as regards the time fixed-effects, the addition of further demand and cost controls relative to the first baseline specification appears to account for the observed downward trend. There is no such trend in the time coefficients from the last regression in table 5 above.

<emplacement tableau>

222. Likewise, the regression of distance on the route fixed-effects remains highly significant and the goodness of fit is high. This positive relationship can be visualized in the scatter plot below:

Graph 4: City Fixed-effects by Distance

<emplacement tableau>

7.2.2.2. Frequency specification

223. The frequency specification is intended to test the fourth hypothesis listed in section 2, namely that the stronger the presence of one of the merging parties the more pronounced the effect on fares of the other.

224. The hypothesis is difficult to test. This is because to measure the strength of a carrier on a route it is necessary to use some capacity variable, such as number of frequencies on the route, which is likely, to some extent to be determined simultaneously with fares and the frequency levels of rival carriers.

225. The results of the baseline frequency specification are reported in the first column in table 14 below.

TABLE 14 - FREQUENCY SPECIFICATION - CITYPAIRS

<emplacement tableau>

226. As anticipated in our intermediate specification the log frequencies of Ryanair has a significant and negative effect on Aer Lingus' prices. This effect is robust to the inclusion in the regression of base destination dummies, demand controls and even the frequencies HHI.

227. In this set of regressions the frequencies of non-flag carriers also have a negative effect on Aer Lingus' fares but much smaller in magnitude (less than a third) than that of Ryanair. The demand control (ln_dest_freq_total) is highly significant, as in the presence specification.

228. Again, the presence of a non-flag carrier with a base at the destination has a very similar effect as in the presence specification (note, however, that the coefficient of Ryanair frequencies and that of the base dummies are not comparable).

229. The same regressions distinguishing between Ryanair at the same airport or at a different airport as in the last column in table 13 give again very similar results, albeit slightly higher in value. In this regression (not reported in table 14) 1% increase in frequencies by Ryanair is associated, respectively, with a 0.044% and 0.026% increase in Aer Lingus' fares when serving the same airport or a different airport in the same citypair. Regressions with Ryanair's monthly average net fares.

7.2.3. Regressions with Ryanair's monthly average net fares as the dependent variable

230. We are not convinced that it is meaningful to apply the fixed-effects method on Ryanair's prices. The reason, as pointed out by Ryanair's expert economists is that there are very few instances where Aer Lingus has entered (or exited) a route previously served by Ryanair or has undertaken a large expansion in capacity. In contrast, Ryanair has entered into a large number of routes where Aer Lingus was already active out of Dublin.

231. To increase the possibility of capturing more events of Aer Lingus entering, exiting, expanding or contracting it is possible to consider the period since Ryanair fare data is available, that is April 1997. There are two problems with this approach. First, for the reasons previously mentioned in section 3.4, the within route effect of Aer Lingus' presence on Ryanair fares earlier than 2002 relative to the more recent past is possibly not an accurate reflection of the competitive constraints Aer Lingus may be exerting on Ryanair. Moreover, important variables controlling for exogenous shocks are available as of January 2002 only. The omission of several of these variables would likely bias the regression results. A further concern is that there has been a gradual but persistent downwards trend in Ryanair prices since the late 90s (even unadjusted for inflation). A regression that relies on comparisons of prices so far back in time is likely to lead to spurious associations.

232. In spite of our reservations and largely for the sake of being exhaustive we run a number of regressions of Ryanair's prices on Aer Lingus presence and frequencies using the full sample that includes Ryanair fares as of April 1997. Because we do not have data for most of our preferred control variables, we only run regressions on the baseline specifications.

233. As shown in table 7 Aer Lingus' presence appears to have a significant negative effect on Ryanair's prices, close to -10%. This is the only rival that appears to constrain Ryanair in any way. The regression in column 3 suggests that this effect is largely on city-pairs where Aer Lingus serves the same airport as Ryanair. Holding the presence of rivals constant Aer Lingus also appears to have a significant and negative effect on Ryanair as shown in column 4.

TABLE 15 - PRESENCE SPECIFICATION - BASELINE

<emplacement tableau>

234. The frequency specification (not reported) confirms the above results. Aer Lingus is the only carrier that has a negative and significant influence on Ryanair, and this effect is observed only when it serves the same airport.

7.2.4. Limitations of the fixed-effects regressions in this case

235. The fixed-effects procedure is subject to two caveats. Firstly, it is based on the assumption that entry and exit decisions are exogenous. This assumption may only be approximately correct. It is possible that a high Aer Lingus price on a route makes it profitable and so encourages entry and expansion by both Ryanair and its rivals. If so entry or expansion would be endogenous. This means that the estimates may be subject to some selection bias since they are conditional on the carrier being present on a given route.

236. A second problem is that the frequency variables may also be possibly endogenous, although it seems sensible to assume that airlines set these frequencies at least a few weeks in advance and then optimize their pricing and load factors conditional on the preset frequencies.

237. In theory, these problems can be addressed by instrumenting the explanatory Variables (69). The Commission has tested a number of candidate instruments included in the data set, such as intra-route frequency rank, own costs or own total frequencies at destination airport. However all these instruments turned out to have very poor properties.

7.3. The parties comments on the Commission's fixed-effects regressions following the Statement of Objections

238. The expert economists of the merging parties were granted separate access to the data and the Commission's computer code in the "data room" in order to allow them to replicate, comment and criticize the Commission's analysis and interpretation of the results. RBB was granted access to the data room on April 4, 10 and 13. CRA was granted access to the "data room" on April 11, 12 and 16. This section reports on the comments regarding the Commission's analysis included in CRA and RBB's "Dataroom report" submitted on April 18 and April 13, respectively.

7.3.1. Regressions with Aer Lingus' monthly average net fares as the dependent variable

7.3.1.1. CRA Comments

239. CRA agrees with the Commission's focus on fixed-effects, which "mirrors" their own analysis. CRA makes a number of comments broadly in support of the conclusion reached by the Commission from its own fixed-effects regressions that Ryanair exercises a strong competitive constraint on Aer Lingus' prices.

240. First CRA points out that with respect to regressions on Aer Lingus' average fares the R2 statistic is about 88 percent in all the tested regressions for both presence and frequencies specifications. This implies that the bulk of the variation in fares is actually explained by the model.

241. Second, in all the presence specifications, the coefficient on Ryanair's presence variable is negative and statistically significant at 1%, which implies that the coefficient is estimated precisely and that competition from Ryanair does have an impact on Aer Lingus fares. CRA notes that the impact is found to be about 7.7% for a range of specifications.

242. CRA also argues that the Commission's discussion of the HHI regressions (70) misinterprets the regression coefficients, and that the reported coefficients are in fact consistent with the magnitudes reported in the presence regressions. CRA correctly points out if Ryanair enters on an Aer Lingus route, this will have two effects: it will both lower the HHI and it will also switch the route status from no-Ryanair presence to Ryanair presence. The coefficient on the HHI variable is positive and significant. Hence if Ryanair enters part of the (negative) effect will come from the effect of the fall in the HHI. Alternatively, if a flag carrier enters with a similar frequency, this will again have the same effect through the HHI, though the effect through the flag presence variable will be different.

243. The Commission agrees with the above argument and with the implication that it is reasonable to interpret the HHI coefficient as the effect of a change in the HHI by an "average" entrant, and the various presence variables as the additional effect of a particular presence relative to the change caused by the change in HHI. According to this interpretation Ryanair's entry reduces Aer Lingus' fares by (a significant) 4.4% below the effect of an "average" entrant (see table 11). Flag carrier entry reduces Aer Lingus fares by (an insignificant) 1.05% less than the "average" entrant and a non-flag entrant by 2.7% less than the "average".

244. In conclusion, according to CRA the Commission thus finds clear evidence that Ryanair significantly constrains Aer Lingus' pricing behaviour. They also note that the results of the Commission's fixed-effect regressions are in line with those submitted by CRA on behalf of Aer Lingus. This pertains to both the magnitudes of the estimated coefficients, and the conclusion that the Commission reached.

7.3.1.2. RBB Comments

245. According to RBB the Commission's fixed-effects regressions do not provide robust evidence to support the view that Aer Lingus and Ryanair constrain each other. RBB raises five criticisms to the Commission's analysis.

??First, the Commission's use of robust regression is not necessary in the case at hand and indeed poses severe shortcomings of the analysis in the face of serial correlation.

??Second, the Commission's specifications are highly sensitive to small changes in the way time fixed-effects are modelled.

??Third, serial correlation of the errors in the Commission's specifications invalidates the standard errors which the Commission uses for hypothesis testing.

??Fourth, the Commission's specifications suffer omitted variable bias in that they fail to control appropriately for route specific demand conditions and/or fail to account for endogenously determined fares and frequencies.

??Fifth, the frequency specification used by the Commission suggests the constraint Ryanair exercises on Aer Lingus is not economically significant.

246. In light of these criticisms, Ryanair in its response to the Statement of Objections concludes that the constraint that Ryanair exercises on Aer Lingus' prices is "very weak". According to Ryanair, its expert economists "demonstrate that the Commission's evidence (and that provided by El) is at best non-robust (i.e., highly sensitive to small changes) and at worst meaningless (i.e., based on misspecified models). In short, the Commission has not put forward robust empirical evidence to show that FR constrains El". The Commission disagrees with this conclusion. In this section we discuss and respond to all the criticisms raised by RBB.

7.3.1.2.1. Use of robust regression estimation

247. RBB correctly points out that all of the Commission's econometric results are based on so-called "robust regression" implemented via rreg, a regression algorithm that differs from standard ordinary least squares (OLS) in that it assigns different weights when calculating the line of best fit. The motivation for this approach is that, to the extent that outliers are present in the data OLS, results can be very much driven by a few extreme observations.

248. RBB claims that the Commission does not explicitly motivate its use of robust regression. According to RBB the Commission must explain why it believes that "robust regression" provides a better approach than OLS or other techniques that deal with outliers such as quantile regression (implemented with the qreg command in STATA - the econometrics software used by both merging parties and the Commission). RBB argues that the use of robust-regression is not necessary and that as a result for most specifications the impact of Ryanair presence is overestimated by robust regression compared to OLS.

249. The Commission first remarks that it is not clear why RBB regards the use of robustregression as unnecessary. First RBB remarks that for "all presence specifications the point estimates using robust regression do not differ substantially from the OLS coefficients or the median regression coefficients" (i.e. using qreg instead of rreg). (71) However in the next paragraph RBB claims that "specifically, for the presence specifications, robust [rreg] and quantile [qreg] regressions differ significantly and OLS lies between the two approaches".

250. Somewhat more clearly RBB argues that "diagnostic plots such as the leverage versus residuals squared plots confirm that there is no reason to believe that OLS results are driven by a few outliers". Thus, according to RBB, the use of robust regression is not justified in the present case, given that the results of the basic OLS estimation seems not to be affected by the presence of outliers, i. e. the problem robust regression is supposed to solve.

251. The Commission does not agree that the presence of outliers can be neglected. The graphs below report the leverage versus squared residuals of an OLS regression and a robust regression both applied to our basic presence specification, as reported in column 1 of table 9. Each of the graphs reports on the horizontal axis the squared residuals, which is a measure of whether the observations in the data set are "well behaved" The leverage, reported in the y axis, gives instead a measure of the role played by each observation in the estimation process: the more this is close to one, the more the observation is determinant in the identification of the fares predicted by the model.

Graph 5 Leverage squared residual - OLS - presence specification.

<emplacement tableau>

Graph 6 Leverage squared residual - Robust Regression - presence specification.

<emplacement tableau>

252. Even if there are not visual extreme cases of outliers, the graphs clearly show the effect of the different computational algorithms used by each of the two models. The fact that robust regression gives different weights to the observation according to their behaviour reduce the effect of odd observations and lead to a concentration of the points next to the axes. This is the objective of using robust regression.

253. An additional, ex-ante, reason to use robust regression is that it would eliminate the influence of errors and inconsistencies in the data used (72). The Commission recalls that it detected several inconsistencies in the data submitted by Ryanair and Aer Lingus also identified a limited number of errors in their first submission. Event though both companies submitted corrected data sets it seems reasonable to use a regression technique that can eliminate the influence of some remaining errors.

254. As mentioned above the robust regression technique used by the Commission (and CRA) has the property of giving a different weight to each observation according to the behaviour of such observation in comparison with the others. Such a technique prevents the results from being driven by a few deviant observations in the dependent variable. In any event, as RBB admits "robust regression is preferable when outliers are present and almost as efficient as OLS when no outliers are present".

7.3.1.2.2. The use of time fixed-effects for each month in the sample

255. Ryanair claims that there are various ways to model time fixed-effects. The Commission reports results whereby each month is allowed its own fixed effect. However RBB proposes using year and month fixed-effects or year and quarters fixed-effects. RBB argues that "a priori there are no reasons to prefer one choice over another", however, using RBB's proposed way to model fixed-effects changes the results of the Commission's model.

256. RBB reports that using instead a set of monthly and yearly dummies (73) one obtains a lower coefficient. With reference to the basic presence specification (74), obtained using robust regression, the coefficient attached the dummy for Ryanair's presence in the route falls from -.077 to -.054, both significant at the 1% level. Looking instead at the basic frequency specification (75), the coefficient of interest goes from -.028 to .0199, with an associated standard error of .0157. RBB conclusion from these results is that the Commission model is highly sensitive to the choice of time fixed effect.

257. The Commission disagrees with the view that a priori there is no correct method to model fixed-effects. Indeed, precisely because using one or an alternative method may lead to slightly different results it is important to choose the right method to model fixedeffects and to motivate it properly.

258. The Commission first notes that the plot of the time fixed-effects (see graphs 1 and 3 in section 6.3) clearly suggest that the month coefficients adequately control for seasonality.

259. Second, usually time period fixed-effects involve one dummy variable for each time period (that is, every month of every year), not for each month (i.e. 11 dummies for the 12 months). The former correct for aggregate shocks that are common across markets, whereas the latter correct for seasonality. If a seasonality correction is desired, it would be preferable to have three sets of monthly variables, one for 'summer' markets, one for 'winter' markets, and one for markets that do not have a marked seasonal pattern. In contrast, if common shocks are to be removed, then one set of dummies (e.g., one for each month over five years) should be used. This is exactly the approach used by the Commission.

260. In all the Commission regressions, a time fixed effect has been introduced into the specification, which means a dummy for each period present in the sample, i.e. 59 dummy variables for a period of five years with monthly observations. The reason for such an econometric strategy is to control for common shocks that are specific to a certain period and common across markets, either city pairs or airport pairs. At the same time, seasonality is taken into account under the assumption that each year could present a slightly different seasonal pattern to the year before.

261. The Commission remarks that the chosen way to model time fixed effects has a wider scope than just to control for seasonal pattern. Using a dummy for each month we can control for exogenous shocks in a way the seasonal pattern proposed by RBB does not allow to. As an example we can take the extreme case of 9/11 (76) and the effect of that event on airline industry as a whole. The Commission's model properly controls for such an extreme event, introducing a dummy for each of the month after that event. In the RBB framework, a monthly dummy will average the effect of 9/11 for each calendar month over all years in the sample. Yearly dummies would even spread the residual effect over the whole 2001. This is clearly inappropriate.

262. The Commission's follow-up analysis indicates that lower coefficients using RBB proposed way to model time fixed-effects is basically due to the introduction of yearly dummies. However from a methodological standpoint, we do not see any clear reason to use such dummies, since it is difficult to think of shocks that are common to a whole calendar year. 263. RBB also argues in their Dataroom Submission, a time fixed effect could account for common seasonality across routes. Given the wide range of destinations covered by the sample, it could be the case that routes are differently affected by seasonality. Fares of Dublin-Salzburg, a typical winter destination, will be affected by seasonality in a different way than, for example, Dublin-Alicante. For this reason, the reported coefficient of the month fixed effect will be a compromise between these different types of routes. The Commission believes that it is unlikely this could lead to an upward bias in the coefficient of interest. This could only happen if the presence (or frequencies) of Ryanair is spuriously associated with uncontrolled route-specific seasonal fluctuations in fares. The Commission can think of no reason why this could be the case.

264. In any event, and in order to control for such an eventuality, we have added a set of additional controls in our basic presence and frequency specification. First, we have divided all the routes on the basis of their nature of summer destinations or winter destinations. This has been done comparing the average winter and summer fares. We have then considered as summer destinations those for which average summer fare is at least twice the one charged in winter (77). Conversely, a winter destination is one with an average winter fare at least twice the summer one. Among the residual routes, we have isolated routes towards big hubs, which could reasonably follow a different seasonal pattern than routes to smaller destination cities (78). Finally, we have interacted the dummies for the four groups obtained with a set of monthly dummies.

265. This procedure has been alternatively replicated through the interaction with seasonal dummies. The following table reports the results obtained for the relevant variables, namely presence of Ryanair and log of the frequencies of Ryanair, in the additional regressions run in comparison with the basic ones presented in the first column of Table 9 and Table 14.

<emplacement tableau>

266. The table provide a synoptic view of the estimates for different models used for the seasonality. Comparing the first column, the original result presented in the SO, with the second and third column, we can easily see our results for the presence specification is robust to the introduction of monthly and seasonal dummies with the type of destination dummies described above. After the removal of the time fixed effect, the coefficient reported is slightly lower. This effect, as explained above, is due to the introduction of yearly dummies, which cannot be methodologically justified.

267. The robustness of the introduction of the new variables is confirmed by the frequencies specification. In this case, the effect observed slightly decreases, but still the results are significant at the 1% level. Again the unjustified introduction of the yearly dummies leads to a slightly lower coefficient and the loss of statistical significance. This is clearly due, as explained above, by the yearly dummies spuriously capturing the overtime increasing scope for overlapping between Ryanair and Aer Lingus. .

7.3.1.2.3. Serial correlation

268. RBB argues that robust-regression should also not be used because it does not allow correcting for the presence of serial correlation. As explained in annex III, serial correlation arises when the prices charged in the past could affect the level of prices today. In other words "serial correlation is present if, over time, a positive error in one month is more likely to be followed by a positive error in the following month rather than a negative".

269. RBB argues that air fares data suffers from positive serial correlation "as the Commission acknowledges at paragraph 30 in Annex 3 of the SO, based on the Durbin- Watson test". However as RBB admits "serial correlation does generally not affect the consistency or unbiasedness of the estimation results" (provided that there is no lagged dependent variable included on the right hand side, as is the case in the Commission's regressions). However, RBB argues, the presence of serial correlation might lead to "systematically lower standard errors". This, in turn, may lead to finding that a coefficient is statistically significant when it is not.

270. To deal with the problem of serial correlation RBB proposes to compare four different estimation procedures: (i) robust-regression (used by the Commission and implemented via rreg in STATA), (ii) OLS with clustered standard errors, (iii) OLS with clustered standard errors (by route) excluding observations that are dropped using robust regression and (iv) median regression (quantile regression implemented via qreg in STATA) (79). RBB produces a table reporting the coefficient estimates and standard errors for the baseline presence and frequency specification obtained by each of these four estimation procedures.

271. The main result, according to RBB, is that rreg and qreg standard errors are substantially lower than the OLS standard errors which control for serial correlation using STATA's cluster option. "in detail, the correct OLS standard errors are twice as large compared to the RREG standard errors. This has the effect that, even under the same modelling assumption that the Commission reports, the significance level of the Ryanair presence coefficient decreases dramatically".

272. The Commission believes that this criticism is inconsequential. First as RBB itself admits "the Ryanair presence effect lies between 7.7% and 6.8% using four different estimation techniques". This clearly implies that irrespective of which estimation procedure the competitive constraint that Ryanair exerts on Aer Lingus is economically significant.

273. Second, with respect to statistical significance, the Commission acknowledges that the use of OLS as proposed by RBB leads to standard errors that are twice as large as those obtained with robust-regression. However, the coefficients for Ryanair presence remain statistically significant at least at the 5% level for all methods proposed by RBB in the case of the presence specification (that is, OLS cluster OLS cluster (restricted) and median regression) (80). In the case of the frequency specification the coefficients remain statistically significant at the 1% level with median regression, and at the 10% level with OLS cluster (restricted) (81). Standard OLS (cluster) leads to a p-value=0.168 in the (baseline) frequency regression. However, this is the only estimator that does not control for outliers (82).

274. Fourth, as acknowledged by RBB, the presence of serial correlation in the errors does not lead to biased or inconsistent results for the Commission econometric specification. The fact that the average fare charged by Aer Lingus on a certain route at a given point in time could to some extent depend on past values of the fare for that route has no effect on the coefficient for variables assessing competition, given the setting of the Commission model.

275. Finally, and quite independently of the above arguments, the Commission objects to the use of the standard OLS with the cluster option proposed by RBB. This technique, which is used in cross-section and panel analysis in presence of observations that can be grouped according to specific characteristic (e. g. members of the same family in crosssection household surveys), allows the error terms to be correlated within each group (cluster) but not between different groups. The cluster option, however, does not in fact address the type of correlation by which the fares of some of the routes are affected, that is serial correlation. The cluster option, indeed, does not pose any restriction on the nature of correlation present within each cluster and consider all the clustered observations correlated with the same magnitude. In our case, this would mean assuming that fares set in May 2002 could affect the fare set June 2006 in the same way as the one of May 2006. For this reason, it is well acknowledged in the literature that the cluster option is a suitable instrument in presence of a large number of clusters and a relatively small number of within cluster observations83. However, this is not the case here. The Commission data set includes 55 city-pairs and an average of 35.6 observations for each city pair. That is, we have a large number of clusters together with a relatively large number of clustered observations. This alone makes the use of the cluster option unsuitable in this case.

276. However, the Commission acknowledges that it is necessary to determine whether serial correlation might be affecting the statistical significance tests in the Commission fixed-effects regressions. Hence, we have performed a detailed analysis of the behaviour of the error term. The diagnostics have been run for single route price series and for the aggregate panel data set and have tested the standard assumptions of no serial correlation and no heteroskedasticity (see also Annex IV.1).

277. Concerning serial correlation, the problem directly addressed by RBB, the Wooldridge test has been performed both on the presence and on the frequency specification. The test has rejected the null hypothesis of absence of serial first order correlation for all the specifications tested. The rejection of this assumption, it is worth stressing out again, does not lead to any bias in the coefficients reported by the Commission - given that no lagged variable in included among the regressors (84).

278. To test the constant variance assumption (i.e. no heteroskedasticity) the Commission has performed the standard Breusch-Pagan test for the presence of heteroskedasticity in the error terms and a more sophisticated likelihood test based on Generalized Feasible Least Square estimation. Both the tests have identified the presence of heteroskedasticity in both the presence and frequencies specification.

279. To correct for the presence of serial correlation and the heteroskedasticity the most suitable approach is use the estimation method proposed by Newey/West (implemented via the newey2 command in STATA). This correction allows the error term to be heteroskedastic and/or correlated within each route, with a coefficient of correlation which decreases the farer in time are the observations. The only limiting assumption imposed by this method is that serial correlation decays relatively fast overtime within each route. This hypothesis is reasonable in the present case.

280. The tables below present the results for the Newey correction model applied to the main presence and frequencies specifications. For each of them, an additional specification including monthly dummies interacted with the type of destination has also been added for a robustness check. The results reported are fully consistent with those reported above in section 7.2. based on the robust regression (implemented via rreg in STATA).

<emplacement tableau>

7.3.1.2.4. Risk of omitted variable bias

281. RBB exposition of their concern related to the inclusion of the Aer Lingus capacity variable (ln_seats_aerlingus) in some of the Commission's presence regressions is somewhat confusing. RBB argues that accepting that the capacity variable is an adequate control for differences in demand across routes and over time is "tantamount to admitting that there is an omitted variable bias". In fact the opposite is true. If "Aer Lingus capacity" is an adequate proxy for demand its inclusion precisely is justified to eliminate the omitted variable bias if no control for demand is included. The use of proxy variables is a standard and simple solution to avoid the risk of omitted variable bias. As explained by Wooldridge (page 295) "loosely speaking, a proxy variable is something that is related to the unobserved variable that we would like to control for in our analysis". Under reasonable assumptions, including the proxy variable in the regression eliminates or at least reduces bias.

282. That said it seems RBB is trying to make two distinct, albeit related arguments. First RBB argues that the inclusion of Aer Lingus' capacity (i.e. frequency) as an explanatory variable of Aer Lingus fares in some regressions will lead to endogeneity bias. This is because RBB argues that capacity is simultaneously (i.e. endogenously) determined with prices. This is exactly the same argument raised against CRA "presence" regressions and already addressed in section 7.1.4, par. 177. Second, given that "capacity" is not a good proxy for demand (because it is allegedly endogenous) the Commission's regression actually fails to include an adequate demand control. Hence the results are likely to suffer from omitted variable bias.

283. Notably RBB argues that monthly frequency on each route varies substantially consistent with the flexibility to adjust frequency easily. The Commission agrees that the parties have flexibility to change frequencies. However flexibility in shifting frequencies does not mean that frequencies are set simultaneously with prices. In general the levels of frequencies depend on expected levels of demand. Hence fluctuations in frequencies across routes and season simply reflect fluctuations in expected demand, for example due to seasonality or anticipated one time events. It is precisely this ability to determine the optimal level of frequencies a few months in advance in light of expected demand that makes the former variable a good proxy for the latter.

284. The second argument is straightforward to address. First note that the Commission's baseline specification includes Aer Lingus capacity as a control or scale factor. However further specifications drop this variable and include an alternative demand control, namely the log of total frequencies offered by all carriers at the destination airport (ln_dest_freq_total) (see for example table 10). As explained above this variable controls for the traffic at the destination airport and is therefore a good proxy for variations in demand (anticipated at least at the time that carriers set their frequencies). This variable turns out to be highly significant in all specifications. More importantly the coefficients of Ryanair's presence are robust both statistically and economically to the use of this alternative demand control.

7.3.1.2.5. Economic significance of impact of Ryanair's frequencies on Aer Lingus' price

285. As shown in section 7.2.2.2 of this Annex, if one measures the strength of Ryanair's presence, it appears that Ryanair imposes a more significant constraint on Aer Lingus when it serves the same airport. Ryanair argues repeatedly in its reply to the Statement of Objections (paragraphs 26, 95 and 113) that the Commission made a significant error in the interpretation of its own econometric results regarding the effect that Ryanair has on Aer Lingus prices (such that the effect was exaggerated by a factor of 100). As a consequence, Ryanair suggests that the Commission's conclusions rely on such mistaken interpretations and thus all the Commission's econometric results should thus be dismissed.

286. Indeed, there was a mistake in § 64 of Annex 4 in the Statement of Objections. The Commission acknowledged and corrected this mistake when it provided the merging parties a version of the annex that revealed the actual coefficients of the Commission's regression analysis. This was only possible when both parties confirmed on writing that they agreed to allow each other to see these results. However, this mistake appears in a paragraph which does not comment on the results. This paragraph describes the frequency specification and provides an illustration of how the coefficient should be interpreted in general. That is, the misleading Statement concerns a purely hypothetical situation and not the interpretation of any of the Commission's results. As a result it is incorrect to argue that the Commission's Statement of Objections was misled by the wrong interpretation of its econometric results.

287. RBB argues that "in the frequency regressions, even taking the Commission's results at face value, the impact of Ryanair on Aer Lingus is very weak - Ryanair would have to triple frequency to induce Aer Lingus to lower fares by 5%". The Commission notes that this way of deriving price effects from the reported coefficients is misleading.

288. The proper way to derive price effects from the coefficients in the frequency specification is described in the academic papers published by the expert economists involved in the FTC vs. Staples/Office Depot merger (87). As mentioned above the econometric approach used by the Commission, is similar to that proposed by the merging parties in that merger, and later also adopted by the FTC. In particular the "frequency specification" is a direct adaptation from the specification used in that case.

289. Here we apply the same methodology used in the Staples/Office Depot case. In that case both the FTC's and the merging parties' economic expert calculated the price effects under the assumption that all Office Depot stores that overlapped with Staples would be closed. In this case the analogous assumption is that Ryanair would withdraw its presence from routes where it competes with Aer Lingus, allowing Aer Lingus to charge higher prices. Under this assumption, the estimated price effect of the merger can be derived as follows. Given the original model:

(formule) (1) (88)

first compute the in-sample prediction of Aer Lingus fares,

(formule) (2)

290. Next compute for each observation the alternative prediction modifying our original vector of regressors xit under the assumption stated above that Ryanair withdraws from overlap routes, allowing Aer Lingus to increase prices. In this way we are able to compute the expected price increase would Ryanair disappear from all routes operated by Aer Lingus. Given i=1,...,N as the total number of routes operated by Aer Lingus, and t=1,..,T the period for which we observe the fare charged by Aer Lingus, we obtain the average percentage price increase as:

(formule) (3)

291. Equation (3) represents the price increase faced by the average Aer Lingus customer. The intuition for this procedure is as follows. First we take the price predicted by the regression pˆ (for each time and route). We then assume that Ryanair withdraws. This corresponds to setting the absence dummy to 1 for all routes and time periods and setting Ryanair frequencies equal to zero. We obtain in this way the alternative set of regressors denoted as x ~ .We then take the data set where Ryanair is always assumed absent and compute the predicted price effect using the estimated coefficients (p ~). We then take the difference ( pˆ -?~p ) and divide by pˆ to derive the percentage price effect for each route and period. Then we average this percentage price effect across all routes and time periods.

292. An alternative assumption also discussed in the Staples/Office Depot case is to consider only the price increase on the formerly overlapping routes only, assessing in this way the effect of the merger in the restricted sample of the routes affected by a decrease in competition compared to the pre-merger situation. To see this, we define: (formule)

293. and then compute the average price only in the routes and dates affected by the merger:

(formule) (4)

294. The formula above generates the average price effect considering the impact of the merger relative to the average price levels over all the periods in the sample. However it could be argued that to compute the price effect one must only consider how the merger eliminates competition from the routes where the merging parties currently overlap. To do this we can compute the price effect only taking into consideration the last month (or months) for which data is available. Again there are two possibilities: to consider the average price effect on overlap routes or the average price effect over all routes where Aer Lingus is present.

295. Table 16 below is based on the frequency specification reported in Table 14. The table reports the average estimated increase in Aer Lingus fares if Ryanair withdraws from all the routes and for all the months in which both Ryanair and Aer Lingus were flying. This increase in computed for the latest three month of the sample, namely September, October and November 2006, and over the complete sample used for the regression which starts as of January 2002.

TABLE 16 PRICE EFFECTS OF THE MERGER FOR THE FREQUENCY SPECIFICATION UNDER ALTERNATIVE ASSUMPTIONS

<emplacement tableau>

296. As expected, the assumptions of the two methods lead to a significantly different estimated price increase. Considering the first method, the average price increase ranges across specifications between 5.7% and 6.5% for September, 5.9% and 6.5% for October and 6.2% and 7.1% for November 2006. The average effect for all the months included in the estimation sample is between 4.7 and 5.1%. This result is consistent across specifications and in line with the results on the presence specification.

297. Method II restricts the averaging sample to the overlapping routes. The average price variation increases significantly on the basis of the share of overlapping routes over the total number of routes operated by Aer Lingus in each and every time period considered (89).

298. The price effect is consistent across specifications and ranges between 10.2% and 12.6% considering single months, while is between 12.2% and 13.1% for the whole sample of overlapping routes. Comparing the results obtained with the two methods, it is interesting to notice that Method II reports a higher price increase in the whole sample rather than the most recent months in time. The opposite happens under method one. This would suggest an average higher Method II price effect in months far back in time to which Method I gives a lower weight because of the reduced scope for overlapping in the previous periods compared to more recent ones.

7.3.1.3. Conclusion

299. The Commission has carefully considered the comments and criticisms made by both CRA and RBB with regard to the Commission's fixed effects regressions. A priori, some of the criticisms were reasonable and relevant. A more in depth analysis however, confirms and thereby strengthens the Commission results. In particular, the finding that the presence of Ryanair has a statistically and economically significant effect on Aer Lingus' prices is extraordinarily robust.

7.3.2. Regressions with Ryanair's monthly average net fares as the dependent variable

7.3.2.1. RBB and CRA comments

300. As explained in paragraph 230 above the Commission agrees with RBB that there is not sufficient variation, within a reasonable time period, in the presence of Aer Lingus in routes previously operated by Ryanair, As a result the fixed-effects regression is unlikely to provide any reliable estimates of the possible impact of Aer Lingus' presence on Ryanair prices.

301. CRA, however, argues the Commission should be placing a greater amount of weight on the results reported in table 15 above which use a sample going back to 1997. CRA accepts that there are very few instances where Aer Lingus has entered (or exited) a route previously served by Ryanair. However they argue that there are a number of routes that Aer Lingus flies on seasonal basis, while Ryanair is present on these routes throughout the year (e.g. Marseille). According to CRA it is possible that Ryanair could price differently whenever Aer Lingus is present on that route, as opposed to when it is not (although of course one needs to take into account that its pricing will also be affected by a lower seasonal demand). These seasonal switches are still very valuable in identifying a potential effect of Aer Lingus on Ryanair and thus the Commission should be placing more weight on the results of these regressions than it does at this point. In conclusion, CRA notes that the "presence" regressions reported in table 15 show some evidence that Ryanair's fares are systematically lower by 9% to 11% whenever Aer Lingus also operates on the same route.

302. However, RBB notes that the Commission's fixed-effects regression for Ryanair fares use a sample that goes back to 1997. Only then is there sufficient variation in the presence of Aer Lingus to obtain meaningful estimates. But RBB remarks that over such a long time period it is not reasonable to presume that unobserved effects are fixed. For example, RBB notes, "the events of 9/11/01 had a fundamental impact on air travel and, following these events, we would expect unobservable behaviour to change"

303. The Commission agrees with RBBs line reasoning and rejects CRA arguments. As explained in detail paragraph 37 in the Statement of Objections (paragraph 51 in this document) it does not appear justified to extend the sample more than 5 years, in particular since prior to 2001 Aer Lingus operated a very different business model before its restructuring started in 2002 (90).

304. RBB criticizes the Commission for reporting these results because they claim that "they are at best meaningless and at worst misleading". The Commission, however, reiterates that these results were reported in Annex 4 of the SO in the interest of full transparency and fully acknowledging that it did not attach them any weight and derive no conclusions in the competitive assessment from this regressions. Had the Commission failed to report these regressions the merging parties, in particular Aer Lingus, would have likely charged the Commission with hiding potentially relevant results.

7.3.2.2. Aer Lingus likely constrains Ryanair also on variables other than price

305. CRA accepts that "in practice the competitive constraint flowing from Aer Lingus to Ryanair is difficult to identify precisely because of limitations in the available data. [...]* given that Ryanair has entered several routes where Aer Lingus was already active over the past few years, the competitive constraint exercised by Ryanair on Aer Lingus can be measured quite accurately. The constraint operating in the other direction (from Aer Lingus to Ryanair) however cannot be measured with the same degree of precision given that there are only very few episodes of Aer Lingus entry/exit on routes where Ryanair is already present".

306. However, CRA correctly points out that this does not imply that Aer Lingus is not acting as a significant constraint on Ryanair on routes where both carriers compete. CRA argues that as a matter of economic theory a finding that Ryanair significantly constrains Aer Lingus implies (by definition) that a sufficient number of customers are prepared to switch from Aer Lingus to Ryanair in response to a more competitive price-quality offering. Switching by this (marginal) group of customers disciplines Aer Lingus fares, inducing it to charge lower fares (as demonstrated by the econometric results). At the same time, CRA argues, the very presence of these "marginal customers" creates incentives for Ryanair to set lower fares, as it can benefit from capturing them.

307. The Commission agrees with this line of reasoning but only up to a point. As acknowledged in the Statement of Objections (see footnote 350) if one firm constrains the other, the reverse is also true at least for some range of prices. However it is not necessary that this constrain operates only (or largely) at the level of prices. The entry of Aer Lingus on a route previously dominated by Ryanair may prompt Ryanair to respond not by lowering prices but by increasing frequencies or by lowering prices for ancillary services. In that case a price regression would not reveal anything meaningful with respect to the competitive constraint that Aer Lingus exercises on Ryanair, even if there were sufficient instances of Aer Lingus entry/exit into Ryanair routes (which is anyway not the case).

308. As explained in the body of the decision Ryanair and Aer Lingus are close competitors. They compete largely for the same pool of customers and competition takes place via a differentiated product offering in which a lower price reflects a lower quality product and a higher price reflects additional features. Price is only one parameter in the analysis. In the presence of Aer Lingus, Ryanair may not lower its prices but may be forced to increase the quality of its service along some of these dimensions. Most importantly Ryanair makes a significant amount of profit from the sale of ancillary services. Ryanair may therefore not be obliged to reduce air fares as much as it would need to reduce the price of these ancillary services or enhance the terms and conditions.

309. In conclusion, the results of the regression analysis do not put into question that the merger would eliminate the competition in quality and in offering of ancillary services that currently exists between Aer Lingus and Ryanair, which is also reflected in the numerous promotions and advertisements of both parties that make reference to each other - see in particular section 7.4.2.

8. Conclusion

310. The Commission's price regression analysis tests the following hypothesis: ???Whether the presence of one of the Merging Parties in the route is associated with a statistically and economically significant reduction in the fares of the other; ???Whether the Merging Parties exert on each other a stronger competitive constraint than any other existing competitor; ???Whether the existence of an actual or potential competitor with a significant presence at the destination airport on a route originating in Dublin has an impact on the Merging Parties' prices; ???Whether a stronger presence of one of the Merging Parties (in terms of number of frequencies) has a more pronounced effect on the other's fares.

311. Cross-section regression analysis, including the two-step approach proposed by RBB, appears inadequate to assess the constraint that the merging parties exercise on each other's prices. In contrast standard fixed-effects regressions allow to properly isolating the constraint that Ryanair's presence or number of frequencies imposes on Aer Lingus prices in routes out of Dublin. The available data does not allow to test the effect that Aer Lingus presence might have on Ryanair's prices.

312. The Commission's fixed-effects regressions validates the following hypothesis set out ex-ante:

* First, depending on the specification, the Ryanair's presence is associated with Aer Lingus charging around 7-8% lower prices when considering city-pairs reflecting the Commission's retained market definition and around 5% lower prices when considering airport-pairs (hypothesis I-A). This effect is economically and statistically significant in all tested regressions. This result is also robust, correcting for the presence of outliers, heteroskedasticity and serial correlation. It is also highly robust to the use of alternative specifications including alternative demand and supply controls. Notably, in practically all cases the control variables in the different regressions have the expected signs and are statistically significant. The explanatory power of the regression is also high with R2 consistently above 80%.

* Second, comparing the coefficients of Ryanair with that of flag-carriers and non-flag carriers, as well as Aer Arann and CityJet, Ryanair's presence or number of frequencies have a much stronger economic impact (at least double) than that of any other type of carrier. In fact, in most cases the regressions indicate that the presence of other carriers has no economic or statistically significant effect on Aer Lingus fares. These results validate hypothesis II-A.

* Third, destination-based flag carriers exert only a very limited constraint on Aer Lingus. Destination-based non-flag carriers exert a higher constraint than flag based carriers. However, their constraint is around half or less than the constraint exerted by Ryanair on Aer Lingus retaining the Commission's market definition. Moreover flag carriers, for instance are only present on 8 of the 37 overlap routes upon which Aer Lingus and Ryanair currently compete, and tend to be much smaller than either Ryanair or Aer Lingus where they are present (especially for point-to-point passengers). Thus, contrary to Ryanair's claim, it cannot be expected that the merged entity would be effectively constrained by flag or other non-flag carriers post-merger. These results validate hypothesis III.

* Fourth, measuring the strength of Ryanair's presence using number of frequencies in the route as a proxy provides further confirmation that Ryanair constrains Aer Lingus. It is possible to examine the price change in overlap market only or across all markets under various assumptions. For example one can focus on the price effect on the last month for which data is available or the price effect on average over the full sample period. Depending on the specification the price effect of the merger implied by the Commission's frequency regressions is around 5-6% (on average over all routes) or 10-12% (if only overlap routes are considered). This adds to the robustness of the results derived from the presence specifications. This validates hypothesis IV. It is also worth noting that, as expected, Ryanair appears to impose a more significant constraint on Aer Lingus when it serves the same airport.

313. The Commission's regression analysis confirms and complements conclusions set out above on the basis of qualitative evidence that Ryanair and Aer Lingus are close competitors. The results indicate that Aer Lingus prices are currently constrained by competition from Ryanair.

314. Furthermore, post-merger as predicted by standard ''non-coordinated effects" analysis (see paragraph 24 in the Horizontal Merger Guidelines), both carriers would internalise the effects of setting higher fares on each other. In particular, the merged entity would have the incentive to set higher fares for Aer Lingus as most of the customers lost would be captured by Ryanair.

315. The loss of post-merger competition on Aer Lingus is in itself a major cause for concern from the transaction, as also recognized by Ryanair Indeed, Ryanair concedes that "the main issue in the merger assessment is whether the merger reduces constraints on Aer Lingus post merger" (91).

316. The fixed-effects regressions with Ryanair's prices as the dependent variable do not allow reaching conclusions with respect to the impact of Aer Lingus on Ryanair prices. This is because there are insufficient instances of Aer Lingus exiting or entering into a route where Ryanair was previously present. In other words there is little variation in the presence of Aer Lingus on Ryanair routes. It should be emphasised, however, this neither validates nor refutes the hypothesis that Aer Lingus exerts a competitive constraint on Ryanair's prices. Indeed the evidence presented in the previous section makes it clear that Ryanair (as well as Aer Lingus) permanently monitor each other's prices. If Aer Lingus were to lower its prices this would tend to attract customers that would otherwise had flown Ryanair. If a sufficient number of customers switch to Aer Lingus (switching costs are very low and in most overlap routes there are no closer alternatives) this would lead to lower load factors. According to its pricing policy Ryanair would react by reducing its prices in order to meet its load factor targets. If Ryanair were not constrained by Aer Lingus there would be little logic in permanently monitoring their prices.

317. Moreover, both economic theory and qualitative evidence suggest that Ryanair might be constrained on parameters of competition other than price. As a result of the merger the ability to increase Aer Lingus fares would further relax the competitive constraint faced by Ryanair, allowing it to increase fares of ancillary services or to reduce quality without risking losing as many customers as in the absence of the merger.

318. Finally, it should be noted that the estimated effect of Ryanair on Aer Lingus prices is likely to be underestimated. As pointed out in Section 7.7 of this Annex, the presence of Ryanair in Dublin exerts a potential competitive constraint on Aer Lingus. On routes out of Dublin where it is the only carrier, it can be expected that Aer Lingus sets prices which are lower than what it would charge if Ryanair had no Dublin base. Since the regression analysis considers only fares' overtime variations within each route and only captures price reductions subsequent to Ryanair's entry, this potential competition constraint does not show up in the empirical results.

Notes

1 See EU Commission "Guidelines on the assessment of horizontal mergers" (2004-C 31-03), paragraph 24.

2 The alternative might be to use anecdotal evidence, which may be less reliable in some cases.

3 The standard error is the estimated standard deviation of a statistic. The formula depends on the statistic in question. For example, the standard error of a sample mean is just the sample standard deviation divided by the square root of the sample size. In multiple regression analysis, each estimated coefficient has an associated standard error. The estimate divided by the standard error gives you a way of testing whether the coefficient of interest equals zero (the null hypothesis). This is known as a t-test because this ratio is compared to a t-distribution to obtain the p-value.

4 In competition cases, it is often not possible to increase the size of the sample given time and resource constraints.

5 The term "preferred" in this context refers to the regression technique or specification which, on the basis of objective criteria can be considered superior to alternatives - for example because the necessary assumptions are more likely to be met.

6 In most cases also at the 1% level.

7 See for example, Ryanair's memo of 31.01.07 on "Yield Management".

8 Statistical significance is a formal way of assessing whether observed associations are likely to be explained by chance alone.

9 The threat of entry (i.e. potential competition) can also exert a competitive constrain on the incumbent which cannot be directly observed or quantified.

10 In theory, it can be expected that if one firm constraints the other the reverse is also true at least for some range of prices. This is because customers respond by switching to a change in relatively prices. When the direction of change of the relative prices is reversed the flow of customers will also be reversed.

11 Most Aer Lingus flights are point-to-point or single segment flying from a single origin to a single destination (for example, Dublin to Frankfurt). However, for periods between January 2002 and December 2003, Aer Lingus operated multi-segment flights on a small number of routes that included Dublin. A multi segment flight is one where the aircraft takes off from an origin airport, lands at an intermediate destination where passengers both embark and disembark before taking off again to land at the final destination (for example Cork, Dublin, Frankfurt).

12 Routes out of Cork or Shannon were excluded from the data set in order not to introduce in the regression analysis additional sources of route heterogeneity, which could lead to bias - see the next section for more details.

13 Most Aer Lingus flights are point-to-point or single segment flying from a single origin to a single destination (for example, Dublin to Frankfurt). However, for periods between January 2002 and December 2003, Aer Lingus operated multi-segment flights on a small number of routes that included Dublin. A multi segment flight is one where the aircraft takes off from an origin airport, lands at an intermediate destination where passengers both embark and disembark before taking off again to land at the final destination (for example Cork, Dublin, Frankfurt). It is not clear what the relevant seat capacity in these few instances is but this has no impact on the regression analysis and the number of affected flights is very small. In any event Aer Lingus has provided its fare and passenger volumes on a point-to-point basis.

14 These data has been aggregated according to the market definition in order to match with the rest on the data used in the analysis based at the city pair level.

15 This problem has also been encountered by Ryanair's expert economists as further explained below.

16 RBB "Bargepole: Ryanair cross-sectional regression results", 10 December 2006.

17 Heteroskedasticity occurs when the variance of the error terms differ across observations. It is a common violation of the standard assumptions in cross-section regressions. The most common response to the (potential) presence of heteroskedasticity of an unknown form is to use a heteroskedastically-robust estimator for the covariance matrix of the regression coefficients. For more immediate reference see also Annex IV.1 below.

18 "Shoulder" months are less influenced by seasonal factors.

19 The estimation technique uses robust standard errors.

20 Another problem is that RBB has not provided any explanation as regards the selection of routes to include in the regression. This is problematic insofar as the results may be sensitive to cherry-picking of routes.

21 In fixed effects regressions this matters less given that (i) the higher number of observations increases the precision of the estimates and (ii) the fixed effects compare prices over time, within the route, before entry and after entry.

22 The market power dummy takes a value of 1 for Frankfurt Hahn, Girona (Barcelona), Charleroi, Glasgow Prestwick and Stockholm Skavsta and 0 otherwise.

23 We are considering here as "flag carriers" all national airlines which were (or still are) state-owned and considered as the countries' only or leading airline. The carriers included in this group usually operate according to a business model, addressed as network model, which is based on the interaction between a few main airports, or hubs, and a larger set of secondary ones. According to this model, mid and long haul routes are usually covered by multi-segment flights (see footnote 11).

24 The variable "Non-flag carriers" aggregates all other competitors present in the routes considered that are not flag carriers (charters excluded).

25 This interpretation is accurate when the coefficient on a dummy variable suggests a small proportionate change in the level of the dependent variable. This is the case in the regressions reported in this annex. When the coefficients are large, say above 0.25, the exact percentage difference can be obtained using a correction. Namely, if b is the estimated coefficient on a dummy variable and V(b) is the estimated variance of b then: g = 100 (exp(b - V(b)/2) - 1) gives an estimate of the percentage impact of the dummy variable on the variable being explained. See Van Garderen, Kees Jan and Shah, Chandra, "Exact Interpretation of Dummy Variables in Semilogarithmic Equations". The Econometrics Journal, Vol. 5, pp. 149-159, 2002.

26 For the purposes of this regression analysis the following airlines were considered as "flag" carriers:, Aeroflot, Air Canada, Air France (and separately Cityjet), Alitalia, Austrian Airline, British Airways, Continental Air, Croatia Airlines, Delta Airlines, Czech Airlines, Finnair, Iberia, Lot, Lufthansa, Luxair, Malev Hungarian Airlines, Scandinavian Airline, Swiss Air Lines, Tap Air Portugal.

27 Another alternative to resolve the technical difficulty that the log of 0 (frequencies) does not mathematically exist is to take the log (number of frequencies + 1). This approach raises difficulties with respect to the direct interpretation of the coefficients of interest. Concerning the cross-section regressions, the introduction of this modification confirms the presented results in terms of signs and levels of significance; as expected, given the small sample, the coefficients for the frequencies variables result modified, either upward or downward for each of the regressions. As already stated, the Commission does not consider cross-section regression as a suitable method to assess the level of the impact. In the case of fixed-effect regressions, he Commission has replicated several of the reported "frequency regressions" replacing the absence dummies and log of frequencies. This approach essentially leads to the same results as those obtained with the specification reported above.

28 See Orley Ashenfelter & David Ashmore & Jonathan Baker & Suzanne Gleason & Daniel Hosken, 2006. "Empirical Methods in Merger Analysis: Econometric Analysis of Pricing in FTC v. Staples," International Journal of the Economics of Business, Taylor and Francis Journals, vol. 13(2), pages 265-279, July.

29 Switching from city pairs to airport pairs, the number of overlapping routes goes from 24 to 13 for May, from 26 to 14 in October.

30 The Commission notes that adding the flag_based and non_flag based variables leads to statistically insignificant coefficients. However, note that both are also statistically insignificant. Potentially it can be argued that the true specification excludes these variables. Leaving statistically insignificant variables out of the regression equation increases the accuracy of the test of statistical significance.

31 The number of observations in the frequencies regressions may differ from the ones which uses presence. In order to control for the presence of outliers, we have used a robust OLS estimation which assigns a weight to each observation with higher weights given to better behaved observations. In extremely deviant cases, the weight is set to missing and the observations are not included in the analysis. Concerning the different number of observations in April and October, Aer Lingus flew in October two additional routes than in May: Lanzarote and Newcastle. Ryanair started operating Dublin-Berlin in June 2006.

32 Note that the regression keeps Aer Lingus' own frequencies constant.

33 See RBB's "Dataroom Submission" 13 April 2007.

34 In the case of Ryanair, correlation is equal to -0.45 for the month of May and -0.43 for October, both significant at the 1% level. They reach -0.58 and -.55 respectively, if we restrict the computation to the overlapping routes with Aer Lingus, which are ones actually included in the Aer Lingus frequency regression sample.

35 Wooldridge, J. (2003), Introductory Econometrics: A modern approach, 2nd edition, Thomson Publisher

36 See also Annex IV.1.

37 See CRA's "data room" report: "Technical Annex: CRA's review of the Econometric results available in the Data Room".

38 The constant term (ai ) should be omitted when estimating fixed effects. However, it is reproduced here as it appears in RBB's submission.

39 Distance is fixed over time, and hence cannot be included in the first stage.

40 See RBB's "Dataroom Submission" 13 April 2007, section A2.2

41 Annex 1 in its "Position Paper on Barriers to Entry" dated 20th February 2006.

42 RBB assumes that if there was no base at the other end of the route in December 2006 then no base existed in the whole period January 2005 to December 2006.

43 It is also impossible to determine the direction of this bias given that so many such factors can affect both Ryanair's prices and Aer Lingus presence in the route, such as the presence of other competitors, the type of destination, the popularity of the route according to purpose of travel, customer awareness, destination airport characteristics, number of alternative airports at destination, safety considerations, duration of travel, population at destination etc. RBB only controls for distance in the second stage regression.

44 See RBB's "Dataroom Submission" 13 April 2007, section A2.5.

45 See Jonathan B. Baker, "Econometric Analysis in FTC v. Staples, Journal of Public Policy & Marketing, Vol. 18, No. 1, 1999.

46 See RBB's "Dataroom Submission" 13 April 2007, Annex 4, page 13

47 It is also disconcerting to find out that the demand proxy that was defined in the RBB 8 February Annex appears to have been constructed in a very different fashion from what was indicated in that Annex. For example, on page 4 of that Annex RBB states that "For all destinations X out of Dublin, we compute an average yield from another destination to X to control for the demand for the flight Dublin to X, which we label pjt. For instance, the route London to Venice is used as a control for Dublin to Venice". However, RBB's Data room submission states that "the demand proxy is calculated for all destinations X out of Dublin by taking an average of the outbound yield of all other routes out of non-Dublin airports to X ..... which we label pjt. We therefore do not use an arbitrarily chosen demand proxy as seems to be suggested by the Commission."

48 This tests whether the regression equation is misspecified. Misspecification can result from omitting an important explanatory variable or form choosing the wrong functional form.

49 See Aer Lingus' response to the Statement of Objections of 18 April 2007, page 31.

50 RBB in fact argues in their Dataroom submission that "In order not to run into a problem of measurement error for the Aer Lingus presence dummy, our analysis excluded observations for the Dublin-Bristol and the Dublin-Liverpool routes. The Commission's criticism at paragraph 108 therefore does not apply. (Note that including these routes has no impact on our results.)"50. Again, the fact that RBB dropped several observations without mentioning it is disconcerting and leads one to wonder if observations were selectively dropped from other regressions.

51 This method to estimate fixed effects model is known as the dummy variable regression.

52 If the unobserved variables are expected to be uncorrelated with the other explanatory variables an alternative approach known as "random-effects" is likely to be more efficient than the fixed-effects method.

53 Robust estimation is used to control for error term's properties.

54 In the Data Annex its report of 9 February 2007, CRA highlights the fact that Aer Lingus operated a limited number of multi-segment flights in the period before April 2004. For these routes, the reliability of the variable measuring capacity is reduced by the unmeasured presence of connecting traffic. Later on CRA informed the Commission that also other routes are affected by the same problem (see par. 26). In relation to the period as of January 2002, the routes in which Aer Lingus operated a substantial number of multi-segments flights are: Dublin-Cork for the period between January 2002 and October 2003, Dublin- Frankfurt for the same period, Dublin- Birmingham (Jan 2002- May 2002). In order to check for the sensitivity of the econometric analysis to this issue, we have replicated all our regressions excluding the above mentioned routes for the relevant months from the sample. The effect of this modification on the final result on the regressions performed is minimal. On 22 March 2007, CRA has also confirmed that its results are not affected by the exclusion of the additional multi-segmented routes detected after the data submission 26 January 2007.

55 Further regressions used capacity share as the explanatory variable of interest with results also roughly consistent with those obtained in the first submission with a shorter panel.

56 CRA Reports of 23 November 2006, 6 December 2006 and 9 February 2007.

57 See RBB's "Dataroom Submission" 13 April 2007, Annex 4.

58 Route and time subscripts are dropped for convenience.

59 See RBB's "Dataroom Submission" 13 April 2007, Annex 4, page 27.

60 See RBB's "Dataroom Submission" 13 April 2007, Annex 4, page 29.

61 In Froeb & Werden (1991) "Endogeneity in the concentration-price relationship: causes and consequences" US Department of Justice, Antitrust Division, EADG Paper 91-7 the authors derive the asymptotic bias in the OLS estimator for a simple model with linear demand, constant marginal cost, Cournot competition and unobserved variation in demand and factor prices. They show that unobserved variation in demand produces a negative bias and observed variation in costs a positive bias. The net bias is most likely to be negative, but it need not be. They also show that the bias may be small, but it may also be larger than the true coefficient.

62 The Commission acknowledges that RBB's use of "matched route" demand proxy is a priori a promising approach to control for demand within the route less likely to raise concerns with endogeneity bias. However, for the reasons explained in section 6.1.3 this approach as implemented by RBB raises serious methodological concerns and likely results in an incorrect specification. The results are therefore not reliable. Furthermore the Commission points out that this or similar approaches to control for demand are not even an option in regressions to explain Aer Lingus' prices given that Aer Lingus does not, during the sample period, operate a single route connecting two non-Irish airports.

63 Given a model and data in which fixed effects estimation would be appropriate, a Hausman test tests whether random effects estimation (i.e. with the ai uncorrelated with the explanatory variables - see paragraph 149) would be almost as good. The Hausman test is a test of the null hypothesis that random effects would be consistent and efficient against the alternative hypothesis that random effects would be inconsistent. It is important to point out that if we believe that ai is uncorrelated with the explanatory variables then the coefficients of interest can be consistently estimated by using a cross-section. Hence, implicitly, the Hausman test also provides an indication of the suitability of the fixed effects model over the cross-sectional approach.

64 The only drawback of the use a fixed effect model after a rejection of the null hypothesis in the Hausman test relies on the larger variance of our coefficient. Given that consistency is not at stake, and rejection concern very few cases, we have not taken into account a random effect model.

65 For a relevant part of the frequency specifications, the standard Hausman test was not able to compute the matrix of the difference of the disturbance variances. In order to get around this problem, we have run the test as an f-test as presented in Wooldridge (2002), pp.290-291.

66 A carrier is considered to have a base at the destination airport if it operates more than 200 frequencies a month from that airport to any other short-haul destination. This roughly corresponds to operating 2 planes with a turnover slightly above 3 return flights per day for every day of the month.

67 Of course both total costs and seat capacity are highly correlated and thus their inclusion together reduces the efficiency of the estimator.

68 In these regressions we include Aer Arann and Cityjet in the non-flag category. This modification has not any effect in the final results.

69 For example, Evans, Froeb & Werden (1993) run a fixed effects IV model on the data of Evans & Kessides (1992). They regress price on route HHIs. As instruments they use a one-year lag of the route HHI. The coefficient more than doubles relative to the fixed-effects without instrumenting.

70 See paragraphs 167-168 of the SO; see also table 11 above.

71 As RBB explains "median regression is implemented in STATA using the qreg command. While OLS estimates the mean of a distribution, qreg estimates the median which is not as sensitive when the sample contains outliers".

72 The Commission has spotted a significant set of clear inconsistencies in the data submitted by Ryanair in reply to the Formal Request for Information of 27 January 2007. In response to the subsequent request for clarification, Ryanair sent an amended version on 21 February 2007. On March 8, Aer Lingus informed the Commission of the presence of a limited number of errors in their first submission and provided a new version of their dataset. Even if the inconsistencies detected did not refer to the fare variable, the Commission considered appropriate to take the eventuality of the presence of minor, unspotted errors in the data into account in performing their analysis.

73 This would mean a set of twelve dummies for each month of the year plus a dummy for each year present in the sample.

74 Presented in Column (1), Table 9 of Annex IV to the Decision.

75 Column (1), Table 14 of Annex IV to the Decision.

76 The example is fictional given that this date is not present in our sample. However, our model is able to capture the mid and long term effect of such a shock from January 2002 over.

77 According to this procedure, we consider as winter destinations Lanzarote, Fuerteventura, Las Palmas and Salzburg.

78 A big hub is defined as an airport which provides over 2000 average weekly frequencies for short and mid haul routes. Unfortunately, we have not such information for long haul routes. However, given that the only destination cities which satisfy this restriction are Amsterdam, Barcelona, Munich, Frankfurt, London, Madrid, Milan, Paris and Rome, we consider our definition absolutely reasonable. For Barcelona, which also satisfies the rule for summer destinations, we have opted for the inclusion in the "big hubs" group only.

79 In addition RBB reports confidence intervals derived using a block bootstrapped procedure

80 The p-values are higher when using RBB's approach to model time fixed effects, i.e. monthly and yearly dummies instead of time fixed effects. However as explained in the previous subsection this alternative does not appear justified. In any event, the coefficients remain significant at least at the 10% level - again irrespective of the model used.

81 Standard OLS (cluster) leads to p-value=0.013 in the (baseline) presence regression and p-value=0.168 in the (baseline) frequency regression. However, this is the only estimator that does not control for outliers. Further the Commission regards the Newey estimator as a superior technique to deal with any potential serial correlation as explained below.

82 RBB ignores the fact that a one-sided test can be considered more appropriate in this case - see the introductory section for further details. This is of limited relevance in the fixed-effects results reported by the Commission since, even on the basis of a two-sided test, the coefficients of Ryanair presence were significant at the 5% level in each and every one of the tested regressions (and in most cases at the 1% level), whether based on the presence or the frequencies specification. Nonetheless, the Commission notes that with a one-sided test all p-values should be halved. This would imply that even with RBB's preferred model, OLS cluster, the coefficient in the frequency specification is significant at the 10% level (with OLS cluster (restricted) the coefficient becomes statistically significant also at the 5% level).

83 See Wooldrige (2002), Econometric Analysis of Cross-section and Panel Data, section 11.5 , page 331, Cameron, A and P. Trivedi (2005), Microeconometrics methods and applications, page 834.

84 In the opposite case, such a variable would be correlated with the error term and report a biased coefficient.

85 (A): Column 1, Table 9. (B): Column 4, Table 9. (C): Column 4 Table 10.

86 (A), (B): Column 1 and 3 of Table 14.

87 See Orley Ashenfelter & David Ashmore & Jonathan Baker & Suzanne Gleason & Daniel Hosken, 2006. "Empirical Methods in Merger Analysis: Econometric Analysis of Pricing in FTC v. Staples," International Journal of the Economics of Business, Taylor and Francis Journals, vol. 13(2), pages 265-279, July. See also Jonathan B. Baker, "Econometric Analysis in FTC v. Staples, Journal of Public Policy & Marketing, Vol. 18, No. 1, 1999.

88 This is the same model presented in par. 189. For simplicity, with are aggregating the competition variables with the other regressors in xit.

89 This is clear considering the price effected estimated for the three month considered using specification (1). The highest average price increase under Method I is reported in November, while, under Method II, October reports the highest price effect. The price effect in overlapping routes (Method II) translates into a price effect at the systemic level (Method I) on the basis of the importance of the overlapping for each time period considered.. Coming back to the example, in the month of November we observe the highest share of overlapping over total routes (25/41=0.61). In October this share is lower (0.52). Concluding, even if the price effect for actual overlapping routes passengers is higher in October, for the average customer the effect is higher in November given the higher probability of incurring in an overlapping route.

90 See also paragraph 185 in the Statement of Objections, that is, paragraph 231 in this document.

91 See Ryanairs reply to the Article 6(1 )(c) Decision, page 34).

Annex IV.1

Ordinary least squares estimation applies to the linear multiple regression model:

(formule)

where Yi is observation i of dependent variable Y, X1i is observation i of independent variable X1, ß0 is the constant term and equals the coefficient of an implicit explanatory variable with value 1, ß1 etc. are the coefficients for the independent variables, ei is the residual. The ordinary least squares estimation method minimizes the sum of the squared residuals. When using a number of assumptions, OLS produces so-called best linear unbiased estimators (BLUE). These estimators are unbiased (i.e. expected to be equal to the true value) and efficient (i.e. estimated with smallest standard errors). The assumptions are:

1. There is no correlation between explanatory variables and residuals (no simultaneity), i.e. cov(Xji,ei) = 0. Failure of this assumption results in biased estimates of the coefficients of explanatory variables.

2. The expected or mean value of the residuals equals zero, i.e. E(ei) = 0. Failure of this assumption results in a biased estimate of the constant term.

3. Residuals are homoskedastic (no heteroskedasticity), i.e. E(ei2) = d2 = constant. Failure of this assumption results in inefficient estimates and biased tests of hypotheses.

4. Residuals are independently distributed (no serial correlation), i.e. E(eiej) = 0. Failure of this assumption results in inefficient estimates and biased tests of hypotheses.

5. Explanatory variables are independent (no multicollinearity), i.e. cov(Xi,Xj) = 0. Failure of this assumption results in inefficient estimates and biased tests of hypotheses. In addition to these well-known standard assumptions, we also have:

6. Residuals are normally distributed, i.e. e ~ N(0, d2). Together with assumptions 2 and 3 and 5, failure of the normality assumption invalidates the use of the Student t-distribution in coefficient t-tests.

7. Explanatory variables are measured without error (no errors in variables). Failure of this assumption results in biased estimates of the coefficients.

8. Variables that are time series must be stationary (no unit roots), i.e. well-defined mean and variance. Failure of this assumption results in spurious regressions (except in the special case of cointegration).

If one or more of these assumptions is found to be rejected in a regression analysis, it is necessary to adjust the estimation approach.

Opinion of the Advisory Committee on Concentrations given at its 151st meeting on 11 June 2007 concerning a draft decision relating to M.4439 - Ryan Air/Aer Lingus

Rapporteur : POLAND

1. The Advisory Committee agrees with the Commission that the notified operation constitutes a concentration within the meaning of Article 3(1)(b) of the EC Merger Regulation.

2. The majority of the Advisory Committee agrees with the Commission that the notified transaction can be deemed to have a Community dimension pursuant to Article 1(3) of the EC Merger Regulation.

3. The Advisory Committee agrees with the Commission's definitions of the relevant markets as stated in the draft decision.

4. The Advisory Committee agrees with the Commission that the proposed transaction would significantly impede effective competition in the common market as a result of the creation of a dominant position of Ryanair and Aer Lingus on 35 routes from and to Dublin, Shannon and Cork.

5. The Advisory Committee agrees with the Commission that the proposed transaction would significantly impede effective competition as a result of the creation or strengthening of a dominant position by the elimination of a credible potential entrant on 15 routes from and to Dublin and Cork.

6. The Advisory Committee agrees with the Commission that the proposed transaction does not give rise to efficiencies that would counteract the significant impediment to effective competition resulting from it.

7. The Advisory Committee agrees with the Commission that the final commitments offered by Ryanair do not address the competition concerns identified by the Commission in a sufficient manner and will therefore not eliminate the significant impediment to effective competition created by the transaction.

8. The Advisory Committee agrees with the Commission that the notified concentration must be declared incompatible with the common market and the functioning of the EEA Agreement.

9. The Advisory Committee asks the Commission to take into account all the other

<emplacement tableau>

FINAL REPORT OF THE HEARING OFFICER IN CASE COMP/M.4439 - Ryanair/Aer Lingus

(pursuant to Article 15 and 16 of Commission Decision (2001-462-EC, ECSC) of 23 May 2001 on the terms of reference of Hearing Officers in certain competition proceedings - OJ L162, 19.06.2001, p.21)

On 30 October 2006 the Commission received a notification pursuant to Article 4 of Council Regulation (EC) No 139-2004 (the Merger Regulation) of a proposed concentration by which Ryanair Holdings Plc (Ryanair) acquires, within the meaning of Article 3.1 (b) of the Merger Regulation control of the undertaking Aer Lingus Group Plc (Aer Lingus) by way of a public bid announced on 4 April 2006.

After an initial examination of the notification the Commission concluded that the notified concentration fell within the scope of the Merger Regulation and that, even taking into account commitments offered by the notifying party on 19 November 2006, subsequently modified on 14 December 2006, the proposed concentration raised serious doubts as to its compatibility with the common market and the functioning of the EEA Agreement.

Accordingly, on 20 December 2006, the Commission decided to initiate proceedings in accordance with Article 6.1(c) of the Merger Regulation. On 22 February 2007 the Commission, in agreement with Ryanair, decided to extend the procedure by 20 days in accordance with Article 10.3, second subparagraph, of the Merger Regulation.

On 27 March 2007 the Commission sent a Statement of Objections to Ryanair. Access to the file was granted by way of a CD-ROM on 29 March 2007 and a subsequent Data-room procedure that had been agreed between Ryanair and Aer Lingus, in which commercially sensitive data could be analysed under conditions that ensured that the confidentiality of the underlying information was maintained. Ryanair submitted a written reply to the Statement of Objections on 17 April 2007. Further access to file was granted on 1, 6 and 8 June 2007.

During access to file Ryanair was inadvertently provided with confidential information of certain other parties. Remedial measures were undertaken and the affected parties were informed, as appropriate.

Ryanair did not request the opportunity to develop their arguments at an oral hearing pursuant to Article 14.1 of Regulation No 802-2004.

On 17 April 2007 Ryanair submitted commitments in order to resolve the competition concerns identified in the Statement of Objections. These commitments were subsequently modified on 3 May 2007, which triggered an extension of the deadline by 15 days in accordance with Article 10.3 of the Merger Regulation. The Commission conducted a market test of these commitments and Ryanair was granted access to the non-confidential replies to this market test on 16, 23 and 30 May. I have not been asked to verify the objectivity of the market inquiry.

After having examined the commitments, as modified by Ryanair on 3 May 2007, the Commission concluded that they did not sufficiently address the competition concerns identified in the Statement of Objections and that therefore, the concentration would significantly impede effective competition on the common market.

In my opinion the draft decision relates only to objections in respect of which Ryanair has been afforded the opportunity to make known their views.

Ryanair has not raised before me any issues with regard to its rights of defence during this procedure.

Other involved parties

Aer Lingus has had a particular position in the proceedings, as it is the target of a hostile take-over bid, and as such considered to be an "other involved party" within the meaning of Article 11 (b) of Regulation No 802-2204. Commensurate with this position Aer Lingus was closely involved at every stage in the proceedings. In particular, certain key documents were provided to Aer Lingus in the course of the Phase II investigation. It received a non-confidential copy of the Statement of Objections. It was also granted access to other documents in the file to the extent considered appropriate by DG Competition in order for it to submit its comments in accordance with Article 17.2 of Regulation No 802-2004. This included, amongst others, access to the Data-room in a manner similar to Ryanair. Aer Lingus also received a non-confidential copy of Ryanair's response to the Statement of Objections.

Aer Lingus claimed that its access to file was unduly limited and that it should have been entitled to full access to the file on the same basis as the notifying party in order to fully exercise its right to be heard. This claim was rejected by DG Competition and Aer Lingus did not pursue this issue subsequently before me.

Aer Lingus did not request the opportunity to develop their arguments at an oral hearing pursuant to Article 14.2 of Regulation 802-2004.

Other third parties

Two interested third parties, within the meaning of Article 11 (c) of Commission Regulation 802-2004, requested to be admitted to the proceedings, namely the Irish Department of Transport (DoT) and SIPTU, a union representing employees of Aer Lingus. By decisions dated 4 April and 12 April 2007 I decided to admit them into the proceedings. Subsequently, DG Competition granted them the opportunity to make their views known.

Conclusion

I view of the above I consider that the rights to be heard of all participants to the present proceedings have been respected.

Brussels, 20 June 2007

(signed)

Karen Williams