EC, August 21, 2007, No M.4523

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(1) thereof, Having regard to the Commission's decision of 3 May 2007 to initiate proceedings in this case, After consulting the Advisory Committee on Concentrations (2), Having regard to the final report of the Hearing Officer in this case (3), Whereas:

(1) On 15 January 2007, the Commission received a request pursuant to Article 4(5), followed by a formal notification on 23 March 2007 pursuant to Article 4 of Regulation (EC) No 139-2004 ("the Merger Regulation") of a proposed concentration by which the undertaking Travelport LLC ("Travelport", USA), a subsidiary of The Blackstone Group ("Blackstone", USA), acquires within the meaning of Article 3(1)(b) of the Merger Regulation control of the whole of Worldspan Technologies Inc. ("Worldspan" USA) by way of purchase of shares.

(2) By decision dated 3 May 2007, the Commission found that the notified operation raised serious doubts as to its compatibility with the common market and the functioning of the EEA agreement. The Commission accordingly initiated proceedings in this case pursuant to Article 6(1)(c) of the Merger Regulation.

1. THE PARTIES

(3) Travelport ("the notifying party"), a subsidiary of Blackstone, aggregates content from airlines, hotels, car rental agencies and other travel suppliers and distributes this content to final consumers. The company operates Galileo - a global distribution system ("GDS") and Gulliver's Travel Associates. In addition, Travelport operates a number of online travel agencies and global websites, including ebookers, Orbitz, Cheaptickets, Octopus Travel, HotelClub and RatesToGo.

(4) Worldspan provides travel distribution services through the Worldspan GDS. The company focuses on providing GDS services to on-line and, more recently, also to traditional travel agencies, primarily in the leisure sector. Furthermore, Worldspan provides certain IT services to airlines (for example, internal reservation systems and flight operations technology services).

2. THE CONCENTRATION

(5) The proposed operation concerns a concentration within the meaning of Article 3(1)(b) of the Merger Regulation, since it concerns the acquisition by Travelport of sole control of Worldspan through the acquisition of the latter's entire share capital.

3. COMMUNITY DIMENSION

(6) The undertakings concerned have a combined aggregate worldwide turnover of more than EUR 5 000 million (4). The aggregate Community-wide turnover of Blackstone exceeds EUR 250 million (5). However, Worldspan's Community-wide turnover (6) does not exceed EUR 250 million, which means that the proposed transaction does not have a Community dimension within the meaning of Article 1 of the Merger Regulation. (7)

4. ARTICLE 4(5) REFERRAL

(7) The proposed transaction would have been subject to mandatory scrutiny under national merger control law in four Member States (Austria, Germany, Italy and Poland) with two additional Member States competent to review the transaction (the United Kingdom and Ireland).

(8) On 15 January 2007, the Commission received a reasoned submission by Travelport in which the company requested a referral to the Commission pursuant to Article 4(5) of the Merger Regulation. No Member State objected to the referral of the proposed transaction to the Commission. The transaction has therefore been examined by the Commission.

5. RELEVANT MARKETS

5.1. Relevant product market

5.1.1. Definition of the product market

(9) In previous cases, the Commission has defined a GDS as a tool provided to travel agents ("TAs") in order to allow them to obtain information and make reservations related to airlines and other internationally operating travel service providers ("TSPs") including car rental companies and hotels. These TSPs supply the GDS with data on the products they provide. (8)

(10) The product market affected by this transaction is the market for electronic travel distribution services through a GDS. A GDS is a two-sided platform through which TSPs such as airlines, car rental companies and hotel chains (9) distribute their travel content to TAs and ultimately to end-consumers. At the same time, TAs can access and book travel content for end-consumers.

(11) GDS providers act as intermediaries in a market of a two-sided nature, connecting two separate customer categories. In the upstream market (the TSP side of the market), TSPs offer GDSs information on their booking inventory and the content, while the GDSs offer TSPs booking capabilities and a distribution channel to TAs. (10) In the downstream market (the TA side of the market), GDSs offer TAs reservation, booking and ticketing services by means of a comprehensive tool which allows comparison of prices and conditions from hundreds of TSPs. The Commission's indepth investigation (11) has confirmed these characteristics of the product market.

5.1.2. General market description

(12) The existence of the GDS platform is justified by the added value it creates. A GDS coordinates the demand of TAs, thereby generating a positive network externality which is internalised by the TSPs. Since they allow access to a broad network of TA outlets (and indirectly to a large number of end-consumers), GDS providers are effective distribution channels for TSPs ("network effect"). In particular, a centralised search for fares in one GDS is more effective and less time-consuming for TAs than multi-channel searches from numerous TSP-specific sources. (12)

(13) Whenever there are several providers of the same type of platform, customers on each side of the platform may choose to subscribe to one provider only ("single-homing") or to several providers ("multi-homing"). In two-sided markets customers do not necessarily make the same choice (that is to say, to opt for single-homing or multi-homing"), either within the same customer group, or across the two groups.

(14) The concept of multi-homing covers different platform choices. A multi-homing customer may either subscribe to all available platform providers or to more than one but not all of them. In addition, all or only some of the customers on each side of the platform may be multi-homing.

(15) In this case, virtually all airlines subscribe to all GDS providers. (13) Other TSPs (car rental firms and hotel chains) tend to do the same, whereas TAs generally tend to use single-homing. (14)

(16) Customers' actual choice of single-homing or multi-homing depends on several factors. Firstly, it depends on the degree of asymmetry in the network effects and in particular on whether or not they arise only/mainly on one side of the platform. If the number of "reachable" TAs is important for the TSPs (network externalities generated on the TA side), multi-homing becomes more attractive for the TSPs. The fact that virtually all TSPs use multi-homing, reduces the network externalities on the TSP side (or removes them altogether) and makes single-homing the most viable option for TAs (because multi-homing does not allow TAs to reach more TSPs). Secondly, the choice depends on the degree of differentiation between the services proposed by each provider over their respective platforms. Thirdly, customer preferences may determine the choice to use single homing or multi homing.

(17) For instance, a TSP targeting a given geographic market will, in order to maximise the coverage of end-consumers, provide content to all GDS providers offering an effective distribution channel in that geographic market. The TSP will therefore opt for multi-homing by distributing its inventory via all (geographically relevant) GDS providers. However, given the fact that contracts between GDS providers and TSPs are normally concluded on a global basis, TSPs will tend to subscribe to all GDS providers. If a sufficient number of TSPs use multi-homing (which means that each GDS provides a broadly similar content), TAs will only need to subscribe to one GDS, because the added value of subscribing to a second GDS will be close (or equal) to zero (therefore not off-setting the additional costs generated by subscribing to two instead of one GDS). (15) Given the fact that most TAs are net receivers (16), possible price differences of GDS providers are not a significant incentive for TAs to use multi-homing. In addition, the cost of subscribing to an additional GDS does not in itself constitute an obstacle to multi-homing for TAs.

(18) If a TSP withholds specific content from one GDS and not from another, an element of differentiation is introduced which is of great relevance to TAs (for example, if the lowest fares of a TSP which is important sales-wise for a given TA are not being distributed via all GDS providers). In such cases, TAs may switch to another GDS providing all fares (including the lowest fares) or even opt for multi-homing. This scenario - albeit simplified - illustrates how the loss of content upstream may cause the GDS provider to lose market share downstream (due to TAs switching away from the platform which no longer provides complete content).

(19) As regards network effects, as for any two-sided platform, demand on one (or both) side(s) tends in principle to vanish if there is no demand on the other side. The size of the network on one side determines the willingness of customers on the other side to pay in order to join the platform. The larger the number of "reachable" TAs, the higher the positive network externalities (in terms of volumes of bookings) generated via a given GDS. The higher the positive network externalities, the higher the price that a TSP is willing to pay to distribute content via that GDS.

(20) Platform differentiation is determined by the different GDS providers' coverage in terms of content (for example, geographic areas covered, inclusion of "non-airline" content), network size, optional services provided (for example, additional functionalities for TSPs and TAs) and the quality of the technical support to TAs and TSPs. In this market, the crucial issue is content. Multi-homing on one side of the market decreases the incentives for multi-homing on the other side. In this specific case, the larger the number of fully multi-homing TSPs, the lower the demand for multi-homing from TAs, given the fact that all GDS providers distribute comparable content.

(21) Having analysed the general functioning of the market for GDS services, the Commission must assess the available alternatives to the GDS platform and determine whether any of these alternative technologies ought to be included in the same relevant product market.

5.1.3. Alternatives to GDS services

(22) According to the notifying party, the GDS platform is only one of a number of alternative technological platforms which distribute travel related content to end consumers. These alternative platforms may be used by TSPs, TAs and end consumers (either directly, or indirectly via TAs) to by-pass the use of the GDS platform. These alternative technologies are (i) "GDS New Entrants", (ii) meta search engines, (iii) direct links and (iv) "supplier.coms".

(23) In order to determine whether any of these alternative technologies should be included in the same relevant product market as the GDS-providers, the Commission has assessed the extent to which these alternatives are substitutable to GDS services.

5.1.3.1. "GDS New Entrants"

(24) The notifying party states that several companies have recently developed alternative technological platforms providing travel content and booking capabilities. (17) Using these alternative platforms - called "GDS New Entrants" ("GNEs") by the notifying party (18) - would allow TAs to by-pass the traditional GDS providers.

(25) At present GNEs are only active in the US market, where they account for a small portion of all bookings made. In the US, TAs tend to use GNEs as complements rather than substitutes to the GDS providers. The market investigation has not revealed any indications that EEA market entry by GNEs is imminent. In the EEA, GNEs must primarily be regarded as potential market entrants only. If they enter the EEA market, judging by developments in the US, GNEs are likely to function as partial substitutes to the GDS platform and become "fringe" players with little impact on the market.

(26) During the market investigation, the Commission contacted a number of GNEs identified by the notifying party. None of the alleged GNEs considered themselves as direct competitors to the GDS providers in the EEA, but rather as software technology providers or complements to the GDS providers. (19) Moreover, certain GNEs are "piggy-backing" on the GDS providers. (20) Parts of the travel content included on such alternative platforms originate from a GDS (21) and bookings are ultimately processed via one of the GDS providers. The "piggy-backing" GNEs - while providing TAs with an alternative interface offering travel content and booking capabilities - are dependent on the traditional GDS providers. GNEs may offer TAs access to certain negotiated fares that are not directly available in a GDS. However, the actual booking of the negotiated fare may subsequently be made via a GDS.

(27) None of the TAs in the EEA which received questionnaires from the Commission indicated that they used the services of a GNE. (22)

(28) None of the TSPs questioned by the Commission currently use GNEs for bookings in the EEA. (23)

(29) The Commission concludes that the competitive constraint exerted by GNEs on the market behaviour of GDS providers in the EEA is currently very limited, if not nonexistent. Moreover, entry to the EEA market is likely to be costly for the GNEs. Not only would the GNEs have to offer TSPs their product at prices lower than the incumbent GDS providers, more importantly they would also have to offset the incentive payments that TAs would forego by not booking via GDS.

(30) Bearing these considerations in mind, it is concluded that in the EEA, GNEs should not be included in the same relevant product market as the GDS-providers.

5.1.3.2. Meta search engines

(31) There are a number of internet web-sites which aggregate and compare travel service fares (for example, from the "supplier.coms" of LCCs and other scheduled airlines), enabling visitors to those sites to find the most attractive fares available online. (24)

(32) However, meta search engines do not have any own booking capabilities which means that consumers (and possibly TAs) will have to go to a "supplier.com" to book their tickets (or, in the case of TAs, possibly a GDS). Meta search engines primarily function as a channelling tool which point consumers towards the travel service provider offering the best fares. Meta search engines are therefore likely to increase end-consumers' use of "supplier.coms" but they do not substitute TSPs as TAs need to use the services of GDS providers (25).

(33) It is therefore concluded that meta search engines should not be included in the same relevant product market as GDS services.

5.1.3.3. Direct links

(34) According to the notifying party, certain TSPs have concluded agreements with larger TAs according to which a direct link between the TSP and the TA is established. Via the direct link, the TA gains direct access to the booking inventory of the TSP and may book tickets directly using this link, thereby by-passing the GDS providers (26). By circumventing the GDS in this manner, the TSP avoids paying booking fees to the GDS provider. Provided that the cost of operating the direct link is lower than the cost of booking via GDS, TSPs ought to have strong incentives to establish direct links to major TAs.

(35) For several reasons, the incentives for TAs to invest in direct links are less clear-cut. First, given the fact that a direct link only covers the inventory of a single TSP, whereas a GDS comprises the inventories of hundreds of TSPs, a single direct link substitutes only a limited portion of the bookings made via GDS (27). In order to become a viable substitute to the GDS, the TA would have to establish direct links with the major TSPs in the geographical area in which the TA is active (28). Second, the TA would need to invest in software which incorporates the direct links and a GDS (in order to cover content not covered by the direct links) into a single user-friendly and efficient interface. The transaction costs for establishing a set of direct links and combining them in a comprehensive user interface are likely to be high (29). Third, TAs booking via a direct link instead of a GDS, would forego incentive payments from the GDS provider. As will be shown below (30), these payments are substantial. In order to make a direct link attractive to the TA, the TSP would have to offer a solution which offsets lost incentive payments from the GDS providers and make investments in a user interface worthwhile for the TA.

(36) None of the EEA-based airlines questioned by the Commission have indicated that they have established direct links with TAs. However, several respondents indicated that they were considering establishing such links with TAs in the future. So far, direct links have primarily been established in the US (31) American Airlines and Delta Airlines have stated that they have established direct links with TAs in the US, via Orbitz SupplierLink. United Airlines maintains a direct link with United's internal booking system in the US.

(37) Given their absence from the EEA market, direct links will not constrain the market behaviour of the merged undertaking in the short term. In the longer term direct links could function as partial substitutes to GDS services for TSPs and TAs. In order to become a viable but partial substitute to the GDS for the TA, multiple direct links must be set up and incorporated into a single, user-friendly interface. Given the fragmented nature of the EEA market (32), which implies high transaction costs, and the current absence of incentives for TAs to substitute GDS bookings with bookings via direct links, a possible future establishment of direct links between TSPs and TAs may be expected to have a very limited impact on the market behaviour of the merged undertaking in the medium term.

(38) Bearing these considerations in mind and for the purposes of this decision, direct links cannot be regarded as belonging to the same relevant product market as the provision of GDS services.

5.1.3.4. "Supplier.coms"

(39) At present, practically all TSPs operate their own internet web-sites, through which end-consumers may book travel services (flights, rental cars, hotel rooms) directly ("supplier.coms").

(40) The fact that the uptake of supplier.coms has been substantial during the last few years is beyond doubt. Using IATA data, the notifying party has estimated the portion of direct sales in relation to total sales for the 20 largest airlines in the EEA. Airlines' direct sales include to a very large extent on-line sales via their own websites and, to a lesser extent, off-line sales via their own call-centres and ticket offices. On average, [20-30%]* of all bookings of these airlines were direct in 2005 (compared to [10-20%]* in 2004 and [10-20%]* in 2003). (33)

(41) In order to determine whether the services provided via supplier.coms are substitutable to GDS services to an extent which would allow the inclusion of the two types of services in the same relevant product market, a complex assessment covering both sides of the market must be carried out. On the upstream side of the market, TSPs have very strong incentives to promote the use of supplier.coms, whereas on the downstream side of the market, TAs have few incentives (or rather have disincentives) to use supplier.coms.

(42) On the upstream side of the market, supplier.coms are used by TSPs as an alternative distribution channel whereby end-consumers may be reached directly, by-passing not only the GDS providers but also the TAs.

(43) TSPs have strong economic incentives to increase bookings via their own distribution platform rather than via GDS. Although the setting-up of a supplier.com web-site and promoting it require large up-front investments, the Commission's indepth investigation has confirmed that the TSPs' average costs per booking via their own supplier.coms are in most cases substantially lower than their average costs per booking via GDSs. The average cost per booking segment for direct bookings made via the airlines' own web-sites was estimated by the respondents to the Commission's in-depth investigation to be between EUR 1 and EUR 5,50. (34) However, when assessing TSPs incentives to steer booking traffic away from GDS providers to supplier.coms, the most appropriate means of doing so is by looking at the marginal costs which are involved in such shifts. Once the initial investments have been made and the supplier.com of a TSP is up and running, the marginal cost (that is to say, the TSP's cost of booking one more segment via its web-site) is negligible, whereas for a "marginal" booking via a GDS, the airline's marginal cost of booking includes the full booking fee charged by the GDS (minus possible volume-based discounts) as well as the possible booking commission paid to the TA doing the GDS booking. (35) In this respect, the marginal cost for the TSP is between EUR 0.20 and EUR 2, according to the respondents to the Commission's in-depth investigation.

(44) The fact that the TSPs' marginal costs for booking via supplier.coms are substantially lower than the marginal cost of booking via GDS explains TSPs' current propensity to promote supplier.coms at the expense of GDS providers. Another reason is the need of conventional airlines to compete with LCCs, whose main distribution channel is supplier.com. An additional incentive for TSPs to promote supplier.com is related to the fact that when the end-user logs on to the web-site of the TSP, the enduser is no longer able to do comparison-shopping (or comparison shopping becomes more difficult because the end user has to consult multiple web-sites), which leads to higher yields for the TSP.

(45) However, the diversion of booking volumes to supplier.com is limited by the loss of network externalities, that is to say, the possibility for TSPs to reach a large number of end-consumers via the distribution channel represented by the network of TAs connected to the GDS provider. This element is likely to be more important on away- from-home markets, where the probability that an end-consumer visits the supplier.com web-site of the TSP is likely to be lower than in the home market.

(46) Having assessed the TSPs' incentives to divert bookings from GDS providers to their own supplier.coms, the Commission must investigate the TSPs' ability to do so. The ability of airlines to divert booking traffic away from the GDS providers depends on a number of circumstances specific to each airline. First, the market investigation has confirmed that economy class tickets are more likely to be booked via supplier.com than business class tickets. Second, unmanaged travel is more likely to be booked via supplier.com than managed travel. (36) Third, simple itineraries are more likely to be booked via supplier.com than more complex itineraries involving connecting flights and flights combined with rental cars or hotels. Consequently, airlines having a large proportion of point-to-point routes are able to divert more bookings to supplier.com than airlines with "hub and spoke" networks and a large number of interconnecting flights. Fourth, tickets booked in an airline's home market are more likely to be booked via supplier.com than tickets booked in other countries. The fact that most airlines - in particular the legacy "flag carriers" - have the largest market shares in their home markets, that airlines' brand image is strongest in the home market and that there are normally no linguistic barriers within the home market, are all circumstances which facilitate booking diversion towards supplier.coms. Internet penetration in the airline's home market is another decisive factor for the ability to divert booking traffic. The higher the degree of internet penetration, the easier it is to convince end-consumers to book via supplier.com.

(47) These circumstances explain why certain airlines have taken the strategic decision to sell tickets exclusively (or almost exclusively) by direct means, the most notable examples being Ryanair and EasyJet. The fact that these LCCs do not issue business tickets, have only leisure travellers or corporate travellers travelling outside travel management programmes, operate point-to-point routes, have no interlining agreements with other airlines (thereby avoiding complex bookings) and operate in countries with high degrees of internet-penetration, explain why these carriers have been able to forego GDS bookings completely (Ryanair) or almost completely (EasyJet).

(48) Among airlines with a limited ability to divert large volumes bookings from the GDS providers, one would expect to find airlines selling primarily business class tickets, airlines operating "hub-and-spoke" networks (37), airlines operating regional networks "feeding" into major airport hubs (resulting in a larger proportion of complex interlining tickets), airlines with a comparatively low proportion of sales made in the home market in relation to sales in the rest of the world and airlines whose home markets are less developed when it comes to internet penetration.

(49) Depending on the business model chosen by each airline and the exogenous factors set out in this section, airlines' ability to shift bookings away from GDS providers varies considerably and it may be concluded that all airlines - with the exception of the certain LCCs - have a portion of bookings that are in fact "captive" to the GDS providers. Most airlines are therefore faced with a "ceiling" above which bookings via supplier.coms may not be increased. A major European airline that currently processes approximately 20% of booking segments via its suppler.com website, has indicated that the company is relatively close to its "ceiling".

(50) It may therefore be concluded that for the vast majority of airlines, supplier.coms function as partial substitutes to GDS services and that large volumes of airlines' bookings are to some extent "captive" to the GDS providers.

(51) The arguments set out in this section mainly concern airlines. However, most of these arguments also apply to rental car firms and hotel companies. Although these companies are far less dependent on GDS services than airlines, a certain proportion of their bookings are also "captive" to the GDS providers. It is important for rental car firms and hotels to be distributed via GDS in order to capture travellers wanting to "add on" a rental car and/or a hotel booking to their GDS flight reservation. A large number of these "add on" bookings concern higher yield corporate travellers.

(52) As regards the downstream side of the market, since bookings via supplier.coms bypass not only the GDS providers but also TAs, it seems reasonable to conclude that TAs have disincentives to substitute bookings via GDS with bookings via supplier.coms. Since supplier.coms are designed for self-booking by end-consumers themselves (38), TAs have strong incentives to keep using a GDS and dissuading customers from booking themselves.

(53) From an economic point-of-view, the disincentives are strong. By booking a client's ticket via supplier.com, the TA has to forego not only the incentive payment paid by the GDS provider but also a possible booking commission paid by the airline. In order to compensate this loss of revenue, TAs need to charge the end-user a service fee (which in its turn gives the end-consumer the additional incentive to book his tickets himself via supplier.com instead of the TA, further decreasing the revenues of the TA).

(54) From a practical point-of-view, the Commission's market investigation has confirmed that TAs do not see supplier.coms as substitutes for a GDS, mainly because they lack versatility and are more cumbersome to use. A supplier.com only covers the inventory of a single TSP, whereas a GDS includes the inventories of hundreds of TSPs. By booking via a supplier.com instead of a GDS, the TA loses his main competitive advantage, namely the ability to compare prices and conditions of all TSPs and to instantly book the option preferred by the client (the belief of the customer that the TA possesses superior information is the reason for consulting the TA in the first place).

(55) Consequently, TAs are likely to book tickets via supplier.com only when they have no alternative, for example, when booking tickets with a TSP that has either completely withdrawn from the use of GDS (like certain LCCs) or reserves certain content for the supplier.com (39).

(56) Considering that substitution upstream is only partial, leaving substantial volumes of TSPs' bookings "captive" for the GDS providers and that downstream substitution is very limited or non-existent, it is concluded that supplier.coms should not be included in the relevant product market in which GDS-providers are active.

(57) Nonetheless, it should be emphasised at this stage that it is likely that in particular the rapid uptake of supplier.coms will have an impact on the competitive conditions in the market for GDS services and constrain the market behaviour of the merged undertaking on the EEA market. This impact and these constraints will be analysed in the competitive assessment.

5.1.3.5. Conclusion on the definition of the relevant product market

(58) It is concluded that a "GDS only" product market best reflects the current competitive conditions for the GDS providers in the EEA.

(59) Contrary to the GDS platform, supplier.coms have end-consumers, instead of TAs, on the "downstream" side. (40) Other platforms connect the same customer groups as GDS providers (that is to say, TSPs upstream and TAs downstream) but are not substitutable because of limited functionalities or higher costs (for example, direct links).

5.2. Relevant geographic market

5.2.1. The upstream side of the market

(60) The notifying party submits that the upstream side of the market (that is to say, TSPs providing booking inventory to GDS providers) is at least EEA-wide.

(61) The majority of the replies to the Commission's market investigation state that the market is global. The main argument put forward by the respondents relates to the fact that they conclude worldwide agreements with the GDS providers.

(62) However, despite these views, there are good reasons to conclude that the market ought to be considered as EEA-wide rather than global.

(63) The market shares of the GDS providers vary substantially depending on region and country concerned. Post-merger in an EEA-wide market, Amadeus will be by far the largest GDS provider in relation to TSPs, with a market share in 2006 of [50- 60%]* (41), whereas Galileo/Worldspan will become the second largest operator with a market share of [20-30%]* (Galileo [20-30%]* and Worldspan [0-10%]*). Sabre's EEA market share amounts to [10-20%]*. On the North-American market the situation is actually the reverse. Sabre is market leader with [40-50%]*, Worldspan holds [20-30%]*, Galileo [10-20%]* and Amadeus has the lowest market share with [0-10%]* (42).

(64) Further, although TSPs tend to conclude worldwide agreements, these agreements normally include separate regional pricing schemes for the EEA, the US and other parts of the world. It appears that list price booking fees paid by TSPs for a booking made in a GDS in the EEA are substantially higher than in the US (43)

(65) These differences in booking fees between the EEA market and the US can be partly explained by the fact that the US market has been deregulated (44) since 2006, whereas the EEA market is currently regulated by the EU Code of Conduct. In 1989, the vast majority of airline bookings were made through GDSs and most of the GDSs were owned and controlled by airlines (45). The Community considered at the time that, in order to deal with the competition concerns arising in the supply chain of air transport products, it would be more efficient in terms of transport policy and market efficiency to develop an ad hoc regulatory framework than merely rely on the generally applicable provisions of competition law. (46)

(66) The Code of Conduct recognised that GDSs required a certain degree of regulation in order to ensure that all TSPs enjoy the same level of access to TAs and consumers. It was established with the aim of improving transparency and preventing discriminatory behaviour both by the system vendors themselves and also by TSPs, especially in the frequent cases where, in the past, airlines had a stake in the ownership and control of a GDS. On the one hand, GDSs were required to deal in an even-handed manner with all TSPs and TAs, while, on the other, airlines with a financial stake in a GDS were required not to favour that system over the others. The Code of Conduct also imposed obligations in terms of neutral display in order to avoid discriminatory treatment of airlines on the system's principal display.

(67) Since the last change to the Code of Conduct in 1999, the airline distribution market has experienced important developments: most airlines have divested their participation in GDSs - although some airlines still hold minority stakes - and alternative distribution channels are gaining ground, mainly due to the development of direct Internet sales. (47)

5.2.2. The downstream side of the market

(68) The notifying party submits that the downstream side of the market (that is to say, GDS providers providing reservation, booking and ticketing services to TAs) is national in scope. In the past, the Commission has considered the downstream side of the market to be national in scope, while recognising the possibility of a wider market evolving due to "the development of Internet-based travel service information and reservation systems that make information and data available across borders." (48)

(69) As can be observed from the following table the market shares of the GDS providers in 2006 varied significantly between Member States (49):

<table>

(70) The responses of TAs to the Commission market investigation fully confirmed that the markets are still predominantly national in scope (50). Almost all TAs - often including online TAs - are still active in only one country, with the exception of a few TAs/Travel Management Companies ("TMCs"), such as American Express and Carlson Wagonlit, which could be considered to have pan-European (or worldwide) activities. The subscription fees that TAs pay for the use of a GDS and the incentive payments they receive vary between countries. In addition, Amadeus and Galileo have established national sales and service points in almost all countries of the EEA, in order to better serve the specific national markets (51).

5.2.3. Conclusion

(71) It is therefore concluded that the geographic market is EEA-wide on the upstream side of the market and national in scope on the downstream side of the market.

6. COMPATIBILITY WITH THE COMMON MARKET AND THE EEA AGREEMENT

6.1. Introduction

(72) The notified transaction reduces the number of GDS providers active in the EEA from four to three and increases significantly the merged undertaking's combined market shares in a number of Member States. The merger could theoretically lead to non-coordinated as well as coordinated effects. As regards non-coordinated effects, the Commission identified three theories of harm which it considered prima facie plausible. In addition to assessing the risk of coordinated effects, the Commission focused its in-depth investigation of non-coordinated effects on assessing whether:

(i) the merger would allow the merging undertakings to use their strong market position downstream vis-à-vis TAs in order to increase prices vis-à-vis TSPs upstream ("vertical cross market effects");

(ii) the merger would eliminate Worldspan as the alleged "pricing Maverick" and therefore lead to post-merger price increases;

(iii) the merger would allow the parties to exploit their post-merger market power vis-à-vis TAs in Member States in which Galileo/Worldspan would have high market shares.

6.2. Non-coordinated effects

6.2.1. Vertical cross-market effects

6.2.1.1. Description of the theory of harm

(73) During the initial stage of the investigation, concerns were raised that Galileo/Worldspan would be able to leverage its post-merger market power vis-à-vis TAs in a number of national downstream markets, in order to strengthen its bargaining power in relation to TSPs operating on the upstream EEA market. This possibility to leverage market power can be labelled as a "vertical cross-market effect". Such an effect could be described as follows.

(74) After the merger, Galileo/Worldspan would obtain large market shares (above 40%) with significant increments on the downstream side of the market in Ireland, the United Kingdom, Italy, the Netherlands, Hungary and Belgium. In these Member States the 2006 market shares ranged from [40-50%]* to [70-80%]*(see section

6.2.3. Non-coordinated effects on the downstream side of the market).

(75) If a TSP has a particular interest in having an extensive distribution network in a Member State where the merged undertaking would have a broad TA network, Galileo/Worldspan could possibly leverage its downstream market share in that Member State in order to gain concessions from the TSP when negotiating a worldwide agreement. In other words, the bargaining position of the merged firm vis-à-vis TSPs could result in greater market power than Galileo/Worldspan's upstream market share of [20-30%]* in the EEA would suggest. This bargaining power could possibly allow the combined entity to raise prices unilaterally post merger.

(76) The occurrence of non-coordinated price effects is a concrete possibility in the two sided market for GDS services, in which (i) downstream customers (TAs) are in a favourable position because each GDS provider must have an extensive network of TAs to be able to offer TSPs an effective distribution channel and (ii) upstream customers (TSPs) are dependent on the GDS providers to distribute their travel content. TSPs have strong incentives to use multi-homing in order to maximise their "reachable" customer base). In such a situation, the GDS providers have strong 17 incentives to maintain and expand their TA network by providing financial assistance to TAs and recoup that investment (and generate their margin) by extracting rents upstream from the TSPs. The merged undertaking thus has an incentive to translate high downstream market shares into unilateral price increases upstream.

6.2.1.2. Market functioning - "multi-homing/single-homing"

(77) In order to fully understand the functioning of the market for GDS services and to assess the likelihood and significance of vertical cross-market effects, it is necessary to understand that the market is characterised by a platform facing multi-homing on one side and single-homing on the other (multi-homing/single-homing). (52)

(78) TSPs generally use multi-homing, as they have to distribute their content via all four GDSs in order to obtain the desired market coverage, whereas most TAs use single-homing, as one GDS suffices in most cases to provide them with the required TSP content. (53)

(79) The two-sided GDS market contains a number of elements which are characteristic of multi-homing/single homing situations as described in economic literature. These elements are (i) a limited degree of product differentiation, (ii) asymmetries in network effects (network externalities are generated mainly on the TA side and GDS providers have to create demand on that side in order to have demand on the TSP side) and (iii) a distribution of prices and revenues skewed towards one side of the platform (GDS providers obtain profit only on the TSP side of the market and partially use this profit to offset net losses on the TA side).

(80) In this multi-homing/single-homing framework, the GDS providers are faced with bargaining on both sides of the platform, with different strengths and weaknesses. Given the fact that most TSPs use multi-homing, GDS providers end up offering relatively homogeneous products on the TA side. Therefore, TAs do not need more than one GDS provider to reach most of the available content. This means that GDS providers only compete for customers on the TA side of the platform (mainly by granting financial incentives to use a particular GDS). The customer base they obtain on the TA side of the platform (and the related positive network externalities) represents the main asset which GDS providers offer customers on the TSP side of the platform.

(81) As long as TAs use single-homing, GDS providers have exclusive access to TAs belonging to their respective TA networks. Each GDS provider therefore has a certain degree of monopoly power in relation to TSPs that need to reach the TAs exclusively connected to one GDS. This monopoly power allows the GDS provider to charge higher prices to TSPs. These "monopoly rents" extracted from TSPs are to a large extent used to cover the financial incentives granted to TAs.

6.2.1.3. Recent market developments affecting the relative bargaining power of TSPs, GDS providers and TAs

Introduction

(82) The Commission investigation has found that the bargaining interaction between GDS providers and customers on both sides of the market has started to change. TSPs and TAs have recently increased their bargaining strength relative to GDS providers. These changes in the relative bargaining power relate to (i) consolidation among TAs, (ii) the introduction of direct bookings via supplier.coms and (iii) surcharges imposed by TSP.

Consolidation among TAs

(83) Recent consolidation among TAs (54) in combination with competition between the GDS providers to maintain and extend their TA networks is one of the factors leading to considerable increases in the financial assistance paid by GDS providers to TAs. In response, GDS providers have attempted to offset the increased expenditure downstream by increasing revenues upstream in relation to TSPs.

Market developments: supplier.com and surcharges

(84) In response to the GDS providers' attempts to exploit further their gatekeeper positions (55), TSPs have made substantial efforts to improve their bargaining positions vis-à-vis GDS providers. TSPs have developed supplier.coms for two purposes. First, TSPs have developed supplier.coms as alternative distribution channels directly accessible to end-consumers (and in part to TAs (56)). By reducing distribution costs, bookings via supplier.com facilitate full service airlines' competition with LCCs. Second, supplier.com provides TSPs with a viable alternative distribution channel for travel content (for example, the lowest fares) that they may wish to withhold or threaten to withhold from the GDS providers. The introduction by TSPs of direct bookings via supplier.com has introduced an element of differentiation in terms of content made available selectively on one platform (supplier.com) and not on another (GDS), or via one but not all GDS providers.

(85) The increasing competition between full service carriers and LCCs may explain the substantial increase in the number of bookings via supplier.coms in recent years. TSPs withholding or threatening to withhold content may explain in part some shifts of market shares among GDS providers on the TA side. (57) This market evolution has subsequently forced GDS providers to revise their strategy towards TSPs and grant discounts in exchange for commitments from TSPs to provide "full content", that is to say, the same content as is available via the websites of the TSPs.

(86) The so-called "Participating Carrier Agreements ("PCAs") concluded by GDS providers and airlines have been complemented by a series of Full Content Agreements ("FCAs"). Galileo and Worldspan as well as other GDS providers, have entered into FCAs with a number of airlines. These agreements provide for significant discounts on GDS booking fees in return for a commitment from the airlines to distribute all public fares and associated inventories through that specific GDS, in particular fares which until then were available only through the airlines' supplier.com websites.

(87) The emergence of these discounts in exchange for full content reflects the changed bargaining positions of the TSPs and the GDS providers.

(88) It may be concluded that content has become the crucial element which determines the bargaining between TSPs and GDS providers. TSPs' development of supplier.com and the ensuing possibility to withhold (or threaten to withhold) content from the GDS providers has destabilised the pattern of rent extraction in the single-homing/ multi-homing framework as GDS providers must be able to provide full content (in particular the lowest fares) to TAs. If a GDS provider is unable to offer TAs full content, the GDS provider risks losing customers, which intrinsically weakens the GDS provider's bargaining position with respect to TSPs.

(89) Full-content agreements have led to a decrease in net booking fees of approximately 10% for certain airlines active in the EEA. (58) A further indication of the GDS providers' loss of relative bargaining power in relation to TSPs (and possibly to TAs) is the merging parties' declining gross margins. Galileo's gross margins for its Community operations declined from [...]* in 2004 to [...]* in 2006, resulting in a gross margin decrease of approximately [...]* in 2005 and approximately [...]* in 2006. Worldspan's margins for the EEA have been [...]* for the past [...]* years. (59)

(90) The market investigation also showed that some TSPs have been able to negotiate additional rebates to the ones provided under a full content agreement with their GDS providers. These rebates are granted under so-called "opt-in" contracts which supplement the existing full content agreements and are conditional to attaining a sufficiently large participation to the "opt in" scheme among TAs. This allows the GDS provider to recoup the costs of the rebate granted to the TSP by applying a variable "opt in" charge to TAs (60). The incentive for TAs to "opt-in" stems primarily from the fact that "opting out" would in some cases also imply a cost in terms of a surcharge imposed by the TSP, which may by far exceed the variable "opt-in" fee.

(91) The market investigation confirmed that the application of "opt-in" agreements in the EEA is currently still rather limited and seems to predominantly concern the United Kingdom and Irish markets. Nonetheless, this is an on-going evolution characterising the EEA market, in which the mere possibility of their application has an impact on the relative bargaining power of TSPs and GDS providers.

(92) Further, the market investigation indicates that "opt-in" schemes seem more successful with TAs located in the home market of the TSP. This is related to the higher brand recognition that any TSP tends to enjoy in its home market, which makes that content the most sought after by TAs. Apart from the threat to withhold travel content via supplier.com, TSPs have developed an additional tool to put pressure on GDSs. By applying, or threatening to apply surcharges to TAs, TSPs may influence the use of a specific GDS and make it lose volumes in favour of either supplier.coms (where no surcharges are imposed) or another GDS (61). An additional, indirect effect of the surcharge is to put pressure on the GDS providers to reduce the distribution costs charged to TSPs, by using as leverage, the position of strength which TAs have in their relations with GDS providers. GDSs are generally susceptible to pressure from TAs, as their network of TAs constitutes their main asset, and switching by TAs can significantly weaken their negotiation position vis-à-vis TSPs.

(93) The market investigation has confirmed these three scenarios for the EEA. First, TSPs could impose surcharges on TAs to stimulate bookings via their supplier.com, deviating volumes from GDSs. A good example of this is the current practice of Brussels Airlines, which reportedly imposes a surcharge on Belgian TAs making certain low value bookings via a GDS rather than via supplier.com. (62) The direct effect of the surcharge in this case is one of deviating volumes from GDS bookings to supplier.coms.

(94) A second reason why TSPs may consider applying surcharges is connected to the "opt-in" agreements. In order to encourage TAs to "opt in", TSPs can impose surcharges on TAs which "opt out". The per segment surcharge will always be higher than the per segment opt-in fees paid by the TAs. The market investigation showed that even the threat of a surcharge may be effective, since, for instance in the case of the United Kingdom, British Airways has been able to convince most TAs to opt in. It seems likely that the use of opt-in contracts may further spread to other Member States. The direct effect of the surcharge, in this case, is to influence the choice of whether or not to opt in as well as the choice of the GDS to be used for bookings, in view of the possible differential in the magnitude of the surcharges and of the differential between surcharges and opt-in fees.

(95) A third reason why TSPs could apply (or threaten to apply) surcharges is the indirect impact on negotiations between TSPs and GDSs. If a GDS is unwilling to improve its terms of contract, the TSP can credibly threaten to impose a surcharge on TAs that would have the potential to let the GDS lose volumes of GDS bookings for that airline, and eventually even induce the TA to switch to another GDS. This puts pressure on the GDS with which the TSP is negotiating and thereby weakens its bargaining position vis-à-vis the TSP. Although the market investigation did not provide concrete examples for the EEA where such surcharges were applied or used as a negotiating tool, the market investigation did confirm that such a practice is increasingly common in the US The use of a surcharge, or threat thereof, in these circumstances could therefore prove to be an efficient negotiation tool to put pressure on GDSs and induce them to offer better terms for all or certain classes of bookings.

Impact of the merger

(96) The in-depth investigation confirms that TSPs are capable of forcing GDS providers to lower their prices in exchange for full content, or alternatively, in order to avoid surcharges being applied on the TAs they have under contract.

(97) In this new market environment, the reduction in the number of GDS providers is unlikely to lead to price increases as a result of "vertical cross-market effects" for the following reasons.

(98) On the TSP side, the in-depth investigation confirms that TSPs are capable of forcing GDS providers to lower their prices in exchange for full content, or alternatively, in order to avoid surcharges being applied on the TAs they have under contract. Airlines in particular have developed and keep on developing a number of bargaining tools (especially, but not only, supplier.coms) allowing them to keep part of their surplus in the negotiation with GDS providers. Even in a situation with only three GDS providers none of them will be able to increase prices because TSPs will maintain sufficient bargaining power, based on (i) the capacity to channel bookings towards the supplier.com websites, (ii) the surcharges imposed on TAs, (iii) the brand recognition in the home market(s) and (iv) the possibility to develop new bargaining tools in the future. (63) Therefore, a reduction in the number of GDS providers from four to three does not increase the likelihood of unilateral price increases as a result of "vertical cross-market effects".

(99) This conclusion also applies to other TSPs, such as rental car companies and hotel chains. Firstly, their dependency on GDSs for the distribution of their travel content is much lower than in the case of airlines (64). This implies that any potential negative effect deriving from this transaction can only affect a limited part of their business. Most of the rental car companies and hotel chains have well developed supplier.coms. Some of them have also established direct links (65), which allows them to bypass GDSs. One of the rental car companies stated that it had been offered a "full content" agreement by one of the GDSs although, in the end, it rejected it (66). This would confirm the great value which GDSs also attach to "content" of the other TSPs and reflects the leverage these TSPs may have vis-à-vis GDSs. One major rental car company confirmed that certain content is currently only available via its supplier.com (67). Although one major rental car company foresees that the merger will lead to increased prices, this view is not generally shared by the other rental car companies or hotel chains (68). Therefore the overall conclusion drawn for airlines also applies to other TSPs, such as rental car companies and hotel chains.

(100) On the TA side, a sufficient number of GDS platforms will remain available to TAs (69) and as set out in section 6.2.3.3., switching costs are not an insurmountable obstacle to choosing another GDS provider. The fact that GDS providers need to create and maintain a sufficiently broad network of TAs in order to generate demand on the TSP side leaves TAs in a favourable bargaining position vis-à-vis GDS providers even after the elimination of one of them.

(101) Taken together, these elements (effective bargaining power of the TSPs, and ongoing or possible development of additional bargaining tools) suffice to counter the potentially detrimental effect of the merger in terms of the potential occurrence of vertical cross-market effects as a consequence of the reduction from four to three GDS providers.

6.2.2. Elimination of Worldspan as a "price maverick"

(102) A second theory of harm investigated by the Commission relates to Worldspan allegedly acting as a "price maverick" in the EEA, charging lower prices than its competitors (Galileo, Sabre and Amadeus). During the market investigation, concerns were expressed that, following the loss of competition between the merging undertakings, Worldspan's prices would be increased and aligned with those charged by Galileo.

(103) The results of the Commission's in-depth investigation show, however, that this theory of harm cannot be upheld. In order to conclude that the merger would be likely to lead to significant post-merger price increases by Worldspan it would have to be shown that, pre-merger, Worldspan's prices are significantly lower than those of its competitors , in particular Galileo, and that the merging parties would have the incentives and ability to increase Worldspan's prices post-merger.

6.2.2.1. Worldspan is not charging lower prices

(104) It is difficult to make direct price comparisons between GDSs due to the complexity of the product offerings and the pricing structures. It appears however, that (i) "net average price per booking" (all kind of bookings included) and (ii) "net price (list price less discounts) for active (70) segments" are the most useful proxies to compare prices and analyse price evolution. "Net average price per booking" takes into account all booking alternatives and the related price differences whereas "net price for active segments" is the most significant standard booking (71), in terms of both number of bookings and GDSs' revenues.

(105) In order to make an analysis of the "net price for active segments", the notifying party submitted a comparison of the merging parties' most basic types of booking for 2006: Galileo's "Active Net Segment" with Worldspan's "Full Service (72)". This shows that Worldspan's list price is actually [...]* than Galileo's list price for [...]* types of "Full Service" booking alternatives:

<table>

(106) If Worldspan's fees for each booking category are weighted as an average across all bookings according to the relative weight of each of the four categories of Worldspan pricing under its "Full Service" functionality level, the result is USD [...]*, while Galileo's Active Net Segment price for 2006 is USD [...]*.

(107) Moreover, the notifying party submitted a comparison of prices actually charged by Galileo and Worldspan over a three year period for six large EEA airlines to show that, for comparable transactions, Worldspan's prices are not consistently lower than those charged by Galileo. The comparison is based on Galileo's Active Net Segment fee and Worldspan's Full Service fee.

<table>

(108) The Commission has asked the most relevant TSPs (both in terms of number of bookings and revenues) if they consider that Worldspan's prices have been generally lower than those of competing GDSs in the EEA over the last five years (73).

(109) Most of the respondents to the Commission's in-depth investigation disagree with this hypothesis. For instance, British Airways, Alitalia and SAS explicitly confirmed that Worldspan has not been consistently lower in price. US airlines, such as Delta, also regard Worldspan as more or less comparable in pricing to the other GDSs.

(110) The data provided by the TSP respondents to the in-depth investigation allowed the Commission to create the following chart (74) on the evolution of the net price per booking per GDS.

<table>

(111) As can be observed from the chart, [...]* GDS services were lower priced than [...]* in 2003, 2004 and 2005. This trend is expected to continue in 2007. As regards [...]*, its GDS services were lower priced than [...]* in 2001 and 2002. However, [...]* GDS services are, in most cases, invoiced in euro. Therefore, when considering the favourable exchange rate between the euro and the US dollar in recent years, (75) [...]* becomes more expensive.

(112) If, instead of comparing net price per booking, the comparison is made on the price per Active bookings/segments (the most significant for TSPs in terms of both the number of bookings and the total amount), the results confirm that other GDSs are cheaper or price their products very close to Worldspan, as shown in the following chart (76) .

<table>

(113) Worldspan's prices for active segments in the home market (home country/domestic region) are similar to those of [...]*. As regards others markets, [...]* appears to be the cheapest, Worldspan and [...]* having similar prices.

(114) Some of the TSPs active in the rental car and hotel sectors confirmed that, although the fees charged by Worldspan were traditionally lower than those of competing GDSs, Worldspan has already started to increase its fees more than the other GDS to catch up with the level of fees applied by the latter.

(115) It can therefore be concluded that although the in-depth investigation has not provided evidence that Worldspan is more expensive than Galileo, it has convincingly shown that, in most cases, Worldspan is not the lowest priced GDS for TSPs. Generally, there is a cheaper alternative present on the market.

6.2.2.2. Worldspan has lost market share

(116) Another reason why, according to the notifying party, Worldspan does not qualify as a price maverick, is because its alleged low pricing policy has not allowed Worldspan to aggressively expand its market presence. The notifying party argues that, on the contrary, as the smallest GDS in the EEA for more than five years, Worldspan's market share in the EEA has not shown signs of growth.

(117) The evolution of Worldspan's market share between 2003 and 2006 shows a decrease of [0-5%]* in the upstream market (EEA). In relation to the downstream market, Worldspan's market shares have remained relatively stable, with annual average increases/decreases of around [0-5%]* or less, with the exception of Hungary, where a sharp growth took place between 2004 and 2005. The reductions in the national market shares have been higher in those Member States where Worldspan was historically strongest, such as The Netherlands or the United Kingdom.

<table>

(118) In 2007 Worldspan is likely to drop to below [0-10%]*, following the recent loss of its two main customers in the EEA, [...]* and [...]*, to Amadeus and Sabre (77).

(119) Therefore, contrary to what one would expect from a company which is alleged to be a maverick, Worldspan's markets shares does not show general signs of growth.

(120) Finally, according to the notifying party, Worldspan can not be considered as a price maverick in the EEA, since it acts rather as a price taker than as a price setter. The notifying party refers inter alia to the fact that other GDSs were first in the EEA in concluding full content agreements with [...]* EEA airlines (78).

(121) This is confirmed by the results of the in-depth investigation. British Airways indicated that in the EEA, Amadeus has led changes in pricing models, which were then followed by some of the other GDSs. As an illustration, British Airways refers to the introduction in 2005 of the "Home Market" concept, which entails a different, lower booking fee for the home market of the TSP. This pricing structure which leads to lower prices for GDS services provided in the home market of the airline was copied by Worldspan in 2006. More generally, British Airways made reference to the fact that Worldspan tends to issue its price increases after the other GDSs publish their prices.

(122) Lufthansa also confirmed that Worldspan has reacted to the steps taken by the other three GDSs in recent years, as regards its pricing policy. Lately, Worldspan has also followed its competitors in terms of ticketing fees (which were previously included in the booking fee). Lufthansa added that Worldspan always presented itself as a follower of the other GDS in this matter.

6.2.2.3. Galileo/Worldspan are not each other's closest competitors

(123) The notifying party considers that the scope for price increases by Worldspan post merger is further decreased by the fact that Galileo and Worldspan are not each other's closest competitors on the EEA market.

(124) The in-depth investigation confirms that Galileo is generally perceived by TSPs as stronger in corporate travel, while Worldspan is stronger in leisure travel and on-line travel agencies (79). Further, for example, British Airways considers that Worldspan has qualities and functionalities which set it aside from other GDSs (for example, in the automation of its fare filing and fare quoting systems) whereas in other respects its functionality is considered to be weaker (for example, in its agency front-end user interface). Downstream, the vast majority of TAs consider Amadeus as the closest competitor of both Galileo and Worldspan (80).

6.2.2.4. The Commission considers it unlikely that Worldspan would increase its prices postmerger and align them to Galileo's prices.

(125) The fact that the merging parties are not each other's closest competitors reduces the likelihood that Worldspan's prices would increase post-merger. Moreover, the decreasing pre-merger margins of the merging undertakings indicate that the scope for higher post-merger prices is very limited. In this regard, the figures provided by the notifying party show that Galileo's gross Community margins declined from [...]* in 2004 to less than [...]* in 2006. Moreover, Worldspan's margins for the EEA have been [...]* for the past [...]* years.

(126) The likelihood of higher post-merger prices upstream is further reduced by the fact that such a price increase would trigger a potential withdrawal of content from Worldspan's GDS by the TSPs or the imposition of surcharges by the TSPs on TAs which use Worldspan.

(127) British Airways (81) confirmed for example that, if a GDS reduced British Airways' discounts by 5% to 10%, it might threaten to withdraw BA.com fares, or impose surcharges on TAs which use the respective GDS. Delta Air Lines (82) confirmed that the threat to withdraw content is an effective means to obtain price concessions from GDSs.

(128) In summary, the results of the Commission's in-depth investigation show that there is insufficient evidence to conclude that Worldspan charges lower prices than its competitors and that it acts as a price maverick. As a result, it is unlikely that the transaction would lead to an increase in Worldspan's prices.

6.2.3. Non-coordinated effects on the downstream side of the market

(129) As regards the downstream side of the market, the transaction would lead to high market shares (above 40%) in six Member States, with significant increments. In those six Member States the market shares in 2006 ranged from [40-50%]* to [70- 80%]*.

<table>

(130) The size of the high combined market shares in those six Member States could potentially allow the merging parties to behave independently of their competitors and customers post-merger and exploit their commercial relationship with TAs.

(131) However, the in-depth investigation shows that the merger will not allow the merged undertakings to exert market power on the TAs in those national markets where the transaction will lead to high joint market shares.

6.2.3.1. Downward trend of Galileo's market share

(132) The notifying party argues that Galileo has lost [...]* market shares in each of the Member States where it traditionally had an important share, due to historical links with national carriers. The following table shows the negative evolution of Galileo's' market share in all six Member States as well as the positive evolution of Amadeus' market shares.

<table>

(133) According to the notifying party, the decrease in Galileo's market shares demonstrates that Galileo's above average market share does not reflect market power. The notifying party states that the transaction is unlikely to reverse the declining trend in Galileo's market share, especially given Worldspan's marginal role at EEA level and its documented inability to improve its market position on the EEA market over the years.

(134) The trend in Galileo's market shares as well as the marginal role played by Worldspan at EEA level have been confirmed by the in-depth investigation. This is in itself, however, not sufficient to conclude that the merger would not lead to non-coordinated effects in the downstream market.

6.2.3.2. Incentives paid by the GDSs to the TAs

(135) TAs are in general net cash receivers as they receive more financial incentives from GDSs than they pay as fees to the GDSs. Their incentives have consistently increased over the last five years, even in the Member States where the merging undertakings have high market shares (above 40%).

(136) The in-depth investigation showed that during the period from 2003 to 2006 TAs' revenues in general increased and the evolution of their gross margins83 was positive.

(137) This development shows the importance of TAs to GDSs and reflects the general view among the respondents to the in-depth investigation that competition between GDSs on the downstream market is strong. Every GDS faces high fixed costs combined with the low cost of processing additional bookings. This creates strong incentives for each GDS to maximize the number of segments processed through its system, and the revenues generated from such bookings. As a result, there is intense competition between GDSs for each TA contract to gain the revenues earned from incremental bookings.

(138) The strong position of TAs can be explained by the fact that (i) the GDSs depend on the TAs for the distribution of their content and (ii) TAs only need one GDS to obtain the necessary content for their customers (single-homing). These circumstances reduce the market power which high market shares may provide GDSs in certain Member States. High combined market shares are therefore unlikely to lead to price increases (reduction of incentives) to TAs.

6.2.3.3. Switching costs

(139) An additional reason why the transaction is unlikely to result in price increases on the downstream market is related to the fact, as confirmed by the in-depth investigation, that the switching costs are not of such a nature as to create an insurmountable impediment to actual switching.

(140) Although the quantification of switching costs is difficult, both in terms of time and training needed, as well as in financial terms, some general conclusions may be drawn from the in-depth investigation. Small TAs need one or more weeks to change from one GDS to another; the training needs are not significant and their productivity is not negatively affected by the change. However, big TAs estimate the time needed to switch as around 12 months, at a significant financial cost (more than EUR 1 million). Also the needs in terms of training are higher. In particular cases (for example, because of technical aspects of the change), the financial cost as well as the time needed may be even higher.

(141) Whilst the in-depth investigation confirmed that switching costs do exist, it also showed that they have not prevented significant switches in the past. In the period 2003 to 2006 several TAs moved from Galileo to Amadeus. In addition, Worldspan lost two of its major customers in this period, [...]* and [...]*.

(142) The notifying party argues that the average financial assistance per segment paid by Galileo increased by [...]* between 2001 and 2006 (from EUR [...]* in 2001 to EUR [...]* in 2006). (84) This shows that intense competition exists between GDSs and also underlines the relative market power of TAs. It can be concluded that switching costs are, in any case, not an impediment for TAs to change from one GDS to another.

(143) In this regard, the PhoCusWright study (85) concludes that [...]* of the participating TAs had changed GDSs at their last contract renewal, and in most of the six relevant Member States, this was as high as [...]* of the agencies interviewed.

(144) To a lesser extent, the current process of TA consolidation (86) can also lead to an increase in the number of TAs that use more than one GDS ("dual-automation"), thereby further facilitating switching.

(145) Even if some switching costs exist, it seems highly unlikely that the transaction would lead to any competition concerns due to the merging parties' high combined market shares downstream, given (i) the negative evolution of the merging parties' joint market shares, (ii) the incentive for GDSs to compete on national markets where they have a high market share downstream by the increase over time in incentives paid to TAs in such national markets and (iii) the general positive opinion of the TAs on the merger based on their assumption that the merger will create a strong alternative to Amadeus.

(146) The results of the in-depth investigation also show that, in general, TAs view the merger as bringing more competition to the market, given that the gap in terms of market power between Amadeus and the entity resulting from the merger will be reduced. Most of the associations of TAs consider the merger will have no negative effects on their market. Although the choice will be reduced from four GDSs to three, competition between Galileo and Amadeus will be stronger, according to the associations of TAs.

(147) It is, therefore, appropriate to take particular account of the following:

(i) the effective bargaining power of the TSPs and on-going or possible development of additional bargaining tools, which suffice to counter the potentially detrimental effect of the merger in terms of the reduction from four to three GDS providers as well as the potential occurrence of vertical cross-market effects,

(ii) the lack of evidence to conclude that Worldspan has charged lower prices than its competitors and that it has acted as a price maverick,

(iii) the incentives for the GDSs to compete on national markets where they have a high market share downstream and the general positive opinion of the TAs on the merger based on their assumption that the merger will create a strong alternative to Amadeus.

(148) On that basis it can be concluded that non-coordinated effects are unlikely to be created as a result of the merger.

6.3. Coordinated effects

6.3.1. Upstream

(149) The proposed transaction leads to a reduction in the number of globally active GDSs from four to three with Community market shares on the upstream market covering the relationship between TSPs and GDSs of respectively [20-30%]* (Galileo [20- 30%]* and Worldspan [0-10%]*) whereas Amadeus holds [50-60%]* and Sabre [10- 20%]*.

(150) This market situation justifies an assessment as to whether the concentration may give rise to concerns of coordinated effects in these markets. The Commission Guidelines on the assessment of horizontal mergers87 present the analytical approach used by the Commission in its appraisal of horizontal mergers.

6.3.1.1. Reaching terms of coordination

(151) Generally, the less complex and the more stable the economic environment, the easier it is for companies to reach a common understanding on the terms of coordination. In this context volatile demand, substantial internal growth by some firms in the market or frequent entry by new firms may indicate that the situation is not sufficiently stable to make coordination likely.

(152) Although no significant GDS market entry has taken place during the last five years, the evolution of market shares over the last five years confirms that the economic environment in which the GDSs compete in the EEA has been subject to considerable change. Whereas Galileo's market share on the upstream market for GDS services to TSPs dropped from [20-30%]* to [20-30%]* in the period from 2001 to 2006, Amadeus' market share increased from [40-50%]* to [50-60%]*. Sabre's market share decreased from [10-20%]* to [10-20%]*. Only Worldspan's market share remained relatively stable at [0-10%]* (88).

(153) Further, the growth of supplier.coms (in particular in the airlines' home markets) during the last five years should be taken into account as a destabilising factor in reaching terms of coordination in the GDS market. Estimates based on IATA data for 2005 show that the percentage of direct bookings worldwide as opposed to GDS generated bookings, for the top 20 EEA carriers, ranges between [...]* for Alitalia to [...]* for BMI (89). The market investigation further confirmed that most airlines foresee a further growth of their direct sales via supplier.com. For rental car companies and hotels the distribution of travel content by means of GDS has traditionally represented a relatively small part of their bookings. The notifying party states that for hotel accommodation approximately [70-80%]* of all bookings in the EEA are made through other sources than a GDS. (90) With respect to car rental services, the notifying party estimates that only [0-10%]* of these services in the EEA are booked through a GDS. (91)

(154) Even if what is set out in this section does not totally rule out the possibility of coordination between the three GDSs that would remain on the upstream market after the merger, these circumstances would render such coordination more difficult and thus unlikely.

6.3.1.2. Monitoring of deviations

(155) Only the credible threat of timely and sufficient retaliation prevents companies from deviating from terms of coordination. This requires markets to be sufficiently transparent to allow coordinating companies to monitor to a sufficient degree whether companies are deviating.

(156) Although the services offered by a GDS are of a rather homogeneous nature, the pricing structure and the product offerings of all GDSs are complex. Currently, in the EEA, GDSs apply different types of agreements in parallel, that is to say, standard Participating Carrier Agreements ("PCAs") and full content agreements, sometimes complemented by "opt-in" agreements. The differences in and variety of both the pricing structures and the product offerings in these agreements make sustained coordination impracticable.

(157) For example, Galileo has a standard model airline (PCA) agreement called "Galileo International Global Airline Distribution Agreement ("GIGADA") (92). This agreement applies [...]*. In the price list a distinction is made between [...]*, [...]*, [...]* and [...]*. Airlines can complement the standard services with additional services, to be chosen from an extensive menu of optional services offered by Galileo in GIGADA, which are all priced separately.

(158) Despite the fact that Galileo applies price lists on a uniform basis for all airlines, fees payable by airlines to Galileo pursuant to GIGADA vary significantly between airlines. This is because each airline's invoice reflects not only the standard services and specific optional services selected by that airline, but is also based on the number and type of transactions requested by TAs for those services on the Galileo GDS.

(159) In addition to the PCA agreement, Galileo has entered into full content agreements with a number of major EEA airlines (93). These agreements constitute amendments to GIGADA and are all individually negotiated between Galileo and the airline. [...]*.

(160) Galileo's full content agreements in certain Member States may further provide for an [...]*. (94)

(161) The market investigation confirmed that other GDSs apply equally complex, but differing, structures for their product offerings and prices. (95) This significantly reduces price transparency. (96) The transparency on the market that remains is further reduced by the fact that GDSs modify product offerings and price structures on a regular basis. (97) One respondent to the market investigation stated that the GDS providers frequently recruit staff from each other and that these recruitments increase market transparency. However, given the frequent price and product revisions of the industry, any transparency enhancing effects would be temporary. (98)

(162) Although it would appear from the market investigation that the contracts between GDSs and rental car companies and hotels are less complex in structure, they do not provide the level of transparency which would make coordination feasible. (99) In relation to rental car companies, Amadeus has a separate pricing for the European market, whereas the other GDSs apply one uniform tariff. Further different levels of connectivity may result in different fees (100) and some GDSs offer cash rebates if certain volume increases are achieved. As contracts are negotiated on an individual basis prices per booking differ from customer to customer (101).

(163) A number of respondents to the market investigation pointed to the existence of socalled "Most Favoured Nation" ("MFN") clauses in their agreements with GDSs. The use of these clauses may increase price transparency. MFN clauses oblige the contracting parties to offer to their counterpart equally beneficial terms as offered to third parties. It was alleged that the obligations following from the MFN clause would increase market transparency. It appeared from the market investigation and the interviews organised with TSPs that MFN clauses are indeed included in a number of GDS agreements with TSPs. It was confirmed, however, that in the large majority of cases the MFN clauses relate to the obligation on the TSPs to provide the GDS content parity, and are therefore merely a reflection of the obligations laid down in the Code of Conduct (102). The market investigation showed that MFN clauses on price parity are rare and seldom enforced. Moreover, [...]*, most contracts provide for the right to an audit carried out by an independent auditor, bound by confidentiality rules, which would limit the degree of price transparency that might result from such a clause.

(164) Considering the characteristics of the relevant markets, and in particular, the limited degree of transparency, it would be difficult for the three GDSs to monitor any potential deviation from planned coordinated action after the merger.

6.3.1.3. Deterrent mechanisms

(165) Coordination is not sustainable unless the consequences of deviation are sufficiently severe to convince coordinating companies that it is in their best interest to adhere to the terms of the coordination.

(166) Immediate retaliation through the reduction by the coordinating GDSs of the fees charged to travel content providers would be ineffective to retaliate against the deviating GDS, since it would not provoke switching behaviour by TSPs, considering that they need the services of all GDSs.

(167) A more realistic retaliatory measure would be for a GDS to offer certain important TAs who use the services of the deviating GDS higher incentives or direct lump sum payments in order to provoke a switch from the deviating GDS to a different GDS. Although such an action would be possible, it could be a costly strategy as the incentive offered to the TAs has to be sufficiently large to provoke the switch of GDS.

(168) It appears from the above that retaliation by means of increased incentive payments to TAs cannot be excluded (103).

6.3.1.4. Reactions of outsiders

(169) For coordination to be successful, the actions of non-coordinating companies and potential competitors, as well as customers, should not be able to jeopardise the expected outcome of the coordination. If coordination was intended to increase prices in the market, this would only hurt consumers if non-coordinating companies were unable or had no incentive to respond to such behaviour and render the coordinated price increase unprofitable.

(170) In this case there would seem to be significant competitive constraints present on the market which would destabilise any attempt at coordination. These constraints stem, in particular, from the possibility of TSPs to withhold content and make this content only available via a direct distribution channel such as supplier.com (104). In addition, coordination leading to price increases could provoke TSPs in the EEA to invest more in the development of alternatives to GDSs such as GNEs and direct links, which are currently still lagging behind in development in the Community as compared to the situation elsewhere, such as the US

(171) Considering the above and the fact that the criteria that have to be fulfilled in order to show coordinated effects are of a cumulative nature, the merger is unlikely to have coordinated effects on the Community market for the supply of GDS services to TSPs.

6.3.2. Downstream

6.3.2.1. Reaching terms of coordination

(172) The issue of coordination could, in principle, also arise on the downstream market which covers the relationship between GDSs and TAs. The downstream markets in the EEA are characterised by significant differences in market shares between the four GDSs on a country by country basis. These differences can partly be explained by historic reasons.

(173) Competition between GDSs on the downstream market is strong and does not show signs of coordinated behaviour. During the last [...]* Galileo and Worldspan have lost [...]* market shares to Amadeus (105). It has further been confirmed by the market investigation, and in particular the interviews with TSPs and GDSs, that competition between GDSs for contracts with TAs is currently intense, which is reflected by the increase in the incentive payments by GDSs to TAs over the last five years.

(174) The circumstances identified in this section suggests that the market positions of the merging undertakings on most downstream markets are relatively unstable, which would complicate reaching agreement on terms of coordination.

6.3.2.2. Monitoring of deviations

(175) The terms and conditions of travel agent contracts are generally non transparent as they are negotiated individually between TAs and GDSs. GDSs have no visibility into the complex terms and conditions offered by competing GDSs (106). Although some price transparency may exist, as a result of a possible exchange of information between TAs and GDSs in the context of contract negotiations, the fact that most contracts are individually negotiated would significantly limit the degree of transparency that might result from this. Therefore, the possibilities to successfully monitor coordinated behaviour seem very small, as it would require monitoring of the level of content, functionalities, services, financial assistance, bonuses and other terms and conditions that each GDS offers to individual TAs.

(176) Considering the limited degree of transparency it would be difficult for the three remaining GDSs to monitor any significant deviation from planned coordinated action.

6.3.2.3. Deterrent mechanisms

(177) The deterrent mechanisms that can be applied are essentially the same as the ones discussed in relation to the upstream market.

6.3.2.4. Reactions of outsiders

(178) It is true that in the event of coordinated behaviour, there are not many readily available alternatives for TAs to turn to. The use of supplier.com is for many too burdensome for the moment and the other alternatives to GDSs are currently not well developed enough in the EEA to provide an adequate alternative.

(179) However, considering that the criteria that have to be fulfilled in order to show coordinated effects are of a cumulative nature, it can be concluded that the merger is also unlikely to have coordinated effects on the downstream market.

7. CONCLUSION

(180) For the reasons set out above, it is concluded that the proposed concentration would not significantly impede effective competition in the common market or in a substantial part of it. The concentration should therefore be declared compatible with the common market and the EEA agreement,

Has adopted this decision:

Article 1

The notified concentration whereby Travelport LLC acquires sole control within the meaning of Article 3(1)(b) of Regulation (EC) No 139-2004 of the undertaking Worldspan Technologies Inc. is hereby declared compatible with the common market and the EEA Agreement.

Article 2

This Decision is addressed to:

Travelport LLC

400 Interpace Parkway, Building A

Parsippany, NJ 07054

USA.

Notes

1 OJ L 24, 29.1.2004, p. 1

2 OJ C ...,...200. , p....

3 OJ C ...,...200. , p....

4 Blackstone: EUR 23 755 million (2006); Worldspan: EUR 700 million (2006).

5 EUR [...]* (2006).

6 EUR [...]* (2006).

7 Turnover calculated in accordance with Article 5(1) of the Merger Regulation and the Commission Notice on the calculation of turnover (OJ C66, 2.3.1998, p. 25).

* 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.

8 See Commission decisions in cases COMP/M.2197 Hilton/Accor/Forte/Travel Services/JV of 16 February 2001, COMP/M.2510 Cendant/Galileo of 24 September 2001 and COMP/M.2794 Amadeus/GGL/JV of 21 May 2002.

9 Airlines account for the largest number by far of bookings on the GDSs. According to the notifying party, the share of hotel and rental car bookings in relation to all GDS bookings in the Community may be estimated at approximately [0-10%]* for the four GDS providers. Notification, paras 148-149.

10 For the purposes of this decision, the terms "the TSP side" or "the upstream side" will be used interchangeably (together with "the TA side" and "the downstream side").

11 As a part of its market investigation, the Commission sent detailed questionnaires to full service airlines, low cost carriers ("LCCs"), car rental firms, hotel companies, travel agencies, associations of travel agencies, competing GDS providers and companies operating alternative distribution channels of travel content. Certain groups of respondents received several questionnaires. In addition, a number of in-depth interviews were conducted with airlines, airline alliances, competing GDS providers, travel agencies, associations of travel agencies and associations of business travellers.

12 More generally, following Evans (2003) ["Some Empirical Aspects of Multi-sided Platform Industries", Review of Network Economics, 2 (3), pp.191-209] n-sided platforms may be (i) "coincident" platforms when they offer substitutable products or services on the same sides, (ii) "intersecting" platforms when this is the case only for some (m < n) sides or (iii) "monopoly" platforms when they have no competing providers on any side. GDS platforms are coincident, 2-sided platforms.

13 With the notable exception of certain LCCs such as Ryanair and certain charter carriers that do not use GDS services at all. Certain regional carriers do not distribute their travel inventories through all GDS providers.

14 It should be noted that some TAs do not use the services of a GDS at all and that a minority of TAs use multi-homing. In exceptional cases, a TA group may subscribe to all GDS providers. However, even multi-homing TA groups, generally use single-homing in individual outlets (that is to say, each outlet only uses one GDS).

15 According to two European associations of travel agents, subscribing to a second GDS may increase the operational costs of a small or medium travel agent by 5% to 10%. However, beyond a certain scale of operation, multi-homing (at least dual-homing) may become interesting in order to reduce the risk of service disruption in case of temporary system failure of one GDS. In particular larger TAs may consider dual homing, as they may be able to generate the booking volumes which are needed to be eligible for the incentive payments from each of the GDSs concerned. Joint reply to Commission questionnaire by ECTAA (Group of National Travel Agents' and Tour operators' Associations within the EU) and GEBTA (Guild of European Business Travel Agents) of 14 June 2007.

16 TAs are net receivers in the sense that the incentive payments paid by the GDS provider to the TA are greater than the subscription fee charged by the GDS provider.

17 Travelport mentions the companies G2 Switchworks, ITA Software, Farelogix, Hitch-Hiker, Ypsilon and Dolphin. Notification, para. 101.

18 An alternative term used in the industry seems to be "Alternative Content Access Platforms" ("ACAPs"). See "GDS Alternatives", press briefing by the Star Alliance of 16 March 2006, annex 1 to Travelport's reply to the Decision pursuant to Article 6(1)(c) of the Merger Regulation of 10 May 2007. For the purposes of this decision, the term "GNE" will be retained.

19 Questionnaire reply by Farelogix, Inc. of 2 April 2007; questionnaire reply by Hitch-Hiker GmbH of 30 March 2007; questionnaire reply by Ypsilon.Net AG of 30 March 2007.

20 Questionnaire reply by Ypsilon.Net AG of 30 March 2007.

21 Travelport states that Galileo as well as other GDS providers deliver content to [...]*. Notification, para 101.

22 See questionnaire to TAs of 21 March 2007, questions 13 and 20.

23 See questionnaires to TSPs of 19 March 2007 and 21 March 2007, questions 10 and 24. In this context, Travelport has submitted to the Commission a press briefing by the Star Alliance of March 2006 in which the Star Alliance announces its intention to conclude agreements with several "GNEs". "GDS Alternatives", press briefing by the Star Alliance of 16 March 2006, annex 1 to Travelport's reply to the Decision pursuant to Article 6(1)(c) of the Merger Regulation of 10 May 2007.

24 Travelport mentions the following meta search engines: Travelsupermarket.com, flightcomparison.co.uk, farecompare.com and Kayak.com.

25 It should be noted that meta search engines in combination with supplier.coms could possibly act as a substitute to GDS-providers for very small TAs that do not subscribe to a GDS. However, booking travel services in this manner is cumbersome and time-consuming and lacks the overview, versatility and expediency of a GDS. The use of meta search engines in combination with supplier.coms, is therefore a poor and partial substitute to a GDS. TAs without GDS account for a very small portion of the total number of bookings.

26 According to the Notification, the company Navitaire has developed an industry standard interface by which TAs may access TSP inventory directly. [...]*. Notification, para. 100.

27 Due to this partial substitution, direct links are likely to be established only between large operators covering large booking volumes, for example, between the largest "flag carrier" and the largest TA in a given country.

28 The feasibility of this task depends on the size and concentration level of the market. The establishment of a network of direct links seems more plausible for very large TAs which are active in a large geographic area with few, large TSPs. In a more fragmented market, transaction costs would be higher.

29 "Translation software" which is able to "translate" commands, etc. of different booking tools into one single system is reportedly being developed.

30 See section 6.2.3 Large market shares of the parties on the downstream side of the market.

31 The introduction of direct links is a more viable option in the US - with its very large, homogenous domestic market dominated by seven large network carriers.

32 Compared to the homogenous US market, the EEA market is very fragmented since it consists of 30 heterogeneous countries, a very large number of legacy carriers and with most TAs operating on a national basis. The transaction costs for establishing direct links partially substituting the use of a GDS are therefore likely to be higher in the EEA than in the US

33 The proportion of direct sales varies between [0-10%]* and [60-70%]*. It should be noted that these figures concern the airlines' worldwide sales. The proportion of direct sales in the individual airline's home country may be higher. Travelport submission of 7 June 2007.

34 One respondent stated that because a supplier.com web-site is the company's own proprietary mainframe system, booking costs are lower via supplier.com than via a GDS.

35 Traditionally, airlines have paid commissions to TAs for each booking they generate. In later years, however, airlines have endeavoured to reduce or even abolish commissions. In order to compensate this loss of revenue, TAs charge end-users a service charge instead. Airlines have been more successful in reducing/abolishing commissions in their respective home markets. According to the market investigation, commission payments have decreased in the Community in recent years, from 9% in certain Member States to a situation in which commissions have disappeared altogether. Several major European airlines have recently abolished commissions to TAs in their home markets.

36 Managed travel denotes corporate travellers whose employers have outsourced their travel booking activities to TAs specialised in corporate travel, so-called travel management companies ("TMC"s).

37 Lufthansa argues that it is harder for Lufthansa than for British Airways to divert bookings to its website. This is due to differences in the airlines' networks. Whereas British Airways' activities are very much concentrated in London, Lufthansa operates multiple hubs in Germany, resulting in a larger proportion of complex bookings with connecting flights. Lufthansa reply of 30 March 2007 to Commission questionnaire.

38 However, certain airlines offer specific supplier.com websites for bookings by TAs, called Business to Trade ("B2T") sites. Joint reply to Commission questionnaire by the Federation of Tourist Industry - Belgium (FTI) and the Belgian Travel Organization (BTO) of 11 June 2007.

39 An example of an airline which has adopted this strategy is Aer Lingus. Minutes from interview with Aer Lingus, 10 June 2007.

40 Certain airlines however offer specific Business to Trade (B2T) web-sites.

41 Market share calculated by volume, that is to say, the portion of all booked segments made through a particular GDS. Booked segments are the different legs of a flight including stop-over, car rental services, hotel reservations, etc. A transaction made by an end-consumer at a TA normally includes several booked segments from different content providers.

42 Data submitted by the notifying party, during the pre-notification process.

43 American Airlines and United Airlines have indicated - in their respective responses to the Commission's (DG TREN) "Consultation paper on the possible revision of Regulation 2299/89 on a Code of Conduct for computerised reservation systems"- that the average booking fees paid in the US (all inclusive fee) for each of the four GDSs are substantially lower than the corresponding fees in the EEA. The Commission market investigation has confirmed these differences. The consultation paper is available at http://ec.europa.eu/transport/air_portal/consultation/ 2007_04_27_en.htm.

44 In recent years, other regulatory authorities, in Canada and the United States of America have reviewed their respective regulatory frameworks regarding GDS Both countries had regulatory frameworks applicable to GDSs which were similar to the Community Code of Conduct. The US opted for a total liberalisation, whereas Canada opted for a partial but still far-reaching de-regulation.

45 Commission (DG TREN) "Consultation paper on the possible revision of Regulation 2299-89 on a Code of Conduct for computerised reservation systems". This document describes developments of the last 20 years on the GDS market.

46 Council Regulation (EEC) No 2299-89 of 24 July 1989 on a code of conduct for computerized reservation systems, OJ L 220, 29.7.1989, p.1 as last amended by Regulation (EEC) No 323-99 (OJ L 40 13.2.1999, p.1).

47 Commission (DG TREN) "Consultation paper on the possible revision of Regulation 2299-89 on a Code of Conduct for computerised reservation systems".

48 Commission decisions in cases COMP/M.2794 Amadeus/GGL/JV of 21 May 2002, para. 17; COMP/M.2510 Cendant/Galileo of 24 September 2001, para. 13 and COMP/M.2627 Otto Versand/Sabre/Travelocity JV of 19 December 2001, para. 20.

49 Notification, annex 21.

50 Commission questionnaire to TAs of 21 March 2007.

51 See their respective we-sites: http://www.amadeus.com/amadeus/x7517.html and http://www.galileo.com/galileo/fr-be/contact/Europe.

52 This situation is identified in the economic literature as "competitive bottlenecks", see for example, Armstrong (2006) "Competition in Two-Sided Markets", Rand Journal of Economics, 37 (3), pp.668- 691.

53 Exceptions can be observed to this pattern especially on the TA side, as a relatively low number of large (and sometimes multinational) TA groups tend to subscribe to more than one platform and exceptionally to all of them. Still, in most cases multi-homing only arises at the group level, while at the level of individual TA outlets within the group, single-homing is largely prevalent. See submission by the notifying party of 7 March 2007 "Galileo/Worldspan: Analysis of TAs Switching and Multi- Automation" by RBB-Economics. Certain TSPs may decide not to provide content at all (for example, Ryanair), but whenever they provide it, they aim to maximise the network they can reach and therefore subscribe to all available GDS providers. Notwithstanding the limited exceptions to the general pattern, it can still be retained as accurately representing the functioning of the industry.

54 For example, Carlson Wagonlit Travel/Navigant and ProTravel, American Express/Rosenbluth, BCD/TQ3 and the Travel Company, Thomas Cook/My Travel and TUI/First Choice.

55 Each GDS provider controls the access to its own network of TAs.

56 Certain airlines operate specific Business to Trade ("B2T") websites. However, the use of supplier.com websites by TAs is limited by the time and costs necessary for multi-channel search, as compared to the GDS platform. This tends to limit their use by TAs as a simple complement to GDS (or a temporary solution to system failure for single-homing TAs). Supplier.com websites are mainly targeted at end consumers.

57 If a GDS provider is unable to provide TAs with the best fares, TAs are induced to switch to another GDS providing that content.

58 Answers to the questionnaire sent to TSPs on 25 May 2007.

59 Submission by the notifying party of 11 June 2007.

60 For a better understanding of the financial flows between TSPs, GDSs and TAs it should be stressed that there are two separate financial flows in the system. The first flow concerns payments directly made by the TAs to the TSPs for the travel service being purchased (for example, the flight, the hotel accommodation or the rental car) and the eventual surcharges which are due by the TAs to the TSPs. The second financial flow concerns the fees paid by the TSP to the GDS for the distribution of its travel content and the net payments by the GDSs to the TAs (for example, incentive payments, minus subscription fees and "opt-in" fees) for their use of that particular GDS.

61 Surcharges could appear prima facie as a price increase directly imposed by TSPs on TAs. However, as confirmed by the market investigation, the fact that average net prices did not considerably increase for TAs, rather indicates that TAs passed them on to GDS providers (by playing one GDS against the other and obtain more financial assistance) and points in the direction of surcharges having served as a further bargaining tool against GDS providers. "Opt-in agreements have mainly occurred as a result of major network carriers threatening to remove content (fares and/or availability) from the GDS or implement a surcharge for GDS bookings (over their own website). [Given that] the content was available prior to the action without a surcharge one can hardly associate them with a value added." Reply by CarlsonWagonLit of 20 June 2007 to Commission questionnaire (question 14). This can actually be seen as the carrier having "actively encouraged agents to use their trade online site". Reply by the Association of British Travel Agents Limited (ABTA) of 12 June 2007 to Commission questionnaire (question 17).

62 Brussels Airlines applies, since January 2006, a surcharge of EUR 2.50 per booking segment for the five lowest booking classes. Joint reply by the Belgian Travel Organization (BTO) and the Belgian Federation of the Tourist Industry (FTI) of 11 June 2007 to Commission questionnaire (question 15); reply by the Dutch Association of Travel Agents and Tour Operators (Algemene Nederlandse Vereniging van Reisondernemingen - ANVR) of 26 June 2007 to Commission questionnaire (question 15); joint reply by the Group of National Travel Agents' and Tour Operators' Associations within the EU (ECTAA) and the Guild of European Business Travel Agents (GEBTA) of 14 June 2007 to Commission questionnaire (question 15).

63 Airlines could possibly develop "joint" supplier.coms (for example, within the framework of airline alliances) that could compete with GDSs. Such solutions would have the advantage, as compared to mono-brand supplier.coms, of decreasing the costs of multi-channel searches and become more competitive with respect to the centralised solution represented by the GDS platform. Another possibility, mainly used in North-America for the time being, would be to unbundle the fares (that is to say, separating the fare for the pure travel from the fare related to ancillary elements such as, for example, for airlines, luggage, catering, web ticketing/check-in, etc). Such fare unbundling makes comparison via GDSs more difficult and therefore decreases the value for TAs of the GDS as a search and booking tool. Joint reply by the Group of National Travel Agents' and Tour Operators' Associations within the EU (ECTAA) and the Guild of European Business Travel Agents (GEBTA) of 14 June 2007 to Commission questionnaire.

64 In its submission of 19 June 2007, the notifying party states that the merged undertaking has a combined share of [0-10%]* in the EEA. The underlying calculations are based on (i) the total car rental revenues in Europe (EUR [...]* in 2005), (ii) an estimated average rental price per car rental booking (EUR [...]*), (iii) the number of car rentals booked via Galileo in the EEA in 2005 ([...]*) and (iv) the number of car rentals booked by Worldspan in the EEA in 2005 ([...]*). Even if the estimated average car rental price was increased to EUR [...]*, the parties' combined share of all car rental bookings in the EEA would be approximately [0-10%]*. It should be noted however, as confirmed by the market investigation, that for the bigger internationally operating rental car companies this percentage may be higher (between 20 and 30%).

65 Avis Europe has negotiated supply agreements directly with major travel agencies and airlines, which includes supply via XML connectivity. Reply by Avis Europe plc of 20 April 2007 to question 12 of Commission questionnaire to TSPs.

66 Reply to question 10 of the second Commission questionnaire to TSPs.

67 Reply to question 11 of the first Commission questionnaire to TSPs.

68 See reply by Avis Europe plc of 20 April 2007 to question 29 of Commission questionnaire to TSPs. Avis states that it does not believe "[the company] would be affected by a change in suppliers from four to three, in this case, given the spread of our business to the remaining GDS suppliers".

69 "[...]* the impact on TAs' position would not be too important, [...]* that the latter would still have the choice between 3 GDSs [...]*". Joint reply by the Group of National Travel Agents' and Tour Operators' Associations within the EU (ECTAA) and the Guild of European Business Travel Agents (GEBTA) of 14 June 2007 to Commission questionnaire.

70 According to information provided in the Notification, an active net segment charge is imposed on an airline on a monthly basis based on the number of bookings minus the number of cancellations made in relation to the airline's inventory in each month. A passive booking is created when a flight segment has been booked directly with the airline by the passenger or by a travel agent and the details need to be recorded for itinerary/ticket printing. A charge is imposed for a passive booking only when the passive segment is ticketed by the travel agent using the Galileo GDS.

71 According to figures provided by the notifying party, standard bookings represent [...]* of revenues and optional bookings [...]*.

72 According to information provided in the Notification, "Full Service" refers to the basic booking capability on the Worldspan GDS, including flight schedules with flight availability display, fare, and fare rules display, and ticketing capability for airline's flights.

73 Question 10 of the questionnaire sent to TSPs on 25 May 2007.

74 The sample includes the answers of 17 TSPs. However, not all of them have provided the Commission with figures for 2001 to 2008 (figures for 2007 and 2008 are estimates). Some of the TSPs have only provided figures for certain years.

75 See the web site of the European Central Bank (ECB) for the evolution of the USD/EUR exchange rate from 4 January 1999 (http://www.ecb.int/home/html/index.en.html )

76 The sample includes the answers of 13 TSPs (12 for Amadeus). The figures relate to 2006, the only year for which comparable figures are available.

77 Submission by the notifying party of 10 May 2007.

78 The [...]* airlines are [...]*.

79 Answers by TSPs to question 10 of the questionnaire of 25 May 2007.

80 Answers by TAs to question 11 of the questionnaire of 21 March 2007.

81 Reply by British Airways of 8 June 2007 to Commission questionnaire (question 16).

82 Reply by Delta Air Lines of 7 June 2007 to Commission questionnaire (question 17).

83 The gross margin is defined as advance payments + incentives + other revenues received by TAs from GDS providers less subscription fees and other costs paid by TAs to GDS providers.

84 Submission by the notifying party of 14 June 2006.

85 "Final Report Europe", prepared for Skadden, Arps, Slate, Meager & Flom LLP by PhoCusWright Inc,. 5 April 2007, as submitted by the notifying party.

86 Examples of travel agency consolidation include Carlson Wagonlit Travel/Navigant and ProTravel, American Express/Rosenbluth, BCD/TQ3 and The Travel Company, Thomas Cook/ MyTravel plc and TUI/First Choice.

87 OJ C 31, 5.2.2004, p. 5.

88 Notification, annex 19, table 1, "Net bookings and Market shares by GDS in the EEA (2001-2006)" (based on MIDT data).

89 Submission by the notifying party of 7 July 2007. According to this submission the percentage of direct bookings for the top 20 airlines located in the EEA lies between the figures of Alitalia and BMI (e.g. Lufthansa [...]*, Air France [...]*, Iberia [...]*, SAS [...]*, Austrian Airlines [...]*, KLM [...]*, and SN Brussels [...]*). It should be noted that these percentages may underestimate the actual importance of direct bookings as these data relate to 2005 and do not include the generally predicted further growth for 2006 and 2007. This is confirmed by the fact that the recent figures submitted by the airlines in their responses to the Commission questionnaires suggest higher percentages for direct sales.

90 Submission by the notifying party of 11 June 2007.

91 Submission by the notifying party of 19 June 2007.

92 Notification, paras 210-227.

93 These airlines include: [...]*. Notification, footnote 55. Worldspan has full content agreements with [...]*. Notification, footnote 59.

94 [...]*. Notification, para. 216.

95 Reply by Amadeus of 19 April 2007 to Commission questionnaire (question 10) and reply by Sabre of 20 April 2007 to Commission questionnaire (question 10). Reply by Amadeus of 15 June 2007 to Commission questionnaire (questions 4, 8 and 9) and reply by Sabre of 26 June 2007 to Commission questionnaire (questions 4, 8 and 9). Amadeus argues in particular in response to question 8 of the second questionnaire that it "designs its pricing towards airlines on a segmented approach, where generally different booking types are subject to different fee levels. We believe that the concept of a single fee for all bookings is no longer adequate in view of the business challenges faced by commercial airlines today. As a result the value and complexity of the distribution services we deliver to airlines differ from one reservation to another. In response to this situation, we decided in 2004 to adapt the framework of our booking fee structure in order to better reflect the value and complexity of the sales we deliver to all our participating carriers [...]*."

96 Lufthansa states that "price transparency is poor because, pricing categories are frequently changed, pricing structure is too complex, invoice verification nearly impossible, unclear notification of the GDS in case of new pricing elements, and new pricing schemes take months until reflected correctly in the invoices". Reply by Lufthansa of 30 March 2007 to Commission questionnaire (question 26). Brussels Airlines considers that "the complexity of the pricing structure does not allow price transparency as too many add-ons are necessary to run the business efficiently". Reply by Brussels Airlines of 13 April 2007 to Commission questionnaire (question 26). Certain airlines argue in general terms that the GDSs have very similar prices and pricing structures. However, this argument is contradicted by the fact that most airlines negotiate individual contracts, the terms of which may significantly differ from each other, taking into account the relative strength and importance of the airline concerned and its ability to shift bookings away from GDSs.

97 The notifying party states that GDS list price revisions generally consist of a complex set of changes. The price may increase for some functionalities or transaction types and may decrease for others. For example, in 2004 Galileo increased list prices for (active, open and passive) ticket coupons in Europe, while its list prices for interactive sell and interactive display decreased and other prices remained unchanged. Further, in some cases price changes are not limited to a modification in the amount charged for existing pricing segments, but rather concern a complete overhaul of the pricing structure. [...]*. Notification, para. 231. Delta Air Lines considers that the transaction would have a very limited effect on price transparency in the market. Reply by Delta Air Lines of 7 June 2007 to Commission questionnaire (question 16).

98 At the request of the Commission, the notifying party has submitted information on recruitments to/from Galileo, Worldspan and other GDS providers (submission of 6 March 2007). There are no indications that such recruitments to/from GDS providers are more frequent than in other industries.

99 One rental car company disagrees on this point. However, this view is not shared by other rental car companies or hotels. Europcar argues that, there is no transparency and that each of the providers are negotiating on an individual basis without access to information of others. Reply by Europcar of 30 March 2007 to Commission questionnaire (question 26). The lack of transparency in the relationship between hotels and GDSs is confirmed by several respondents. Reply by Louvre Hotels of 24 April 2007 to Commission questionnaire (question 26) and reply by Intercontinental Hotels Group of 11 April 2007 to Commission questionnaire (question 26).

100 In the case of Galileo, [...]*. Submission of the notifying party of 11 June 2007.

101 Galileo applies the following booking fees for 2007: [...]*. Worldspan's booking fees for 2007 are as follows: [...]*. Similarly, the fees that Galileo charged to hotels for 2007: [...]*. Submission by the notifying party of 11 June 2007. Worldspan's fees for Hotels are as follows: [...]*.

102 See Article 4.1 of the Code of Conduct and the submission of the notifying party of 1 June 2007.

103 Delta Air Lines considers the likelihood of coordinated effects to be very limited. Reply by Delta Air Lines of 7 June 2007 to Commission questionnaire (question 16).

104 This possibility also applies to other TSPs than airlines. At least some of them retain a certain part of their best content to their own websites. In addition, the other TSPs also seem to be open to new technology allowing them to (partially) bypass GDSs. AVIS Europe plc has negotiated direct supply agreements with major travel agencies and airlines, by means of XML connectivity which allows the travel agent to bypass the GDS entirely. Reply by Avis Europe plc of 20 March 2007 to Commission questionnaire (question 12). Finally, rental car companies and hotels are far less dependant for their distribution on the use of GDSs than airlines.

105 Notification, paras 186-192, annex 22 (Analysis of TAs switching and multi-automation, study by RBB Economics).

106 Amadeus argues that "there is no price transparency either at the travel services provider level or at the travel agencies level. GDSs negotiate contracts on an individual basis and there is no publicity regarding prices". Reply by Amadeus of 19 April 2007 to Commission questionnaire (question 26). Sabre submits that it "does not believe that there is a large degree of price transparency regarding GDS travel agency tariffs. While travel agencies negotiate with GDS providers regarding tariffs and, in the course of such negotiations, may indicate the rates being charged by other GDS providers, GDS providers cannot verify such claims, and have no other source that would provide their competitors' tariffs". Reply by Sabre of 20 April 2007 to Commission questionnaire (question 26).

Draft final report of the hearing officer in case COMP/M.4523 - Travelport/Worldspan

(pursuant to Article 15 of Commission Decision (2001-462-EC, ECSC) of 23 May 2001 on the terms of reference of Hearing Officers in certain competition proceedings - OJ L 162, 19.6.2001, p. 21)

On 15 January 2007, the Commission received from Travelport LLC a request for referral pursuant to Article 4 (5) of Council Regulation (EC) No 139-2004 (Merger Regulation). No Member State competent to examine the concentration under its national competition law (Austria, Germany, Italy and Poland) objected to the referral. Accordingly, the concentration was deemed to have a Community dimension pursuant to Article 4 (5) of the Merger Regulation and should therefore be notified to the Commission.

Subsequently, on 23 March 2007, the Commission received a notification of a proposed concentration by which Travelport LLC acquires, within the meaning of Article 3.1 (b) of the Merger Regulation, sole control of the undertaking Worldspan Technologies Inc. by way of purchase of shares.

After a preliminary examination of the notification, the Commission found that the transaction raised serious doubts as to its compatibility with the common market and the functioning of the EEA agreement. Accordingly, the Commission decided, on 3 May 2007, to initiate proceedings pursuant to Article 6(1) (c) of the Merger Regulation.

Travelport LLC was, upon request, granted the opportunity to review, on 15 May 2007, a key document in the file in accordance with DG Competition's "Best Practices on the conduct of EC merger control proceedings".

Following an in-depth market investigation the Commission concluded that the proposed transaction would not significantly impede effective competition in the common market or a substantial part of it and is therefore compatible with the common market and the EEA Agreement. Accordingly, no Statement of Objections was sent to the notifying party.

No queries or submissions have been made to me by the merging parties or any third party.

The case does not call for any particular comments as regards the right to be heard.

Brussels, 26 July 2007

(signed)

Karen Williams

Opinion of the Advisory Committee on Mergers given at its 153rd meeting of 2 August 2007 regarding a draft decision relating to Case COMP/M.4523 Travelport/Worldspan

Rapporteur : The Netherlands

1. The Advisory Committee agree with the Commission that the notified operation constitutes a concentration with a Community dimension within the meaning of Articles 1, 4(5) and Article 3(1)(b) of the EC Merger Regulation.

2. The Advisory Committee agree with the Commission's definitions of the relevant product markets as stated in the draft decision.

3. The Advisory Committee agree with the Commission's definitions of the relevant geographic markets as stated in the draft decision?

4. The Advisory Committee agree with the Commission's assessment that the transaction does not lead to any competition concerns due to non-coordinated effects, or coordinated effects in GDS services.

5. The Advisory Committee agree with the Commission that the concentration as notified does not raise serious doubts as to its compatibility with the common market in accordance with Article 10 (2) of the Merger Regulation.

6. The Advisory Committee agree with the Commission that the notified concentration should be declared compatible with the Common Market and with the functioning of the EEA Agreement in accordance with Articles 2(2) and 8(1) of the Merger Regulation and Article 57 of the EEA Agreement.

7. The Advisory Committee agree with the publication of its opinion in the Official Journal of the European Union.

<table>