EC, July 16, 2003, No 38-233

THE COMMISSION OF THE EUROPEAN COMMUNITIES,

Having regard to the Treaty establishing the European Community, Having regard to Council Regulation No 17 of 6 February 1962, the first Regulation implementing Articles 85 and 86 of the Treaty, (1) as last amended by Regulation (EC) No 1216-1999, (2) and in particular Articles 3 and 15(2) thereof, Having regard to the Commission decision of 27 July 1999 to initiate an investigation into the telecommunications sector, Having regard to the Commission decision of 19 December 2001 to initiate proceedings in this case, Having given the firms concerned the opportunity to make known their views on the objections raised by the Commission in accordance with Article 19(1) of Council Regulation No 17 and with Commission Regulation (EC) No 2842-98 of 22 December 1998 on the hearing of parties in certain proceedings under Articles 81 and 82 of the EC Treaty, (3) Having consulted the Advisory Committee on Restrictive Practices and Dominant Positions, Having regard to the report of the Hearing Officer, Whereas:

I. THE FACTS

A. INTRODUCTION

(1) High-speed Internet access is a key factor in the development of the information society in Europe. By making the Internet quicker, easier and more convenient to use, and by enlarging the range of applications and uses which it makes possible, highspeed Internet access contributes to the development of e-commerce and the emergence of new electronic forms of distribution, which lends it a crucial economic and strategic dimension. The spread of high-speed Internet access in Europe depends to a significant extent on digital subscriber line, or DSL, services being made available on the local access networks run by telecom incumbents. These local access networks, which link end-users to the general telecom network, are known as local loops. Commercial DSL service provision really got started in Europe in 1999.

(2) Against this background, the Commission decided, in July 1999, to launch an inquiry into the sector pursuant to the powers conferred on it by Article 12(1) of Regulation No 17, which focused in part on the provision of local loop access services and residential local loop services. (4) The inquiry's main aim was to find evidence of anticompetitive pricing for services using the local loop, of entry barriers and of predatory pricing.

(3) Further to information obtained from France Télécom and its subsidiary Wanadoo Interactive and evidence submitted by several of Wanadoo Interactive's Internet service provider (ISP) competitors within the framework of this sector inquiry, the Commission decided to take a close look at the prices which Wanadoo Interactive charged its residential customers in France for high-speed Internet access and launched own-initiative proceedings in September 2001. In addition to these proceedings, it sent several requests for information to France Télécom, Wanadoo Interactive and its competitors and conducted an inspection on the company's premises in April 2002. The Commission also sent Wanadoo Interactive two successive statements of objections, dated 19 December 2001 and 9 August 2002 respectively, to which Wanadoo Interactive replied on 4 March and 23 October 2002 respectively.

(4) The period covered by this Decision runs from January 2001 to October 2002 and coincides with a tremendous boom in the high-speed Internet access market in France. In less than two years, the number of subscribers to high-speed Internet access services increased fivefold from around 200 000 to more than one million. The main beneficiary of this growth was Wanadoo Interactive, a France Télécom subsidiary with a market share of more than [70- 80]* % at the end of the period.

(5) The facts set out below concern Wanadoo Interactive's strategy of below-cost pricing as practised between January 2001 and October 2002 on the high-speed Internet access market. These practices concerned two of Wanadoo Interactive's high-speed products for residential customers, Wanadoo ADSL and eXtense, both of which are based on ADSL technology, (5) which itself is a variation on DSL technology. (6) The following paragraphs provide a detailed analysis of the revenue generated by these services and the related overheads and clarify the method chosen by the Commission to determine whether the prices charged could be regarded as constituting predatory pricing, in the light of the following legal analysis. They also set out evidence which was found when the on-site inspection was carried out and which sheds light on Wanadoo Interactive's in-house pricing deliberations and its high-speed access strategy.

B. THE COMPANY

(6) Wanadoo Interactive is part of the France Télécom group. 99.9% of its capital is held by Wanadoo SA. France Télécom's shareholding in Wanadoo SA fluctuated between 70% and 72.2% during the period covered by this Decision. France Télécom is 56% state owned. Before the beginning of 2001, Wanadoo Interactive was known as France Télécom Interactive. (7) The group formed by Wanadoo SA and its subsidiaries ("the Wanadoo group") encompasses all France Télécom's Internet activities and its telephone directory business. The Wanadoo group's activities therefore include Internet access, general or theme-based portals, e-commerce, printed, Minitel and Internet directories and Internet services for business and industry. Within the group, Wanadoo Interactive covers operational and technical aspects of Internet access services in France, including the ADSL services discussed in this Decision.

(7) Basically, the Wanadoo group's activities consist of two major segments: access, portals and e-commerce on the one hand and directories and services for business users on the other. Between 2000 and 2002, the first segment's performance was particularly impressive. Two factors acted as a motor for growth: a substantial increase in the number of active subscribers and rising revenue from individual subscriptions as a result of the development of high-speed Internet access services.

(8) In 2000 the Wanadoo group recorded consolidated pro forma turnover of euro 1 111 million, EBITDA (8) of - euro 66.4 million and a net consolidated result of - euro 102 million. (9) Turnover on directories and services for business users amounted to more than euro 741 million in that year, i.e. more than two-thirds of the consolidated turnover of Wanadoo SA and its subsidiaries. (10) Income from the access, portals and e-commerce segment totalled euro 370 million. Turnover from Internet access accounted for 22% of total consolidated turnover. (11) While the access, portals and e-commerce segment recorded negative EBITDA of euro 285 million, directories and services for business users generated positive EBITDA of more than euro 225 million.

(9) The Wanadoo group realised an operating result (EBITDA) of - euro 63.534 million in 2001 and of + euro 28.5 million in the first half of 2002. The net consolidated result for the two periods amounted to - euro 193.2 million and - euro 38.5 million respectively. The contribution of the group's two main segments can be summarised as follows. Turnover on the group's Internet activities was euro 715 million and euro 516 million respectively (i.e. 45% of the group total) in 2001 and the first half of 2002; the segment's operating result for those two periods was - euro 359 million and - euro 112 million respectively. Directories and services for business users generated turnover of euro 848 million (54% of the group total) in 2001 and euro 401 million (43% of the group total) in the first half of 2002. The segment's operating result in these two periods amounted to euro 218 million and euro 125 million respectively.

(10) Wanadoo's consolidated accounts therefore showed significant losses in 2000 and 2001; the profits generated by the second segment did not completely offset the major losses recorded by the first. Better operating conditions for Internet access activities enabled the group to break even in the first half of 2002.

(11) The Wanadoo Interactive subsidiary, to which this Decision is addressed and which specialises in Internet access activities in France, recorded turnover of euro 402 million, an operating result of - euro 92 million (i.e. - 23% of turnover) and a net result of - euro 89 million in 2001. In the first half of 2002 it recorded turnover of euro 305 million, an operating result of euro 31 million and a net result of euro 26 million.

(12) At 30 June 2001 Wanadoo Interactive had 2.33 million active Internet subscribers in France. The Wanadoo group's UK subsidiary, Freeserve, which had been acquired at the start of the year, had 2.03 million subscribers in the United Kingdom. At 1 January 2002 Wanadoo had more than six million customers in Europe; by September 2002 it had more than seven million.

C. THE SERVICES

1. Introduction: Internet access in France

(13) Internet access services consist in an Internet service provider (ISP) offering Internet access via a terminal and a wide range of services such as navigation, email, downloading of files and applications, hosting of personal pages, user networking, etc. In the vast majority of cases Internet subscribers are connected to their ISP either via a telephone line, (12) by cable (in areas served by cable networks) or via a dedicated link. During the period covered by this Decision and at the time of its publication, Internet access via mobile terminals (13) accounted for only a very marginal share of the market.

(14) In the first half of 2001 the number of Internet subscriptions in France was estimated at more than 6.1 million. By 31 March 2002 this number was estimated at 7.72 million, 29% up on the previous year. By 30 September of that year there were more than 8.46 million active Internet subscribers in France. (14)

2. The services: high-speed Internet access for residential customers

(15) High-speed Internet access is a specific Internet access service for which access modes enabling high-speed traffic are required. The respective characteristics of low-speed and high-speed access are clearly identified by service providers and consumers alike.

(16) The main advantage of high-speed access over dial-up access is that it makes the Internet much more convenient to use by virtue of:

- upload and download speeds which are significantly higher than those offered by low-speed Internet access,

- the possibility of an always-on connection and being able to use the same access line for telephone calls or other communications (such as fax messages) while still linked up to the Internet. (15)

(17) From a practical viewpoint, residential users can currently obtain a high-speed Internet connection in two ways involving relatively minor differences in performance but different technical constraints: (16)

- either via a telephone line adapted to the needs of ADSL technology, with equipment (modem, filters) which separates voice traffic frequencies from data traffic frequencies, thus enabling the telephone and Internet to be used simultaneously. The shorter the distance between the subscriber's terminal and the distributor in which the operator's multiplexer is located, the better the performance; (17)

- or via cable with a modem linked to the cable operator's network. However, precisely because of the "cascade" structure of cable networks, saturation problems may arise where large numbers of subscribers in the same area access the network at the same time.

In both cases, the options offered by service providers reflect the risks inherent in the access mode in question in terms of the undertakings given on speed of access and service availability. (18) Although watertight speed guarantees are not always forthcoming, service providers generally undertook to deliver a download performance of 512 kbit/s (network to subscriber) and an upload performance of 128 kbit/s (subscriber to network) during the period covered by this Decision.

(18) To benefit from high-speed Internet access, residential users must have a telephone line or high-speed access cable and an ISP subscription enabling them to access the web. Consumers can opt to buy high-speed access separately from a provider such as France Télécom or a cable operator and to take out an Internet subscription from a specialist provider. Buying the service from two separate providers means they can cancel their subscription to one of the two components of unlimited access. Alternatively, consumers may prefer a single transaction in which they purchase a package comprising high-speed access and Internet service.

(19) As the operator of the local telephone access network known as a local loop, France Télécom occupies a special position on the market. The telecom incumbent offers the high-speed access line component only, leaving Wanadoo Interactive to market Internet access subscriptions, portals and contents. Since end-1999 France Télécom has offered an ADSL service targeted on residential customers known as Netissimo 1. France Télécom customers who have taken out a subscription to Netissimo 1 can buy Internet access from the ISP of their choice if they so wish.

(20) The main high-speed ISPs using ADSL technology during the period covered by this Decision were Wanadoo Interactive, Club Internet (T-Online France), Nerim, Easynet, Mangoosta, (19) Infonie, (20) Tiscali-Liberty Surf and 9Online. These companies started marketing ADSL right at the end of 1999.

(21) The main cable operators involved in providing high-speed Internet access are Est Vidéocommunication, France Télécom Câble, a France Télécom subsidiary which provides technical back-up for Câble Wanadoo, NC Numéricâble, Noos and UPC France. All these operators link sales of Internet access per se to sales of high-speed access. Noos subscribers also buy Noos's Internet service, UPC subscribers use Chello's, NC Numéricâble customers subscribe to AOL's and France Télécom Câble customers buy Wanadoo Interactive's. Internet access via cable modem was first made available at the end of 1997.

D. WANADOO INTERACTIVE'S "GENERAL PUBLIC" ADSL PRODUCTS AND CORRESPONDING REVENUE

(22) Wanadoo Interactive sells ADSL-based high-speed Internet access services and cablebased access services. The latter, which are marketed under the Wanadoo Câble trademark, are not covered by this Decision, which is solely concerned with Wanadoo ADSL and eXtense services.

1. Wanadoo ADSL et eXtense

(23) Wanadoo Interactive started providing high-speed Internet access under its own trademark at the end of 1999. Until January 2001 the company simply offered, under the "Wanadoo ADSL" brand name, its own high-speed Internet access service complementing France Télécom's Netissimo 1 service (i.e. provision of an ADSL line). Under these arrangements, Wanadoo ADSL customers pay a monthly subscription to France Télécom for the service, (21) the rental of the ADSL modem from France Télécom, as well as a subscription to Wanadoo Interactive as the ISP (i.e. FRF (French francs) 135 or euro 20.58 per month inclusive).

(24) In addition to this initial offering, since 8 January 2001 Wanadoo Interactive has offered a package targeted on the same customers. Since that date it has sold France Télécom's Netissimo 1 service and unlimited flat-rate Internet access as a package known as "eXtense" or "kit eXtense" (22) for a monthly subscription of FRF 298 or euro 45.42, taxes included. Subscribers who opt for this solution do not need to maintain a specific contractual relationship with France Télécom for the ADSL service per se because they simply buy a modem at a cost of FRF 990 or euro 150 (taxes included) instead of renting one from France Télécom. (23)

(25) The launch of eXtense boosted the development of the ADSL services marketed by Wanadoo Interactive. It will be seen from Table 1 that since the launch Wanadoo ADSL's customer base has expanded at a relatively slow rate in comparison with the rapid growth recorded by the eXtense package. However, in the second quarter of 2002 [...]*as a result of the French Competition Council's decision to take action against France Télécom for marketing the package in its shops. (24)

Table 1: Wanadoo Interactive' ADSL subscriber base (512 kbit/s)

<emplacement tableau>

2. Total and unit revenue from Wanadoo ADSL and eXtense products

1. Total revenue from services (2001)

(26) Wanadoo Interactive claims that in 2001 eXtense generated revenue from subscriptions of around euro [...]*, against euro [...]* for Wanadoo ADSL. (25) In the first half of 2002 eXtense generated euro [...]*; the corresponding figure for Wanadoo ADSL was euro [...]*.

2. Theoretical unit revenue from subscriptions (26)

(27) In the case of Wanadoo ADSL, the monthly unit charge invoiced to each customer during the period under consideration amounted to FRF 113 or euro 17.20 (taxes excluded). eXtense produces a theoretical monthly unit revenue of FRF 249 or euro 37.98 (taxes excluded). In both cases, this is recurrent revenue. In addition, customers who do not have a modem buy the starter pack when they take out a subscription. The sales price of the starter pack is not treated as proceeds in the following analysis, but losses on cost-price will be treated as non-recurrent net acquisition costs. Special deals on subscriptions or starter pack purchase prices during the period in question have been treated by the Commission as non-recurrent acquisition costs and not as losses on recurrent revenue.

3. Actual average unit revenue from subscriptions

(28) The month-on-month figures for actual unit revenue from Wanadoo Interactive's ADSL subscriptions differ from the monthly nominal subscription charge figures referred to in paragraph 27 for at least two reasons.

(29) First, average net monthly revenue from subscriptions is mechanically lower than theoretical nominal revenue because of the time-lag between registering customers as subscribers and the point at which their subscription starts functioning on the one hand and the entry in the accounts of the first corresponding subscription payments on the other. The time-lag between registering new subscribers and receipt of the first payments amounts to [...]*on average. As such, a new subscriber registered in any given month generates on average [...]* % of the theoretical revenue from an existing subscriber over the whole month. (27) The impact of this time-lag - which Wanadoo Interactive refers to as a "side effect"- is all the more significant where the subscriber base is growing rapidly, as Wanadoo Interactive's has been since the beginning of 2001.

(30) Second, [...]*in 2001,[...]*. (28) The resulting losses were evaluated in July 2001 at approximately FRF [...]*for the first six months of 2001, (29) i.e. the equivalent of FRF [...]* (euro [...]*) per month on average per Wanadoo ADSL or eXtense subscriber. At end-October 2001 these losses were evaluated at FRF [...]* for the first ten months of the year, (30) i.e. on average FRF [...]* (euro [...]*) per month and per ADSL subscriber. At the time Wanadoo Interactive estimated that, on average, [...]* % of subscribers had been affected [...]* since the beginning of the year.

(31) Wanadoo Interactive contested the accounting value of the aforementioned amounts with the Commission but was unable to provide a detailed evaluation of [...]*during the period in question. (31) Nor was it able, on a more general note, to explain the discrepancies between theoretical average turnover and average turnover as per the accounts, which would leave the so-called side effects unexplained. However, for the last five months of 2001 and the first six months of 2002 it is possible to approximate these unexplained discrepancies by comparing the average actual turnover which an ADSL subscriber represents against theoretical turnover corrected by the mechanical effects of growth in the subscriber base.

(32) These discrepancies are only partially explained by the purely mechanical phenomena underlying the difference in turnover described in paragraph 29. A comparison between actual average turnover and theoretical turnover (32) reveals an unexplained differential of approximately euro [...]* per month for the eXtense package for the period from 1 August to 15 October 2001, of euro [...]*for the period from 15 October 2001 to 15 February 2002 and euro [...]* for the period from February to June 2002. Any ex post assessment of the services in question must take those losses into account.

(33) From January 2001 to October 2002 the actual average revenue (taxes excluded) from subscriptions to Wanadoo ADSL and eXtense was therefore significantly lower than the average monthly theoretical revenue, mainly because of the two phenomena referred to in paragraphs 29 and 30. After agreeing on a method for calculating the unexplained differentials in turnover with the Commission, (33) Wanadoo Interactive eventually challenged the idea of taking these differentials into account on the basis that it was impossible to predict them. (34) However, the Commission has pointed out that Wanadoo Interactive takes account of this average monthly revenue as per the accounts in its own management indicators, as shown by several of the company's budget presentations. (35) A company document dated December 2001 indicates growth in average monthly turnover per subscriber from FRF [...]* (i.e. euro [...]*) in January 2001 to FRF [...]* (i.e. euro [...]*) in October 2001. (36) A document dated January 2002 refers to an average turnover in 2001 of euro [...]*per eXtense subscriber, i.e. [...]* % less than theoretical average turnover per subscriber. (37) A document from November 2001 indicates average turnover per subscriber of euro [...]*in 2001 and euro [...]* in 2002 for eXtense customers, i.e. still about [...]* % below the theoretical average turnover. (38) Hence Wanadoo Interactive cannot argue with any degree of justification that it was not aware of this type of differential or that its financial analyses did not contain any evidence thereof.

(34) Below we examine average theoretical revenue corrected to reflect losses in turnover unexplained by side effects and corresponding in part to compensation for dissatisfied customers. However, apparent losses relating to the time-lag between the registration of new subscribers and the entry in the accounts of the corresponding subscription payments will not be investigated further because they are solely attributable to rapid growth in the subscriber base.

4. Content-related revenue

(35) In addition to generating revenue from subscriptions and, where appropriate, starter pack purchases, ADSL customers represent a less direct source of revenue in the form of online advertising, e-commerce and audience fees. Although these products relate to activities distinct from Internet access, they are inseparable from it, and accordingly the Commission has incorporated the resulting revenue into its analysis. The corresponding figure for 2001 was estimated on an ex ante basis at euro [...]* (taxes excluded) per subscriber per month. Actual market trends produced a much lower result that year; around euro [...]*per subscriber per month. These rather disappointing results were confirmed before the start of 2002. (39) The more modest forecast for 2002 was between euro [...]*and euro [...]*per month and per subscriber. (40) The result for the first half of 2002 was approximately euro [...]*per month per subscriber. (41)

E. COSTS INCURRED BY WANADOO INTERACTIVE

1. Costs included in the analysis

(36) Traditionally, a distinction is drawn between variable (or operational) costs and fixed (or structural) costs. The 1982 French General Accounting Plan defines variable costs as those costs which fluctuate according to company output alone, it being understood that cost fluctuations are not necessarily directly proportionate to fluctuations in output. This definition is in line with that accepted by the Court of Justice of the European Communities, which defines these costs as varying "according to the quantities produced." (42) It should be noted that this definition does not require the charge to be made strictly at the same time as the good or service is produced.

(37) In addition to this first distinction, a second is made between direct and indirect costs. The 1982 French General Accounting Plan defines the direct costs of a good or service as the costs directly attributable to it (these are usually variable or operational costs) and the variable or fixed costs which may be unambiguously associated with those costs even if they transit into the cost accounts via analysis centres corresponding to company activities which are not exclusively dedicated to producing the good or service in question. Direct costs are frequently confused with variable costs but may also include fixed or structural costs, like the costs associated with fixed assets specifically dedicated to producing the good or service in question. Indirect costs, on the other hand, are costs which are not directly attributable to a given product, but which must be broken down according to formulas determined for the various products and which reflect, as closely as possible, the indirect causal relationships. Direct costs plus indirect costs equal full costs. However, full costs generally exclude non-attributable costs, such as general financial costs and extraordinary costs which are not linked to production of the good or service in question.

(38) Lastly, a distinction must be made, in cost analysis, between recurrent and nonrecurrent costs. With the framework of a service like a high-speed Internet subscription, recurrent costs are costs which arise on a periodic basis. In activities of this type, non-recurrent costs are costs which arise just once (or on a very occasional or accidental basis) per subscriber, for instance customer acquisition costs.

(39) Broadly speaking, Wanadoo Interactive identifies the following costs for its ADSL products:

- network costs;

- customer acquisition costs;

- other production costs.

(40) In 2001 those three categories accounted for [...]* %,[...]* % and [...]* % respectively of the costs of the two services combined. Acquisition costs account for [...]* % of the total costs of the eXtense service.

(41) Paragraphs 42 to 69 look at the three cost types and their characteristics in the light of the various cost categories referred to in paragraphs 36, 37 and 38.

2. Network costs

(42) For residential customers to have an end-to-end high-speed Internet access service, Internet traffic must be kept under control on three levels. Wanadoo Interactive therefore uses three different services supplied by France Télécom and based on the requisite network architecture to ensure Internet connectivity.

(a) First, Wanadoo Interactive uses a service known as IP/ADSL (ADSL access service), which is the wholesale equivalent of France Télécom's Netissimo 1 retail service. This service ensures the ADSL link between subscribers to France Télécom's telephone network and a broadband access server (BAS) which is generally located at regional level. This section covers the public switched telephone network's local access segment, i.e. the telephone line, local link distributor fitted with an Asynchronous Transfer Mode (ATM) multiplexer and an ATM backbone regional segment linking the multiplexer to a BAS. Wanadoo Interactive concluded the first IP/ADSL 1 access service contract on [...]*, just before launching its eXtense package. (43)

(b) The service marketed as the IP/ADSL routing service ("routing service" or "IP routing service") enables traffic to be routed and conveyed from BASs to a central site which acts as a service platform for Internet access subscribers. France Télécom connects the provider's platform to its network via a very high-speed connection at the end of which it has installed a router.

(c) Lastly, beyond its service provision point, Wanadoo Interactive must be able, as a matter of course, to link subscribers to all worldwide Internet networks, and for that purpose it has taken out an international connectivity contract with France Télécom.

(43) A distinction must be drawn between the Wanadoo ADSL product on the one hand and the eXtense service on the other. In the first case, users are directly linked to France Télécom, which provides them with the ADSL line (via its Netissimo 1 service), i.e. the first segment of the service referred to in paragraph 42. In this instance, Wanadoo Interactive covers the second and third services only, i.e. IP routing and international connectivity. In the second case, however, Wanadoo Interactive provides a continuum of services to its customer and uses France Télécom's wholesale services to cover the three segments referred to in paragraph 42. In this case, Wanadoo Interactive resells the technical service provided by France Télécom.

(44) Wanadoo Interactive has purchased IP routing and international connectivity from France Télécom since [...]*, and ADSL access since [...]*. In addition, Wanadoo Interactive has been linked to France Télécom by an international Internet connectivity contract since [...]*.

1. Access costs

(45) France Télécom invoices the ADSL access service by multiplying the number of ADSL subscribers by the basic charge calculated per subscriber line. Each individual subscriber generates a recurring subscription cost for this service (recurring direct variable costs).

(46) Three periods must be distinguished in this connection:

- between November 1999 and July 2001 the IP/ADSL access service was invoiced [...]*; the cost for Wanadoo Interactive was FRF [...]* (euro [...]*);

- between 1 August 2001 and 15 October 2002 the IP/ADSL access service was invoiced uniformly per customer, irrespective of the number of customer connections. The cost was lower than in the previous period (FRF 140 or euro 21.34 per subscriber);

- as of 15 October 2002, the charge for the service was reduced again to FRF 101.67 or euro 15.50 per subscriber on the basis of the same invoicing principles as in the previous period.

2. Routing costs

(47) Each customer of Wanadoo Interactive's ADSL services has an impact on the volume of traffic processed by France Télécom as part of the IP routing service, partly because of engineering rules and the statistical connection between the number of subscribers and the speed used, and partly because of the invoicing rules applied by France Télécom.

(48) Since early 2001 the service has been invoiced in several different ways. Initially, until August 2001, the charge was calculated [...]*. This [...]*, it being assumed that each subscriber used approximately [...]*per busy hour. During the second period, from August 2001 to March 2002, the service was invoiced as [...]* of [...]*per online subscriber. Lastly, during the third period, routing service invoices were drawn up on the basis of [...]*. In all cases, over the whole of the period covered by this Decision, invoiced traffic is very largely determined by the number of subscribers to ADSL services and represents a direct variable cost.

(49) The invoiced service is made up of several elements: a "connection" component, with installation charges and a monthly licence fee (44) and an IP bandwidth component for which a monthly fee is charged, corresponding to traffic between the ISP platform and final users. Until October 2002 there were two routing service variants linked to two separate pricing structures. The provider could subscribe to the regional routing service to supply customers located in the same region as its service provision platform. At the same time, it could subscribe to the national routing service for customers based outside the region. In the second case, a higher charge was made. Since October 2002 the price for the two types of routing has been the same.

(50) For the period from August 2001 to March 2002, the invoicing rules for bandwidth use were applied [...]* (45). For end-2001, [...]*. Various items of evidence discovered at Wanadoo Interactive in the course of the inspection suggest that from [...]*2001 bandwidth use per online subscriber was approximately [...]*, rather than [...]*. (46) Information from competitors and France Télécom itself show an underlying, regular growth in traffic in the first few months of 2002 to levels in excess of 40 kbits/s. [...]* measures taken by France Télécom since March 2002 report speeds of between [...]* and [...]* for Wanadoo Interactive between April and June 2002. (47)

(51) The invoices issued to Wanadoo Interactive by France Télécom for the routing service at the end of 2001 were drawn up [...]* (48). Connection rates were below Wanadoo Interactive's forecasts to autumn 2001, and average speeds per subscriber per busy hour invoiced by France Télécom were also below forecasts, varying between [...]* and [...]* kbit/s between August and September 2001 and between [...]* kbit/s and [...]* kbit/s between February and June 2002. (49) Actual costs as invoiced by France Télécom to Wanadoo Interactive for its routing service for the last five months of 2001 and early 2002 were therefore significantly below forecasts.

(52) The trend in routing service charges in 2002 was less favourable than anticipated. In October 2001 Wanadoo Interactive expected these charges to fall to euro [...]* per Mbit/s on average as from the start of 2002, [...]* % down on the then price. By April 2002 it had already revised this estimate in its target profit and loss accounts for eXtense and Wanadoo ADSL products up to euro [...]* per Mbit/s on average from the date on which the new charges came into force. Lastly, the new routing charges applicable to Wanadoo Interactive for its traffic in France were set in July 2002 at euro 448 per Mbit/s as from 15 October 2002.

(53) Since March 2002 France Télécom has invoiced Wanadoo Interactive [...]* (50) [...]* (51). France Télécom's decision to switch invoicing methods resulted in supplementary invoicing for [...]*. However, France Télécom invoiced Netissimo 1 and Netissimo 2 type traffic [...]*. For that reason, it is difficult to identify the exact proportion of the supplementary invoice accounted for by Netissimo 1 and the Commission has discounted the slight increase in costs per subscriber which appears to have resulted which, incidentally, is minor in scale. (52)

(54) The impact of the routing service costs associated with the connections described in paragraph 49 (setting-up of connection and monthly subscription) appears relatively marginal in comparison with other costs. These recurring connection rental costs will be excluded from the analysis of variable and full costs.

(55) The cost per ADSL subscriber of the routing service can be estimated at euro [...]* (FRF [...]*) for the period from January to July 2001, at euro [...]* (FRF [...]*) for the period from August to mid-October 2001, at euro [...]* (FRF [...]*) for the period from mid- October 2001 to February 2002 and at euro [...]*for the period from February to October 2002. (53)

3. The costs of international connectivity

(56) The purpose of the international connectivity service is to convey all traffic generated by Wanadoo Interactive subscribers when navigating outside Wanadoo Interactive's platform. Although, in contrast to the other two services already described, this service is not specific to the provision of high-speed Internet access, there is no denying that each subscriber to Wanadoo ADSL and the eXtense package generates a certain volume of traffic and that this traffic is added to the total download requested by Wanadoo Interactive. In its in-house forecasts, the company includes these costs in the network costs of its ADSL services. (54) The fee paid by Wanadoo Interactive to France Télécom is a variable cost because it depends on the number of Wanadoo Interactive subscribers.

(57) The costs associated with international connectivity, which were stable in 2000 and 2001, fell at the beginning of 2002 following [...]*. As such, it can be estimated that the cost of international connectivity per ADSL subscriber has declined from euro [...]*to euro [...]* on average since February 2002. (55)

(58) However, it should be noted that the cost assessment of international connectivity at end-2001/beginning-2002 referred to in paragraph 57 is based on assumptions which are favourable to Wanadoo Interactive. It is assumed that international connectivity traffic per subscriber is equal to [...]* of invoiced traffic per subscriber for the routing service.[...]*, France Télécom's international connectivity service is invoiced on the basis of the download actually measured and observed at point of presence (POP level),. [...]* (56) [...]*Accordingly, applying a coefficient of [...]* % of theoretical bandwidth invoiced for the routing service in order to estimate international connectivity traffic results in actual traffic being underestimated. However, subject to that reservation, the Commission accepts the estimate provided by Wanadoo Interactive for the period up to April 2002 pending more accurate information on actual traffic.

(59) For the period after April 2002, the Commission based its cost assessment of international connectivity on a coefficient of [...]* % of bandwidth noted by France Télécom in April, May and June 2002. (57)

4. Refunds [...]

(60) [...]*. These sums should be deducted from the network costs actually incurred by Wanadoo Interactive in 2001 in the ex post analysis of costs.

3. Customer acquisition costs

(61) ISPs refer to costs incurred for activities designed to attract potential customers to services or to encourage them to take out subscriptions and costs associated with setting up subscriptions and services as acquisition costs (or customer conquest costs to use Wanadoo Interactive's terminology). These are one-off costs for each new subscriber, incurred at the point in time when the contractual relationship is entered into with the provider. Wanadoo Interactive identifies several items constituting customer acquisition costs: advertising, sales network fees, access costs, special offers and similar activities, the cost of the eXtense starter pack and other structural costs.

1. Advertising, marketing activities and special offers

(62) Advertising has a direct impact on the number of new subscriptions. In this instance, only advertising and marketing specific to Wanadoo Interactive's ADSL services were taken into account, to the exclusion of generic advertising on high-speed access or general advertising for Wanadoo or its services.

(63) Wanadoo Interactive does not accept that advertising should be regarded as a variable cost. In particular, it has invoked Commission merger decisions in which advertising costs are presented as sunk costs. (58)

(64) First, the fact that advertising costs are regarded as constituting sunk costs in the Commission decisions to which Wanadoo Interactive refers does not mean that they cannot be treated as variable costs. Not all sunk costs are necessarily fixed costs. (59) In addition, the advertising costs associated with a given product are generally regarded as direct variable costs for cost accounting purposes. (60) Even if advertising costs may be regarded as fixed costs in the very short term because they do not have an absolutely immediate effect on sales, they do have a clear impact in the short term - in any event less than a few weeks. This is particularly true in an expanding market characterised by information campaigns focusing on new products or on seasonal special offers. It might be different for general communication expenditure merely designed to publicise the company and its trade mark in general, as opposed to a specific product. In the case in point, the advertising costs are quite specific and correspond to successive campaigns which focus strongly on sales promotion. There is accordingly a strong correlation between information campaigns and trends in Wanadoo Interactive's sales, with a correlation coefficient of [>0,53]*. (61) The fact that, in chronological terms, advertising is the event which generates sales rather than a cost resulting from each sale taken individually does not alter this conclusion (62) any more than the fact that the cost of advertising campaigns is not necessarily directly proportionate to each product unit. (63) In this case the scale of the advertising and information campaigns was geared to the desired result, and as such the relevant costs were determined by a customer acquisition target.

(65) In practice, in addition to questioning the nature of the advertising costs, Wanadoo Interactive disputes their inclusion in a predatory pricing test. The Commission takes the view that this objection is irrelevant in the case in point. This is because, in so far as these costs determine the development of the relevant company's subscriber base in a growing market and, as such, the market power it acquires or reinforces, they represent an essential variable of the analysis.

(66) Special offers are motivated by the same customer acquisition logic as advertising campaigns. The related costs should be regarded as variable costs, particularly since by their very nature they are directly linked to the purchase of each new product unit sold (discounts, refund offers, etc.).

2. Payments to the sales network

(67) Payments to the sales network are also strictly proportionate to the number of new units sold. A commission for each new subscription is paid to the sales outlet (sales outlets forming part of the France Télécom network or other distribution points). [...]*. The average variable cost of paying sales outlets is therefore determined by [...]*.

3. Service access costs

(68) Service access costs are invoiced by France Télécom to Wanadoo Interactive for each new subscription to the eXtense service. For the period as a whole, these costs amounted to euro 53.40. However, between 15 October 2001 and 15 February 2002 and after September 2002, France Télécom exempted all ISPs from these access costs.

4. Other production costs.

(69) The other production costs identified by Wanadoo Interactive are: platform costs; customer service costs, the bulk of which are direct costs associated with the hot line service; customer administration costs, relating inter alia to invoicing and debt recovery; and customer loyalty costs. The Commission's analysis incorporates all the explanations provided by Wanadoo Interactive on the structure, breakdown and evolution of these costs and their classification as variable and fixed costs respectively.

F. THE EXTENT TO WHICH COSTS WERE COVERED BY REVENUE

(70) In this section the Commission explains and comments on various approaches that a test of predation might adopt, and the results arrived at when those approaches are applied to the revenues and costs considered in sections I.D and I.E:

- recovery of the full instantaneous costs as recorded in the undertaking's accounts;

- recovery of the full costs and of the variable unit costs as recorded in the accounts after they have been adjusted by spreading certain costs over several years;

- recovery of the variable costs for the different generations of subscribers separately;

- recovery of the costs which the undertaking would have foreseen ex ante, excluding any incidental or fortuitous components that may have affected initial forecasts.

(71) For each of these approaches the Commission begins its analysis in March 2001. (64) Although Wanadoo Interactive's ADSL services were launched commercially at the end of 1999, and made significant losses in 2000, the Commission takes the view that at that time the high-speed market had not yet developed sufficiently for a test of predation to be significant.

(72) The merits and conceptual limits of the different approaches are outlined below. At the end of the analysis the Commission finds that only the adjusted costs approach allows any valid conclusions to be drawn; the others can help to throw further light on the matter, but no more.

1. Recovery of full instantaneous costs

(73) The instantaneous costs and revenues considered in this approach are the costs and revenues arising over time as they are entered month by month in the undertaking's cost accounts. Over the period considered, the total revenue from and the full costs incurred in respect of Wanadoo Interactive's two ADSL services were at approximately the levels shown in Table 2.

Table 2: Revenue and full costs of ADSL and Wanadoo Interactive services (euro million)

<emplacement tableau>

(74) It will be seen that the rates of recovery of full instantaneous costs were low, at [...]*% and [...]*% in the two halves of 2001. In the first half of 2002 the rate was better. (65)

(75) There is an important limitation on the method of analysing the recovery of instantaneous costs which makes it impossible to draw conclusions on this basis. In an expanding market, where the costs of acquiring customers form a substantial proportion of expenditure, a firm operating normally cannot set out to recover its full costs immediately. Thus the rates of recovery shown in Table 2 do not by themselves prove that prices were predatory, and are given here only for information.

2. Recovery of adjusted costs

(76) This approach is based on the consideration that it is not the firm's objective to produce an instantaneous profit. Rather the firm will seek to achieve a level of recovery of recurrent costs (network costs and production costs) which is sufficient to ensure that the margin between revenue and recurrent costs will, within a reasonable time, also cover the non-recurrent variable costs invested in the commercial development of the particular product, on items such as advertising, promotion, marketing etc. The non-recurrent variable costs are accordingly adjusted and spread over a certain period in line with the principle of the depreciation of assets. (66) This method supposes that the firm seeks to secure a return on its investment within a reasonable time, rather than to recover all its costs at once. It may be that its prices will not fully cover its costs in the first few years of business, without driving off the market competitors with less financial stamina who are likewise investing with a view to reasonable profitability.

(77) To assess the economic equilibrium of Wanadoo Interactive's ADSL services, the Commission has spread the costs of acquiring customers over 48 months. A subscription to the eXtense service ties the customer for a year at least. But to spread the acquisition costs over just one year would not reflect ordinary market conditions, since on average subscribers stay with the same service provider for longer than that.

(78) Wanadoo Interactive has accepted this approach, though in its reply of October 2002 it says that in its view the average customer life is now something more like [...]*. (67) In its letter of 4 March 2003, Wanadoo Interactive maintains that four years is a "[...]*", and that a duration of "[...]* would not be unreasonable", (68) basing itself on customer turnover between August 2001 and June 2002. (69) For the reasons set out in its statement of objections of 19 December 2001, the Commission takes the view that the turnover rates observed over a period in which the clientele has been growing vigorously cannot be considered representative: the only such rates that would be useful in calculating the average customer life would be rates observed among a stabilised clientele.

(79) In their business plans Wanadoo Interactive's competitors likewise spread nonrecurrent variable costs over time, but the periods chosen are shorter than those used by the Commission in this case. (70) In its simulations of margin squeezes the French regulator, the Autorité de régulation des télécommunications (ART), also uses a duration lower than the four years applied here. (71) It should be pointed out, lastly, that in the period considered in this Decision the length of the period over which the nonrecurrent costs are to be spread is immaterial to the finding that the full costs are not covered, because the full costs are not covered even if depreciation takes place over an infinitely long time. (72)

(80) In this Decision the "adjusted unit costs" means the sum of the recurrent unit costs, recorded on a monthly basis, and the (non-recurrent) acquisition costs divided by 48. (81) In paragraphs 82 and 83 the Commission considers the recovery of adjusted variable unit costs, and in paragraphs 84 to 86 the recovery of adjusted full unit costs. The Court of Justice has accepted two tests of cost recovery, depending on whether the actions of the dominant firm formed part of a plan to eliminate competitors. (73) In the present case the Commission takes the view that Wanadoo Interactive's intentions are beyond doubt, (74) and that the results obtained by both tests ought to be considered.

1. Recovery of variable unit costs

(82) Under the method of adjusting acquisition costs used by the Commission, the rates of recovery of variable unit costs are as shown in Table 3. (75) The table also shows an average for the two services together, (76) weighted in line with their respective shares of turnover in the period in question. (77)

Table 3: Recovery of adjusted variable costs

<emplacement tableau>

(83) For the period ending August 2001, recurring revenue was not sufficient even to cover recurring expenditure, irrespective of non-recurring variable costs: recovery of recurring costs was [...]*% for the eXtense service, and [...]*% for the Wanadoo ADSL service. (78)

2. Recovery of variable unit costs

(84) An analysis of full costs is necessary in the light of the Community case-law on predation. It is all the more justified in the present case because. [...]* (79) The use of full costs was particularly necessary in that Wanadoo Interactive showed a net loss from 1999 to 2001. It could hardly avoid seeking to ensure that each product in the range covered its full cost, because during these years there was no product that showed a profit sufficient to free the other lines of business in the portfolio from the need to meet this objective. (80)

(85) In the first seven months of 2001 there is no doubt that full costs were not covered, given the level of recovery of variable costs. In the period after 1 January 2001 the rates of recovery of full costs were as follows. (81)

Table 4: Recovery of full unit costs

<emplacement tableau>

(86) In the period January 2001 to October 2002, then, the adjusted full unit costs were never recovered. It should be pointed out that the choice of the length of time over which the non-recurrent costs are to be written off has no effect on the finding that the company was unable to cover its full costs in this period. In the case of eXtense, the margin over total production costs was always negative throughout the period, so that there could never be a contribution to the recovery of non-recurrent acquisition costs even if they were to be spread over an infinite time. (82) In the case of Wanadoo ADSL, the margin over total production costs was likewise negative until December 2001; from January 2002 onward it became slightly positive, but would cover acquisition costs only over a period of more than 96 months.

3. Analysis by successive generations of new subscribers

(87) In its reply of 23 October 2002, Wanadoo Interactive criticises the Commission's method of adjusting acquisition costs by spreading them over 48 months. (83) In order to analyse cost recovery, Wanadoo argues that the Commission should have calculated not in terms of the entire duration of the line of business but in terms of the successive generations of new subscribers. Wanadoo Interactive suggests two methods: first, to analyse the recovery of variable costs and full costs looking only at the new generations of subscribers in each period, and second, to carry out a discounted cash flow analysis for each new generation of subscribers.

1. Analysis of adjusted costs for new generations of subscribers

(88) For each new generation of subscribers, the method suggested by Wanadoo Interactive ignores the costs of acquiring previous generations. (84) This method of assessing the equilibrium of a particular activity minimises the role played by the losses borne during the initial development phase. For example, Wanadoo Interactive's theory means that the assessment of the economic equilibrium of the ADSL business at the end of 2001 should not take account of the acquisition costs associated with the new subscribers recruited in the period January to July 2001, because a fresh generation of new subscribers has arrived in the meantime. Wanadoo Interactive is here calling into question the very principle of a spreading of acquisition costs. But despite this objection of principle, its theory does deserve study, as part of an analysis aimed at determining the period during which Wanadoo Interactive's pricing practices constituted a barrier to entry and to the expansion of competitors on the relevant market.

(89) In a supplementary analysis, therefore, the Commission has considered the costs really borne by each new generation of subscribers. There is no reason to look once again at the period from January to August 2001: here it is not disputed that the variable costs were not covered. For the three succeeding periods the results are actually very close to those for the subscriber population as a whole which are set out in Table 3 (within 1.5%). (85) If the recovery rates calculated for the two services are weighted in line with their respective shares in the progression of turnover, the result is actually less favourable than that obtained using the previous method in the period August to October 2001 and again in the period February to September 2002.

2. A cash flow analysis would be invalid in this case

(90) In its reply of October 2002 Wanadoo Interactive suggests that it should be considered whether in each new generation of subscribers the cash flow per customer, cumulatively over the average life of that subscriber, is positive. (86) The method Wanadoo Interactive outlines involves a number of difficulties of principle, in the light of which it should be rejected.

(91) First, this analysis does not allow any conclusion to be drawn regarding predation. In the most common configurations a firm pursuing a predatory policy hopes that it will be able to recover its initial losses in some form in the medium or long term. The initial loss intended to eliminate weaker competitors has to be offset by positive margins later. This means that there might very well be predation even though, despite heavy losses to the company at the beginning, the subscriber does generate a positive discounted cash flow over the duration of the subscription.

(92) Second, an analysis by population of subscribers artificially segments the successive waves of customers, whose economic and financial significance will ultimately be apparent only when they are looked at together. Even if one were to accept the method suggested by Wanadoo Interactive with respect to the generations of subscribers acquired in the past, one cannot overlook the impact of future generations of subscribers. To assess whether Wanadoo Interactive is acting reasonably in acquiring new subscribers, it is not enough to look at the discounted cash flows of a group of subscribers recruited in a stated period: one also has to consider the prospects of growth in later periods. A firm's strategy is not based exclusively on the pricing of a particular product, but incorporates other factors such as the rate of growth and the commercial objectives that reflect it. The method suggested by Wanadoo Interactive ignores this dimension.

(93) Let us suppose, for example, that the subscribers recruited in month t of year N allow the company to envisage a positive cash flow within a time frame compatible with the average life of a subscription. On the face of it it might be concluded that the recruitment of subscribers in month t of year N is profitable. But even if these subscribers individually may generate a positive cash flow in time, the growth of sales in the succeeding months (t+1), (t+2) etc. entails acquisition costs that reduce the profitability of the business as a whole. A segmented analysis of new recruitment overlooks the overall equilibrium of the business, which for the company is the vital question. The dynamics of sales and acquisition costs are such that the acquisition of new subscribers who may apparently be profitable individually in the medium term actually worsens the net cumulative position of the business as a whole in the short term. This aspect deserves special attention here, because Wanadoo Interactive's ADSL services are growing rapidly, and the company intends this growth to continue at least until 2004. Even though an individual subscriber might be expected to generate a positive discounted cash flow over three years, therefore, Wanadoo Interactive's ADSL business might nevertheless continue to be in overall deficit for many years more.

(94) In the short term, the recruitment of apparently profitable new subscribers will improve the net overall position of the activity in question only if the following inequality is satisfied:

(Xt-1 + xt).m > a.xt

where "Xt-1" is the cumulative total number of subscribers in month (t-1), "xt" is the number of new subscribers in month t, "m" is the margin per subscriber, and "a" is the acquisition cost per new subscriber. The condition can be restated (Xt-1 + xt).m - a.xt > 0. To put it another way, the recruitment of new subscribers will have a positive effect on the firm's net position only if the rate of growth of the business is less than m/(a-m). It is interesting to note that in the period from January 2001 to June 2002 the condition was never satisfied for the eXtense product. (87) Thus regardless of the medium-term value of the successive subscribers taken individually, the recruitment of new subscribers constantly worsened the net position of the business over that period.

(95) The Commission takes the view, therefore, that an analysis of discounted cash flows cannot take the place of an analysis of the overall position of the activity, because it gives an artificially optimistic picture of the terms of entry and expansion on the relevant market.

(96) The third difficulty in the approach suggested by Wanadoo Interactive is the absence of the information that would be needed to reconstruct the margins anticipated by the company. Here and there in the company's internal papers one can find assessments of trends in costs, but scattered information of this kind is not sufficient to allow a precise ex post reconstruction of the future revenue flows anticipated by Wanadoo Interactive at the time when it decided its commercial development strategy. Not even the prices of Wanadoo Interactive's offerings in the years 2003 and 2004 were plotted out with any certainty in advance. For example, several of the company's documents refer to a target price below FRF [...]*, taxes included, for the eXtense package in 2004, (88) but the development over time of the margin anticipated by the company was not clear.

4. Recovery of the adjusted costs foreseeable ex ante

(97) In order to analyse Wanadoo Interactive's tariff policy one has to reconstruct the recovery of variable and full costs at the end of 2001 and the first few months of 2002 as it could reasonably have been estimated on the basis of the assumptions used at the time by Wanadoo Interactive for new generations of subscribers. (89) As explained in paragraph 99, numerous incidental factors arose which affected the real outturn for the end of 2001 and the beginning of 2002. In the summer of 2001 it would not have been possible to anticipate with certainty the factors that actually affected the end of the year. And the factors that affected the year 2002 could not reasonably have been anticipated before the beginning of that year.

(98) Wanadoo Interactive has not contested the principle of such an examination of foreseeable costs, which was put forward in the supplementary statement of objections of 9 August 2002. It has merely said that some drops in price were not incidental but desired, although they were not referred to in budgetary assessments and accounting targets. (90) It quotes just one example, that of the costs of [...]*, which were the subject of commercial negotiation in [...]*. The Commission considers that these negotiations concluded too late [...]* for Wanadoo International to have been able to allow for them at the time it decided its commercial development strategy for the second half of 2001. (91)

(99) It is safe to assume that in the analyses of profitability it made in summer 2001, Wanadoo International could not have taken account of the following incidental factors:

- the fall in the costs [...]*;

- the refunds [...]*, (92) and the revenue foregone as a result of [...]*;

- the billing for bandwidth consumption at a level substantially lower than [...]*;

- the real distribution of subscribers between Île-de-France and the rest of the country [...]*. (93)

(100) On the other hand, Wanadoo Interactive expected that access to the service would be free in [...]*, whereas in fact it became so only in on 15 October 2001. As regards advertising revenue, Wanadoo International anticipated a recovery in the market in online advertising in the autumn, and it can be supposed that it envisaged advertising and audience revenue in line with its initial budgetary forecasts.

(101) On the basis of the assumptions used by Wanadoo Interactive at the time, therefore, if the company had been using a method of adjusting acquisition costs similar to that used by the Commission, the rates of recovery of adjusted variable costs that it could have anticipated in the second half of 2001 would have been as follows:

Table 5: Recovery of foreseeable variable costs at the end of 2001

<emplacement tableau>

(102) Turning now to the foreseeable full costs for the end of 2001, the target business plan for the two services for the end of 2001, referred to in the statement of objections of 19 December 2001, (94) showed levels of cost recovery of [...]*% before acquisition costs, as envisaged at the time by Wanadoo Interactive following its own methods. (95) These levels confirm that in the middle of October 2001 Wanadoo Interactive was still unable to anticipate recovery of the adjusted full cost in the last weeks of 2001, either for eXtense or for Wanadoo ADSL.

(103) To reconstruct the costs that could have been anticipated by the company at the end of 2001 for the year 2002 is a more delicate exercise. The company had put forward very favourable hypotheses for the development of certain costs, but it knew that these drops in costs would not take place at once at the beginning of the year. It knew that the year would be divided into two periods, before and after the lowering of France Télécom's access and routing tariffs.

(104) At the beginning of 2002, therefore, before France Télécom's new wholesale tariffs entered into force, Wanadoo Interactive could reasonably anticipate only a continuation of the tendencies in operation at the end of 2001, corrected to take account of a number of important factors:

- access to the service would be free to Wanadoo Interactive, either because these expenses would be billed to the customer or as a result of periodic France Télécom promotion campaigns; - the costs of the hot line would fall;

- the costs of [...]*; this was known [...]*;

- the net cost of the modem to Wanadoo Interactive would be reduced to [...]*;

- the costs of international connectivity would be reduced, though the bandwidth consumption assumed would be the same as in 2001. (96)

(105) On the basis of these hypotheses it can be presumed that at the beginning of 2002, before the new routing tariffs entered into force, Wanadoo Interactive could work on the assumption that the profitability of new subscribers, calculated according to the adjusted costs method used by the Commission, would be as shown in Table 6. As eXtense accounted for the bulk of the 2002 sales objectives, (97) Wanadoo Interactive could not expect fully to cover its variable costs.

Table 6: Recovery of foreseeable variable costs at the beginning of 2002

<emplacement tableau>

(106) In the second period in 2002, after the point when the sharp fall in prices that Wanadoo Interactive anticipated had taken place, Wanadoo Interactive could on the basis of its forecasts envisage the recovery of about 100% of the full costs as adjusted by the Commission's method.

5. Conclusion on recovery of costs

(107) It will be seen that for the whole of the period from January 2001 to October 2002 Wanadoo Interactive never covered the variable costs or the full costs relating to its ADSL services as they appeared in its own accounts. Even in 2002, when profitability improved, every additional month, bringing its cohort of new subscribers, worsened the net position of the ADSL business. This straightforward finding is repeated here only for the sake of clarity.

(108) The Commission has applied a method designed to assess whether Wanadoo Interactive's ADSL services had a production margin sufficient to cover the costs of acquiring new subscribers if the acquisition costs are depreciated over 48 months. To supplement this analysis, given that fortuitous circumstances arose at the end of 2001, the Commission has also evaluated the extent to which Wanadoo Interactive could have made a positive forecast of the profitability of its ADSL services in the second half of 2001 and at the beginning of 2002 on the basis of the working hypotheses available in the company at that time.

(109) The Commission has concluded from this method and the supplementary assessments referred to in this section that:

- from 1 January 31 July 2001 Wanadoo Interactive did not cover the variable costs of its ADSL services;

- from 1 August 2001 to 15 October 2002 Wanadoo Interactive never covered its full costs, but barely or almost covered its adjusted variable costs, though until March 2002 it could not have foreseen this;

- after 15 October 2002 Wanadoo Interactive covered its costs as adjusted in accordance with the method used by the Commission.

G. INTERNAL COMPANY PAPERS CONCERNING A STRATEGY OF PREEMPTION OF THE HIGHSPEED MARKET

1. The intention to preempt the high-speed market was formulated

(110) Several documents found on Wanadoo Interactive's premises attest the existence of a strategy of preemption of the high-speed market or "ADSL market". In particular:

- a document dating from July 2000 (98)[...]* (99) [...]*

- [...]*; (100)

- ; [...]* (101)

- [...]*;

- [...]*speaking of the period from 2001 to 2003, [...]* (102)

(111) This objective of preempting the high-speed market is a reflection at Internet service provider level of the objective set for France Télécom as a whole by its chairman with respect to ADSL, namely "to preempt the market by stealing a march on our competitors". (103)

(112) A translation of this concept of preemption into market share terms can be seen in several documents, which state that Wanadoo Interactive is trying not just to be the market "leader" but to take and hold very large market shares.

(113) [...]*. (104)

(114) [...]*:

- [...]*; (105)

- [...]* (106) [...]*.

(115) [...]* (107) [...]*. (108) [...]* (109)

(116) [...]*. (110) [...]* (111) [...]*

(117) [...]*. (112) [...]*

(118) The strategy pursued on the high-speed market formed part of a broader effort to increase market shares for all forms of Internet access. [...]* (113) [...]*

(119) Wanadoo Interactive has contested the significance of the papers cited here and of the word "preemption" they use, and the relevance of the ambitions expressed in terms of market share. The words quoted by the Commission, it says, merely reflect the dialectic of the decision-making process in a big organisation; they represent the views of their authors, and not of the company as a legal person. And they are informal and impromptu manners of speaking, or in some cases off-the-cuff messages not thoroughly thought through. (114)

(120) As a matter of strict fact the Commission feels it ought to refute this contention.

(121) First, the words quoted regarding the objective of preemption were for the most part not uttered in the course of "aimless conversations", as Wanadoo Interactive has claimed. (115) Three of the five passages referring to preemption which have been quoted here are taken from formal presentations to management meetings. The document from which the fourth passage is taken is [...]* not a document prepared by Wanadoo Interactive's sales management as a means of motivating its sales staff on the ground, as the company has maintained. (116)

(122) Some documents, [...]*, could indeed be described as informal, but this makes it all the more instructive to observe the way the objective of preempting the market permeated the company, appearing in papers of very different kinds. Looking beyond the diversity of the documents, therefore, one can see practical evidence of the company's unity of purpose. Contrary to what Wanadoo Interactive claims, the authors of these papers never do actually discuss the objective of preempting the market: it is quite clearly the initial assumption they all make.

(123) Nor are the quotations given here from the words of junior employees in the company: they are all from senior managers, and indeed [...]*. It is hardly likely, therefore, that they should be off-the-cuff messages not thoroughly thought through. (117)

(124) Wanadoo Interactive has rightly pointed out that the market share targets announced are not absolutely identical: the estimates of Wanadoo Interactive's share of the highspeed market range from [...]*% to [...]*%, with the exception of one document giving a figure of [...]*%. The company has complained that the Commission has not tried to rank these evaluations in the order of their importance. (118) Let us therefore consider the document with the greatest authority, [...]*. It was this presentation that announced market share targets of [...]*% for dissociated ADSL offerings, like Wanadoo ADSL, and [...]*% for packaged ADSL offerings, like eXtense. A document immediately subsequent to that presentation confirms the objective of taking [...]*% of the dissociated ADSL offerings segment by 2004. (119)

2. Internal thinking on the price level and Wanadoo Interactive's knowledge of the economic and financial impact of the development of its ADSL services

(125) In the course of its investigations the Commission has been able to take stock of the internal process of reflection on the setting of prices for the services in question and the economic and financial effects of the tariffs decided. The material gathered on the company's premises is outlined below.

(126) A large number of company documents, [...]*confirm that the company was aware of the dangers associated with its pricing of ADSL services from a very early stage. These documents clearly show that Wanadoo Interactive knowingly weighed a shortterm profitability objective against an objective of vigorous penetration of the market, and that it deliberately sacrificed the first to the second. [...]*

(127) [...]*the price of the Wanadoo ADSL service was set at FRF 135 (or euro 20.58), taxes included. (120) [...]*. (121) [...]*. It was originally a promotional offer, (122) but became a permanent price.

(128) From January 2001 the price of the eXtense package subscription was set at FRF 298 per month, taxes included. That price and the price of the "pack" containing the modem had been the subject of different hypotheses for more than a year.

(129) Much earlier, [...]*, a price target of FRF [...]*, taxes included, for a subscription to the projected product, then called [...]*, had been mentioned [...]*. (123)

(130) In spring 2000, when prospects for the launch of a packaged offering by Wanadoo Interactive were hardening, there was mention of a subscription price of FRF [...]*, taxes included. (124) [...]*. [...]* (125) But that scenario lost ground in summer 2000, after [...]*.

(131) From summer 2000 onward, several alternative scenarios for combining the price of the pack and the monthly subscription were under discussion. (126) Under the hypotheses being studied, subscription prices for 2000 and 2001 ranged from FRF [...]* to FRF [...]*, taxes included, and the price of the pack from FRF [...]* to FRF [...]*, taxes included. (127) One of these scenarios was known as the "central scenario". Here the monthly subscription would be FRF [...]* , taxes included, and the pack FRF [...]*. The combination of prices that was ultimately adopted was not considered in the course of that exercise; (128) but on the hypotheses that Wanadoo Interactive was using at the time, it can be calculated that the prices actually set in January 2001 would have generated very heavy deficits and a net value which updated to 2003 would have been negative by FRF [...]* . (129)

(132) In [...]* the discussion of the level of prices considered new forecasts of the costs of the service, in particular involving more favourable assumptions for network costs. A presentation made on [...]* as part of the budgetary discussions speaks of a subscription price of FRF [...]* (130) for the whole of 2001; this was the price then entered in Wanadoo Interactive's draft budget. The price of the subscription was finally adjusted slightly in the course of the autumn, and the price charged in January 2001 was FRF 298, taxes included, with the price of the pack set at FRF 990, taxes included.

(133) Thus discussion of the level of prices for the eXtense package was hesitant for a fairly long period in 2000. But it should be noted that all the scenarios envisaged expected heavy losses in 2001, even supposing that favourable tariffs were to be applied for routing and for Netissimo immediately on 1 January 2001.

(134) [...]* (131) [...]* (132) [...]* (133)

(135) The choice of a subscription price for the eXtense package slightly below FRF[...]*, taxes included, reflects an intention to penetrate the market rapidly, regardless of the short-term financial disadvantages. [...]* (134)

(136) The version of the 2001 budget dated [...]* provided for a gross margin over the production costs of the eXtense package and Wanadoo ADSL together equal to FRF [...]*, and a net margin, after taking account of acquisition costs, of FRF [...]*. (135) The version of the 2001 budget dated[...]*, (136) after pointing out that ADSL would have a gross margin of FRF [...]* in 2000, provided for a gross margin per unit of FRF [...]* per month in 2001, equal to a total gross margin of FRF [...]* and a net margin of FRF [...]*. (137) The version of the 2001 budget dated [...]* provided for a production margin of FRF [...]* on the same products, and a net margin of FRF [...]* when account was taken of acquisition costs. [...]*

(138)

(137) At the end of 2000 and the beginning of 2001, it was clear that in Wanadoo Interactive's portfolio of products ADSL services stood out on account of the extent to which they were in deficit. This is particularly instructive with reference to the way in which Wanadoo Interactive weighed considerations of market penetration against respect for the short-term economic fundamentals. The definitive version of the 2001 budget provided for a margin over production cost of FRF [...]* per ADSL subscriber per month; [...]*. (139)

(138) For part of the first half of 2001 these figures gave rise to discussion within the company regarding the pricing of the eXtense package. The expected fall in the tariffs for traffic routing and Netissimo had failed to materialise, and the cost assumptions made in the budget were proving overoptimistic. [...]* (140) [...]* (141) Between March and May 2001 possible monthly subscription prices of euro [...]* (FRF [...]*) (142) or FRF [...]*, taxes included, were spoken of within the company, as was an increase in the price of the pack. (143) [...]* (144) [...]*.

(139) The changes in the tariffs for IP/ADSL traffic routing and Netissimo access which were made in August 2001 produced some improvement in the situation. But internal papers show that serious concern nevertheless persisted within the company regarding the viability of ADSL at the end of 2001 and even at the beginning of 2002. [...]*

(140) [...]*. (145) [...]*. (146) [...]*. (147)

(141) [...]* (148) [...]* (149) [...]* (150) [...]* (151) [...]* (152) The wording reflects an extremely reserved assessment of the financial logic of Wanadoo Interactive or its competitors developing their high-speed business to any significant extent in 2002.

3. Wanadoo Interactive was aware that there were legal dangers

(142) The internal papers gathered on the company's premises show that Wanadoo Interactive was made aware in advance of the legal dangers associated with pricing its retail services below cost.[...]* (153) [...]* (154) [...]* (155) [...]*. (156)

(143) In its reply of October 2002 Wanadoo Interactive argues that the documents cited were valueless because they did not originate in its legal department. In the Commission's view the situation is quite the reverse: it is instructive that concern over the legitimacy of the practices was widespread in the company, and was not confined to legal specialists. [...]* Wanadoo Interactive has observed (157) [...]*. (158) But this observation is of no value here, because the words in question did not in any way refer to the gap between the retail prices of Wanadoo Interactive's products and the underlying costs, but rather to the relationship between France Télécom's wholesale offering ADSL Connect ATM, which was intended for alternative telecommunications operators, and wholesale prices for the resale of services to Internet service providers.

(144) The Commission takes the view that without necessarily knowing that its price might constitute predation, (159) Wanadoo Interactive was well aware of the problem of a squeeze between the wholesale and retail prices of the provision of ADSL services to end users. As early as the summer of 2000 Wanadoo Interactive had identified the legal risk [...]* This awareness should have encouraged it to observe a measure of discipline in its pricing policy, and to price its services at a level above its costs.

4. The context provides evidence of a strategy designed to contain and drive off competitors

(145) The setting of prices below cost is not the only method used by Wanadoo Interactive and its main shareholder to hold back the development of competitors and to divert the growth in the high-speed market to its own advantage. Other means also seem to have been used. Although these actions can be imputed to Wanadoo Interactive itself only in part, and although they are not the subject of objections made against Wanadoo Interactive, they are very important for an understanding of the context of the case. This is because they show how Wanadoo Interactive's policy of selling below cost in order to preempt the market formed part of an overall strategy.

(146) First, Wanadoo Interactive was given privileged access to information regarding the suitability of telephone lines for ADSL technology, and generally regarding the operational implementation of ADSL connections. Its competitors in the ADSL segment were asked to go through a long procedure that was difficult to reconcile with mass marketing, especially via large retailers. Internal company documents attest that Wanadoo Interactive was well aware of the discriminatory advantage conferred on it by the immediate verification of the suitability of lines, because it was informed of an instruction given to France Télécom shops not to mention this. (160) The situation was referred to the French Competition Council by T-Online France in an application for interim measures. By decision of 27 February 2002 the Competition Council held that there was a danger of structural discrimination in favour of Wanadoo Interactive in the operation of the IP/ADSL contract which would be a serious handicap to mass marketing, and asked France Télécom to put an end to it.161 That decision was upheld by the Paris Court of Appeal on 9 April 2002.

(147) On this point Wanadoo Interactive does not deny the facts adduced by the Commission, and confines itself to observing that in July 2002 the Competition Council decided that its injunction of February 2002, which led to the suspension of the sale of eXtense packs, was being complied with, and that the ban could be lifted. (162) The facts adduced by the Commission therefore remain true of the period before summer 2002, and Wanadoo Interactive has not contested them with respect to that period.

(148) Second, the supply of ADSL modems was probably used jointly by France Télécom and Wanadoo Interactive to slow the development of competitors in the start-up phase at least in the first few months of 2001. France Télécom had authority to approve ADSL modems for use on its network. This power seems to have been used to postpone any widening of the range of potential modem suppliers, and to consolidate the shortage that obtained on the market at the beginning of 2001. On the one hand, France Télécom delayed the approval of modems manufactured by ECI, which it finally granted on 19 March 2001, and waited for a decision of the ART on a dispute proceeding before allowing Internet service providers to install modems not approved by that date; the regulator's decision came in November 2001. (163) On the other hand, orders for modems placed jointly by France Télécom and Wanadoo Interactive had the effect of taking up almost all of the production capacity of the only supplier authorised at the time, Alcatel, and this made it difficult for competitors to obtain supplies. (164) [...]*, (165) [...]*. (166) [...]*. (167) The "closing off" of the modems market in the first few months of 2001 had a strong inhibiting effect on the initial development of competitors. By way of example, in December 2000 Wanadoo Interactive already had at least [...]* modems in stock at France Télécom shops, while in January 2001 TOnline had succeeded in obtaining only one-tenth of that figure.

(149) Wanadoo Interactive has not commented on the facts adduced or the quotations used by the Commission. It has merely referred the Commission to a decision taken by the French Competition Council in June 2002, dismissing an application for interim measures lodged by Olitec. (168) Olitec, a manufacturer of terminal equipment, complained of an absence of transparency in the procedure for the approval of modems and the tenders organised by France Télécom. The Competition Council took the view that since the decision of the ART in November 2001, which had been followed by the amendment of France Télécom's IP/ADSL contract in March 2002, the difficulties in the approval procedure were now surmountable. As regards the time before that the Council considered that Olitec had not been able to produce sufficiently convincing evidence to show that the methods of approving modems reflected discriminatory practices. On this point the Commission would observe that when it took its decision the Competition Council had no knowledge of the existence of the documents referred to in paragraph 148, which seem to indicate that the approval of modems was based not solely on technical criteria but also on considerations of advantage.

(150) Lastly, in autumn 2001 France Télécom put a "partnership offer" to Internet service providers, which was to apply with effect from December 2001; (169) the offer brought with it a serious danger of restriction of competition. At that time the relative improvement in the tariffs for IP/ADSL access services and routing made it less unrealistic economically for competitors to come forward with ADSL offerings, and several Internet service providers were trying to arrive at a policy for marketing via large retailers. The principle of the partnership offer was that France Télécom, in association with providers, would market ADSL access in packaged form at a fixed price (FRF [...], taxes included, to be paid on purchase, and FRF [...], taxes included, to be paid as a monthly subscription, including FRF [...], taxes included, as a monthly subscription to the service provider). This would give Internet service providers the benefit of France Télécom's brand image and distribution network. In return the retail prices were to be fixed, as was the payment to be made to the sales outlet, which was FRF [...] per subscription activated. The object of this operation was doubtless to discourage competing telecommunications operators from offering services similar to Netissimo, (170) and thus to reduce the possibility that service providers might offer products technically different from those of Wanadoo Interactive. By circumscribing retail prices and the commissions to be paid to distributors, the arrangement would also have the effect of depriving service providers of much of their freedom of manœuvre in their commercial policy. This partnership offer, although the responsibility for it lay not with Wanadoo Interactive but with Wanadoo Interactive's majority shareholder, is witness to an objective of control of the development of the market (171) and a desire to dissuade competitors from differentiating their offerings from those of the company in question..

(151) On this point Wanadoo Interactive does not contest the existence of the partnership offer, and confines itself to observing that the partnership offer was never put into effect. (172) The Commission would point out that the offer was not put into effect because of the negative reaction of the firms to which it was made, and because of the action taken by the Competition Council. (173) The absence of implementation was not in any way due to a unilateral withdrawal of its commercial proposal by France Télécom.

H. PROCEDURE

1. History

(152) By letter of 18 September 2001 the Commission informed France Télécom and Wanadoo Interactive that the information gathered in the course of the sector inquiry into the unbundling of the local loop, from those two undertakings and from their competitors, gave grounds for the initiation of an investigation specifically into the pricing of the ADSL products marketed by Wanadoo. The Commission then sent various requests for information in order to complete its information.

(153) This stage of the investigation culminated on 19 December 2001, when the Commission sent Wanadoo Interactive a statement of objections. Wanadoo Interactive submitted its observations on the statement of objections in a reply dated 4 March 2002, and at a hearing held on 18 March 2002. Several competitors - AOL France, Mangoosta (in liquidation under court supervision), Tiscali France, Easynet and Noos - were recognised as interested third parties, and were allowed to attend the hearing.

(154) At a meeting held on Commission premises on 19 March 2002, representatives of France Télécom, (174) while denying any infringement, nevertheless put forward a plan to remedy the failure to cover costs by thoroughly recasting France Télécom's IP/ADSL access and routing charges. (175)

(155) On 4 and 5 April 2002 Commission officials visited the company's premises under Article 14(2) of Regulation No 17. On this occasion Wanadoo Interactive presented them with its estimates of revenue and costs for the business in question in 2001 and 2002. The Commission officials took copies of numerous company papers.

(156) On 9 August 2002 the Commission sent Wanadoo Interactive a supplementary statement of objections. This completed a number of points of fact and law in the analysis set out in the statement of objections of 19 December 2001, in particular by incorporating the information that the Commission had obtained since the beginning of 2002 in reply to requests for information made to Wanadoo Interactive and its main competitors, and the conclusions the Commission had drawn from its visit on 4 and 5 April 2002.

(157) Wanadoo Interactive submitted its observations on this supplementary statement of objections in a reply of 23 October 2002, and did not request an oral hearing.

(158) On 16 January 2003 the Commission sent Wanadoo Interactive a letter setting out facts which had not been mentioned in the statements of objections and of which account might be taken in a Decision (hereinafter the "letter on the facts"). Most of this material consisted of updatings of tables that had been sent to Wanadoo Interactive with the supplementary statement of objections. The letter on the facts explicitly stated that the information it contained was not such as to extend the duration of the suspected infringement or in any way to alter the Commission's assessment of the degree of gravity of the abuse.

(159) In a letter to the Hearing Officer dated 31 January 2003, Wanadoo Interactive asked for an extension of the time allowed for a reply, which had initially been set at four weeks. The Hearing Officer did not accept the grounds the company put forward in support of its request, but allowed Wanadoo Interactive a further ten days to enable it to seek any clarification of the letter on the facts that might be necessary from the Commission departments. On 26 February 2003 the representatives of Wanadoo Interactive did ask the Commission for clarification of a number of aspects of the letter on the facts. The Commission replied by letter of 28 February 2003. By letter of 4 March 2003 Wanadoo Interactive replied on the substance of the letter on the facts of 16 January

2. Access to the file

(160) In the course of the proceedings Wanadoo Interactive exercised its right of access to the file on several occasions: on 7 and 23 February, after the first statement of objections was sent; on 20 and 21 August 2002, after the supplementary statement of objections was sent; and on 23 and 27 January 2003, after the sending of the letter on the facts. By letter of 6 June 2003, lastly, the Commission sent Wanadoo Interactive numbered copies of the papers not covered by business secrecy that had been included in the file since the preceding January. (176)

(161) In several letters Wanadoo Interactive has argued that it has not been given proper access to the file. (177) First, it says that the Commission has not given it access to the sector inquiry into local loop unbundling; it takes the view that the contributions made by competing firms might provide information that would be useful to its argumentation. The Commission, it says, has selected some documents from the sector inquiry and has deliberately ignored others. Second, Wanadoo Interactive disputes the inaccessibility of papers provided by some of its competitors, and criticises the Commission for failing to give any indication of the content of those documents. Third, Wanadoo Interactive argues that it has not had full access to the whole of the Commission's file on this investigation, because some papers are not entered on the numbered list of documents that was supplied to it.

(162) On the first question, it is true that the sector inquiry was in a sense a catalyst for the investigations that led to proceedings being initiated specifically against Wanadoo Interactive; this is part of the function of any inquiry into a sector of the economy. But the sector inquiry was looking at the question of local access in general, and especially offers of fully unbundled or shared access and offers of indirect access aimed at operators. Thus it covered an area much wider than the subjects investigated in the present proceeding. In the interests of sound administration the Commission opened two separate files, with some documents from the sector inquiry being copied and included in the new file on Wanadoo Interactive. The company has accused the Commission of discrimination in the selection of the papers included in the Wanadoo Interactive file. On 18 February 2002, to enable Wanadoo Interactive to form an opinion of the objectivity the Commission had exercised in selecting the papers to be included in the new file, the Commission sent Wanadoo Interactive an exhaustive numbered list of the documents in the sector inquiry in France. Wanadoo Interactive has not considered it necessary to ask to see specific documents on the list sent to it. The Commission would also point out that in March 2002 it published a report on the question of the unbundling of the local loop, with a detailed annex on France, setting out all the conclusions to be drawn from the inquiry. (178)

(163) On the question of accessibility of documents, the Commission feels it has to observe a fair balance between the rights of the defence and the protection of the interests of the firms which have supplied it, at its request, with information constituting business secrets. The Commission notice on the internal rules of procedure for processing requests for access to the file in cases pursuant to Article 85 and 86 of the EC Treaty, Article 65 and 66 of the ECSC Treaty and Council Regulation (EEC) No 4064-89, (179) and especially paragraph II.D.2 thereof, lays down a line of conduct to be followed by the Commission which is intended to preserve this balance, more especially in cases of abuse of a dominant position. The notice points out that in cases of abuse of a dominant position competitors face a special danger of retaliatory measures, because of the considerable commercial pressure that the dominant firm is able to exercise on them; this calls for special precautions. In general, therefore, the Commission has treated information sent by competitors as confidential wherever they so requested. At the same time, by giving Wanadoo Interactive systematic access to the content of the requests for information which the Commission sent to the firms, whether as part of the sector inquiry or in the course of the investigation of the present case, it has enabled the company to assess the object and scope of the information provided by its competitors. When in their replies these firms added information going beyond the scope of the questions asked, a brief indication of the nature of those comments was entered on the list of the documents in the file that was supplied to Wanadoo Interactive. When Wanadoo Interactive so requested, giving its reasons, the Commission obtained non-confidential versions of their replies from the competitors, and sent them to Wanadoo Interactive. (180)

(164) The unnumbered documents not included in the enumerative list are all entirely public documents, such as press articles and the published reports of international institutions. None of them is used as evidence or incriminating matter in this case. These documents are freely accessible to the public. And although it did not number them, the Commission did give Wanadoo Interactive access to all these documents; the company's legal representatives made complete copies in January 2002, August 2002 and January 2003.

(165) The Commission believes that on these various points it has followed an approach that respects the legitimate interests of competitors and allows Wanadoo Interactive fully to exercise its right to defend itself. The Commission has explained this approach to the company repeatedly, both in conversation and in various letters. (181)

3. Wanadoo Interactive's objection to the letter on the facts

(166) Wanadoo Interactive drew attention to what it saw as a difficulty of procedure, namely the significance and scope of the letter on the facts which the Commission sent on 16 January 2003. In a letter to the Hearing Officer dated 31 January 2003, Wanadoo Interactive contested the very principle of the letter on the facts, contending that its nature and content required that a third statement of objections be sent in order to enable it to defend itself properly.

(167) In his reply of 18 February 2003 the Hearing Officer observed that the letter from the Commission departments did not in any way add to the scope of the two earlier statements of objections, or modify the Commission's assessment in a manner unfavourable to Wanadoo Interactive; an assurance to that effect was given in several places in the letter in question, and in those paragraphs where it was not explicit it could be deduced sufficiently clearly from the context and nature of the comments made. The Hearing Officer concluded that it was not necessary that the Commission send Wanadoo Interactive a further supplementary statement of objections.

II - LEGAL ASSESSMENT

A. ARTICLE 82 OF THE TREATY

(168) Article 82 of the Treaty prohibits as incompatible with the common market any abuse by one or more undertakings of a dominant position within the common market or in a substantial part of it insofar as it may affect trade between Member States. Such abuse may, in particular, consist in directly or indirectly imposing unfair purchase or selling prices.

B. THE RELEVANT MARKET

1. The service market

1. Introduction

(169) In its statement of objections of 19 December 2001, the Commission defined the relevant market as that for high-speed Internet access for residential customers. On that market, Internet service providers offer customers, consisting mainly of households, Internet access for a single terminal, with download speeds up to 512 kbit/s182 and upload speeds of 128 kbit/s during the period covered by this Decision. A flat-rate amount is charge for these services. They offer unlimited use and allow simultaneous use of the telephone line to make and receive calls. These Internet access services allow hosting of a website and the provision of email facilities, including a number of email addresses. The offerings available on the market comprise very comparable functionalities. Prices are very similar. The service providers' marketing information is based on two key messages: download speed and the convenience of simultaneous telephone and Internet use.

(170) This market definition includes both offerings based on ADSL technology and offerings based on cable modem technology. Although the geographic penetration potential of the two technologies is very different, with ADSL offering much wider scope, and although the technical performances are not wholly comparable, because of the greater risk of congestion on cable networks, it was apparent that, on the demand side, substitutability between these different offerings was sufficient to justify their inclusion within the same service market here. It should be emphasised, however, that the options offered by the cable operators have since early 2001 consistently had a much lower growth potential than the options based on ADSL technology. On 1 January 2001, the cable television networks served around 8 million French households, whereas on the same date the ADSL technology deployed on the fixed telephone network allowed some 11 million telephone lines to be covered. In addition, of the 8.4 million marketable cable sockets, only half had undergone the necessary adjustments to be able to take high-speed Internet connection. Lastly, the cable operators operate on the basis of municipal or local franchises and their presence throughout the country is fragmentary, whereas service providers using ADSL technology are subject only to the constraints imposed by the roll-out of ADSL technology by France Télécom. The service providers' penetration capacity was thus a priori greater with ADSL as from the end of 2000.

(171) The Commission has also taken the view that high-speed Internet access for residential customers must be distinguished from high-speed Internet access for business customers. There are at least three differences: very different price ranges; generally much superior operating capacities for business customers; and a wider range of technical options for access. In the first place, prices are frequently three to five times higher than those charged to residential customers. These higher charges obviously reflect more advanced functionalities. The products designed for business customers offer download speeds at least twice as great as those described above for residential customers, greater web-page hosting capacities and the possibility of multi-terminal use and networking operation. Their technical specifications and charges are sometimes tailored to the customer, which is impossible on the mass market. Lastly, from a technical point of view, high-speed Internet access for business customers can be achieved through a greater variety of solutions than in the case of residential customers. Leased lines, wireless local loop or satellite transmission are not realistic options at present in the case of residential customers, given the costs involved, but they are viable in the case of business customers.

(172) The inclusion in the service market of offerings provided by cable operators and the exclusion of offerings intended for business customers were not contested by Wanadoo Interactive.

(173) However, Wanadoo Interactive did object to the Commission's distinction between high-speed Internet access and low-speed Internet access. Wanadoo Interactive identifies only one market for Internet access as a whole. In its view, Internet access is growing steadily, albeit a growing proportion of consumers is being attracted, in the period 2000 to 2010, by high-speed Internet access offerings, at the expense of lowspeed Internet access offerings. High-speed and low-speed, in its view, thus simply constitute two segments of one and the same market. Following the migration process, the low-speed segment, it argues, is now of only residual importance. In Wanadoo Interactive's view, what we are witnessing therefore is simply a gradual move upmarket, with the old version of the service (low-speed) tending to be replaced by the new version (high-speed), this being a basic market trend. (183) Wanadoo Interactive's argument is based on the fact that, since the specific content of high-speed Internet access had not matured, high-speed Internet access uses available in 2001 and 2002 were in fact similar to low-speed uses. Wanadoo Interactive emphasises in particular the similarity, in its view, between the flat-rate offerings for low-speed access beyond 20 hours, which it considers are comparable to unlimited packages, which it claims are offered by numerous competing providers, though it cites only two examples relating to AOL. (184) In fact, it argues, it is not the speed which high-speed users are seeking, but rather the flat-rate charge and unlimited use available in such offerings. For the reasons set out in paragraphs 174 to 202, the Commission takes the view that the relevant service market is restricted here to high-speed Internet access alone, excluding the low-speed market.

2. Differences in use

(174) Wanadoo Interactive argues that no distinction can be made between "high-speed" and "low-speed" because the uses made of them are not radically different and because most users use the same type of Internet access applications and functions irrespective of the mode of access.

(175) Clearly, high-speed and low-speed Internet access offer one and the same basic range of common uses: email, information searches, navigation, chatting, downloading of files, etc. However, it is important to note that some of the high-speed uses are possible with low-speed Internet access only under extremely dissuasive conditions of loss of quality and convenience, while some applications available with high-speed access are simply not feasible with low-speed access. For example, the downloading of very voluminous video files or interactive network games are very difficult or indeed impossible to perform with low-speed Internet access. Moreover, Wanadoo Interactive acknowledges, in its reply of 4 March 2002 to the statement of objections, that "[...]* % of audiovisual/multimedia activities are specific to ADSL". (185) Wanadoo's chief executive officer confirmed in the autumn of 2002 that, for network games, a download speed of 128 kbit/s was insufficient and that consequently such applications were possible only with higher download speeds. (186) It may be noted that the fact that some applications are not feasible with low-speed Internet access is one of the elements taken into account by other competition and regulatory authorities (187) in identifying two separate markets. (188)

(176) The ease of use provided by high-speed Internet access explains why subscribers have quite distinct user profiles. Almost 58% of high-speed users in France are reported to use audio-video applications, as against less than 25% of Internet users as a whole; more than 45% of high-speed users in France are reported to use the Internet for file transfer, as against less than 21% of Internet users as a whole. (189) Wanadoo Interactive estimated in October 2001 that [...]* % of eXtense subscribers downloaded audio files. Internet users with high-speed Internet access are generally ranked as large consumers. (190)

(177) High-speed Internet access subscribers are connected much more frequently than lowspeed Internet users. One of the advantages of high-speed offerings is the elimination of connection time, which can take 45 seconds in the case of low-speed access. This feature encourages one-off and frequent Internet consultations, for example to look up a particular item of information. According to Wanadoo Interactive, [...]* % of ADSL users use the Internet several times a day, as against [...]* % of its customers in all categories taken as a whole. (191)

(178) Another particular feature is that the high-speed subscriber is connected for a much longer time on average than the low-speed user. Wanadoo Interactive's figures confirm that the duration and frequency of Internet use increased considerably once the switch is made from low-speed to high-speed access. According to an internal company document dated October 2001, the monthly duration of declared actual use rises from [...]* hours to [...]* hours, (192) whereas [...]* % used the Internet several times a day as against [...]* % of those who had not yet subscribed to eXtense. Another document dated 13 July 2001 shows that in April 2001 Wanadoo Interactive's ADSL subscribers used the Internet on average [150-250]* hours a month, as against an average of [0- 100]* hours in the case of customers who had subscribed to the low-speed package "Intégrales 100 heures". (193)

(179) According to Wanadoo Interactive, high-speed Internet access subscribers are not seeking speed or convenience of use so much as unlimited access. The company cites as evidence the frequent reference to the unlimited access facility which users identify as their reason for choosing high-speed access. (194) In point of fact, this feature simply reflects the fact that, up to August 2002, the conditions did not really exist for the development of unlimited low-speed access facilities under acceptable cost conditions and at attractive price levels for the final customer, which prompted low-speed users attracted by the prospect of unlimited access to turn to high-speed access facilities. (195)

(180) Lastly, another aspect of the particular profile of high-speed users is the bandwidth used by this type of service. Wanadoo Interactive's ADSL competitors point out that the real subscriber bitrates observed at the beginning of 2002 were on average greater than 20 kbit/s per connected subscriber, as against 1 kbit/s approximately on average in the case of low-speed users. (196) There is thus a ratio of 1 to 15 at least between the bandwidth consumed per high-speed user and a low-speed user. This gap reflects the very wide differences in use, which is much heavier in the case of high-speed Internet access.

3. Technical features and performances

(181) An important technical feature of high-speed Internet access is the specific nature of the modems used. A high-speed Internet access modem cannot be used for low-speed Internet access and vice versa.

(182) In its reply of 4 March 2002, Wanadoo Interactive argued that a market definition could not be based on performance differences identified between products, since uses are relatively convergent and similar. On the contrary, any such differences in performances would simply be symptomatic of a process of moving upmarket in a market which was constantly developing. Higher-performance products gradually superseded earlier generations of less sophisticated products. To back up its arguments, Wanadoo Interactive drew an analogy in the Internet access sector with the personal computer and server sectors. It cited three Commission Decisions in merger cases concerning these sectors. (197) According to Wanadoo Interactive, in these various Decisions, the Commission decided not to distinguish different markets for products that involved developing technology. (198)

(183) Even assuming the analogy with the IT equipment sector is relevant, it should be noted firstly that, in actual fact, in the Decisions cited by the company, the Commission did not reach any conclusion on market definitions. The extract quoted by Wanadoo Interactive from the Fujitsu/Siemens Decision (199) merely reports the opinion stated by a number of third parties during the investigations, and the Commission went on in the very next sentence of its Decision to state that, for the purpose of the case in point, it was not necessary to reach any conclusion on the question of the relevant product market. The HP/Compaq Decision recognises that three separate server markets should be distinguished on the basis of different functionalities, reflected more or less in the price bands. (200) It may also be noted that, in a very different sector, that of passenger cars, though it also left the market definition open, the Commission recognised that it was possible to identify several separate markets on the basis of a number of objective criteria such as engine size, (201) even though the ultimate purpose of owning a car, namely the individual transport of passengers on public roads, is always the same. Accordingly, technical differences in performance, though not necessarily the sole basis of the market definition, may be a factor in it. (202) It is therefore useful to examine in greater detail the performance differences between high-speed and low-speed Internet access.

(184) There is no absolute technical definition of the dividing line between "low-speed" and "high-speed" access. However, there is a general consensus that "high-speed" refers to services that allow download speeds appreciably greater than those made possible by ISDN technology. (203) The Organisation for Economic Cooperation and Development (OECD) considers high-speed access to mean any access facility that allows a download speed of more than 256 kbit/s. (204) The Federal Communications Commission (FCC) defines high-speed access as download speeds in excess of 200 kbit/s, i.e. well in excess of the performance allowed by "primary ISDN" technology. The UK regulator Oftel has defined high-speed Internet access as downstream capacity in excess of 128 kbit/s. (205) In the case of the French market being examined here, it should be noted that, during the period under review, from January 2001 to October 2002, all the offerings allowed download speeds of the order of 512 kbit/s, i.e. around four times greater than that allowed by ISDN technology on low-speed networks. The offerings on the French market are thus well above the speeds used as criteria both by the OECD and by the FCC.

(185) In the case of high-speed Internet access, the offerings available from French operators during the period covered by the infringement involved download speeds equal or superior to 512 kbit/s. In the case of standard Internet connection, the download speed on an analogue line is 56 kbit/s. ISDN technology does, it is true, allow download speeds on the telephone line to be boosted to 128 kbit/s. But, in addition to the fact that it offers lower potential than that required for the generally-accepted definition of high-speed access referred to in paragraph 184, ISDN technology is reported to account in France for scarcely 0.3% of domestic Internet access. (206) It played a wholly marginal role during the relevant period. There are no unlimited Internet access offerings available using ISDN. Wanadoo Interactive draws attention to the launching of ADSL offerings with download speeds of 128 kbit/s in the autumn of 2002 which, it argues, bears witness to the continuity between low-speed and high-speed. These offerings became available at the end of 2002, at the end of the period covered by this Decision. There is no need for the Commission to decide whether they belong to the high-speed or low-speed market. The Commission does not possess the information that would enable it to evaluate the degree to which they can be substituted for 512 kbit/s offerings. At first sight, however, factors militating against their inclusion in the high-speed Internet access market are their technical performances and a speed which is four times less than that provided by 512 kbit/s offerings, and a range of prices which is much closer to unlimited low-speed offerings. (207)

(186) In the vast majority of cases, it may be said that during the period covered by this Decision, high-speed Internet access offered the possibility of download per second at virtually ten times the rate of low-speed Internet access. This was indeed the commercial angle which the service providers constantly focused on during the relevant period. For the purposes of defining a market, any such performance ratio of 1 to 10 amounts to a very significant difference.

(187) The differences in performance between the two categories of products are clearly perceived by consumers. According to Wanadoo Interactive's marketing services, ADSL is chosen by Internet users primarily on account of [...]*. (208) [...]*. (209) In addition, according to Wanadoo Interactive's internal documentation, "ADSL is preferred to unlimited access packages because of [...]*. (210) It cannot therefore, be argued that performance differences are regarded as marginal by users. Furthermore, analysis of price differences between low-speed and high-speed offerings shows that consumers are prepared to pay a premium for extra performance and convenience.

4. Price differences in respect of services and differences in revenue per subscriber

(188) In return for the speed and convenience which it provides, high-speed Internet access entails a different charge system than that of conventional access. It is difficult to establish a strict comparison between the range of charges for conventional Internet access and high-speed Internet access, because of the wide range of options provided by the different ISPs and the scale of offerings not involving subscription and of charges calculated per minute. However, a number of comparison exercises allow some idea to be formed of the price differences between high-speed and low-speed Internet access during the period covered by this Decision.

(189) First, comparison between the average monthly revenue procured by Wanadoo Interactive from all its subscribers, on the one hand, and the average monthly revenue from its ADSL subscribers, on the other, gives an approximation of the difference between the average price for low-speed Internet access and the average price for a high-speed Internet access subscription. In the first half of 2001, Wanadoo Interactive's revenue per Internet customer (including ADSL subscriptions) averaged euro [...]*,211 whereas, at that time, it may be estimated that the average annual revenue from Wanadoo Interactive's ADSL subscribers, weighting the two services in question, was euro [...]* per month and per subscriber, i.e. [...]* as much as the overall average.

(190) A second method can be used for comparing prices, based on the suggestion made by Wanadoo Interactive in its reply of 4 March 2002 to the statement of objections, that the relevant comparison with high-speed Internet access should be packages offering unlimited low-speed Internet access. Unlimited flat-rate packages in the strict sense, apart from the trial initiated by AOL in 2000 and 2001, developed in France only as from August 2002. However, packages offering 20 hours, 30 hours or 40 hours were already available in 2001. If one compares, in the case of each provider, the rate for its high-speed Internet access offering with the charge for its most expensive low-speed Internet access flat-rate or unlimited offering, it may be seen that the ratio is never less than 1.8 and that in the great majority of cases it is well above 2. (212) It may thus be said that, in general, in the period from January 2001 to October 2002, high-speed Internet access offerings for residential customers were approximately twice as expensive as low-speed access offerings which are closest within the range of low-speed access services, (213) not to mention cheaper low-speed access offerings.

(191) Wanadoo Interactive explains in its reply that the present prices for high-speed Internet access offerings are too high and harm the spread of this access mode and deduces from this that prices should fall in the near future. (214) This hypothesis of a future reduction in a price differential cannot be taken as a criterion in assessing the definition of the relevant market in 2001 and 2002.

(192) A price and revenue ratio of at least one to two between two services cannot be regarded as minor. It can be taken as a basis for defining two separate markets.

5. Assessment of the degree of substitutability

(193) Low-speed and high-speed Internet access undoubtedly present some degree of substitutability. According to the market analyses carried out by Wanadoo Interactive's marketing departments, a change in the charges for the packages referred to as "unlimited" or "integral" (20, 30, 40, 50, 60 or 100 hours a month) can result in a change in the growth trajectory of high-speed offerings. Several documents found on the premises of Wanadoo Interactive show that, for example, a subscription charge for the eXtense package of the order of FRF [...]* a month would have adversely impacted the development of that service, to the benefit of the "integral" packages. (215) Wanadoo Interactive points to the risk of "cannibalisation" between flat-rate or unlimited access offerings on the one hand and ADSL offerings on the other in the acquisition of new subscribers. The Commission considers that the effect of the company's argument is merely to show that it is highly probable that subscribers will migrate from low-speed facilities to high-speed Internet access facilities, a matter which it has never doubted.

(194) However, this observation on probable migrations from low-speed to high-speed Internet access does not in itself allow the conclusion to be drawn that there is sufficiently significant substitutability between the services. Although some form of substitutability exists, its operation is extremely asymmetrical, given the value attached by users to the intrinsic features of high-speed Internet access. (216)

(195) In this respect, as will be seen from Table 7, the information obtained from Wanadoo Interactive shows that customer migrations from high-speed to low-speed facilities are [...]* compared with migrations from "integral" to "high-speed" facilities. (217) The rates of migration are very dissimilar. If, however, the products were perfectly substitutable from the point of view of demand, the rates of migration should be identical or at least comparable.

Table 7: migration of Wanadoo subscribers between unlimited offerings and ADSL

<emplacement tableau>

(196) These figures tend to show that the propensity of "integral" subscribers to migrate to "high-speed" facilities is on average [...]* times as great as that of "high-speed" subscribers to migrate to "integral" facilities. Clearly, if there were highly symmetric substitutability between the services, the differences in rates of migration would not have been so disproportionate.

(197) The low rates of migration of ADSL subscribers towards "integral" facilities is all the more revealing as, during the period covered by this Decision, the catalogue price per hour for these flat-rate offerings was generally on a downward trend, (218) while prices for the eXtense and Wanadoo ADSL services remained stable. If Wanadoo subscribers had actually considered the products to be interchangeable, any such change in relative prices should have resulted in a switch to low-speed packages. (219) However, this was not the case, Wanadoo's ADSL offerings continued their strong growth with only limited migratory movements. (220)

(198) In general, Wanadoo subscribers switched to an extremely limited extent from ADSL to other Wanadoo offerings; the figure may be estimated at an average of [...]* % of the customer base per month for the eXtense service between August 2001 and June 2002 and at [...]* % of the customer base per month for the Wanadoo ADSL service between January and June 2002. (221) However, these rates (in particular for Wanadoo ADSL) overestimate switches from ADSL to low-speed, since they include switches between the ADSL products themselves that should be disregarded if a proper assessment is to be established. (222)

(199) On the question of customer transfers from high-speed to low-speed Internet access where there was a moderate price change, the Commission had a survey carried out (223) among a representative sample of residential subscribers (224) to these high-speed Internet access services. (225) The survey involved asking the sample for its reactions to a general high-speed subscription price increase of 10%. The survey shows that 80% of subscribers would keep their high-speed Internet subscription if there were a 10% price increase, 7% would cancel all Internet connections and 13% would switch to low-speed Internet subscription packages, with half of this latter category opting for unlimited access packages. In order to assess whether the increase in prices would be profitable to the company and evaluate more fully the degree of competitive constraint exercised on high-speed services by low-speed services, the impact of any such increase on the margin on recurring variable costs should be examined. On the one hand, in the case of 80% of its subscribers, the company would, through a 10% price increase, achieve a net gain of some euro 3.80 for eXtense and euro 1.72 for Wanadoo ADSL. On the other, the company would loose 20% of its subscribers and the relevant margins. Given the margin levels on recurring variable costs for eXtense and Wanadoo ADSL products from 2001 up to the date of this Decision, the differential between additional profits and profit losses is always positive. (226) Conversely, for this condition not to be met, the level of the margin on recurring costs would, for the two products in question, had to have reached 40%, which has never occurred since ADSL products first appeared and which seems difficult to achieve in the near future given the characteristics of the market.

(200) It must be noted that, amongst the reasons which the Internet users surveyed gave for keeping the subscription, the fact of having entered into a one-year subscription with the service provider, though a very widespread practice, (227) was seen as a decisive constraint by only a very small minority. (228) Nor is there any significant difference in reaction between subscribers who have leased and those who have purchased their modem. Finally, those surveyed cite the unlimited access aspect of the package and speed of connection as the reasons for keeping their subscription despite the price increase, rather than objective constraints (for example, the fact of having taken out the subscription for one year). It would thus seem that users attach a price and a particular value to the specific qualitative features of high-speed Internet access.

(201) The results of the survey referred to in paragraphs 199 and 200 (229) do not allow the conclusion to be drawn that there is sufficient substitutability between low-speed and high-speed offerings for a significant price change to prompt subscribers to switch in large numbers from high-speed to low-speed. On the contrary, the results show that high-speed subscribers have a distinct preference for the convenience and performance offered by high-speed access. The results also invalidate Wanadoo Interactive's suggestion in its reply of 4 March 2002 "that a price increase for high-speed access would contribute to customers switching to low speed". (230) On the contrary, the Commission must conclude that there is no significant interchangeability between the two categories of service.

(202) Lastly, it has to be noted that, quite clearly, Wanadoo Interactive's in-house analyses prompt its relevant departments to take the view (231) that "ADSL is a [...]* (232) [...]* market for unlimited access".

6. Conclusion on the service market

(203) In the light of the analysis it has carried out the Commission considers that the relevant service market to be used in analysing Wanadoo Interactive's conduct is the market for high-speed Internet access for residential customers.

(204) This market definition is broader than that used by the ART. (233) It identifies a highspeed Internet access market for residential customers, (234) but, within that market, distinguishes a high-speed Internet access market using ADSL technology, (235) notably because of the difference in its penetration potential compared with Internet access by cable.

2. The geographic market

(205) The Internet service providers operating on the high-speed Internet access market and the telecommunications operators active on the ADSL services market operate on a national basis. Even if at present some of their high-speed Internet access offerings are limited to specific areas because of technical deployment constraints, it is clear that all the providers pursue the same deployment and growth objectives at national level. Furthermore, prices are set at national level. Internet service providers provide a service aimed at the resident population within the national territory and are subject to national rules and regulations.

(206) The relevant geographic market is accordingly deemed to be the French national market. This definition is also that used by Wanadoo Interactive. (236)

C. THE DOMINANT POSITION

(207) In looking at Wanadoo Interactive's position on the relevant market, the Commission will examine several factors in turn: the market shares held since the beginning of 2001, the effects of the link-up with the France Télécom group, and the pre-eminent position held in conventional Internet access and in directories.

1. General

(208) Wanadoo Interactive points out that the Court of Justice has defined a dominant position as "a position of economic strength enjoyed by an undertaking which enables it to prevent effective competition being maintained on the relevant market by giving it the power to behave to an appreciable extent independently of its competitors, customers and ultimately of its consumers". (237) According to Wanadoo Interactive, the charges it had decided to make were determined by the attitude of its competitors, and the company was in reality "under pressure from existing and potential competitors". (238)According to this analysis, the level of prices charged is itself a demonstration of the fact of this competitive pressure.

(209) The Commission considers that Wanadoo Interactive misinterprets the concept of behaving independently referred to in the Court's judgment. The power to behave independently does not mean that the dominant undertaking is in a position of complete invulnerability vis-à-vis existing or potential competitors. Any such invulnerability requirement would ultimately render the very concept of abuse of a dominant position irrelevant, since, in that case, the dominant undertaking, protected from any competitive risk, would have no need to engage in practices designed to eliminate rival firms. Enjoying a dominant position does not mean that there is no competitive pressure. The dominant position simply enables the relevant undertaking "if not to determine, at least to have an appreciable influence on the conditions under which that competition will develop, and in any case to act largely in disregard of it so long as such conduct does not operate to its detriment". (239) The Commission considers that, in this case, the various elements of dominance described in paragraphs 211 to 252 gave Wanadoo Interactive the means to engage in a large-scale market penetration strategy that was untenable for its competitors, which were unable to follow suit.

(210) Wanadoo Interactive made a number of points on some of the dominance factors identified by the Commission, its view being that none of the factors taken on its own was evidence of dominance. The Commission would point out at the outset, before looking in detail at the points raised by the company, that it did not itself, in its analysis, look at any single factor, but examined a combination of elements which, taken together, are such as to give Wanadoo Interactive a dominant position on the relevant market.

2. Market shares

(211) Table 8 shows the growth of Wanadoo Interactive's market share since the beginning of 2001 amongst ADSL and cable modem high-speed Internet service providers. (240) Wanadoo Interactive's market share includes both Wanadoo Interactive's ADSL subscribers and Câble Wanadoo's subscribers. In its own market analyses, the company includes subscribers to Câble Wanadoo, which carries its brand name, although technically the service is provided by France Télécom Câble. (241)

Table 8: Wanadoo Interactive's market shares on the high-speed market

<emplacement tableau>

(212) Very large market shares, in excess of 50%, must be regarded as serious, and indeed sufficient, evidence of the existence of a dominant position, save in exceptional circumstances. (242) However, since the market was still at a fairly immature stage at the beginning of 2001, two other key considerations must be taken into account in addition to absolute market shares.

(213) Wanadoo Interactive's relative position vis-à-vis its next largest competitors needs to be examined first. (243) On 1 January 2001, Wanadoo Interactive had [...]*times as many subscribers as its largest competitor, the cable operator Noos. (244) As from the second half of 2001, Wanadoo Interactive opened up a considerable gap vis-à-vis its largest competitor. In the first quarter of 2001, its growth rate was more than [...]*times that of Noos. At the end of March, Wanadoo Interactive had [...]* times as many subscribers and at the end of June [...]* times as many. One year later, at the end of June 2002, Wanadoo Interactive had about [...]*times as many subscribers as Noos. As far as its other competitors were concerned, whether cable modem or ADSL service providers, none of them ever exceeded a market share of 8% during the period under review. Thus, as will be seen from Table 9, not only did Wanadoo Interactive have a clear lead over its largest competitor throughout the period, it also widened its lead over it to a striking extent.

Table 9: ratio between Wanadoo Interactive's market shares and those of its main competitor

<emplacement tableau>

(214) In addition, particular attention needs to be given, within the relevant market, to the market segment involving ADSL offerings. The cable modem offerings segment has less potential in terms of geographical deployment capacity than the ADSL segment. (245) If one looks only at the segment involving ADSL Internet access services for residential customers, Wanadoo Interactive's market shares were very large: consistently close to [90-100]* % from the beginning of 2001 to the end of the summer 2002, and still [80-90]* % in August 2002. (246) As will be seen from Table 10, since the end of 2000 Wanadoo Interactive has had more than [...]* times as many ADSL subscribers as its largest competitor in this segment.

Table 10: ratio between Wanadoo Interactive's market shares and those of its main competitor in the <emplacement tableau>

(215) Relative market shares, and the relationship between the market shares of the undertaking concerned and of its competitors, especially those of the next largest, are held by the Court of Justice to be relevant factors in assessing a dominant position. (248) Consequently, the market shares held by Wanadoo Interactive since the beginning of 2001 are, in view of the above considerations, serious evidence of dominance.

(216) Wanadoo Interactive considers that market shares do not provide reliable evidence in the context of an emerging market characterised by a still limited customer base. Wanadoo argues that it is necessary to adopt a forward-looking approach on market shares. In addition to this assessment on the validity of market shares, Wanadoo Interactive expresses doubts on whether a dominant position can be determined on an emerging market (249) and argues that the Internet access sector is a part of the new economy in which market shares provide only very unreliable evidence. (250) These remarks call for the following comments.

(217) Firstly, as a general point, the fact that a sector is a growth sector does not in itself mean that it is not covered by the competition rules, and in particular Article 82 of the Treaty. Analysis of the market position must of course take account not only of market shares, but also of the specifics of a dynamic market. This is precisely what the Commission set out to do in this Decision. However, the Commission cannot debar itself from examining the position of an undertaking on a growth market pending its final consolidation, since this would mean ex post acceptance of any abuses committed.

(218) Secondly, while it cannot be ruled out that the term "emerging market" may properly be applied to the high speed Internet access market, (251) the Commission takes the view, at all events, that it is impossible to regard the relevant market as entirely new or in a phase of pure experimentation. At the end of June 2000, the high-speed Internet access market for residential customers already had some 100 000 subscribers, and at the end of 2000, there were more than 180 000 subscribers in France. As from the first quarter of 2001, the market grew by more than 5 000 new subscribers a week. Although the term emerging market may be used to describe this key period, it is quite clear that, at that date, high-speed Internet access services had gone well beyond the stage of simple commercial or technical experimentation. In other countries, high-speed Internet access services already had a large number of subscribers at the end of 2000. Germany, for example, had nearly 450 000 subscribers to high-speed Internet access services at the end of 2000, while the United States had more than 5 million. (252) In addition, although high-speed does constitute a separate market, Internet access services and certain uses linked to high-speed services were not new. The high-speed market developed in France as from 1997 and as an extension of an existing market, that of low-speed Internet access offerings. At the end of 2000, there were some 5 million Internet subscribers in France. Even though high-speed Internet access displayed some features of an emerging market, it cannot be asserted that Wanadoo Interactive's development on the relevant market as from 2001 took place against a background of uncertainty and instability associated with an entirely new type of product.

(219) Thirdly, the risk of a fluctuating market share that could, according to the company, be a feature of a market in a high growth phase does not seem to apply to Wanadoo Interactive in the case in point. The company's market share grew consistently and steadily throughout the high-speed market and the company's penetration on the ADSL offerings segment is surprisingly stable over the period. There is thus no evidence of wide and erratic fluctuations in market shares on the high-speed market in France as from the beginning of 2001, a feature which would indicate that there was no dominant position.

(220) Fourthly, Wanadoo Interactive suggests that a forward-looking or dynamic analysis of market shares should be used in addition to the observation of the market shares that existed in 2001 and at the beginning of 2002. According to Wanadoo Interactive, not only current competition, but also potential competition from future competitors should be examined. To understand the company's perception of the reality and scale of the threat posed by future competitors, reference may be made to the simulations carried out by Wanadoo Interactive itself. In March 2001, the company foresaw a customer base of [>1.5]* million by the end of 2004, both ADSL and cable access. (253) For the same period, according to internal company documents, the number of subscribers for the high-speed access market could be estimated at [>2.5]* million by the end of 2004. (254) Wanadoo Interactive's forecast was thus equivalent to [50-60]* % of the whole of the market at the end of 2004. This market penetration forecast was re-assessed by Wanadoo Interactive in June 2001 at [...]* million customers by the end of 2004, not including cable access. Wanadoo Interactive thus believed at that time that it would have [...]* of the ADSL segment at the end of 2004, and thus at least [50- 60]* % of the residential high-speed Internet access market. (255) At the time when these forecasts were made, the company did not foresee any significant erosion at all in its share of the high-speed market. On the contrary, the company was forecasting a steady increase in its market shares in ADSL. (256) Wanadoo Interactive's own forward-looking analyses tend to show that the company was confident in its capacity to hold on to its very high market shares in the medium term and did not feel any threat to its dominant position. At all events, a slight decline in market shares which are still very high cannot constitute proof of the absence of a dominant position. (257)

(221) Lastly, in its reply of 23 October 2002, Wanadoo Interactive cites the marketing campaigns carried out by its competitors at the time of the Internet Festival in March 2002. However, the company does not prove that these marketing campaigns could have had a destabilising effect on the market shares which it held. On the contrary, on the basis of the figures set out in Table 8, it may be seen that the company's market share remained intact in the first half of 2002, despite the efforts of its competitors. Furthermore, the only credible competitor explicitly cited by Wanadoo Interactive is T-Online France, which operated Club Internet, whose "declared intention is to take 15% of this segment [ADSL]". (258) Even if T-Online France were at some time in the future to win 15% of the ADSL segment on the high-speed market, Wanadoo Interactive's position does not seem to be such as to be undermined by this level of penetration. In its reply of 23 October 2002, Wanadoo Interactive also describes the campaigns and objectives of its competitors, on the basis of their advertising in the autumn of 2002. In actual fact, even during this period in which its competitors made exceptional efforts, Wanadoo Interactive accounted for more than [50-60]* % of the growth of the high-speed market. Although the marketing campaigns of competitors in a context of healthy competition are liable to lead to some erosion of the relevant company's market shares, its dominance in terms of market shares on the relevant market does not seem threatened.

(222) In view of the above, it cannot be argued that there was any volatility, fragility or instability in Wanadoo Interactive's position on an emerging market that would invalidate an examination of market shares and their inclusion amongst the factors determining dominance.

3. The link-up with France Télécom

(223) Amongst Internet service providers in France, Wanadoo Interactive occupies a special position because it forms part of the France Télécom group. The fact that Wanadoo Interactive belongs to the France Télécom group gives it commercial and technical deployment facilities and potential financial support which are liable to be of crucial importance in gaining massive penetration of a developing market involving a new type of activity.

(224) As a general point, Wanadoo Interactive disputes that it benefits from the support provided by France Télécom to its subsidiary, pointing out that other providers of high-speed Internet access are backed by powerful global companies. (259) More specifically, Wanadoo Interactive cites the case of AOL France and T-Online France, which form part of the AOL-Time Warner and Deutsche Telekom groups respectively, and Tiscali.

(225) It should be noted in this connection that, according to the Court of Justice, the size, financial strength and degree of diversification of competitors at world level do not necessarily deprive the dominant undertaking of its privileged position. (260) The situation of undertakings affected by Wanadoo Interactive must therefore be analysed in the light of the circumstances of the present case.

(226) Tiscali is a group which specialises in Internet access, but does not have the diversified resources that would give it some security of revenue, in contrast to the France Télécom group. T-Online France is a subsidiary of Deutsche Telekom, but Deutsche Telekom does not have a position on the relevant geographic market comparable to that of Wanadoo's parent company, with the result that its backing is necessarily only financial. AOL France is linked up to a group with a relatively diversified portfolio, but one which until the end of 2002 held only a very weak position on the high-speed Internet access market in France. (261)

(227) Wanadoo Interactive's argument regarding the claimed power of its competitors and their strategic support from large international groups is all the less relevant as none of them occupies a pre-eminent position on the low-speed market, which constitutes one of the possible growth reservoirs of the high-speed market. On the contrary, Wanadoo Interactive is by far the leading low-speed Internet access provider in France. Wanadoo Interactive's share of the market for low-speed Internet access was estimated in 2001 at between [30-40]* %. (262) By comparison, the next largest competitor on the low-speed market, AOL, held scarcely more than [10-20]% of the market at that date, i.e. [...]* less than Wanadoo Interactive. (263) A survey of a sample of Internet users published in November 2001 on the question "Which is your main ISP?" gave Wanadoo 39%, AOL 16%, Tiscali 9% and Club Internet 6.5%. (264) Thus, the positions held on the low-speed market by the relevant competitors were insufficient to act as a counterweight to the advantages enjoyed by Wanadoo Interactive on the relevant market.

(228) Whatever the willingness of the relevant groups to back the investments and commercial initiatives of their French subsidiaries, none of them could pretend to provide their subsidiaries with technical backing and backing in terms of commercial network in France on such a decisive scale as those provided by France Télécom to Wanadoo Interactive, as outlined in paragraphs 229 to 246.

1. Synergies within a large group

(229) In its stock exchange listing prospectus issued in July 2000, Wanadoo described the general advantages it would acquire from the support of the France Télécom group as follows: "The Wanadoo group benefits from strong synergies with France Télécom. At distribution level, France Télécom distributes the Wanadoo connection kits in its 700 France Télécom sales outlets and its strong regional representation ensures that the group's name is well known. The group benefits directly from France Télécom's knowhow in marketing to the general public and industry. The group draws on France Télécom's infrastructure in the IP area and in France Télécom's investments in promoting the new access technologies, such as (...) ADSL. Lastly, the group benefits from the formidable reservoir of professional know-how which France Télécom represents". (265)

(230) In addition to this general statement by the company itself, a more precise description should be given of the nature of the benefits which Wanadoo Interactive derives from the link-up with France Télécom in the deployment of its ADSL offerings for the mass residential market.

2. Technical support

(231) In technical terms, Wanadoo Interactive benefits from the support of France Télécom's facilities. France Télécom is the incumbent telecommunications operator in France. It operates either directly or indirectly, through its subsidiary Transpac, long-distance networks in France which are used to carry Internet traffic. Above all, France Télécom is the owner of the local telecommunications access network linking all telephone subscribers to its network. The use of France Télécom's local access network is essential in order to provide an ADSL service. Throughout the period covered by this Decision, France Télécom controlled virtually all the ADSL connections in France.

(232) Specifically, the technical support of France Télécom gave Wanadoo Interactive a considerable competitive edge throughout 2000 and the first seven months of 2001. Unlike the other Internet service providers using ADSL technology for high-speed Internet access offerings, Wanadoo Interactive received preferential treatment from the incumbent telecommunications operator throughout that period. France Télécom offered Wanadoo Interactive a bespoke national and regional IP routing facility as from the end of 1999. During the same period, the other Internet service providers were offered only the Turbo IP routing facility, which imposed greater constraints on them since it obliged them to have points of presence in each of France Télécom's 41 geographic areas in order to achieve maximum territorial coverage. (266) Consequently, the only constraint on Wanadoo Interactive's geographical presence was the rate of roll-out of ADSL technology by France Télécom, whereas the other ADSL service providers had to establish numerous points of presence in order to be present outside Paris, which few of them were prepared to do. [...]* (267) [...]* (268) During the same period, Wanadoo Interactive's competitors, and notably Noos, were represented almost exclusively in the Paris region, or indeed, in the case of providers using ADSL technology, in just a few districts of the capital.

(233) The bespoke facility finally disappeared in August 2001, with the entry into force (269) on 20 July 2001 of tariff decision No 2001480 concerning developments in IP/ADSL and Collecte IP/ADSL offerings and the extension to all Internet service providers of the facility for national and regional IP/ADSL routing. However, in the spring of 2001 as in the summer of 2001, Wanadoo Interactive was confident of France Télécom's willingness to grant it a second bespoke facility to reduce its production costs if need be. (270)

(234) Another technical advantage lies in Wanadoo Interactive's ability to have real-time access to information allowing potential customers' addresses to be linked up with France Télécom's corresponding distributors, so that Wanadoo Interactive was able to inform any interested customers within a very short period of time whether the area within which they lived was covered by an ADSL service. Wanadoo Interactive's competitors, by contrast, have to send a specific request to France Télécom to establish potential customers' eligibility for ADSL. This means a loss of time for them which may make it more difficult to clinch the deal with the potential customer. (271)

(235) The technical advantages referred to in paragraphs 232 to 234 (bespoke facility and access to files on high-speed-convertible lines) reflect the vertical integration between Wanadoo Interactive and France Télécom, which is far and away its main supplier. The Court of Justice has held that a vertical integration effect could be regarded as one of the factors contributing to a dominant position. (272) The fact that, in the case in point, Wanadoo Interactive is legally a separate entity distinct from France Télécom does not detract from the effects of the vertical integration of the group as a whole.

(236) Wanadoo Interactive has not disputed the fact that the bespoke service provided in 2001 could have constituted a major advantage in the roll-out of its services. However, it did dispute the fact that access to knowledge of France Télécom's ADSL network and to the eligibility of phone lines could have constituted an advantage. It considers that, as from October 2001, all competitors had the use of a server of equivalent eligibility. It acknowledges that the Competition Council in France decided on interim measures in February 2000, (273) but emphasises that that decision does not affect the substance. The Commission for its part considers that the conditions relating to the eligibility server have no longer been discriminatory since July 2002, when the French Competition Council held that the interim measures it had imposed had been complied with. The fact remains that Wanadoo's privileged treatment continued for almost a year and a half, at an important stage in the development of the market.

(237) France Télécom's technical support gives Wanadoo Interactive an advantage to which its competitors do not have access, which must be regarded as constituting a strategic entry barrier and, consequently, as a factor of dominance.

3. Commercial support

(238) Wanadoo Interactive benefits from the commercial backing of France Télécom, whose network of agencies gives it a comprehensive physical presence throughout the national territory. The marketing of the relevant products does of course entail remuneration for France Télécom's sales outlets. The fact remains, however, that the presence of Wanadoo's name in France Télécom's 700 commercial agencies and the exclusivity enjoyed by its ADSL offerings within this network gave it a degree of visibility which allowed it easily to reach potential customers throughout France. (274) The density of the network makes it easier to identify sales prospects, on the one hand, and to guide customers interested in ADSL services to Wanadoo Interactive's Internet access offerings on the other. This aspect is all the more important as, in order to be connected to a "dissociated" ADSL service, and regardless of which Internet service provider they initially envisage being connected to, consumers are in all cases obliged to apply to the France Télécom agency to which they are attached for the provision of an ADSL line. (275) When they do so, it is quite possible that, in practice, and especially since the marketing of the eXtense package combining an ADSL line and Internet access provided by Wanadoo, [...]*

(239) In order to gain a clearer idea of the advantage afforded by this commercial presence in agencies, one may compare the network of 700 France Télécom agencies in which Wanadoo Interactive was represented with the 26 sales outlets belonging to Noos, its main competitor at the beginning of 2001. (276) The distribution medium did expand during the course of 2001, particularly as from September 2001. At the end of 2001, Wanadoo Interactive's high-speed services were marketed in [...]* sales outlets, both France Télécom shops and large-scale retail outlets and specialised retailers. (277) Competitors also gradually developed their sales outlet networks. The fact remains, however, that at the beginning of 2002, the leading competitor, Noos, was represented in fewer than [...]* sales outlets, (278) i.e. more than [...]* times fewer than Wanadoo Interactive. (279)

(240) Furthermore, the efficiency of the distribution networks accessible to competitors is much lower than that provided by the network of France Télécom's agencies. None of Wanadoo Interactive's competitors enjoys an exclusive relationship with a distributor dedicated to telecommunications services and Internet access comparable to the exclusive relationship between France Télécom and Wanadoo Interactive. [...]*: (280) the exclusivity enjoyed by Wanadoo Interactive's high-speed services in France Télécom's agencies thus represented a very significant means of dominating the market.

(241) Thus, throughout 2001 France Télécom's agency network provided crucial support for the marketing of Wanadoo Interactive's ADSL services, particularly the eXtense package. During 2001, several in-house company documents reflect the very strong commitment of France Télécom's agencies to the eXtense product. (281) At the beginning of 2001, the company estimated that, for 2001 as a whole, France Télécom's agencies would place between [...]* % of the eXtense packages and ADSL subscriptions sold, (282) i.e. a proportion which is very much higher than the average observed for products in the Wanadoo Interactive range as a whole. (283) At the end of 2001, Wanadoo Interactive still regarded the France Télécom agencies as the main channel for distributing ADSL products on the residential market, with [...]* % of the sales anticipated in 2002. (284)

(242) It must also be noted, as far as this commercial support is concerned, that Wanadoo Interactive benefits indirectly from France Télécom's advertising relating to highspeed Internet access. Its advertising tries to persuade potential consumers to apply to their France Télécom agency, where they are likely to be channelled towards Wanadoo Interactive's services. (285) France Télécom is reported to have spent some euro [...]* on press and radio advertising of this type in 2001, (286) alongside the approximately euro [...]* spent by Wanadoo Interactive on ADSL advertising campaigns.

(243) In Michelin, the Court of Justice considered the question of commercial networks giving direct access to consumers and the possibilities of commercial synergies with other elements in the product range of a company or of the group to which it belongs as a factor contributing to the establishment of a dominant position. (287) In the case being examined here, the support which Wanadoo Interactive derives from its exclusive presence in France Télécom's agencies is a particularly important factor liable to underpin its dominant position, in a situation where products have entered a phase of rapid growth. The cost and time required for a company to set up its own distribution network or a network involving an exclusive commercial relationship comparable to that enjoyed by Wanadoo must be regarded as an obstacle to market penetration by competitors and as an important factor in Wanadoo's dominant position, particularly at the start of the period under review.

4. Logistical and financial support

(244) Lastly, Wanadoo Interactive benefits from logistical and financial support from France Télécom. A look at the way in which eXtense starter packs are acquired and marketed is revealing. France Télécom buys the modems and puts together the starter packs, placing orders with different suppliers. [...]* (288) The assumption of these cost items by France Télécom during the period between the entry of the starter packs into stock and their sale to customers, which is generally several weeks, represents a significant advantage for Wanadoo Interactive, since it reduces its working capital requirement, an advantage which its competitors did not enjoy.

(245) More anecdotally, as Wanadoo Interactive's main supplier, France Télécom allowed Wanadoo Interactive favourable payment terms [...]* (289) whereas the other Internet service providers were invoiced by France Télécom at regular dates, or indeed in some cases in advance. (290) Wanadoo Interactive [...]*confined itself to noting that the IP/ADSL access service had been invoiced late in 2001 to a number of competitors, (291) which was not relevant to the issue raised by the Commission.

(246) The Commission considers that the logistical support provided for the procurement of eXtense packs, the savings in terms of working capital requirements for marketing them and the invoicing periods which France Télécom occasionally applied to its subsidiary constitute, in the context of a growth market, very appreciable advantages that contribute to Wanadoo Interactive's dominant position.

4. Wanadoo Interactive's position on the directory publishing markets

(247) In 2000, France Télécom brought together within Wanadoo its telephone directory publishing activities and its activities involving the sale of advertising space in business directories, which had previously been divided between its own departments and the Office d'Annonces (ODA). These activities bring in very considerable revenue and are very profitable activities of the Wanadoo group. In a decision adopted in March 1998, the Competition Council took the view that France Télécom and the ODA held a dominant position respectively on the business directory publishing market and on the market for the sale of advertising space in business directories. (292)

(248) Although, for the purposes of this proceeding, it is not necessary to examine whether Wanadoo SA and its yellow pages (Pages Jaunes) subsidiary, the successors of France Télécom and the ODA in the relevant activities, hold a dominant position, it should be noted that the inclusion of these highly profitable activities in the group's portfolio is liable to considerably reduce the impact which the Wanadoo Interactive subsidiary's policy of selling at a loss on the high-speed Internet access market has on the group's worth and stock exchange value. In 2000 and 2001, Wanadoo achieved, in its "directories and services for business" segment results before depreciation of respectively euro 225 million and euro 238 million. (293) Quite clearly, Wanadoo Interactive would have viewed the very heavy losses recorded on the high-speed Internet access market in a very different light if the impact of this severe loss had not been offset, at Wanadoo group level, by the very substantial profits earned from the directory publishing business, and in particular from the sale of advertising space. The composition of the Wanadoo group's business portfolio thus gives it considerable financial power that helps to establish its dominant position on the relevant market.

(249) Wanadoo Interactive disputed the advantage associated with the revenue earned for the group from its directory business. On the contrary, it takes the view that the only real item that counts is the euro [...]* raised when Wanadoo was introduced on the stock exchange in 2000. (294)

(250) The Commission [...]* considers that the cash raised in 2000 represented a very appreciable advantage in providing financial underpinning for the loss-making highspeed Internet access business. Wanadoo Interactive claims that its competitors are "all linked up with financially powerful groups", (295) without providing any factual evidence for its assertion. The Commission considers, by contrast, that this level of cash assets gave rise to substantial financial asymmetry in relation to certain operators such as Mangoosta.

(251) Furthermore, the Commission considers that the level of cash assets at the beginning of the period is not the only parameter to be taken into consideration. A company cannot manage its cash without taking account of the manner in which the level of its cash assets is affected by its current operating conditions. Thus, the room for manœuvre afforded to Wanadoo by its cash assets would have been less if the deterioration in its cash created by its loss-making Internet activities had not been offset by the net cash contributions generated by a profit-making activity.

(252) The Commission considers that the benefits deriving from the directory publishing business were such as to reinforce the dominant position on the market for high-speed Internet access, particularly during a period of cyclical downturn and the drying-up of the financial resources available for the Internet sector. (296)

5. Conclusion

(253) A dominant position may derive from a combination of several factors which, taken separately, are not necessarily determinative. (297) In the case under examination here, the Commission considers that the market shares held by Wanadoo Interactive since the beginning of 2001, the multiple affects of the link-up with the France Télécom group and the position held by the Wanadoo group on the directories market are, taken together, liable to give Wanadoo Interactive a dominant position on the French market for high-speed Internet access for residential customers.

D. ABUSE OF A DOMINANT POSITION

(254) This section shows how the below-cost pricing practised by Wanadoo Interactive formed part of a deliberate strategy of predation aimed at preempting a burgeoning market, and hence in what way it constitutes an abuse of a dominant position within the meaning of Article 82 of the Treaty.

(255) The arguments advanced by Wanadoo Interactive concern the applicability of Article 82 of the Treaty to an emerging sector, the objective justification for below-cost pricing, and the question of barriers to entry and the recoupment of initial losses. They are discussed in detail in this section.

1. Nature of the abuse: the non-recovery of costs as part of a plan to preempt the market

(256) Community case-law has given rise to two separate methods of analysis when it comes to examining whether an undertaking has practised predatory pricing, one based on variable costs, and the other on full costs. (298) Thus, the existence of predatory prices is established in the following situations:

- the non-recovery of average variable costs per unit, the establishment of which in itself suffices to justify a finding of abuse;

- the non-recovery of average full costs where this is accompanied by a plan indicative of an intention to eliminate competitors.

1. The recovery of costs

(257) On the basis of the information set forth in section I.F, broadly speaking a distinction can be made between three separate periods since the beginning of 2001:

- from 1 January to 31 July 2001, Wanadoo Interactive was far from recovering the (adjusted) variable costs of the services at issue;

- from 1 August 2001 to 15 October 2002, Wanadoo Interactive came close to satisfying the test applied by the Commission, but without ever actually doing so: Wanadoo did not recover its (adjusted) full costs; (299) nor - although it did recover its (outturned) adjusted variable costs for part of the period - was it able before March 2002 to forecast the achievement of such recovery in advance;

- from 15 October 2002 onwards, Wanadoo Interactive clearly satisfied the cost recovery test applied by the Commission, both for full costs and for variable costs, even though its instantaneous revenue may have been lower than its accounting costs.

(258) There can be no doubt about the lack of any short-term rationality in market deployment for the first seven months of 2001.

(259) For the following period, although the accounting costs were never recovered and until October 2002 each additional month saw a further worsening of the cumulative result for the activity, the company was hovering close to, but always slightly on the wrong side of, the borders of economic rationality as referred to by the Commission in this Decision. Its forecasts did not allow it to take a favourable view of the short-term financial impact of the commercial development of its ADSL products before France Télécom's new routing and access charges came into force in 2002. Only the introduction of the new France Télécom charges on 15 October 2002, coupled with a decision by the telecoms operator to offer all Internet service providers free access to its services, enabled Wanadoo Interactive to satisfy without ambiguity the test used by the Commission and to experience short-term financial rationality.

(260) Wanadoo Interactive has criticised the Commission for having overlooked in its analysis the features peculiar to the launching of a new product.

(261) The criticism is unfounded. In its analysis, the Commission has on the contrary made a major adjustment to the application of the Akzo predation test in the sense of greater flexibility.

(262) First, instead of simply examining costs and revenue as entered in the company's accounts, as the Akzo judgment suggests it might do, the Commission has spread customer acquisition costs over 48 months, treating as it were these costs as a commercial investment to be written off over a customer's realistic lifetime. The flexibility thus introduced into the application of the Akzo test is significant in that it has the effect of greatly softening the impact of customer acquisition costs, which nevertheless accounted for [...]* % of the costs incurred by Wanadoo Interactive in 2001. (300) Moreover, in its assessment of the revenue generated by subscriptions, the Commission has taken into account a theoretical, nominal subscription revenue, and not the average revenue actually observed, which is constantly between [...]* % and [...]* % below the theoretical revenue. In so doing, it has neutralised the mechanical effect linked to the growth in the subscriber base explained in paragraph 29. The combined effect of these two corrections is to improve by more than [10-30]* points the cost recovery rates used in this Decision compared with what they would have been had they simply been based on the unadjusted accounting data.

(263) Moreover, the Commission has ignored the cost of capital in its analysis. Lastly, as regards the assessment of certain cost components, when in doubt the Commission has always worked on the basis of whichever assumptions are most favourable to the company concerned.

(264) The method of assessing the recovery of the adjusted costs has thus proved entirely respectful of the growth market context and has made for a much more nuanced approach than would have been achieved simply by noting the accounting costs recorded during the period. The Commission considers that it has gone as far as it can in the present case in adapting the method without distorting the logic behind a predation test.

2. The questioning by the company of the interpretation of its pricing practice

(265) In the opinion of Wanadoo Interactive, the present case lacks several ingredients which alone would enable the strategy it pursued to be characterised as predatory. There would above all have to have been a sudden fall in prices, that fall would have to have been selective and have favoured a given category of customer, and it would have to have been accompanied by signs of predation on the market, that is to say, by threats against competitors to dissuade them from operating on the market. (301) Lastly, the fall would have to have been sufficiently long lasting to have had an impact.

(266) The Commission considers that the legal precedents do not cover every possible predation scenario. Predation can take forms other than the radical elimination and wholesale ousting of competitors from the market. (302) More generally, predation may simply consist in dictating or inhibiting the competitive behaviour of an existing or potential rival. (303) Here, a predatory price is simply one which leads to a maximisation of profits through its exclusionary or other anticompetitive effects.

(267) Wanadoo Interactive evokes the need for a fall in prices before one can talk of predation. The Commission considers that the argument as to the need for a fall in prices is largely fallacious inasmuch as what is involved here is the market roll-out of new products, the focus being therefore more on below-cost pricing. (304) From a strictly factual point of view, moreover, Wanadoo Interactive's argument is questionable in the present case: the launch of packaged offerings at the end of 2000 meant a further fall in costs for the subscriber of the order of 21% compared with the separate ADSL offerings that existed previously. (305)

(268) Economic doctrine in no way limits predatory behaviour to selective strategies aimed at winning back a specific customer. Such a restriction of the analysis would be meaningless in the present case. The relevant market is a mass residential market. It is therefore by no means a market for the supplying of a limited number of large customers each generating a substantial turnover as in the precedents Wanadoo Interactive cites in its defence, which contain scenarios where selective pricing may be rational.

(269) Lastly, according to Wanadoo Interactive, the price of its ADSL services never constituted an element in its growth strategy, any more than it did an instrument wielded with a view to eliminating competitors. Wanadoo Interactive thus points out that its 2001 and 2002 marketing plans made no mention of prices, but focused instead on other strategic factors. The price of services constituted as it were an exogenous market variable and not a component of the company's strategy. In its pricing policy, Wanadoo was, so it claimed, guided only by the desire to make its offerings profitable. (306)

(270) The Commission considers that, on the contrary, several internal company documents establish a link between the level of prices and Wanadoo Interactive's growth strategy, both at the time of the discussions prior to the setting of the price and during subsequent talks on possible price adjustments held throughout 2001. In a letter of July 2000, [...]* (307) [...]* (308) A month earlier, [...]* (309) [...]* The Commission considers, however, that what matters in the present case is not so much the level of prices per se as the difference between prices and costs.

3. The intention to drive out competition

(271) In the Commission's view, proof of the intention to drive out competition may take the form of internal documents of the dominant undertaking which reveal the existence of a plan to damage competitors.310 In the absence of documents providing direct evidence, the intention to eliminate competitors may be deduced from the circumstances of the case and from a series of important and convergent factors. (311)

(a) The plan to preempt the market

(272) The internal company documents described in section I.G.1 refer to a deliberate strategy of preemption by Wanadoo Interactive of the high-speed market, and in particular of the ADSL segment.

(273) The Commission considers that, in the context of a new market which has entered a period of dynamic growth, and in view of the advantages derived by Wanadoo from its membership of the France Télécom group, this preemption plan must itself be treated as an intention to drive out competition.

(b) The choice of a prices and volumes strategy not replicable by the competition

(274) The intention to contain competition is apparent also from the ex ante assessment by Wanadoo Interactive of the conditions of profitability of the services in question at a time when the company was fixing its market penetration objectives. During the first seven months of 2001, Wanadoo Interactive deliberately set its charges for and promoted its sales of high-speed Internet access services knowing full well that the prices were much lower than the variable costs. (312) As can be seen from the provisional profit and loss account for 2001 and various exchanges within the company, when, in June-July 2001, Wanadoo Interactive revised its quantitative sales targets for the second half of 2001 strongly upwards, (313) it was likewise not in a position to envisage a positive cash flow before the end of the year (314) despite [...]* Similarly, when it fixed the quantitative objectives for 2002, again during summer 2001 - then set at a level [10-20]* % higher than originally envisaged (315) - Wanadoo Interactive was unable to reach a favourable conclusion about the viability of ADSL, at least not for the first few months of 2002, on the basis of the information and forecasts then available.

(275) Clearly, given their degree of knowledge of the cost of the services of IP/ADSL access and IP routing of ADSL traffic, Wanadoo Interactive's competitors in the ADSL segment were a fortiori not in a position to envisage recovering to any greater extent their variable costs throughout the latter part of 2001. It was only from December 2001 at the earliest (316) that it gradually dawned on competitors that, on an outturned cost basis, the economics of ADSL were less unfavourable to them than expected owing to the rules for measuring traffic consumption finally adopted by France Télécom, rules which the terms of the IP/ADSL routing contract did not make it possible to foresee.

(276) Knowledge of the cost of France Télécom's services was so piecemeal at the beginning of 2002, even among well-informed observers, that, in its opinion of 30 April 2002, the ART still considered service providers' payments to France Télécom for ADSL access and the routing of ADSL traffic to be higher than their revenue. (317)

(277) In the analysis of Wanadoo Interactive's intentions towards its competitors, the ex ante perception by the company itself of the profitability prospects at the time of fixing of its quantitative objectives is more important than the outturn, which was not finally known in its definitive form until several weeks after the 2001 financial year had ended. It was in fact this perception of economic conditions throughout 2001 that governed how competitors in the ADSL segment viewed Wanadoo Interactive's determination to frustrate their expectations of growth, market share acquisition or restoration of their margins. (318)

(278) It was clear throughout 2001 that Wanadoo Interactive had deliberately chosen a growth strategy in relation to high-speed access which was problematical as far as its own results were concerned. The company thus clearly and knowingly made a tradeoff between an objective of minimal profitability which would have made it possible to recoup within a reasonable period the cost of acquiring customers, and an objective of ambitious market penetration to the detriment of its competitors.

(279) In the course of 2001, Wanadoo Interactive observed on several occasions that its own strategy, which combined unprofitable charges with significant sales volumes, was economically unsustainable for its competitors and hence could not be reproduced, at least not on a comparable scale. [...]* (319) [...]* (320) It also knew that the reason why a competitor like AOL, a major operator in the low-speed market, was not entering the high-speed market was that it was impossible to emulate Wanadoo's retail tariffs without incurring losses. (321)

(280) [...]* its competitors in the ADSL segment were systematically several profitability points behind it. This disadvantage for competitors was mainly due to the structure of routing service price scales, which was unfavourable to competitors with fewer than 20 000-30 000 subscribers, and to the need for service providers serving a small number of subscribers to spread connection costs (which are fixed in the short term) over a limited customer base. These factors had been diagnosed by Wanadoo Interactive, which had carried out detailed estimates of the gross margin gap according to traffic throughput volumes, highlighting the advantages it enjoyed as market leader. (322) The cost disadvantage suffered by competitors also stems from other, less readily quantifiable factors. For example, unlike Wanadoo Interactive, not all Internet service providers have access to financial facilities limiting the need to finance lossmaking activities externally, and they therefore had to add to their operating costs the financial expenses resulting from the losses incurred. (323)

(281) For all these reasons, while Wanadoo Interactive itself was unable to recover its adjusted full costs, not only did its competitors not recover their adjusted full costs but also there was every chance of their not recovering their variable costs either. Even when Wanadoo Interactive expected, at the end of 2001, to secure a slightly positive margin on its production costs the following year, (324) it was clear that its competitors would be unable to recover their variable costs and hence give economic meaning to the development of their high-speed business.

(282) It is accordingly considered that throughout 2001 Wanadoo Interactive's strategy was aimed at recruiting the largest possible number of subscribers despite the fact that the economic conditions were not in place. [...]* Wanadoo Interactive's strategy thus had the effect of eliminating or marginalising competitors incapable of aligning themselves on it. The strategy was continued with in 2002, [...]* (325)

(283) Notwithstanding this, Wanadoo Interactive did not alter its commercial policy one iota. Prices remained unchanged. (326) The quantitative market penetration objectives were not modified. [...]* (327) Thus, despite Wanadoo Interactive having received the Commission's first statement of objections a few weeks earlier, the idea of going allout to win customers was still being considered for subsequent years.

(284) In this context, the announcement by Wanadoo Interactive in 2001 and at the beginning of 2002 (328) of particularly ambitious commercial objectives, unattainable by a non-dominant undertaking under the unfavourable profitability conditions then prevailing, had the effect of discouraging rival companies and contributed towards the attainment of the company's objective of driving out or containing competition. It is impossible to subscribe here to the analysis proposed by Wanadoo Interactive, which sees in the expression of its preemption strategy a mere subjective intention. (329) On the contrary, the expression of the preemption intention is conveyed and corroborated by indisputable economic factors and by the translating of the intention into commercial policy. It is impossible to hold that a commercial growth strategy which puts competitors' capacity for financial resistance to the test has anything to do with competition based on merit. Consequently, the Commission must consider that the dynamic and scope of the below-cost selling during the relevant period are in themselves evidence of the intention to drive out competition. (330)

(c) A contextual element: the overall policy of the France Télécom group towards competition in the high-speed market

(285) In order better to assess the scope of Wanadoo Interactive's policy and how it fits into an overall plan, the subsidiary's behaviour may usefully be viewed against the background of that of the France Télécom group as a whole. What follows in paragraphs 286 to 290 is not a list of objections directed against Wanadoo Interactive, but the strategy pursued by the subsidiary cannot be completely dissociated from the objectives of the parent company.

(286) First of all, reference may be made to the behaviour of France Télécom on the wholesale market. The charges for and commercial terms of the wholesale access solutions called, according to the French regulatory terminology, options 1, 2 and 3 are not covered by this proceeding. (331) With regard to these, the Commission would simply point out that the ART considered until summer 2002 that the charges adopted by France Télécom did not enable the various potential operators to enter the value creation chain of high-speed Internet access, referring to a situation of "blockage of competition" on the ADSL market. (332)

(287) France Télécom's pricing strategy concerning option 5, which according to the French terminology covers the combination of the services of IP/ADSL access and IP routing of ADSL traffic outlined in paragraph 42, is of the utmost importance in this context. [...]* Table 11, which sets out the forecast results for (wholesale) IP/ADSL access and IP routing services, illustrates this point.

Table 11: Forecast operating figures for France Télécom's high-speed ADSL Internet access services (euro million) (333)

<emplacement tableau>

(288) The cumulated profit earned by France Télécom from its wholesale services, [...]* can be contrasted with Wanadoo Interactive's cumulated losses from its ADSL services between 2000 and 2002 [...]* (334) [...]*Wanadoo Interactive, marketed the retail product by reselling the same wholesale products at a loss under highly unfavourable profitability conditions.

(289) Wanadoo Interactive has not commented on the substance of the evidence in the table. (335) It simply points out, firstly, that France Télécom's wholesale tariffs are approved by the national telecommunications regulator - something which in its view limits France Télécom's margin for manœuvre - and, secondly, that France Télécom and Wanadoo Interactive are separate legal entities with their own shareholder bases. (336) On the first point, the Commission would observe that, while wholesale tariffs are indeed approved by the Minister for Telecommunications on the basis of an ART opinion, the national regulator has no power of initiative as to their level and simply formulates an opinion on proposals from France Télécom without being able to amend them in any way. (337) At best, the regulator may reject en bloc, as it did in spring 2002, a tariff proposal deemed unsatisfactory and ask France Télécom to submit a new proposal, without, however, being able to dictate its content. It is therefore wrong to say that France Télécom does not have sufficient margin for manœuvre to set its tariffs on its own initiative. On the second point, concerning the legal separation of France Télécom and Wanadoo Interactive, the Commission would simply stress the strength of the shareholder links and the cohesion within the group through its executive committee.

(290) Lastly, the Commission has set out in section I.G.4 various additional elements reflected in company documents which constitute evidence of an intention to restrict the competitive scope of rivals of its subsidiary Wanadoo Interactive: the policy of "closing off" the modems market during the first few months of 2001; the procedure for informing Internet service providers as to whether lines are suitable for ADSL, which was discriminatory for two and a half years; and the proposed agreement on the remuneration of distribution networks and on retail prices for the sale of ADSL packages. These facts do not fall within the scope of the objections examined as part of this proceeding. Accordingly, the Commission will not comment on their compatibility with Community competition law. Nevertheless, these elements do form part of an overall strategy aimed at containing for as long as possible competitors' growth within narrow limits and hence they are useful in that they give a better grasp of the context and of Wanadoo Interactive's intentions. They serve to illustrate a multifaceted strategy aimed at containing the growth of competition within very narrow bounds. (338)

(291) Thus, the intentional strategy of preempting the market in high-speed Internet access for residential customers pursued in 2001 and 2002 by Wanadoo Interactive formed part of a set of practices employed within the France Télécom group which throw the policy pursued by its subsidiary into sharper relief. Wanadoo Interactive's commercial policy of below-cost selling, unsustainable by competitors, was part of a deliberate plan to divert demand for high-speed services to the benefit of the France Télécom group as a whole, at different stages of the value chain, and to corner as it were the first few years of market growth.

4. Objectives pursued by Wanadoo Interactive using the preemption strategy

(292) The focusing by Wanadoo Interactive on the objective of preempting the high-speed market at a time when the economic conditions were not in place found its rationale in the medium-term objectives described in paragraphs 293 to 298. Preempting the high- speed market by below-cost pricing was designed to maximise revenue and margins on ADSL subscriptions, stabilise that same revenue by increasing subscriber loyalty, and maximise revenue on neighbouring markets, notably those in advertising and electronic commerce. It was also intended as a means of acquiring a position of strength on the market in the acquisition of specific high-speed content.

(293) First of all, the position occupied on the high-speed market was intended to make it possible in the long term to rebuild the company's margins on its ADSL products. Various company documents give different time horizons, both near and distant, for the restoration of positive margins on the products in question. [...]* (339) [...]* In its reply of 23 October 2002, Wanadoo Interactive included a table showing margins of euro [...]* a month on average for the eXtense package as from November 2002, thereafter remaining at this level until [...]*. (340) The company has also stated that this estimated margin of euro [...]* was in itself conservative, pointing to a "rising trend in terms of margins" which its simplified estimate did not properly reflect, without, however, providing any simulations of trends in margins in the years ahead. (341) At all events, irrespective of the exact level of margins and the precise date on which this target was to be met, it is clear that Wanadoo did expect, over a time horizon of several years, to earn sizeable margins on this pure access activity. In the light of these internal company reflections, it is clear that the rationale of a predation policy for Wanadoo Interactive lay less in an unlikely increase in Internet access subscription prices than in a recoupment of initial losses through a gradual rebuilding of margins on access products.

(294) Not only was it intended that subscribers to high-speed services should ultimately generate a turnover and margins higher than those generated by subscribers to lowspeed services, but it was also intended that they should constitute a more stable source of revenue. The market in low-speed Internet access is seen by Wanadoo Interactive as a "churn" market, i.e. a market with high customer turnover in which customers switch readily from one service provider to another. By contrast, according to Wanadoo the churn rate for subscribers to high-speed services is [...]* times lower than for low-speed subscribers. (342) The acquisition of a more stable customer base is clearly a safety factor as far as the company's revenue is concerned (343) and it contributes towards the attainment of an objective of increased subscriber loyalty and of maximisation of the profitability of customers over their lifetime.

(295) Secondly, the potential for generating turnover from ADSL customers should, according to Wanadoo, lie not only in access revenue but also in portal revenue and electronic commerce revenue linked to more intensive use of the Internet, and hence should stem from markets neighbouring the relevant market. (344) As an internal company document explains, [...]* (345) A Wanadoo Interactive document, speaking of ADSL subscribers, [...]* (346) Wanadoo Interactive's [...]* marketing plan [...]* (347) With ADSL, Wanadoo Interactive's intention was [...]* (348) [...]* (349) The intention with ADSL was also to improve advertising revenue prospects. With this in view, the acquisition of substantial market shares is strategic inasmuch as, as was pointed out in the course of internal company discussions about the 2004 strategic plan, [...]* (350) Moreover, over the period 2001-2004, Wanadoo Interactive foresaw [...]* increase in revenue from content, electronic commerce and advertising, the total volume of which [...]* between 2000 and 2004, and the average level of which per subscriber was [...]* over that period. (351)

(296) Thirdly, an early presence in the high-speed Internet access market has a direct impact on content supply. The fact of having gained a considerable head-start in the deployment of commercial Internet access offerings confers on the service provider an important advantage when it comes to acquiring content. (352)

(297) Lastly, preemption of the high-speed market was seen as an important factor in consolidating the service provider's image. (353) The objective of an early presence in the relevant market arose from the need [...]* (354) [...]* (355) [...]* (356) [...]* (357) The strategy's objective was clear: to create, in the eyes of the consumer, a strong association between the Wanadoo trade mark and the high-speed concept, so as to be assured subsequently of the lion's share of any market growth. In this respect, 2001 was a decisive year: thanks to its positioning in terms of price, [...]* (358)

(298) The overall economic reasoning underlying the market preemption strategy is finally summed up as follows in [...]* memo [...]* (359) [...]* The Commission considers that this analysis sums up perfectly the logic of predation and preemption applied by Wanadoo Interactive, a logic which was perfectly rational strategically as far as the company was concerned.

5. Conclusion on the intention to preempt

(299) The Commission considers that Wanadoo Interactive deliberately embarked upon a strategy of expanding its high-speed services, combining below-cost prices and substantial sales volumes, which was unsustainable by its competitors.

2. Applicability of the predation concept to a growth sector

(300) In its reply of 23 October 2002, Wanadoo Interactive maintains that it is at all events impossible to make an allegation of predation in a business sector which does not possess a sufficient degree of maturity, and that therefore such an allegation cannot be levelled against it. (360) It questions the relevance of the application of the traditional case-law of the Court of Justice to a sector which it describes as "emerging". (361) In the context of a new service, the incurring of significant losses is, in its view, inevitable, and instant profitability is "by definition impossible" (362) on a market characterised by great economic uncertainty. (363)

(301) In this connection, it must be pointed out that nothing in Article 82 of the Treaty or in the Community case-law on the subject provides for an exception to the application of the competition rules to sectors which are not yet fully mature or which are considered to be emerging markets. To subordinate the application of the competition rules to a complete stabilisation of the market would be to deprive the competition authorities of the power to act in time before the abuses established have exerted their full effect and the positions unduly acquired have thus been finally consolidated. It follows, on the contrary, from the case-law that it must be possible to penalise predatory pricing whenever there is a risk that competitors will be eliminated, as the aim pursued by the Treaty, which is to maintain undistorted competition, (364) rules out waiting until such a strategy leads to the actual elimination of competitors. (365)

(302) Even if it cannot be ruled out that the high-speed Internet access market must indeed be considered an emerging market, it is at all events a specific type of market justifying particular vigilance on the part of the Community competition authorities. The market has strong links with the market for local access in the telecommunications sector, which for its part possesses none of the features of an emerging market. The telecommunications sector, and local access in particular, is still deeply affected by the preponderance of the former telecommunications monopoly, which controls almost all access to the final consumer. Even if high-speed Internet access were to be treated as an emerging market, its links with a market in the throes of deregulation and the resulting risks of a leverage effect make it impossible to grant it any exemption from the ordinary rules of Community competition law.

(303) The Commission would point out further that, in the context of the development of ADSL products, the objective of rapid profitability was not beyond reach despite the newness of the products in question. [...]* (366) [...]* France Télécom's figures show therefore that, contrary to what Wanadoo Interactive maintains, the newness of ADSL services in no way precluded the attainment of an objective of fairly rapid profitability. The incurring of substantial losses on a new type of product such as ADSL can therefore by no means be considered either inevitable or necessary.

(304) In this context, the Commission considers that Article 82 of the Treaty is applicable and that a charge of predation cannot be averted in the present case on the ground that the market is an emerging one.

3. Lack of objective justification for below-cost pricing

(305) According to Wanadoo Interactive, the rule established by the Akzo predation test is not absolute and must be assessed in the light of the economic circumstances of each case. The company maintains that below-cost pricing may be perfectly justified by aims other than a wish to oust competition. Firstly, it says, it is seeking to achieve economies of scale, and secondly its below-cost pricing benefits the market as a whole. Below-cost pricing which has the effect of subsidising the customer is, it argues, even socially desirable. (367) Lastly, even if it were dominant, the company is, it maintains, obliged by existing or potential competitive pressure to align itself on the prices set by other competitors. These various arguments are examined in paragraphs 306 to 331.

1. Economies of scale and learning effects

(306) Wanadoo Interactive states that the variable cost of some products diminishes with the quantity produced and depends on the volume of demand, while a high production volume may help to gather experience and therefore to reduce the costs of future production. (368)

(307) This argument, instead of legitimising predation objectively, highlights on the contrary one of the rational objectives thereof. One of the objectives of below-cost pricing may be to reserve for the company engaging in the practice the benefit of economies of scale on the market and to delay accordingly for competitors their arrival at the same volume threshold allowing the economies of scale. For an argument based on efficiency gains to be admissible, it must be possible to prove that such gains could not have been achieved by means other than a below-cost selling strategy. In the present case, there is no guarantee that such gains could not have been achieved had the market developed in a balanced manner. Moreover, a combination of being at a higher point on the learning curve than competitors and having higher output thanks to below-cost pricing may have exclusion effects capable of consolidating the dominant company's hegemony. (369)

(308) In the present case, the conquest of market shares at the beginning of 2001 regardless of the losses incurred enabled Wanadoo Interactive to achieve economies of scale early on. The learning effects are difficult to measure, except perhaps in the reduction of certain customer administration costs, which was noticeable from the end of 2001. The scale economies are much more apparent in an area such as the routing of ADSL traffic. Until October 2002, routing charges were sharply tapered according to traffic volume. (370) Depending on the traffic volume, the charge per Mbit/s for the regional traffic routing service varied on a scale of [...]* and the charge for the national routing service on a scale of [...]*. The charge is a function of the number of subscribers. Thus, at the end of 2001 or the beginning of 2002, on the basis of the routing service price scales, a new operator serving some 8 000 subscribers divided evenly between the Paris region and the rest of metropolitan France bore an average routing cost per subscriber [...]* higher than that borne by Wanadoo Interactive. (371) Wanadoo Interactive had for its part already achieved by summer 2001 a customer base that attracted the lowest charges. In view of the tapering nature of the price scale, an operator did not qualify for the lowest routing charge until it had more than 16 000 or so ADSL subscribers in the Paris region and more than 81 000 subscribers in the provinces. (372) In August 2002, none of Wanadoo Interactive's competitors in the ADSL segment had yet attained these thresholds. Wanadoo Interactive, for its part, had crossed them as early as April 2001 thanks to its rapid growth and its below-cost pricing.

(309) Thus, while the search for scale economies and learning effects may be included among the rational justifications for predatory behaviour, it may not serve to legitimise that practice from the point of view of competition law since it has the effect of conferring a more favourable cost structure on the dominant undertaking to the detriment of its competitors.

2. Lack of justification in the form of positive externalities

(310) Wanadoo Interactive suggests that it chose initially to sacrifice its profitability on the altar of market growth with a view to helping to increase awareness of high-speed Internet, which would in turn have benefited its competitors and the market in specific high-speed content, and to reducing network costs.

(311) This argument cannot be upheld. An undertaking in a dominant position has no need to practice predatory pricing in order to attract new customers and draw consumers' attention to the product. (373)

(312) Above all, Wanadoo Interactive's argument is deficient in one essential respect: there is no proof that the strategy pursued by the company would alone have made it possible to attain the desired objective of increased broadband use in France. The positive effects linked to market growth could have been brought about had the market developed under conditions of equilibrium among service providers. If it had really been the France Télécom group's intention to develop the high-speed market for the benefit of all operators, France Télécom could have priced all its wholesale products - from shared or full unbundled access to the local loop to IP/ADSL access and routing services - at low levels encouraging the entry of competitors. The France Télécom group chose instead to confine the losses associated with developing high-speed access to its retailing subsidiary, thereby diverting the market growth to its advantage. It cannot therefore cogently be maintained that the France Télécom group, and Wanadoo Interactive in particular, were guided by a desire to develop the market for the benefit of all stakeholders. (374) If such had been the case, France Télécom could have applied much sooner the remedy that was finally proposed once proceedings had been initiated against Wanadoo Interactive.

(313) Wanadoo Interactive also maintains that its pricing policy has had a positive spillover effect in bringing about a reduction in the cost of modems and ADSL equipment in France. (375) This argument does not withstand scrutiny. In 2000 and 2001 Alcatel, the leading supplier of ADSL equipment in France, produced at its plant in Belgium more than six million modems. (376) By the end of 2001 there were some 15 million DSL connections worldwide. (377) The total number of ADSL connections in France in 2001 was of the order of 392 000, and only some of these were equipped with Alcatel modems. The expansion of the French market has therefore not been decisive as regards this supplier's production volumes or as regards the level of its manufacturing costs and selling prices, since all modems of the same type are technically identical. (378)

3. The argument as to alignment on competitors' prices

(314) On a number of occasions, Wanadoo Interactive has claimed that, as a would-be new entrant into the relevant market, it was forced to align itself on the prices charged by competitors. (379) This claim elicits several series of observations, on the validity of the alignment argument from a point of view of principle, on the market position actually occupied by the companies concerned, and, lastly, on matters of fact which factually refute Wanadoo Interactive's contention.

(315) First of all, from a point of view of principle, it is true that new entrants or undertakings which are not in a dominant position are entitled to charge promotional prices for limited periods. Their sole aim is to draw the consumer's attention to the very existence of the product, more persuasively than by a mere advertisement, (380) and such offers do not have any negative impact on the market. On the other hand, alignment by the dominant operator on the promotional prices of a non-dominant operator is not justified. Whilst it is true that the dominant operator is not strictly speaking prohibited from aligning its prices on those of competitors, (381) this option is not open to it where it would result in its not recovering the costs of the service in question. Whilst the fact that an undertaking is in a dominant position cannot deprive it of its entitlement to protect its own commercial interests when they are attacked, such behaviour cannot be countenanced if its actual purpose is to strengthen that dominant position and abuse it. (382) The dominant undertaking thus has a special responsibility not to allow its behaviour to impair genuine undistorted competition on the common market. (383)

(316) Wanadoo Interactive claims it was obliged to align the price of its eXtense service, set in autumn 2000, on those of the offerings of three competitors: the cable operator Noos using cable modem technology, and the ADSL Internet service providers Club Internet (T-Online France) and Mangoosta. None of these three operators can be characterised as dominant or be deemed to have objectively posed a serious threat to the interests of the dominant undertaking. Their respective situations at the end of 2000 and the beginning of 2001 are commented on briefly in paragraphs 317 to 319.

(317) The cable operator Noos cannot be characterised as the dominant operator on the domestic market. It is restricted in its scope to the Paris region, and what is more a significant number of communes in that region, especially in the departments immediately surrounding Paris, are not covered by it. (384) Having almost no distribution outlets of its own, its marketing network is small. It enjoys none of the advantages that Wanadoo Interactive derives from its many links with France Télécom. It had no experience with low-speed Internet access and hence it had not only to enter a market that was new to it but also to learn the rudiments of a trade that it had never practised. [...]* starting in 2000 and the first quarter of 2001 Wanadoo Interactive enjoyed a growth rate invariably [...]* times faster than Noos's in this market. (385) It was clear as soon as ADSL technology was launched in France at the end of 1999 that Noos would quickly be outdistanced and that it would not occupy a dominant position on a national scale on the market in high-speed Internet access.

(318) T-Online France, for its part, displayed at the relevant time none of the features of a dominant undertaking on the relevant geographic market. Its entry into the high-speed market in France was no more than symbolic. In June 2000 the company had 24 times fewer high-speed Internet subscribers than Wanadoo Interactive. Both in September 2000 and at the end of 2000, the ratio between the total number of T-Online France subscribers and of Wanadoo Interactive subscribers was [...]*, and at the end of the first quarter of 2001 [...]*. (386) During the second half of 2000, the total number of subscribers to Wanadoo Interactive's high-speed Internet access services grew [...]* times faster than that of T-Online France, and in the first quarter of 2001 [...]* times faster. (387)

(319) Mangoosta, the third competitor, likewise did not occupy a dominant position on the relevant market. Unlike Wanadoo Interactive, which had been an established player on the market in low-speed Internet access for more than five years, Mangoosta was an entirely new company without any track record or logistical resources to match those of Wanadoo Interactive. Its ADSL subscriber base never grew beyond the 5 000 mark.

(320) Secondly, it should be pointed out for the sake of completeness that Wanadoo Interactive's statements are factually inaccurate. The cases of the two Wanadoo Interactive products to which this Decision relates and the chronology of the price decisions of the various market operators are examined in paragraphs 321 to 331.

(321) The case of the Wanadoo ADSL service is easily dealt with. As indicated in paragraph 127, the price of this service was set before April 1999. At that time, the question of alignment did not arise. Club Internet for its part did not announce its selling price until 22 October 1999 and, as pointed out in paragraph 127, had to align itself on the price set by Wanadoo Interactive. It is therefore not possible to claim any alignment on the prices of a competitor for this product, which, in mid-2002, accounted for over a third of the company's high-speed residential subscribers.

(322) The case of the eXtense service merits more detailed examination. The price of this service will be scrutinised in the light of the prices charged by Noos, Club Internet (TOnline France) and Mangoosta, in that order.

(323) From May 2000, (388) Noos offered a subscription to the Noosnet high-speed access service for a total of FRF 378 (or euro 58), taxes included, per month. (389) This was 27% more per month than Wanadoo Interactive was to charge several months later for a subscription to the eXtense service. In addition, Noos demanded FRF 1 200 (or euro 183) up-front from the subscriber, which was 21% more than it cost to buy an eXtense starter pack from Wanadoo Interactive. (390) Noos's offering was thus appreciably more expensive than the one which Wanadoo Interactive was to propose several months later.

(324) In reality, Wanadoo Interactive pitched the price of its eXtense service slightly below a temporary promotional offer by Noos. Noos offered a subscription at FRF 299 taxes included for a period of one year (391) to customers newly recruited between 28 August 2000 and 21 January 2001, and to those customers only. (392) This promotional offer was subsequently extended until 1 March 2001. However, it was clearly a short-lived offer, presented to the public as such, and the advantage it conferred was limited in time. Moreover, it covered only the subscription, and not the up-front payment referred to in paragraph 323, with the result that it was always far less advantageous than Wanadoo Interactive's offer. Furthermore, in February-March 2001, just a few weeks after eXtense was launched, all of Noos's prices were increased, with the result that they were now much higher than Wanadoo's, both because of the promotion period coming to an end and because of the billing for any exceeding of the upload limit, (393) coupled with the marketing of the "Noosnet forfait 1 Go" offering aimed at large customers, henceforth the only Noos offering that can properly be compared to eXtense. (394)

(325) The monthly subscription fee for Noos's offering has therefore never been lower than that of Wanadoo Interactive. Even during Noos's promotional period, which overlapped only with the first seven weeks of the marketing of the eXtense service, the up-front payment required of the customer by the cable operator was invariably 20% higher than the price payable for the eXtense starter pack.

(326) T-Online France, for its part, announced the launch of its offering several weeks before Wanadoo Interactive, on 22 November 2000, with a view to commencing sales on 1 December 2000. (395) Wanadoo Interactive's internal documents reveal that, after some hesitation, the price was in reality set internally at FRF [...]* taxes included (before being [...]* FRF 298 taxes included) during the first half of October 2000, i.e. more than six weeks before T-Online made public its packaged service offering and more than two weeks before T-Online set its price internally. The price mentioned momentarily during the summer of FRF [...]* taxes included no longer appears in any budget documents after 5 October 2000. It would appear therefore that Wanadoo Interactive's argument is factually incorrect.

(327) Wanadoo Interactive claims that the budget presentations obtained by the Commission from its finance department are not formal decisions in the way that a price decision is. (396) In its reply of 23 October 2002, the company maintained that it had planned up to the last minute to charge FRF [...]* a month for its eXtense service. (397) The subscription fee was, it said, set once and for all on 22 November 2000, the day on which Club Internet made the details of its offering public. To back up this assertion, it produced an email drafted by [...]* on 22 November 2000 [...]*, in which it is stated: "I confirm the subscription fee of FRF 298 taxes included (instead of FRF [...]* taxes included) from 1 January."

(328) The Commission would point out first of all that, in a company the size of the one at issue here, a budget presentation is an extremely serious exercise. The price assumptions made during the first half of October 2000 were by no means exploratory scenarios, but instead quasi-definitive assumptions. At all events, the extremely succinct email produced by Wanadoo Interactive cannot be considered to be a document of greater evidential value than the budget presentations referred to in paragraph 326 or to attest to the fact that Wanadoo Interactive was still seriously thinking about setting the price at FRF [...]* taxes included. It does not prove at all that the company had decided on a firm launch price at that level. It indicates at most that there had been some hesitation about the definitive price level up to that point, and that it was only then that that hesitation was finally swept away. The company has, moreover, not produced any other evidence of the FRF [...]* price hypothesis surviving, other than residually, during the period from [...]*.

(329) Lastly, as regards Mangoosta, on 23 October 2000 this competitor launched its packaged service at a price of FRF 330, taxes included, per month, modem rental included. The launch was announced in the press on 6 October 2000. (398) Wanadoo Interactive claims that it aligned itself on this competitor also.

(330) The argument employed by Wanadoo Interactive elicits a number of responses. First of all, the nominal price proposed by Mangoosta was 11% higher than the price at which Wanadoo Interactive's offering was to be pitched several weeks later. Moreover, except in the case of the first 1 000 subscribers recruited during a 19-day promotion period, (399) Mangoosta, unlike Wanadoo Interactive, billed to the consumer installation costs of FRF 500 taxes included. Wanadoo Interactive points out that Mangoosta's subscription fee included the modem rental and the installation kit. A simple calculation (400) shows, however, that it was cheaper to subscribe to Wanadoo Interactive's offering for any period of at least 15 months. (401) Lastly, in February 2001, barely weeks after Wanadoo Interactive launched its eXtense product, Mangoosta increased its monthly subscription fee by 20%, making it highly uncompetitive compared with that of Wanadoo Interactive. If Wanadoo Interactive had in fact set its prices out of fear of being ousted by Mangoosta, there was nothing to prevent it from revising its prices upwards.

(331) The documents cited by Wanadoo Interactive concerning the in-house perception of the potential pressure exerted by competitors (402) lead at most to the conclusion that during the second half of 2000 the company may have overestimated the reality of the potential competitive threat and have defined a strategy of preventive occupation of the ground by way of riposte. [...]* Such a mistaken perception cannot, however, serve to justify below-cost pricing. It would be tantamount to acknowledging that a dominant undertaking is entitled to determine its commercial policy in such a way as to stymie the ambitions of purely virtual competitors. [...]* In such a context, while the argument based on alignment on competitors' prices would have been admissible in principle, it lost all factual foundation as from [...]*. This Decision therefore finds fault with the company not so much for setting prices at the end of 2000 at a belowcost level as for subsequently maintaining those prices at that level as part of a wideranging strategy of market preemption deployed at national level as from the beginning of March 2001.

4. Probability of a recoupment of losses

(332) Wanadoo Interactive maintains that a strategy of predation may be considered viable and realistic only if the structure of the market and the presence of sufficient barriers to entry enable the dominant undertaking subsequently to enjoy high, stable margins on the relevant market and thus to recoup any losses initially incurred. (403)

(333) It should be pointed out that neither the case-law of the Court of Justice (404) nor the decision-making practice of the Commission (405) requires proof that initial losses were actually recouped before a finding can be made of abuse through predatory pricing. Action by the competition authorities is justified as soon as there is a risk that competition may be distorted, regardless of any prospect of loss recoupment.

(334) However, while the recoupment of losses initially incurred may constitute a rational objective associated with predation, other scenarios are perfectly conceivable. In certain specific cases, the undertaking may embark upon a strategy of predation with aims other than the achievement of operating margins higher than those which would prevail in a competitive context. For example, in certain highly specific share ownership scenarios, the undertaking may attach only secondary importance to recoupment of its losses. (406) It may also abandon the idea of recouping all its initial losses and concentrate instead on balancing its future costs and revenues. (407) Lastly, it may aim at recoupment in the long term by means other than its operating results. (408)

(335) The Commission accordingly considers that proof of recoupment of losses is not a precondition for a finding of abuse through predatory pricing.

(336) Subsidiarily, it must be pointed out that the recoupment of losses is rendered plausible in the present case by the structure of the market and the associated revenue prospects.

(337) The more the market's functioning is characterised by the existence of significant entry barriers capable of facilitating loss recoupment, the more a predation strategy will pay dividends. The probability of loss recoupment will be less strong in a market where entrants have access to the same technology as the incumbent operator, where there are fewer irreversible costs and where firms can enter and exit the market without difficulty and without extra cost. (409) Conversely, it is not necessary for purposes of loss recoupment that the entry barriers should be absolutely insurmountable, (410) as in that case the entry of competitors is altogether improbable and recourse to a predation strategy with a view to dissuasion is no longer warranted. For loss recoupment to become a likely prospect, it is sufficient that the obstacles to entry guarantee the dominant undertaking the maintenance in the long term of a large degree of market concentration in its favour. The barriers may take many forms. They may be absolute, insurmountable barriers, such as regulatory, legal or technical obstacles. Or they may be barriers which, without necessarily being insurmountable, nevertheless constitute obstacles to entry slowing down new entrants' progress. These obstacles exist where the new entrant into the relevant market is placed at a disadvantage compared with the incumbent solely by reason of the latter's already being in business. (411) Lower unit production costs thanks to economies of scale, brand image, privileged access to information on costs and cost trends, and the dominant undertaking's entrydiscouraging behaviour itself (412) are capable of constituting obstacles to entry. The higher the non-recoupable entry costs, the more the market is characterised by strong obstacles to entry, as the risks associated with unsuccessful entry are high.

(338) Although proof of the possibility of actually recouping the initial losses is not required by Community law, the Commission intends, accessorially, to examine the entry barriers and entry costs which characterise the relevant market and which render plausible a recoupment of the losses by the dominant firm in the long run. It is necessary to identify, firstly, the factors forming an obstacle to the redistribution of subscribers existing at the time when the abuse comes to an end and, secondly, the factors making it difficult for competing companies to secure a significant share in the market growth resulting from the ending of the abuse.

(339) The strategic barriers related to the effects of the link-up with the France Télécom group, together with the many synergies resulting therefrom, were set out in section II.C.3. A number of other obstacles to the entry and growth of competitors, which are likely to assume particular importance when the abuse comes to an end, must be discussed, namely: the disincentives to mobility on the part of existing subscribers; the cost of acquiring sufficient notoriety in a mass market; and the cost of rolling out alternatives to France Télécom's wholesale offering. In addition, the level of the margins recorded by Wanadoo Interactive since October 2002 is a sign of the company's capacity to recoup its losses in the medium term.

1. Disincentives to mobility on the part of existing subscribers

(340) It will first of all be examined whether existing subscribers to Wanadoo Interactive's ADSL services can switch easily to another provider.

(341) As pointed out in paragraph 294, the market in high-speed Internet access differs from that in low-speed access in that it is not a market with a high customer turnover, having a churn rate three times lower than the average for all Internet users.

(342) Besides brand loyalty and customer satisfaction with the quality of the service provided, a number of objective factors may help to explain the low mobility of subscribers.

(343) First, practically all market offerings, and the subscription to the eXtense service in particular, involve commitment periods of either one year or two years and provide for financial penalties in case of termination of the contract before its expiry.

(344) Secondly, assuming that this first obstacle has been overcome and the subscriber has been released from his commitment, two situations may arise. In the first, the subscriber wants to switch from an ADSL provider to a broadband cable service provider, or vice versa: he then has to change modems and incur a potentially dissuasive cost. In the second, the subscriber wants to switch from one ADSL service provider to another ADSL service provider. It is then not necessary to change modems if the equipment is compatible with the technical specifications of the telecoms operator administering the service. Two scenarios must, however, in turn be considered. In the first scenario, the Internet service provider to which the consumer has chosen to migrate is a telecoms operator and controls the value chain: there is then a risk of several days' interruption of service as there is as yet no procedure for migrating from a France Télécom ADSL offering to an ADSL offering of another telecoms operator. In the second scenario, the consumer turns to an ADSL provider who is in reality just a reseller of the service proposed by France Télécom: the access provider then has to pay France Télécom euro 53.40 by way of commissioning costs. All this helps to explain why the temptation for Wanadoo Interactive's existing subscribers to migrate to competing service providers, and in particular to service providers proposing cheaper offerings, is in reality not very strong. (413)

(345) Owing to the disincentives to mobility on the part of existing subscribers, the bulk of the effort involved in the conquest of market shares by competitors must therefore be concentrated on new customers.

2. The costs of entering and acquiring a critical size in a mass market

(346) It is common practice to consider that early entry into a market confers a major advantage on a firm where it has been able to establish a significant preference for its brand in the eyes of the consumer, not by providing an objectively better service than its competitors, but simply because it was first on the market or because it has spent heavily on its image. (414) In such a context, a new entrant must not only invest the technical resources needed to produce and market the good, but it must also incur substantial promotional expenditure if it is to make its offering well enough known.

(347) Contrary to what Wanadoo Interactive has claimed, Internet access markets are characterised by significant barriers to entry. In support of its position, Wanadoo Interactive has cited the July 2000 stock exchange listing prospectus, in which the market is described as having low barriers to entry. This brief comment in a public document must be supplemented both by an observation of developments on the relevant markets since that position was taken and by internal company documents, likewise dating from after that time. Any validity Wanadoo's analysis as set out in the abovementioned prospectus may have had in the first half of 2000 may have diminished with time in view of the behaviour of the operators present in this sector.

(348) It would be wrong, moreover, to consider the structure of a market to be an entirely exogenous, immutable given. On the contrary, the behaviour of firms on a market, and in particular the behaviour of the firm first present thereon or enjoying other historical advantages, is likely to modify the conditions governing entry or growth by competing companies on that same market. A company's early commitment to a market appearing at first sight to be relatively open may be intended not only to derive benefit from more favourable production conditions, but also to increase the costs of entry for potential competitors. (415)

(349) [...]*, Wanadoo Interactive [...]*mentions . (416) It is thus difficult for a new, unknown operator to become established quickly as a recognised service provider. Even service providers with recognised brands have to spend very substantial sums on promotion, as can be seen from Table 12, which is based on an internal Wanadoo Interactive document, (417) setting out the promotional expenditure of the leading French service providers in 2001.

Table 12: promotional expenditure of Internet service providers in 2001 (low-speed and high-speed combined)

<emplacement tableau>

(350) Advertising expenditure of such orders of magnitude, both in absolute terms and as a proportion of turnover, especially on the part of companies already well known as service providers, bears witness to the scale of the cost of acquiring sufficient notoriety and visibility on a mass residential market. For a new operator on the highspeed Internet access market with no previous experience of low-speed Internet access, like Mangoosta, it is not technically impossible to enter the market, but it is extremely doubtful whether a new entrant of this type will be able to make up for its lack of image through a wide-ranging promotion campaign and arouse the interest of a sufficient number of consumers to threaten the position of a dominant operator. A fortiori, a new, unknown entrant which does not embark upon a massive promotion campaign can hope to acquire only a small or negligible share of the market. Such was the case on the French market with two operators which started up in 2002. (421) Thus, wholly new, unknown entrants, without being prevented from penetrating the market altogether, are forced to employ niche strategies if they are not to incur promotional expenditure out of all proportion to their commercial investment capabilities. This advantage in terms of notoriety for the incumbent constitutes a high entry barrier for potential new entrants into the Internet access market.

(351) With regard to the high-speed Internet access market in particular, [...]*. (422) Service providers must, during this high-speed market development phase, build an image as the default supplier of a product viewed by the consumer as technically sophisticated and become large enough to benefit from economies of scale.

(352) In this process, the chronological sequence of entry into the market is far from neutral. Clearly, a service provider that has a considerable head start over its competitors during the initial phase of market growth is able to capitalise on the momentum thus gained. By contrast, laggards must make a much bigger effort to acquire customers if they wish to make up for lost time and bridge the resulting image gap and confer on their high-speed service the same notoriety as that of the dominant undertaking's flagship offering. In these circumstances, new competitors are confronted with the need not only to carry out the expenditure technically necessary in order to provide the service but also to undertake substantial advertising and promotional expenditure both to raise their product's profile and to undermine loyalty to the dominant undertaking's brand.

(353) An industry in which advertising expenditure is highly significant, especially where the amounts spent on advertising are of the same order of magnitude as the revenue earned from the associated sales, must be considered an industry characterised by substantial entry costs. (423) The advertising expenditure already incurred confers an advantage on the incumbent and creates a same-sized handicap for new entrants, whose advertising and promotion costs are generally higher and less effective than the incumbent's. (424) Although the advantages derived by the incumbent from its already being present on the market may gradually be eroded with time, new market entrants are obliged to bear higher advertising costs for a longer period than the incumbent if they are to be able to challenge its market position. (425) The present case provides an illustration of this type of situation.

(354) Wanadoo Interactive sees the upsurge in promotional offers from October 2002 as proof of the strength of competition and of the return to healthy market conditions. The Commission interprets it differently. These large-scale promotional campaigns by competitors are, at this stage in the market's development and in view of the head start gained by Wanadoo Interactive, the only means of avoiding outright eviction and of achieving a certain visibility with a view to acquiring a critical size. The analysis of competitors' promotional expenditure426 attests to the fact that this expenditure represents a highly significant entry cost. Three salient points in particular emerge from it.

(355) Firstly, during the period 2001-2002, Wanadoo Interactive spent nearly euro [...]* on advertising, promotion and various marketing activities related to its eXtense and Wanadoo ADSL services, of which [...]* (427) and more than [...]* of the total promotion budgets of Tiscali-Liberty Surf and T-Online France - Club Internet for their entire range of services. (428) Leaving aside cases of new entrants without any experience or notoriety in the low-speed Internet access market, even firms well ensconced in that market will probably hesitate before committing themselves in the high-speed market to advertising and promotion campaigns on the same scale as those which Wanadoo Interactive undertook in 2001 and above all in 2002. Although advertising expenditure focused on acquiring new customers is in the nature of a variable cost, and although it ceases if the company decides to stop recruiting new subscribers, it none the less represents a high entry cost which is liable to prove dissuasive should the company set its sights on a non-negligible market share. In order to benefit from scale economies in the activity in question, including qualifying for the lowest IP routing charges, and in order to gain separate growth momentum through having a critical size, (429) it is necessary to acquire several tens of thousands of customers, (430) which in itself calls for a very substantial advertising and promotion outlay on top of the other customer acquisition costs.

(356) Secondly, the advertising and promotion expenditure incurred is very substantial compared with revenue. It can be roughly estimated that in 2002 the advertising and promotion outlay of Wanadoo Interactive's competitors was equivalent to approximately 38% of their total high-speed turnover and to 102% of the additional turnover generated by the growth in their high-speed subscriber base in 2002. For Wanadoo Interactive, the corresponding ratios were [...]*% and [...]*%. These ratios are very high, especially for Wanadoo Interactive's competitors, and they attest to the fact that large-scale market entry is particularly costly.

(357) Thirdly, not only is the advertising and promotion expenditure competitors must incur if they are to acquire a certain visibility substantial, but it is less effective than that of Wanadoo Interactive owing to the advantages in terms of notoriety on the relevant market already enjoyed by that company. In 2001, Wanadoo Interactive spent on advertising, promotion and various marketing activities related to its ADSL services approximately euro [...]* per new subscriber. During the first half of 2002, the company devoted under the same heads approximately euro [...]* per new subscriber. During the last quarter of 2001 and the first half of 2002, to acquire each additional customer competitors for their part devoted [...]* to [...]* times more resources to advertising, direct marketing, promotion and publicity than the amounts indicated for Wanadoo Interactive in 2001 and spring/summer 2002. In the second half of 2002, the average amounts per subscriber acquired converged as between Wanadoo Interactive and its competitors. However, if one subtracts from the sample the service provider Free, whose product, which is appreciably cheaper than that of the competition, requires less in the way of advertising to set it apart,431 competitors' advertising and promotion expenditure is still [...]* times greater than that of Wanadoo Interactive. Consequently, while it is not impossible for operators with a high profile in the lowspeed market to win market share in the high-speed market, they will have to put in a much greater effort than Wanadoo Interactive did back in 2001.

(358) Thus, the structure of the market in the wake of Wanadoo Interactive's predation strategy is not such that it acts as an absolute bar to entry into and growth on the highspeed market for service providers which already enjoy notoriety in the low-speed market. It does, however, affect very strongly the scale and volume of their entry into the market and their capacity to attain a critical size. It is, therefore, not so much the nature and unitary amount per new subscriber of the advertising and promotion expenditure as the overall volume of the commitments needed to acquire a critical subscriber base that turns advertising and promotion expenditure into a potentially dissuasive entry cost. Even companies with a firm foothold in the low-speed Internet access market may hesitate to embark upon advertising and promotion campaigns on the same scale as Wanadoo Interactive. (432) In this context, the effort required of competitors in order to become more than mere "background noise" has to be seen as a barrier to entry and to re-entry for new entrants of the Mangoosta type, or as a significant entry cost for operators already present on the low-speed market, both of which promote a high degree of market concentration. (433) The foreseeable difficulty for competitors to acquire a critical size on a mass market enabled Wanadoo to envisage as perfectly plausible a market sufficiently concentrated to allow a recoupment of losses.

3. Cost of alternatives to the retail offering proposed to service providers by France Télécom

(359) Instead of having recourse to France Télécom's wholesale offering, which combines both IP/ADSL access and IP/ADSL routing services, competitors in the ADSL segment may choose to build their own telecommunications network with a view to freeing themselves from the economic constraints associated with that retail offering and, possibly, proposing to the final consumer a product characterised by a better quality/price ratio.

(360) Such a choice of strategy exposes a company, however, to obstacles which do not arise in the case of the mere retailing of the wholesale access and routing services provided by France Télécom.

(361) First of all, the service provider's national and regional telecommunications network must attain a sufficient degree of capillarity to serve the whole country. The operator Free, which has held a telecommunications operating licence since early 1999, has for example spent more than three years setting up a several thousand kilometre long optic fibre network costing between euro 15 and 30 million, (434) only to finish up covering just the country's main population centres. Investment on such a scale cannot be carried out quickly and without a very substantial initial capital outlay.

(362) Moreover, in order to master the various parts of the telecommunications service, a service provider making this investment choice must resort, in order to reach the final customer, to the local loop unbundling solution, which involves high fixed costs. The leasing by France Télécom of the copper pairs to other companies entails for the latter not only recurring charges but also very substantial, non-recoupable fixed costs related inter alia to the transfer of the lines and to the service of providing space for the companies' equipment in France Télécom's premises. The initial capital outlay is substantial, and at all events out of reach for an entrant which does not have very considerable financial resources. In Free's case, it is put at euro 10 million initially for unbundling alone (435) and at euro 70 million for all the investment linked to ADSL in 2003, on top of the transport network infrastructure costs already referred to. (436) In addition, the actual local loop unbundling process is particularly long and technically difficult.

(363) Wanadoo Interactive has pointed out that, even if they are not network operators along the lines of Free, Internet service providers can use the infrastructure of the alternative telecommunications operator LDCom. (437) This is factually accurate inasmuch as LDCom offers network services to service providers using the solutions known according to the French regulatory terminology as option 1 and option 3, but two points have to be made. First, recourse to the services of a wholesaler introduces an intermediary into the value chain, creating an extra cost compared with direct administration of the network by the service provider itself, whose freedom it limits technically and financially. Moreover, LDCom is itself faced with the difficulties and financial needs outlined in paragraph 362, with the result that the technical solutions it proposes are at present partial and geographically highly circumscribed. (438)

(364) The construction of an alternative telecommunications network and recourse to the local loop unbundling solution result in considerable delay and the initial tying-up of large sums of money. In these circumstances, this model, which alone makes it possible to break free from the technical and financial constraints, is being implemented only very gradually and, at the time of this Decision, very marginally.(439) The cost and time needed to set up an ADSL network comparable to that of France Télécom are such that they limit considerably the scope for competing service providers to propose an economic model that is more attractive to the consumer. They must therefore also be interpreted as barriers to entry.

4. The rebuilding and increasing of Wanadoo Interactive's margins, evidence of the likely recoupment in the long run of the initial losses

(365) Wanadoo Interactive confirmed in its October 2002 reply that as from 2003 its gross production margin on the eXtense package service will be more than euro [...]* per month (giving a margin rate of [...]*%), and that this margin shows an upward trend over time. (440) Such a margin level would thus make it possible to recover customer acquisition costs for each new subscriber to the eXtense product in about a year. Overall, assuming - on the basis of the information supplied by Wanadoo - that the margin levels for the eXtense and Wanadoo ADSL services are maintained throughout 2003 at euro [...]* and euro [...]* respectively, subscribers recruited up to 31 March 2003 would alone generate a margin on recurring costs in excess of euro [...]*, (441) more than enough to cover the non-recurring costs linked to the acquisition of new customers, (442) and making a significant contribution towards recouping the losses of previous financial years. Although it is difficult to predict when the initial losses will have been recouped, it is likely that recoupment will occur within the space of a few financial years.

(366) The level of the margins upon termination of the abuse as indicated by Wanadoo Interactive, and their maintenance or progression over time, are firm evidence of the company's probable ability to recoup its initial losses over a reasonable time horizon through rebuilt margins.

5. Conclusion on the likelihood of loss recoupment

(367) The Commission considers that there are indeed significant entry barriers and high entry costs to be paid for the acquisition of a critical size, even if these obstacles are not absolute in nature. Contrary to what Wanadoo Interactive maintains, entry (or reentry) into and the acquisition of a critical size on the relevant market are costly and time-consuming. These features create an environment conducive to the maintenance by Wanadoo Interactive of a position of very strong dominance and prevent the growth of competitors representing a danger to Wanadoo Interactive. The recoupment by Wanadoo Interactive of its initial losses is therefore a likely scenario. The predatory strategy introduced in 2000 appears pertinent in this context.

5. Conclusion on the abuse

(368) In view of the above, the Commission considers that the pricing of the eXtense and Wanadoo ADSL services at levels which did not make it possible to recover the variable costs from March to August 2001 or to recover the full costs from August 2001 to October 2002, as part of a plan to preempt the high-speed Internet access market at an important stage in its development, constitutes an infringement of Article 82 of the Treaty.

E. REPERCUSSIONS OF THE ABUSE ON COMPETITION

1. Effects on the market in 2001 and 2002 of the strategy followed by Wanadoo Interactive

(a) Evolution of market shares and growth rates

(369) The first effect of Wanadoo Interactive's predatory strategy was to drive out of the market and cause the definitive disappearance in the first half of 2001 of the company which was at the time its most determined rival and most advanced competitor in rolling out ADSL services, Mangoosta.

(370) More generally, the comparative growth in sales by the different players on the market in high-speed Internet access during 2001 and until the autumn of 2002, shown in Table 13, is a revealing indicator of the effect of Wanadoo Interactive's predatory pricing in containing and eliminating competition.

Table 13: Weekly growth of Wanadoo Interactive and its competitors in terms of numbers of subscribers

<emplacement tableau>

(371) Between January and December 2001, Wanadoo Interactive recruited over [...]* new subscribers per week to its Wanadoo ADSL and eXtense services. The residential ADSL customer base of all its competitors put together grew at the rate of some [...]* subscribers per week in the same year, while its competitors on the cable modem highspeed segment for their part recruited fewer than [...]* new subscribers per week during that period. Above all, the growth rate of the cable operators fell off markedly from the first quarter of 2001 onwards, just when Wanadoo Interactive's sales were speeding up. Wanadoo Interactive thus grew in 2001 at a weekly rate that was [...]* times higher than all its competitors put together. The gap in sales growth rates widened in the third and fourth quarters of 2001, with Wanadoo Interactive expanding on the relevant market respectively [...]* times faster than all its competitors put together. (443) Overall, Wanadoo Interactive absorbed [80-90]*% of the growth in the high-speed market during the whole of 2001.

(372) During the first eight months of 2002, Wanadoo Interactive recruited more than [...]* new subscribers per week, [...]* times more than all its competitors put together. During that period, the cable operators continued to grow slowly; the competitors on the ADSL segment, while growing [...]* times more slowly than Wanadoo Interactive, nevertheless managed to narrow the growth gap somewhat with the dominant provider, thanks to the very considerable commercial efforts they made during the Internet festival (February-March 2002) and above all the Competition Council's decision banning France Télécom from continuing to sell its subsidiary's eXtense packages through its sales outlets until July 2002. Overall, during this period Wanadoo continued to absorb more than [70-80]* % of market growth.

(373) As will be seen from Table 14, the result of the distortions in growth rates brought about by Wanadoo Interactive's pricing policy was a very steep decline in the market shares of competing cable operators during the period covered by this Decision, and the maintenance of competitors in the ADSL segment at very low levels of market penetration.

Table 14: High-speed subscriber bases and market shares of Wanadoo and competitors

<emplacement tableau>

(374) In September 2001 competitors' market shares in the ADSL segment were so small that Wanadoo Interactive at the time regarded the cable operators as its only genuine rivals on the market in high-speed Internet access. (444) In the ADSL segment, it saw the "market at the end of 2001 [as] dominated by Wanadoo but not very active overall". (445)

(375) At the end of 2001, of a market in high-speed Internet access estimated by the Commission at 568 000 residential subscribers, Wanadoo Interactive held a [70-80]*% share, with its ADSL access alone serving [60-70]*% of the market. Wanadoo Interactive's share of the ADSL segment was [90-100]*%. The situation changed little during the first eight months of 2002: Wanadoo Interactive's share of the ADSL segment fell to [80-90]*%; overall, Wanadoo Interactive's ADSL products increased their share of the high-speed market from [60-70]*%, and if Câble Wanadoo is included the market share of the company as a whole remained stable at [70-80]*%.

(376) Not only did Wanadoo Interactive's market share rise very sharply in 2001 and thereafter remain high, the competition also became more fragmented. Wanadoo Interactive considerably increased its lead over its first rival, from [...]* to [...]*. While at the end of 2001 the leading competitor still had a market share of over 15%, by August 2002 no competitor had more than 10% of the market. Except for T-Online France, with a market share of nearly 8% at that time, none of the competing ADSL providers then had more than a 2.5% share of the market. Two trends can therefore be observed on the market between the beginning of 2001 and the autumn of 2002: a steep rise in Wanadoo Interactive's market share and increasing fragmentation of the surviving competitors.

(b) Interpretation of market trends

(377) ADSL customer preference for Wanadoo Interactive's services cannot be ascribed to any intrinsic superiority of the latter. The ADSL access part of the services offered by competitors was strictly equivalent to that provided by Wanadoo Interactive (in the case of the eXtense package) or France Télécom (in the case of the Wanadoo ADSL service), since in fact it corresponded in all cases to the Netissimo 1 service until the autumn of 2002. Furthermore, for the Internet access part proper, Wanadoo Interactive's competitors were infinitely less well placed on the high-speed market than on the dial-up Internet access market, although the service provided is not technically different.

(378) Wanadoo Interactive has ascribed the slow growth of its competitors up until autumn 2002 to a passive, wait-and-see attitude, pointing in support of this argument to their advertising expenditure for low-speed services in 2001. (446) The Commission considers, on the contrary, that the competitors' delay in rolling out their services in no way resulted from a failure to recognise the strategic importance of high-speed Internet access: they were simply waiting until France Télécom's wholesale prices and Wanadoo Interactive's retail prices were set at levels enabling competitors to draw up coherent business plans that would be acceptable to investors. AOL had as early as 1999 begun testing the water with a view to providing an ADSL service but abandoned the plan on grounds of insufficient profitability. Liberty Surf, which subsequently became Tiscali France, had also begun in 2000 to put together an offering based on an indirect access service called ADSL Connect ATM (447) but withdrew on account of the unsatisfactory level of wholesale prices, which it claimed created a margin squeeze. Easynet was for its part the first operator to sign up to the IP/ADSL service. (448) Nerim launched its offering at the beginning of autumn 1999, (449) at the same time as Club Internet. (450) It therefore cannot be argued that Wanadoo Interactive's competitors did not recognise what was at stake in high-speed services from the outset.

(379) Wanadoo Interactive's competitors in the ADSL high-speed Internet access segment confirmed that they were unable to align their prices on those charged by Wanadoo Interactive, given the costs involved, without incurring heavy losses. The price decisions taken by France Télécom in August 2001 led to an improvement in the situation. But the evidence collected by the Commission and the cost and price simulations submitted by the Internet service providers concerned demonstrate that, even after that date, aligning prices on those charged by Wanadoo Interactive was liable to generate losses. (451)

(380) The gradual emergence of different offerings from spring 2001, and above all autumn 2001, onwards reflects in most cases a strategy of mere passive presence on the market, the aim for the service providers concerned being to respond to spontaneous demand and upgrade their existing customers who were already subscribing to lowspeed services, without canvassing for and recruiting new customers. During the hearing held on 18 March 2002, the service providers AOL France and Tiscali France agreed that their strategy fitted that description. [...]*, Wanadoo Interactive referred [...]*. (452) [...]*, (453) [...]*.

(381) Wanadoo Interactive's pricing practices in fact forced its actual or potential competitors in the ADSL segment to choose between the following options:

- to align their prices on those of the dominant provider (or charge slightly less) and incur losses;

- or to continue to charge higher prices and risk being driven completely out of the market; (454)

- or to refrain from penetrating the market or a particular market segment, or enter it passively, minimising certain strategic items of expenditure such as the costs of customer acquisition, in particular advertising, (455) and adopting minimal quantitative sales targets. (456)

(382) The objective of the Internet service providers who took the first option was to gain or retain some visibility on an emerging market, without actively marketing the products concerned, (457) so as to avoid incurring excessive losses, since those service providers often did not have the scope for cross-subsidisation between business areas that the Wanadoo group had. Their aim was to respond to spontaneous demand from their existing customers, not genuinely to stimulate demand.

(383) The competitors that took the second or third options gave preference to financial security at the expense of their market presence. (458)

(384) Certain operators combined different options or switched between them over time. The example of Mangoosta is instructive here. Mangoosta was an independent service provider seeking considerable expansion nationwide on the residential high-speed Internet market. Initially, in late 2000, it launched its products on a slightly higher price base than Wanadoo Interactive, (459) with some commercial success. To stem the resulting losses, Mangoosta finally raised its prices by nearly 20% in late winter 2000-2001, (460) but this did not prevent it being placed under administration on 2 August 2001. Mangoosta's insolvency is a typical example of how a new entrant is driven out of the market by predatory pricing, where the new entrant, after having for a time set its prices at below costs, then adopts a more restrictive commercial policy and ends up being completely marginalised. (461)

(385) The growth strategy pursued by Wanadoo Interactive in 2001 and 2002, far from being based on competition on merit, involved on the contrary putting the financial strength of its competitors to the test. The effect of Wanadoo Interactive's pricing policy was thus to bring about a steep decline in its leading competitor's market share and prevent all its other competitors expanding, or even bar them completely from the market in high-speed Internet access for residential customers until October 2002. (462) The fact that, during the period while the abuse lasted, some of the surviving competitors increased their market share slightly does not mean that no abuse within the meaning of Article 82 of the Treaty existed or that it had no effects since, in the absence of the behaviour imputed to Wanadoo Interactive, the market shares of those competitors could have grown more significantly. (463)

2. Effects at the end of the abuse and in the medium term

(386) In its arguments, Wanadoo Interactive stressed the great vitality of the market in highspeed Internet access from September 2002 onwards. The Commission in no way disputes the fact that competitive conditions were restored after the entry into force on 15 October 2002 of the new charges for France Télécom's access (464) and routing (465) services, which had been presented as a remedy at the meeting of 19 March 2002. (466) On the contrary, the different pricing decisions taken by France Télécom both in the wake of these proceedings and in response to the opinion of 16 April 2002 issued by the ART Authority have had the effect not only of evening out the prices for the different technical solutions for implementing ADSL, including via local loop unbundling, (467) but also of very significantly reducing the cost of retail supply. Consequently, combining the IP/ADSL access and routing services that go to make up option 5 enables ISPs to provide a service based on an economically viable model, and there is also a wider range of access solutions which offer a realistic alternative to option 5. On peut donc estimer que la prise d'effet de ces différentes mesures tarifaires le 15 octobre 2002 a conduit à un déverrouillage du segment de l'accès à haut débit par la technologie ADSL.

(387) In the last four months of 2002, when its competitors were making unprecedented promotional and commercial efforts, Wanadoo Interactive continued to absorb [50- 60]*% of growth in the high-speed market, the balance being shared between ten or so competitors. During that period, Wanadoo Interactive in no way slackened its marketing efforts, stepping up the rate of sales and deciding, following the example of several competitors, to distribute modems free of charge from October 2002 onwards. Club Internet/T-Online France did not at the end of this period represent the major threat claimed by Wanadoo Interactive. (468) By the end of 2002, Wanadoo Interactive's market share had declined to [60-70]*% but, despite very great commercial efforts, none of its competitors had managed to acquire more than a 9% market share. In the first quarter of 2003, competitors stepped up their commercial efforts on the occasion of the Internet festival. At 31 March 2003 Wanadoo Interactive's market share stood at some [60-70]*%. A single competitor had succeeded in gaining a market share slightly exceeding 10%. Despite the breakthrough achieved by this player, whose subscriber base was still more than four times smaller than that of Wanadoo Interactive, competition was fragmented. Over the period between August 2002 and March 2003, Wanadoo Interactive still absorbed [40-50]*% of market growth, placing it well ahead of its leading rival.

(388) It is difficult to speculate on how market shares will evolve in the medium term on the basis of the trends observed during the months immediately following the abuse. However, even adopting extremely conservative growth assumptions for Wanadoo Interactive and optimistic assumptions for its immediate rivals, the chances of competitors catching up with Wanadoo Interactive appear to be practically nil, since no other company is recruiting new subscribers at a faster rate than the dominant company. (469)

(389) The most likely scenario is that Wanadoo Interactive will retain a market share persistently close to [50-60]*%. Consequently, although in the medium term the growth of competitors may lead to some erosion in Wanadoo Interactive's market shares in high-speed Internet access, it is quite possible that competitors will never be able to catch up and that Wanadoo Interactive's market power has been sustainably established thanks to the head start it gained during the existence of the abuse. Nevertheless, given the growth potential of this market in the years ahead, it is likely that the effects of the strategy pursued by Wanadoo until the end of 2002 will not be irreversible. On the one hand, the majority of the subscribers recruited by Wanadoo during the period at issue will probably remain loyal to that provider in the long term; on the other hand, it is likely that new subscribers will be distributed more evenly between the different providers, on condition that Wanadoo Interactive does not repeat its previous behaviour of below-cost selling.

(390) Wanadoo Interactive also argued that it ultimately mattered little whether its competitors would one day grow sufficiently to draw level with it, but what was important was to determine whether they were able to exercise competitive constraint on its behaviour.

(391) In terms of the principles at stake, Wanadoo Interactive's argument cannot be endorsed as it would run counter to the objective of undistorted competition enshrined in Article 3(g) of the Treaty. Furthermore, irrespective of any consideration to do with protecting the interests of undertakings established in the common market, the Commission takes the view that it is by no means neutral from the standpoint of collective welfare whether competitors are ever able to catch up with Wanadoo Interactive's market share, all the more so because the effects of the latter's position on the relevant market spill over onto related markets and revenues, such as on-line advertising, the sale of pay content and electronic commerce.

F. EFFECTS ON TRADE BETWEEN MEMBER STATES

(392) The effect of Wanadoo Interactive's predatory pricing policy was to make it difficult, if not impossible, to penetrate the market in high-speed Internet service provision for residential customers. Under these circumstances, it was difficult for Internet service providers established in other Member States to carry on high-speed Internet business in France without incurring heavy losses. In addition, the conditions in which this market had developed favoured Wanadoo Interactive on related markets linked to advertising and electronic commerce. Wanadoo Interactive's behaviour had repercussions on the structure of competition in the common market. (470) Trade between Member States may thus have been affected. (471)

G. ARTICLE 3 OF REGULATION NO 17

(393) Under Article 3 of Regulation No 17, where the Commission finds, upon application or on its own initiative, that there is an infringement of Article 82 of the Treaty, it may by decision require the undertakings concerned to bring such infringement to an end.

(394) Although some evidence suggests that the relevant products have been sold at above costs since October 2002, it is the Commission's duty to take every step to ensure that Wanadoo Interactive has genuinely and permanently put an end to its predatory policy on the market in high-speed Internet access and that it will in future refrain from repeating such behaviour. (472)

(395) It should be borne in mind here that in late March 2002, when Wanadoo Interactive was aware that it was the subject of proceedings initiated by the Commission in the previous December on account of predatory pricing practices, a scenario based on a return to below-cost pricing was being considered within its [...]* department. (473) This scenario involved [...]*, (474) [...]* (475) [...]*. (476) A resumption of such below-cost selling behaviour by Wanadoo Interactive would have the effect of seriously jeopardising the improvement in the competitive situation observed in the autumn of 2002, when the abuse stopped. Under these circumstances, the Commission considers it appropriate to adopt a decision. (477)

G. ARTICLE 15 OF REGULATION NO 17

(396) Under Article 15(2) of Regulation No 17, the Commission may by decision impose fines not exceeding 10% of turnover in the preceding business year where an undertaking has either intentionally or negligently infringed Article 82 of the Treaty.

(397) The extent to which Wanadoo Interactive was aware of the costs incurred, as shown among other things by the business plans for the services in question, is sufficient proof that it was fully conscious of the effects its prices would have on its short-term profitability, in view of the scale of its capture strategy. The company's internal documents also show that it was not unaware of the legal risks involved in below-cost selling. The company consequently knew that its behaviour could have the effect of distorting competition in the common market. The infringement was therefore committed intentionally. (478) In determining the amount of the fine, the Commission must take account of the gravity and duration of the infringement, in accordance with Article 15(2) of Regulation No 17, and of any aggravating or mitigating circumstances, as provided in the guidelines on the method of setting fines imposed pursuant to Article 15(2) of Regulation No 17 and Article 65(5) of the ECSC Treaty ("the guidelines"). (479)

1. Gravity of the infringement

(398) The assessment of the gravity of the infringement must take into account the nature of the abuse, its actual impact and the extent of the relevant geographic market.

1. Nature f the infringement

(399) In determining the gravity of the infringement, the Commission will take into consideration the fact that the infringement was deliberately aimed at preempting a booming market which is playing a key role in the advent of the information society. Both the Court of Justice and the Commission have repeatedly found against predatory pricing, i.e. below-cost selling by a dominant company.

2. Impact on the market

(400) Wanadoo Interactive's share of the market in high-speed Internet access for residential customers grew from [40-50]*% to [70-80]*% during the period in question. One competitor was eliminated, the market shares of several competitors fell very steeply, and those of the other surviving competitors grew very slowly or stagnated at an insignificant level. (480) Although it is not proven that the developments observed on the market can be ascribed exclusively to Wanadoo Interactive's behaviour, the latter's predatory pricing policy has undoubtedly had adverse repercussions on competition.

(401) The impact of the infringement is reinforced by its timing, coinciding with a period of strong market demand for high-speed Internet access. The first high-speed access services by cable modem became available in France in the course of 1997, but conditions on the retail market were radically affected by the commercial introduction in late 1999 of ADSL high-speed access solutions. The infringement coincided with the boom in ADSL in France and a period in which alternative telecom operators were endeavouring, by means of local loop unbundling, to implement technical solutions that were substitutable for France Télécom's access network. (481) The infringement has radically reshaped the market structure.

3. Extent of the relevant geographic market

(402) The anticompetitive effects of the behaviour under examination extended over the entire territory of one Member State, France.

4. Conclusion on the gravity of the infringement

(403) The behaviour under examination constitutes an infringement with a distinct anticompetitive dimension in a sector that is regarded as playing a strategic role in the development of the information society. It is the reflection of a plan to preempt a new market with strong potential, occurring at a key stage in its development. By its very nature and by the impact it has had on the market of an entire Member State, the conduct in question constitutes a serious infringement.

(404) In view of the foregoing, the basic amount of the fine, determined according to the gravity of the infringement, should be set at euro 9 million.

2. Duration of the infringement

(405) Wanadoo Interactive takes the view that the infringement, in so far as there was any infringement, began in late March 2001, when mass marketing of eXtense was launched, and ended in August 2001, after which the company only just covered its variable costs adjusted according to the Commission method. Wanadoo thus argues that the period of below-cost selling did not exceed four months and was consequently too short to have had appreciable effects on the market structure.

(406) The Commission would begin by pointing out that 2001 was preceded by a period of over a year in which Wanadoo Interactive sold its Wanadoo ADSL service at a loss. Moreover, from January and February 2001, Wanadoo Interactive embarked on a growth path for its high-speed subscriber base which was around twice as fast as its growth rate in 2000. The mere announcement of the roll-out of the eXtense package had immediate effects on the market, in particular on orders received by Mangoosta. (482) Although such sales, particularly after January 2001, displayed some features of abusive behaviour, the Commission does not consider it appropriate to find that they constituted an infringement, in view of the characteristics of the market during the period in question. (483) The infringement is therefore found to have begun on 1 March 2001.

(407) Given the intention to preempt the market, which took the form of a strategy for driving out competitors, the Commission considers itself bound to apply a variable cost recovery test prior to August 2001 and a full cost recovery test after August 2001, in line with the relevant case-law. Under such tests the abuse came to an end on 15 October 2002, when the remedy presented by France Télécom in March 2002 came into effect.

(408) The Commission therefore finds that the infringement lasted nineteen and a half months. It is therefore an infringement of medium duration. In accordance with the guidelines, the Commission considers it appropriate to increase the fine by 15% for the period between March and October 2002.

(409) It follows that the basic amount of the fine should be euro 10.35 million.

3. Aggravating or mitigating circumstances

(410) Wanadoo Interactive called for no fine to be imposed on the ground that the cost standard applied departed from the precedents established by Community case-law. The Commission cannot accept this argument. The method used by the Commission is not new: it is based on the rule established by the judgment in Akzo. In applying the Akzo tests the Commission has simply spread the costs of customer acquisition over a period similar to the average customer lifetime. This is an adaptation to the circumstances of the case which is in fact favourable to Wanadoo Interactive, in comparison with the conclusions that would have been reached by means of a pure accountancy-based analysis, more in line with the approach taken in previous decisions. (484)

(411) Neither can the extent of the wholesale price adjustments made by France Télécom in summer 2002 be regarded as a mitigating circumstance. The scale of the effort made by France Télécom in respect of its wholesale charges is attributable to the combined action of the regulatory and competition authorities at national and Community level, not to any spontaneous change of policy on its part. France Télécom has furthermore challenged before the national courts the new unbundling charges, which make a significant contribution to restoring healthy competitive conditions.

(412) Wanadoo Interactive lastly sought to rely on its openness and full cooperation in the course of these proceedings as mitigating circumstances. The Commission notes, however, that the company merely complied in a normal manner with its obligations under Regulation No 17 with regard to the provision of information to the Commission.

(413) The Commission therefore considers that there are no aggravating or mitigating circumstances in this case.

(414) In conclusion, the amount of the fine to be imposed in accordance with Article 15(2)(a) of Regulation No 17 should be set at euro 10.35 million,

HAS ADOPTED THIS DECISION:

Article 1

From March 2001 to October 2002, Wanadoo Interactive infringed Article 82 of the EC Treaty by charging for its eXtense and Wanadoo ADSL services predatory prices that did not enable it to cover its variable costs until August 2001 or to cover its full costs from August 2001 onwards, as part of a plan to preempt the market in high-speed Internet access during a key phase in its development.

Article 2

Wanadoo Interactive shall immediately bring to an end the infringement referred to in Article 1, in so far as it has not already done so. It shall, in the context of the ADSL services referred to in Article 1, refrain from any behaviour having an object or effect identical or similar to that of the infringement.

Article 3

Wanadoo Interactive shall forward to the Commission, at the end of each year up to and including 2006, the revenue account for its different ADSL services, showing its accrued income, operating costs and customer acquisition costs.

Article 4

A fine of euro 10.35 million is hereby imposed on Wanadoo Interactive for the infringement referred to in Article 1.

The fine shall be paid in euros, within three months of the date of receipt of notification of this Decision, to the following bank account: Account No 001-3953713-69 (IBAN Code: BE 71 0013 9537 1369; SWIFT Code: GEBABEBB) of the European Commission at Fortis Banque/Fortis Bank, Rue Montagne du Parc/Warandeberg 3, 1000 Brussels.

After expiry of that period, interest shall become automatically payable at the rate applied by the European Central Bank to its main refinancing operations on the first day of the month in which this Decision was adopted, plus 3.5 percentage points, namely 5.6%.

Article 5

This Decision is addressed to Wanadoo Interactive, 48 rue Camille Desmoulins, F-92791 Issy les Moulineaux Cedex.

This Decision shall be enforceable pursuant to Article 256 of the EC Treaty. Done at Brussels, [...]

For the Commission Notes :

1 OJ 13, 21.2.1962, p. 204/62.

2 OJ L 148, 15.6.1999, p. 5.

3 OJ L 354, 30.12.1998, p. 18.

4 Decision C(99)2435 of 27 July 1999.

5 Asymmetric Digital Subscriber Line.

6 ADSL is technology enabling the terminal part of the telephony network to be digitised using highspeed bandwidth over a single copper pair. The basic telephony service occupies a bandwidth of a few kilohertz. By setting up a modem and specific ADSL filter at each end of the copper pair, bandwidths ranging from 20 kHz to more than 1 megahertz can be used. The enriched line can process bidirectional data traffic ranging from a few hundred kilobit/s to several megabit/s while also enabling simultaneous telephone use.

7 For the sake of simplicity it will be referred to in what follows as "Wanadoo Interactive" irrespective of the period in question.

8 Earnings before interest, taxes, depreciation and amortisation.

9 Information for that year is taken from Wanadoo's 2000 annual report.

10 "Paper" advertising alone generated revenue of more than euro 558 million, i.e. half total turnover.

11 With revenue of almost euro 250 million.

12 There are no systematic data on this subject. However, on the basis of the figures published by the Association des Fournisseurs d'Accès or the Autorité de régulation des télécommunications (ART), Internet access via a fixed telephone line was still thought to account for more than 95% of total residential customer logons at the end of the first quarter of 2001.

13 WAP services on mobile telephony networks.

14 Source: Association des fournisseurs d'accès à Internet, data as at 30 June 2001, 31 March 2002 and 30 September 2002.

15 Low-speed Internet access entails making a specific connection every time the user logs on, which means he cannot make or receive telephone calls simultaneously on the same line.

16 Optic fibres and leased lines are two further options for business users. At present, rapid penetration of the mass customer base appears to be beyond the capacity of wireless local loops and satellite for technical reasons.

17 Saturation problems may also occur in view of the fact that customers share some traffic merger infrastructure, [...]*.

18 High-speed Internet services via cable modem are slightly less attractive than ADSL, at least for very heavy users, because a supplementary charge is usually made once users exceed a certain monthly traffic threshold (or a certain number of hours online each month).

19 Company taken over by Nerim in August 2001.

20 Infonie's ISP activities were taken over by Liberty Surf on 23 November 2001.

21 The retail price of the service provided by France Télécom to final users varied from euro 30.18 inclusive at the beginning of 2001 to euro 25 inclusive as of October 2002.

22 This offering was referred to as Xtense during the first few months of 2001. It was then renamed eXtense, the name which will be used hereinafter.

23 The starter pack sold by Wanadoo Interactive comprises a USB plug and play modem, a configuration CD Rom, a modem installation CD Rom, three distributed filters, an RJ 11 connection cable, an installation manual and user guide and a 30-minute helpline card.

24 Decision No 02-MC-03 of 27 February 2002 on the referral and application for interim measures lodged by T-Online France.

25 [...]*.

26 Real average revenue differs from nominal revenue as explained in points 28 to 34 below.

27 In point 7 of its reply dated 24 April 2002, Wanadoo Interactive provides a detailed explanation of the reasons for this time-lag between registering customers in the subscriber base and incorporating subscriptions in turnover (file, p. 4714 et seq.). According to Wanadoo Interactive, this analysis is also valid for 2002 (file, p. 6096).

28 Article 5 of the general conditions of sale of the eXtense package service provides for customers to be paid compensation equivalent to one month's subscription if Wanadoo Interactive is unable to resume service during the business day following the day on which the interruption is notified. Over and above this legal obligation, dissatisfied customers are offered free subscriptions lasting several months with the aim of retaining them.

29 [...]* (file, p. 4657 et seq.).

30 [...]* (file, p. 4915 et seq.).

31 Lastly, in its reply of 4 March 2003 (paragraph 77), Wanadoo Interactive disputed ever having[...]*, but this is directly contradicted as regards 2001 by the in-house documents referred to in footnotes 29 and 30 to this Decision and as regards 2002 by Wanadoo Interactive's reply of 13 December 2002 to a request for information from the Commission (file, p. 6097).

32 See Annex 16. For 2002, the discrepancies calculated in this annex do not take account of subscriptions of several months offered free of charge to customers when they sign up, which are regarded in the Commission's analysis as costs relating to special offers. No special offers in the form of free monthly subscriptions for new customers were made available in 2001.

33 [...]* (file, pp. 4994-4995).

34 Reply of 4 March 2003, paragraph 78.

35 These documents and the Wanadoo Interactive in-house documents referred to below were obtained in

the course of the inspection carried out on the company's premises on 4 and 5 April 2002.

36 [...]* (file, p. 2745.)

37 [...]* (file, p. 2766), [...]*.

38 [...]* (file, p. 2870).

39 In any event, Wanadoo Interactive learnt of these worse-than-anticipated results before [...]* file, p. 2768).

40 Letter from Wanadoo Interactive of 27 September 2002, point 2.

41 Letter from Wanadoo Interactive of 28 November 2002, point 4.

42 Case C-62-86 Akzo Chemie v Commission [1991] ECR I-3359, paragraph 71.

43 Contract No 00000031 between France Télécom Interactive and France Télécom for the provision of IP/ADSL 1 and IP/ADSL 2 services enclosed with the letter from France Télécom dated 20 February 2001 in reply to the Commission's request for information dated 29 January.

44 However, it should be noted that the bespoke routing facility made available to Wanadoo Interactive, which remained valid until 1 August 2001, did not provide for any monthly subscription for the connection component.

45 Point 1 of the letter from France Télécom dated 24 April 2002 in reply to the request for information from the Commission dated 12 April 2002.

46 [...]*file, p. 4645 et seq.). [...]* (file, p. 4503 et seq.) [...]* (file, p. 3545 et seq.).

47 Letter from France Télécom dated 20 December 2002 (file, p. 6107). Measures taken between July and September 2002 (letter from France Télécom dated 5 February 2003) point to[...]*.

48 Annexes 1 and 2 to the letter from France Télécom dated 24 April 2002 (file, p. 4722 et seq.).

49 Source: tables provided by France Télécom on 3 and 29 May 2002 in reply to the Commission's request for information dated 14 March 2002; letter to the Commission from Mr Parfait, the CEO of Wanadoo Interactive, dated 7 May 2002, p. 3 (file, p. 4897 et seq.);[...]*.

50 [...]* (file, p. 5904).

51 [...]* (file, p. 6807 et seq.).

52 [...]* (File, p. 5904),[...]*.

53 See Annex 17 to this Decision.

54 Point 6 of France Télécom's reply (dated 6 December 2001) to the Commission's request for information dated 23 November 2001.

55 See Annex 18 to this Decision.

56 Explanations provided by France Télécom in point 4 of its letter to the Commission dated 13 November 2002.

57 Letter from France Télécom dated 20 December 2002 (file, p. 6108).

58 Commission Decision IV/M.2817- Barilla/BPL/Kamps of 25 June 2002 (OJ C 198, 21.8.2002, p.4) and Commission Decision IV/M.430 - Procter and Gamble/VP Schickedanz of 21 June 1994 (OJ L 354, 31.12.1994, p. 32), referred to in paragraphs 260 and 261 of Wanadoo Interactive's reply of 23 October 2002 to the statement of objections dated 9 August 2002. The 1994 Decision clearly states that a campaign in a particular magazine or media channel has a fixed cost irrespective of turnover, but it does not draw the conclusion that all advertising and promotional campaigns in general consist wholly of fixed costs.

59 William J. Baumol and Robert D. Willig, 'Fixed Costs, Sunk Costs, Entry Barriers and Sustainability of Monopoly', Quarterly Journal of Economics, August 1981, e.g. p. 407.

60 See A. Mikol, J.C. de Guardia and H. Stolowy, Comptabilité analytique et contrôle de gestion, Dunod, 1991, p. 100.

61 See Annex 21. The linear correlation coefficient is calculated according to the usual mathematical formula, as is the correlation between covariance and the square root of total variance. Correlation rates always range between -1 (inverse correlation) and +1 (complete correlation), with values around 0 indicating zero correlation. Wanadoo Interactive claims in its reply of 4 March 2003 (paragraph 60) that a coefficient of [>0,53]* is not significant, but contradicts itself by providing, in support of its comments, an annex setting out elements of statistical theory which identifies precisely that figure ([>0,53]*) as the boundary between "average correlation" and a "strong correlation". In addition, in view of the range of available data pairs ([...]* variable pairs), a coefficient of [>0,53]* must be regarded as a particularly significant indicator of a statistical correlation between the two series of data (advertising expenditure on the one hand and growth in the subscriber base on the other), with a very low risk of error. It can be demonstrated mathematically that, for a range of variable pairs of this type, these pairs must be regarded as significantly correlated with a risk of error of less than 1% where the correlation coefficient exceeds 0.53. That condition is met in this instance.

62 In paragraph 60 of its reply of 4 March 2003 (file, p.6871), Wanadoo Interactive claimed that a correlation link did not necessarily indicate a relationship of cause and effect because of what statisticians refer to as a "confounding factor". The examples quoted by Wanadoo Interactive (a strong correlation between UK tax revenue and crime in Japan, or between wheat prices and the rodent population) are not relevant here. Above all, the company is unable to identify a potential "confounding factor" in the case in point.

63 Wanadoo Interactive stated that, in its analysis, advertising cannot be regarded as a variable cost because it does not maintain a constant proportional relationship with growth in the subscriber base (paragraphs 63 to 65 of the reply of 4 March 2003, file, pp. 6873-6875). In the case in point, the Commission takes the view that growth in the subscriber base is not strictly proportional to advertising expenditure (which would suggest a function of the type y=a.x), but considers that it is linked to that expenditure by a linear relationship of the type y=a.x + b (see values of a and b as calculated in Annex 21). In practical terms, as simple common sense would suggest, that means that growth in the subscriber base is partly proportional to advertising expenditure and partly generated by exogenous factors.

64 The period covered by this Decision begins on 1 March 2001, but the figures given in Tables 2 and 3 include the months of January and February 2001. This is because even though the figures specifically for January and February 2001 were not relevant to the analysis, the information available was not differentiated finely enough to make it possible to separate and disregard them. The fact that those figures are included in Tables 2 and 3 does not affect the assessment of the duration of the infringement.

65 At the time of this Decision it is not possible to establish the real costs for the first half of 2002 with any precision. In the data assembled from the figures supplied by Wanadoo Interactive, which come directly from the company's cost accounts (Wanadoo Interactive's letter of 13 December 2002, p. 6 (file, p. 6099)), there are certain cost components which are not included. The real rate of recovery of costs was probably appreciably lower.

66 It should be pointed out, however, that Wanadoo Interactive does not carry out any depreciation of this kind in its own accounts, where the expenditure in question is entered at the time it arises, in the same way as ordinary current expenditure.

67 See paragraphs 321 and 322 of Wanadoo Interactive's reply of 23 October 2002 (file, p. 6314).

68 Wanadoo Interactive reply of 4 March 2003, paragraph 29 (file, p. 6866).

69 Wanadoo Interactive arrives at an annual turnover rate of [...]* % for eXtense and [...]* % for Wanadoo ADSL. In the case of Wanadoo ADSL, however, the firm has mistakenly taken account only of cancellations, without also including customer migration to other products offered by the same firm. Wanadoo ADSL's real rate of customer turnover in annual equivalent ((out-migration + cancellations) ÷ average clientele) is [...]* %, which according to Wanadoo Interactive would be the equivalent of an average customer life of [...]* years.

70 Oreka writes down its spending on acquisition over an average period of 18 months (file, p. 6113), while Tiscali and T-Online France spread the net costs of their special offers on ADSL modems over twelve months and two years respectively (file, pp. 5365 and 5378).

71 The ART depreciates the euro 53 charged by France Télécom for access to the service over a period of three years. See ART opinion No 02-346 of 30 April 2002 on tariff decision No 2002033 relating to the development of the IP/ADSL routing offering and tariff proposals for IP/ADSL offerings announced by France Télécom, paragraph III.1.

72 See paragraph 86.

73 Akzo, paragraph 71.

74 See section I.G.

75 The detailed calculations are set out in Annexes 1 to 8.

76 See Annex 15 to this Decision.

77 In its reply of 4 March 2003, paragraphs 14-20 (file, pp. 6861-6863), Wanadoo Interactive contests this method of weighting, arguing that the Commission earlier weighted the figures by reference to the customer bases of the two services. But the Commission considers that in order to calculate an overall rate of cost recovery (the total revenue of the two services divided by the costs of the two services together), the rates of recovery observed for the two products should be weighted by reference either to the total revenue or to the total cost of each of them. The Commission does not possess information on the total cost over the entire period, and has accordingly taken the total revenue as the weighting factor here. To choose the customer base as the weighting factor would be arithmetically wrong: it would seriously overestimate the weight to be given to the Wanadoo ADSL product, treating it as if its costs and revenues were the same as those of the eXtense service, whereas in fact its costs and revenues per subscriber are only half those of eXtense. The Commission takes the view that it cannot be bound by an error of calculation made at an earlier stage in the proceedings, provided that in order to enable the undertaking to defend itself it has given it the opportunity to submit its observations on the rectification of the mistake, as it did in this case in its letter of 16 January 2003.

78 Annexes 1 and 2.

79 This fact is evidenced in all of Wanadoo Interactive's in-house documents relating to profit-and-loss forecasts, product budgets and the financial results of the various offerings. An exhaustive list of the documents gathered at Wanadoo Interactive would be long, and the Commission will here confine itself to citing, by way of example, [...]* (file, pp. 2869 and 2870), [...]* (file, p. 2770). [...]* (file, p. 2982)).

80 The need for a gross margin that makes a sufficient contribution to the recovery of the company's full costs is pointed out in [...]* (File, p. 2885)).

81 For the rate of recovery shown for the two together, the individual figures are weighted in line with the respective share of turnover, as explained in paragraph 82 (see Annexes 13.2, 14.3 and 15.1 to this Decision).

82 For example, if acquisition costs are spread over ten years, the rate of recovery of the full costs of eXtense is [...]*% from August to December 2001 and [...]*% from January 2002. In the case of Wanadoo ADSL, with costs spread over the same period, the rate of recovery is [...]*% for the period August to December 2001 and [...]*% thereafter.

83 Wanadoo Interactive's reply of 23 October 2002, paragraph 289 and Appendix D, Annex 9.

84 Reply of 23 October 2002, paragraphs 282 and 286.

85 Annexes 3.2, 4.2, 5.2, 6.2, 7.2, 8.2 and 15.2.

86 Wanadoo Interactive's reply of 23 October 2002, paragraphs 333 and 341 (file, pp. 6316 and 6317), and reply of 4 March 2003, paragraphs 23-25 (file, p. 6865).

87 See Annex 20 to this Decision.

88 [...]* (File, pp. 2901 and 2907).

89 The short-term predation test advocated by Areeda and Turner is likewise based on the concept of pricing below the costs than can be "reasonably anticipated".

90 Reply of 23 October 2002, footnote 101 (file, p. 6272).

91 [...]* (Letter from Wanadoo International to the Commission, 7 May 2002, note 2 (file, p. 4904)).

92 [...]* annexed to the reply of 7 May 2002 to a Commission request for information (file, p. 4913)).

93 It will be remembered that up to the autumn of 2002 the distribution of subscribers between the Paris region and the rest of the country had an appreciable effect on costs, because France Télécom's routing charges were [...]* higher in the provinces.

94 Paragraphs 49 and 56 of the statement of objections of 19 December 2001.

95 These costs are unadjusted, unlike those used under the Commission's method. But the difference does not affect the comparison of the results, because Wanadoo Interactive's evaluation does not include acquisition costs, and the effect of acquisition costs is greatly minimised when they are spread over four years in accordance with the Commission's method. The figures in the Commission's estimates and in the target profit-and-loss account are consequently fairly similar.

96 At the very beginning of 2002 Wanadoo Interactive was not in a position to know that the bandwidth per subscriber would ultimately be evaluated on a [...]* basis[...]*. Indeed this point is not even referred to in Wanadoo Interactive's reply of 4 March 2002 to the statement of objections of 19 December 2001.

97 In the first two months of 2002, eXtense accounted for [...]*% of Wanadoo Interactive's sales of ADSL products on the residential market, as compared with [...]*% in the third quarter of 2001. [...]* only in February 2002, when the Competition Council banned the marketing of eXtense packages in France Télécom shops. This [...]* was not foreseeable.

98 [...]* (file, p. 3827).

99 This expression refers to the product that was later called eXtense.

100 [...]* (file, p. 3124).

101 [...]* (file, p. 3764).

102 [...]* (file, p. 4261).

103 Quoted in the judgment of the Paris Appeal Court in France Télécom v T-Online, 9 April 2002, p. 4.

104 [...]* (file, p. 3764).

105 [...]* (file, p. 3276).

106 Wanadoo Interactive's assessment of its market share here is well below its real share at that time.

107 [...]* (file, p. 4155).

108 [...]* (file, p. 4157).

109 [...]* (file, p. 3095).

110 [...]*

111 [...]* (file, p. 2708 et seq.).

112 File, p. 3316.

113 [...]* (file, p. 2897).

114 Reply of 23 October 2002, paragraph 181 (file, p. 6289).

115 Reply of 23 October 2002, paragraph 182 (ibid.).

116 Reply of 23 October 2002 to the statement of objections of 9 August 2002, paragraphs 162 and 175.

117 It should be noted, too, that their tone is measured and quite unlike the examples of belligerent language and rousing rhetoric cited by Wanadoo Interactive (reply of 23 October 2002, footnote 141).

118 Reply of 4 November 2002, footnote 150.

119 [...]* (file, p. 3316): [...]*.

120 [...]* (file, p. 3155). [...]* (file, p. 3367), [...]* (file, p. 3345).

121 [...]* (file, p. 2790).

122 [...]*

123 [...]* (file, p. 3362).

124 [...]* (file, p. 2598); [...]* (file, p. 2600).

125 [...]* (file, p. 3134).

126 These scenarios are presented in simplified form in Annex 10.

127 File, pp. 2606-2651.

128 [...]* (file, pp. 3133-3134).

129 See Annex 19 to this Decision.

130 [...]* (file, p. 3055).

131 [...]* (file, p. 2642).

132 In these estimates the total cost of acquiring a customer is allocated in its entirety to the year in which the expenditure arose.

133 [...]* (file, p. 2626).

134 [...]* (file, p. 2652 et seq.).

135 [...]* (file, p. 2974).

136 [...]* (file, p. 3323).

137 This version was based in particular on the following assumptions: cost of Netissimo access FRF[...]*, cost of conveyance of IP traffic FRF [...]*, and cost of international connectivity FRF [...]* per month. The last component proved to be an overestimate. But the first two components proved to be underestimates, as compared even with the average costs actually noted after July 2001. These assumptions were reproduced in the definitive 2001 budget at the beginning of 2001.

138 [...]* (file, p. 3275).

139 [...]* (file, p. 4351).

140 [...]* (file, p. 3020). [...]*

141 [...]* (file, p. 2803).

142 [...]* (file, pp. 3233 and 3234).

143 [...]* (file, p. 3211).

144 [...]* (file, p. 3106). [...]* (, p. 2754).

145 [...]* (file, p. 2921 et seq.). [...]* (file, p. 3110 et seq.).

146 [...]* (file, p. 3110 et seq.).

147 [...]* (file, p. 3511).

148 [...]* (file, p. 2978).

149 [...]* (file, p. 2870).

150 [...]* (file, p. 2880). [...]*

151 [...]* (file, p. 3097).

152 [...]*

153 [...]* (file, p. 2826 et seq.).

154 This expression referred to the ADSL access service IP ADSL, also known as Netissimo 1.

155 This refers to the IP routing service for ADSL traffic.

156 [...]* (file, p. 3080 et seq.).

157 Reply of 23 October 2002, paragraph 101.

158 [...]* (file, p. 3294 et seq.).

159 It should be pointed out, however, that in June 2001 the ART, in a public opinion on Wanadoo Interactive's ADSL services, referred quite directly to the possibility of a predatory policy: "the terms of the offer made [by France Télécom] to Wanadoo are such that Wanadoo has to bear substantial losses, given the level of prices charged on the residential market since packages have been marketed. The situation might be evidence of predatory behaviour on the part of France Télécom and its subsidiary Wanadoo on the market in ADSL services for the general public" (ART opinion No 01-548 of 19 June 2001 on tariff decisions Nos 2001482 on the development of Netissimo 1 and Netissimo 2 services and 2001480 on the development of the supply of IP/ADSL access provision and IP/ADSL routing provision).

160 [...]* (file, p. 3632).

161 Decision No 02-MC-03 of 27 February 2002 regarding a reference and an application for interim measures submitted by T-Online France (file, pp. 1557 to 1569.)

162 Competition Council decision No 02-D-46 of 19 July 2002 regarding the application of Article 3 of decision No 02-MC03 of 27 February 2002, cited in paragraph 247 of the reply of 23 October 2002 (file, p. 6300).

163 [...]* (file, pp. 1673 to 1688).

164 [...]* (file, p. 5023).

165 [...]* (file, p. 2792).

166 [...]* (file, p. 5096)).

167 [...]* (file, p. 3109). [...]* (file, p. 3214)). [...]*

168 Decision No 02-D-40 of 25 June 2002 regarding a reference and an application for interim measures lodged by Olitec, referred to in paragraph 249 of Wanadoo Interactive's reply of 23 October 2002 (file, p. 6300).

169 File, pp. 1580 to 1590.

170 [...]* (file, p. 3036).

171 In opinion 02-35 of 9 January 2002 to the Competition Council on an application from T-Online France, paragraph 3.4, the ART states that by this partnership offer "France Télécom would very strictly circumscribe the Internet service providers' prices (selling prices to end users) and the ISPs' commercial policy by fixing the price to be paid to the distribution network" (file, p. 1553).

172 Paragraph 250 of the reply of 23 October 2002 (file, p. 6300).

173 In Article 4 of its decision No 02-D-46, already referred to, the Competition Council prohibits the implementation of the partnership offer in the following terms: "France Télécom is required to suspend any multi-ISP offer intended to be marketed in large-scale retailing which associates the provision of its ADSL Netissimo service with an Internet service provision offering with the same characteristics as those described in Annex 10 to [T-Online's] reference" (file, p. 1569).

174 Lawyers representing Wanadoo Interactive were also present at this meeting.

175 France Télécom's proposed charges for IP/ADSL access and routing were made public on 11 April 2002 (file, p. 2564 et seq.).

176 This information was accompanied by an exhaustive list of all the documents in the Commission file, updated since January 2003. Wanadoo Interactive submitted observations on some of these new papers in a letter of 18 June 2003.

177 Letters from Wanadoo Interactive's lawyers dated 14 January 2002, 22 January 2002, 8 February 2002, 10 September 2002 and 30 September 2002 (file, respectively pp. 847, 871, 5546, and 5854 et seq.).

178 The report can be consulted freely on the Internet site of the Directorate-General for Competition (http://europa.eu.int/comm/competition/antitrust/others/sector_inquiries/local_loop/local_loop_unbundl ing_inquiry.pdf). The Commission's letter of 17 January 2002 informed Wanadoo Interactive that the report and its annexes would be published on the site (file, p. 858).

179 OJ C 23, 23.1.1997, p. 3.

180 Commission letters to Wanadoo Interactive's lawyers, 11 and 15 October 2002 (file, pp. 5888 and 5893).

181 Letters to Wanadoo Interactive's lawyers from Commission departments, 17 and 28 January 2002, 18 February 2002, and 20 September 2002, conversation of 23 January 2002, and letter to the same lawyers from the Hearing Officer, 4 March and 15 October 2002 (file, pp. 852, 870, 909, 927 bis, 5561 and 5897).

182 The download speed up to 768 kbit/s made available by Freesurf on 23 October 2001 was, until the autumn of 2002, the only one which differed from all the other, hitherto relatively similar, download speeds on offer. As from the autumn of 2002, at the end of the period covered by this Decision, a greater variety of download speeds became available on the market.

183 Reply of 4 March 2002, p. 22 et seq. (file, p. 1727 et seq.).

184 Reply of 4 March 2002, p. 20 (file, p. 1725).

185 Reply of 4 March 2002, p. 18 (file, p. 1723).

186 Reuters article transmitted on 11 October 2002 at 11.23, "Wanadoo launches an ADSL facility at 30 euros a month".

187 See, for example: Federal Communications Commission, Applications for consent to the transfer of control and licences and section 214 authorizations by Time Warner Inc. and America Online Inc., Transferors, to AOL Time Warner Inc., transferee, CS Docket No 00-30, FCC, 11.1.2001, points 69 and 71 in particular.

188 This view is also supported in the academic world. See Jerry A. Hausman, J. Gregory Sidak and Hal J. Singer, 'Cable Modems and DSL: Broadband Internet Access for Residential Customers', American Economic Association Papers and Proceedings, Vol. 91, 2001; Robert W. Crandall, J. Gregory Sidak, Hal J. Singer, 'The Empirical Case against Asymmetric Regulation of Broadband Internet Access', March 2002.

189 These figures are taken from a study drawn up by NetValue in October 2001, cited in Le Journal du Net.

190 Wanadoo Interactive's chief executive officer quoted on 20 August 2001 in Le Journal du Net.

191 [...]*. Document found during the inspection.

192 [...]* study carried out by [...]* (file, p. 4147).

193 File, p. 4586.

194 Paragraph 37 of the reply of 23 October 2002 (file, p. 6255).

195 AOL had launched an unlimited low-speed access package in 2000, but had to double the prices charged for it in July 2001 because of costs. It was only as from August 2002, thanks to changes in interconnection charges, that Tiscali and AOL were able to introduce unlimited low-speed access facilities that were more attractive and economically more profitable. Wanadoo Interactive's low-speed facility known as "Intégrale 100 heures" is not strictly speaking an unlimited access facility, since it provides for a maximum amount of use that is less than the average period of use of high-speed users. Indeed, this package met with a fairly limited take-up, with [...]* subscribers on 31 March 2002, which is [...]* less than Wanadoo Interactive's ADSL subscribers on that date. Moreover, this facility was replaced at the end of 2002 in the catalogue by "Intégrales 60 heures". In Annex 20 to its reply of 23 October 2002, Wanadoo Interactive provides [...]* witness document 27 attached to the reply of 23 October 2002 quoting the following remarks by the president of Club Internet: "Unlimited packages make sense only in high-speed access, for two reasons: it is in this type of intensive Internet use that the convenience and speed of high-speed access are fully justified and, in addition, only this type of use allows the telephone line to be left free, which is crucial if one wants to be able to continue to use the telephone".

196 Statement made by the representative of Easynet France at the hearing held on 18 March 2002.

197 Decision of 26 May 1997 in Case IV/M.920 - Samsung/AST (OJ C 203, 3.7.1997, p. 3); Decision of 30 September 1999 in Case IV/JV.22 - Fujitsu/Siemens (OJ C 318, 5.11.1999, p. 15); Decision of 31 January 2002 in Case IV/M.2609 - HP/Compaq (OJ C 39, 13.2.2002, p. 23).

198 Reply of 4 March 2002, paragraphs 80 to 84.

199 The quotation used by Wanadoo Interactive (paragraph 82 of the reply of 4 March 2002) is: "a distinction between the different categories of personal computers is not justified" (file, p. 1729).

200 HP/Compaq Decision, paragraphs 20 to 22.

201 Decision of 12 May 1999 in Case IV/M.1519 - Renault/Nissan (OJ C 178, 23.6.1999, p. 14), paragraphs 19 and 20.

202 Applications for consent to the transfer of control and licences and section 214 authorizations by Time Warner Inc. and America Online Inc., Transferors, to AOL Time Warner Inc., transferee, CS Docket No 00-30, 11.1.2001, point 63.

203 Integrated systems digital networks. This technology allows download speeds up to 128 kbit/s, appreciably lower than ADSL performance. Internet access using this mode is generally charged by minute of use, in contrast to high-speed offerings.

204 OECD report dated 29 October 2001 entitled The Development of Broadband Access in OECD Countries, p. 6.

205 Office of Telecommunications, Direction to Resolve a Dispute between BT, Energis and Thus concerning xDSL Interconnection at the ATM Switch, 21.6.2002, paragraph E 12.

206 Source: BDRC Ltd, "The Development of Broadband Access Platforms in Europe", August 2001, p. 79.

207 It cannot be ruled out that these new low-speed ADSL offerings are examples of disguised unlimited low-speed access packages, presented more attractively as new technology, and offering the advantage of leaving the telephone line available for use. One of the objectives of these offerings would be to develop unlimited offerings not affording high-speed performances, but allowing service providers to avoid having to pay contributions towards financing the universal service, which until 2003 were based on the amount of low-speed traffic, and imposing a strain on their economic equilibrium. In its publication Internet, un bilan du marché of 18 March 2003, the ART uses the term "medium speed" to describe this type of offering.

208 [...]* file, p. 3104 [...]* file, p. 3121 [...]* pp. 3143 et seq. of the file [...]* file, pp. 4074 et seq. [...]* file, pp. 4136 et seq.

209 [...]* file, p. 3222

210 [...]* file, p. 2083

211 Wanadoo press release, 26 July 2001.

212 See Annex 24.

213 The differences shown in Annex 24 underestimate the average price difference, since they are based on calculation of an arithmetical average. Calculation of the price difference on the basis of an average weighted to take account of the size of each provider's subscriber base would have produced greater differences.

214 Reply of 4 March 2002, p. 22, paragraphs 68 to 72 (file, p. 1726).

215 It should be noted that Wanadoo Interactive's "Intégrale 100 h" offering which, at euro 9 a month less than the eXtense package, is the offering which in terms of unlimited access and price is closest to the highspeed access offerings in Wanadoo's range, had only [...]* subscribers on 31 March 2002 (i.e. [...]* times less than the ADSL offerings) and was growing at a rate more than [...]* less than that of the ADSL offerings (file, pp. 2577 to 2579).

216 In the case of the United Kingdom market, this idea of asymmetric substitutability was put forward by the Office of Telecommunications in its Direction to resolve a dispute between BT, Energis and Thus concerning xDSL interconnection at the ATM switch, referred to above, paragraph E 16.

217 Information taken from [...]* (file, pp. 4132 et seq. and 4240 et seq.).

218 Between September 2001 and March 2003, the price per hour (leaving aside the fidelity option which brought a 20% discount) for the "integral" ten-hour package fell from euro 1.04 to euro 1 (- 4 %). The price for the entry package (three hours at the end of 2001, then five hours at the beginning of 2003 fell from euro 1.93 to euro 1.6 (- 17%). The prices for the twenty hours a month package remained stable. The thirtyhour package at euro 0.75 an hour was replaced by forty-hour and sixty-hour packages at euro 0.5 and euro 0.42 an hour ( a reduction of 34% and 45% respectively compared with the previous thirty-hour package).

219 The phenomenon of differing trends in prices for low-speed and high-speed offerings is not confined to France. Between 2001 and 2002, ADSL prices in Europe fell by an average of 11.4%, whereas during the same period the price for 40 hours of low-speed use fell by more than 33% (Teligen Total Research, "Report on Internet Access Costs via a Standard Telephone Line, ADSL and Cable Modem", July 2002, p. 60). However, throughout Europe, high-speed access offerings continued to grow at a much higher rate than low-speed offerings. In the United States, econometric analyses show that the incremental price for low-speed Internet access is not correlated with the price for high-speed Internet access and that low-speed Internet access does not therefore exercise any constraint on high-speed tariffs (see Hausman, Sidak and Singer, 'Cable Modems and DSL: Broadband Internet Access for Residential Customers').

220 This pattern is not specific to Wanadoo, but is common to all the players on the French market. Between January 2001 and January 2002, high-speed Internet access prices remained stable. During the same period, the flat-rate prices for 20 hours, 40 hours and 50 hours a month fell by 5.3%, 5% and 9.6% respectively (source: ART, "Internet, un bilan du marchè français, mars 2003"). The relative price of flat-rate low-speed offerings closest to unlimited-access formulas thus fell between January 2001 and January 2002. However, during 2001, the rate of growth of the high-speed market, measured in terms of the number of new subscribers per week, continued to accelerate throughout 2001 (+ 74 % between the second half of 2000 and the first quarter of 2001, + 19% in the second quarter of 2001, + 4% in the third quarter of 2001 and + 91% in the last quarter of 2001).

221 The number of customers switching from these services is shown in Annex 16. During the period under review, an average of 418 customers per month switched from the eXtense service (out of an average customer base of 263 000 during the period). In the case of Wanadoo ADSL, the average number of customers migrating was 985 a month (out of a customer base of 120 000).

222 In practice, the vast majority of these changes involved customers switching from Wanadoo ADSL to eXtense.

223 Survey carried out by Novatris in June 2002 for the Commission (file, pp. 5115 to 5174).

224 The sample included 789 residential ADSL or cable high-speed Internet access subscribers.

225 In its reply of 23 October 2002, Wanadoo Interactive challenged the choice of the sample and argued that a sample covering the whole of the French population should be used in order to assess more accurately the behaviour of intended users who had not yet subscribed to one or other formula. The Commission had decided not to use this method, since it considered that, to gain a valid assessment of the propensity of high-speed users to switch to low-speed, it was sufficient to examine a sample of individuals who had already tried out the functionalities and features of high-speed Internet access services. [...]* file, p. 3234

226 See Annex 25 for the detailed calculations.

227 Two thirds of those surveyed stated that they had taken out a subscription for one year or more.

228 Only 6% of those surveyed who had kept their high-speed Internet access subscription where there was a 10% price increase cited their one-year subscription as a motivating factor, and only 2% gave it as their main reason.

229 The results of this survey are corroborated by another survey carried out amongst 955 internet users in August 2002, which showed that 0% of high-speed users would be prepared to revert to low-speed (article published in Le Figaro on 18 October 2002: 'Le boom de l'Internet à haut débit', set out in Annex 2 to Wanadoo Interactive's reply of 23 October 2002, (file, p. 6450).

230 Reply dated 4 March 2002, p. 22, paragraph 72 (file, p.1727).

231 [...]* (p. 3224 of the file). [...]*

232 [...]*

233 The Competition Council has not so far given a ruling on the delineation of the relevant market. But it has stated - and this view has been accepted by the Paris Court of Appeal - that "it cannot be excluded that the France Télécom group holds a dominant position ... in the provision of high-speed Internet access" (decision No 02-MC-03, already referred to).

234 See in particular ART opinion No 02-35 dated 9 January 2002 to the Competition Council on T-Online France's application against France Télécom's practices on the markets for the provision of Internet access by ADSL.

235 See ART opinion No 00-28 dated 7 January 2000 on the Competition Council's request for an opinion on the case brought by 9 Télécom against some of France Télécom's practices on the high-speed Internet access market using xDSL technologies, section 3.1, competition analysis; ART opinion No 02-346 dated 30 April 2002 on tariff decision No 2002033 concerning the IP/ADSL routing offering and on the tariff proposals concerning IP/ADSL offerings announced by France Télécom, in particular section II.2; ART opinion No 01-548 dated 19 June 2001 on tariff decision No 2001482 concerning the development of Netissimo 1 and Netissimo 2 services and tariff decision No 2001480 concerning the development of the IP/ADSL access provision offering and the IP/ADSL routing offering, section II.1.

236 Reply dated 4 March 2002, p. 30 (file, p. 1735).

237 Case C-27-76 United Brands and Others v Commission [1978] ECR 207, paragraph 65.

238 Reply dated 4 March 2002, p. 32, paragraph 110 (file, p.1737).

239 Case 85-76 Hoffmann-La Roche v Commission [1979] ECR 461, paragraph 39.

240 Percentages calculated on the basis of the answers to various requests for information provided by Wanadoo Interactive, Noos, NC Numéricâble, UPC France, Est Vidéocommunications, T-Online France, Tiscali France, AOL France, 9Télécom, Easynet, Nerim, BD Multimédia, NetUltra, Infonie and Oreka.

241 See, for example, the following in-house documents obtained on Wanadoo Interactive's premises: [...]* (file, p. 4236 et seq.); [...]* (file, p. 4128 et seq.); [...]* (file, pp. 4155-4156); [...]* (file, p. 4254 et seq.); [...]* (file, p. 4263); [...]* (file, p. 4278 et seq.); [...]* (file, p. 4350 et seq.).

242 Judgment in Akzo, paragraph 60.

243 Wanadoo Interactive's position in relation to all its other competitors is examined in section II.E.1.

244 See the subscriber bases of Wanadoo Interactive and Noos at the end of 2000 and beginning of 2001 in Annex 25.

245 On 1 January 2001, there were in France 4.9 million cable sockets capable of taking high-speed services, as against more than 11 million lines directly connectable to ADSL. The 16 largest cable operators had geographically fragmented and technically disparate networks, whereas the telephone network which carries ADSL is uniform in technical specifications. The Paris region itself, in which cable operators are strongly established, presents a mosaic of areas served by different operators. Lastly, cable operators are prohibited by law from serving more than 8 million inhabitants, which imposes a severe constraint on the extension of their networks. The weaknesses of the cable networks compared with ADSL in the deployment of high-speed Internet access are analysed in a number of decisions and public documents issued by the ART (see, in particular, its opinion No 00-28 of 7 January 2000, and the report on the situation of the cable operators which can be consulted at its website: http://www.arttelecom. fr, clicking on "les grands dossiers", "réseaux câblés").

246 Percentages calculated on the basis of the raw figures provided by the service providers' replies referred to in footnote 240.

247 The relative improvement in the main ADSL competitor's situation in the first quarter of 2002 is mainly due to the slowdown in Wanadoo Interactive's sales as a result of the ban on marketing eXtense packages in France Télécom agencies imposed by the Competition Council in the decision of 27 February 2002 already referred to.

248 Hoffmann-La Roche, paragraph 48.

249 Reply of 4 March 2002, paragraph 154 (file, p. 1749).

250 Ibid., paragraph 121.

251 However, it should be noted that Wanadoo Interactive bases its assessment (reply of 4 March 2002, footnote 80 - file, p. 1739) on a Commission non-opposition merger Decision adopted in May 2000, in which the xDSL equipment market was described as an emerging market (Commission Decision of 19 May 2000 in Case IV/M.1908 - Alcatel/Newbridge Networks (OJ C 169, 17.6.2000, p. 7), at paragraph 12). That Decision was adopted more than seven months before the beginning of the period covered by this Decision. Furthermore, it applies only to the segment using xDSL technology, which developed later than cable modem technology.

252 BDRC, "The Development of Broadband Access Platforms in Europe", August 2001, pp. 83 and 126. It is known that Wanadoo Interactive was following the situation abroad during the initial stage of development of ADSL services. The [...]* (file, p. 3146) includes the following paragraph: "In the United States (source [...]* study, January 2000) demand for high-speed is soaring in the towns in which ADSL is deployed (one third of demand). This demand is growing more rapidly on the residential market and is driven by customers' need for speed, value-added content and convenience of use".

253 [...]* (file, p. 4254).

254 pp. 7 and 13 of Annex 31 to Wanadoo Interactive's reply of 4 March 2002 to the statement of objections (file, pp. 2076 and 2082).

255 [...]* (file, p. 2907).

256 [...]* (file, p. 2907), which indicates a [...]* % market share for Wanadoo in ADSL by 2004.

257 Court of First Instance of the European Communities in Joined Cases T-24-93, T-25-93, T-26-93 and T- 28-93 Compagnie maritime belge v Commission [1996] ECR II-1201, paragraph 77.

258 Reply of 4 March 2002, paragraph 118 (file, p. 1740).

259 Reply of 4 March 2002, paragraphs 123 and 149 (file, p. 1748).

260 Case 322-81 Nederlandsche Banden Industrie Michelin v Commission [1983] ECR 3461, paragraph 59.

261 Up to the end of 2002, AOL held less than 3% of the high-speed Internet access market in France.

262 Reply of 4 March 2002, paragraph 114 (file, p. 1738).

263 Ibid., footnote 76.

264 Study on Internet access carried out by the Benchmark Group among 6 000 Internet users, reported in Le Journal du Net of 15 November 2001.

265 Wanadoo's stock exchange listing prospectus, July 2000, p. 52.

266 Up to October 2000, the Turbo IP facility was indeed not only offered to, but imposed on anyone wishing to subscribe to one of the IP/ADSL 1 or IP/ADSL 2 offerings.

267 [...]* file, p. 255 et seq.).

268 At the end of January 2001, Wanadoo Interactive's total reserved bandwidth for Netissimo 1 was about [...]* Mbit/s for Paris and Île-de-France, as against [...]* Mbit/s for the rest of France (Annex to France Télécom's reply of 16 May 2001, file, p. 113 et seq.).

269 Following the approval of this tariff decision by the Minister for Economic Affairs, Finance and Industry.

270 [...]* (file, p. 2909), [...]* (file, p. 3317).

271 [...]*

272 United Brands, paragraphs 70 and 71.

273 Abovementioned decision of the Competition Council adopted on 27 February 2002.

274 Wanadoo Interactive's products are also marketed in [...]* general or specialised retail outlets and through a network of small independent retailers comprising [...]* outlets (France Télécom's letter of 22 November 2001, file, pp. 689 et seq.).

275 It will be borne in mind that, for the purposes of high-speed Internet access using ADSL, residential consumers can choose between acquiring a package (combining the provision of the high-speed line with the provision of Internet access proper), and entering into a separate contract with France Télécom for the provision of the Netissimo 1 service (ADSL access line) and with the Internet service provider of their choice (Wanadoo Interactive or another ISP) for Internet access proper.

276 Email sent by Noos to the Commission on 7 October 2002, file, p. 5338. At that time, Club Internet did not have any shops or retail network of its own (letter from T-Online France dated 28 February 2002, file, p. 1359).

277 France Télécom's letter to the Commission of 22 November 2001, file, p. 706.

278 Email sent by Noos to the Commission on 30 August 2002, file, p. 5240.

279 At the beginning of 2002, the products of T-Online (Club Internet), which was at that time the second largest competitor on the market in numerical terms, were stocked by a network of fewer than 500 general or specialised retailers (T-Online's letter to the Commission of 28 February 2002, file, p. 1359), i.e. four times fewer than Wanadoo Interactive.

280 [...]* (file, p. 4180).

281 [...]* (file, p. 3211); [...]* (file, p. 3859). [...]*

282 [...]* (file, p. 3853). [...]* (file, p. 4329). [...]* (file, pp. 2434 et seq.).

283 [...]* (file, p. 3954).

284 [...]* (file, p. 3104).

285 It should be noted here that, even amongst the customers who finally opted for a "dissociated" offering, from January to September 2001, the period during which the Netissimo service was almost exclusively marketed through agencies, more than [...]* opted for the Wanadoo ADSL service, the rest being divided amongst all the other providers. While this does not constitute hard proof, it seems to indicate that consumers applying to a France Télécom agency were directed towards Wanadoo's service.

286 Source Sécodip, cited by one of Wanadoo Interactive's competitors in its reply to the Commission's request for information of March 2002.

287 Michelin, paragraphs 55, 56 and 58.

288 [...]* (file, p. 3637).

289 [...]* (file, p. 251 et seq.), [...]* (file, p. 633) . [...]* (file, pp. 4727 et seq.) [...]* (file, pp. 4765 et seq.), [...]* (file, p. 5904), [...]*

290 See France Télécom's reply of 1 August 2001 to the information request of 10 July 2001. In that reply (Annex 2a), it is clear that [...]*. France Télécom's replies of 14 March 2002 and 4 November 2002 show that an equivalent phenomenon occurred in 2002.

291 Reply of 4 March 2002, paragraph 142.

292 Decision No 98-D-16 of 3 March 1998 concerning practices implemented by the Office d'Annonces in the marketing of advertising space in telephone directories.

293 Annex to Wanadoo's consolidated accounts as at 31 December 2001, p. 14 (document transmitted by Wanadoo Interactive on 15 March 2002, file, p. 2369).

294 Point 3 of appendix C to Wanadoo Interactive's reply of 23 October 2002 (file, p. 6366 et seq.).

295 Point 2 of appendix C to the reply of 23 October 2002.

296 In its abovementioned publication "Internet, un bilan du marché français/mars 2003", the ART describes the phenomenon as follows: "as from mid-2000, particularly because advertising revenue had not reached forecast levels, investors reduced the amounts invested in Internet companies. The raising of funds and the introduction of companies onto the stock exchange became more difficult. This drying-up of funding led to a change in economic models ... Internet companies could not continue pursuing a strategy of acquiring customers at any price and were forced to come up with a viable economic model".

297 United Brands, paragraph 66.

298 Akzo, paragraphs 71 and 72; Case C-333-94 P Tetra Pak v Commission [1996] ECR I-5951, paragraph 41.

299 It should be pointed out, in addition, that over the same period Wanadoo Interactive likewise did not cover the unadjusted variable costs as entered in the accounts.

300 The apparent "weight" of customer acquisition costs in the analysis is, depending on the period, reduced in the calculation of the recovery rates in section I.F to between [0-10]* % for the eXtense service and between [0-10]* % for the Wanadoo ADSL service, i.e. some [10-30]* points or more below their share in the actual costs posted.

301 Paragraphs 205 to 215 of appendix C to Wanadoo Interactive's reply of 23 October 2002 (file, p. 6293 et seq.).

302 David E.M. Sappington and J. Gregory Sidak, 'Are Public Enterprises the Only Credible Predators?', The University of Chicago Law Review, 271, 2000, pp. 274 and 275.

303 P. Bolton, J. Brodley and M. Riordan, 2000, 'Predatory Pricing: strategic theory and legal policy', 88, Georgetown Law Journal, pp. 3 and 36: "The disciplining of rivals is itself exclusionary since its object is to exclude the growth and expansion of the prey or the prey's entry into new markets."

304 Ordover et al., op. cit., p. 9.

305 With the Wanadoo ADSL formula, the cost at the end of 2000 came to FRF 378 taxes included, not counting the cost of installation and commissioning (FRF 768.57 taxes included). The cost of the eXtense packaged formula came for its part to FRF 298 taxes included, or 21% less than the former.

306 To back up its assertion, Wanadoo Interactive refers to two documents, one dated July 2000 and the other May 2001 ([...]* (file, p. 3142), and [...]* (file, p. 3106), cited in paragraph 89 of Wanadoo Interactive's reply of 23 October 2002.) However, these two documents are not really relevant to the analysis of this case. The first document dates from summer 2000, when the company preferred to price the subscription at FRF [...]* taxes included; it therefore predates by several months Wanadoo Interactive's decision to set the retail price of the subscription to the eXtense package at FRF 298 taxes included. The second document sets out the personal views of two Wanadoo Interactive employees on billing the subscriber for the cost of accessing the service (and not on the subscription price). These views were, however, not acted upon [...]* Thus, while thoughts of improving profitability may have been uppermost in the minds of some executives, they had no impact on the company's actual pricing choices.

307 [...]*

308 [...]* (file, p. 3109)

309 [...]* (File, p. 3234).

310 See Commission Decision No 85-609-EEC of 14 December 1985 in Case IV/30.698 - ECS/Akzo (OJ L 374, 31.12.1985, p. 1), at paragraph 81.

311 Court of First Instance in Case T-83-91 Tetra Pak v Commission [1994] ECR II-755, paragraph 151.

312 See section I.G.2 tracing the history of the internal company thinking on the price and margin levels for ADSL services during 2000 and 2001.

313 [...]* (file, p. 3316). [...]*

314 See section I.G.2.

315 [...]* Wanadoo Interactive mentions a target of [...]* (file, p. 2907)). Mention is subsequently made of a penetration target of [...]* (file, p. 3074).

316 [...]* (file, p. 2293 et seq.) [...]*

317 ART opinion No 02-346 of 30 April 2002 and the press kit made available on that occasion. According to the Authority's estimates at that time, "based on an average speed of 23 kbit/s per subscriber, as per current operating conditions", Internet service providers "currently pay France Télécom a monthly fee of euro 39.70 (before tax) per subscriber for IP/ADSL offerings (routing + access), an amount greater than their revenue", which is put at euro 38 (before tax) on the basis of the price charged by Wanadoo Interactive (file, 4956). The ART's opinion states just as explicitly that ISPs had had to pay France Télécom a fee of euro 39.70 per subscriber per month, forcing them to incur losses in order to gain a foothold on the market and compete with Wanadoo.

318 See Commission Decision in the Akzo case, paragraph 79.

319 [...]* (File, p. 3213).

320 [...]* (file, p. 3158).

321 [...]*[...]* The enclosure contains a statement by Mr Treppoz, CEO of AOL France: "In the days when we were owned by Cégétel, we launched an offering with Monaco Telecom and had 500 subscribers. We did not launch it in France as France Télécom's ADSL retail offering is currently not a money-maker. Technically, we are ready, but we are not in the business of losing money" (file, p. 3065).

322 The document [...]* analyses in detail, in the light of the speed and size of connections, the margin on network costs hypotheses resulting from the entry into force of the new contract for national and regional IP routing of ADSL traffic (file, p. 2921 et seq.). It is clear from this document that a competitor with less traffic than Wanadoo Interactive will enjoy margins on network costs several points lower than those of Wanadoo Interactive. Thus, a very small competitor offering connection speeds of only [...]* Mbits/s earned a negative margin on network costs [...]* points below that of Wanadoo; a medium-sized competitor which had a larger subscriber base and offered connection speeds of [...]* Mbits/s was still [...]* margin points below Wanadoo; and a bigger competitor offering connection speeds of [...]* Mbits/s managed to reduce the gap between it and Wanadoo to [...]* margin points.

323 In this connection, it can be estimated that the cost of externally financing the deficit incurred by Wanadoo Interactive in 2001 as a result of its ADSL business (losses in the region of euro [...]*) would have entailed financing costs of about euro [...]* per subscriber per month.

324 For Wanadoo Interactive, the adjusted full cost of the eXtense service was during the last few months of 2001 and the first half of 2002 approximately [...]* % above the adjusted variable cost. It can be considered, moreover, on the basis of the network cost differentials that an average competitor is invariably [...]* profitability points below Wanadoo Interactive. Thus, while during that period Wanadoo recovered [...]* % of its full costs (or [...]* % of its variable costs), competitors invariably achieved a rate of recovery of their variable costs necessarily lower than [...]* %.

325 See paragraph 139.

326 [...]* (file, pp. 3216, 3217 and 3233).

327 [...]* (file, pp. 3534 et seq.)

328 The commercial objectives and achievements of Wanadoo Interactive are occasionally communicated to the press. In an article in Les Echos dated 7 September 2001, Wanadoo's CEO, Nicolas Dufourcq, is quoted as mentioning a target of 300 000 ADSL subscribers by the end of the year and the acquisition of 1 000 new ADSL subscribers a day. An article in Le Journal du Net dated 4 October 2001 referred to the 210 000 ADSL subscribers then registered, 10 000-a-week growth in the number of subscriptions and a target of "300 000 by the end of the year". An article in Les Echos dated 21 February 2002 gave a figure of 457 000 ADSL subscribers and mentioned a target of one million subscribers by the end of 2002. In an article in the same newspaper dated 23 January 2002, mention was made of the recruitment of 140 000 subscribers between October and December of the previous year.

329 See paragraphs 117, 118 and 133-137 of the reply of 23 October 2002 (file, p. 6276 et seq.)

330 Court of First Instance in Tetra Pak, paragraphs 151 and 190.

331 Option 1 refers to full unbundled access to the copper pair, option 2 refers to shared access to the copper pair, and option 3 covers an offer of indirect access to the copper pair via the incumbent operator's ATM network.

332 See paragraph III.1.3 of ART opinion No 02-346, referred to above. See also the press release and transparencies published on the ART's website on 7 May 2002. In its reply of 4 March 2003 (paragraph 103 - file, p. 6883), Wanadoo Interactive states, moreover, that [...]*

333 Taken from the forecasts drawn up by France Télécom for its wholesale IP/ADSL and IP routing services in 2001, paragraph 3 of France Télécom's letter to the Commission dated 9 April 2002 (file, p. 2573).

334 [...]*

335 The table was communicated to it under cover of the Commission's letter dated 16 January 2003.

336 Paragraphs 104-107 of Wanadoo Interactive's letter dated 4 March 2003 (file, p. 6884).

337 Pursuant to Article L.36-7, paragraph 5, of the French Posts and Telecommunications Code, the ART "shall issue a public opinion on (...) tariffs for services for which there are no competitors on the market, prior to their approval, where appropriate, by the Minister for Telecommunications and the Minister for the Economy." It is thus clear from the French legislation that the ART has no competence to deliver an opinion on Wanadoo Interactive's retail tariffs. With respect to France Télécom's wholesale offerings, in its opinion of 19 June 2001 on the IP/ADSL and routing offerings the ART considered, not that the incumbent's tariff proposal was satisfactory on the whole, but only that it should help to "significantly improve the situation on the retail market", about which it had expressed concern in the same opinion. In its opinion of 7 May 2002, the ART did not impose on France Télécom any offering price levels, but merely indicated the minimum conditions which new offerings should meet.

338 In its decision No 02-D-38 of 19 June 2002 concerning a referral and an application for interim measures from Liberty Surf, the Competition Council, despite not having in its possession the same evidence as the Commission, talks of a "strategy common to" both France Télécom and Wanadoo and of a "community of interests", and states that the degree of autonomy of Wanadoo Interactive vis-à-vis France Télécom affects "the ascribability of the practices but not their materiality".

339 File, p. 2957.

340 Annex 9 to Appendix D to Wanadoo Interactive's reply of 23 October 2002 (file, p. 6379).

341 See paragraphs 327 and 336 of Wanadoo Interactive's reply of 23 October 2002 (file, p. 6315).

342 In 2000, for all Wanadoo subscribers, all access services combined, the annual average churn rate came to [...]*% [...]*file, p. 3438)). In 2001 and 2002, the annual average churn rate for eXtense was of the order of [...]*% (paragraph 31 of the reply of 4 March 2003 (file, p. 6867)); the eXtense service thus had a churn rate more than [...]*times lower than that which low-speed products had in 2000. In January 2002, the monthly churn rate was [...]*% for dial-up low-speed offerings, [...]*% for "integral" offerings and [...]*% for broadband; in February 2002, the monthly churn rates were [...]*%, [...]*% and [...]*% respectively [...]*file, p. 4134)). The churn rate for low-speed offerings is thus [...]*times higher than that for high-speed offerings.

343 [...]* (file, p. 3412).

344 The economic literature acknowledges that initial losses are recouped not only on the relevant market but also on closely related markets. M. Canoy, P. de Bijl, Ron Kemp, "Access to Telecommunications Networks", mimeo, CPB, October 2002, p. 30.

345 [...]* (file, p. 3223).

346 [...]* (file, p. 2655).

347 [...]* (file, p. 3409).

348 Ibid.

349 A [...]*study [...]* states (p. 10) that "[...]*have already made purchases on line." (file, p. 4145).

350 [...]* (file, p. 4249).

351 [...]* (file, p. 3437 et seq.).

352 It is not without reason, for example, that in October 2002 Microsoft chose as vehicles for its new interactive gaming platform, Xbox Live, the service providers Wanadoo Interactive and Noos, which at that time ranked first and second on the French market, and not other service providers such as TOnline France or Tiscali (Reuters, "Microsoft to pilot Xbox Live in Europe in Nov", dispatch dated 25 September 2002, 4.21 hours).

353 File, p. 3415.

354 [...]* (file, p. 3409).

355 [...]* (file, p. 3415).

356 The abovementioned [...]* study states that, in the residential market, [...]*% of new subscribers to the eXtense product learned about the Wanadoo offering by word of mouth, twice as many as did so through the press or online advertising. The same document indicates that [...]*% of subscribers to the eXtense package would recommend the product to their friends and acquaintances, compared with [...]*% for all Wanadoo Interactive subscribers, all services combined.

357 [...]* (file, p. 3086).

358 [...]* (file, p. 4157).

359 [...]* (file, pp. 2958 and 2959).

360 See paragraphs 188 and 191 of the reply of 23 October 2002 in which Wanadoo Interactive refers to the necessity of "a mature market on which the operator complained of has been in a dominant position for a significant length of time" (file, p. 6291).

361 See paragraphs 193 and 196 of Wanadoo Interactive's reply of 4 March 2002 and paragraph 348 of its reply of 23 October 2002 (file, p. 6319).

362 Paragraph 57 of the reply of 23 October 2002 (file, p. 6262).

363 Paragraphs 84-87 of the reply of 23 October 2002 (file, p. 6270).

364 In accordance with the objective set out in Article 3(g) of the Treaty.

365 Court of Justice in Tetra Pak, paragraph 44.

366 See [...]*

367 Paragraphs 227 and 228 of the reply of 4 March 2002 (file, p. 1766).

368 Reply of 4 March 2002, paragraphs 161 and 162 (file, p. 1752).

369 See, for example, Bolton et al., op. cit., p. 51.

370 [...]* (file, p. 274) [...]*

371 Assuming a bandwidth consumption of the order of 15.8 kbit/s per subscriber.

372 Making the same assumption as in footnote 371.

373 P. Areeda and D. Turner, 'Predatory Pricing and Related Practices under Section 2 of the Sherman Act', Harvard Law Review, vol. 88, p. 714.

374 The ART stated, moreover, in a document dated April 2002 entitled "l'accès à Internet, premier bilan", p. 10, that "if alternative ISPs really had the means to compete with Wanadoo's offerings, then the rate of penetration [of high-speed access] might well be much higher."

375 Submission by Wanadoo Interactive at the hearing on 18 March 2002.

376 Point 5 of Alcatel's letter to the Commission dated 12 June 2002 (file, pp. 5026 and 5027)

377 Statement by Easynet's representative at the hearing on 18 March 2002.

378 Point 6 of Alcatel's abovementioned letter.

379 Reply of 4 March 2002, paragraphs 185, 186, 199 to 202, section 3.4.6.2.

380 Areeda et al., op. cit., pp. 715 and 716.

381 Judgment in Akzo, paragraph 135.

382 See Court of First Instance in Case T-65-89 BPB Industries and British Gypsum v Commission [1993] ECR II-389, paragraph 117; United Brands, paragraph 189; Compagnie maritime belge, paragraphs 146 and 147.

383 Court of Justice in Michelin, paragraph 57; Court of First Instance in Case T-228-97 Irish Sugar v Commission [1999] ECR II-2969, paragraph 112.

384 Wanadoo Interactive points out in its reply of 4 March 2003 (paragraph 37 - file, p. 6868) that Noos had been granted an exclusive concession in Paris. The exclusive nature of this concession is immaterial here since the concession related to the cable TV network infrastructure and not to the provision of high-speed services. Noos therefore enjoyed no territorial exclusivity with regard to broadband, the telephone network being at least as extensive as Noos's cable network.

385 A detailed comparison between Noos's subscriber bases at the end of 1999 and during the first quarter of 2001 is to be found in Annex 23 to this Decision. 386 See Annex 23 to this Decision.

387 Ibid.

388 Historically, Noos's offering, then known as Cybercâble, was launched in 1998. During the second half of 1999 and for the first two months of 2000, Noos's offering was suspended for technical reasons. It was resumed in March 2000 with a dual pricing structure depending on whether or not the Internet subscriber was also a cable TV subscriber: in the former event, the price was FRF 299 taxes included, and in the latter, FRF 389 taxes included. In May 2000, a unified pricing system was introduced, including a new price.

389 This amount can be broken down into FRF 299 (or euro 45.58), taxes included, for Internet access and FRF 79 (or euro 12.04), taxes included, for modem rental.

390 This up-front payment consisted of a FRF 700 installation charge and a FRF 500 modem deposit. The deposit is repayable at the end of the contract period, but the FRF 700 installation charge is not. By contrast, the purchaser of the eXtense package is not out of pocket as Wanadoo Interactive does not charge for installation and the customer owns the modem, which constitutes a saleable asset.

391 Noos provided the modem to the customer free of charge for one year from the date on which he signed up.

392 Customers recruited before 28 August 2000 paid the old subscription fee of FRF 378 taxes included.

393 This upload limit billing lasted from February to November 2001.

394 From March 2001, large customers were offered a special version of the Noosnet offering known as "Noosnet forfait 1 Go" for FRF 499 plus a FRF 79 modem rental fee. [...]* (file, p. 3266)).

395 Article in Le Journal du Net, 23.11.2000, entitled "Pack ADSL: Club Internet tire le premier".

396 Footnote 86 to the reply of 23 October 2002 (file, p. 6268).

397 Paragraph 77 et seq. of the reply of 23 October 2002 (file, p. 6268).

398 Article in Le Journal du Net, 6.10.2000, entitled "Yahoo France se lance dans l'ADSL avec Mangoosta".

399 To subscribers recruited between 6 and 23 October Mangoosta offered the first month's subscription plus the installation costs.

400 The calculation consists in finding the value for the variable X in the following equation: 500 + 330X = 995 + 298X.

401 In addition, by acquiring the starter pack, the Wanadoo subscriber became owner of a saleable good, unlike the Mangoosta subscriber, to whom the up-front payment represented a net loss.

402 See paragraphs 61 and 62, 71 to 75 and 166 of the reply of 23 October 2002 (file, pp. 6203, 6267 and 6287).

403 Cf., for example, paragraphs 253 to 300 of the reply of 4 March 2002; paragraphs 197 to 200 and pp. 14 to 18 of Annex 1 to the reply of 23 October 2002.

404 Court of Justice in Tetra Pak, paragraph 44.

405 Commission Decision 2001-354-EC of 20 March 2001 in Case COMP/34.141 - Deutsche Post AG (OJ L 125, 5.5.2001, p. 27).

406 Some economists take the view, for example, that in the case of a public enterprise a strategy of predation without recoupment of initial losses is perfectly conceivable, as the need for ultimate recoupment of the losses is less strong owing to the ultimate guarantee of viability provided by the state shareholder (see in this connection David E.M. Sappington and J. Gregory Sidak, Competition Law for State-Owned Enterprises, December 2002). In the present case, Wanadoo Interactive, which is owned more than 70% by France Télécom, which is itself state-owned, is a public enterprise. Loss recoupment is even less likely to be an objective where the public enterprise has in its portfolio of activities other, highly profitable products reducing the need to offset the losses induced by the predation, as is the case here.

407 P. Bolton, J. Brodley and M. Riordan, "Predatory Pricing: response to critique and further elaboration", The Georgetown Law Journal, vol. 89, 2001, pp. 2512 and 2513.

408 It is also possible to envisage a predation strategy the purpose of which is not to maximise revenue and operating profit through supracompetitive prices or margins, but simply to increase the company's value by enlarging its customer base and hence the potential future value of its goodwill. This increase in value has financial implications for the company's shareholders when the business is sold or whenever its market value increases (e.g. as a result of an exchange of shares).

409 See Baumol, 'Contestable Markets: an uprising in the theory of industry structure', American Economic Review, vol. 72, p. 1 et seq. (March 1982).

410 Because of the potential risks of an over-restrictive approach to entry barriers, some economists prefer the terms "obstacle to entry" or "conditions of entry" (See Joskow et al., op. cit., p. 227).

411 Ordover et al., op. cit., p. 11.

412 See Scherer and Ross, Industrial Market Structure and Economic Performance, pp. 360 and 361.

413 The cases of the competing ADSL service providers Free and Oreka are significant in this regard. From August 2002, Oreka's offering was priced at euro 39.95 taxes included, or almost 12% less than what most providers of comparable products were charging. The price of Free's offering launched in September 2002 came to euro 29.9 taxes included, or approximately 34% below the prices of its competitors. Oreka, whose pricing policy Wanadoo Interactive describes as "aggressive" (paragraph 105 of the reply of 23 October 2002 (file, p. 6274)), gained in all only 1 500 ADSL subscribers during the four months following the launch of its offering (letter from Oreka to the Commission dated 2 January 2003 (file, p. 6112)). Free, for its part, recruited more than 100 000 customers during the same period, but only a small proportion of them came from competing ADSL providers and hence, a fortiori, from Wanadoo Interactive. The number of Free subscribers originally with competitors has been divulged to the Commission but is a business secret (file, p. 5386); it is less than 10% (file, p. 6838). Between September and December 2002, fewer than [...]* subscribers to Wanadoo's ADSL or eXtense services, or less than [...]*% of Wanadoo Interactive's average subscriber base during that period, migrated to Oreka and Free. It should be pointed out in this connection that, at that time, more than [...]* Wanadoo subscribers, be they customers of the Wanadoo ADSL service or customers of the eXtense offering for more than a year, were free to migrate to a different service provider. Therefore fewer than [...]*% of Wanadoo Interactive's subscribers who were free to migrate left this service provider for a cheaper competitor.

414 See, for example, Paul Joskow and Alvin Klerovick, 'A Framework for Analyzing Predatory Pricing Policy', Yale Law Journal, vol. 89, Nb 2, December 1979, pp. 228 and 229.

415 Elie Appelbaum and Chin Lim, 'Contestable Markets under Uncertainty', Rand Journal of Economics, vol. 16, No. 1, spring 1985, p. 28.

416 [...]* (file, p. 4249).

417 [...]* (file, p. 4068).

418 AOL France has not published its 2001 turnover.

419 Turnover of Internet business only, excluding telecoms products.

420 T-Online France has not published its 2001 turnover.

421 In its reply of 4 March 2002 and at the hearing on 18 March 2002, Wanadoo Interactive claimed that it was easy even for a wholly new entrant to penetrate the market. It gave the example of Dixinet (replies dated 4 March 2002, paragraph 299, and 23 October 2002, enclosure 1, p. 16), which had been offering an ADSL service since the beginning of 2002. However, apart from the fact that this service was priced at euro 46 taxes included, i.e. more than that of Wanadoo Interactive, Dixinet was in reality merely experimenting with a subscriber base which as yet had no statistical significance on the French market at the end of 2002. At the end of August 2002, Dixinet had only 10 subscribers to its ADSL and telephone services (letter from BD Multimédia to the Commission dated 26 September 2002 (file, p. 5327)). The case of Net pratique, which was also highlighted by Wanadoo Interactive, is equally revealing. The company launched its service in summer 2002 and at the end of 2002 it had only 1 400 or so subscribers (letter from Net Ultra to the Commission dated 12 December 2002 (file, p. 5393)).

422 File, p. 3414.

423 Scherer and Ross, Industrial Market Structure and Economic Performance, pp. 436-438.

424 Cf., for example, William S. Comanor and Thomas A. Wilson, 'Advertising Market Structure and Performance', Review of Economics and Statistics, volume XLIX, November 1967.

425 William S. Comanor and Thomas A. Wilson, 'The Effect of Advertising on Competition: a survey', Journal of Economic Literature, vol. XVII (June 1979), especially at pp. 453 and 454.

426 See Annex 22.

427 Only Wanadoo Interactive's ADSL and eXtense services offering download speeds of 512 kbps are referred to here, to the exclusion of its other ADSL products.

428 [...]* (file, p. 4068).

429 The network effects linked to a broad subscriber base play a part in the recruitment of new subscribers through "word of mouth".

430 As explained in section II.B.3.1, in order to qualify for the lowest IP routing charges, it was necessary before October 2002 to build up a subscriber base in excess of 95 000.

431 The economic model of Free's offering cannot be reproduced by just any operator, as will be shown in paragraphs 361 and 362.

432 An analyst commented on the situation at the end of 2002 as follows: "In France, Tiscali no longer has the financial resources to cut its prices aggressively and to launch vast marketing campaigns" (article in Les Echos dated 9 September 2002 entitled "Wanadoo et T-Online se ruinent à l'extérieur").

433 In its Barilla-BPL-Kamps Decision, already referred to, at paragraph 21, the Commission had occasion to acknowledge that advertising may constitute a significant barrier to entry, as Wanadoo Interactive admits in its reply of 23 October 2002, paragraph 201 (file, p. 6293). Likewise, in its Decision in the Procter and Gamble/VO Schickedanz case, paragraph 145, the Commission described a process similar to that observed in this case as follows: "In terms of barriers to entry, the marked increase in the level of on-going promotional support has significantly raised the level of expenditure necessary for a new entrant wishing to stand out above the 'background noise' of this on-going brand support, so that the absolute amount of money required to overcome the promotional barrier to entry has been significantly increased. Furthermore, the effect of this increase in the underlying level of advertising expenditure is also to increase the minimum viable market share on entry since the potential entrant knows that in order to fund competitive levels of advertising, it must achieve sufficient sales ... Retaliation by incumbents with significant market shares against new entrants is thus cheap and easy, and until the new entrant achieves a similar market share, incumbents have a cost advantage on one of the key competitive parameters."

434 Online article dated 13 June 2002 (http://qualitysteam.fr/actualités/juin02/13-06-2002-1.html).

435 Article in Les Echos entitled "Free propose un accès ADSL à 30 euros", 23 September 2002.

436 Article in La Tribune entitled "Le fournisseur d'accès Free tire la croissance d'Iliad", 5 March 2003.

437 Reply of 4 March 2003, paragraph 103 (file, p. 6883).

438 At the end of 2002, LDCom offered more than [...]* times fewer ADSL lines than France Télécom.

439 As at 30 June 2002, there were estimated to be in France between 500 and 750 unbundled lines out of a total of more than 33 million telephone lines (See European Commission, "Eighth Report on the Implementation of the Telecommunications Regulatory Package", Annex 3, p. 72). As at 1 February 2003, there were approximately 10 400 unbundled lines in France (http://www.art-telecom.fr, heading entitled "actualité du dégroupage"). Even assuming that all of these lines carry ADSL, this figure is negligible (less than 1%) compared with the total ADSL subscriber base at that time.

440 Paragraphs 327 and 338 of Wanadoo Interactive's reply of 23 October 2002 (file, pp. 6312 and 6313).

441 This figure is little more than a guesstimate of the total margin generated during the year as there should be added to it that produced by customers recruited between now and the end of the financial year.

442 It will be recalled that in 2002 Wanadoo Interactive spent some euro [...]* on advertising, marketing and promotional activities related to its ADSL services.

443 The rate of growth of sales rose to more than [...]* new subscribers per week in September and to more than [...]* in October and November 2001 (see Annex 3).

444 In the document drawn up by Wanadoo Interactive [...]* (file, p. 4005).

445 [...]* (file, p. 3073).

446 Reply dated 4 March 2003, paragraph 102 (file, p. 6883).

447 Article in Le Journal du Net, 13.4.2000, entitled "Liberty Surf s'essaye à l'ADSL".

448 Statement made by the representative of Easynet France at the hearing held on 18 March 2002.

449 Article in Le Journal du Net, 5.10.1999, entitled "Nerim.net mise tout sur l'ADSL".

450 Article in Le Journal du Net, 22.10.1999, entitled "Une offre ADSL à 440 F tout compris pour Club Internet".

451 In a letter to the Commission dated 26 October 2001, Easynet states: "Before the IP/ADSL charges ... and the IP/ADSL routing fees and technical conditions changed ... it was totally impossible for us to offer an ADSL package at the market price (reference: Wanadoo's eXtense package at FRF 298 taxes included) under viable economic conditions. We sold only the IP service at FRF 120 per month taxes included. The new price conditions ... enabled us in late September to offer the ADSL Easyconnect package at near to market conditions (although for more than Wanadoo's package)" [emphasis added]. The Easyconnect subscription fee was FRF 327.98 per month taxes included (i.e. 10% more than the eXtense subscription), the starter pack being sold for FRF 1 402.54 taxes included, "much more than the eXtense starter pack, selling at FRF 999 taxes included since January 2001". In another letter dated 5 November 2001, Easynet states that it had to charge FRF 327.98 per month in order to break even. Furthermore, in the tables of figures transmitted to the Commission on 7 September 2001 (the precise content of which is a business secret), Liberty Surf tried to demonstrate that an Internet service provider could not break even when charging the same retail price as Wanadoo Interactive, even after the entry into force of the summer 2001 price decisions. The Commission has received other evidence predating the pricing decisions taken by France Télécom in July 2001 which stress the still more unfavourable conditions prevailing at that time.

452 [...]* (file, pp. 4165 and 4167).

453 Ibid.

454 Price is an important criterion in the choice made by residential customers. According to a survey [...]*.

455 According to the [...]* survey [...]* and cited in footnote 454, [...]*.

456 Service providers are able to gear their advertising and promotion efforts to their quantitative targets in terms of the recruitment of new customers.

457 For example, by concentrating efforts on the Île-de-France region and making only limited advertising investments.

458 The need for financial security and minimum profitability is a key concern for Internet service providers that do not have the backing of a group capable of absorbing losses. In a letter to the Commission dated 29 June 2001, an independent ADSL service provider thus stressed that the low retail prices imposed by Wanadoo Interactive caused "absolutely enormous ... harm ... in terms of the ability to mobilise investors in this business".

459 Charging a monthly subscription fee of FRF 330 taxes included for a bundled offering combining ADSL access and the Yahoo service.

460 Mangoosta's monthly subscription fee was raised from FRF 330 to FRF 390 taxes included on 19 March 2001.

461 Bolton et al., "Predatory Pricing: strategic theory and legal policy".

462 It should also be pointed out that, according to the survey [...]*, [...]*% of eXtense customers were former subscribers to low-speed Internet services operated by other providers (as against [...]*% who were former subscribers to Wanadoo's low-speed services). It thus appears that the eXtense package also affects the market in low-speed Internet access and can shift the balances on that market.

463 Judgment in Compagnie maritime belge, paragraph 149.

464 The new IP/ADSL access charges are [...]* % lower than those applied previously.

465 The new routing charges have since 15 October 2002 been on average 45% lower than before. They also decrease much less with traffic volume, so that small providers now suffer only a small cost handicap.

466 See section I.H.1.

467 France Télécom nevertheless brought an appeal before the French supreme administrative court in June 2002 (application No 247.866) against all aspects of the ART's decision of 16 April 2002 relating to the charges for partial and total unbundling and submitted its complete certified statement to the Council of State on 16 December 2002. The current charges for local loop unbundling are therefore pending on the decision of the administrative court, which creates a climate of uncertainty.

468 In its reply dated 23 October 2002, paragraph 16, Wanadoo Interactive claimed that T-Online France's penetration target was to gain a [...]*% share of the high-speed market in 2003. But the press article from which Wanadoo Interactive selectively quoted (file, p. 6246) merely stated that T-Online France expected that by the end of 2003 high-speed Internet access would account for 35% of Internet access as a whole, without making any reference to its own penetration targets. The quotation used by Wanadoo Interactive is therefore invalid.

469 If the growth trends (both overall and for each provider) observed over the period between September 2002 and March 2003 were to continue until the end of 2004, and ruling out the eventuality of any amalgamation of competitors, Wanadoo Interactive would then have a market share of around [40- 50]*%, more than [...]* that of its leading competitor, with none of the other players having a market share of more than 9%.

470 United Brands, paragraph 201; Case 22-79 Greenwich Film Production v SACEM [1979] ECR 3275, paragraph 11.

471 See Court of Justice in Case 56-65 Société Technique Minière v Maschinenbau Ulm [1966] ECR 281, in particular at p. 303; Joined Cases C-215-96 and C-216-96 Carlo Bagnasco and Others [1999] ECR I- 135, paragraph 47; Case C-35-99 Arduino [2002] ECR I-1529, paragraph 33; Case C-309-99 Wouters [2002] ECR I-1577, paragraph 95 with other references. See also Court of First Instance in Case T- 143-89 Ferriere Nord SpA v Commission [1995] ECR II-917, paragraph 20.

472 Court of Justice in Case 7-82 GVL v Commission [1983] ECR 502, paragraph 24 et seq.; Court of First Instance in Case T-62-98 Volkswagen v Commission [2000] ECR II-2707, paragraph 199.

473 [...]* (file, p. 3534 et seq.).

474 [...]*.

475 [...]*.

476 [...]*.

477 See the Commission's Deutsche Post Decision, paragraph 48.

478 Court of Justice in Case C-279-87 Tipp-Ex v Commission [1990] ECR I-261; BPB Industries and British Gypsum Ltd v Commission, paragraph 165.

479 OJ C 9, 14.1.1998, p. 3.

480 See section I.E.

481 Parliament and Council Regulation (EC) No 2887-2000 of 18 December 2000 on unbundled access to the local loop (OJ L 336, 30.12.2000, p. 4) entered into force on 1 January 2001 in France as in the other Member States.

482 See the presentation by the representative of Mangoosta's official receiver at the hearing of 18 March 2002 tracking the impact of roll-out of the eXtense service on the customer contact and order withdrawal rates registered by Mangoosta between week 44 of 2000 and week 11 of 2001 (file, pp. 1493 and 1494). Mangoosta's representative described the impact which the mere announcement of the eXtense package launch had on its expansion plan. Over and above its commercial effects, reflected in an immediate drop in orders received by Mangoosta, the implementation of Wanadoo Interactive's commercial strategy was claimed to have caused negotiations between the company and its creditors to be broken off, thereby seriously undermining its viability.

483 In 2000 the market consisted of less than 200 000 subscribers and was growing relatively slowly. Furthermore, it was only after March 2001 that Wanadoo Interactive left Noos well behind by launching its national marketing campaign for the eXtense package.

484 See section I.F.1.

Annexes

Annex 1: Adjusted variable costs of the eXtense service: recovery from 8 January to 31 July 2001 (actual)

Annex 2: Adjusted variable costs of Wanadoo ADSL: recovery from 1 January to 31 July 2001 (actual)

Annex 3: Adjusted variable costs of the eXtense service: recovery from 1 August to 15 October 2001 (actual)

Annex 4: Adjusted variable costs of Wanadoo ADSL: recovery from 1 August to 15 October 2001 (actual)

Annex 5: Adjusted variable costs of the eXtense service: recovery from 15 October 2001 to 15 February 2002 (actual)

Annex 6: Adjusted variable costs of Wanadoo ADSL: recovery from 15 October 2001 to 15 February 2002 (actual)

Annex 7: Adjusted variable costs of the eXtense service: recovery from 15 February to 30 September 2002 (actual)

Annex 8: Adjusted variable costs of Wanadoo ADSL: recovery from 15 February to 30 September 2002 (actual)

Annex 9: Adjusted variable costs of the eXtense service: forecast recovery for the second half of 2001, only new subscribers

Annex 10: Adjusted variable costs of Wanadoo ADSL: forecast recovery for the second half of 2001, only new subscribers

Annex 11: Adjusted variable costs of the eXtense service: forecast recovery from January to March 2002

Annex 12: Adjusted variable costs of Wanadoo ADSL: forecast recovery from January to March 2002

Annex 13: Complete costs: actual for 2001

Annex 14: Complete costs: estimates for 2002

Annex 15: Summary of cost-recovery indicators

Annex 16: Reconstitution of the turnover on eXtense and Wanadoo ADSL from August 2001 to June 2002

Annex 17: Cost of the bandwidth for IP traffic invoiced by France Télécom from September 2001 to June 2002

Annex 18: Evaluation of the cost of international connectivity from February 2002 to June 2002

Annex 19: Business plans for 2001

Annex 20: Estimate of the margin trend on the combined direct costs of the eXtense and Wanadoo ADSL services from January 2001 to December 2002

Annex 21: Advertising expenditure and growth in total number of subscribers

Annex 22: Advertising and promotional expenditure for Wanadoo Interactive and its highspeed competitors (512 kbit/s)

Annex 23: Comparison of total number of subscribers of Noos, Club Internet and Wanadoo Interactive from December 1999 to March 2001

Annex 24: Comparison of high-speed and low-speed offers

Annex 25: High-speed/low-speed interchangeability

Annex 26: Growth-rate trend for Wanadoo and its competitors

ANNEX 1 - Adjusted variable costs of the eXtense service: recovery from 8 January to 31 July 2001 (actual)

<emplacement tableau>

ANNEX 2 - Adjusted variable costs of Wanadoo ADSL : recovery from 1 January 31 July 2001 (actual)

<emplacement tableau>

ANNEX 3 - Adjusted variable costs of the eXtense service: recovery from 1 August to 15 October 2001 (actual)

<emplacement tableau>

ANNEX 4 - Adjusted variable costs of Wanadoo ADSL : recovery from 1 August to 15 October 2001 (actual)

<emplacement tableau>

ANNEX 5 - Adjusted variable costs of the eXtense service : recovery from 15 October 2001 to 15 February 2002 (actual)

<emplacement tableau>

ANNEX 6 - Adjusted variable costs of Wanadoo ADSL : recovery from 15 October 2001 to 15 February 2002 (actual)

<emplacement tableau>

ANNEX 7 - Adjusted variable costs of the eXtense service : recovery from 15 February to 30 September 2002 (actual)

<emplacement tableau>

ANNEX 8 - Adjusted variable costs of Wanadoo ADSL : recovery from 15 February to 30 September 2002 (actual)

<emplacement tableau>

ANNEX 9 - Adjusted variable costs of the eXtense service : forecast recovery for the second half of 2001, only new subscribers

<emplacement tableau>

ANNEX 10 - Adjusted variable costs of Wanadoo ADSL : forecast recovery for the second half of 2001, only new subscribers

<emplacement tableau>

ANNEX 11 - Adjusted variable costs of the eXtense service : forecast recovery from January to March 2002

<emplacement tableau>

ANNEX 12 - Adjusted variable costs of Wanadoo ADSL : forecast recovery from January to March 2002

<emplacement tableau>

ANNEX 13 - Complete costs: actual for 2001

<emplacement tableau>

ANNEX 14 - Complete costs: estimates for 2002

<emplacement tableau>

<emplacement tableau>

ANNEX 15 - Summary of cost-recovery indicators

<emplacement tableau>

ANNEX 16 - Reconstitution of the turnover on eXtense and Wanadoo ADSL from August 2001 to June 2002

<emplacement tableau>

<emplacement tableau>

ANNEX 17 - Cost of the bandwidth for IP traffic invoiced by France Télécom from September 2001 to June 2002

<emplacement tableau>

ANNEX 18 - Evaluation of the cost of international connectivity from February 2002 to June 2002

France Télécom carried out an accurate measurement of ADSL traffic flows from April 2002 but only on the basis of Netissimo 1-type flows

For the period 15 February to 31 March 2002, international-connectivity traffic is assumed to be [...]* of ADSL routing traffic invoiced by France Télécom. [...]*

This traffic is shown in Annex 17.

For the period 31 March to 30 June, the data supplied by France Télécom in its reply of 20 December 2002 is used.

The average is calculated on the basis of a weighting of traffic by the total average number of ADSL subscribers

<emplacement tableau>

ANNEX 19 - Business plans for 2001

<emplacement tableau>

ANNEX 20 - Estimate of the margin trend on the combined direct costs of the eXtense and Wanadoo ADSL services from January 2001 to December 2002

<emplacement tableau>

ANNEX 21 - Advertising expenditure and growth in total number of subscribers

<emplacement tableau>

ANNEX 22 - Advertising and promotional expenditure for Wanadoo Interactive and its high-speed competitors (512 kbit/s)

22.1. ADVERTISING AND PROMOTIONAL EXPENDITURE OF COMPETITORS (euro '000, unless otherwise indicated)

Aggregate data taken from the replies from Club Internet, AOL, Tiscali, Easynet, 9Online, Free, Net Pratique, NC Numéricâble, UPC France, Noos, Nerim

Free access to the services offered by some providers has not been regarded as a promotion

<emplacement tableau>

<emplacement tableau>

ANNEX 23 - Comparison of total number of high-speed subscribers of Noos, Club Internet and Wanadoo Interactive from December 1999 to March 2001

<emplacement tableau>

ANNEX 24 - Comparison of high-speed and low-speed offerings

<emplacement tableau>

ANNEX 25 - High-speed/low-speed interchangeability

<emplacement tableau>

ANNEX 26 - Growth-rate trend for Wanadoo and its competitors

<emplacement tableau>

[chart]*

The graph illustrates the weekly growth trend in the number of subscribers of Wanadoo and all its competitors (via cable modem and ADSL)

The sales of cable operators crumbled in the first quarter of 2001. Sales of competing ADSL providers only took off from October 2002 (end of the abuse). The drop in Wanadoo's sales in the spring of 2002 was due to the effects of the prohibition on marketing eXtense in FT's sales outlets following the French competition council's decision in February 2002.

NOTES

In the published version of this decision, some information has been omitted pursuant to Articles 20(2) and 21(2) of Council Regulation (EC) No 17-62 (replaced by Articles 28(2) and 30(2) of Council Regulation (EC No 1-2003) concerning non-disclosure of business secrets and other confidential information. The omissions are shown thus [...]*. Where possible the information omitted has been replaced by ranges of figures or a general description.