EC, October 17, 2007, No 38.606

THE COMMISSION OF THE EUROPEAN COMMUNITIES,

Having regard to the Treaty establishing the European Community, Having regard to Council Regulation (EC) No 1-2003 of 16 December 2002 on the implementation of the rules on competition laid down in Articles 81 and 82 of the Treaty (1), and in particular Article 7 thereof, Having given the undertakings concerned the opportunity to make known their views on the objections raised by the Commission in accordance with Article 27 of Regulation No 1-2003 and Article 12 of Commission Regulation No 773-2004 (2), Having regard to the final report of the Hearing Officer in this case, After consulting the Advisory Committee on Restrictive Practices and Dominant Positions (3), Whereas:

1. INTRODUCTION

1. This case concerns a decision by Groupement des Cartes Bancaires "CB" ("the Groupement") introducing a series of measures having the object and effect of restricting the competitive advantage of some of its members (primarily the banking arms of large retailers and online banks) who are new entrants (4) in the market for issuing bank cards in France, to the benefit of "main" members of the Groupement. The adoption of these measures was made possible by the distinction drawn in the Groupement's articles of association (Formation Agreement and Rules of Procedure) between "main" members (members of the Board of Directors who have been granted exclusive powers enabling the measures to be adopted) and other members of the Groupement. By increasing the cost of the cards issued by the new entrants, the measures adopted by the Groupement - despite formally applicable equally to all members of the Groupement - in practice have discriminatory effects which help to maintain the price of bank cards in France above the level that would result from free competition and to limit the number of cards supplied, to the detriment of consumers.

2. PROCEDURE

2. On 10 December 2002 the Groupement notified to the Commission (see section 7 below) pricing measures consisting of a mechanism for regulating the acquiring function (the "Merfa"), a change in the membership fee for the Groupement, an additional membership fee and a fee for "dormant members", as well as non-price measures consisting of new formulae for allocating voting rights among members of the Groupement and rights to the Groupement's assets and to the revenue from the membership fees and the dormant members' fees. The notified measures entered into force on 1 January 2003.

3. A notice inviting third parties to submit their comments on the agreements notified by the Groupement was published in the Official Journal (5).

4. On 20 and 21 May 2003 inspections were carried out under Article 14(3) of Regulation No 17 (6) on the premises of the Groupement and on those of eight members of the Board of Directors of the Groupement (7) and of companies specialised in electronic banking (8) (a term which refers in particular to payment by bank card).

5. On 1 May 2004 the notification lapsed following the entry into force of Regulation No 1-2003 (9).

6. By letter dated 10 June 2004 (10) the Groupement informed the Commission that its Board of Directors had decided unanimously, on 8 June 2004, "not to implement in their current state the notified measures for which it was responsible, i.e. the Merfa, the new arrangements for the membership fee and the additional fee applicable to 'dormant members'" pending a Commission decision on their compatibility with competition law (11).

7. On 8 July 2004 the Commission addressed a Statement of Objections to the Groupement and to nine banks members of its Board of Directors (BNP-Paribas, Caisse Nationale d'Epargne et de Prévoyance (hereinafter "CNCEP"), Crédit Agricole, Crédit Mutuel, Crédit Industriel et Commercial (hereinafter "CIC"), Crédit Lyonnais, La Poste, Natexis-Banques Populaires and Société Générale). The Statement of Objections of 8 July 2004 related to an agreement between these nine banks and the Groupement.

8. The addressees of the Statement of Objections sent their observations to the Commission on 8 November 2004.

9. A hearing was held on 16 and 17 December 2004.

10. During the Combined General Meeting held on 11 March 2005, a proposal for amending the Groupement's Formation Agreement (12) was submitted to the members of the Groupement for approval. That amendment removed from the Formation Agreement the amendments passed by the General Meeting of 20 December 2002 (see recitals 155 and 161).

11. A second Statement of Objections - replacing the previous one, which was withdrawn - was sent to the Groupement on 18 July 2006. That Statement of Objections, like the present Decision, which is a follow-up to it, concerns a decision by an association of undertakings (a decision of the Groupement) establishing a series of measures having an anticompetitive object and effect.

12. On 19 October 2006 the Groupement sent the Commission its observations in reply to the Statement of Objections of 18 July 2006, followed on 17 November 2006 by supplementary annexes (13).

13. A hearing was held on 13 November 2006.

3. THE RELEVANT SECTOR

3.1. General

14. Payment cards can be used to pay for purchases from merchants, whether outright, by deferred debit or on credit, or to withdraw cash from an automated teller machine (ATM). They can also be used to make distance payments (by telephone or over the Internet).

15. The transactions of payment or withdrawal by card can take place only downstream of an issuance activity and a merchant acquiring activity. Issuance is an activity whereby a financial institution issues and delivers cards, generally in return for payment by cardholders of an annual fee, commonly called "cardholder fee". Merchant acquiring consists in financial institutions offering merchants the service of processing card payment transactions. These two activities, though complementary and interdependent, constitute two distinct occupations.

3.2. Card payment systems in France

16. Among the cards used in France, a distinction can be made between "CB" bank cards, proprietary or store cards (cards issued by distributors, as a rule in cooperation with institutions specialised in consumer credit), and charge cards, the best known of which are American Express and Diners Club.

17. The CB system is by far the biggest payment system in France. CB cards are considered universal cards in France, such is the number of points (payment terminals and ATMs) that accept them, which sets them apart from proprietary cards and charge cards, which are less widely accepted.

18. In France, banks issue either "national" CB cards - which cannot be used outside the country - or "international" CB cards. Whenever an "international" CB card is used to make a payment in France, it is the "CB" function that is used. The transaction is processed by the Groupement network, without any technical or financial intervention by Visa or MasterCard, whose logo appears on the card. The "Visa" or "MasterCard" function is used only when the cardholder makes payments outside France.

19. There are a large number of banks which practise merchant acquiring (87 banking groups were acquiring merchants for CB cards in the fourth quarter of 2002) (14). However, the bulk of the merchant acquiring agreements with large networks (hypermarkets, fuel distributors, travel agents) are concentrated in the hands of a few operators (see recital 205 below). The agreements proposed to merchants by members of the Groupement provide not only for the acceptance of CB cards but also systematically for the acceptance of Visa and MasterCard cards.

20. The number of cards is not a sufficient indication of the importance of the various types of card since numerous proprietary cards, for instance, are rarely, if ever, used. The value of transactions, on the other hand, can serve as a reference point. The figures for France are as follows:

Type of card: Value of transactions (€ million) Market share

CB cards (including CB/Visa and CB/MasterCard) 190 000 78.35%

American Express/Diners Club 12 500 5.15%

Aurore, other merchants' proprietary cards 40 000 16.49%

Total 242 500 100%

Source: RBR report, "Payment Cards in Europe 2004", file p. 25733.

3.2.1. CB cards

21. CB cards are issued by the members of the Groupement. The approximately 149 (15) members of the Groupement are French banks and credit institutions, including all the retail banks in France, and foreign banks.

22. Under the rules of the Groupement adopted in 1995 and amended in 2000, Visa or MasterCard cards issued in other countries by a bank that is not a member of the Groupement can be used in the CB system only if they are used more than 50% (in terms of volume or the number of transactions) in the CB system (16).

23. CB cards are the only universal payment cards in France: they are distinguishable among other things by their extremely large acceptance network in that country and by the possibility of using the Visa or MasterCard function abroad. CB cards are accepted in France by over a million sales outlets and terminals (17) and 46 151 ATMs (18). The number of CB cards in circulation is over 51.2 million (19). In 2005, 6.27 billion payment and withdrawal transactions were carried out, for a total amount of €325.4 billion (20).

3.2.2. Proprietary or store cards

24. Proprietary or store cards bear the trade marks of shops or private businesses. Their use is often limited to effecting payment in those shops or in "partner" businesses. They are therefore not universal payment cards as they do not make it possible to effect payments on premises other than these shops, businesses or partner networks and are only rarely usable abroad, save in a partner network. They are issued and managed by credit institutions on behalf of businesses (while a CB card, on the other hand, can be issued by any institution belonging to the CB system). Mention may be made, for example, of the Aurore card, issued by Cetelem (a BNP-Paribas subsidiary) and its subsidiaries and partners, the Pass card of the Carrefour hypermarket group (issued by its banking subsidiary S2P), or the Auchan card (issued by Banque Accord). Although in terms of numbers they account for 44.38% of all payment cards in France - as compared with the 53.42% which are CB cards (21) - their market share in terms of both volume and value of transactions is much smaller (3.8% and 15.4% respectively) (22).

3.2.3. American Express and Diners Club cards

25. American Express cards are issued by the American Express Company or its accredited partners throughout the world. In France, they are issued by a subsidiary, American Express Carte France SA, and by a number of financial institutions under licence (23). There are approximately 1.9 million American Express cards in circulation in France (24). The American Express card is accepted by 9.6 million merchants worldwide, in particular those associated with travel (hotels, restaurants, airlines, etc.) (25). Acquiring is effected solely by American Express Europe Limited. American Express has concluded agreements with certain French banks (Société Générale, Crédit Lyonnais, Crédit Mutuel, CIC and Caisses d'Épargne), which make their ATMs available for withdrawals using the American Express card.

26. Diners Club International is a subsidiary of the banking group Citicorp, which operates through franchisees across the world. There are approximately 30 000 Diners Club cards in circulation in France (26). Some large French banks have made their ATMs available to Diners Club (e.g. La Poste, Société Générale, Crédit Agricole and Caisses d'Épargne).

27. Together, American Express and Diners Club cards account for slightly more than 5% of payments in value terms in France (and 1.6% in volume) (27).

4. GROUPEMENT DES CARTES BANCAIRES "CB"

28. The principles according to which the Groupement operates are described in its Formation Agreement and Rules of Procedure (28). The Formation Agreement is concerned mainly with the legal and organisational arrangements of the Groupement, while the Rules of Procedure lay down the broad lines of the Groupement's practical and operational arrangements.

4.1. Main members and secondary members; affiliates

29. Members of the Groupement are either main members or secondary members attached to one of the main members (29).

30. A "main member" is a member of the Board of Directors of the Groupement (definition given in the Preamble to the Rules of Procedure (30)). There are 11 main members, listed exhaustively and by name in Article 11.2 of the Formation Agreement (Banque Fédérale des Banques Populaires (representing the Banques Populaires group), BNP Paribas, CNCEP, CCF, Confédération Nationale du Crédit Mutuel, Crédit Agricole et Fédération Nationale du Crédit Agricole (hereinafter "FNCA"), Crédit du Nord, CIC, Crédit Lyonnais, La Poste and Société Générale).

31. The main members of the Groupement are a closed group of members. The Groupement's articles of association (Formation Agreement and Rules of Procedure) do not provide for any procedure for acquiring the status of main member bank, with the result that a competitor that succeeded in significantly expanding its electronic banking business could not acquire the status of main member except in the hypothetical event of its succeeding in securing an amendment to the Formation Agreement, which is in practice impossible without the consent of the main members (31).

32. A "secondary member" is any member of the Groupement that is not a member of the Board of Directors and is attached to a main member (Preamble to the Rules of Procedure). According to Article 7.1 of the Formation Agreement, "Membership of the Groupement requires the institution to be attached for the entire period to one of the Main Members".

33. The term "affiliate" refers to an institution that is not a member of the Groupement but is authorised to engage in CB activities on account of its affiliation to a member of the Groupement (32).

34. Pursuant to Article 7.1 of the Formation Agreement: "The request to join the Groupement must be submitted to the Chief Executive Officer by a Main Member. It is subject to the approval of the Board of Directors".

4.2. The Chief Executive Officer

35. The Groupement is managed by a Chief Executive Officer (Article 15 of the Formation Agreement). While enjoying very extensive powers for managing the Groupement, the CEO is expected to comply with and implement the instructions issued by the Board of Directors, to which he reports (Article 16 of the Formation Agreement).

4.3. The Board of Directors

36. It is clear from paragraphs 2 to 4 of Article 13 of the Formation Agreement, listing the tasks of the Board of Directors, from other articles of the Formation Agreement and the Rules of Procedure, and from practice within the Groupement that the Board of Directors is the body of the Groupement which has general decision-making power. By contrast, the General Meeting has only limited powers (33) and the CEO is an executive subordinate to the Board of Directors (34).

37. The Groupement's Formation Agreement lays down the principle that "the Board of Directors is invested with the widest powers to take all decisions concerning the major options facing the Groupement, subject to the powers conferred ... on the General Meeting ... and the Chief Executive Officer" (Article 13.1 of the Formation Agreement) and is "responsible ... for determining the general policy followed by the Groupement and accordingly for issuing instructions to the Chief Executive Officer" (Article 11.1 of the Formation Agreement).

38. Under the powers formally conferred on the Board of Directors, it alone may adopt pricing measures applicable to all members of the Groupement (new (35) Article 13 of the Rules of Procedure (36) and Article 14 of those same Rules (37)). The Board of Directors may also shape and steer the general policy of the Groupement (Article 13.4 of the Formation Agreement); amend the Rules of Procedure of the Groupement (38); propose amendments to the Formation Agreement (39); approve the accession of a new member, it being possible for a candidate bank to present an application for membership only through the medium of a main member (40); determine the sanctions applicable to members (41); exclude a member from the Groupement (42); authorise the issuing of cards by any member of the Groupement (43); determine the conditions of the card payment guarantee (44); and lay down the conditions for merchants to accept payment by card, approve the conditions of access for cardholders to payment terminals and ATMs and, approve the rules requiring members to respond to all requests for authorization (45).

39. Depending on the type of decision, decisions may be taken by a simple majority (46), by a majority of two thirds of the votes (47), or by a majority of three quarters of the votes (48) of members present or represented. However, if three members so request, decisions of the Board of Directors may require a dual majority. In that case, in addition to the majorities mentioned above, a simple majority of the members present is also required (49).

40. Neither the Formation Agreement nor the Rules of Procedure requires prior consultation of those that are not main members concerning draft pricing measures.

41. The pricing measures which vary according to the number of cards issued (Merfa, membership fee per card, additional membership fee and "dormant members" mechanism) (50) were drafted by the main members and then formally adopted by the Board of Directors, without the non-main members being able to participate in their preparation or even simply being consulted in advance, not having been informed of the existence of the pricing measures until after their adoption (51) (see recital 133, section 6 below).

4.5. General Meetings

42. All members have the right to participate in General Meetings. Pursuant to Article 24 of the Formation Agreement, it is the responsibility of the Ordinary General Meeting to approve or rectify the accounts and appoint or dismiss the Chief Executive Officer, the Management Auditors and the Statutory Auditor. The Extraordinary General Meeting may amend the Formation Agreement and call for the premature winding-up of the Groupement or the prolongation of its duration.

43. In the General Meeting, each member has one vote plus a number of votes equal to the sum of the number of transactions initiated by cards within the scope of the Groupement (Article 23.3 of the Formation Agreement). As the members of the Board of Directors account for most of the transactions carried out by the Groupement (52), in practice motions put to the General Meeting with the Board's agreement are certain to be carried.

4.6. The Electronic Banking Steering Committee

44. The composition and functions of the Electronic Banking Steering Committee ("the COM") are mentioned neither in the Rules of Procedure of the Groupement nor in the Formation Agreement. In its reply of 7 November 2003 to the Commission's request for information of 24 October 2003, the Groupement described the COM as an "informal body" that had been meeting since February 1999 and brought together "[...] certain banking institutions particularly active in the management of electronic means of payment" and "[...] the Groupement on the basis of their experience of administering a payment system". The organisations represented in the COM are in fact always and exclusively main members.

45. The COM is composed of [...] the Groupement and [...] the main member banks, apart from CCF, a subsidiary of the UK bank HSBC. [...] the Groupement's articles of association (Formation Agreement and Rules of Procedure) make no mention of the COM, nor a fortiori do they confer on it any decision-making powers or any status as a body of the Groupement.

46. The COM prepared the pricing measures in question in the course of many meetings where these highly practical issues were debated, on the basis of simulations and economic studies produced by consultants (see section 6). The measures prepared here were then formally adopted by the Board of Directors.

47. Several documents obtained during the inspections show that, despite its lack of formal decision-making powers, the COM exercises de facto powers that are absolutely crucial in the decision-making process. For example, it was concluded during the meeting of the COM on 11 October 2002 that "the entire dossier was accepted" (53). A statement was also made to the effect that the "policy decisions adopted" by the COM had simply been "put into effect" by the Board of Directors (54). At all events, no discussion of substance ever took place at meetings of the Board of Directors, and even less at the General Meeting, as is explained in section 6 below.

5. THE SYSTEM OF INTERCHANGE FEES IN FRANCE

48. In the CB system, according to the Groupement:

- for each payment using a CB card, the bank that issued the card receives a commission (Merchant Interchange Fee, or CIP) from the bank of the merchant on whose premises the payment is made (55);

- for each withdrawal using a CB card, the bank that issued the card pays a commission (ATM Interchange Fee, or CIR) to the bank operating the ATM at which the withdrawal is made (56). In accordance with Article 14 of the Rules of Procedure of the Groupement, the rules governing these commissions are laid down by the Board of Directors.

49. According to the Groupement, the interchange fees remunerate the services (57) which the banks provide each other to enable them to carry on their electronic banking business (payment or withdrawal transactions), but also constitute a mechanism for regulating the system (58).

5.1. The CIP

50. The purpose of the CIP is to remunerate the services provided by the issuing bank to the acquiring bank (59). According to the Groupement, in so doing it also performs a function of regulating the system by balancing the issuance and acquiring functions of the various members (60).

51. A CIP is calculated for each "CIP Group", in other words for each group of banks composed of a member of the Groupement (referred to as "head of CIP Group") and those of the institutions that are at least 51% controlled by it and "so wish" (61). It can be seen from the formula adopted for calculating the CIP [CIP = €0.1067 + amount of the transaction x (0.21% + TICO)] that this commission consists of three separate components corresponding to the three types of services it is said to be intended to remunerate:

- a (fixed) flat amount (€0.1067) for each payment transaction to cover the fixed processing costs borne by the bank that issued the card; and

- two variable amounts calculated ad valorem by applying a different rate to the amount of the transaction:

- The first is a fixed rate of 0.21%. This is said to be intended to cover the cost of collective security measures aimed at preventing and containing fraud and managing incidents resulting there from.

- The second is the TICO (Interbank Blocked Cards Rate) intended to offset the risk taken by the issuing bank in guaranteeing for the benefit of the acquiring banks payment of the transactions carried out with its cards at the premises of merchants linked to the acquiring banks. Since the risk taken in this respect by an issuing bank is assessed with regard to each of the acquiring banks, there are as many different TICOs as there are bilateral relations between such an issuing bank and each of the acquiring banks. The TICO is equal to the ratio - for each combination of issuing bank and acquiring bank - between the amount of fraudulent transactions (those carried out using cards on the blocked cards list) and the total amount of transactions recorded (62).

5.2. The CIR (63)

52. Each withdrawal transaction gives rise to the payment of a cash advance and possibly also to the payment of other commissions, referred to as the "Withdrawal Service Commission" (CSR) and the "Card Service Commission" (CSC). According to the Groupement, the cash advance, the CSR and the CSC are component parts of a single commission, referred to as the ATM Interchange Fee (CIR) (64).

53. More specifically:

- The cash advance, amounting to € [...], is payable by the bank that issued the card used. The cash advance is transferred to the banks operating the ATMs from which the withdrawals were made. The purpose of the advance is to remunerate the service which the bank renders to the issuing bank by dispensing banknotes to the holders of cards issued by the latter (65). The cash advance is calculated "for each main member, including in respect of its secondary members" (66).

This may be supplemented by:

- either a Withdrawal Service Commission (CSR), paid only by banks whose issuing activity exceeds their acquiring activity, that is to say, where there is less than one withdrawal from a bank's ATM network using other banks' cards (withdrawals "serviced" by the bank) for every two withdrawals from other members ATM networks using cards issued by that bank (withdrawals "foreign" to the issuing bank); in other words, the CSR is payable in respect of each foreign withdrawal by banks whose "servicing rate" (67) is less than 1/3, i.e. by issuing banks that have few, if any, ATMs; its maximum amount is € [...] per foreign withdrawal (68) and it is transferred to the banks operating the ATMs from which the foreign withdrawals were made (69).

- or a Card Service Commission (CSC), paid only by banks whose acquiring activity exceeds their issuing activity, that is to say, where there are more than two withdrawals serviced by those banks for every one foreign withdrawal; in respect of each serviced withdrawal, therefore, the CSC is payable by banks whose servicing rate is more than 2/3, i.e. by banks operating ATMs (acquiring banks) that issue few, if any, cards; the maximum amount of the CSC is € [...] per serviced withdrawal and it is transferred to the banks that issued the cards used in their ATMs (70).

54. Therefore, whereas the cash advance and the CSR "counter-commission" are paid in respect of each foreign withdrawal (payment by the issuing bank to the acquiring bank), the CSC "counter-commission" is paid in respect of each serviced withdrawal (payment by the acquiring bank to the issuing bank).

55. The CSR was added to the cash advance in 1990. The CSC was added in 1996 (71).

56. According to the Groupement, the "CIR is based on a self-regulating mechanism ... and represents a balance between the divergent aims of [card] issuers and ATM operators [i.e. banks acquiring the withdrawal function] (72)". Again according to the Groupement, the purpose of the CSR and the CSC is to allow balanced development of the interbanking components of the CB system constituted by the number of cardholders and the number of ATMs. Payment of the CSR by those banks whose contribution in terms of ATMs to CB interbanking is quite insufficient in comparison with their contribution in terms of cardholders, and payment of the CSC by those banks whose contribution in terms of ATMs is by far excessive in comparison with their contribution in terms of cardholders, is a weighting and compensation mechanism that avoids the balance and viability of the withdrawal function being jeopardized (73).

6. WORK DONE ON THE CS 2002 PROJECT

6.1. Introduction

57. The "CS 2002" project was conceived within the COM, with the participation of all the main members except for Crédit Commercial de France (CCF), a subsidiary of the UK bank HSBC, which was not invited to the meetings (74). The project, the aim of which was to draw up new operational rules for the Groupement, - notably in relation to pricing - resulted in the establishment of the mechanism for regulating the acquiring function (or "Merfa"), the new membership fee (comprised of the membership fee per card and the additional membership fee), the "dormant members wake-up mechanism" and new formulae for calculating voting rights, rights to the assets of the Groupement and rights to proceeds from membership fees and wake-up fees. It was notified by the Groupement to the Commission on 10 December 2002 (see section 7 below).

58. Work on the CS 2002 project began within the COM in 2001 and continued until January 2003. During that period, the COM met on several occasions: 12 April 2001, 6 July 2001, 29 August 2001, 9 November 2001, 9 January 2002, 26 February 2002, 12 April 2002, 29 May 2002, 25 June 2002, 19 July 2002, 11 October 2002 and 7 January 2003 (75). A budget of between €1.8 million (76) and €2 million (77) was allocated to the project.

6.2. Chronological development of the CS 2002 project

59. The initial discussions began between March and November 2001. The principles underlying the project, one of the fundamental features of which was the introduction of a pricing system to prevent a reduction in the cardholder fee caused by the new entrants, were discussed at the COM meeting on 9 November 2001. After many simulations and fine-tunings organised by the [...] meeting within the COM from May 2002 onwards, the pricing measures involved were finally adopted by the Board of Directors on 8 and 29 November 2002.

6.2.1. Initial discussions on reform of the CB system prior to November 2001

60. As explained below, since March 2001, the banks sitting on the Board of Directors of the Groupement were concerned that card issuance by the new entrants would result in a fall in prices and considered that a collective reaction to the arrival of new entrants would be the best response.

61. In March 2001, [representatives] of the Groupement "promised" BNP-Paribas that they would convene a working group bringing together "CL, SG, CA, CM, La Poste, BP" and BNP-Paribas to discuss a problem raised by BNP-Paribas. In an internal memo (78), BNP-Paribas had identified the problem that, although the "new banks" such as ZeBank (as Egg was formerly known) or Bipop were required to pay the full charge for withdrawals as a result of the application of the CSR and the fact that they had no ATMs, they were nevertheless able to charge prices described as "highly aggressive in terms of cardholder fees". According to BNP Paribas's analysis, "As for the new discount banks, it is the reduced annual cardholder fee that worries us the most; it is to be feared that charging more for cash withdrawals will not suffice to make them change their marketing strategy" (79).

62. At a meeting of the COM held on 29 August 2001, it was decided "to reconsider all the arrangements for access to and use of the CB system" (80). A memo from Crédit Agricole to Cedicam (its electronic banking subsidiary) dated the following day also refers to the fact that "discussions will be held on improving the net margin on electronic banking (81).

63. A memo dated 25 September 2001 explains the "risk", the "objective" and the "response" (82).

"THE RISK

Companies along the lines of Zebank, S2P (Carrefour), Casino (even Amex) are converting their proprietary cards into CB cards.

All such a company has to do is to find a CB main member, negotiate a set of charges with it and replace its cards with new ones carrying the CB logo (cost: FRF 30?) and therefore accepted everywhere, while retaining its fidelity principle etc., since those CB cards are multi-purpose.

Not having made any investment (ATMs, GIEs (83), etc.) for years, it offers its cards at a knock-down price and thus poaches customers from bank X that was one of the founders of electronic banking since by signing up to such a card the customer enjoys the advantages of a CB card, but at a lower cost, plus the advantages of a proprietary card.

NB: there are approximately 30 million proprietary cards in France.

THE OBJECTIVE

The long-standing founders/investors want to:

- stop this fall in the selling price of the card, while holding on to their customers

- and (possibly?) recoup the amounts invested

Important: it can be seen from these first two paragraphs that the problem arises only in relation to issuance.

THE RESPONSE

Three simple principles:

- to avoid slashing prices, selling prices have to be pegged: one (or more) issuance fees therefore have to be introduced ...

- but to avoid making the long-standing members pay, this fee should be due only on new cards, on the delta for the year (it is therefore paid once only)

- lastly, the total levied by the GIE should be returned to the 'founder' main members, and NOT the secondary members and affiliates at the end of the year (issue to be discussed, with special reference to possible differentiation between secondary members and affiliates).

Two advantages/consequences:

- the one who pays is the one who issues a large number of cards, including the 'predator' who changes his cards

- the increase in the proportion of free cards is avoided, even in the case of the longstanding members ...

Introduction of the measures:

No change is made to the GIE's existing pricing system: RCB, EMA and SICB on the basis of unit costs, the balance according to the collective formula, but this 'tax' is added.

At the end of the year (the GIE is not allowed to make a profit), any surplus is distributed among the main members exclusively in proportion to the delta (*) of their number of cards in the year and NOT among the secondary members and affiliates.

* the ideal solution would be to deduct free cards from this in order to penalise them.

It can be seen that the deltas on the long-standing members are refunded to them. The GIE remains a GIE.

Only problem: does this have to be justified, and if so, to whom and how? In effect, can we do what we like?

Overall scheme

Main members/affiliates/secondary members

?

GIE

?

Main members

In reality, the only question, before asking how, is whether we have the right (legally, from a tax standpoint, etc.) to differentiate the first level from the second level invoices.

How much to charge/how to justify:

We will never be able to say, to use the argument, that the GIE is working harder, that it therefore needs more resources, etc., so we can only tell the truth:

- protecting ourselves against...

- recouping past investments

We can therefore only justify ourselves on the basis of the principle of recouping" (84) (emphasis in the original).

64. A handwritten note taken by [...] Société Générale [...] on 19 October 2001 refers to the project timetable: "The launch window is narrow but action has to be taken urgently (see above the problem of the banking arms of retail groups)" (85). The same document also contains statements on the aim of the reform (the introduction of a "peg stopping the annual cardholder fee from falling", "to try and organise transfers between players so as to allow remuneration of the investments made by those who contributed to setting up and operating the system, transfers either from the new entrants (Zebank) or from existing members using the system in order to develop more rapidly") (86).

6.2.2. Policy decisions taken at the meeting on 9 November 2001

65. A key meeting of the COM was held on 9 November 2001. According to the record of that meeting (87), [...] all the banks sitting on the Board of Directors of the Groupement (the "main members"), except CCF, were present. While agreeing on the need to study further the acceptability of the measures and the consequences thereof, the participants in the meeting on 9 November 2001 decided on the basic approach of the reform, namely:

- the introduction of a pricing system to shield participants from the reduction in the cardholder fee caused by the new entrants, and a sharing between the participants of the proceeds from the pricing system; and

- the need to impede the new entrants.

6.2.2.1. Establishment of a pricing system to shield participants in the COM from loss of income due to a reduction in the cardholder fee caused by the new entrants. The proceeds from the pricing system to be shared between the participants.

66. Throughout the meeting on 9 November 2001, [...] stressed the threat of new entrants arriving, with the losses that would result there from, and the need to protect themselves against such a development and to be "reimbursed" for their investments (or to have them "paid for"). Accordingly (88):

(a) [...] the Groupement said that:

" everyone is also aware that new entrants have very precise plans for becoming more active in CB ... According to the few calculations that we can make, it can be estimated that, out of the 40 to 50 million proprietary cards in circulation, if 2 million of these cards were to cannibalise CB cards, the community would lose FRF 0.5 billion.

Second example: if, through the actions of new competitors, there was an FRF 50 reduction in the annual cardholder fee (something that is not far-fetched and much less than the price proposed by Zebank) the impact would be FRF 2 billion on the collective margin. What is more, I personally think that we should reckon instead on FRF 100, in other words a loss of FRF 4 billion. The idea of introducing a charging system would shield us from such misfortunes".

"[O]ur approach can have two effects:

(a) New entrants are deterred when they see the figures

(b) they are determined to enter whatever the price and in that case we recoup our investments".

(b) According to the Groupement's [...]: " [L]arge retailers have armed themselves to besiege the CB system. They are ready. We have questioned Casino and they tell us they want to issue 2 million cards; so cannibalisation is on the way. We must urgently make progress on this topic".

(c) [...] Société Générale asked: "Once the machinery is in place, what will happen in five/six years' time? What do we do with the 'kitty'? How do we share it out?" and considers that "Lots of people will want to get their hands on it or give it back to the consumers" and that it was therefore necessary "To add to point 2 [of the remit] that we want to make this whole operation profitable".

(d) According to [...] Crédit Agricole, "[T]he aim of this economic study is to generate cash. But I fear that we will not see much of it come back and that it is more a means of protecting ourselves". "For us, it's a winner in both cases:

- either it deters new entrants and the growing cake is shared among ourselves

- or the cake is shared with new players and in that case we're still OK".

6.2.2.2. Impeding new entrants

67. The participants in the COM discussed the way in which new entrants could best be hindered. The parties refer in graphic terms, on several occasions, either to the creation of barriers to be placed in the way of new entrants and the need to ensure that those barriers are robust, or to the need to deter new entrants. For example:

"Will the barriers be real barriers and not little kerbstones?"; "Creating barriers is all well and good, but we will open up the system if our barriers are hedges" ([...] Caisses d'Épargne ).

" not to move unless the level of the wall to be built is high; it's not a kerb that we need" ([...] BNP-Paribas).

" we will find it difficult to be able to charge those prices without being caught. It's a problem of substance: the height and acceptability of the barriers" "The whole of the question has to be looked at closely ... 1. see if the barrier is robust and does not create any destructive secondary side-effects. 2. We must be certain that the external battle that we are going to fight will not get inside the castle " ([...] Crédit Agricole).

- " On the principle I agree and blocking new entrants: OK" ([...] Crédit Mutuel).

- "The pricing parameters have to be twisted to make them dissuasive and they seem to me to be so dissuasive that they will be very difficult to justify" ([...] Banques Populaires).

6.2.2.3. Remit entrusted to the management of the Groupement

68. By mutual agreement, the participants decided to entrust [representatives] of the Groupement with a three-point remit, which was read out towards the end of the meeting, to:

- "Evaluate all common assets, products and services (including the CB brand) linked to the bank card;

- Devise the principles and arrangements for charging any user of these assets, products and services in order to derive an optimum return there from;

- Build the legal, accounting and tax architecture and the tools necessary for implementing this system of charges".

69. At the end of the meeting, the participants delegated to [...] the task of "assessing the work" that was to be done (89). These two people did indeed continue to work on the steering of the project between meetings of the COM (see recitals 82 and 83 below).

6.2.3. Work by [consultancy firm] on the CS 2002 project in November- December 2001

70. In order to present to the COM by 9 January 2002 a draft response to the remit entrusted to it at the COM meeting on 9 November 2001, the Groupement requested the assistance of [consultancy firm] at a meeting held on 14 November 2001 (90).

71. In the context of this cooperation, [consultancy firm] prepared in December 2001 a presentation entitled "CS 2002 project. Steering Committee" (91). This presentation, which began by claiming that "the unlimited availability of the CB system free of charge is undermining the founders' control of the system" (92), contained an explanation of the context (threats to the CB system and fall in income from electronic banking) and of the reform of the CB system under consideration. The main points of the presentation were:

(a) Identification of the "threats", whether internal (such as the policy of "offering discounts on card issuance and opening up ATMs to proprietary cards" (93)) or external (such as "the direct issue of Visa and/or MasterCard cards usable in France, the emergence of alternative systems offered by computer service firms or telecoms operators and the large-scale issue of free-riding CB bank cards by large retailers").

As regards the internal threat of the "offering of discounts on card issuance" (the issuing of cards by new entrants at prices appreciably lower than those charged by the traditional banks), it was explained that the drop in card prices would lower revenues: "The issuing of free cards and price cutting on card issuance are pushing down issuance charges. A fall in cardholder fees of €7.5 per card engenders a total annual loss of €300 billion and a fall of €15 per card engenders a total annual loss of €625 billion" (94).

- The recent fall in income from electronic banking. For one thing, merchant commissions had fallen "on average by 10% between '96 and '99, which leaves the CB system prone to destabilisation: that represents a cumulative loss of nearly €300 million over the last five years" (95). This threatened the acquiring banks' income: "The continuation of aggressive behaviour on acceptance is directly threatening the €1.1 billion income of the acquiring banks".

(b) The risk of issuance income collapsing because of the new players, who were already present and "account for 7.8% of the total volume of activity of the CB system in 2000". The issuing of significant volumes of cards by these new entrants was resulting in substantial losses in the CB system: "The issuing of a million CB cards by one or more new entrants results, from the first year, in a loss of €100 million on total revenue from card issuance" (96). "The continuation of aggressive behaviour on acceptance is directly threatening the €1.1 billion income of the acquiring banks".

(c) The "contamination" or "contagion" effect on the traditional banks of the appreciably lower prices of the new entrants:

"The issuing of a million CB cards by one or more new entrants results, from the first year, in a loss of €100 million on total revenue from card issuance ... Description of the scenario:

- a new entrant issues 1 million cards at €15/card (price assumption: Intermarché card).

Consequences of the scenario:

- The founders issue 1.5 million cards.

- The founders issue:

- 50% of the new cards at €29.80/card.

- 50% of the new cards at €15/card.

- Contamination of annual cardholder fees on the cards in circulation:

Annual fee reduced to €15/card on 10% of the number in circulation.

A rapid fall in issue prices, a strong contagion effect on the stock of cards in circulation" (97).

(d) Explanations of the reform contemplated, which would involve introducing a pricing system and setting up one or more commercial structures. "The reform is based on:

- The introduction of charging for services provided

- The establishment of one or more commercial structures" (98).

According to the presentation, "charging is justified by the immediate provision of all the resources developed through the activity of the more senior members, which have devoted substantial investments to them" (99).

72. The risk of loss of income for the banks that founded the CB system was also discussed in another [consultancy firm] presentation on the challenges to be faced by the CB system founder banks (100). According to that presentation, the CB system founder banks must "safeguard the income from electronic banking in the face of the threats looming on the bank cards market: offensives by competing systems, 'new entrants'". In this connection, it is again explained that the new entrants could easily become large issuers of CB cards by converting their existing base of proprietary cards into CB cards and offering them to cardholders at an annual fee lower than that proposed by the main members: "Initially issued to encourage customer loyalty or to finance purchases, the functions of proprietary cards are being extended to that of a discounted bank card" (101).

73. In December 2001 some new banks were already offering cards at lower prices than those of the main members:

- Zebank was charging an annual cardholder fee of €21.9 for a Visa deferred debit card; Banque Directe was charging between €0 and €30, depending on the amount of purchases made, for the same type of card (102) (the amount charged at the time by BNP-Paribas, Société Générale and Crédit Lyonnais was around €40 (103)).

- Before the Groupement notified the new pricing system to the Commission, Banque Accord was offering customers of Auchan supermarkets a B international Visa card for €25, with a reduced annual cardholder fee (€15) for their spouse, six months free of charge and new functions such as cash back (104).

6.2.4. January-April 2002: initial attempts at adjusting the pricing system and work leading to the creation of an "industrial and commercial company" (SIC)

74. Another meeting of the COM was held on 9 January 2002. According to the record of that meeting:

"At the meeting on 9 January 2002, the members of the Electronic Banking Steering Committee (COM) validated the work carried out by [representatives] of Groupement des Cartes Bancaires. That work provided an appropriate response to the questions formulated by the COM on 9 November 2001. Transformation of the technological, commercial and geographic environment surrounding electronic banking and the current conditions in which the CB system operates call for an urgent reform of the system. The members of the COM should make arrangements for introducing and charging the 'fee for access to and use of CB interbank services' and establish a new industrial and commercial structure to promote this system, maintain its competitive differentiation and develop the new products and services" (105).

75. In February 2002, the representatives of the banks participating in the COM did not yet have a description of the system under consideration. For example, according to an internal FNCA memo dated 11 February 2002, "given the risks involved in such an arrangement from the standpoint of competition law, GCB does not seem willing to make anything available in writing on the system it is contemplating" (106). This confirms the description given by [...] Société Générale [...] of the discussions as taking place "within a very narrow circle" (107). An FNCA memo dated 11 February 2002 points out that the reform is the Groupement's response to the development of card issuance by large retailers:

"The development of card issuance by large retailers represents a threat to the banks ... the large retailers are in a position to promote the issue of cards at low prices and thereby to upset the balance in the card system's economy, which is largely based on the billing of a flat charge.

- However, the existing barriers to entry are flimsy and therefore allow these new entrants to benefit from all the investments of the CB community financed by the banks. ...

The response contemplated by GCB: it would be based, although I have not been able to obtain precise details, on charging a dissuasive fee representing the value of the banks' investments in the CB community. The fee would be charged by a limited company whose capital would be open only to the banks" (108).

76. An internal BNP Paribas memo dated 20 February 2002 contains a summary on the aim of the reform. According to the memo,

"The central objective of the reform is to remunerate - on future activities - the collective investments made in the past by the founders of the CB system. To that end, three types of action are necessary:

- optimise the CB system as it is today

- introduce mechanisms to charge for new activities, remunerating the value of the investment

- create one or more commercial structures that become owners of the CB system".

On the subject of the commercial structure(s), the memo explains:

"The legal setup has not been definitively decided. The arrangement should be based on three principles:

- the CB founders constitute the pool(s) for the companies to be established. They consequently become the real owners of the CB system. The formula for apportioning shares has yet to be worked out.

- these companies are designed to collect the income from the pricing system ... and redistribute it in the form of dividends.

- the dividends enable the CB founders to offset the additional costs they incur as a result of the new pricing system. ...

Through the entry fees it would introduce, the proposed reform would significantly reduce the scope for slashing prices. If the cannibalisation of our volumes were kept steady, the price effect alone would allow us to regain some of the growth in our net banking income from cards (around €5 million). It is furthermore plausible to expect that there would be less cannibalisation of our volumes, since the new entrants' offerings should be less attractive in terms of prices" (109).

77. The first simulations began in late February 2002 (110) and a further meeting of the COM was held on 26 February 2002. It is clear from the record of the meeting (111) that it was convened to review progress. It was first noted that the project was going well, the necessary adjustments ("additions to the structure") were then reviewed, and the meeting ended with the adoption of a work schedule.

78. In March 2002 several studies and detailed presentations of the reform were carried out (112). One of these presentations covered the principles for setting up the "SIC" (a newly created "industrial and commercial company") and charging the entry fee and the fee for using the CB image.

79. In a presentation dated 18 March 2002, entitled "CS 2002 project - Characteristics of the simulation" (113), simulations were made with and without the new structure. These simulations took account of the threat of electronic banking activities being launched by new entrants, which would result in market shares being captured and a fall in (a) the price of new CB cards, (b) the price of part of the stock of CB cards ("contamination") and (c) in income from the CIP (new activity "on us"). The simulations were made on the basis of data from the "audit of the GIE Cartes Bancaires 'CB'", which forecast "saturation around [...] in the number of CB cards in France (with [...] million cards in circulation at that time)" and "saturation around [...] in acceptance (with a pool of 46 000 ATMs at that time and more than 1.2 million agreements with merchants) ... > The CB system should grow further to reach saturation around 2010". On that basis,

- the simulations without the new structure gave as a result a potential loss of €892 million in net banking income for the current members between 2002 and 2008 (114) as a result of the capture of market shares by new entrants;

- the simulations with the new structure (which involved the levying of two charges: entry fee and fee for the right to use the CB image) led to the conclusion that the current members would succeed in limiting the fall in their net banking income (potential loss reduced to €386 million) (115).

80. The simulations dated 18 March 2002 referred to "new entrants" in contrast with "current members". Nevertheless, it was clear that by "new entrants" those attending meetings of the COM meant in fact existing members of the Groupement who were increasing the number of their cards and were thus "entering" directly into a card issuance market in which their presence had so far been limited, and that the aim of the simulations was to ensure that the new pricing system would benefit the main members. This is clearly explained in an internal Caisses d'Épargne email dated 25 March 2002:

"[T]he aim of the model is to see that the system enables the income losses caused by the arrival of new entrants to be reduced ... The fear is the arrival of new entrants (already CB members) (and this is the only scenario examined) who would:

- capture market shares

- cut prices".

"Such a pricing system is almost entirely beneficial to the founder banks.

Dividends are distributed according to a fixed formula (for five years): which means that capturing market shares is costly but is not profitable to the same extent" (116).

81. The simulations carried out in March and April 2002 also cover the effects of the reform contemplated on the main members sitting on the COM (for example, for Société Générale (117), Crédit Agricole (118) and BNP Paribas (119)).

82. On 28 March 2002 [...] "delegates" of the COM ([...] Société Générale and La Poste), met to review progress made on the project. One of the documents on the agenda is entitled "Founding principles of the CS 2002 project" (120). The other is a presentation on the SIC ("Plan to set up an industrial and commercial company (SIC)"). The handwritten notes taken by [...] Société Générale in connection with that presentation reflect a number of questions:

" - allow in on the basis of what formulae?

- same rule (moving average over n years) ...

- or not same rule:

- take a snapshot of last year on entry

- then vary it by moving average (penalise those on the way up) at given intervals (4 years?)

- how (two classes of shares, shareholder agreement ...) make sure with the 8 [i.e. the main members taking part in the meetings] (97%) - penalise" (121).

83. On 3 April 2002 [...] ([...] Société Générale and [...] La Poste) met again (122) to take stock of (1) the legal aspects (viability of the principle of charging a fee for access; "[N]eed to launch the preparation of a dossier to support the arguments put forward (stability of the value of the system)"); (2) the tax aspects of the SIC, including the "dividends generated"; (3) the shareholder agreement; (4) the draft agreement; and (5) the agenda for the COM meeting on 12 April (123).

84. Spreadsheets making it possible to prepare and finalise the simulations coordinated by the Groupement were provided to the participants in the COM (124).

85. On 4 April 2002, in an internal note of CNCEP, it is explained that:

"The creation of an industrial and commercial company (SIC) [...] makes it possible to react to various threats, in particular the fear of powerful entry of partners already "CB" members, who would take market shares and discount prices" and that the fundamental principles are "a tariff [...] according to the level of participation in the activities" while the revenues "are brought up to the shareholders according to the held shares" (125). Another meeting of the COM was held on 12 April 2002. According to handwritten notes taken by Société Générale's representative on the COM, [a Groupement representative] stated that "we are making progress without calling into question the fundamental objectives" (126). These handwritten notes also refer to the request made by [a CNCE representative] for "priority to be given to the inertia criteria" (127). The meeting dealt among other things with "charging principles". The document introducing this item mentions that applying a fee for access "only to new activities makes it necessary to demonstrate that these contribute much less to the value of the CB system than the old ones" (128). The levels of charges, the links between the SIC and the GIE and the principles for apportioning capital were also reviewed (129).

86. According to a document dated 12 April 2002 ("CS 2002 project: Levels of charges") (130), "With the charging system, the main members will have strong protection against free-riding behaviour on the part of certain new entrants ... With the charging system, the main members will succeed in limiting the fall in their net banking income ... With the charging system, the impact of the new entrants on the main members' net banking income from new electronic banking business will be limited to 43% (instead of 80%)". The latter document confirms that the aim during this period was to work on the fine-tuning of the system in order to implement the policy decisions already taken: "The ongoing discussions with the representatives appointed by the eight founder banks are also aimed at ensuring that the levels of the fee for access reflect the policy decisions taken under the CS 2002 project".

6.2.5. Decision taken in May 2002 to give priority to the "non-structural aspects" of the project (131)

87. A presentation dated 22 May 2002 states that the annual cardholder fees for cards issued by the new entrants could vary between €15 and €20: "Given the present levels of the annual cardholder fee, a charge of €15-20 for universal payment cards issued by retailers appears reasonable" (132).

88. The next meeting of the COM took place on 29 May 2002. Several studies and analyses of the reform contemplated and of the SIC in particular were carried out between the COM meetings of 12 April and 29 May 2002. In view of the conclusions drawn from those studies, and more specifically the possibility that the SIC might not be acceptable under competition law, the main members sitting on the COM eventually decided to abandon the idea of setting up an SIC. Several notes written by representatives of the main members in April and May 2002 refer to the need to abandon the project of creating the SIC (133).

89. On 1 May 2002 the Groupement's [consultant], also recommended abandoning the creation of an SIC (134).

90. The conclusion having been reached that the SIC would be unacceptable under competition law, a new set of proposals "pursuing the same objective" (135) and not requiring any changes to the legal structure was developed in late May 2002.

91. This is apparent from a number of documents discovered during the inspections, such as an internal CE email dated 24 May 2002, according to which the second "defensive measure" successively proposed by the Groupement (the first one having been the SIC) involved "penalising, without changing the present legal structure, new entrants by introducing new 'facially' objective criteria at the level both of the CIP and of the GIE's apportionment formula". The "facially" objective character distinguishes the new direction taken by the SIC proposal, considered "too explicit (agreement between the founders the goal of which was not to share in economic progress and the improvement of the service)" (136), (137).

92. In May 2002 various documents drafted by the main members continued to refer to the foreseeable increase in the prices of bank cards offered by the new entrants as a result of the system of charges contemplated. For example, a presentation on progress made in the CS 2002 project (138) stressed that the new proposal (modulation of the CIP) enabled the price of cards offered by retailers to be increased: "The mechanism for modulating the CIP makes it possible on its own to levy around €7.50 per year on cards issued by the retailers ... The introduction of modulation of the CIP corresponds for the founders to a permanent annual gain of €7.50 per card issued by a retailer with no acquiring activity (or which does not acquire its own sales outlets)". Again, according to an internal Crédit Lyonnais communication dated 28 May 2002, "The new entrants will effectively be held back in their desire to engage in dumping, since the initial macro evaluation points to a shortfall of around €7.50 per card" (139).

6.2.6. The COM meeting on 29 May 2002: general agreement on the direction taken by the CS 2002 project

93. A further meeting of the COM was held on 29 May 2002 at which the main members gave their overall agreement to the direction taken by the CS 2002 project and at which new proposed measures were presented. According to the record of that meeting sent to the [...] members of the COM, the main members endorsed the direction taken by the CS 2002 project:

"[A Groupement representative] opened the meeting and thanked [...] for finding the time to enable him to outline recent developments with the CS 2002 project. After briefly reviewing the background, he said all the institutions had given general agreement to the course being taken, with the exception of [the representatives of the Crédit Mutuel and CIC in the COM], whom they had been unable to contact" (140).

6.2.7. Fine-tuning of the measures contemplated to ensure that new entrants were penalised but that none of the main members taking part in the COM would suffer (June-October 2002)

94. In line with what was agreed at the COM meeting on 29 May 2002, the participants in the COM examined the mechanism under consideration in greater depth by carrying out simulations and fine-tunings (141).

95. On 12 June 2002, in an internal Groupement email ("Progress report on the CS 2002 project") (142), the aim of "balancing acquiring/issuance", following the proposal put forward by [the consultant of the Groupement], appeared for the first time as a justification for the mechanism under consideration:

"(1) Formula for allocating voting and ownership rights and determining contributions: a dossier has been compiled by [...] and will be presented as it stands to the banks.

(2) Entry fees: a proposal drawn up by [the Groupement's consultant] (on which there have been discussions with [a Groupement representative]) will also be put to the banks. Its economic impact is attractive.

(3) Mechanism for balancing acquiring/issuance: different arrangements have been studied. We are working on finalising the proposal by [the Groupement's consultant] described in the attached document in order to include improvements".

96. The "attached document" referred to in the above-mentioned email from [...] was a memo from [the Groupement's consultant] to [representatives] of the Groupement bearing the same date (12 June 2002) in which [the Groupement's consultant] proposed "a system that penalises the members of the GIE whose relative acquiring/issuance activity is low and increases much less (or declines much more) than the average for the members" (143). [The Groupement's consultant] sketched out on that occasion the arrangement that was to become the Merfa.

97. Two days later, on 14 June 2002, the balancing arrangement was dubbed "mechanism for regulating the acquiring function" (144); the CS 2002 project then included the following measures (145):

"- Reform of the formula for allocating ownership and voting rights ...

- Mechanism for regulating the acquiring function: the work conducted with DAJB and [the Groupement's consultant] has made it possible to devise an arrangement that fulfils the objectives set by the COM.

- Increase in the entry fee: [the Groupement's consultant] proposes applying an entry fee by blocks of cards issued that is payable by any new member but also by any existing member whose activity suddenly becomes much greater than observed hitherto ('wake-up of a dormant member')".

98. According to one section in a "progress report memo" of 14 June 2002, "the justification of the mechanism for regulating the acquiring function:

- is based not on the need for a relative balance between issuance and acquiring

...

- ... but on the potentially free riding on the part of players developing issuance activity exclusively, on the understanding that development of issuance activity requires a far less significant commercial (and financial) effort than development of acquiring activity" (146).

As of 14 June 2002, the initial simulations on all the members had been carried out but needed to be supplemented (147). The first indications were nevertheless reassuring for the main members and their subsidiaries (148).

99. A further "progress review meeting" of the COM was held on 18 June 2002. Although the principles underlying the measures had been laid down, participants still had to finalise some of the arrangements for implementing them in the light of the simulations under way:

"Although the principles ... have now been laid down, some of the arrangements for implementing them still have to be finalised in the light of the consistency checks and simulations under way" (149).

100. At the same time, several of the banks' internal communications confirm the parties' growing satisfaction regarding:

- the effects that the measures would have in hindering card issuance by the new entrants;

- the fact that an appropriate justification had been found;

- the fact that all the main members taking part in meetings of the COM would be spared by the measures; and

- the fact that those of the eight participants in the COM which did not have a large acquiring activity (La Poste and Caisses d'Épargne) - but which would not be penalised on that account - would not turn towards acquiring, thereby competing with the other main members (150).

101. Here are some examples:

(a) According to an internal Crédit Lyonnais memo on the "CB Steering Committee meeting on 25 June 2002":

"The proposed mechanism meets the objective initially set, which was to reduce the cost to the founder members of any large-scale issuing of bank cards at a reduced price (the impact is about €10 per year and per card, compared with an amount of €66 over five years in the initial proposal) ... Crédit Lyonnais is not adversely affected by these measures; it would even benefit slightly: ....

It should be stressed that the regulating mechanism has been fine-tuned so that the main members are not penalised: La Poste and CNCE are therefore not affected despite the small scale of their acquiring activities (151).

(b) In an internal BNP Paribas memo in preparation for the COM meeting on 25 June 2002, it is noted that the Merfa, the entry fee and the dormant members wake-up mechanism would have effects on card price setting by the new entrants but no impact on any entity belonging to the group; it would even leave them "huge scope for expansion" in card issuance. It is also explained that the threshold for the Merfa has been calculated so as not to penalise La Poste and Caisses d'Épargne:

"I - Mechanism for regulating the acquiring function ...

- An initial simulation gives, for a 'pure' issuer, an average annual levy per card issued of €10 (on the basis of 50 transactions per card per year). That would represent 20% of the margin on a credit card (annual cardholder fee at €15; margin calculated after risk cost), and would have a favourable influence on the setting of annual cardholder fees for such cards.

- Given our positioning, there would be no impact on any entity belonging to the group: the BDF ratio is approximately 1, which leaves us a great deal of scope for potential expansion of the number of CB cards in circulation (Cetelem, for example).

- The pain threshold has been set at 5% because of the desire not to charge La Poste and Caisses d'Épargne. That is a very significant concession made to them.

- The impact on online banking is obvious.

As for large retailers, the choice of the number of merchant SIRENs as an index of acquiring activity significantly reduces their room for manœuvre (even Pass, Carrefour's acquirer, currently has few active SIRENs).

- As an initial approximation, CB estimates the amount thus levied at around €20 million ...

II - Entry and wake-up fees

... Inactive members that joined three years ago or more will be charged the same rates of entry fees as new entrants if they embark on a significant programme of card issuance. This measure is essential vis-à-vis large retailers, since the banks created by them have been CB members for a long time

... As it stands, this measure adds €5 per card up front to the issuing cost for new entrants and the like. Added to the measures in point I, the extra cost thus amounts to €15 per card in the first year, which is very significant. For us to be affected, the number of CB cards issued by us would have to exceed 16 million cards at group level over the next three years, which leaves us huge scope for expansion (152)."

(c) The following passage can be found in an internal Société Générale communication in preparation for the COM meeting on 25 June 2002:

"1/ FORMULAE

The reform intends to leave the power in the hands of the long-standing members vis-à-vis the new entrants ... Overall we can support this.

2/ ENTRY FEE

The proposed mechanism goes in the right direction since it makes the new entrants pay for past investments that they will use while still being acceptable under competition law (and, hopefully, to Brussels, which will have to be contacted) since it is non-confiscatory and can be explained.

According to the GIE, this fee would yield €15-20 million per year for the CB community; and, on the basis of official commercial prospects, Egg would have to pay €10 million.

A mechanism that is simple and difficult to evade.

3/ REGULATION OF THE ACQUIRING FUNCTION

The mechanism proposed here is justified by the idea that it is essentially the affiliation of new merchants that enables the CB system as a whole to be developed. It therefore involves making the new entrants pay on the basis of their issuing activity.

We have verified according to the figures supplied by the GIE that this mechanism:

- would be difficult to evade; if Auchan were to make Banque Accord act as its acquirer (as S2P does for Carrefour), it could not reduce the fees it pays as an issuer.

- would not have any undesirable effects. We would have run the risk of inciting the large banks which are more issuers than acquirers (such as La Poste and CE), to try to strengthen themselves in the acquiring of large turnovers and to increase further the level of competition on this market, which could really do without it. But to rebalance their position, these banks will try instead to affiliate small traders who count individually, in the mechanism, for as much as large retailers. In addition, the threshold for the application of this mechanism has been set at a level such that it exempts all the main members at the outset" (153).

102. The COM meeting on 25 June 2002 was a "progress review meeting". According to the "Summary report" (154), the COM "asked the Groupement to present to it by 19 July 2002 a final dossier for submission to the Board of Directors on 26 July 2002". According to that document, among the matters raised were "the issue of the SIRENs" and the need to "study in detail the collateral effects of the mechanism". During the meeting, [...] asked for a memo on ATMs: "The Electronic Banking Steering Committee requested a memo setting out the state of progress in the work on ATMs so that this issue could be discussed separately at the next meeting" (155).

103. The consensus reached on 25 June 2002 between the participants in the COM concerning the principles of the reform is evidenced by documents drawn up after the meeting. For example:

- A memo on the meeting on 25 June 2002 sent two days afterwards by the Crédit Lyonnais representative summarises the situation as follows:

"Atmosphere generally in favour of the new mechanism proposed by the GIE, namely:

- a system for "taxing" the activity of issuers that have no acquiring activity.

- a mechanism for charging membership fees that can be substantial for new entrants

The proceeds will be redistributed among the banks involved in acquiring in accordance with a formula based on the number of SIRENs for which they act as acquirers.

The technical arrangements still have to be worked out, as well as an additional legal validation from the standpoint of competition law". (156)

- According to an internal CNCEP communication:

"At its meeting on 25 June 2002 the Electronic Banking Steering Committee validated the proposal aimed at reacting to various threats, in particular the fear of the large-scale entry of partners who would capture shares of the card issuance market and cut prices" (157).

- According to a memo on a working meeting held on 12 July 2002:

"The three key principles presented and validated at the [...] meeting on 25 June were endorsed by all the banks" (158).

104. Between the COM meetings on 25 June and 19 July 2002, as is explained in the paragraphs which follow, work focused on the following:

- the risks of the system being circumvented;

- performance of simulations;

- definitions of active cards and active SIRENs (159);

- distribution of the proceeds;

- the legal soundness of the arrangement with regard to competition law;

-the impacts in terms of taxation of changing the formula for apportioning ownership rights.

105. In July 2002 some of the representatives of the main members sitting on the COM voiced their fears regarding what could be considered the anticompetitive nature of the reform. For example:

- On 2 July 2002 [...] Banques Populaires wrote to [...] the Groupement confirming his reservations regarding both "the economic justification for this project and, consequently, the legal soundness of the arguments put forward in defence of the measures" and "the effectiveness of this concept, which does not seem to me to be likely to correct the original imbalance created by the CIP in favour of the role of issuer". (160)

- In an internal La Poste email dated 11 July 2002 (161), one of the La Poste employees wrote:

"This plan seems to me, at first sight, to be anticompetitive and does not appear to me to correspond to a normal commercial approach to solving the problem ... Such a mechanism looks to me more like an eviction device than a conventional commercial strategy (incentive premium, marketing budget, etc.) aimed at encouraging the development of a customer base composed of affiliated merchants. If I have understood properly, this mechanism does not take account of the size of the affiliated merchants concerned (turnover in particular) or the number of card transactions carried out through them, only the number of affiliated merchants. Lastly, the other members of the GIE could partly benefit from the financial penalties imposed on the members that do not have sufficient affiliated merchants among their customers ...

For that reason, subject to additional information on the project, it is liable in my view to be considered a restrictive agreement between certain members of the GIE who wish to set this mechanism in place, to the detriment of other members. It is unlikely on the face of it that such a solution could convince the European Commission to exempt this agreement under Article 81(3) of the EC Treaty".

106. A preparatory meeting ahead of the COM meeting on 19 July 2002 was held on 12 July 2002, at which the legal risks and the economic justification were analysed. According to a working document: "It emerged clearly that the project was highly presentable and capable of being defended before the EU and national competition authorities and that any risks related more to the way in which the measures were deployed than to their principle" (162).

107. The economic justification that the parties had to provide in order to avoid risks was set out in a record of the meeting on 12 July 2002 drawn up by La Poste (163). It was agreed at that meeting that "CB, via Sofres, will carry out a survey of merchants in order to consolidate the argument concerning development of the acquiring market". It was also agreed on 12 July 2002 that "[the Groupement's consultant] must supplement his paper with answers to the questions raised by [the adviser] and the banks, with special reference to the arguments relating to competition" (164).

108. At the meeting of the COM on 25 June 2002 those present discussed among other things (a) the distribution of the proceeds of the MERFA (165), (b) the risks of circumvention by the "targeted issuers" (on the subject of which an ad hoc meeting was to be held on 17 July 2002 (166)) and (c) the definition of active cards and active SIRENs (167). A complete dossier and simulations were to be sent to the participants in the COM ahead of their meeting on 19 July 2002 (168).

109. The record of the meeting on 12 July 2002 (169) drawn up by La Poste confirms that there was a broad consensus among the parties concerning the principles, even if doubts remained as regards the concrete measures, including the Merfa:

"1. Round-up of views

CA (Crédit Agricole): In favour of the project, which nevertheless requires a number of clarifications concerning the justification for certain choices and other practical implementing arrangements.

BP (Natexis Banques Populaires): In favour of the new entry fees and voting rights. Strong reservations on the Merfa, for which they do not see the economic justification ... To favour the acquiring function, they would have preferred an overhaul of the CIP.

CL (Crédit Lyonnais): Satisfied with the direction taken by the project. Aspects to do with the practical arrangements and the justification for the thresholds must nevertheless be examined further.

CM (Crédit Mutuel): Satisfied with the direction taken by the project ... Like BP, CM finds the Merfa 'risky both economically and legally'.

SG (Société Générale): No objections to the objectives pursued. Is nevertheless doubtful about the Merfa and its method of allocation (CB/acquirers), which cannot in its view be kept as it stands. The incentive mechanism thus devised and shared between operators in a dominant position could cause problems in the courts. It proposes that the Merfa should not be based on the number of SIRENs but on the evolution of flows of SIRENs.

La Poste: In favour of the project. Two aspects should now be examined further: the active implementing arrangements and developing the line of argument concerning the risk inherent in the project from the standpoint of competition.

CE (Caisses d'Épargne): In favour of the project, but has reservations on the method of allocating proceeds. Endorses SG's proposal that flows rather than the numbers of SIRENs should be taken into account. The risks and potential penalties under competition law should also be clarified.

BNPP (BNP Paribas): The proposals are consistent with the CEOs' objectives. Is nevertheless doubtful about the legal soundness with regard to the competition rules".

110. Ahead of the COM meeting on 19 July 2002, the banks taking part in it continued to express favourable reactions. For example:

(a) In an internal CNCEP memo in preparation for the meeting on 19 July 2002, it was noted that the measures contemplated made it possible to prevent the fall in the price of cards and safeguard the main members' incomes:

"An impact study on these measures shows that the new entry fees and the mechanism for regulating the acquiring function will reduce by 25% the gains expected by the large retailers through card issuance. This reduction in the expected gains would reduce the effect of the discount on the prices of these new entrants, and consequently the contamination of our own prices.

The loss of net banking income for the founder members, in the event of a large-scale entry of new players, would be reduced by one third by these mechanisms ...

In conclusion, these new measures appear to contribute to the objectives pursued" (170).

(b) Another internal CNCEP communication points to the growing number of ATMs and the need for pricing criteria deterring new entrants (banking arms of large retailers) from installing ATMs, something which is the exact opposite of the objective of the Merfa declared to the Commission:

"Given the Electronic Banking Steering Committee's finding in 2001 that the number of ATMs was rising and that the CB members needed to make their investments profitable, on 15 March 2002 the Board of Directors instructed the Groupement to set up an expert group to examine a system of differentiated charges based on relevant and auditable criteria for the eligibility of ATMs.

The aim of this system of charges was to deter large groups, particularly large retailers, from installing large numbers of ATMs in public places such as shopping malls" (171).

111. A meeting on "the risks of circumvention of the Merfa by certain banks" (172) took place on 17 July 2002 on the Groupement's premises and was attended by representatives of the Groupement, Crédit Agricole, Natexis and Société Générale. According to a memo on that meeting, the fine-tuning of the Merfa - to avoid La Poste and Caisses d'Épargne being subject to it - had the undesirable effect that the threshold was too low and could be easily attained by the banking arms of large retailers:

"2. Internal risk: linked to the SIRENs

On the basis of the lowest activity ratio (that of La Poste: [...] SIRENs/[...]cards) the GIE has calculated the level to be reached for a candidate seeking to escape the Merfa.

For S2P, for example, which would like to issue between [...] and [...] million cards, it would have to have [...] SIRENs.

This threshold appears to CA and SG to be very low and on the face of it easy to attain ... The GIE is launching research to try to determine the number of SIRENs per large retailer.

An alternative: raising this threshold of [...] SIRENs/[...] cards would mean making La Poste pay the Merfa ... unless a means can be imagined of increasing the number of its active SIRENs" (173).

112. At the next meeting of the COM, on 19 July 2002:

"[...] ask for the number of ATMs to be included in the regulating mechanism. The Groupement will therefore draw up fresh proposals along these lines for the next meeting" (174).

113. The reason for introducing ATMs into the calculation of the Merfa is clear when one reads the comments made by the main members sitting on the COM. For example, according to an internal Société Générale communication dated 24 July 2002 the introduction of ATMs into the calculation of the Merfa would contribute to "inertia" for the new entrants (175).

114. The need to introduce the ATM criterion as an additional feature led to the formal adoption of the measures being postponed, "[T]he aim being to present a new project to the CEOs in early September" (176).

115. The CS 2002 project was briefly discussed for the first time at the meeting of the Board of Directors of the Groupement on 26 July 2002, which was attended by all the main members apart from CCF (177).

116. Between late July and September 2002, several bilateral meetings with the Groupement took place to discuss the preparation of the measures:

"Late July/August/September

Improvements were to be requested and then discussed with the members of the Board of Directors for validation and finalisation, always by means of bilateral meetings, and a working session to finalise the measures was scheduled for 4 October 2002" (178).

117. An internal BNP Paribas memo dated 13 August 2002 (179), after noting that "The two figure growth in electronic banking flows has kept up over the last ten or so years", reiterates the aim of the measures:

"New entrants

• Today the CB system - networks of merchants and ATMs - offers free ride access (entry cost around €150 000) to potential new entrants.

• The potential new entrants are well-known: credit card issuers from the English-speaking countries, banking arms of large retailers, online banks, telecom operators.

• The main risk identified is large-scale entry of these operators into our market together with deep discounting of the cardholder fee: ...

• However, the profitability of the operating account for electronic banking in France is 50% dependent on cardholder fees: ...

- Any threat to cardholder fees constitutes a major risk of decline in net banking income and profitability in electronic banking.

• The main response at the present time is the ban on co-branding, which chiefly hinders the English-speaking issuers in their strategy of entering into alliances with non-banks in order to penetrate the French market (but Air France has got round the ban by signing with American Express).

• This response is deemed insufficient with regard to the other potential entrants. Hence the CB project aimed at:

- strengthening the political alliance between founder members

- increasing the costs of access to the CB system (entry fee, enhancement of the merchant and ATM acquiring function, hiving off the technical network to a subsidiary)".

118. A "CS 2002 meeting" took place on 13 September 2002 at the premises of the Groupement to present the state of progress on the CS 2002 project. The participants agreed to redistribute the proceeds from the Merfa among the acquirers in proportion to their contribution to acquiring. They noted that certain aspects still had to be clarified, in particular the criteria for "eligible ATMs", the integration of interbank transactions into the calculation of the Merfa and the risks of circumvention: "this involves studying the possible scenarios in which a new entrant could avoid being subject to these different mechanisms: cross-border issuance, etc." (180).

119. The participants in the meeting on 13 September 2002 discussed a "working document". This was a draft motion for submission to the "Board of Directors meeting on XX October 2002" deciding to implement three measures (replacement of the single formula laid down in the Formation Agreement by three new formulae (181), modification of the system of membership fees and creation of a "mechanism for regulating the acquiring function") and instructing the Groupement to prepare the ground for their introduction and implementation so that they would enter into force by 1 January 2003 at the latest. This document envisages a number of justifications in support of the future decision of the Board of Directors to introduce the above measures, such as the "positive externalities" concept and the need to "create for members that are card issuers an incentive to develop their acquiring activities so as to continue to contribute to the increase in the value of the system" (182). These justifications constitute a draft of what was ultimately to become the notification of the measures to the Commission.

120. These justifications contrast with the real concerns expressed by the participants at the meeting in late September 2002. For example, a document dated 17 September 2002 (183) on the strategy adopted by large retailers reveals that the reform contemplated was prompted by a concern to counter the negative impact of the entry of new members from the large retailer sector:

"CS 2002 assumption: 2 million CB credit/debit cards issued by the French retailers in two years at a 50% discount

(assuming a total of 4.5 million CB cards issued by all players)

A potential loss of €140 million in net banking income for the traditional banks Factors impacting on the net banking income of the traditional banks:

- Drop in the volume of issuance (2 million cards issued by retailers)

- Average discount of 25% on the price of new cards issued and on 1.5% of the number of cards in circulation

- Activity 'on us' of the retailers' cards amounting to €660 per card per year.

Challenges for the retailers: Reduce the commissions paid to the banks. Certain retailers currently achieve 10-20% of their turnover via their proprietary payment cards and believe that they can double that figure in the medium term

...

Benchmarking the retailers: Pricing of payment cards. Given the present levels of the annual cardholder fee, a charge of €15-20 for universal payment cards issued by retailers appears reasonable".

121. A "general overview" dated 25 September 2002 (184) stresses the consequences of the issuance by new entrants of significant numbers of low-cost cards for the incomes of the founders and for profits from electronic banking, and explains that the reform under consideration would have the effect of limiting the reduction in the price of cards proposed by new entrants:

"(1) Threats and challenges

1.1. Large retailers

- early warning signs (presentation of visual appearance of cards, press statements, installation of ATMs, etc.)

- the objective reasons for entering the card issuance market: development of loyalty cards, consumer credit, the policy of positioning mass-market financial products on the market

- the discount policy, which is shifting the price reference: the CB/Visa card at €15 or €16 and not at €30-40

- the possibility of finding offers of services on the market (subcontracting, etc.) at knock-down prices

- the banking arms of large retailers are already CB members.

1.2. Online banks

...

1.3. Telecoms operators

...

1.4. Consequences in terms of the operating account for electronic banking

The irruption of issuers of cards at discount prices would have far-reaching effects on the incomes of the founders of the CB system:

- direct loss resulting from customer transfers (card commissions affected by new entrants)

- related losses on derived incomes (interchange fees, etc.)

- fall in card commissions on new customers (need to align commercial terms)

- discount on commission to be granted in order to keep long-standing customers

It has thus been estimated that the issuance of 4.5 million cards in two years by large retailers (10% of the number in circulation) could lead to a loss of €140 million in income for the founders over those two years".

122. This general overview analyses the impact of the Merfa on new entrants and considers that new entrants would find it extremely difficult to avoid the Merfa; The Merfa will result for them in an "additional cost per card of around €4, which is important for the price reference for cardholder fees". "The impact is above all on the investments the 'predators' will have to make, since it can be expected that in the long run they will be able to install ATMs and make their investments profitable (although the development of charges for foreign withdrawals makes this more difficult).

But launching an ambitious ATM programme makes it necessary both to raise extensive financial resources and to find suitable locations for installing them" (185).

123. Simulations of the effects of the Merfa continued to be carried out in September 2002 (186) with a view to ensuring that the main members would not be penalized (187).

124. An internal Natexis-Banques Populaires memo dated 8 October 2002 contains a summary of the positions taken by the participants in the COM on the measures that were to be adopted on 11 October 2002 (188). It was explained at the meeting that the introduction of ATMs into the calculation of the Merfa strengthened the effectiveness of the mechanism: "Introducing the ATMs, while apparently not penalising the Group (Banques Populaires), raises the liability threshold and therefore strengthens the effectiveness of the mechanism whereas, before, banks like La Poste or CE required a very low threshold so as not to be liable" (189). It was also explained that the Merfa had a "positive impact" for the banking community since it limited the loss of income resulting from card issuance by new entrants:

"3.2. A positive impact for the banking community

The Merfa acts as a brake on the loss of income if the new entrants issue cards

...

According to these estimates, if the new entrants were to issue 5 million cards over five years and install ATMs so as to avoid overall paying the CSR to another bank, the losses of the founder banks would be €71 million rather than €1 011 million without the Merfa.

If the new entrants were to install larger numbers of ATMs in order to avoid paying the Merfa, the banks' losses would be €906 million ...

It is acknowledged by the Committee of Experts that the proposed mechanism contains the seeds of a risk to the founder banks. Introducing the number of ATMs into the calculation may encourage the new entrants to develop their pool, among other things in order to avoid paying the CSR.

This measure is aimed at discouraging the new entrants from going any great distance down this path and is designed to encourage them to fall back if necessary on other institutions that would then be regarded as comparatively more profitable (and in particular on new establishments abroad)" (190).

125. It can be seen from a number of documents obtained during the inspections that [...] were aware of the penalising effect of the measures in question before they were adopted at the COM meeting on 11 October 2002. For example:

(a) An internal CNCEP memo preparing the ground for the meeting of 11 October 2002 on the CS 2002 project notes that the new entrants would have to make a huge acquiring effort in order to escape the measures' application. The risk of the measures being circumvented through direct card issuance via Visa and the need to be vigilant with regard to other types of "circumvention" (co-branding and/or agreements with Amex-type issuers) was also identified:

"All the principles have been endorsed by all the banks.

GCB has carried out a number of simulations, demonstrating that, to escape the Merfa, new entrants will have to make very substantial acquiring efforts (ATMs in particular).

A study has also been carried out on the risk of circumvention of the system through the issue of cards direct via the Visa network: the study shows that the costs involved in the volume of transactions shifted and in fraud would in that case be extremely high.

In conclusion, these new measures appear to contribute to the objectives pursued. Their legitimacy vis-à-vis the competition authorities will, of course, have to be checked. These measures will only be able to hold back the arrival of new entrants: ...

The implementation of these measures will require extreme vigilance regarding the circumvention systems it is liable to lead to:

- pressure on co-branding

- agreement with AMEX-type issuers, which are thought to have negotiated bilateral acceptance contracts" (191).

(b) an internal Société Générale memo dated 9 October 2002 reviews the different aspects of the measures:

"1.2. The new entry fee

It is based on a lump sum of €50 000 (instead of €38 000 at present) plus a fee of €12 per card issued during the three years following the date of entry. It is supplemented by a complex, but essential, mechanism penalising any existing dormant members that start to issue massively ...

Agreement on this proposal.

1.3. The Merfa

...

1.3.1. SIRENs

and not SIRETs, since the latter criterion would facilitate circumvention of the Merfa. The large retailers could easily increase their number of acquiring contracts based on the concept of SIRET premises.

1.3.2. 'Approved' ATMs

according to criteria complying not only with the security decree but also ... with CB security requirements ... and with requirements ... relating to the service provided ...

These criteria are not crippling for large retailers. But they undoubtedly increase the cost, for them, of installed ATMs ...

Agreement on these criteria"

126. Ahead of the COM meeting on 11 October 2002, the Groupement carried out simulations of the effects of the measures, with special reference to the price of cards. The aim of the simulations was both to check that new entrants would indeed be prevented from offering cards at prices appreciably lower than those charged by the large banks, hampered as they would be by the new measures, and to ensure that participants in the COM would not be penalised. For example:

(a) In the simulations annexed to a Natexis Banques Populaires communication, the Groupement forecast that the price of bank cards could fall to as low as €15 without charging. It also estimated that with the Merfa the cardholder fee would be €20 (193).

(b) A conclusion on the "results of the simulations and comments", annexed to the above-mentioned document, contains simulations of the losses or gains for the main members resulting from the development of acquiring activities by new entrants with or without the Merfa. The conclusion is that "in all cases the Merfa minimises the loss for the main members!"

6.2.8. The COM meeting on 11 October 2002

127. At the COM meeting on 11 October 2002, the [...] the main members attending accepted the measures forming part of the CS 2002 project, namely the Merfa, the new membership fee and the reform of the apportionment formulae. According to the summary report on the meeting:

"The entire dossier is accepted:

1. As regards ATMs, the Groupement is to provide a precise wording of the definition of CB-approved ATMs to be used in calculating the Merfa.

2. The reform of the apportionment formulae is accepted. The base of the moving average over seven years is adopted for the voting formula.

3. The membership fee is accepted.

4. The Merfa is accepted, it being suggested that the fee per card be set at €11" (194).

128. The handwritten notes taken by [a] Groupement [representative] on the meeting of 11 October 2002 show that the project was "adopted unanimously without any reservations" (195).

129. The work on the definition of approved ATMs continued in October 2002. This is established by several documents drafted by the main members, for example an internal BNP Paribas communication welcoming the definition of approved ATMs since it excluded the ATMs operated by large retailers: "The definition of CB approved ATMs is acceptable: it makes it possible to exclude the 'lightweight' ATMs that are widespread in large retail outlets" (196).

6.2.9. The formal adoption of the pricing measures in question by the Board of Directors (decisions of 8 and 29 November 2002)

130. [T]he COM [...] is not a body provided for in the Groupement's Formation Agreement. Only the Board of Directors has the authority formally to adopt pricing measures (and changes to the Rules of Procedure). The amendments to the Formation Agreement introducing the non-price measures had to be adopted by a decision of the General Meeting.

131. According to the minutes of the Board of Directors meeting on 8 November 2002:

"The Board of Directors unanimously approved the following measures:

1. Replacement of the existing single formula laid down in the Formation Agreement by three formulae determining:

- contributions to the Groupement's operating costs (calls for funds) and members' liability towards third parties

- voting rights within the Board of Directors of the Groupement

- allocations in the event of liquidation, winding-up or the distribution of assets or the acquiring of holdings by the Groupement, and established in accordance with the rules described in the dossier submitted to the Board of Directors;

2. Modification of the system of membership fees for joining the Groupement, as proposed in the dossier;

3. Creation of a 'mechanism for regulating the acquiring function' as described in the dossier submitted to the Board of Directors" (197).

132. According to the same document:

"The Board of Directors requested the Groupement to prepare for and implement the application of the above described measures at the earliest opportunity ... These amendments were to be brought to the attention of the European Commission. Measures 1 and 2 [formulae and membership fees] were to be operational by 1 January 2003 at the latest, while the creation of a mechanism for regulating the acquiring function (Merfa) was to apply forthwith (i.e. it was to be based on 2002 data).

The Board of Directors requested the Groupement to prepare all the documentation explaining these measures for all CB members and to distribute it among the latter at the earliest opportunity".

133. While confidentiality prevailed during the preparation and fine-tuning of the measures (198), once adopted they had to be rapidly communicated to all the members of the Groupement. On 13 November 2002, an information memo to CB members was sent by the Groupement to the main members (199).

134. Another meeting of the Board of Directors was held on 29 November 2002. There were two items on the agenda:

(a) Approval of a change in the dormant members wake-up mechanism. The Groupement had ascertained that the data required for applying this mechanism could not be collected as from the date certain members had joined. It was accordingly decided to limit the reference period for the data to be collected in order to gauge the activity of members liable to be classed as dormant to the years 2000, 2001 and 2002 (200). It was also decided to allow banks that joined the Groupement in 2002 to choose between paying the new membership fee (as if they had joined on 1 January 2002) or paying the dormant members wake-up fee (201).

(b) Approval of the amendments to the Formation Agreement that were necessary in order to implement the non-pricing measures. These amendments were to be submitted to the General Meeting on 20 December 2002.

135. The above clearly demonstrates that the pricing measures in question were prepared by the COM, then formally adopted by the Board of Directors, without secondary members (in particular the new entrants) being able take part in their preparation or even being consulted beforehand.

136. Between November 2002 and January 2003, several internal communications between staff of the main members sitting on the COM noted that new entrants would indeed be penalised by the measures and could not escape them, or "or it will cost them dearly" (202). Encouraging acquiring by new entrants appeared to be of secondary importance, the satisfaction expressed by the members sitting on the COM and on the Board of Directors being directed above all at the fact that the measures should make new entrants less competitive. For example:

(a) On 5 November 2002, a BNP Paribas [representative] wrote: "I confirm to you that Egg would indeed be penalised by the new CB rules ... Egg would have to pay €11 per card issued over the first three years, not to mention the commission on cash withdrawals (€1.66 per withdrawal + remuneration of the cash advance +/- €0.8) if more than a third of their transactions are the subject of an authorisation request. In conclusion, their business plan is likely to be more than tight" (203).

(b) On 9 December 2002, Egg met the Groupement's management and subsequently sought explanations from its main member, Société Générale. According to an email from [...] Société Générale:

"Egg (level N-1) invited us to lunch ... to present to us its objections:

1/ We, as main member, should have informed them in advance of what was brewing to enable them to incorporate these new facts into their business plan.

2/ The immediate application of the adopted measures is virtually retroactive for Egg.

3/ The basis for the measures favouring an acquirer to the detriment of an issuer (who also helps to add value to the CB system) is questionable.

On 1/ and 2/, we replied that there was nothing new, that it was not usual practice to open up a discussion with secondary members on such a complex topic ...

On 3/, after reiterating the principle of balanced contributions to interbanking that underpins the rules and regulations governing CB, we called on them to consider the possibility of embarking on an acquirer activity basing themselves on computer service companies in the same way as our other secondary member acquirers.

That gave them something to think about and comforted them a little. Wrongly, as, even with the help of Experian or First Data, it will not be easy for them to be competitive in the eyes of merchants or it will cost them dearly" (204).

(c) An internal Crédit Lyonnais email dated 12 November 2002 comments on the impact of the measures on certain new entrants:

"Of the three arrangements, two have short-term impacts on S2P, Egg and Banque Accord:

1. a new membership fee is introduced from 1 January 2003 (€50 000 plus €12 per card); although S2P is not a new entrant, a wake-up mechanism is also planned for members that suddenly engage in significant issuance activity;

(the calculation for S2P is in hand)

2. a mechanism designed to compensate for any imbalance between acquiring and issuance;

where acquiring activity is non-existent (as in the case of Egg), the annual fee is set at €11 per withdrawal and payment card;

as long as the bank's acquiring activity is lower than its issuing activity, the annual fee is charged in proportion, and it is cancelled out where the two activities are in balance.

(the calculation for S2P is in hand)" (205).

(d) An internal Société Générale memo dating from January 2003 stated that the simulations carried out by the Groupement confirmed that the measures adopted would penalise new entrants:

"1. INTRODUCTION OF THE Merfa

... With a few simple data, the GIE has carried out simulations for the retailers and Egg. The situations are as follows:

• Carrefour: with the pool of ATMs already installed, Carrefour has a cardissuing capacity of 1.2 million units.

• Auchan: does not have any ATMs. Issuing 100 000 cards in 2003 and 100 000 in 2004 would 'cost' €4.6 million.

• Casino: with 200 ATMs, Casino apparently has the capacity to issue 800 000 cards.

• Egg: assuming 200 000 cards sold in 2003 and 400 000 in 2004, the 'cost' over the two years would be €11.4 million. Egg reacted strongly and was obviously not in the picture, therefore has to redo its business plan.

- NB: Axa made strong representations to BNP after the sale of Banque Directe and BNPP had to make valuation concessions.

- the Leclerc group, not much interested in cards, is satisfied with the system, which holds back untimely initiatives". (206)

(e) In a memo on the record of the COM meeting held on 7 January 2003, the CNCEP representative expressed satisfaction at the penalising effect the new measures would have on Egg: "the change in the dormant members wake-up mechanism is effective: Egg would have to pay €11.4 million in 2003-04 with the new system, which should change its business plan" (207).

7. THE NOTIFIED MEASURES

7.1. Introduction - The "principles underlying" the notified measures

137. On 10 December 2002, the Groupement notified to the Commission the new pricing measures of the CB system adopted by the Board of Directors of the Groupement on 8 and 29 November 2002 together with the amendments to the Formation Agreement that were to be adopted by the General Meeting of the Groupement on 20 December 2002. These measures, which entered into force on 1 January 2003, established:

- a "mechanism for regulating the acquiring function" (the "Merfa") (see section 7.2 below),

- a reform of the Groupement membership fee applicable to new members only (the membership fee as such and an additional membership fee) (see section 7.3),

- "as a measure accompanying the membership fee" (208), a fee labelled the "dormant members wake-up mechanism" (see section 7.4) and

- a new formula governing the distribution of voting rights among Groupement members, the sharing between them of rights to Groupement assets and the distribution of the proceeds from membership and dormant-member fees (section 7.5).

138. The section of the Groupement's notification concerning the "principles underlying" the notified measures is reproduced in full below:

"The principles underlying the notified measures

(22) The actual contributions from the members of the Groupement are calculated taking into account the fact that the value of the CB system, like that of any payment system, is greater than the sum of the investments made by each of those taking part: there is here a network effect, recognised by economists under the name of 'positive externalities', which is the fruit of the cooperation between members.

(23) In the CB system as it presently stands the acquiring function generates more 'positive externalities' than the issuance function: widening acceptance of the CB card, especially in sectors where it is underdeveloped, is likely in return to attract new types of CB cardholder or to encourage more intensive use of existing CB cards, thereby producing such 'positive externalities'; similarly, enhancing the security of ATMs will likely allay the fears of some potential users of the CB card. On the other hand, increasing the already high number of CB cardholders is unlikely to induce merchants who are currently not affiliated to the CB system to sign up or banks to open up their ATMs to CB interbanking".

7.2. Mechanism for regulating the acquiring function (Merfa) (209)

139. According to the notification, the Merfa's purpose is twofold: (i) to encourage those members of the Groupement that are more issuers than acquirers to develop their acquiring activities so as to make a balanced contribution to the increase in the value of the CB system; and (ii) to take into account financially the efforts of members with a large acquiring activity in comparison with their issuing activity (210).

140. For the purposes of the Merfa, a bank is defined as "any group composed of a member of the Groupement and subsidiaries in whose capital it owns at least a 51% stake". Banking groups include, in addition to subsidiaries, entities belonging to a central body within the meaning of the Monetary and Financial Code (211).

141. Banks whose relative acquiring activity is considerably lower than (less than 50% of) their relative issuing activity will be subject to the MERFA (212).

142. For each bank (i), two ratios reflecting its contribution to acquiring activities will be calculated each year:

- the bank's share of ATM-acquiring activity throughout the system:

R (i) = (number of active type-approved ATMs of bank (i) / total number of active type-approved ATMs);

- the bank's share of merchant-acquiring activity throughout the system:

S (i) = (number of active SIRENs (213) of bank (i) / total number of active SIRENs).

From these two ratios, an overall index will be calculated for the contribution to acquiring activity:

A (i) = 1/2 x (R(i)+S(i))

The bank's share of CB card issuances throughout the system will also be calculated:

E (i) = (number of active CB cards (214) of bank i / total number of active CB cards).

The following formula will then be calculated: T (i) = A (i) / E (i)

Any bank for which T (i) < 1/2 will be liable for the Merfa in respect of a given year (215).

143. Banks subject to the Merfa will have to pay a levy of between €0 and €11 per payment and cash withdrawal card and between €0 and €3 per cash withdrawal card, according to the imbalance between their relative weight as acquirers and their relative weight as issuers:

P(i) = 1 - 2 x T (i)

The amounts levied per card will be:

- per active payment and cash withdrawal card: P (i) x €11 before tax;

- per active cash withdrawal card: P (i) x €3 before tax.

144. The proceeds of the Merfa will be redistributed among the members of the Groupement that are not subject to it, in proportion to each member's overall index for its contribution to the acquiring activity (216); amounts received under the Merfa can be used as the recipients wish (217).

7.3. Reform of the membership fee (218)

145. According to the Groupement, the aim of the reform of the membership fee is to take greater account of the investments and efforts made by the members of the Groupement and the immediate advantages afforded to new entrants as a result of those investments and efforts. It also takes account of new entrants' actual use of the CB system by modulating the fee they have to pay according to the volume of cards they issue.

146. The reform includes:

- the membership fee as such, consisting of:

- an increase in the fixed component of the membership fee, for all new entrants, from €38 000 to €50 000 before tax; and

- the introduction of a fee of €12 before tax per active card issued during the first three years after joining the Groupement, based on the difference between the number of CB cards in circulation at the beginning and end of the year in question;

as well as, if necessary, an additional membership fee: where the number of cards in stock during or at the end of the sixth year is more than three times the number of cards in circulation at the end of the third year, the member concerned must pay an additional fee of €12 before tax per new card (fee per card over the number which the new entrant would have had to have issued so that the number of cards at the end of the sixth year is not more than three times the number of cards at the end of the third year: €12 x 1/3 "excessive" cards) (219). The fee to be paid is based on one third of the difference between the two figures (220).

7.4. Dormant members wake-up mechanism (221)

147. According to the Groupement, "In order to eliminate any discrimination between new entrants and 'dormant members' (i.e. members who joined the Groupement prior to implementation of the new arrangements without developing any significant CB activity), such dormant members will be charged an equivalent fee of €12 per card" (222). The arrangement applies to dormant members whose share in the issuance activity of the whole "CB" system during one of the years 2003, 2004 and 2005 (223) is more than three times greater than their share in the overall "CB" activity (issuance + acquiring) of the whole "CB" system during one of the years 2000, 2001 or 2002. To determine whether or not this is the case, the following nine ratios must be calculated:

A. Ratios for active cards in each of the years 2000, 2001 and 2002:

A2000 = member's number of active cards at the end of 2000 / number of active cards in the CB system at the end of 2000.

A2001 = member's number of active cards at the end of 2001 / number of active cards in the CB system at the end of 2001.

A2002 = member's number of active cards at the end of 2002 / number of active cards in the CB system at the end of 2002.

B. Ratios for active ATMs in each of the years 2000, 2001 and 2002:

B2000 = member's number of active ATMs at the end of 2000 / number of active ATMs in the CB system at the end of 2000.

B2001 = member's number of active ATMs at the end of 2001 / number of active ATMs in the CB system at the end of 2001.

B2002 = member's number of active ATMs at the end of 2002 / number of active ATMs in the CB system at the end of 2002.

C. Ratios for active SIRENs in each of the years 2000, 2001 and 2002:

C2000 = member's number of active SIRENs at the end of 2000 / number of active SIRENs in the CB system at the end of 2000.

C2001 = member's number of active SIRENs at the end of 2001 / number of active SIRENs in the CB system at the end of 2001.

C2002 = member's number of active SIRENs at the end of 2002 / number of active SIRENs in the CB system at the end of 2002.

148. The highest of these nine ratios is called "Mi". In addition, the member's share of card issuance activity throughout the system during each financial year 2003, 2004 and 2005 is calculated:

P2003 = member's number of active cards at the end of 2003 / number of active cards in the CB system at the end of 2003.

P2004 = member's number of active cards at the end of 2004 / number of active cards in the CB system at the end of 2004.

P2005 = member's number of active cards at the end of 2005 / number of active cards in the CB system at the end of 2005.

149. Members for which any one of the P2003, P2004 or P2005 indices is more than three times the value of the Mi ratio will be regarded as "dormant members". The member in question will pay the fee once only in respect of the first year in which its P(i) ratio is more than three times higher than its Mi ratio (224).

150. The dormant members wake-up fee is calculated as follows

- At the end of each year, the number of cards that can be issued without paying this fee (Ei) is calculated:

Ei = 3 x Mi x number of active cards in the CB system at the end of the same year

- Where, at the end of one of those years, the number of active cards of member I exceeds Ei, the member will have to pay a fee equal to:

€12 x (member's number of active cards at the end of the year in question - Ei) In the case of members that joined the Groupement during the course of 2002, the indices for 2002 will be calculated in relation to the total number of cards, ATMs and SIRENs in the CB system, reduced in proportion to the number of full months in which they were CB members.

Members that joined the Groupement during the course of 2002 have the option of asking to be treated as new entrants rather than possibly being subject to the dormant members wake-up arrangements (225).

151. Although, according to the Groupement, the proceeds from the membership fees and dormant members wake-up fees were not intended to be redistributed among the members immediately,, they were meant to be distributed subsequently. Members were then free to use as they saw fit the amount they received as their share, which was allocated according to the formula for distributing members' rights to the Groupement's assets (226).

7.5. New formula governing the distribution of voting rights among the members, rights to the Groupement's assets and proceeds from membership and dormant members wake-up fees (227)

152. Prior to the changes that entered into force on 1 January 2003, the same formula was used to determine members' contributions to the costs of the Groupement, their voting rights in its decision-making structures and their liability towards third parties.

153. As of 1 January 2003, this single formula is to be used only for calls for funds to be paid into the Groupement to cover its annual operating expenses and investments and for apportioning liability (e.g. for any debts following liquidation of the Groupement).

154. The new formula is based on the average number of transactions over the last seven years, irrespective of the date on which each of the members joined (228). It will be used to calculate (1) the number of voting rights on the Board of Directors and at the General Meeting and (2) members' rights to Groupement assets and in the distribution of the proceeds from membership and dormant-member wake-up fees among the members.

8. ADVANCEMENT OF THE CS 2002 PROJECT FOLLOWING SUBMISSION OF THE NOTIFICATION TO THE COMMISSION

155. Following the notification to the European Commission on 10 December 2002, the amendments to the Formation Agreement - and consequently the adoption of the nonprice measures (new formulae governing distribution of voting rights among the members, rights to Groupement assets and proceeds from membership and dormant member fees) - were unanimously approved by the General Meeting of 20 December 2002, with one abstention (Banque Accord) (229). Although the non-price measures were formally approved by all Groupement members, they, like the pricing measures adopted by the Board of Directors, were prepared by the main members (see section 6 on the COM's preparation of the measures) without the participation of secondary members and without their being informed.

156. The background to this approval by the General Meeting is as follows:

(a) 17 members were present at the General Meeting: the 11 main members, two institutional representatives (from the French national bank and the tax authorities) and only four non-main members (namely CPR Billets, a Crédit Agricole subsidiary; Cofinoga; Confinoga's subsidiary Sygma Banque; and Banque Accord). The main members represented 72 secondary members, and 85 members were neither present nor represented (230).

(b) The documentation transmitted to the General Meeting makes no mention of the link between the amendments to the Formation Agreement submitted to that meeting and the pricing measures already adopted (231), although the main members had received from the Groupement on 13 November 2002 a memo intended to be used to inform secondary members of the measures. This lack of any explanation at the General Meeting is all the more remarkable since, during the preparatory work, the pricing measures were the centre of attention.

157. In January 2003, two technical amendments were made to the measures adopted by the Groupement's decision-making bodies. According to the file on preparations for the COM meeting on 7 January 2003, the entry into force of the Merfa was postponed from 1 January 2002 to 1 January 2003 (i.e., the Merfa was to be paid in 2004 on the basis of the data for 2003). The method for calculating the dormant members wake-up fee was also changed. The calculation was henceforth to be carried out as follows:

"A study will be made of the relative positions of a member as issuer (its number of cards/total number of cards), acquirer (its number of SIRENs/total number of SIRENs) and manager of ATMs (its number of ATMs/total number of ATMs) at the end of 2000, 2001 and 2002, and the best of these nine indices will be selected.

If at the end of the following three years (2003, 2004 or 2005) its ratio of cards issued to the total number of cards in circulation is more than three times the index selected, then the institution will pay €12 per card for the 'surplus' cards (and not on all the cards it has in circulation)" (232).

158. In addition, members joining in 2002 had the choice between being considered new entrants or dormant members (233).

159. A study by [the Groupement's consultant] on the economic justification for the measures was sent to the Commission in March 2003. Prior to this, [a Groupement representative] exchanged a number of emails with [the Groupement's consultant] (234). [The Groupement representative] pointed to the contradiction that the economic study was still in preparation when it was supposed to have provided objective grounds for the adoption of the measures: "Is it necessary to update the statistical data for the report??? I think not, as the report should have been drawn up before the Board reached a decision on the Merfa and the membership fees" (235).

160. On 8 June 2004 (that is to say, after the Commission's inspections), the Board of Directors decided "not to implement in their current state the measures for which it is responsible, i.e. the Merfa, the new arrangements for the membership fee and the additional fee applicable to 'dormant members'" (236). According to the Groupement, "the notified measures have been suspended pending the Commission's decision. The Groupement will then take a decision as to any amendment or repeal" (237).

161. The Groupement's General Meeting on 11 March 2005 decided to amend Articles 7, 10, 12, 13 and 23 of the Formation Agreement of Groupement des Cartes Bancaires so as to remove the amendments made on 20 December 2002 following the adoption of the notified measures.

9. RELEVANT MARKET

9.1. The product market: the payment cards issuance market

9.1.1. Payment card issuance, acquiring of payment and withdrawal transactions and competition between systems

162. As is apparent from the statement of facts in the preceding sections, the payments for the measures at issue in this Decision must be settled when certain thresholds are exceeded by a member issuing CB cards.

163. A distinction should be made between payment card issuance, acquiring of payment and withdrawal transactions, and competition between systems.

164. As in other card payment systems, four parties are involved in the processing of payment transactions in the CB system (apart from the Groupement, which manages the network): card-issuing financial institutions, merchant-acquiring financial institutions, cardholders and merchants equipped with payment terminals. In the case of a withdrawal from an ATM, the parties are card-issuing financial institutions, ATM-managing financial institutions (acquirers) and cardholders.

- Payment card issuance

165. Issuing banks operate in the payment card issuance market, where they invite private individuals, in return for payment of an annual fee, to become holders of the cards they issue, which enables them to pay using their card or to withdraw cash. To this must be added services linked to use of the card, such as the sending of transaction statements, or assistance in the event of loss or theft of the card.

- Acquiring of payment and withdrawal transactions

166. Acquiring banks compete with one another in the acquiring market, where they are responsible vis-à-vis their merchant customers for completing transactions and providing certain ancillary services (such as, for example, guaranteeing payment subject to the merchant complying with all the security measures for which it is responsible).

- Competition between systems

167. Payment card systems compete with one another ("intersystem" competition) to induce financial institutions to opt for membership of their system and issue their cards rather than those of competing systems and to ensure that their cards are actually used rather than those of other systems.

168. Competition between banks may take place within a system, both over the issuance of payment cards and over the acquiring of (merchant) payment or withdrawal transactions. It does not always do so, however, as banks may be members of more than one system and offer issuance or acquiring services involving more than one brand of card.

169. In the present case, the competition between French banks takes place predominantly within the CB system, given the importance of this system in France. This is confirmed both by the discussions between members of the COM and by the internal bank documents obtained during the inspections, in that neither the discussions between the banks nor the internal deliberations concerned any competition from banks belonging to a system other than the CB system.

170. The weakness of intersystem competition in France increases the need for robust intrasystem competition. In other words, the stronger the position of a system in intersystem competition, the more serious is any weakening of competition inside it. In the present case, as is explained in recital 20 above, CB cards are used to pay more than 78% of the total value of card payment transactions in France. The Visa and MasterCard systems do not exert any significant competitive pressure in France: they have no network infrastructure of their own there, and institutions issuing Visa and MasterCard cards in France belong to the Groupement and display the Groupement's CB logo alongside the Visa or MasterCard logo. The Visa and MasterCard cards issued in France are used, not as cards belonging to a competing system, but as cards belonging to the CB system. As regards cards issued in the Visa or MasterCard networks abroad, the Groupement takes it upon itself, by virtue of a motion of its Board of Directors of 28 July 1995 as amended on 22 September 2000, to render them unusable should they be used primarily in its system without the issuing institution being a member of the Groupement (238).

9.1.2. The relevant market does not include means of payment other than payment cards

171. In the notification and up to the time of its reply to the Commission's July 2006 Statement of Objections, the Groupement took the view that the relevant market was that for all means of payment, cash and cheques included (239). However, since its October 2006 reply, by referring exclusively to the competition it would face from other card payment systems (proprietary or store cards and charge cards) without mentioning the competition from any means of payment other than cards, the Groupement seems to imply that by "market for payment systems" it henceforth means "market for card payment systems".

172. It must be made clear from the outset that, while the Groupement considers that other means of payment such as cash and cheques are "clearly substitutable to "CB" cards" to varying but significant degrees" (240), the relevant market includes neither notes and coins nor cheques.

173. In its past decisions, the Commission has found that there is no relevant market grouping together all means of payment, but instead more restricted product markets. It has already had occasion to observe the following:

- Distance payments (transfers and so on) are not substitutable for payment cards because they cannot be used to pay for items across the counter in shops (241), which is a basic feature of payment cards.

- Cheques are not substitutable for payment cards (242). Whereas the possibility of using cards for payments abroad is a feature much appreciated by cardholders (243), cheques are not usable abroad. The withdrawal function of payment cards is usable 24 hours a day, seven days a week, which is not the case with cheques. Cheques can be used for larger amounts than cards (244). The cost to banks of processing cheque payment transactions is much higher than that of processing card payment transactions (transport of the paper cheques, need to transcribe the data, employment of counter staff) and cannot be passed on, even partly, to the direct user (banks are prohibited under French law from charging for the costs of cheque issuance). By contrast, cardholders must pay an annual commission to the issuing banks and merchants must pay a commission to their acquiring bank on each payment transaction (on top of the purchase or rental and maintenance of the payment terminals, software, user licences, etc.). What is more, the automatic processing of card payments and withdrawals from ATMs involves a significant saving for banks, notably in terms of human resources. The use of cards for distance payments is expanding (including use at home via the Internet), whereas sending cheques by post is unsafe, more expensive and slower.

- Notes and coins are not substitutable for cards (245). Their use is highly unsafe compared with cards, where the need to type in a confidential code (or to check the cardholder's identity) before carrying out payment or withdrawal transactions affords protection against the harmful consequences of any loss or theft of the means of payment. Unlike notes and coins, cards make it possible to effect payment at a distance. Lastly, the sending of a statement for card payment transactions is automatic, which is not the case with payments by notes and coins.

9.1.3. The question whether or not proprietary or store cards and charge cards (Amex and Diners) should be included in the relevant market can remain open

174. Proprietary or store cards (246) do not cover a sufficient number of sales outlets or ATMs to confer on them a degree of universality comparable to that of the CB system, thereby reducing their capacity to compete, as a system, with the Groupement, whose cards are accepted in approximately one million sales outlets and 46 000 ATMs (247). For example, the Aurore card - of the various proprietary cards, the most widely accepted among merchants - may be used in some 139 000 sales outlets, or five times fewer than CB cards, and can be used to withdraw cash from the ATMs of BNP Paribas and Caisses d'Épargne but not from those of the other large retail banks (248). Of the other proprietary card issuing institutions, those able to rely on the largest number of acceptance points manage cards accepted by, at most, some 30 000 sales outlets, or almost 25 times less than CB cards. At best, their cards make it possible to effect withdrawals from only one large retail bank (249).

175. In the case of so-called private label card networks, American Express cards are accepted three times less than CB cards (payment by American Express is possible only in 300 000 sales outlets) (250), while Diners Club is a very minor player in France (251).

176. Despite their lower acceptance rate, average expenditure using proprietary and private label cards is higher than with CB cards (252).

177. Although the above elements are strong arguments for excluding proprietary cards and private label cards from the relevant market, the question whether these cards should be excluded from the relevant market can be left open. Of a total value of card transactions of €242.5 billion in 2002, more than 78% was accounted for by CB cards, compared with 16.5% by proprietary cards (the remainder being accounted for by American Express or Diners private label cards) (253). Consequently, the restriction of competition caused by the measures at issue affects a substantial part of the relevant market, whether it includes, besides CB cards, proprietary and private label cards, or whether the market is limited to CB cards alone.

9.1.4. The market for issuance, being that in which the anticompetitive effects under examination occur, constitutes a market separate from those for acquiring and for card payment systems

178. According to the Groupement:

- Issuance does not constitute an autonomous market, because cardholders view as the determining factor in their choice of bank not the conditions peculiar to the issuance service but the banking and support services proposed to them as a whole (254).

- Nor is there a market for acquiring, because merchants view as the determining factor in their choice of acquiring bank not the conditions peculiar to the acquiring service but the banking and support services proposed to them as a whole, including a host of other conditions (255).

- The Commission fails to take account of the existence of interdependences between the activities of issuance and acquiring, each of which generates positive externalities for the other - that is to say, of "the 'two-sided' nature of card payment services". The Groupement suggests that it follows from this "two-sided" nature that issuance and acquiring constitute a single market. On this point, the Groupement does not really specify whether it considers such a market to be separate from that for payment systems or whether issuance and acquiring are mere components of a payment systems market (256).

- The "first and primary relevant market" is the market for payment systems, including proprietary and charge cards (257). The measures at issue should be analysed solely in the context of the payment systems market because the Groupement is "active" in that market alone, and not, unlike the banks members of the Groupement, in issuance and acquiring (258).

179. First of all, contrary to what the Groupement maintains, it does not follow from the fact that the Groupement is not itself an issuer (it being the banks members of the Groupement that are the issuers) that the measures cannot be examined in the context of the issuance market. Community competition law is applicable to the conduct of an undertaking or association of undertakings restricting competition in a market other than that in which it provides its services and for the benefit of undertakings other than itself. Such is the case here, where the Groupement - a card payment system providing services to the banks belonging to it - has taken measures restricting competition between banks in the card issuance market for the benefit of the large banks involved in preparing the measures.

180. Secondly, with regard to possible interdependences between the activities of issuance and acquiring, each of which is claimed to generate positive externalities for the other (the "two-sided" nature of card payment services), the Commission does not dispute that payment by card has two "sides" linked by the existence of network effects. The activities of issuance and acquiring are each indispensable to the other and to the functioning of the card payment system in general. Merchants would not agree to join the card payment system if the number of cardholders were insufficient. Similarly, consumers would not wish to possess a card if it were not usable on a sufficient number of merchants' premises. However, the "two-sided" nature of an economic activity by no means signifies that the system concerned constitutes a single market. Such a nature in no way prevents issuance, acquiring and the payment system from constituting separate markets. The fact that there may be an interdependence between the activities of issuance and acquiring (and between those activities and the activities of the payment system) by no means signifies that issuance and acquiring form part of a single, wider market.

181. According to the Commission's decision-making practice, the two-sided nature of an economic activity is compatible with the existence of separate markets. For example:

- In the Visa International case, where it distinguished between the market for card payment systems ("intersystem market") and the markets for issuance and acquiring ("intrasystem markets") (259), the Commission stated that it "does accept that ... there is interdependent demand from merchants and cardholders, but not that there is joint supply of a single product. Visa card issuers and acquirers each offer a distinct service to a distinct customer. Issuing and acquiring are fundamentally different activities, involving different specialisations and costs" (260).

- Several merger decisions have distinguished between the market for the written press - where the demand side is made up of newspaper readers - and the market for advertising - where the demand side is made up of advertisers (261). However, the Commission points out in them that demand in each of these markets may be interdependent.

- In the Microsoft case, the Commission distinguished between the market for client PC operating systems and the market for media players (a form of software for client PCs), while pointing to the existence of network effects rendering the two markets interdependent (262).

182. As to issuance, acquiring and the payment system each constituting a separate market, it should be pointed out that, according to the Commission's practice, a relevant product market comprises all those products and/or services which are regarded as interchangeable or substitutable by the consumer, by reason of the products' characteristics, their prices or their intended use (263).

183. Clearly, there is no such single, overall service as a "banking and support service" offered indiscriminately to all banking customers - be they cardholders, merchants or whatever - and constituting a single market. There are many separate banking products, which are neither interchangeable nor substitutable, forming the subject matter of a separate supply and a separate demand.

184. For example, it cannot be seriously maintained that the demand by a prospective cardholder for a payment card (with the accompanying services) is identical to the demand by a merchant for the provision of a service enabling him to accept card payments in his store, or, for that matter, to the demand for a mortgage loan. Nor does the provision of one of these services automatically involve the provision of every other banking service in existence. One is not substitutable for the other and the demands for these various services are separate. The customer neither requires nor considers all these services to be interchangeable or substitutable.

185. In the light of the various criteria which distinguish a market - characteristics of the products or services sold, identity of the suppliers or customers, pricing of the products or services - the issuance of cards, the acquiring of merchants and the provision of a card payment system are clearly distinguishable from one another as being so many separate markets:

- in the case of card issuance, there is the sale of cards and the provision of associated services (processing of payments by the cardholder, debiting of the cardholder's account, issuing of statements listing the payment and withdrawal transactions carried out with the card, etc.) by the cardholder's bank (the supplier) to the cardholder (the demander) in return for payment by the cardholder to his bank of an annual fee;

- in the case of merchant acquiring, there is the sale or rental of payment terminals and the provision of services associated with payment by card (processing of payments by the merchant, crediting of the merchant's account, provision of software and technical backup, etc.) by the merchant's bank (the supplier) to the merchant (the demander) in return for payment of a commission for each transaction;

- in the case of the card payment system, the organisation of the system by the Groupement - which defines the general architecture of the system, the technical standards, the contractual rules and the interbank procedures linking the member institutions and manages the resources common to the latter - confers on the system its interbanking character by enabling acquiring banks and issuing banks to work together to perfect the services each bank provides to its customers. This constitutes a service provided by the Groupement to its members in return for payment of an entry fee and a contribution towards the costs of running the system.

9.2. The relevant geographic market: France

186. According to the Groupement, CB cards are issued primarily to holders resident in France, who use them mainly in France (264).

187. The market share held by cards issued in other countries and usable in France is only marginal. Reference should be made here to the Groupement's rule prohibiting the use in France of Visa or MasterCard cards issued abroad by banks not belonging to the Groupement (non-CB cards) where such cards are used primarily on the French network (for more than 50% of payment or withdrawal transactions) (265).

188. The geographic market is therefore France. The Groupement shares this conclusion (266).

9.3. Conclusion

189. It follows from the above that the relevant market in which the anticompetitive effects falling to be examined occur is the market for payment card issuance in France, there being no need to specify whether the market is confined to CB cards alone or also comprises other cards such as proprietary or charge cards.

10. LEGAL ASSESSMENT: INFRINGEMENT OF ARTICLE 81(1) OF THE EC TREATY

190. Article 81(1) of the EC Treaty prohibits, as incompatible with the common market, all agreements between undertakings, decisions by associations of undertakings or concerted practices which may affect trade between Member States and which have as their object or effect the prevention, restriction or distortion of competition within the common market, and in particular those which directly or indirectly fix purchase or selling prices or share markets or sources of supply.

10.1. Decision by an association of undertakings

191. The concept of an association of undertakings, which is not defined by the Treaty, seeks to prevent undertakings from being able to evade the rules on competition on account simply of the form in which they coordinate their conduct on the market (267). As a general rule, an association consists of undertakings of the same general type and makes itself responsible for representing and defending their common interests vis-à-vis other economic operators, government bodies and the public in general (268). Such is the case of the Groupement, which consists of banks and other financial institutions (269), and makes itself responsible for representing them and defending their common interests regarding the interbank system of "CB" payment cards (270).

192. Groupement des Cartes Bancaires "CB" is therefore an association of undertakings within the meaning of Article 81(1) of the EC Treaty, which the Groupement does not dispute (271) and which has already been pointed at by the Court of First Instance of the European Communities (272). The tariff measures in question (namely the Merfa, the per card membership fee, the supplementary membership fee and the fee known as "dormant members wake-up mechanism"), adopted by the Groupement's Board of directors, therefore constitute a decision of an association of undertakings.

10.2. Restriction of competition

10.2.1. Existence of a restriction of competition by object

193. According to the case law of the Court of Justice, in order to determine whether an agreement has as its object the restriction of competition, it is necessary "to examine the aims pursued by the agreement as such, in the light of the economic context in which the agreement is to be applied" (273). It may also be necessary to consider the actual behaviour of the parties on the market. The way in which an agreement is actually implemented may reveal a restriction by object even where the agreement does not contain an express provision to that effect (274). The same reasoning applies to decisions by associations of undertakings.

194. The history of the measures at issue (see section 6 above) clearly demonstrates that the reform embarked upon as part of the CS 2002 project was aimed at countering the threat to the main members' income posed by the arrival of new entrants offering cards at prices which significantly undercut those of the large banks (see, among others, recitals 80, 103 and 110(a)). The threat of the large-scale issuance of bank cards by new entrants at prices significantly undercutting their own was raised both at the meeting on 9 November 2001 (see recital 66) and throughout the work on CS 2002 (see, among others, recitals 61, 63, 71, 72, 79, 117 and 121).

195. The aim of the measures was to penalise new entrants (see recitals 82, 91, 96, 125(a), (b) and c)) by imposing on them additional costs at the point of issuance in order to limit their ability to offer cards at prices significantly undercutting those of the main members (see, for instance, recitals 63, 66(a), 92 and 110(a)). In this way, the main members could avoid having to lower prices themselves - thus preserving their income - or lose market share (see, inter alia, recitals 63, 71, 76, 110(a), 124, 126(a) and 126(b)). The measures were fine-tuned in order to penalise the new entrants while sparing the main members, especially La Poste and Caisses d'Épargne, who were the main members most threatened by the new measures (see, for example, footnote 137 and recitals 101, 111, 121 to 124, 126(b) and 238). The reform was intended instead to place the main members at an advantage.

196. The measures at issue are the result of a new orientation given to the proposals after the project of creating a "SIC" (industrial and commercial company) was given up and are in continuity with it. The SIC already had the aim of obstructing new entrants and of "bringing up" the amounts collected to the most important members (275). However, although the SIC was given up after coming to the conclusion that it was in violation of competition law, it was nevertheless decided to confer on the new orientation given to the proposals which resulted, in the end, in the measures at issue the "same purpose" as that of the SIC (276).

197. According to the Groupement:

- The Commission is concluding that there is an anticompetitive object solely on the basis of the presumed intentions of certain members of the Groupement, whereas the object of an agreement (or a decision by an association of undertakings) must result from the terms of that agreement (277).

- It is not possible (since it is contradictory) to claim that there is a decision by an association of undertakings - which would require the decision to reflect the collective will of all the Groupement's members - while building the case on the intention of just a few members (namely some of the main members) (278). By doing so, the Commission has taken the alleged anticompetitive intention of nine main members as an expression of the will of all members of the Groupement (279).

- The Commission has failed to consider the true object of the agreement, which is to combat free riding (280).

- By taking the view that it is for the Groupement to show that the measures are justified, under Article 81(3) of the EC Treaty, in that they do indeed serve to combat free riding, the Commission has evaded to prove that there is no free riding which is the duty imposed on it by Article 81(1) of the Treaty.

198. In order to carry out an in-depth analysis of the true object of this decision by an association of undertakings, the object of the measures at issue will be analysed with regard to the measures' very formula, which contradicts the objectives stated in the notification and does indeed restrict competition (section 10.2.1.1). The object of restricting competition identified in the formula of the measures is perfectly consistent with the true objectives of the measures as they emerge from the preparatory work within the COM (section 10.2.1.2). It should be pointed out here that the Board of Directors, the body which decided the pricing measures in question, includes only main members, and that the main members (with the exception of CCF, a subsidiary of the UK bank HSBC) prepared the measures in the COM, before they were adopted without discussion by the Board of Directors (see section 6 above).

10.2.1.1. The measures' very formula contradicts the objectives stated in the notification and does indeed restrict competition

199. According to the Groupement, the Merfa is intended "(i) to encourage those members of the Groupement that are more issuers than acquirers to develop their acquiring activities so as to make a balanced contribution to the increase in the value of the CB system; and (ii) to take into account financially the efforts of members with a large acquiring activity in comparison with their issuing activity" (281). The objective of the "dormant members wake-up mechanism" would be, according to the Grouping, to "eliminate any discrimination" by sanctioning the members which share in CB card issuance as compared to the total number of CB card issued of the whole system, during one of the years 2003, 2004 and 2005 would be more than 3 times higher than their share in the total "CB" activities (acquiring + issuance activities) of the whole system years during one of the years 2000, 2001 or 2002 (282).

200. In fact, the measures do not make it possible to attain such aims. Financial institutions which do not wish to be liable for the Merfa are faced with the alternative of reducing issuance or increasing acquiring. As regards the "dormant members wake-up mechanism", banks only limit issuance (283). However, the option of increasing acquiring is scarcely feasible in practice (see section A(1) below). The measures are not appropriate instruments for encouraging acquiring and therefore result either in an additional cost (payment of the fee) or a limitation of card issuance (A.). Moreover, the alleged encouragement of acquiring by the measures is in contradiction with the function ascribed by the Groupement to other commissions in the CB system (B.).

A. The measures (particularly the Merfa) are not appropriate instruments for encouraging acquiring and therefore result, either in an additional cost (payment of the fee), or in a limitation of card issuance

201. While a number of obstacles make it very difficult, in practice, for a new entrant to develop acquiring (1.), due to its very formula the Merfa and "dormant members wake-up mechanism" are not appropriate instruments to encourage acquiring (2.).

(1) A number of obstacles make it very difficult, in practice, for a new entrant to develop acquiring

202. The very formula of the Merfa is such that anyone wishing to escape it is obliged to develop acquiring (of SIREN merchants and ATMs) or to limit card issuance.

203. However, the development of acquiring by a new entrant is made very difficult in practice (a) owing to the fact that this market is almost exclusively in the ands of the main members, (b) owing to the vital importance of a local branch network that makes it possible to develop an overall, individualised and sustained relationship with merchants and (c) owing to the fact that it is reasoanable to believe that the most profitable merchant sectors and areas to be equipped with ATMs have already been acquired by the main members.

204. Accordingly, the Merfa and the "dormant members wake-up mechanism" (284) basically drive anyone wishing to avoid it to directly limit the issuance of cards. Otherwise the fees will have to be paid, which represents an additional cost potentially affecting the price of cards.

(a) The acquiring market is almost exclusively in the hands of the main members

205. The development of acquiring by a new entrant is made very difficult by the fact that this market is almost exclusively held by the main members. Although the number of acquiring banks is relatively high in France compared with the other Member States (see recital 491), most of the acquiring market is concentrated in the hands of the main members. While there are some 90 banks involved in acquiring (285), almost all - more than [...]% - of this activity is exercised by the main members, the remaining 80 acquiring banks sharing the remaining [...]% (286):

<table>

(b) The importance of a local branch network that makes it possible to develop an overall, individualised and sustained relationship with merchants is essential

206. It is very difficult for a bank which is not a main member to gain access to the merchant acquiring market (or to consolidate its presence on that market). Owing to the very high fixed costs of investing in the infrastructure needed to engage in acquiring and the uncertainty of a return on such investment, a bank must first attain a critical mass before it can carry on such business. For some banks, the number of SIRENs to be acquired would be very high (according to [the Groupement consultant]'s additional study, between 1 071 and 1 052 SIRENs for Axa Banque and Egg (287)). The difficulty of launching acquiring activities without a large number of branches was highlighted by a number of new entrants (see, for example, the comments by Citigroup) (288).

207. In particular, a local branch network making it possible to develop an overall, individualised and sustained relationship with merchants is essential to the pursuit of an acquiring activity.

208. The profitability of banks' acquiring activity depends on their overall relationship with the customer as a whole and not only on their acquiring performance in purely electronic banking terms. This assessment is shared by a number of banks, including both banks which the Groupement claims it wants to encourage to increase their acquiring activity (Citigroup, Covefi and Axa Banque) and main members such as CNCE):

"By tying card issuing activities to merchant acquiring activities and to the operation of ATMs, the incumbent banks link the cost of card issuance to the existence of an established banking network, thus raising the barriers to entry or expansion for players that have no or limited operations in France" (289).

"The Merfa favours banks which provide services both to individuals and to tradespeople. For the latter, the banking relationship must be global and not restricted to the acquiring function" (290).

"It is therefore difficult to imagine embarking on the acquiring of small merchants without offering them the full banking service" (291).

"Should a bank decide no longer to equip a merchant, the latter would probably transfer his account to a bank which agreed to equip him. Ceasing acquiring can therefore be very costly for a bank in turnover terms" (292).

209. The banks in a position to offer merchants an overall, individualised local relationship covering the whole range of banking services are the large retail banks, namely the main members which hold [vertually all] of the acquiring market. On the other hand, banks which carry on quite different business from the acquiring of merchants (Internet banks, banking arms of large retailers, etc.), cannot easily become acquiring banks, as the new entrants attest (293).

(c) It is reasonable to believe that the most profitable merchant sectors and areas to be equipped with ATMs have already been acquired

210. Moreover, it is reasonable to think that the profitable areas are already equipped with ATMs and therefore that new machines would have to be installed either in areas which are not or only marginally profitable or in areas already equipped with ATMs. According to Cofidis "ATMs could be installed only in the best places, to the detriment of the profitability of those which are already there" (294).

211. Faced with major obstacles to developing acquiring, new entrants wishing to avoid having to pay the Merfa will prefer to focus their "efforts" on limiting card issuance. Those which potentially would have to pay the "dormant members wake-up mechanism" do not even have this theoretical possibility of increasing acquiring since their share in acquiring activities is exclusively taken into account for the years already passed (295). If they choose instead to pay these fees so as to be able to issue as many cards as they really wish, the sums paid under the Merfa are not allocated to the maintenance and development of the system but merely redistributed among the members not liable for it (primarily the main members, i.e. the members of the Groupement's Board of Directors) (296), which are free to use them as they choose, (as the Groupement acknowledges (297)). Payment of the fees, therefore, to a cost linked to cards for new entrants and an additional source of income for the main members, but this income is not spent on developing acquiring.

212. The Merfa formula and the dormant members wake-up mechanism in themselves thus limits the issuance of cards by new entrants and affects their prices.

213. The same can be said of the new membership fee per carda and the additional membership fee, which both consist of a fee to be paid on each card issued and are payable only by new entrants to the issuance market (new members of the Groupement or establishments which, though already members of the Groupement, have so far issued very few cards and now start issuing a great many), whereas all the longer-standing members of the Groupement - and in particular the main members - avoid them because they are already present on the market and are not exposed to a risk of increasing their market share by a proportion that would trigger the application of these fees.

(2) By their very formula, the measures are not appropriate instruments for encouraging acquiring

214. The Merfa does not encourage members to conform to a balance of acquiring and issuance identified as best for the system, but requires each member to conform to the prevailing acquiring/issuance ratio - that of the main members - with no guarantee that it is in the interests of the system.

215. The business and transactions of each member are not compared with an acquiring/issuance activity level considered optimal for the system. What is compared in the Merfa formula is a member's share in the CB system's total acquiring activities (298) (these activities being measured in terms of number of SIRENs and ATMs) and its share in the system's total issuance activities (299), with the ratio between the two ratios having to be no lower than one half (see recital 142 above).

216. Compliance with this maximum deviation depends, moreover, less on the relevant member alone than on the main members taken as a whole. Inasmuch as the latter hold by far the greatest number of SIRENs, ATMs and active CB cards, compliance with a maximum permissible deviation of one half between the member's share in the system's total "acquiring" and its share in the system's total "issuance" depends on the number of SIRENs, ATMs and active CB cards attached to the main members. The Merfa formula does not dictate a balance. It dictates that the member should not deviate from the behaviour of the main members as regards the division of their effort between, on the one hand, the acquiring of SIRENs and, on the other, the issuance of cards. It is all the more difficult, moreover, for a member to determine what acquiring effort is sufficient in order to escape the Merfa (or how many cards to issue before reaching the limit) because the number of SIRENs and ATMs acquired and the number of cards issued by the other members are communicated to the Groupement only retrospectively, at the end of the accounting period used to calculate the Merfa. New entrants therefore find themselves in a situation of uncertainty, at the mercy of the main members which have prepared and fine-tuned the measures at issue and whose acquiring/issuance profile they have to follow.

217. The function of the Merfa is to force each member not to stray too far from the acquiring/issuance ratio prevailing among the main members, without any guarantee that this is the best ratio for the system. The Groupement therefore imposes mimicry, forcing new entrants to follow the dominant profile of the main institutions even though there is nothing to show that the system's equilibrium would be undermined by the coexistence of opposing strategies (with some establishments carrying out the acquiring function and others focusing on issuance). The same is true of the dormant members wake-up mechanism (300).

218. Moreover, in principle, an establishment's issuing activity generates positive externalities which benefit the system as a whole.

219. GE Money Bank and Banque Accord neatly sum up how, through its very formula and the fact that it is nigh-on impossible for the members targeted by the measures to develop acquiring, the Merfa does not serve to develop acquiring but does indeed impede the freedom of the members targeted to issue as many cards as they would like.

220. According to GE Money Bank:

"a bank subject to the Merfa has three options for limiting the Merfa's financial impact:

(i) it can introduce an acquiring function;

(ii) it can install additional automatic teller machines (ATMs);

(iii) it can reduce its bank card issuing activities" (301).

GE Money Bank passes in review each of these three options before concluding that a member wishing to limit the Merfa's financial impact can actually use only the third option, namely limiting the number of cards issued:

"For GE Money Bank (and probably most banks other than the main members), the first option would require it to launch a new activity from scratch for a new clientele of merchants to whom the bank is currently unknown.

It is hard to see how the second option could be reconciled with a sound business strategy given the saturation of the ATM market and the strong presence of the main members in this sector.

The third option (reducing bank card-issuing activities) will in a good number of cases undoubtedly result in an increase in the relative share of the activity without actually enhancing acquiring activity per se.

Let us take the example of GE Money, which went for the third option in order to minimise the impact of the Merfa and substantially reduced the number of payment cards in 2004 compared with the two preceding years ... We do not see how this decision contributed to the main objective of the Merfa.

In conclusion ... we believe that the criteria used to achieve the objective of the Merfa system are inappropriate and impose a disproportionate and unjustified burden on the banks subject to it" (302).

221. According to Banque Accord: "the thinking behind [the Merfa] is simple: if you want to issue bank cards without being willing or able to acquire a sufficient number of new merchants accepting the card or without opening new cash distributors, you either pay a compensatory contribution and increase fees to holders of your cards or decide not to issue as many bank cards as planned" (303).

222. Intrinsically, through its very formula, the Merfa dictates a behaviour - limiting the issue of cards or choosing to bear an additional cost not borne by the main members - which hinders the freedom of new entrants to compete freely with the main members, without this measure being justified by a function of regulating acquiring/issuance activities (see section 11.1.3 below): the Merfa's very object is anticompetitive.

B. The function of the Merfa is in contradiction with that ascribed by the Groupement to the interchange fees and those of the other measures

223. The alleged regulatory function of the Merfa (that of encouraging the development of acquiring) is contradicted by the existence of interchange fees penalising acquiring.

224. The Merfa is paid by banks primarily engaged in issuance and redistributed to acquiring banks (304) on the grounds that acquiring activity should be encouraged whereas issuance activity should not.

225. Yet, as the Groupement [...] pointed out when opening the COM meeting of 9 November 2001, "the CIP incorporates a 'natural' stabiliser" (305).

226. In this respect, the Groupement claimed in 1995 that the CIP interchange fee (merchant interchange fee) and the CSC component of the CIR interchange fee (ATM interchange fee) were mechanisms regulating the system between issuers and acquirers, and more specifically in favour of issuance, i.e. in direct contrast to the Merfa, since the CIP and CSC are paid by acquirers to issuers:

"the CIR is based on a self-regulating mechanism ... The amount of the CIR, like that of all interchange fees, including the Merchant Interchange Fee or CIP notified to the Commission on 20 December 1990, represents a balance between the divergent aims of the different members of the system, each of which pursue their own commercial strategy. ... The system therefore incorporates its own regulating mechanism which is born of the conflict of interests between issuers and ATM operators [acquirers], with most of the members of the Groupement generally combining these two roles." While the business strategy of some banks may prompt them to favour one or other component of interbanking, their ambitions cancel out those of banks with the opposite business strategy.

... The CSR and the CSC, for their part, constitute the indispensable weighting mechanism ... while preserving the overall balance of the withdrawal function in the system" (306).

227. It is therefore contradictory to claim to be encouraging acquiring by means of the Merfa when at the same time other mechanisms (the CIP and CSC) are, according to the Groupement itself, aimed at restoring a balance by encouraging issuance. The regulatory functions ascribed by the Groupement to, on the one hand, the Merfa and, on the other, the CIP and CSC are diametrically opposed to each other: the former is paid by issuers to stimulate - it is claimed - acquiring while the latter are paid by acquirers to issuers.

228. A number of banks stress the contradiction between the Merfa and the CIP and CSC commissions: Banque Accord, for instance, states that "a bank which expands its pool of ATMs to escape the Merfa will see its CSC increase" (307). Likewise, GE Money Bank reports a "scissor effect" between the CSC and the MERFA (308). If the Merfa diminishes with the installation of ATMs, then there is a threshold beyond which the installation of additional ATMs no longer has any effect on the Merfa (for which the bank has ceased in effect to be liable), whereas the bank starts to have to pay the CSC for having acquired "too many" ATMs.

229. It was only after the Commission suggested, in the course of investigating this case, that the Merfa might be in contradiction with certain interchange fees and at the very least duplicated the interchange fees in its function as an instrument for balancing the system and combating free riding, that the Groupement began contesting what it had previously claimed (that interchange fees could be regarded as a mechanism for regulating the externalities generated by issuing and acquiring activities) (309). Since then the Groupement has argued that interchange fees are intended exclusively to remunerate the interbank services which the banks provide to each other for each payment or withdrawal transaction and certainly not to perform a regulatory function between issuing and acquiring activities.

230. Yet the justification of interchange fees as remuneration for interbank services in no way precludes a regulatory effect (310): whereas the payment by the acquirer to the issuer of an interchange fee for each transaction (the CIP and the CSC component of the CIR) is an incentive for issuers and a disincentive for acquirers, developing issuing activities - which generate positive externalities on the acquiring side - can, for this very reason, be considered an interbank service provided by the issuing bank to the acquiring bank and warranting remuneration by the CIP interchange fee and the CSC component of the CIR interchange fee. The Groupement claimed exactly this when notifying the CIR in 1995, justifying the CSC as financial compensation for the service represented by the network effect, namely the positive externalities generated by issuing activities for acquiring (the installation and management of ATMs):

"the system's withdrawal function can operate only if ... the number of cardholders and the number of ATMs from which they make withdrawals grow in a generally balanced manner and if the imbalances observed at certain banks are corrected by way of financial compensation. The greater the number of cardholders making withdrawals from CB ATMs, the more inclined banks are to install such ATMs ...

- commission No 2, the newly created card service commission or CSC takes account of the contribution to CB interbanking of members of the Groupement whose cardholders make withdrawals from ATMs managed by other members" (311).

231. Furthermore, as well as being in contradiction with the regulatory function ascribed by the Groupement to the CIP and CSC interchange fees, the allegedly regulatory function of the Merfa is also in contradiction with some of the other notified measures, namely the new additional membership fee and the dormant members wake-up mechanism, which penalise banks that have not issued sufficient cards in the recent past. This is because:

- a new entrant which issues in the course of the first three years less than a third the number of cards during or at the end of the following three years is liable for an additional membership fee, and

- a member whose share of the Groupement's issuing activity for one of the years between 2000 and 2002 (312) is less than a third of its share of the Groupement's issuing activity for one of the years between 2003 and 2005 is liable for the dormant members wake-up mechanism.

On the other hand, a member whose share of the Groupement's issuing activity is more than double its share of the Groupement's acquiring activity is liable for the Merfa for each year.

232. The Groupement has not provided any explanation for this contradiction between penalising low issuance in the recent past (through the additional membership fee and the dormant members wake-up fee) and penalising what is deemed excessive issuance after 1 January 2003. There is a real conflict of objectives here, since the additional membership fee is a "rolling" system measured over time (the six years taken into account in the calculation are the first six years following entry into the system, in whatever year that falls), which means that in the first years of its entry to the market a new entrant will face the contradictory logic of the Merfa, which on a yearly basis discourages issuance deemed "excessive" by the Groupement, and the additional membership fee, which discourages inadequate issuance in the first three years of joining the CB system as compared with the subsequent three years.

233. It is clear from the above that:

- the Merfa's object is not that claimed in the notification, namely to encourage acquiring and discourage what is considered to be excessive issuance by certain members, but to restrict competition (section A. above);

- the pursuit of such an objective is doubtful in that it is contradicted by the objective of other measures (interchange fees and others) (section B).

234. This anticompetitive object of the Merfa is, moreover, corroborated by statements made by the main members while the measures were being prepared (see section 10.2.1.2 below).

10.2.1.2. The object of restricting competition identified in the very formula of the measures is perfectly consistent with the true objectives of the measures, as expressed in statements made by the main members while the measures were being prepared

235. Examination of the documents gathered during inspections and dating from the period during which the Groupement and the large banks that are its main members prepared the measures confirms that the measures' anticompetitive object - which is inherent in their formula, in particular as regards the Merfa (see sections 10.2.2.1.A.1 and

10.2.2.1.A.3., recitals 265 to 278, 282 to 286, 291 to 295 and 299 below and section

10.2.1.1.A.1, recitals 201 to 217 above) - is indeed to hinder competition from new entrants to the French market for the issuance of bank cards (A.) in order to limit the fall in the price of bank cards (C.) and enable the main members involved in preparing the measures to safeguard their market shares and their income (B.).

A. The wish to hinder competition from new entrants and to penalise them

236. The object, inherent in the very formula of the measures, of hindering competition from new entrants in the French market for the issuance of bank cards reflects the wish, as expressly attested by a number of documents gathered in the course of inspections.

237. First of all, the participants in the COM expressed their wish to "hold back the arrival of new entrants" (see sections 6.3.2.1-6.3.2.3 and recital 125) and/or to penalise them (313).

238. Not only would the participants in the COM not be subject to these fees, they would benefit from them directly as the fees paid would then be redistributed to them. The banks participating in the preparation of the measures at issue were not to be penalised; quite the contrary, the intention was that they should benefit from them through the distribution of the fees borne by the new entrants (see recitals 144 (concerning the Merfa) and 151 (concerning the other fees) above) (314).

239. Lastly, the objective of penalising new entrants is corroborated by the fact that several studies and fine-tunings were carried out (see, for example, recitals 77, 0 and 126 above) in order to find those measures which would penalise new entrants the most, without the main members being themselves subjected to them in practice (in particular La Poste and Caisses d'Épargne, which were the main members most exposed to the contemplated measures). The SIREN and ATM criteria were used in order to penalise new entrants while sparing the main members:

- The SIREN criterion was used in order to penalise the banking arms of large retailers, since the acquiring of a large retailer counts as only one SIREN for the purposes of the Merfa and the dormant members wake-up mechanism, with the formula taking no account of the number of card transactions acquired or the size of the affiliated merchants concerned (see recital 105). Thus, according to statements by representatives of the main members, the SIREN criterion significantly reduces the large retailers' room for manœuvre (see recital 101(b)) and makes it more difficult for new entrants to escape the measures (see recitals 125(b) and 287).

- The taking into account of ATMs in calculating the Merfa was also used to penalise new entrants in order to make it more difficult for them to find ways of escaping the Merfa (see recitals 122 and 124). This factor was even described as contributing to "inertia" (see recital 113(a)).

- Lastly, the measures at issue were drawn up in such a way as to prevent any of the main members which took part in their preparation from being made subject to them. Several documents obtained during the inspections show that the measures were designed in such a way as to prevent the weakest main members, namely La Poste and the CNCEP, from being made subject to them (see recitals 111 and 101(b)). This was confirmed in a CNCEP internal memo referring to La Poste and the CNCEP: "The calculation rule is to be further developed so as not to penalise those two banks" (315).

240. It is clear from the above that the anticompetitive object of the measures at issue reflects the wish to hinder competition from new entrants and penalise them.

B. The wish to safeguard the main members' incomes

241. The measures at issue - which by their very formula (316) have as their overall object to hinder competition from new entrants in the market for the issuance of bank cards - enable the market shares and incomes of the main member banks that prepared them to be safeguarded.

The Groupement and the main members sitting on the COM had identified the new entrants' plans to issue bank cards as a threat and wished to limit the resulting loss of income (see references in recital 241 above, and in particular the statements of [...] the Groupement and [...] the Société Générale at the COM meeting of 9 November 2001, quoted in recital 66(a) and (c) above).

242. The desire on the part of the Groupement and the main members sitting on the COM to safeguard the main members' income and to reduce the losses that the issuance of bank cards by new entrants would mean for the main members was expressed on a number of occasions during the period of preparation of the measures at issue, as can be seen from several documents obtained during the inspections (317).

243. One way of preventing a loss of income for the main members participating in the COM would be to limit the fall in the prices of bank cards, which account for a large part of banks' income: "Any threat to cardholder fees constitutes a major risk of decline in net banking income and profitability in electronic banking" (318).

An estimate of the losses that would be caused by a reduction in cardholder fees clearly shows the importance of such fees:

"A reduction in cardholder fees of:

- €7.5 per card generates an overall annual loss of €300 million.

- €15 per card generates an overall annual loss of €625 million" (319).

244. Lastly, the amounts paid under the Merfa must be redistributed to the members which are not subject to it, in other words for the virtually exclusive benefit of the main members (320).

C. The wish to limit the reduction in the price of bank cards

245. Limiting a fall in prices compared to the level which would result from free competition constitutes a restriction of competition prohibited by Article 81(1) of the EC Treaty, as the Commission's consistent practice attests (321).

246. The Groupement had foreseen the effects on the founder banks of a large-scale issuance of bank cards by new entrants. According to these forecasts, the founder banks' income would fall and they would be obliged to reduce the price of their cards by the contagion effect of the new entrants' card prices (322).

247. The measures at issue were adopted in order to hinder the fall in the price of bank cards announced by new entrants. They imposed additional costs on the issuing of cards on new entrants, which would be obliged either to pass them on to cardholders or to avoid them by reducing their supply.

248. The existence of an objective of limiting the fall in the price of cards announced by new entrants (and so limiting the impact which the fall in prices brought about by the arrival of new entrants would have on the main members' cardholder fees) is corroborated by statements made on several occasions by the representatives of the main members participating in the COM when the measures at issue were being drawn up (323) (324).

249. As for the Groupement's argument that it is not possible (in that it is contradictory) to claim a decision by an association of undertakings - which would require the decision to reflect the collective will of all the Groupement's members - and to build a case on the intention of just a few members (namely some of the main members) (325), the Commission would stress the following points:

- Firstly, it does not base its conclusion that the object of the measures is anticompetitive solely on the intention expressed by the main members when preparing the measures. It bases its conclusion primarily on the anticompetitive object resulting from the very formula of the measures and finds, moreover, that this object corresponds perfectly to the objectives pursued, as expressed when the measures were being prepared.

- Secondly, it cannot be concluded, simply because a decision by an association of undertakings was taken by the directing body of an association of undertakings on behalf of all its members (or because the members could simply be deemed, under the terms of their initial membership application, to have approved all decisions to be taken on their behalf, although, in practice, some of them in fact disapprove the decisions) (326) that such a decision can in no way, as is the case here, manifest the wish of the few members able to exercise decision-making powers in the matter on their own and benefit them by penalising certain other members.

Accepting such a conclusion would be tantamount to authorising anticompetitive practices against the member undertakings of an association on the sole ground that the practices had officially been drawn up by the association. For anticompetitive behaviour to escape the penalties laid down by Community competition law, it would be enough for it to take the form of a decision by an association of undertakings and for one of the undertakings penalised to be a member of the association. The case concerned by this decision is a good illustration of how certain undertakings exercising decision-making powers within an association of undertakings can penalise their competitors and thereby gain a competitive advantage by means of a decision of the association which they entirely dominate.

- Thirdly, it should be pointed out that, according to the Court of Justice, for a decision to qualify as a decision of an association of undertakings it is not required that each and every member of the association approves the very decision taken by the association (that is to say, no unanimity is required). It is enough that the decision was taken by the body qualified to coordinate the members' activities (327). Accordingly, the possible disagreement of certain members does not disqualify the decision as a decision of an association of undertakings under Article 81, paragraph 1, EC, since this decision was taken by the Board of Directors of the association.

250. As for the Groupement's argument that the true object of the measures at issue is not to penalise new entrants but to combat free riding, and that the Commission, by not taking account of this objective of combating free riding under Article 81(1) of the EC Treaty, but simply under Article 81(3), has failed in its duty to prove that there is no free riding (328), the Commission considers:

- that it has shown in section 10.2.1 that the true object of the measures is indeed to restrict competition;

- that it cannot seriously be argued, as the Groupement claims, that Article 81(1) of the EC Treaty requires the Commission to prove that there is no pro-competitive object.

It is the Commission's duty to enforce the provisions of the Treaty establishing the European Community, Article 81(1) of which prohibits decisions by associations of undertakings which have as their object the restriction of competition. This means inter alia that the Commission may establish, where relevant, the existence of an anticompetitive decision by an association of undertakings.

It is not until paragraph 3 of Article 81 that it is stated that the provisions of paragraph 1 (and therefore of the prohibition it contains) may be declared inapplicable where the decision has benefits meeting the conditions laid down in paragraph 3, and it is settled case-law that the burden of proving the efficiency gains generated by the agreement or decision by an association of undertakings lies with the undertakings or association in question (329).

But in any event, even though the burden of proof does not lie with it, the Commission has not failed to set out the reasons and evidence that cause it to doubt the existence of free riding and consider that the object or effect of the measures cannot be to combat the alleged free riding (330).

Lastly, as set out below (see section 10.2.2.3), the measures in question cannot be considered ancillary restrictions since they are not directly related and necessary to the very working of the CB system or another main operation not restricting competition (and certainly not proportionate to such an objective).

10.2.1.3 Conclusion on the existence of a restriction of competition by object

251. It follows from all the above that the measures have an anticompetitive object consisting in hindering current and future competition from new entrants. As regards the Merfa, those wishing to escape it have, in principle, two options: they can either develop acquiring or limit issuance. Nevertheless, in practice, the Merfa cannot encourage new entrants to develop acquiring because, as explained above (see recitals 201 to 212), the latter are faced with major obstacles to developing acquiring. New entrants will therefore be obliged to either reduce their issuance activities or pay the Merfa. In both cases they will be penalised and the main members will be the beneficiaries. This objective of hindering competition from new entrants is corroborated by several statements made by the Groupement and the main members during the period of preparation of the measures at issue.

10.2.2. Restriction of competition by effect

252. Although, according to Community case-law, there is no need to take account of the concrete effects of an agreement when it is established that it has as its object the restriction of competition (331), the Commission will examine the anticompetitive effects of the measures at issue in this section.

253. According to Community case-law, the analysis of the effects within the framework of Article 81(1) of the Treaty must take account not only of the actual effects but also of the potential effects. In the European Night Services case, the Court of First Instance stressed that "the examination of conditions of competition is based not only on existing competition between undertakings already present on the relevant market but also on potential competition, in order to ascertain whether, in the light of the structure of the market and the economic and legal context within which it functions, there are real concrete possibilities for the undertakings concerned to compete among themselves or for a new competitor to penetrate the relevant market and compete with the undertakings already established (Delimitis, cited above, recital 21)" (332).

254. An agreement may be considered anticompetitive even if its actual effects have not yet been established (333). In a decision concerning the UK Agricultural Tractor Registration Exchange (334), the Commission found that the agreement in question was anticompetitive in so far as "it can be said with sufficient certainty that the market shares and market positions of the members would have been different in the absence of the Exchange and are likely to be different in the future if the Exchange does not continue", even though it was impossible in practice to provide proof of actual differences. It should be pointed out that the Commission considered in this case that the agreement had produced effects in the market despite the fact that its operation had been suspended following the dispatch of a statement of objections. The Court of Justice followed the same line when it took account of the effects of the agreement in question despite its having ceased to produce any (335).

255. The measures to which this Decision relates affect both actual and potential competition. As is explained in recitals 265 et seq., the measures at issue impose additional costs on new entrants. On the other hand, the large banks which participated in the preparation of the measures receive the largest part of the new fees (336) - the part corresponding to the Merfa - and are free to dispose of the amounts thus allocated to them as they see fit (337), which widens the cost differential to the detriment of new entrants. Moreover, these additional costs are not easily avoidable, as is explained above (see recitals 201 et seq.)

256. The imposition of these additional costs has actual and potential effects on the price of new members' cards and/or on their card issuing plans. The measures at issue have the actual and potential effect of higher prices for the cards proposed by new entrants (compared with the price of about €15 which the Groupement and the main members had on several occasions considered likely in the absence of the measures (338)), unless they limit the issuance of cards. New entrants are obliged to come to terms with two variables - the price and the number of cards issued - and to give up competing with the traditional banks on both aspects at once.

257. The measures also have actual and potential effects on the prices of the main members' cards, whether new entrants increase their prices or reduce the number of cards issued:

- If new entrants increase the price of their cards, then there is less competitive pressure on the prices charged by the main members participating in the COM. In this case, the price differential with the price charged by the large banks is reduced or negated, which lessens the competitive pressure to which the large banks would otherwise be subject. Consequently, the adoption of the measures at issue enables the main members to maintain or increase the price of their cards.

- If the new entrants choose to remain below the issuance thresholds that, in each case, trigger the application of the charging measures against them, then this also leads to less competitive pressure on the prices of the main members sitting on the COM.

258. As will be explained below, the measures at issue are anticompetitive in that they:

- have or may have the actual and potential effect of preventing the erosion of market shares and income which the main members sitting on the COM would have suffered following the issuance by new entrants of significant numbers of cards at prices significantly undercutting those of the large banks;

- actually and potentially limit or may limit technical development in so far as new entrants are or may be impeded in their issuance of cards with new functions;

- have the actual and potential effect of walling off the French market for the issuance of bank cards inasmuch as they hinder or may hinder access by new entrants (which, according to the Groupement, will in future be mainly foreign banks (339)).

259. Any analysis of the measures' effects must take into consideration the fact that, on 8 June 2004, the Board of Directors decided to suspend the operation of the Merfa, the new membership fee (including the additional fee) and the dormant members wake-up fee (340) pending a possible decision by the Commission as to their compatibility with competition law, and that the amounts payable have never been recovered or received by the Groupement. The decision of the Board of Directors stipulates a mere suspension of the measures and in no way their outright cancellation.

260. The Commission will therefore analyse in turn:

- the effects the measures would have if they were implemented (if the suspension were lifted and the amounts recovered (section 10.2.2.1.) and

- the existence of tangible effects (effects that the measures have had and continue to have despite the suspension of the measures) (section 10.2.2.2.).

10.2.2.1. The effects the measures would have if they were implemented (if their suspension were lifted)

261. If the measures were actually applied, the service offered by the new entrants would be significantly less attractive (A). The consequences of a less attractive service offered by new entrants would not be limited to effects on the prices (B.) or volume of cards issued by new entrants (C.). Much more than that, a less attractive offer from new entrants would have the effect of reducing the competitive pressure to which the large banks would be subject in the total absence of the measures, and enabling those large banks to maintain a much higher level of prices than would result from free competition, thereby also preserving their income (D.). Application of the measures would have the effect of slowing down appreciably the erosion of the large French banks' market shares that would occur if the attractiveness of the new entrants' offerings were not reduced.

262. Any application of the measures - which have only been suspended, and could be immediately reactivated - will be harmful to consumers, since the latter will not benefit from the offer of cards at prices significantly undercutting those of the large banks, as would have been the case without these measures. This consideration applies not only to the customers of new entrants and to the customers of the large banks, which avoid the substantial competitive pressure that the new entrants would have brought to bear, but also to consumers in general, who suffer in an environment in which free competition cannot produce all its positive effects.

263. In order to establish the existence of a restriction of competition by effect, the Commission must be able to foresee with sufficient certainty that there will be negative effects in the relevant market in terms of prices, output, innovation or the variety or quality of goods and services (see recitals 253 and 254 above). In the present case, in addition to its own analysis, the Commission has in its possession very precise witness statements from the main member banks on the effects the measures would have if they were reactivated. Moreover, the replies from new entrants to the Commission's requests for information largely coincide with the views of those who designed the measures.

A. The substantial lessening of the attractiveness of the offerings of new entrants through the fine-tuning of the measures

264. The measures substantially reduce the attractiveness of new entrants' offerings.

1) The additional cost imposed on new entrants

a) The amount of the additional cost

265. If the suspension of the measures were lifted, a new entrant to the bank card issuance market which is not active in the acquiring market (a 'pure' issuer) would have to pay (in addition to the fixed membership fee of €50 000, which the Commission is not calling into question):

- the Merfa (a maximum amount of €11 per active card issued, in the absence of any acquiring activity);

- either a one-off payment of €12 per card issued during the three years after joining the Groupement plus - if during or at the end of the period running from the fourth year to the end of the sixth year after joining, the new entrant issues more than three times the number of cards issued during the first three years - an additional fee of €12; this fee, which is also payable only once, is calculated on the basis of the number of cards that it "should have" issued during the first three years in order for the stock of cards issued at the end of the sixth year not to be more than three times the stock at the end of the third year; (beyond the sixth year after joining, if the new entrant does not generate a sufficient number of acquirings it may still be subject to the Merfa);

- or a fee, known as the "dormant members wake-up fee", of €12 per card issued in excess of the number of exempt cards (determined by reference to the relative rate of payment/withdrawal issuance or acquiring effected by the new entrant during each of the years 2000-02 and to the new entrant's share of the active card market for each of the years 2003-05), again in addition to the Merfa, where applicable.

266. Although, formally, the measures at issue are applicable equally to all the Groupement members, in practice (because of the criteria chosen in measures' formula to determine if a given member has to pay and to define the amount due) the large banks that took part in their preparation do not have to pay them, while new entrants do and - as the impact of the measures varies according to the number of issued cards - are thereby imposed significant additional costs. Moreover, those banks which are not subject to the measures receive the lion's share of the new fees - the part corresponding to the MERFA (341) - which increases the cost differential generated by the measures to the detriment of new entrants.

267. Thus, any new entrant has to pay €12 per card issued during the first three years after joining. A new entrant issuing 100 000 payment cards during the first three years after joining therefore incurs an additional cost of €1.2 million, payable only once. Furthermore, if this new entrant is a pure issuer without any acquiring activity - which is the case with most of those subject to the MERFA (342) - and consequently must pay €11 per active card by way of the Merfa, it incurs a further additional cost of €11 per active card. Consequently, a pure issuer incurs an additional cost of €23 during the year of issuance of a card and an additional cost of €11 in each subsequent year. During the first period of three years after joining, this amounts to an average annual additional cost of €15 per payment card issued and active ((€23 the first year + €11 the second year + €11 the third year) / 3).

268. Any new entrant which, between the third and the sixth year after joining, issues more than three times the number of cards issued at the end of the third year is subject to the additional fee of €12 per card issued in excess of three times its stock of cards at the end of the third year. For example, a new entrant which, at the end of the third year, has issued 100 000 cards may issue up to 300 000 cards before the end of the sixth year after joining. If it issues 500 000 cards, it has to pay €12 for the third (343) of the 200 000 cards issued in excess, or € 800 000 on top of the annual €11 by way of the Merfa. The average surcharge for issuing cards in excess of the permitted threshold is €15 a year between the third and the sixth year, which in turn represents 53% of the cardholder fee proposed by new entrants.

269. As is explained in Section 7.3 above, the dormant members wake-up mechanism has an effect similar to the membership fee: a dormant member which, during the years 2003, 2004 or 2005, holds a stock of active cards more than three times the size of the exempt stock - calculated by reference to its stock of active cards, ATMs and SIRENs for the years 2000, 2001 and 2002 - is subject to a fee of €12, payable only once, per card issued in excess of the number of exempt cards.

270. These additional costs are significant compared with the prices applied by new entrants. The €23 which a pure issuer new entrant has to pay when issuing a card represents an additional cost of 81% compared with the proposed card price (of about €28.5 on average among the new entrants which will be subject to the Merfa if they fulfil their issuance plan (344)). In subsequent years a pure issuer has to pay €11 per active card, or an additional cost of the order of 39 % compared with a cardholder fee of the order of €28.5.

271. If a new entrant which is a pure issuer passes on these additional costs to its cardholders, it has to offer its cards at a price higher than that of the traditional banks (345), namely €51.5 for the first year (346) and €39.5 thereafter (347) or €43.5 if account is taken of the average additional costs during the initial three-year period, on average €15.

272. The following table shows the above estimated additional costs per card for a new entrant which is a pure issuer. Similar additional costs are borne by "awakened" dormant members (348).

273. Pure issuer new entrants therefore incur the following additional costs:

- a surcharge of €15 on average during each of the years 1-3 after joining;

- if they are not subject to the additional fee, they incur a surcharge of €11 per subsequent year under the Merfa;

- if they are subject to the additional fee, new entrants will have to pay, for years 4- 6, €15 for cards liable for the additional fee (they will still pay €11 for the other cards), and €11 a year after the sixth year for as long as they are liable for the Merfa.

274. According to the Groupement (349), the Merfa has no effect on the prices of new entrants in so far as only [...] of a membership of some 149 are liable to pay it. The Group also argues that only [more than 65% of the institutions subject to the Merfa] were liable to pay the maximum of €11 per card. Banque Accord (the bank of the large retail group Auchan) was liable for only €4.77 a card; Banque Chabrières (the bank of the large retail group Intermarché) of €6.17 a card and Banque Bipop for €7.52 a card.

275. It should be pointed out here that:

- According to the Groupement, two thirds of the members liable for the Merfa ([...] out of [...] (350), more than 65%) would have to pay the maximum amount of €11 per card. It is therefore wrong to claim that the members liable will in most cases be liable for less than the maximum of €11 per card.

- The measures are focused on the new entrants likely and seen as likely to exert competitive pressure on the traditional banks: the banks having to pay the largest amounts are mostly new entrants attached to large retail groups and/or engaged in internet banking (351). A number of these new entrants were named by the main members when they were preparing the measures (352).

- The fact that some members attached to large retail groups (353) do not figure on the list of those that would have had to pay the Merfa in 2003 does not mean that the Merfa had no effect. On the contrary, the Merfa serves as a deterrent by inhibiting the issuing plans of the main members' potential competitors. Some banking arms of large retailers not subject to the Merfa declared that they had scaled down (or prepared to scale down) their issuing plans with a view to escaping a charging measure that they feared (354) (even after the measures were simply suspended (355)) and were cited by the main members as needing to be penalised in their issuing projects (356). When, quite exceptionally, this was not the case, it was because the Groupement and the main members saw, during the preparation of the measures, that they were not interested in issuing significant numbers of CB bank cards, with the result that they were not targeted by the measures and penalised by them (357).

- While many members of the Groupement do not figure on the list of members liable, those targeted are listed (with the exception, as explained above, of certain banks that scaled down their issuing plans to avoid it), whereas the main members, the member establishments controlled by the main members and the members not perceived as a threat by the main members are not. Far from showing that the Merfa is not anticompetitive in its object and effect, the fact that members not perceived as a competitive threat to the main members - and not therefore targeted by the measures during their preparation - ultimately escape the Merfa would tend to illustrate the opposite.

276. While the Groupement highlights the very small number of new entrants not paying the Merfa at the maximum rate of €11 a card, it should be pointed out that the amounts of €6.17, €7.22 and €4.77 a card payable by Banque Chabrières, Axa Banque and Banque Accord nevertheless represent a substantial extra cost. For Banque Accord, paying neither at the maximum rate, the Merfa and the dormant members wake-up mechanism together represent, in the first year, no less than 43% of the selling price of its card (358).

277. It is clear therefore from the list of the Groupement's members liable for the MERFA (359) that the Merfa spares the main members and their subsidiaries and specifically targets, among the Groupement's other members, new entrants capable of exerting competitive pressure on the main members by issuing cards at more competitive prices.

b) The additional cost is not easily avoidable

278. The additional cost to new entrants resulting from the measures at issue is not easily avoidable:

- Firstly, the €12 surcharge implied by the membership fee cannot be avoided because any new entrant or dormant member subject to the dormant members wake-up fee will have to pay €12 per card issued.

- Secondly, as regards the Merfa, no member of the CB system can foresee with certainty the acquiring effort needed so as not to be liable for the Merfa. It is only at the end of the year that the members of the Groupement liable for the Merfa are identified on the basis of the acquiring and issuing activity (in terms of the number of ATMs, SIRENs and active CB cards) they declare at the end of the year (360) in relation to that of all the other members of the system. Unable to know beforehand the scale of acquiring and issuing activities by other members of the Groupement, a member cannot really predict the level of acquiring and issuing activity that would enable it to escape the Merfa at the end of the year.

- Thirdly, the installation of ATMs entails significant additional costs for new entrants without their being certain that it will exempt them from the Merfa. Thus, according to the Groupement, the installation of an ATM costs approximately €40 000-50 000 (361) to which must be added other costs linked to the operation of ATMs. According to the Groupement's estimates, in order to escape the Merfa Axa Banque and Egg would have had to install [...] and [...] new ATMs respectively (362); each of the two banks would therefore have had to spend between € [...] and € [...] million in an attempt to escape the Merfa, without any guarantee of success. An additional difficulty concerning the installation of ATMs is that it is reasonable to think that the locations not yet served are not profitable, while the most profitable places are already served by (363).

- Fourthly, as regards payment acquiring, it is difficult for a small or medium-sized bank to gain access to the acquiring market or to consolidate its position in that market. Owing to the high fixed costs of investing in the infrastructure needed to engage in the business of acquiring, a bank must attain a critical mass before it can carry on such business. It is unlikely, moreover, that new entrants will manage to acquire SIRENs where the large banking groups (with the largest share of the acquiring market) have failed (see recital 205). And new SIRENs will probably be acquired for the most part by the main members; indeed, these groups are in a privileged position when it comes to carrying on acquiring activities, able as they are to offer merchants the whole range of banking services, including local ones. This is because most merchants prefer to maintain an overall relationship with a single bank rather than with several and, for the bank, the profitability of its acquiring activity depends on its overall relationship with the customer and not only on its acquiring performance in purely electronic banking terms (see recital 206).

2) The penalising of new entrants

279. The measures penalise new entrants. This is clear from the facts of the case, as examined above, and the statements of the main members, the new entrants themselves and the consumer associations BEUC (the European Consumers' Organisation) and UFC (Union Française des Consommateurs) - Que Choisir, all of which have expressed converging opinions on this point.

280. It is clear from remarks made by main members while the measures were being prepared that new entrants will be penalised by the measures in such a way as to reduce significantly their competitive advantage (see section 6 above, and in particular the quotes from Caisses d'Épargne, BNP Paribas and Société Générale in recitals 125, and 136(a), (b), (d) and (e) above).

281. Several new entrants argued that they had indeed been penalized (364) and that the technical development of cards had thereby been limited, the offer of new functions having been reduced (365), an analysis shared by the consumer associations BEUC and UFC-Que Choisir (366).

3) The criteria used for fine-tuning and its effects

282. The penalising of new entrants on the one hand and the positive effects for the major banks participating in the COM on the other are attributable to, among other things, fine-tuning when the measures were being prepared, in other words the choice of criteria used in formulating the various pricing measures at issue.

283. As mentioned above in section 6.3 (see for instance recitals 110, 111 and 122), the most significant types of fine-tuning are having recourse to SIREN and ATM criteria in the Merfa formula and the so-called dormant members wake-up fee, and the use of the "group of banks" criterion.

a) The SIREN criterion

284. As explained above, the formula determining the applicability or otherwise of the Merfa uses the criterion of the number of SIRENs to measure acquiring. The SIREN criterion is also used in the formula for the dormant members wake-up mechanism (367). The use of the SIREN criterion rather than the SIRET criterion penalises new entrants.

285. For the purposes of calculating the interchange fee payable by the acquiring bank in respect of each payment, the Groupement measures the electronic banking activity of each member in terms of number of transactions per SIRET number (and not per SIREN number) "acquired" by that member. However, for the purposes of the Merfa and the dormant members wake-up mechanism the Groupement decided to depart from this principle. Under French regulatory provisions (368), a SIREN number is assigned to each enterprise (that is to say, to each natural or legal person engaged in a trading activity) and a SIRET number is assigned to each establishment. Thus, a merchant whose enterprise is divided up into several establishments is allocated a single SIREN number but several SIRET numbers. Choosing the SIREN number rather than the SIRET number (or the number of payment transactions) has the effect of reducing the share of new entrants from the large retail sector in the acquiring activity of the whole system taken into account by the Merfa. The acquiring activities of new entrants from the large retail sector naturally relate to payment transactions with large enterprises, each made up of a large number of establishments.

286. If the number of card payments at merchants acquired by each bank is compared with the number of active SIRENs of that same bank in 2003, it becomes clear that the acquiring activity of a large retailer's banking arm is greatly reduced where the number of SIRENs is used as reference criterion. Witness the gap between the acquiring activity of a large retailer's banking arm and that of a main member, which is much wider when that activity is measured in terms of number of SIRENs than in terms of number of payments: whereas the number of interbank payment transactions acquired by [banking arm of a large retailer] represents 29% of that of [a main member bank of the Groupement], the percentage in the case of SIRENs falls to 0.16% (percentage compared with [a main member bank of the Groupement]: 85% of transactions, as against 0.27% for SIRENs; compared with [a main member bank of the Groupement]: 38% of transactions, as against 0.007% for SIRENs) (369). The choice of the SIRET criterion better reflects the true scale of a bank's acquiring activity since a very large merchant owning several establishments represents in reality for its acquiring bank a much larger volume of payment transactions requiring processing than a small merchant with only one establishment.

<table>

287. Furthermore, documents obtained during the inspections show that the SIREN criterion (rather than SIRETs or the number of transactions) was chosen with the specific aim of penalising new entrants. For example:

- In an internal Société Générale memo, it is written that taking SIRETs into account might have rendered large retailers less liable to the Merfa:

"1.3.1 SIRENs

and not SIRETs, since the latter criterion would facilitate circumvention of the Merfa. The large retailers could easily increase their number of acquiring contracts based on the concept of SIRET premises" (370);

According to BNP Paribas: "As for large retailers, the choice of the number of merchant SIRENs as an index of acquiring activity significantly reduces their room for manœuvre (even Pass, Carrefour's acquirer, currently has few active SIRENs)" (371).

288. The main members' views tally with those of new entrants. Thus, several new entrants stress the penalising effect for large retailers of the choice of the SIREN criterion as a measure of acquiring activity for purposes of calculating the Merfa:

- According to S2P, taking SIRENs into account when calculating the Merfa is not conducive to the development of acquiring activities:

"The acquiring function is assessed on the basis, not of volumes of transactions processed, but of a SIREN count, the SIREN being a mere legal identifier of a company.

As a result, a retail company consisting of 90 hypermarkets accounting for 80 million transactions a year is put on the same footing and represents for the GIE a weight equivalent to that of an independent jeweller's effecting 1 000 CB transactions a year, and this despite the fact that the interchange fees paid to the GIE that same year amounts to €17.4 million for the retailer compared with less than €1 000 for the small trader" (372).

- Banque Accord states that the Auchan group generates very substantial transaction volumes which are acquired by Société Générale and account for [...] % of the latter's acquiring activity. However, this activity amounts to only seven SIRENs (373). Banque Accord concludes from this that:

"the SIREN criterion adopted by the Merfa is an anticompetitive criterion when it comes to acquiring hypermarkets as it takes no account of the volumes acquired" (374).

289. At the hearing of 13 November 2006, when asked by BNP Paribas's representative why it did not itself conduct the acquiring of CB card payments in the hypermarkets of the group to which it belonged (Auchan hypermarkets) - as, he said, the Carrefour group's S2P bank did - Banque Accord replied that:

- even if S2P did conduct the acquiring of payments at Carrefour hypermarkets, that would amount to at most 15 SIRENs, which was far too few to enable it to escape the Merfa without cutting back its issuing plans (as the same BNP Paribas representative had, moreover, pointed out when the measures were being prepared (375)); and

- for its part, if Banque Accord were to conduct the acquiring of payments at its Auchan hypermarkets (even though it already handled all the technical side of acquiring), that would amount to at most four SIRENs or the possibility to issue barely 200 cards without paying the Merfa.

290. Lastly, in the opinion of a participant in the meeting on 12 July 2002 preparing the ground for the COM meeting on 19 July 2002: "the SIREN concept does not reflect the reality of an acquiring activity (the SIRET being more appropriate)" (376).

(b) The inclusion of ATMs in the Merfa formula

291. The penalising effect on new entrants of the taking into account of ATMs in the Merfa formula was also foreseen by the main members sitting on the COM. Thus:

- The participants in the COM noted that bringing ATMs into the calculation of the Merfa introduced an inertia factor for new entrants (see recital 113). They also noted that new entrants would have great difficulty in equipping themselves with ATMs in order to escape the Merfa: "The impact is above all on the investments the 'predators' will have to make, since can be expected that in the long run they will be able to install ATMs and make their expected investments profitable (although the development of charges for withdrawals makes this more difficult). But launching an ambitious ATM programme makes it necessary both to raise extensive financial resources and to find suitable locations for installing them" (377).

- According to a memo preparing for the COM meeting on 11 October 2002, taking ATMs into account "raises the liability threshold and therefore strengthens the effectiveness of the mechanism" (378).

- In an internal Caisses d'Épargne memo, it is explained that "to escape the Merfa, new entrants will have to make very substantial acquiring efforts (ATMs in particular)" (379).

c) The "groups of banks" criterion

292. The groups of banks taken into account for the purposes of the Merfa are the "CIP Groups" (380). According to the Groupement, "'CIP Group' is the term used to designate the group of banks composed of a member of the Groupement (head of CIP Group) and those of the institutions that are at least 51% controlled by it and so wish" (381).

293. Having recourse to this criterion enables the main members to avoid a situation in which some of their subsidiaries which issue a lot but acquire little are liable for the Merfa, by diluting their issuance/acquiring activities in the activities of the rest of the "CIP Group" to which they belong. Cetelem (a 99%-owned BNP Paribas subsidiary) thus confirmed in March 2003 that it did not carry out any acquiring activities but still did not have to pay the MERFA (382). This situation contrasts with that of the - primarily issuing - new entrants competing directly with the main members' subsidiaries which, unlike the latter, cannot escape the Merfa in this way.

B. The effects on the price of cards

294. As is explained in section A.1, recitals 265 to 273 above, on the basis of the measures' very formula, new entrants which issue cards without acquiring merchants or installing ATMs are subject to an additional cost of €15 on average during each of their first three years of activity, €11 or €15 for each of years 4-6 after joining (383) and €11 for each subsequent year. This hiking up of costs has an impact on prices:

- either because the new entrants increase the price of their cards, which leads to less competitive pressure on the prices charged by the main members sitting on the COM;

- or because the new entrants remain below the issuance thresholds that, in each case, trigger the application of the measures against them, which also means less competitive pressure on the prices of the large main member banks sitting on the COM.

295. In other words, actual application of the measures will inevitably be reflected by effects on the prices charged by the large main member banks, whether new entrants increase their prices or whether they reduce the number of cards issued.

296. These considerations are supported by the views of the main members, new entrants and consumer associations. In this respect, the Commission would stress that the statements of the Groupement and its main members on the desired effects, the statements of the new entrants liable for the Merfa on the effects experienced, the conclusions reached by the Commission - on the basis of the measures' very formula - that effects are inevitable if the suspension of the measures is lifted and the Commission's findings concerning the measures that the effects have already had all concur (the Commission does not hold the statements of new entrants to prove the measures' anticompetitive effect but rather to confirm its conclusions; by way of proof, it has conducted its own analysis of the actual effects on the market (see sections 10.2.2.2.A.2 and 10.2.2.2.B.2, recitals 331 to 339 and 348 to 352) and set out, on the basis of the measures' very formula, the effects they would inevitably have if their suspension were lifted (see section 10.2.2.1 above)).

297. When preparing the measures, the Groupement and the main members expressly intended them - and especially the Merfa - to reduce the heralded cut in the price of cards by new entrants or result in a price increase (see inter alia the statements of the Groupement, BNP Paribas, Caisses d'Épargne and the Société Générale quoted in recitals 92, 101(b), 110(a), and 126(b)).

298. For their part, the new entrants are unanimous in the view that the measures in question generate additional costs that will affect the price of their cards (cardholder fee) (384), and the consumer associations BEUC and UFC-Que Choisir also see the measures as increasing the price of cards.

C. The inhibiting effect on new entrants' issuing plans

299. As explained in sections A. and B. above, new entrants would incur additional costs when issuing bank cards if the new measures were applicable. These additional costs increase proportionally with the volume of new cards issued, which gives new entrants an incentive to issue fewer cards. This is because:

- In the case of the Merfa, for the same acquiring ratio, the amount payable under the Merfa by a new entrant increases proportionally with the increase in the number of active cards issued by that new entrant.

- In the case of the membership fee of €12 per card issued during the first three years after a new entrant joins, the more cards the new entrant issues, the more it must pay. Thus, a new entrant will have to pay €12 000 if it issues 1 000 cards and €120 000 if it issues 10 000 cards.

- The additional fee of €12 per card also has the effect of limiting or reducing the card issuing plans of new entrants in so far as the additional fee is not payable if the number of cards issued between the third and the sixth year is less than three times the number of cards issued during the first three years after joining. If, on the other hand, new entrants exceed this ceiling, they will have to pay the additional fee, the amount of which increases in line with the number of cards issued in excess of the maximum number of exempt cards (385).

- The dormant members wake-up fee also leads to a downward revision of new entrants' card issuing plans. The fee is due only when dormant members issue cards in excess of the number of exempt cards (386).

300. In order to limit the increase and the accumulation of costs, new entrants therefore have the possibility of maintaining issuance below the thresholds which in each individual case trigger the application of the measures against them (387). This option has in fact been used by several new entrants. It will be explained below how, during the 17 months during which the measures were in force, new entrants' issuing programmes were sharply curtailed (see section 10.2.2.2.A.1 "The reduction of issuing plans" below).

301. The main members had anticipated the effect of a reduction or limitation of the issuance of cards by new entrants failing which the latter would have to incur the additional cost stemming from the measures:

(a) In an internal BNP Paribas communication, the negative effects of the new measures on Egg's card issuing plan were stressed: "I confirm to you that Egg would indeed be penalised by the new CB rules presented to the Board of Directors ... In conclusion, their business plan is likely to be more than tight" (388).

the changes made to Egg's issuing plan following the adoption of the measures at issue were also foreseen in an internal Caisses d'Épargne email: "The change in the dormant members wake-up mechanism is effective: Egg would have to pay €11.4 million in 2003-04 with the new system, which should change its business plan" (389).

D. The beneficial effects of the measures for the participants in the COM, including the safeguarding of their income

302. As explained in sections A(1) and B above, the measures at issue, which are based on card issuance, increase the costs per card issued by new entrants who have no acquiring activity by €15 per card per year on average during the first three years, and by €11 thereafter, and this increase in costs has or at least may have negative repercussions on prices. Applying these measures allows a significant limitation of the income erosion that would be suffered by the main members in the event that new entrants were to issue a considerable number of cards at appreciably lower prices. The estimates that were made by the participants in the COM confirm the benefits the new measures bring them. Thus, the participants in the COM considered that:

(a) The measures would work to their advantage (see recital 101(a) as regards Crédit Lyonnais and recital 101(b) as regards BNP Paribas) and would not apply to any of the participants in the COM. None of the main members sitting on the COM were on the list of those liable for the Merfa in 2003 (390). Moreover, in so far as the proceeds from the Merfa were to be distributed among those who were not subject to it in proportion to their share of acquiring activity, the main members participating in the COM would benefit from it as they had decided when the measures were being prepared. According to the estimates drawn up by the Groupement for 2002, the CB members deriving the greatest benefit from the Merfa were the main members sitting on the COM (see the table in footnote 406). Even La Poste and Caisses d'Épargne would not be penalised, "despite the small scale of their acquiring activities" (see recitals 101(a) and 101(c)).

(b) Their subsidiaries would not be liable, as is shown by the example of Cetelem, the BNP Paribas subsidiary (see footnote 148 and recital 293).

(c) The measures did make it possible to reduce the losses they would have incurred as a result of the issuance of cards at appreciably lower prices by new entrants. For example:

- in the "results of the simulations and comments" annexed to the internal Société Générale memo preparing for the COM meeting on 11 October 2002 it is stated that "in all cases the Merfa minimises the loss for the main members!" (391);

- according to an internal CNCEP memo, thanks to the new measures the expected losses of €140 million would be reduced to €93 million (392);

- according to a general overview document on the CS 2002 project dated 25 September 2002: "Without charging, the impact of new entrants on the existing members will amount to €893 million between 2002 and 2008 ... With the charging system, the existing members will succeed in limiting the fall in their GDP" (393);

- according to an internal Natexis memo preparing for the COM meeting on 11 October 2002, the Merfa has the effect of reducing the losses resulting from the issuance of bank cards by new entrants:

" 3.2. A positive impact for the banking community

The Merfa acts as a brake on the loss of income if the new entrants issue cards...

if the new entrants were to issue 5 million cards over five years and install ATMs so as generally to avoid paying the CSR to another bank, the losses of the founder banks would be €71 million rather than €1 011 million without the Merfa" (394).

303. The beneficial effects of the measures for main members have also been stressed by consumer organisations. Thus, BEUC and UFC-Que Choisir have stated that "the 'current' members of GCB, and in particular those with extensive networks, which seems to be the case with the 'main members', would receive additional income as a result of the new charging system" (395).

E. The insulation of the French market for the issuance of bank cards

304. The competitive situation of foreign banks - which is already weakened owing to the fact that, if they wish to issue payment cards in France, they must in practice issue CB cards (396) and that the Groupement's rules limit cross-border issuing (397) - is made even more difficult by the fact that, when they issue CB cards in France, the measures at issue increase their costs. Foreign banks are ill-placed to avoid the measures' application by setting up a merchant acquiring or an ATM network in France.

305. This view is shared both by foreign banks and by other card payment systems. For example:

(a) According to Citibank "The payment cards business in France is currently controlled by a limited number of large banking groups, with well established national banking networks, that also operate and control the CB organization ... Citigroup is concerned that the new fee structure proposed by CB will perpetuate this situation ... In particular, the new fee structure will increase barriers to entry and expansion, created by CB's control over the French payment cards business, and will further entrench the competitive advantages enjoyed by the large incumbent issuers by increasing the direct and indirect costs of new and small issuers and reducing the costs of incumbent banks ... we view the notified CB fee proposals as a device proposed by the incumbent banks to deter entry by new issuers and expansion by smaller players, and thus preserve the existing oligopolistic structure for payment cards in France ... by tying card issuing activities to merchant acquiring activities and to the operation of ATMs, the incumbent banks link the costs of card issuance to the existence of an established banking network, thus raising the barriers to entry or the expansion for players that have no or limited operations in France" (398).

(b) According to Visa, "A significant impact of the new fee structure will be on financial institutions outside France (for example, internet banking institutions) as it is such institutions who are likely to be seeking access to the French market in the future. The proposed new rules may also limit the likelihood of new market entrants generally. In this respect, [Visa] note[s] that any bank, including foreign banks which issue cards used mostly within France (more than 50% of total usage), must join the CB system and would thus be subject to its rules and fees.

Therefore, it is likely that the revised fee structure for new members will have an adverse impact on cross-border issuing by increasing the barriers to entry to the French market and, consequently, have an adverse impact on competition between payment systems in France" (399).

306. The Groupement expected that most new applications for membership of the CB system would come from foreign institutions ("it is probable that future new memberships will come mainly from foreign institutions, since the number of French institutions likely to engage in CB activities that are not members of the Groupement seems to be limited" (400)). This has indeed been the case in recent years: "the number of foreign institutions applying for membership is growing. Of the 32 institutions that have joined the Groupement since 1998, about half have been of foreign origin" (401).

307. It should be recalled in this connection that two foreign banks, Egg and Capital One, ceased their card issuing activities in France after the adoption of the measures at issue (although it is not possible to establish a direct and exclusive causal relationship between the measures' adoption and the abandonment of the activity by the banks).

308. The fact that some foreign institutions (such as Caixa Bank, Banco Popolare di Verona, Arab Bank and Monte Paschi Banque) are not affected by the measures at issue does not invalidate the Commission's conclusion. The insulation of the French market does not have to be total for it to be concluded that the measures have an insulating effect.

309. Consequently, by penalising and hindering competition from foreign banks, the measures have the effect of insulating the French market for the issuance of bank cards.

10.2.2.2. The existence of tangible effects on the market

A. Effects produced during the period 1 January 2003 - 8 June 2004

310. As is explained above, the measures at issue in this Decision were adopted by the Board of Directors on 8 and 29 November 2002 and were applicable from 1 January 2003 until 8 June 2004, when the Board decided to suspend them.

311. The measures at issue were brought to the attention of members of the Groupement by an information memo dated 13 November 2002. It was from that date, therefore, that the measures produced effects, as will be explained below, in that new entrants had to make sweeping changes to their plans in order to avoid the measures or at the very least to reduce their cost somewhat.

312. It will be explained in recital 320 that in 2003 several new entrants considered that on the basis of the 2003 figures they were liable to the Merfa but would be able to change their business plans so as to avoid the measures or reduce their negative impact (402).

313. Members were reminded of the application of the new measures by letter from the Groupement dated 23 December 2003. According to that letter, "Certain contradictory reports having appeared in the press claiming that the Groupement has decided not to apply the measures adopted by its Board of Directors in December 2002, namely the new system of membership fees, the dormant members wake-up fee and the Merfa, your Board of Directors has asked me to confirm to you that these measures will be applied in 2003. Consequently, some members will receive invoices from the Groupement for payment either of the Merfa or of the dormant members wake-up fee, according to their contribution to optimising the CB system. These invoices should be sent to you in April 2004 and will be payable immediately".

314. Although a list of those liable for the Merfa in 2003 was drawn up (403), the Groupement did not collect the sums that were due under the measures linked to the issuance of cards (404).

315. During the period of application of the measures, the new members owed €12 per card issued during their first year of membership and (depending on the imbalance between their relative weight as acquirers and their relative weight as issuers) a maximum of €11 by way of the Merfa. Moreover, during this period, any dormant member who "woke up" owed €12 per card issued in excess of the cards it was entitled to issue "free of charge" on the basis of its issuance and acquiring activities between 2000 and 2002, in addition to the Merfa that might be applicable to it.

316. On the other hand, none of the participants in the COM were subject to the Merfa. Neither La Poste nor Caisses d'Épargne, the main members most vulnerable to the negative effects of the measures envisaged, figure among those liable for the MERFA (405), precisely because, as is explained above (see section 6.3.7), the measures finally adopted had been fine-tuned to prevent this. Quite the reverse, if the Merfa had been collected the main members would have benefited from it, since, as already mentioned (see recital 144), the Merfa is distributed among the members who are not subject to it, in proportion to their overall index figure for their contribution to acquiring activity (406).

317. During the period in which they applied, therefore, the measures rendered new entrants liable for the additional costs while sparing the main members. Consequently, the measures at issue prevented cards from being offered at prices appreciably below those of the big banks, and prevented other services linked to cards from being offered free of charge or at lower prices; as a result the measures contributed to maintaining the status quo (preserving income and market shares). The measures at issue also had the effect of discouraging the issuing of cards by new entrants (the more cards they issued, the more they had to pay), and hence limited the production of bank cards. Moreover, in so far as new entrants intended to issue cards with new functions, the measures at issue limited the technical development of bank cards. Lastly, in so far as new members of the Groupement were foreign institutions, the measures also had the effect of insulating the French market for bank cards.

318. The suspension of the measures linked to the issuing of cards on 8 June 2004 did not mean that the charging measures that had been taken vis-à-vis new entrants during the previous period were deprived of their effects. Thus, the behaviour of new entrants between November 2002, when the charging measures at issue were adopted, and June 2004 was conditioned by the certainty that the measures were going to apply on the basis of the number of cards issued during that period. That certainty had a negative impact on competition despite the fact that, at a later date, the Board of Directors decided "not to implement in their current state" the measures at issue.

319. In particular, the adoption of the measures led new entrants to make big cuts in their plans for the issue of cards. And the main members did not have to reduce their prices, while some new entrants did have to increase theirs.

1) The reduction of issuing plans

320. During their application the measures at issue had the effect of reducing the scope of the new entrants' card issuing programmes, and even of throwing into question their plans to issue new cards at all. For example:

(a) After the measures at issue came into force, Banque Accord greatly reduced its CB bank card issuing programme. In fact, it issued only 48% of the cards planned. This reduction, which was due in particular to the impact of the measures at issue (407), confirms what Banque Accord stated in 2003, namely that the Merfa had had a major impact on its CB bank card issuing plan: "The new charging system will induce us to modify, or even abolish, the current offering, which was economically viable before the decision was announced" (408).

Following the adoption of the Merfa, Banque Accord decided to reduce its CB card issuing plan rather than pass on the new costs to its customers:

"In view of the referral of the Merfa dossier to the Commission of the European Communities, we have not passed on the new CB costs to our customers. However, we have reduced our CB card issuance volumes because the Merfa penalises issuing activity. We are no longer going to offer self-service blister subscriptions in Auchan supermarkets ... We are converting fewer proprietary cards to CB cards and are reserving our bank card for our best customers, but without for the time being passing on to them the additional financial cost stemming from the MERFA (409)". "The introduction of the Merfa fees by Groupement des Cartes Bancaires CB at the end of December 2002 has had a very significant impact on our CB card issuing plan. In view of our profitability level, we have scaled back our bank card issuing ambitions in order to be able to cover the cost of the Merfa" (410).

(b) GE Money Bank (formerly GE Capital Bank) also had to revise its card issuing strategy radically following the adoption of the Merfa: "the introduction of the new charges (including the Merfa) prompted GE Capital Bank to revise its CB card issuing strategy" (411). It proceeded to correct its card issuing targets strongly downwards (initially by about 9% in 2003, and then by about 70% for the period 2004-06) (412), as it had decided to do following the adoption of the MERFA (413). Only the launch of a Visa Plus CB card, limited to cash withdrawals, enabled GE Money Bank to lessen the planned reduction in card issuance, as most of the cards issued by that bank in 2004 were cash withdrawal cards (the Merfa being lower in the case of cash withdrawal cards (between €0 and €3) than in that of cards with a payment and a withdrawal function (between €0 and €11)) (414).

(c) S2P states that it was forced into a drastic 85% reduction of its card issuing plan, from an initial 2004 target of 1 000 000 down to 132 043 on 31 December 2004, which was only 13% of the initial forecast (415). S2P has also stated its intention of remaining below the number of cards which in its case would trigger the Merfa:

In view of the new Merfa charges, we have in effect been obliged to revise our CB card issuing plan, not to mention our overall strategy in this business area.

Originally, our idea was to convert our stock of proprietary cards (three million cardholders) into bank cards en masse, thereby following the practices already implemented in the English-speaking countries ... On this basis and as indicated in our previous submission, we had set our sights on 1 000 000 customers by the end of 2004.

On the current bases of the Merfa and in view of our pool of ATMs, we would be eligible for the charge from 450 000 to 500 000 cards.

We have therefore revised our marketing and communication plan so as to remain below the threshold for the next two to three years ...

At this stage we would confirm that we have rethought our strategy in France, concentrating on promoting among the Carrefour group's clientele a fidelity programme which 4.4 million new customers have signed up to since April 2004, rather than on expanding our bank card issuing programme by 150 000 cardholders.

Without wishing to draw hasty conclusions, we would like to point out that in Spain, where prices, regulations and the commercial context are admittedly very different but where the market is nevertheless more open, our subsidiary Servicios Financieros Carrefour EFC SA has in the same time, i.e. in less than two years, issued more than [X.X] million bank cards to its domestic customers (416).

(d) According to Cofidis: "The new charges introduced by the Merfa are liable to jeopardise the supply of CB cards by Cofidis. This is because the cost of issuing CB cards would become prohibitive, and if customers were to be charged for all the costs borne by Cofidis there would be an increase in the cost of the card to the customer, who would switch from the Cofidis card to the cards offered by our competitors not subject to the Merfa" (417). "[A]t the end of 2003, our plan was to issue 20 550 CB cards. In mid-October, the number of CB cards issued was 7 000. We estimate that the number will be 8 000 between now and the end of this year 2003" (418). In 2003 Cofidis issued only 34% of the cards planned, and in 2004 only 13% of the cards planned (419).

(e) Banque Casino states that: "The Merfa makes it, moreover, economically extremely difficult on the face of it to transform existing proprietary cards into bank cards" (420). "The additional costs linked to the Merfa will unquestionably induce the Casino group bank to reconsider its CB plans. In particular, any programme for the issuing of substantial volumes of CB cards henceforth seems to us to be economically difficult to implement. The very system of compensation introduced by the Merfa has the effect, besides increasing our operating costs, of reducing the operating costs of those banks which benefit from the Merfa" (421).

321. According to the Groupement no substantial effect on issuance is discernible, because out of the eleven banks questioned by the Commission only two, AXA Banque and GE Money Banque, alleged that there had been any effect on their issuing targets (422).

322. The Groupement makes the following specific points:

(a) The answers referred to by the Commission are too few (423).

(b) Statements running counter to the Commission's argument are in the majority. Barclays Bank, Bipop and Covefi said that the Merfa had had no impact on the number of bank cards they had issued (424), while Banque AGF contradicted the Commission's statements by saying that "the projections had not been achieved essentially for commercial reasons" and that the revision of these objectives "derived mainly from marketing activity and demand on the part of customers, and were not the outcome of a change of direction in the strategy of the Bank with regard to the number of cards to be issued that was in strict relation to the decision of the Board of Directors of the Groupement" (425).

(c) The number of cards issued by S2P did not really change after the suspension of the Merfa, because it was already below the number of cards that would have made it liable to the Merfa. This shows that the target S2P had set was overambitious (426).

(d) The Groupement suggests that Banque Accord's statement of 2005, according to which "the market had not come up to [its] expectations", invalidates its statement of 2003, according to which it had "scaled back [its] bank card issuing ambitions in order to be able to cover the cost of the Merfa" (427).

323. It should be pointed out, first of all, that the Groupement here seeks to reply to the Commission's arguments only with regard to the effects actually produced on issuance and prices, that is to say the data collected empirically from the banks questioned in 2003 and 2005 (outlined in the present section 10.2.2.2): the Groupement ignores the Commission's observations on the measures' potential effects, that is to say the effects which, from a straightforward analysis of the form of the measures themselves, it can be said would certainly be produced if the suspension of the measures were to be lifted and the measures actually enforced (see section 10.2.2.1, "The effects the measures would have if they were implemented (if their suspension were to be lifted)").

324. It should be borne in mind that the implementation of the measures was first postponed (428) and then suspended indefinitely, and that the sums due under the measures have never actually been collected.

325. The potential effects of the measures are not purely "theoretical", as the Groupement claims: they are the effects that the measures would produce if their suspension were to be lifted, and the sums due were actually to be collected.

326. Although the sums due have never yet been collected, the replies from the banks questioned in 2003 and 2005 describe effects that were produced by the suspended measures, as a result in particular of the prudence exercised by the liable banks, who anticipated the recovery of the sums due, and before they knew that the measures would be suspended took decisions whose effects lasted beyond the suspension. It is reasonable to suppose, therefore, that these effects are less far-reaching than those that the measures would produce if the suspension were to be lifted (the "potential" effects).

327. Thus the observations submitted by the Groupement in response to the Statement of Objections skirt around the principal effects of the measures, which are the potential effects.

328. Turning to the arguments put forward by the Groupement regarding the issuance of cards, the Commission sent requests for information to a substantial number of banks to which the measures applied, including the nine banks liable to the largest amounts under the Merfa (over €150 000) (429) and others that the Groupement and the main members had referred to as a threat at the time the measures were being prepared (430). The Statement of Objections of 2006 cited no fewer than fourteen replies (431).

329. Most of the banks the Commission questioned described the measures as having a negative impact on issuance (432).

330. Moreover:

(a) The Commission sent requests for information to Barclays and Bipop in 2005 only because they were on the Groupement's list of CB members liable for the Merfa in 2003 (Barclays being one of the banks liable for more than €150 000). But when the Groupement included these banks among those liable to the Merfa it was making a mistake. Barclays had decided "to put an end to its business in France at the end of 2001 or the beginning of 2002 [i.e. before the measures were adopted] and consequently to recruit no new customers; the number of customers fell from 35 000 to 15 000 between 2000 and 2004". Barclays concluded that it "would not be subject to the new Merfa as it stood, unless it were to issue more than 100 000 cards ... which was difficult to imagine with the exit of Barclaycard" (433). The same applies to Bipop. At the time the measures were adopted, Bipop was a small financial institution in financial difficulties which later led to its closure. Bipop never envisaged "going retail", or issuing cards, except in isolated cases and "only to customers who expressly asked for it" (434). The fact that Barclays and Bipop did not change their card issuing programmes after the measures were adopted does not show that there were no effects on the liable banks, because Barclays and Bipop were not liable banks.

b) It is true that in 2005 Covefi, which via Cetelem was 34% controlled by BNP Paribas, one of the main members, said that the adoption of the Merfa had not had any effect on the number of cards it had issued. But Covefi has always maintained that the measures had no effect on the volumes it issued only because it chose to offset the extra cost imposed by the Merfa by increasing its prices or reducing its services, rather than by reducing issuance: as far as Covefi was concerned, growth in card issuance was an "imperative necessity" (435). In its reply of 2005 Covefi made it clear that the Merfa had had no impact so far only because Covefi was "convinced that the Merfa mechanism would not be implemented given the questionable character of the decisions incorporated in it". Covefi went on, however, to say that if the Merfa were to be implemented as it stood it was certain "that the cost of the Merfa would be passed on to the cardholder: this would amount to an annual cost of about €10 per cardholder per year" (436).

Thus the Groupement cannot cite Covefi's reply in support of the argument that the measures had no effect on issuance, because Covefi would have been penalised in terms of issuance if it had not chosen to pass on the cost of the measures in its prices.

(c) Lastly, it is true that Banque AGF says that it was unable to achieve the planned number of cards "essentially" or "mainly" for commercial reasons, and that the change of direction was "not in any strict relation" to the adoption of the measures, but it clearly states that as the implementation of the Merfa "would induce a change in the criteria of financial equilibrium" Banque AGF would have to "revise its business plan and review its commercial strategy with regard to the issuance of CB cards" if the Merfa were to be implemented as it stood (437).

(d) S2P said in July 2005 that while it was below the number of cards that would render it liable to the Merfa, this was precisely because as a result of the Merfa it had already been forced to make drastic cuts in its plans for the issuance of cards over the whole of 2004, from a planned stock of 1 000 000 cards to only about 150 000, a reduction of 85%. S2P also said that although the Merfa had already been suspended for over a year it had revised its marketing plans in order to remain below the number of cards that would trigger the Merfa for the two or three years following (438). Lastly, S2P does not suggest that its target of 1 000 000 was overambitious: quite the reverse, it points out that at the same time, in less than two years, its Spanish subsidiary issued [X] million cards to domestic customers, and that the conversion of millions of proprietary cards into bank cards is a practice already followed by the big supermarket chains in the United Kingdom and the United States. The Groupement consequently has no basis for the claim that S2P's statement of July 2005 shows that the Merfa had no effect on card issuance.

(e) Banque Accord did say in 2005 that the market had not come up to its expectations, but it made it clear that the gap of 52% between the stock of cards planned and the stock of cards achieved between 2003 and 2004 (439) was also due to the fact that "our shareholders have asked us to be prudent in view of the impact of the measures taken by GIE Cartes Bancaires on this activity; we have therefore reduced our prospection marketing activities by about 20%" (440).

This statement of 2005 thus confirms the statement of 2003 in which Banque Accord likewise spoke of a reduction in the volume of cards it was issuing as a result of the Merfa, which had led it to abandon its plans to prospect in shops: "we have reduced our CB card issuance volumes because the Merfa penalises issuing activity. We are no longer going to offer self-service blister subscriptions in Auchan supermarkets. The presence in shops which we had announced for September 2003 came to nothing because of the Merfa charges" (441).

2) The effects on the price of cards

331. When the measures were being prepared the effect sought by the Groupement and the main members was to maintain the prices of their cards (442). To achieve this the Groupement and the main members sought to keep new entrants from issuing cards at prices appreciably lower than their own, in order to prevent a general fall in the price of both new entrants' cards and their own cards (as a result of what was known as the "contagion" effect).

332. In December 2001 some new entrants were already offering cards at prices appreciably below those of the traditional banks (443): for example, Zebank (Egg) was charging an annual cardholder fee of €21.9 for a Visa deferred debit card, and Banque Directe was charging €0-30 a year, depending on the amount of purchases made, for a deferred debit CB Visa card (444). Similarly, Banque Accord was offering customers of Auchan supermarkets an international Visa CB card for €25, with a reduced annual cardholder fee (€15) for their partner, six months free of charge, and new functions such as cash back (445).

333. Throughout the period when the measures were being prepared, the Groupement and the representatives of the main members (except CCF) who attended meetings of the COM and the working parties operated on the basis of estimates according to which the new entrants would be issuing bank cards at prices between €15 and €20. For example, the forecasts drawn up by the Groupement and submitted to the banks who were members of the COM in May 2004 said that "Given the present levels of the annual cardholder fee, a charge of €15-20 for universal payment cards issued by retailers appears reasonable" (446); and in simulations carried out by the Groupement in October 2002, the price taken as a basis without the Merfa charges was likewise €15: "for the variant without the Merfa, a price of €15 is taken as a basis (cf. prices observed among large retailers)" (447).

334. When preparing the measures, the Groupement and the representatives of the main members reckoned that, with the new charging system that was being drawn up, new entrants would have to pass on 50% of the charges in the card price, which would oblige new entrants to increase the price of their cards to €22 instead of €15: "The new entrant passes on 50% of the charges in the issue price, as follows: €15 + (€13.30) = €22" (448).

335. And the new entrants were indeed unable to issue their cards at a price of €15-20. In the years 2002-2004 their prices were higher:

- Axa Banque began issuing Visa cards in 2003 at a price of €24 for its immediate debit Visa card and €30 for its deferred debit Visa card. These prices remained the same in 2004, but increased strongly the year after (449).

- AGF offered its immediate debit Visa card at €30 (450), and increased its prices in 2005;

- Covefi offered an immediate debit international CB card at €22.50 (451);

- GE Money Bank marketed its classic Visa card at €23, the first year being free (452);

- Banque Cofidis marketed its cards at €24 (453);

- S2P began issuing in 2003 at €25 (454);

- Banque Accord offered its Visa card at €25, its MasterCard at €25 and its Visa Premier card at €45 (455); and

- ING offered its international Visa card at €40 (456).

336. It is clear from the foregoing that the measures in question had effects on new entrants' card prices, and more precisely the effect sought by the Groupement and the main members, the effect of preventing the issue of cards at prices of €15-20 by the new entrants who were targeted by the measures.

337. When the measures were being prepared, the Groupement and [...] the main members considered that, without the new charging system, they would be forced to reduce the prices of their cards because of a "contagion" or "contamination" effect of the reduction in prices initiated by the new entrants. As a result of the measures, however, the main members would be able to maintain the level of their prices.

338. The adoption of the measures at issue did enable the main members to increase or at least to maintain the price of their cards. In no case was the price of their cards reduced - "contaminated" by the new entrants' prices - during the period of application of the measures, as they had feared would happen without the measures. Here are a few examples of the prices of immediate debit international bank cards (457):

- Caisses Régionales du Crédit Agricole increased the price of its classic Visa card on average from €40.19 in 2002 to €42.42 in 2003 and to €43.60 in 2004 (458). Its subsidiaries Crédit Lyonnais, Finaref and Sofinco did not reduce the price of their cards during the period of application of the measures. First of all, Crédit Lyonnais did not reduce the price of its international Visa cards following the adoption of the measures in question. Crédit Lyonnais offered an immediate debit international Visa card at €32 in 2001, €34 in 2002 and €34.5 in 2003, 2004 and 2005 (459). Finaref offered international Visa payment cards at €24 in 2003, 2004 and 2005 (460). Sofinco, which reduced the price of its international Visa cards before the adoption of the measures in 2001 (from €25.15 between 1996 and 2001 to €24) likewise did not reduce the price of its cards following the adoption of the measures at issue, offering international Visa cards at €24 between 2002 and 2005 (461).

- Nor did Société Générale reduce the price of its cards during the period of application of the measures. In 2001 and 2002 the prices of its Visa and MasterCard cards were €31 and €41 respectively, and during the period of application of the measures they were €32 and €42 (462).

- CNCEP also increased the price of its immediate debit classic Visa cards from €30 before the adoption of the measures (in 2001 and 2002) to €31.5 in 2003 and to €32.5 in 2004 (463).

- BNP Paribas increased the price of its classic Visa card. Before the measures in question were adopted, it offered its classic Visa card at €31.70 in 2001 and at €33 in 2002. During the period of application of the measures, BNP Paribas increased the price of this card to €34 in 2003 and to €35 in 2004 (464).

- Before the measures were adopted, the price of a La Poste immediate debit Visa card was €29 in 2001 and €30.5 in 2002. During the period of application of the measures, the price of these cards was increased to €31.5 in 2003 and to €32.5 in 2004 (465).

- CIC increased the prices of its cards, and specifically its immediate debit international Visa and MasterCard cards, which went up from €29 in 2001 and €32 in 2002 to €33 in 2003 and 2004 (466).

- Natexis Banques Populaires did not reduce the price of its classic Visa card, which it issued at €40 from 2001 onward (467).

339. It follows from the above, first, that the adoption of the measures at issue meant that the main members did not have to reduce the prices of their cards, as they and the Groupement feared they would have to do in the absence of the measures, but instead were able to increase or maintain them; and, second, that the new entrants were not able to charge lower prices of €15-20, but instead maintained their prices or even increased them.

340. The Groupement, however, takes the view that the measures had no effect on the price of cards (468):

(a) The fact that during the preparation of the measures the main members may have estimated that the new entrants would be able to offer CB cards at prices appreciably below their own is irrelevant, and did not bind the Groupement in any way (469). The price of cards remained the same (470), and no increases were observed (471): "None of the alleged effects ... has taken place or is likely to take place" (472).

(b) The Commission's analysis is based on the abstract, hypothetical and unrealistic estimate in the table given in recital 272 above, according to which, in the Groupement's view, all the new entrants would be subject to all the price measures in combination and at their maximum levels (473), whereas the sums due would in reality be much lower.

In support of this argument the Groupement refers to sums to be paid by certain members which are lower than the maximum (474), and states that the great majority of members of the Groupement will not be liable to the Merfa: "of the roughly 160 members of the Groupement at the end of 2003, only [XX] would have been liable" (475).

In particular, according to the Groupement, the additional cost of only €5.98 per card per year that Banque Accord would incur under the Merfa, and the €6.05 it would incur under the wake-up mechanism, would be less than the total additional revenue of €11.02 per card per year that Banque Accord would gain by converting a proprietary card into a CB card (476).

(c) The Groupement argues that the Commission cannot conclude that the measures have a real negative effect on the prices charged by new entrants, because the new entrants' replies never mention actual increases, but speak only of hypothetical price rises and the possibility that the new entrants might need to increase the price of their cards in the future, as can be seen from the use of the future and conditional tenses in their replies (477).

(d) The Commission has engaged in "excessive disclosure avoidance" in respect of some answers given by new entrants to a request for information of 29 June 2005 (478), and "selected the banks to which it addressed the request for information of 2005 on the basis of the view of the effects of the measures that it wanted to throw into relief" (479), whereas a "rigorous and precise examination [of the statements made by the new entrants in 2005] would have shown that the market was manifestly impervious to price reductions and that pricing behaviour varied from one bank to another depending on the bank's particular commercial strategy" (480).

(e) The Commission cannot "give credit to the preconceived, simplistic and false" idea that the new entrants are offering their CB cards at prices appreciably below those of the traditional banks. The Commission fails to take account of the following:

- the prices of CB cards issued by the traditional banks are very competitive when account is taken of the services linked to their cards (481); and

- the prices charged should be "seen in the light of the nature of the card, which may be a payment card or a credit card" (482): the CB cards issued by the new entrants are for the most part credit cards, and the real cost is not confined to the annual subscription but derives mainly from the very high cost of credit (483).

341. The Groupement considers, lastly, that the Commission is drawing the conclusion that the new entrants are being penalised in competition on the sole basis of the prices of cards, whereas it should have examined the role played in the preferences of cardholding consumers by criteria other than the price of cards alone, and especially the CB card's qualities in terms of universality, security and reliability, and should have evaluated the sensitivity of cardholders to changes in price (484).

342. The Groupement's remarks can be answered as follows:

(a) The papers relating to the preparation of the measures clearly show that [representatives of the Groupement] itself, and not just the main members, repeatedly took the view that the new entrants would offer CB cards at prices appreciably lower, and that this view was never contested (485).

It was after a meeting of the COM at which it was said that the much lower prices that the new entrants would be charging were a threat to the traditional banks that the directors of the Groupement were asked to work out the principles of a system of pricing that would be "relatively dissuasive" (486). The Groupement did take the view, therefore, that the new entrants would be offering cards at prices appreciably below those of the traditional members.

In the view of the Groupement itself, then, the prices charged by new entrants would in the absence of the measures at issue have been appreciably below those of the traditional members; so that the fact that the new entrants were not able to charge prices as low as they had hoped, but instead had to maintain or even to increase (487) their prices, or to issue fewer cards than they had hoped at prices appreciably lower than those of the main members, constitutes a negative effect on the prices charged.

The anticompetitive effect that the measures had on prices can also be seen in the fact that the main members did indeed prove able to increase or maintain their prices: (488) one of the objects of the measures was to alleviate the contagion effect (489) that would have been produced by cards issued by new entrants at appreciably lower prices, that is to say the need for the traditional banks to lower their own prices in turn.

Lastly, for the measures to produce anticompetitive effects it is not in any way necessary that the majority of the members should be liable to the measures. The measures produced anticompetitive effects because they were targeted at the new entrants who were in a position to exert competitive pressure, and because they did in fact affect them.

(b) By applying the mechanisms of the various measures to the realistic hypothesis of a pure issuer (490), the table in recital 272 (491) gives an idea of the additional cost generated by the measures themselves.

According to the Groupement (492), of [...] members liable to the Merfa, [more than 65 %] would have paid the maximum of €11 per card. The Groupement cannot claim, therefore, that the hypothesis that forms the basis of the table, that of a member who is a pure issuer and is consequently liable to the maximum amount of Merfa, is abstract, hypothetical and quite unrealistic. The hypothesis describes the position of a number of members greater than the small number of establishments which according to the Groupement were not liable to the maximum level of MERFA (493).

In addition, contrary to what the Groupement contends, the Commission has never claimed that new entrants would all pay the maximum amounts possible, or that they would always be liable to all the measures together. And the Commission does not need to show that the new entrants would be penalised at the maximum rates in order to conclude that there would be restrictive effects on competition.

It is surprising that the Groupement should criticise the alleged absence of any consideration of the purpose of the decision of the association of undertakings as it is set out in the clauses of the decision itself (494), while at the same time claiming that the figures in the table in recital 272 are too abstract to demonstrate the real nature of the measures. The table sets out precisely to provide an objective demonstration of the impact of the tariff mechanisms taken in themselves, by applying them to the simplest hypothesis, the situation most common among the members liable to the Merfa, that of a member who issues only.

But the Commission has not neglected to examine the practical effects of the measures, and has inquired into the numbers of cards issued by the new entrants principally concerned and by the main members (495).

As regards the additional cost borne by Banque Accord, which according to the Groupement was small, the following points should be noted:

- The situation of Banque Accord is not that of the majority, because according to the Groupement most of the members subject to the Merfa are liable for the maximum possible amount of €11 per card, whereas Banque Accord would have to pay only €4.77 under this measure.

- Even for Banque Accord, which does not have to bear the maximum amounts of €11 for the Merfa and €12 for the wake-up mechanism, but instead only € [...] and € [...] respectively, the combination of the two pricing measures in the first year of application of the measures represents an additional cost of [...] % of the selling price of its CB card ([...]).

- The Groupement describes the additional cost of the measures passed on by the new entrants to their customers as "additional revenue" generated by the conversion of a proprietary card into a CB card. This revenue is taken to be equal to the difference between the prices of the two cards (the annual fees); but the price of a card is not a net profit to the issuing bank, so that the gain on the conversion of a card is not simply the difference between the prices of the two types of card. More fundamentally, it is not justifiable to claim that the "profit" or "additional revenue" will offset the loss of attractiveness of new entrants' offerings occasioned by the measures at issue.

(c) The Groupement cannot argue that the impact of the measures on prices is purely hypothetical, and that no increase has ever been observed, and at the same time point out that several new entrants answered that they had indeed increased their prices (496).

Nor can it be argued that the use of the future tense in some of the new entrants' statements shows that the effects they attribute to the measures are hypothetical: a reading of the statements clearly shows that the effects they outline are considered by their authors to be real. The use of the conditional in some of the new entrants' statements is not surprising given that the statements were made at a time when the recovery of the sums due was becoming more improbable, following the inspections of May 2003 and the fact that the Groupement had postponed the implementation of the measures for a year (497), after which they were suspended.

(d) The Commission does not engage in disclosure avoidance, and it has not done so here. When firms give it confidential information or information containing business secrets, claiming that disclosure to third parties would cause them injury, the Commission treats the information as confidential, or as a business secret, after first satisfying itself in consultation with the firm that the information is indeed of this kind, and as far as at all possible obtaining a non-confidential summary to which it allows access on the part of the firm against which allegations are being made. The Groupement had access to the Commission's file, and the Commission answered all the Groupement's requests for clarification with regard to the non-confidential versions of the replies to the request for information of 29 June 2005 (498). After the Commission had replied the Groupement did not advance any further remark, question or difficulty regarding the content of any confidential aspect of the replies to the request for information of 29 June 2005. It is somewhat surprising, therefore, that the Groupement should put forward alleged difficulties of this nature in response to the Statement of Objections of 18 July 2006 (499).

Furthermore, the Commission did not "select the banks to which it addressed the request for information of 2005 on the basis of the view of the effects of the measures that it wanted to throw into relief", contrary to what the Groupement groundlessly alleges.

As for the Groupement's claim that a "rigorous and precise examination" of the statements made by the new entrants would have shown that the market was impervious to price reductions and that pricing behaviour varied from one bank to another depending on the bank's particular commercial strategy (500), the Groupement does not itself carry out any such "rigorous and precise examination", or put forward any ground whatsoever for this claim.

It cannot be argued, as the Groupement seeks to argue, that because no bank has reduced its prices the market must be impervious to price reductions. Quite the reverse, when the measures were being prepared, the Groupement, the consultants it approached and the main members repeatedly expressed their conviction that there would inevitably be a contagion effect on the prices of the traditional banks if the new entrants were left free to issue CB cards at appreciably lower prices (501).

Lastly, the new entrants have argued that the sensitivity of demand to the prices of cards prevents them from charging prices above those of the traditional banks. According to Cofidis, for example, "if customers were to be charged for all the costs borne by Cofidis there would be an increase in the cost of the card to the customer, who would switch from the Cofidis card to the cards offered by our competitors not subject to the Merfa" (502).

(e) That new entrants issuing cards at prices below those of the traditional banks might be prevented from issuing as many as they would wish, far from being a preconceived idea of the Commission's, was seen as a real and serious threat by the Groupement, by the consultants it approached and by the main members throughout the time when the measures based on that threat were being prepared. This is quite objectively shown by the case of banks such as Covefi, GE Money Bank and Cofidis, who charged prices of €22.50, €23 and €24 respectively as compared with the traditional banks' prices of from €32 to €40, and who had to reduce their issuing plans considerably (503).

Even suspended, the measures produce the desired effect of limiting the number of cards issued by the new entrants (504). The Groupement and the main members also stated that the measures would make it possible to prevent a reduction of the new entrants' prices to €15 to €20 (505). The Groupement's argument that the new entrants' prices are not significantly lower than those of the traditional banks confirms that the desired effect of preventing a reduction of the new entrants' prices was achieved.

Moreover, the measures are targeted at the new entrants, who are seen by the main members as a threat, and the main members receive [virtually all] of the payments made by the new entrants (506); so that the additional cost of paying for the measures is a burden on the new entrants and a gain to the main members, whether or not the additional cost is recovered in the selling price and whether or not consumers are sensitive to a difference in the price of cards. Given the scale of the sums to be paid as a proportion of the price of cards, even by the rare members liable to the Merfa who, as the Groupement argues, do not pay the maximum amount (507), the measures are a considerable competitive handicap to new entrants, and an advantage to the main members.

In order to show that the measures have indeed had an anticompetitive effect is not necessary that all of the liable banks should have increased their prices significantly at the same time, though some of them did: the effect also took other forms. In line with the purpose of the measures, some liable banks increased their prices (508), while others decided to issue fewer cards (509) or to reduce the benefits associated with the card (510).

As for the Groupement's argument that the price charged by the traditional banks is highly competitive given the services associated with their cards, whereas the new entrants' prices are not competitive, because their CB cards are essentially credit cards whose real price to the cardholder includes the very high cost of credit in particular, it should be pointed out:

- that the additional cost generated by the measures is a handicap in itself, whether or not the prices charged by the different classes of bank (traditional bank or new entrant) are competitive;

- that the traditional banks themselves issue credit cards;

- that the grant of credit is also a cost to the bank and a service to the cardholder, who will not necessarily make use of the credit facility associated with the card;

- that the Groupement's argument is merely an assertion on its part: the Groupement does not provide any clarification of the nature of the services alleged to vary from one bank to another (that is to say services separate from those making up the CB added value, which are common to all CB cards including those of the new entrants, such as interbanking, universality and security);

- that the Groupement does not show why the price charged by the traditional banks should be seen as "very competitive" in relation to these services.

343. Turning to the Commission's alleged failure to consider the role of criteria other than price, such as the CB card's qualities in terms of universality, security and reliability, and the sensitivity of cardholders to variations in prices, the following points need to be made.

(1) As the measures at issue all take the form of a sum to be paid per CB card issued, the new entrants' cards penalised by the measures have the same CB qualities, in terms of interbanking, universality, security, etc., as the CB cards issued by the main members. Thus the qualities of the CB card cannot justify the penalising effect of the measures at issue, or reduce its importance.

(2) The Groupement's reference to the existence of unspecified services distinct from than the qualities common to all the CB cards issued by the various institutions (511) does not in any way invalidate the demonstration of the existence of anticompetitive effects.

(3) It is not true that the Commission has concluded that there are restrictive effects on competition on the sole basis of the measures' impact on the price of cards, or that it assumes that consumers are sensitive only to this one factor of differentiation between the products offered. The Commission maintains that the additional cost imposed on new entrants by the measures at issue has effects other than its impact on card prices. In particular, the Commission draws attention to the inhibiting effect the measures have on the volume of the card issuing plans of new entrants, and indicates that these cards, issued by Internet banks or linked to large retailers, have new or special functions such as the combination of payment and loyalty cards or cash back, of which consumers will consequently be deprived (either directly, because the number of such cards has had to be scaled down, or indirectly, because the additional cost induced by the measures and the need to maintain reserves (512) affect capacity to invest in other factors of competition such as the functions and services associated with the cards).

B. Effects produced since the suspension of the measures on 8 June 2004

344. As explained above (see recital 259), the Groupement's Board of Directors decided, on 8 June 2004, to suspend the Merfa, the membership fee and the dormant members wake-up fee "pending the Commission's decision. The Groupement will then take a decision as to any amendment or repeal" (513). According to the Groupement, the sums which "would have been" payable by the banks governed by the measures in question will be required to be paid if the Commission subsequently declares the measures to be compatible with Article 81 of the EC Treaty (514): the members liable will then have to pay the sums due not only in respect of the years following a lifting of the suspension, but also in respect of the years preceding it.

345. Not only did the measures affect the market before their application was suspended: they continue to do so, in so far as the suspension may be lifted. There are two reasons for this.

(a) First, there are considerations of prudence, because the Groupement has maintained the uncertainty as to whether the suspension of the measures may be lifted, a step that would require only a decision on the part of the Board of Directors. The new entrants have been issuing fewer cards than they would if they were sure that the measures had been cancelled definitively, because if the suspension is lifted they will have to pay the sums due on the basis of all the cards they have issued from 2003 onward, which would increase the cost of issuing the cards very substantially. The new entrants have consequently been building up reserves (515), and cannot use these resources to compete with the main members and offer prices as low as possible; and they have continued to issue fewer cards than they would wish. This emerges very clearly from the replies to the Commission's request for information (516).

(b) The measures also have an inertia effect, because decisions to reduce the volume of cards issued and not to charge prices as low as had been projected that were taken before the measures were suspended continue to have effect after the suspension.

(1) The effects on issuing plans

346. As a result of a form of inertia, the effects of a reduction decided before the Merfa was suspended persist despite the suspension of the Merfa. Although the new entrants targeted by the measures took the decision to reduce their issuing plans before the measures were finally suspended, their decision was for a duration going beyond the date of the suspension. For example:

- According to Banque Accord, "Since the end of December 2002 we have had to scale back our bank card issuing objectives to [XXX XXX] cards for 2003-2006, a difference of [XXX XXX] cards compared with the issuing plan anticipated before the introduction of the Merfa" (517).

- GE Money Bank reduced its card issue targets substantially, by about 70% for the period 2006-2006 (518).

- The reduction in S2P's car issuing plans, from 1 000 000 cards to about 150 000, by about 85% in other words, continued after the measures entered into force on 8 June 2004. On 31 December 2004 S2P had issued 132 043 cards, which was 13% of what it had been aiming at before the measures were approved. In July 2005 S2P emphasised that the effects of the Merfa would persist for several years: "We have therefore revised our marketing and communication plan so as to remain below the threshold for the next two to three years" [the threshold of 450 000 to 500 000 cards that would trigger the Merfa in S2P's case] (519).

347. As a precaution, the new entrants have been moderating their ambitions for the issue of cards, because the measures have only been suspended, and the suspension may be lifted, so that they are not sure whether the measures may be applied in future:

- GE Money Bank has clearly explained the restrictive effect produced by the Merfa at a time after it had been suspended, as a result of the prudence exercised by the banks liable to the Merfa: "We would like to make it clear that the suspension ... leaves the firms liable to the Merfa in a position of some legal uncertainty. The suspension does not mean that the sums will never be due under any circumstances. The banks liable to the Merfa could be obliged to pay ... It is only to be expected, therefore, and is in line with the requirements of accounting law, that the banks should set aside reserves for the payment of the Merfa. That is what GE Money Bank has done" (520). As well as building up a reserve for payment of the Merfa, therefore, GE Money Bank has stuck to its strategy of reducing card issuance and has cut by half its bank card issuing targets compared with the number of cards it would have issued had the Merfa not been applicable: "Admittedly, the Groupement's Board of Directors has decided to suspend implementation of the Merfa, but we do not know how it will ultimately be applied. In these conditions of total uncertainty, GE Money Bank is maintaining its restrictive CB card issuing strategy, which is essentially limited to withdrawal cards .... In 2005 we intend to issue 18 000 new CB withdrawal cards. That is half the target we might envisage if the Merfa were not applicable" (521).

- Banque Accord states that it has issued only 48% of the cards it had intended to issue, and that this is due to "the reduction in our commercial investment: our shareholders have asked us to be prudent in view of the impact of the measures taken by GIE Cartes Bancaires on this activity (522) ... If it [the Merfa] were to be implemented, this would call into question our development model, which is based on the conversion of our portfolio of proprietary cards into bank cards" (523).

- S2P says that until the Merfa has been abolished definitively it will have to go on limiting the number of its cards to keep it below the threshold at which it would be liable to the Merfa: "If the Merfa were to be maintained, we would limit our programme to 500 000 cards, a figure which, it should be said, constitutes our untaxed ceiling ..." (524);

- Cofidis has stated that the introduction of the measures would have very negative consequences for its development strategy: "if the Groupement were to replace the Merfa by other measures with the same effects, we would have to review our development strategy, as nowadays many partners ask us to extend the service we offer them to include the use of the CB card" (525).

- At the hearing of 13 November 2006, Monabanq (formerly Covefi) and Banque Accord stated: "we have had to make big cuts in the programmes we had launched. The number of cards we issued grew by only 15% in 2004, 2005 and 2006 - and this will be the case again in 2007 - whereas we had hoped for much more rapid progress. Whether or not we are asked to pay it, therefore, the Merfa has made things difficult for us, is still making things difficult for us and will go on making things difficult for us until the situation is finally clarified" (526).

(2) The effects on the price of cards

(a) On new entrants

348. It was explained in section B.1 that during the period before the suspension of the measures, as the Groupement and the main members participating in the COM had planned, the new entrants were unable to issue cards at a price of €15-20. Since the measures were suspended they have continued to produce effects, as new entrants have had to keep their prices above this level or even increase them.

349. Barclays had to continue to increase its prices despite the suspension of the measures, and the price of its Visa card went from €43 in 2004 to €44 in 2005 (527).

350. AXA and AGF also had to increase their prices. AXA Banque's immediate debit CB Visa card went from €24 in 2004 to €30 in 2005, and its deferred debit CB Visa card went from €30 in 2004 to €36 in 2005 (528); AGF's immediate debit CB Visa card went from €30 in February 2004 to €32 from January 2005 (529).

351. The statements of the other banks show that they had to keep their prices at a level above the €15-20 feared by the Groupement and the main members:

- Citibank offered its Visa card at €35 in 2004 and 2005 (530);

- The price of GE Money Bank's CB Visa card was held at €23, the first year being free (531);

- after the measures were suspended Banque Accord continued to offer its CB Visa, CB MasterCard and CB Visa Premier cards at €25, €25 and €45 respectively (532);

- Covefi did not modify its prices following the suspension of the measures and since 2002 has marketed its cards at €22.50 (533);

- Cofidis continues to sell its card at €24 (534); and

- ING still markets its international CB Visa card at €40 (535).

352. It follows from the above figures that, even after the measures were suspended, new entrants continued not to offer their cards at the lower prices lower which, according to the Groupement and the main members, they would have offered in the absence of the measures (536).

(b) On the main members

353. After their suspension the measures at issue also affected the prices of the cards of the main members. After the measures were suspended the main members did not have to reduce the prices of their cards, as they and the Groupement feared they would have to do in the total absence of the measures as a result of the contagion effect of the prices of cards issued by new entrants (see in particular recitals 71, 79, 110(a), 126(a), 241, 242 and 302(c).

354. Several main member banks, such as La Poste, Crédit Agricole and CIC, increased the prices of their cards after the measures were suspended on 8 June 2004:

- La Poste's immediate debit CB Visa card rose from €32.50 in 2004 to €33.50 in 2005 (it had already risen by comparison with 2003, when it cost €31.50) (537).

- The CB classic Visa card issued by Caisses Régionales du Crédit Agricole rose from €43.60 in 2004 to €44.07 in 2005 (it had already risen by comparison with 2003, when it cost €31.50) (538). Nor did Crédit Agricole's subsidiaries reduce the price of their cards following the suspension of the measures. Thus, Crédit Lyonnais continued in 2005 to offer an immediate debit international Visa card at €34.50 (the same price as in 2003 and 2004) and a deferred debit international CB Visa card at €43.50 (the same price as in 2003 and 2004) (539). Sofinco maintained the price of its classic Visa card at €24 in 2005 (540). Finaref also maintained the price of its Visa cards at €24 in 2005 (541).

- The immediate debit CB Visa and CB MasterCard cards issued by Crédit Industriel et Commercial rose from €33 in 2004 to €34 in 2005 (542) (they had already risen from €32 in 2002 to €33 in 2003, the year the measures entered into force).

355. Nor did the other main members have to reduce their prices after the measures were suspended on 8 June 2004: they held them steady:

- BNP Paribas maintained the price of its CB classic Visa cards at €35 in 2004 and 2005 (before the measures were suspended it had increased the price from €34 in 2003 to €35 in 2004) (543).

- Natexis Banques Populaires did not have to lower the price of its CB classic Visa/CB MasterCard, which it has issued at €40 since 2001 (544).

- After the suspension of the measures Société Générale did not have to reduce the price of its CB Visa and CB MasterCard cards, which remained at €32/42 from 2003 onward (having risen from €31/41 in 2002 to €32/42 in 2003, the year the measures entered into force) (545).

- The price of the immediate debit CB classic Visa card issued by Caisses d'Épargne did not fall after the suspension of the measures, remaining at €32.50 (it had increased before the suspension of the measures from €31.50 in 2003 to 32.50 in 2004) (546).

356. It follows that after the measures were merely suspended, the main members were able to maintain and even to increase the prices of their cards.

357. It should be borne in mind, too, that the measures had effects other than the reduction in issuing plans and the maintenance or increase in the prices of new entrants and traditional banks: the banks subject to the measures, which had merely been suspended, were obliged to constitute reserves (547), review their marketing and communication plans (548), cancel loyalty schemes (549), cancel plans to market their cards in their hypermarkets (550), and issue cards for withdrawals only, rather than payment and withdrawal cards (551).

358. Nor can the Groupement seriously claim that the Commission has not considered what the state of competition would have been in the absence of the decision by the association of undertakings (552): the Commission has said that in the absence of the measures the new entrants would in particular have issued more cards, at more competitive prices, enabling them to enter into competition with the big main member banks and obliging those banks to issue cards at prices determined by the free interplay of competition. This is corroborated by what was said at the meetings at which the measures were prepared (which were attended by representatives not only of the main members but also of the Groupement itself) and by the statements of the new entrants.

10.2.3. The restrictive effects of the measures are not outside the scope of Article 81(1) of the EC Treaty

359. In its notification the Groupement advances various arguments in support of its contention that the measures have no restrictive effect on competition in the market (553). According to the Groupement, the notified provisions are not liable to restrict the participants' freedom to take autonomous commercial decisions. Like all CB rules, the notified provisions are interbank rules intended to govern only the relations of members of the Groupement with one another. None of these rules is directed at relations between members of the Groupement and their customers, cardholders or merchants, or consequently liable to limit the commercial autonomy of its members.

360. The Groupement argues that the Merfa is a pricing measure that does not restrict competition in any way, for the following reasons (554).

- A payment system like the CB system can operate in the long term only if the issuance and acquiring functions are assured in a balanced fashion.

- The application of the Merfa levy to institutions concentrating on issuance is, moreover, justified on the ground that that activity generates fewer "positive externalities" than the acquiring function (555). The Merfa merely makes allowance for this economic reality, without however influencing the strategy of any of its members.

- The amount of the levy introduced by the Merfa is not likely to prevent a bank wishing to concentrate on issuance from participating in the CB system. It is merely consideration for the benefit that such a bank derives from the development by the other members of the Groupement of the CB card acceptance network for payments and withdrawals, which development enables it to carry on its issuance activities.

- The criteria governing eligibility for the Merfa are, moreover, defined objectively, on the basis of data declared by the members of the Groupement and verifiable by them, and consequently cannot lead to any discrimination between the members.

361. The Groupement further states in the notification that the new membership fee arrangements do not constitute a barrier to entry to the CB system and involve no restriction of competition:

- The amount of the membership fee is reasonable and proportionate both to the investments made by the Groupement's existing members in developing the CB system and to the advantage afforded to any new entrants by the fact of having access to the system.

- The introduction at the expense of new entrants of a membership fee based on issuance volume involves no discrimination between new entrants and the current members of the Groupement: the fee is merely remuneration for immediate access to the CB system, which has been developed by the current members through major technical and commercial efforts and investments, with the current members also having taken risks that new entrants will not have to take; it is rather the absence of any remuneration that would be liable to give rise to discrimination, by enabling new entrants to behave as free riders.

- The introduction of an equivalent fee payable by dormant members developing a large activity as from the entry into force of the new membership fee does not give rise to any discrimination between those members who have to pay it and the others. Quite the reverse, the members who will be subject to it, who are in a comparable, if not identical, economic situation to that of new entrants and who differ from them only in that they are formally members of the Groupement, will be treated like new entrants, for the same reasons.

362. The measures at issue are in reality intended to ensure that each member is rewarded for the efforts that that member has made in the interests of the others. The measures take proportionate account of the efforts made and benefits gained by each member, and are inherent to the operation of the CB system, in that they are essential to the development of interbank cooperation, constituting the very heart of the system. They are therefore in no way restrictive of competition.

363. These arguments advanced by the Groupement are concerned with the alleged existence of a free rider problem, the need for new entrants to compensate for or remunerate past investments, and the need to balance the functions of issuance and acquiring via an equilibrium to be attained by each member and considered indissociable from the argument regarding the alleged superiority of acquiring over issuance. The arguments are examined in section 11 below, on Article 81(3) of the Treaty.

364. But the Groupement is here arguing that the Commission has not shown that the measures have an anticompetitive object or effect, having failed to take account under Article 81(1) of the EC Treaty of the pro-competitive impact that the measures would have on the market in payment systems (intersystem competition) (556), in particular because they would counteract the alleged free rider problem and balance the externalities generated by the activities of issuance and acquiring.

365. The measures at issue cannot be considered to be ancillary restrictions.

366. The concept of an ancillary restriction falling outside the scope of the prohibition in Article 81(1) of the Treaty covers any restriction which is directly related and necessary to the implementation of a main operation, that is to say, which is objectively necessary for the implementation of the main operation and proportionate to it (557).

367. As the Court of First Instance held in Métropole, however, it would be wrong, when classifying ancillary restrictions, to interpret the requirement for objective necessity as implying a need to weigh the pro- and anticompetitive effects of a measure in the framework of Article 81(1) of the Treaty. According to the Court, when examining the objective necessity of a restriction in relation to the main operation, it is not a question of analysing whether the restriction is indispensable to the commercial success of the main operation but of determining whether, in the context of the main operation, the restriction is necessary to implement that operation. If, without the restriction, the main operation is difficult or even impossible to implement, the restriction may be regarded as objectively necessary for its implementation (558).

368. In the present case, neither the creation nor the survival of a card payment system such as the CB system would be jeopardised by the absence of the measures at issue, as can be seen from the success and steady development of the CB system over more than 22 years, the fact that it has continued operating even though the measures have been suspended since 8 June 2004, and the fact that other systems operate in Europe without anything resembling the Merfa. The argument as to the alleged necessity of the measures, and in particular of the Merfa, will therefore be examined in relation to Article 81(3) of the Treaty, in section 11 below.

10.2.4. Appreciable restriction of competition

369. The restrictions of competition described above are appreciable, first of all because membership of the Groupement, which is by far the largest payment system in France (see recital 17), is unavoidable for banks wishing to issue CB bank cards (559), and secondly because Visa and MasterCard cards issued in France are in practice CB cards (560).

370. According to settled case law, the fact that the undertakings concerned hold almost 40% of the relevant market would be sufficient proof of the appreciable nature of the restriction of competition (561). In the present case, since the institutions concerned hold a much larger market share (see recital 20, with for example a 78% market share in 2002), the appreciable nature of the restriction of competition is beyond doubt.

10.2.5. Trade between Member States is affected to an appreciable extent

371. According to the Court of First Instance, "Article 85(1) [now Article 81 (1])] of the Treaty does not require proof that [...] agreements [or, as in the present decision, decisions of associations of undertakings] have in fact significantly affected trade between Member States, which, moreover, is difficult to establish to a sufficient legal standard in most cases. It requires that it be established that the agreement or practice was capable of having that effect. The condition that trade between Member States be affected is satisfied where it is possible to foresee with a sufficient degree of probability on the basis of a set of factors of law or fact that the agreement or practice found to exist may have an influence, direct or indirect, actual or potential, on the pattern of trade between Member States" (562). In any event, Article 81(1) of the Treaty does not require that the agreements it refers to must have appreciably affected trade between Member States, but it does require that it be established that the agreements are capable of having that effect (563).

372. In the present case there is indeed an appreciable effect on trade between Member States. As shown in recitals 304 to 309, the measures at issue have (or, at least, have potentially) an insulating effect on the French market for payment cards. This insulating effect is in addition to the fact that since 1995 any issuer of bank cards for use primarily in the framework of the CB system is obliged to become a CB member and comply with the rules of the Groupement (564). Several foreign banks are already members of the CB system and other foreign banks will be joining it. These foreign banks will be affected by the measures in question.

373. The Groupement's decision accordingly affects trade between Member States to an appreciable extent, a fact which is not disputed in the notification submitted by the Groupement (565).

10.2.6. Conclusion on the applicability of Article 81(1)

374. It follows from all the above that the measures at issue are contrary to Article 81(1) of the EC Treaty.

11. ARTICLE 81(3) OF THE EC TREATY

375. The decision by an association of undertakings at issue here, being anticompetitive within the meaning of Article 81(1) of the EC Treaty, is automatically void pursuant to Article 81(2) unless Article 81(3) is applicable.

376. Article 81(3) of the EC Treaty stipulates that Article 81(1) may be declared inapplicable to any agreement which contributes to improving the production or distribution of goods or to promoting technical or economic progress, while allowing consumers a fair share of the resulting benefit, and which does not impose on the undertakings concerned restrictions which are not indispensable to the attainment of those objectives, or afford such undertakings the possibility of eliminating competition in respect of a substantial part of the products in question.

377. It is for the parties to an agreement - and, in the case of a decision by an association of undertakings, for the association concerned - to show that it qualifies for the applicatio of pargraph 3 of Article 81 (566), and to produce any evidence that might substantiate the economic grounds put forward for such exemption (567). Moreover, all four tests in Article 81(3) must be satisfied if the measures in question are to benefit from the legal exception under it (568).

378. The Groupement has explained why it takes the view that the tests for the application of paragraph 3 of Article 81 of the EC Treaty are satisfied (569) and submitted economic studies in support of its arguments (570).

379. However, after examining these arguments in detail, the Commission has concluded that the four tests in Article 81(3) are not all satisfied.

11.1. The first test in Article 81(3) of the EC Treaty: no contribution to improving the production or distribution of goods or to promoting technical or economic progress

380. According to the Groupement, the measures in question contribute to economic and technical progress in so far as without them the CB system would face the threat of free riding, resulting in a slowdown in investment and in the financial, technical and commercial efforts which are indispensable if the system is to be maintained and developed (571).

381. The Groupement cites the need to combat free riding as justification for the measures as a whole, while arguing that the Merfa specifically helps to combat it by "encouraging those members that are more issuers than acquirers to develop their acquiring activities" (572) "in order to ensure that the system is balanced" (573) as "a payment system like the CB system can operate in the long term only if the issuance and acquiring functions are equally assured" (574). The Merfa is said to be "the best way" of widening acceptance of the CB card in sectors where acceptance can still be developed (575), which on the issuance side is likely to result in "new types of CB cardholder or ... more intensive use of existing CB cards" (576).

382. The Commission does not in any way deny that competition law allows steps to be taken to combat free riding. However, in this case free riding has not been shown to exist, and even if it had been the measures are not suited to combating it.

383. The justification for the measures given by the Groupement lacks clarity owing to interchanging its different types of argument which are mutually incompatible (see section 11.1.1 below). Nevertheless, in order to make a structured analysis, the Commission will distinguish between:

- free riding in the form of new entrants' benefiting from investments made (or from the "value" brought to the system) by the other members (see section 11.1.2 below); and

- the arguments related to the Merfa concerning the alleged superiority of acquiring over issuance (see section 11.1.3 below).

In any event, there is no free riding (see section 11.1.4 below).

The economic justifications other than that of combating the alleged free riding are likewise not proven (section 11.1.5) and the measures have negative economic effects (Section 11.1.6).

11.1.1. The argument based on alleged free riding lacks clarity

384. The description of the alleged free riding tends to vary. According to the Groupement, the measures address a problem of free riding both:

- in that they consist in remuneration for the value of the system (or that afforded by it) ("first source" of free riding); and

- in that they are a mechanism for balancing the acquiring and issuing functions ("second source" of free riding).

385. According to the Groupement, "the value of the CB system ... and the proven superiority of the positive externalities generated by acquiring over those generated by card issuance" constitute "two separate sources of free riding" (577).

386. With regard to the "first source" of free riding, the Groupement refers to the different concepts of investment costs, risks, efforts and "immediate advantages", both past and future (578).

387. As is explained below, the Groupement has shown neither that there is any need for offsetting the cost of past investments not yet fully amortised (i.e. past investments, or the part of them, which still have value and are still used by new entrants joining the system, while other past investments can no longer be the target of any free riding) nor that there is any value derived from using the system that is not taken into account in the interchange fees paid in respect of each transaction.

388. The "second source" of free riding is said to derive from the fact that new entrants mainly involved in issuing benefit from the positive externalities generated on the issuing side by the acquiring activities of the banks which created and developed the system, without the new entrants' issuing activities generating equivalent externalities on the acquiring side. This justification, expressly put forward in support of the Merfa, also seems to underpin some of the arguments in support of the other charging measures (the membership fee and additional fee, or the dormant members wake-up mechanism), while arguments borrowing from the "first source" of free riding are also adduced in connection with this "second source" of free riding (579).

11.1.2 The alleged free riding on investments or on the "value derived from using the system" ("the first source" of free riding) is not proven

389. The argument as to alleged "free riding" on investments or on the "value derived from using the system" lacks clarity of definition and assessment (section 11.1.2.1.). Moreover, several reasons militate against justifying the measures by an alleged need to "protect" investments (section 11.1.2.2.).

11.1.2.1. The argument based on alleged free riding oo investments or on the "value derived from using the system" lacks clarity

390. The Groupement does not state clearly what the free riding on investments or on the "value derived from using the system" consists of, simply claiming that the measures are remuneration in some instances for past investment costs (increased or not as the case may be by the risks taken by the founder members (580)), in others for the "efforts" made by non-fee-paying members (581), and in others still for the "immediate advantages" afforded by access to the system (582), without specifying the exact meaning of each of these concepts in the present case. The measures are also said to be justified variously in terms of the past and in terms of the future: justified in terms of the creation and past development of the system, when the Groupement argues that the measures compensate for the investments (costs, efforts, risks, etc.) made by the founder members in setting up and developing the system, but justified also in terms of its future development (583), or of its very survival (584), there being a serious risk of the system's collapsing (585).

391. The description of free riding remains confused in the Groupement's comments on the Commission's Statement of Objections of 18 July 2006. For example:

In response to the Commission's argument (set out in the preceding recital) that there was no clear definition of what the alleged free riding consisted of, the Groupement undertook in its comments of 2006 to produce a single definition combining the various concepts which it had thus far scattered throughout its arguments:

"the free riding which the Groupement's decision is designed to combat results from quasi-free access to the CB system, which allows new entrants to derive immediate benefit from the advantages that the system offers thanks to the investments (or efforts, the terms being synonymous) of the members that preceded them. These investments and efforts made by the Groupement's members since the creation of the CB system 20 years ago are reflected in the value which it is recognised as having today, as reflected in the scale of the immediate benefits sought by those who wish to have access to it" (586).

The definition of what precisely the free riding consists of nevertheless remains obscure in so far as the Groupement still does not define the concepts used in the above definition and cites two separate "points of reference", attaching greater importance to the value derived from using the system "in the eyes of those who wish to have access to it" than it does to the argument relating to investment:

"The investments made to develop the CB system are one point of reference, but not the only, or the main, one. It is the value of the system in the eyes of those who wish to have access to it that is the primary reference" (587).

The Groupement gives no explanation for this distinction and ranking, and fails to set out precisely how these "point of reference" are the target of free riding (still less how to reconcile the two concepts of free riding based on using the system and free riding on investments), confining itself to alleging that access to the system is quasi-free, without going into any details to support this argument.

392. The Groupement remains very vague when it comes to assessing the scale of the alleged free riding. According to the estimates of [Groupement's consultant], the Groupement's economist, the total amount of the "development costs" borne "in connection with the CB system since it was set up", comes to close on €4 billion, a figure which covers not only the costs borne "by GIE Cartes Bancaires 'CB' but also those borne by "the member banks" (588) (thus including new entrants which are already members). However, [Groupement's consultant] specifies that "this figure should probably be reduced because, while part of the cost corresponds to expenditure benefiting the CB system as a whole, part of it also directly benefited the institutions concerned without necessarily having an impact on the CB system as a whole" (589).

393. Neither the Groupement nor its [consultant] actually indicates the value of the investments in the system allegedly targeted by the free riding by distinguishing them clearly from the costs benefiting directly or primarily the banks themselves (or partially the other banks, but not the system proper). The Groupement does not specify what data were used (that is to say, what the figures cover) or what method was employed in order to obtain the figure of €4 billion.

394. In this connection, in reply to a request for information from the Commission asking them to indicate the amount they had invested in CB interbanking, a number of main members stated that it was impossible to reply, because the data were not available, and/or stressed that the data transmitted were pure estimates; moreover the criteria and the estimation methods used by the different banks varied considerably (590). These indications and differences show the lack of valid uniform criteria enabling the Groupement to distinguish at the level of its members between investment which is devoted to interbanking and that which is not.

395. The impossibility of producing a valid estimate of the investments in the system is confirmed by the Groupement itself, which states: "As it is the Groupement's policy not to interfere with this freedom of choice, it has no way of assessing the costs borne by its members in connection with their CB activities, except where it is informed accordingly. This being the case, determining average, marginal or incremental costs from the individual costs of its members, apart from the fact that it is not part of its remit, is strictly impossible" (591)..

396. Nevertheless, in its reply to the 2006 Statement of Objections, the Groupement refers to a "value derived from using" the system, which is said to vary from one bank to another ("it is the value of the system in the eyes of those who wish to have access to it that is the primary reference") (592), which for a given bank is said to be at least equal to the difference in price (annual fee) of its CB card and its proprietary card (593).

397. The Groupement thus appears to take the view that the alleged free riding corresponds to the difference between a "value derived from using the system" and the lower value of certain members' contribution to the system, and that the free riding can be measured by the difference between the price of proprietary cards and that of CB cards issued by the same bank. The Groupement then concludes solely on the basis of the "value of using the system" that the current membership fee is inappropriate and ought to be increased (594). Having decided that the membership fee should be increased, the Groupement then concludes that banks which are already members but which are not active enough ("dormant members") would be free riding on the value derived from using the system if a pricing measure were not also applied to them, namely the dormant members wake-up mechanism (595).

398. However, there is no evidence that the difference in price between the proprietary card and CB card of a given bank corresponds to the difference between a "value derived from using the system" (which the Groupement neither defines nor calculates) and the allegedly lower value of certain members' contributions (which the Groupement likewise does not define, still less show how it is actually lower than the value derived from using the system). In any event, the Groupement does not explain why or how the "value derived from using the system" varies from one member to another, nor does it prove its other statements, and it leaves a number of questions unanswered. For example, the Groupement does not explain how the value of the system depends, for a given bank, on whether it already had a stock of proprietary cards and on their price, or how the price difference reflects the value derived from using the CB system despite the fact that belonging to the system also entails costs, which in all likelihood are higher than the costs of a proprietary card system.

399. Nor does the Groupement explain the change in the membership fee from a fixed entry fee to an amount per card issued during a three-year period (which comes in addition to an increase in the fixed entry fee which this decision does not contest).

400. As regards the alleged free riding practised by "dormant" members:

(1) It is contradictory to link the free riding concerned (that related to the "value derived from using the system") to use of the system on the one hand, and on the other to argue that "dormant" members are free riders because during the years preceding the introduction of the measures they did not make enough use of the system.

(2) The Groupement describes the contribution of the "dormant" members to the costs of the system as insufficient on the basis of the volume of their CB activity, but does not specify what that contribution is. The only contributions based on the volume of activity in the system are the interchange fees paid by each member, including those classed by the Groupement as "dormant", on each acquiring transaction (the CIP being paid by the acquiring bank for each card payment to a merchant) and on each issuance transaction (the CIR being paid by the issuing bank for each withdrawal from another bank's ATM).

However, the Groupement also argues in its comments of 2006 that the CIP and CIR interchange fees, since their "sole purpose is to remunerate the services which the banks provide each other" (596), have nothing to do with a member's contribution to the issuance/acquiring activities within the system. The Groupement does not identify the "contribution to the costs of the system based on the volume of CB activity" with which it justifies the "dormant members wake-up mechanism", and if the interchange fees are meant, this would contradict the Groupement's arguments regarding those commissions.

401. The Groupement does not see any need to take account, in referring to free riding based on the value derived from using the system, of the fact that such value is already at least partially covered by the interchange fees paid in respect of each transaction. It argues that the use of the system, which is said to be subject to free riding, is a concept quite separate from the services paid for by the interchange fees (597).

402. However, in using the CB system, each member contributes to the system's operation, something that the Groupement appears to ignore, and at the same time benefits from a number of services, which it pays for by means of the interchange fees applicable to each transaction: processing payment transactions, collective security measures, guaranteeing payments, and the immobilisation of funds paid to the holder making a withdrawal are all services linked to use of the CB system already paid for by way of the interchange fees.

403. The Groupement uses contradictory concepts, fails to define what the concept of use of the system is meant to cover, and fails to indicate any methodology for determining its value or to provide any estimate of that value. It accordingly has not shown that there is any free riding targeting the value derived from using the system, nor has it shown that the services already paid for by way of the interchange fees, and therefore not subject to free riding, have been duly excluded from the "value derived from using the system".

11.1.2.2. There are several reasons that militate against justifying the measures (in particular the membership fees that vary according to the number of cards issued) by a need to remunerate investments

404. Despite the absence of a definition and of a serious estimate of the investments or "value derived from using the system" allegedly targeted by the free riding, and despite its admission that it has "no way of assessing the costs borne by its members in connection with their CB activities, except where it is informed accordingly" (598), the Groupement argues that the measures (and in particular the membership fees that vary according to the number of cards issued) are necessary to "protect" past and future investment, failing which the CB system would "collapse".

A. The Groupement cannot justify the measures by costs far exceeding those borne in the interests of the system

405. Although the Groupement has avoided giving any credible estimate (despite the burden of proof lying with it), there is no doubt that the cost of the investments strictly attributable to the CB system proper is very much lower than all of the costs which the various players may have borne, under one head or another, as part of the numerous activities attributable to electronic banking, but the Groupement seems to have included these costs wholesale in its €4 billion estimate of the costs of investment in the system ("the costs of developing the system since it was set up").

Thus, certain documents obtained on the premises of the Groupement refer to:

- an "investment value" of the Groupement of FR 232 million (€35.5 million) covering "studies", the RCB/RSB network and "other fixed assets" (599), and

- a "GIE CB cost: the bank's participation in the GIE since 1986" of between €172 and 472 million (600), covering "non-security-related costs and investments [and] 100% of the cost of developing the RCB network" (i.e. a sum representing no more than between 4.3 and 11.8% of the €4 billion).

406. Besides the cost of investments attributable to the establishment of the system proper, many other investments by the members seem therefore to have been included in the concept of investments targeted by free riding. These other costs include, for example, the cost of employing IT staff in banks and all the costs related to the installation of ATMs, the manufacture of cards, electronic payment terminals, containing fraud rates and guaranteeing payments to merchants (601).

407. Such costs cover activities benefiting only the banks concerned (and hence not necessarily related to the system and to "interbanking", as they benefit neither the system itself nor any other bank). Thus, while IT staff are not employed exclusively in doing interbank electronic banking work, and a not inconsiderable number of withdrawals are made with cards issued by the same group of banks as that which manages the ATM used, the costs of installing ATMs, manufacturing cards, combating fraud and guaranteeing payments have already given rise to the payment of cardholder fees and merchant commissions charged to banks' customers. The Groupement is ignoring the fact that these costs are offset - if not entirely, then at least largely - by the income from cardholder fees and merchant commissions, which is estimated by the Groupement at around €2.3 billion (602).

408. The Groupement's estimates even include investments not made by the CB members, since it is merchants who have to finance the payment terminals used.

409. The Groupement ignores the fact that the new entrants also carry out investments (in some cases very heavy ones (603)) and contribute to the Groupement's operating costs. The Groupement does say that "the contribution of new members and 'new entrants' to present and future investments, on the same footing as all other Groupement members, does not constitute a contribution to the past investments in the CB system from which they derive immediate benefit simply on joining or, in the case of dormant members, on increasing their CB activities". Section B below addresses the justification on the grounds of remunerating past investments. However, it should be noted at this point that (a) the Groupement sets no limit on the past investments from which new entrants are said to derive benefit (although many such investments are already amortised and, having already been replaced, have not been of any benefit to new entrants; see recitals 411 and 416 below), (b) several new entrants, in particular those targeted by the measures at issue, had already contributed to past investments as they were already members of the Groupement and (c) banks remunerate other banks in the system by way of the interchange fees, thereby also remunerating the past investments of the banks which receive such commissions.

410. The investments in the CB system proper are limited to the activities for which a card payment system must take responsibility. However, according to the Groupement itself, "the GCB's scope is more limited than that of most of its European opposite numbers", with the Groupement's activities covering only the routing of payment and withdrawal transactions, whereas other systems are also responsible for numerous other activities in which the Groupement takes no part (604). Even though the Groupement performs fewer activities than other systems, the other card payment systems in Europe operate without measures comparable to the measures at issue in this case.

411. Moreover, no limit seems to have been applied with respect to how long ago the past costs included in the €4 billion investment estimate were incurred (the figure of €4 billion is said to cover "all costs in connection with the CB system since it was set up", which was in 1984), although a number of past investments must already have been amortised in that they have already been replaced or have become obsolete or the worse for wear and are therefore no longer usable by new members of the system.

412. Besides the estimate of €4 billion, [Groupement's consultant] makes two other types of estimate:

- The first is an estimate of profit per card per year, obtained by dividing the estimated "net profits" from issuing activity (€ [...] million in 2001) (605) by the number of CB cards ([...] million), which gives "some € [...] per card per year. Assuming ... that customer relations have an average 'lifetime' of more than five years, the overall profit generated by issuing a card is greater than € [...]."

- The second divides the amount spent on advertising by the bank Egg (€ [...] million) by the number of cards it expected to issue ([...] million), which for [Groupement's consultant] meant "that for Egg, issuing a CB card 'is worth' at least € [...]".

413. These two other estimates are not based - any more than the first one was - on reasonable methods. They do not concern all past "development costs" or the value of the system proper. Neither the average of members' net profits nor the advertising expenditure of one of them relates to the cost of setting up and developing the system itself. The fees at issue cannot be justified on the basis of such cursory estimates of an alleged "value of joining" (606), which are more a matter of improvisation than of real economic study. Furthermore, taking as the point of reference in justifying the amount of the membership fees factors such as the advertising costs incurred by new entrants is tantamount to allowing the Groupement to tax such new entrants on the basis of the costs they consider it necessary to incur so as to be able to enter the market.

B. The remuneration of past investments or the "fair return on investment"

414. The Commission considered whether the remuneration of past investments that had not yet been amortised could justify the measures at issue, and in particular the membership fees that varied in accordance with card issuance activities.

415. As indicated at recitals 393 et seq. above, the Groupement does not demonstrate which past investments should be remunerated by new entrants, even though many such investments, since the creation of the Groupement in 1984, have lost all or some of their value for any new entrants in that they have been replaced or have become obsolete or worse for wear. This consideration is enough in itself to render the argument regarding the alleged need for new entrants to remunerate the past investments of the main members irrelevant from the point of view of the contribution to any technical or economic progress.

416. However, in the interests of clarity, the Commission would point out that the alleged problem of free riding cannot exist only with reference to the past, without any dynamic perspective. The fact that investments took place in the past indicates that free riding did not pose any economic obstacle when the investments were made. The argument of the "fair return" on investment is irrelevant in so far as the Groupement has failed to show that the system's "traditional" members have not already obtained a remuneration, or that they have contributed to the CB system more than the new entrants have, while it acknowledges that part of the investments by traditional members served to develop their own business (607). As explained in section 11.1.4. below, the Commission does not believe either that a problem of free riding has emerged or could emerge subsequently.

417. Unrelated to remuneration for any service (this being, where appropriate, the role of the interchange fees, as the Groupement acknowledges (608)), payment by a new entrant of the fees at issue amounts to the payment of a fee for acquiring market share, especially since the remuneration increases as the new entrant issues cards and thus contribute to the expansion of the system.

418. The Groupement makes the following comments on the above arguments (sections A and B):

(1) The Commission, it says, considers it illegitimate to remunerate by way of a membership fee past investments that have not been amortised (609).

(2) The Commission is wrong to take the view that investments that have already been amortised should be excluded from the estimate of the investments in the system from which new entrants derive benefit, as even investments amortised in full still have a real value (the Groupement argues that amortisation is purely an accounting concept) (610).

(3) The Commission requires a calculation of the exact value of the investments in the Groupement, which means that in the Commission's view the value of a system is exactly equal to the cost of the investments made to set it up and develop it, and that the membership fee should reflect the investment costs exactly (611).

419. In reply to these comments, the Commission would make it clear that:

(1) The Commission is far from considering that remuneration for an investment used by third parties is illegitimate. But in this case the Groupement confines itself to alleging that all the costs borne by it since its creation amount to €4 billion, without showing that this estimate is correct and that it corresponds to past investments in the system which are actually the target of free riding by new entrants.

(2) As the Groupement has failed to give any evidence showing that the figure of €4 billion actually corresponds to investments that are the target of free riding, the Commission takes the view (for the reasons set out above) that the amount includes costs not borne in the interests of the system, and costs corresponding to investments that have already been replaced or have become obsolete, so that they cannot be included in a calculation of the investments from which new entrants derive benefit; in any event such investments have not been shown to be the target of the alleged free riding given that the new entrants at which the measures were directed were already members of the system before the measures were adopted, and have therefore also made "past" investments, and that the new entrants also contribute, by way of their investments, to the establishment and continuity of the system.

The Commission is not arguing that investments amortised in accounting terms have no economic value whatsoever. But the accounting value of an asset after amortisation at least shows how its value has dropped as a result of wear and tear.

The Commission would have been happy to receive an estimate of the value of the system (from which both the big, main-member banks and the new entrants derive benefit) that was more reliable than an estimate using the accounting concept of amortisation to take account of wear and tear and obsolescence of the investments.

However, the estimate produced by the Groupement, which puts the investments in the system from which new entrants derive benefit at a total of €4 billion, without any indication of the data or the method used to arrive at this figure, falls far short of the rigour and objectivity of the accounting method.

(3) The Commission does not require the value of the investments in the Groupement to be calculated exactly, nor does it require the membership fee that varies according to the number of cards issued to reflect the exact investment cost. It would be satisfied with a sufficiently serious estimate of the value of the benefits of the system for the members and an explanation of what the main-member banks contribute already in comparison with new entrants that serves to demonstrate objectively that there is free riding on the part of the latter. The Groupement has provided neither such a demonstration nor the evidence that make it possible.

C. The protection of future investments and the risk of "collapse" of the system

420. The Groupement argues that if the anti-free-riding measures are not introduced, further investments might be abandoned (612) and certain members might leave the CB system to join or set up a competing system, thereby exposing the CB system to the risk of "collapse": "By demonstrating that the notified measures are needed in order to prevent the development of free riding, the Groupement considers that it has demonstrated the risk of collapse it faces ... The members of the Groupement which have contributed [to the investments] and contribute the most will therefore necessarily be tempted to leave if free riding is tolerated" (613).

421. However, the fact that the members are continuing to invest in the system is evidence that there is no free riding such as, according to the Groupement, would "[cause] investment to dry up as those whose efforts are subject to free riding are induced to abandon them" (614).

422. Moreover, far from providing evidence of the risk of the system's collapsing, the Groupement confines itself to inferring such a risk from the existence of free riding on the part of the new entrants that will induce the main members to leave the Groupement, whose size and financial capacity would therefore be undermined to the extent that its very survival would be threatened (615).

423. The Groupement fails to provide valid proof of the existence of such free riding, and merely states that the additional investment by new entrants is proportional to their activity in the system and that it is minimal (investment by all non-main-members, of whom only a minority are the new entrants targeted by the measures at issue, is said to account for less than 10% of investments in the system) (616): this is not a sufficient basis for holding the new entrants responsible for a risk that certain main members might leave the Groupement, nor for infering the existence of such a risk.

424. Lastly, the Groupement argues (a) that the introduction of the single euro payments area (the SEPA), combined with the continuation of free riding if the measures at issue are not introduced, will act as a strong inducement to CB members to abandon the system for the Visa or MasterCard systems, which are already in a position to offer services that comply with the SEPA rules (617); and (b) that "several other national payment systems have disappeared or are on the point of doing so" (618).

425. As regards the Groupement's first argument:

- the measures were not designed to counter an alleged incentive to leave the Groupement, as the facts in this case show;

- as they are all members of the Visa or MasterCard systems, whose services comply with the SEPA rules, the Groupement's members do not need to leave the latter in order to offer SEPA-compatible services; and

- the Groupement cannot justify the measures by reference to the introduction of the SEPA since, even if the Groupement's hypothetical inability to comply with the SEPA rules in time were potentially to cause banks to leave it, such departures would be due not to the alleged free riding (the existence of which the Groupement fails to demonstrate, as it fails to demonstrate that the measures at issue are apt to counter it) but to the incompatibility of the CB system with the SEPA rules.

426. As regards the Groupement's second argument, it provides no evidence that free riding and the lack of means adequately to counter it are behind the fact that MasterCard's Maestro system has replaced the national systems referred to (or may be in the course of replacing them).

D. The basis of assessment of the fees at issue is incompatible with the principle of a remuneration of investment

427. The fact that the amounts payable under the measures increase in line with the number of cards issued (see recitals 143, 146 and 150 above) is incompatible with a justification based on a need to remunerate investment in order to prevent the investment from becoming a target for free riding. If the measures were really intended to remunerate investment, then the amounts due would not increase in line with the number of cards issued.

11.1.2.3. Conclusion on the free riding on investment

428. Despite the Commission's requests for clarification, the Groupement has not clearly defined the substance or scope of the alleged free riding on investment on the part of new entrants who contribute to the system ("first source" of free riding). Despite the fact that the burden of proof lies with the Groupement, it has furnished no empirical evidence of the existence of such a phenomenon and of its necessarily dynamic character.

429. Since the new entrants (both those which were already Groupement members before the adoption of the measures and new members) contribute to the CB system by issuing CB payment cards, it is reasonable to suppose that the Groupement alleges free riding on their part because it considers that new entrants focus on issuance, which is said to generate fewer benefits (positive externalities) for the system than acquiring ("the second source" of free riding). The measures - chief among which is the Merfa - are justified by the Groupement as being a mechanism for balancing the acquiring and issuing functions, which is examined below.

11.1.3. Justifying the measures, and the Merfa in particular, as a mechanism for balancing the acquiring and issuing functions (remedy to the "second source" of free riding) is not acceptable

430. The Groupement submits that it is necessary to encourage those members that are more issuers than acquirers to develop their acquiring activities, basing this conclusion on two studies carried out by [the Groupement's consultant] in March 2003 (619) and October 2004 (620) respectively. These studies are said to prove the existence of greater positive externalities generated by acquiring than by issuing; they thus conclude that the Merfa is justified on economic grounds. [The Groupement's consultant] has also produced a brief memo on the reform of the membership fee containing an assessment of all the costs borne in connection with the system since it was set up and an estimation of the value of joining (621).

431. The measures are also said to be designed to respond to the rapidly changing conditions in which the CB system operates, namely the establishment of a European payments area and a single euro payments area (SEPA) (622), the pace of technological innovation, which is leading to increased investment, and the growth in security requirements (623).

432. In the Commission's view, the Merfa cannot be justified on the grounds that free riding must be counteracted. First, the conclusions of the economic studies commissioned by the Groupement, identifying a situation which the Merfa is supposed to rectify, are not valid. Second, the Merfa cannot be justified on the grounds of its alleged function of regulating acquiring and issuing activities.

11.1.3.1. The conclusions of the economic studies justifying the Merfa are not valid

433. The Commission takes the view that the Groupement's economic studies are based on questionable data (A.), are marred by methodological errors (B.) and therefore reach questionable conclusions (C.). Moreover, the notified measures were not adopted as a result of the economic studies (D.). Lastly, the Commission's doubts about the validity of the conclusions drawn by the economic studies justifying the Merfa remain even after submission of additional economic studies by the Groupement following its receipt of the Statement of Objections of 18 July 2006 (E.).

A. The studies are based on questionable data

434. According to the Groupement, the need to encourage those members that are more issuers than acquirers to develop their acquiring activities in order to ensure that the system is balanced (624) arises from the fact that the acquiring function generates more positive externalities than the issuance function (625).

435. This is said to have been shown by [the Groupement's consultant]'s studies, according to which an increase in the number of acceptance points also gives rise to an increase in the number and value of payment and withdrawal transactions - thus adding value to CB cards - while the issuance of new CB cards does not generate additional payment or withdrawal transactions. The conclusion that acquiring generates positive externalities while issuance generates none (or few) is based on data indicating that the issuance market is saturated whereas the acquiring market is not. The latter will continue to generate positive externalities on the issuance side, while the former can no longer, or to a far lesser extent, generate externalities on the acquiring side (626).

436. After a detailed analysis of the data and underlying hypotheses in [Groupement's consultant]'s studies, the Commission has concluded that they are not valid.

437. The data used by [the Groupement's consultant] contain "outliers" (627) that should have been eliminated, which would have led to the conclusion that externalities on the acquiring side were not significant either.

438. Moreover, data other than those selected by [the Groupement's consultant] seem to indicate that bank card issuance is not saturated in France, given, among other things, the potential for growth in the number and volume of transactions (628) as a result of the transformation of proprietary cards into bank cards, the issuance of corporate credit cards, the replacement of cheques and cash payments and reductions in cardholder fees (629). These objective data confirm not only the statements made by a number of new entrants (630), but also, and especially, the estimates used by the Groupement and the main members in preparing the measures: according to these, "The audit of the CB bank cards GIE, based on observation of the system's history, forecasts saturation around [...] in the number of CB cards in France (with [...] million cards in circulation at that time)", which means that issuance is not saturated, nor was it when the measures were being prepared in 2001-2002. On the other hand, the Groupement observes (contrary to the studies and justifications submitted in this case) that acquiring is already very developed in France, referring to the "supercritical size" of the ATM pool (631). Likewise, one of the main members explains that "ATM investments are no longer necessary since the national territory is already well covered" (632). Even the data appearing in the magazine published by the Groupement confirm that the ratio of POS terminals to cards is far higher than in any other European country, at 161% of the European average (633).

B. The studies are marred by methodological errors

439. [The Groupement's consultant]'s studies are marred by methodological errors. First, [the Groupement's consultant] uses a simple linear equation containing independent variables, one for payment business and the other for withdrawal business. This theoretical model used by [the Groupement's consultant] is largely incomplete, as only a system of simultaneous equations can reflect the interdependence of externalities in a two-sided market A. "simultaneous" system should be understood to mean a system of two equations - one for issuing activities and one for acquiring activities - which should be solved at the same time to take account of the interdependence of the externalities of issuance and acquiring activity. Second, [the Groupement's consultant]'s empirical results are biased and are not reliable (634).

C. The studies reach questionable conclusions.

440. The conclusion that acquiring generates significant positive externalities but that issuance does not is questionable because, as indicated above, it is based on questionable data and methodology.

441. While contesting the suitability of the methodology and data used (see recital 439), the Commission nevertheless wished to test the robustness of [the Groupement's consultant]'s results using two further econometric tests (635). Taking specifications different from those used by [the Groupement's consultant], the Commission reaches different, quite opposite, conclusions:

- Using lagged variables - the lagged model is often used to measure externalities over time - gives results opposite to those obtained by [the Groupement's consultant]: the effects on the issuance side for payment transactions become significant while those on the acquiring side become statistically insignificant.

- Using [the Groupement's consultant]'s model with raw data instead of growth rates, in relation to flows but also to stocks, the Commission obtains results different from [the Groupement's consultant]'s: If raw data based on flows are applied to the "payment activities" model, the results are the opposite of those obtained by [the Groupement's consultant] (insignificant variables for acquiring, but significant for issuance): For the "withdrawal activities" model, neither the acquiring nor the issuance variables are significant. The raw data based on stocks give contradictory results.

442. The conclusion of [the Groupement's consultant]'s studies, to the effect that acquiring generates more positive externalities than does issuance, is also empirically contested by the experience of other card payment systems. According to Visa, for example, "the Groupement's argument that acquiring activities create more positive externalities than issuing activities do is without foundation. In addition, the significantly increased membership fee for new members and the fee per CB card issued for new members may have an adverse impact on cross-border issuing ... Visa does not believe that imposing the Merfa or issuing fees on banks who only or predominantly issue is the appropriate means of encouraging members to continue to contribute to the CB system as, in Visa's experience, issuing is more important, or at least as important, to the success of four-party payment systems, as acquiring."

"Visa believes that the statement [of the Groupement] " acquisition activities create more positive externalities than issuing activities do " is without foundation. [...] In addition, Visa believes that the significantly increased membership fee for new members and the fee per CB card issued for new members may have an adverse impact on cross-border issuing [...] Visa does not believe that imposing the Merfa or issuing fees on banks who only or predominantly issue is the appropriate means of doing so as, in Visa's experience, issuing is more important or at least as important, to the success of four-party payment systems, as acquiring."

"Visa considers that CB's assertion that "acquisition activities create more positive externalities than issuing activities do" is not a valid or plausible generalisation and therefore not a plausible basis for imposing the Merfa or issuing fees. Visa believes that until now the CB system has prospered and grown without the Merfa.

Based on the functioning of other four-party payment systems (including Visa's own experience), Visa believes it is clear that, in the past, issuing has been at least as important as acquiring, if not more important to the functioning of four-party card payment systems and consequently, that issuing without acquiring is of benefit to the system. In the absence of any evidence from the CB to the contrary, Visa submits that it is likely to still apply to the CB system.

Therefore, in Visa's view, the motivation of CB in establishing the Merfa and issuing fees is unclear" (636).

On the basis of the functioning of its own system, Visa considers that:

"by issuing cards, a member is virtually certain to contribute to the strengths of the system [...] There is a great deal of scope for innovation in promoting card usage (and cardholding) in order to win a greater share of the market from cash, cheques and other cards. Increased card usage will normally happen even where cards are issued to existing [...] cardholders because the terms and/or "type" of the card is considered to be more advantageous to the offeree. Further, involvement in card issuing not only promotes competition on the issuing side, but also encourages the development of the merchant network through providing access to a larger card base, i.e. generates positive network externalities.

In contrast, a member which only acquired merchants would not necessarily increase the volume of Visa card transactions and thereby benefit the system [...].

For this reason, Visa believes that a contribution measured in terms of cardholders (and, therefore, card usage) is the simplest and best method of securing a contribution to the system from which all members benefit and which adds to the strengths of the system and ensures the development and growth of the system on an ongoing basis.

[...]

To the best of Visa's knowledge, four-party card payment schemes other than Visa also acknowledge the importance of issuing. For example, Visa believes Switch and MasterCard members must issue a sufficient number of cards to add value to the scheme before they may commence acquiring activities" (637).

443. Likewise, the absence of measures resembling the Merfa in the systems in place in other Member States has not prevented those systems from developing, even in countries where many issuing banks have entered the market. For example, an analysis of the situation in the United Kingdom, a Member State comparable to France (in terms both of population and of the number and value of card payments (638)) shows that large-scale entry of issuing-only institutions in no way undermines the smooth functioning, still less the very existence, of the system. On the contrary, without any supposedly regulatory system comparable to the fees introduced by the Groupement, ATMs have increased at an even faster rate than in France (639).

D. The measures were not adopted as a result of the studies

444. [The Groupement's consultant]'s first study dates from March 2003 (640), after the measures were notified in March 2002. Even though the study was preceded by a number of reports (641), an analysis of the various versions of the study of March 2003 shows that in reality it was adapted to take account of developments in the preparation of the measures (642).

445. Moreover, examination of the facts of the case shows that the negotiations within the COM on the preparation of the measures were based on other considerations besides taking account of the positive externalities generated by acquiring.

E. The Commission's doubts about the validity of the conclusions drawn by the economic studies justifying the Merfa remain completely relevant

446. After receiving the Statement of Objections of 18 July 2006, the Groupement produced two economic studies (643) supposed to refute the Commission's doubts about the conclusion to the effect that significant positive externalities were generated on the issuance side by acquiring, while almost no positive externalities were generated on the acquiring side by issuance (conclusion of [the Groupement's consultant]'s studies justifying the Merfa).

447. However, far from refuting the Commission's doubts about [the Groupement's consultant]'s conclusion, one of the two studies confirms that it cannot be concluded that the externality generated by acquiring is greater than that generated by issuance: "Another question ... is why acquiring should be encouraged more than issuing. Ultimately, the question boils down to estimating the value of the membership externality from retailers to cardholders (how much economic value is created for a cardholder when a new retailer affiliates with the network?) and to compare it with the symmetric externality (how much economic value is created for a retailer when a new card is issued?). Estimating such externalities is extremely difficult, as it would require a structural model of the "CB" System and very detailed data at the level of individual banks. There is no reason to believe a priori that one externality is systematically bigger than the other: this may depend on the maturity its degree of penetration of both sides of the markets and on available substitutes." (644)

"The need to encourage acquiring more than issuing has not been established ... As rightly pointed out by the [Commission], measuring these externalities would necessitate estimating a complete structural model.

This would require large and rich data sets that are not available yet." (645)

448. The other study (646), while it makes a number of criticisms of the Commission's economic study (i.e. the two additional econometric tests referred to in recital 441 at C. above, carried out in order to test the robustness of [the Groupement's consultant]'s results) (647), acknowledges that [the Groupement's consultant]'s study is marred by a number of methodological errors (see B. above), and in particular that the economic model used by [the Groupement's consultant] is incomplete with respect to the payment aspect and that there are outlying values. As explained above (see recital 437), these values should have been eliminated, which would have led to conclusions different from those of [the Groupement's consultant].

449. The Commission maintains its objection that the justification of the Merfa is invalid in so far as it is based on an economic study whose methodology is not sound.

11.1.3.2. The Merfa cannot be justified on the grounds that it regulates issuance and acquiring activities

450. The Merfa cannot be justified on the grounds that it regulates issuance and acquiring activities in the interests of the system, because it does not have this function (A.) and because this alleged function contradicts the other measures (B.).

A. The Merfa does not perform a regulatory function

451. Since (for the reasons set out above) the statement that acquiring generates more positive externalities than issuance is not valid, the Merfa, which is supposed to remedy a non-existent situation (an imbalance in the system, where issuance is excessive and acquiring insufficient), cannot perform the regulatory function ascribed to it by the Groupement.

452. Moreover, even supposing that acquiring does generate more positive externalities than issuance, the Merfa is prevented by its very formula from producing the best overall issuance/acquiring balance for the system (1.). In addition, there are a number of obstacles preventing banks primarily engaged in issuing from developing acquiring activities in practice in the sectors and areas where an expansion of acquiring is alleged to be in the interests of the system (2.). Lastly, being unable, on account of these obstacles, to develop their acquiring activities in order to escape the Merfa, certain banks might prefer paying the Merfa to restricting card issuance. However, in such an eventuality, the sums paid under the Merfa are not assigned to purposes which would meet the first condition of Article 81(3) (3.).

1. The Merfa's formula prevents it from producing the best overall issuance/acquiring balance for the system

453. Its very formula means that the Merfa is not designed to produce the best overall issuance/acquiring balance for the system. The reference issuance/acquiring ratio is that of the main members (a), and the Groupement does not show that the criteria used produce the best balance for the system (b). Lastly, the Groupement does not show that the best balance for the system as a whole requires all the members to align themselves on the same ideal (c).

(a) The reference issuance/acquiring ratio is that of the main members, not the best balance for the system

454. The Merfa does not encourage acquiring but penalises new entrants which do not conform to the acquiring/issuance ratio prevailing throughout the system, which is that of the main members inasmuch as the acquiring and issuance activities of the latter (measured, according to the Merfa formula, in terms of SIRENs, installed ATMs and issued cards) account for most of the total acquiring and issuance activities in the CB system. There is no guarantee that the main members' acquiring/issuance ratio - from which new entrants cannot deviate too much at the risk of having to pay the Merfa - corresponds to the best balance for the system as a whole (see recitals 214 to 232 above).

(b) The Groupement does not show that the criteria used (SIRENs, ATMs and number of cards) produce the best balance for the system

455. The Groupement does not satisfactorily explain what makes the numbers of SIRENs and ATMs the most appropriate criteria for measuring withdrawal and payment activity. In this connection, it should be noted that the SIREN number (which penalises new entrants from the large retail sector and benefits the main members, see recitals 285 and 286) does not reflect the extent of the SIREN holder's electronic banking business.

456. Interchange fees, on the other hand, are based on the number and value of withdrawal and payment transactions and, for the purposes of processing electronic banking transactions on which these commissions must be paid, merchants are identified by their SIRET number; moreover, according to the Groupement (see recital 492 below), the CIP and CIR interchange fees act as regulators of the acquiring/issuance functions.

457. [The Groupement's consultant] explains that the solution of a mechanism based on the number and value of transactions was rejected because it would have stepped up competition for major merchants and ATMs already acquired rather than acting as an incentive to acquire merchants not yet in the CB system and to install new ATMs in areas where there were few of them (648). However, he does not explain why increased competition for major merchants and ATMs should be avoided or how the new entrants targeted by the measures might succeed where the major banks have failed (see recitals 201 to 222, and in particular 205 to 213, which show that the acquiring market is almost exclusively in the hands of the main members and that the most profitable merchant sectors and areas to be equipped with ATMs have already been acquired).

458. Nor does the Groupement explain why it is best for the system that a member's share in the payment acquiring market (member's number of SIRENs as a proportion of total number of SIRENs in the system) and its share in the withdrawal acquiring market (member's number of ATMs as a proportion of total ATMs in the system) have the same weighting in the formula to determine whether the Merfa applies.

(c) The Groupement does not show that the best balance for the system as a whole requires all the members to align themselves on the same reference threshold.

459. According to the Groupement, the fact that acquiring generates more positive externalities than issuance means that members primarily engaged in issuing must be encouraged to develop their acquiring activities so that acquiring and issuance are ensured in a balanced fashion throughout the system. The Merfa formula penalises members whose share of acquiring is less than half their share of issuance.

460. The Groupement does not show that the balance of the system requires each member to align itself on the same reference threshold, in respect of which there is no evidence that it is the best for the system. While some banks are engaged primarily in issuance activities, others have major acquiring activities and can (at least in part) offset the activities of the former. The Groupement does not show that the balance of the system requires a mechanism that forces each member to aim for similar ratios. On the contrary, the Groupement has expressly recognised that the opposing strategies of issuing banks and acquiring banks offset one another: "While the business strategy of some banks may prompt them to favour one or other component of interbanking, their ambitions cancel out those of banks with the opposite business strategy" (649).

461. Moreover, by stating that acquiring generates "more" positive externalities than issuance, the Groupement acknowledges that issuance also generates some externalities, even if, in its view, fewer than those generated by acquiring. Banks and competing card payment systems claim that card issuance also generates positive externalities - at least as much as acquiring, quite possibly even more (650). However, in order to escape liability for the Merfa, a bank may either expand its acquiring business or rein in its card issuance business. Should a bank choose to reduce issuance rather than boost acquiring, the positive externalities on the acquiring side (generated by issuance activities), which would otherwise have benefited the system and all its members, would be lost, and the loss would not be offset by any positive externalities (on the issuance side) generated by the acquiring business of that bank. Conversely, a drop in the price of cards could stimulate card demand and - consequently - card issuance, which also very likely produces positive externalities on the acquiring side.

462. Similarly, Banque Accord does not agree that each bank should be forced to develop acquiring, as issuance also generates positive externalities: "It is moreover wrong to maintain that each bank must, if it wishes to avoid being taxed, develop the two markets [acquiring and issuance], which require separate specialisms: although developing the acquiring market certainly allows banks to offer consumers a larger number of points of sale where they can pay with their CB cards, which is obviously positive, nevertheless issuance likewise allows each merchant with a payment terminal to expand its business and the services it offers its customers, at the same time allowing the bank to benefit from the revenue directly related to this expansion of the customer base. Issuance therefore benefits acquiring as much as acquiring benefits issuance. There is therefore no imbalance justifying the regulatory criteria applied by the Merfa, which are based on the premise that acquiring alone constitutes a positive externality. Issuance is just as much of a positive externality." (651)

2. In practice, the Merfa does not encourage banks primarily engaged in issuing to expand their acquiring activities in the sectors and areas identified by the Groupement as being in the interests of the system

463. The development of acquiring by a new entrant is extremely difficult owing to the fact that this market is highly concentrated, being almost exclusively in the ands of the main members, owing to the vital importance of a local branch network that makes it possible to develop an overall, individualised and sustained relationship with merchants; and owing to the fact that the most profitable merchant sectors and areas to be equipped with ATMs have already been acquired by the main members (see statements by new entrants (652) and recitals 205 to 213 above).

464. Moreover, the Groupement does not show how the Merfa would widen acceptance of the card in sectors where it is not yet widely accepted (health professions, public services, home deliveries, taxis, local businesses, or markets) (653), especially since new entrants liable for the Merfa do not have the means to reach these highly specific sectors. There is therefore reason to believe that, supposing the Merfa does actually lead to an increase in acquiring (which is doubtful for the reasons already given (654)), it encourages ATM installation more in areas that are already (over)equipped than in sectors and rural areas where there is a lack of equipment. This view is shared by Cofidis, which states that "ATMs could be installed only in the best places, to the detriment of the profitability of those which are already there" (655).

465. Reasons other than an alleged slowdown (or threatened slowdown) in acquiring may explain the low penetration of the CB card in these sectors: a choice not to incur the cost of joining the system, a desire to evade taxes, the conservative attitude of customers who prefer to continue paying with cash (local services), a judgment that payment would be impractical (especially at markets), etc.

3. The sums paid under the Merfa are not assigned to purposes which would meet the first condition of Article 81(3)

466. The sums paid under the Merfa are not allocated to the maintenance and development of the system, but are merely redistributed among the members not liable for it (primarily the main members, i.e. the members of the Groupement's Board of Directors), which are free to use them as they choose (656).

467. In the Commission's view, even assuming that the Merfa might have as its object or effect the development of acquiring (which is not the case), the Merfa has no positive economic effect on the system because (i) according to the Groupement, members engaged primarily in issuance are free not to contribute to the system's acquiring/issuance balance by expanding their acquiring business (657); and (ii) the sums paid under the Merfa are not allocated in the interests of the system.

B. The allegedly regulatory function contradicts the function of the other measures

468. As set out above, the allegedly regulatory function of the Merfa is in contradiction with that of the CIP and CSC interchange fees paid by acquirers and of certain other of the measures in question penalising those who have not issued sufficiently in the past (see section 10.2.1.1. B "The function of the Merfa is in contradiction with that ascribed by the Groupement to the interchange fees and those of the other measures", recitals 223 to 231).

11.1.4. There is no free riding

469. Since it does not give a precise definition of what the alleged free riding involves, and in particular of the nature and value of its target, and since it does not specify the value of the investments which benefit the shared infrastructure as opposed to those which benefit the members who made them, the Groupement has failed to give valid proof of the existence of the alleged free riding.

470. In any event, there is no free riding, for the following reasons:

(a) All members of the system (including, therefore, new entrants) already pay a consideration whenever they use the system, in the form of the CIP or CIR interchange fees.

These commissions pay for services provided and cover the costs of those services (658). According to UFC-Que Choisir (659), the real total cost per withdrawal for ATM managing banks (banks acquiring withdrawal transactions) is around €0.182, an amount far lower than the only fixed part (cash advance) of the CIR at € [...].

However, as set out in section 5 above (660), according to the Groupement the interchange fees perform a regulatory role. The Groupement states "the CIR is based on a self-regulating mechanism ... The amount of the CIR, like that of all interchange fees, including the Merchant Interchange Fee [CIP] ... represents a balance between the divergent aims of the different members of the system, each of which pursue their own commercial strategy ... The system therefore incorporates its own regulating mechanism which is born of the conflict of interests between issuers and ATM operators ..." (661).

(b) The members described as "free riders" have also contributed to the development of the CB system (since they are already members who have paid their membership fee, issued cards and paid the interchange fees) and will continue to do so.

(c) As regards the free riding alleged by the Groupement to consist in the advantage derived by banks engaged primarily in issuance from the supposed superiority of the positive externalities generated by acquiring, the Commission does not accept the initial hypothesis that acquiring is superior to issuance in terms of generating benefits for the CB system (see section 11.1.3) and therefore it does not accept the existence of any alleged free riding on these grounds. Moreover, according to the Groupement, the CIP and CIR interchange fees already perform the function of regulating the insufficient contribution of some members to the acquiring or issuance function.

11.1.5. The other economic justifications are likewise not proven

471. Besides the alleged need to combat free riding and balance issuance and acquiring, the Groupement does not show how the measures ensure that the system keeps pace with innovation, copes with the establishment of a European payments area and meets new security requirements (662). This is all the more true since there is nothing to guarantee that the amounts paid under the measures are allocated to the system. On the contrary, amounts paid under the Merfa are redistributed among the members which are not liable for it. As regards the amounts paid under the other measures, there is no evidence that they are used for the purposes of keeping pace with innovation, coping with the establishment of a European payments area or meeting new security requirements (663). Nor does the Groupement draw a distinction between the costs of investments required to combat the alleged free riding and those required to keep pace with technological innovation and to meet security requirements.

472. It is clear from the above that the contribution to technical or economic progress of the measures at issue has not been demonstrated and that the arguments put forward by the Groupement as regards the need to prevent free riding by new entrants or to encourage acquiring by way of the Merfa cannot be accepted.

11.1.6. The measures have negative economic effects

473. The measures have negative economic effects, such as limiting the supply of CB cards; an increase in the price of cards for holders, or, at the very least, a brake on the reduction in the price of cards; and a more restricted supply of cards with new functions such as the combination of payment and loyalty or cash-back cards (see section 10.2.2.)

474. As regards the Merfa in particular, rather than encouraging the expansion of acquiring, it puts a brake on issuance, an activity which could generate positive externalities of which the system will be deprived.

475. Issuance is penalised on two fronts: not only by application of the formula determining whether or not the Merfa applies, which discourages issuance (or leads to reduced issuance) in order to avoid liability for the Merfa, but also by payment of the Merfa itself, which consists in an amount per active card.

476. Moreover, the effect of the Merfa is to reduce interbanking cooperation, because banks will choose not to join, or will leave the system (although a causal relationship with the measures at issue cannot be proven, it is a fact that Egg and Capital One did not join the French banking network with a pool of ATMs and merchant acquiring agreements but, on the contrary, disappeared from the French banking scene). As explained in section 10.1.2.2. above, the measures tied to card issuance will have the effect, among others, of restricting the number of cards issued, stifling innovation (in so far as new entrants would have supplied cards with new functions) and blocking the reduction in card prices.

477. For the above reasons, the measures in question do not satisfy the first condition laid down in Article 81(3) of the EC Treaty.

11.2. The second test: consumers (cardholders and merchants) not allowed a fair share of the resulting benefit

478. According to the Groupement, cardholders benefit directly from the Merfa because it has the effect of widening acceptance of the cards for both payments and withdrawals. It adds that the benefit to all users, including cardholders and accepting merchants, derives from the fact that the measures in question ensure the maintenance and development of the CB system and the continued quality of service.

479. The Groupement argues that, in so far as the additional costs faced by a bank with excessively unbalanced activities within the CB system are legitimate, their impact on the price of its CB cards cannot in any way be attributed to the measures in question and thus cannot prevent the application of Article 81(3) of the EC Treaty. To argue the contrary would be to contest the right of systems targeted by free riding to take the appropriate measures to combat it, a right which has been recognised by the Commission on many occasions.

480. In the Commission's view, since there is no contribution to technical or economic progress, consumers cannot receive a fair share of any resulting benefit.

481. Although in the study by [the Groupement's consultant] dating from October 2004 (664) it is stated that a bank liable for the Merfa will "save" some € [...] per year if it installs a new ATM (and € [...] if it acquires a new SIREN), it should be noted that this is simply a saving on the amount due under the Merfa. By definition, the "saving" would be the greatest possible if there were no Merfa at all.

482. Far from allowing consumers a fair share of the benefit, the measures in question have negative effects on consumers (or at least might have if they were applied) by hampering the reduction in card prices and/or card issuance (see section 10.1.2.2. above), as is confirmed by the consumer associations UFC-Que Choisir and BEUC, in whose opinion the effects of a "reduction in margins for new entrants" and "a parallel increase in income for existing operators" result for consumers in "a restriction of competition over annual fees": the "new players are jostling the traditional banks in the area of the universal payment card ... The GCB's new financial rules applied to these players seem bound to disrupt this (infant) competition of benefit to consumers" (665).

483. GE Money Bank, one of the banks targeted by the Merfa, also highlights the negative effects of the measure for consumers: "the question arises whether the Merfa does not have the effect of diverting a company's investment policy away from a choice for growth, competitiveness and a wider range of services for consumers towards a choice to limit costs in order to avoid facing higher costs than its competitors" (666).

484. In addition, the installation of further ATMs on a market which is already over equipped will not significantly increase the number of withdrawals. If they are not profitable, the excessive installation of ATMs could even result in an increase in costs for consumers by way of higher charges for withdrawals. In this connection, several members expressed concern that installing new ATMs would not be profitable as the ATMs likely to be sufficiently used have already been installed by the main members (667).

485. It is clear, therefore, that not only do consumers not derive any benefit from any contribution to technical and economic progress by the measures in question but they also suffer from anticompetitive effects on the prices, volume and functions of cards. The second test in Article 81(3) of the Treaty is not met.

11.3. The third test: imposition of restrictions which are not indispensable

486. The third test laid down in Article 81(3) of the EC Treaty stipulates that restrictions on competition contained in an agreement (or a decision by an association of undertakings) which contributes to technical or economic progress, while allowing consumers a fair share of the resulting benefit, must be indispensable. Since the decision in this case does not meet these requirements, it is not necessary to examine whether the restrictions are indispensable. Nevertheless, for the sake of completeness, the Commission has examined this condition and concluded that it is not met.

487. According to the Groupement, the measures are indispensable in order to ensure the continuation of the system, which otherwise, without the means to combat free riding, would "collapse". It argues that if the anti-free-riding measures are not introduced, certain members might leave the CB system and set up a competing system. It claims that the restrictions which could result from the measures in question have been shown to be indispensable because of the need to combat free riding (in particular by ensuring that the system is balanced by encouraging banks engaged primarily in issuing to expand their acquiring activities), while preserving the freedom to act of the system's members. Referring to the studies by [the Groupement's consultant], the Groupement argues that the Merfa is the most objective and least restrictive means of attaining its objective. The measures meet the proportionality condition by taking account, in an objective and non-discriminatory manner, of the real contribution made by each member to the operation and development of the CB system, and of the benefits that that contribution confers on the other members. The Groupement argues that its objective could not be met without the measures (668).

488. The Commission would comment that the new membership fee, the additional fee and the dormant members wake-up mechanism introduced by the Groupement are not indispensable to combat the alleged free riding, in so far as there is no free riding in the CB system, as explained in section 11.1.4 above.

489. The measures at issue, while formally applied without distinction, in practice have discriminatory effects (see section 10.2.1.1.A, recitals 199 to 234 and recitals 282 to 293), on account of the charges to be paid and the fact that, at the same time as allegedly encouraging acquiring, they discourage what is considered to be excessive issuance by certain banks. The choice left to banks to expand acquiring rather than paying the various charges under the measures and/or to restrict issuance is largely restricted by the obstacles to entering the acquiring market, with the result that the measures are discriminatory in practice in so far as they force certain banks to pay the charges and/or cut back their issuing plans.

490. If, as the Groupement claims, the system's interest was in developing acquiring in merchant sectors where the CB card was not yet widely accepted (669), it could have adopted other measures which, unlike the measures at issue that penalise a particular business model, could have attracted merchant sectors not yet affiliated: one example among others of such a measure would be low interchange fees payable on each transaction involving this type of merchant. In this connection, it should be noted that the Visa and MasterCard systems (outside France), and some national systems, set special interchange fees rates for technologies or types of merchant (airlines, service stations, electronic commerce, etc.) in relation to which they wish to stimulate the penetration of bank cards (670).

491. Moreover, the Merfa is not indispensable to balance issuance and acquiring activities. Competition is keen on the French acquiring market, as can be seen from the large number of acquiring banks in France; the number is far lower in the other Member States (671). This leads, according to the [consultancy firm], [...], to a drop in revenue for the founder banks ("The continuation of aggressive behaviour on acceptance is threatening the €1.1 billion income of the acquiring banks" (672)). This point is confirmed by [the Groupement's consultant], who states that the highly competitive nature of acquiring, an activity that consequently yields little direct profit, is "compounded in the CB system by the fact that, unlike what happens in many other systems, a large number of banks have an acquiring business while in other systems, only a few banks offer such services" (673). The need to encourage acquiring by measures such as the Merfa therefore seems questionable.

492. Besides the Merfa, banks engaged primarily in issuing already have to pay the CIR system's CSR counter-commission (in addition, of course, to the CIP in relation to any acquiring activities they may have). According to the Groupement, the function of regulating acquiring and issuance which it attributes to the Merfa is already accounted for by the interchange commissions, the CIR and CIP, with the CIR's counter-commission already compensating for some members' low rate of participation in acquiring. The regulatory function of the two interchange commissions (CIP and CIR) has been expressly and clearly stated by the Groupement: "it goes without saying that the system's withdrawal function can operate only if the number of cardholders ... grows in a balanced manner and if the imbalances observed at certain banks are corrected by way of financial compensation ... users of means of payment are all the more inclined to subscribe to a CB card if it allows them to withdraw cash from a large number of ATMs ... the counter-commission No 1, the withdrawal counter-commission or CSR ... takes account of the contribution to CB interbanking by members of the Groupement whose ATMs are used for withdrawal by holders of CB cards issued by other members"; "The CIR is based on a self-regulating mechanism ... Its amount, like that of all the interchange fees, including the Merchant Interchange Fee or CIP ... represents a balance between the divergent aims of the different members of the system, each of which pursue their own commercial strategy. ... The system therefore incorporates its own regulating mechanism" (674).

493. Although the Groupement claims (only after the Commission had suggested during its investigation of the case that the measures at issue might duplicate the interchange fees in terms of being justified on the grounds of balancing the system and combating free riding (675)) that the CSR component of the CIR was designed to "adjust the remuneration for interbank services that the banks provide one another" while the Merfa was designed to "remunerate the benefit to the CB system and its members as a whole provided by the acquiring business of those who operate such business" (676), justifying the interchange fees on the grounds that they remunerate inter-bank services does not means that they have no regulatory function: the generation by acquiring activities of positive externalities on the issuance side can be regarded as an interbank service provided to the issuing bank by the acquiring bank which should be remunerated by way of the interchange fees.

494. The Groupement also stated in its notification that the positive externalities generated by the acquiring function, which the Merfa is supposed to encourage, include "increased ATM security" (677), while in its notification of the CIR, it stated that the second element of this interchange fee paid for the service of introducing collective security measures provided by the cardholder's bank to the merchant's bank (678). The "security" justification for the Merfa therefore overlaps with and contradicts the justification for the CIR (in the case of the CIR the compensation for security is paid by the acquiring bank to the issuing bank, while the Merfa is paid by issuing banks).

495. Lastly, according to the Groupement, a reduction in the CIP would only exacerbate competition to acquire the large retailers, which generate major payment flows, without expanding affiliation in the sectors where the card is not yet widely accepted. It would lead only to a surplus of ATMs in the most profitable areas without resolving the lack of equipment in rural areas.

496. However,

- The Groupement states that its cards are increasingly accepted in the sectors that the Merfa was supposed to encourage, even though the measures have been suspended (679). This indicates that the measures are not necessary to the penetration of the CB card in these sectors.

- It is evident that the main members participating in the COM wished precisely to prevent the measures from increasing competition for the acquiring of merchants that already accepted CB cards (see for example recital 101(c) above).

- The Commission does not recognise the superiority of the positive externalities generated by acquiring which the Merfa is designed to take into account (see for example recitals 439 to 442).

- The Commission does not share the Groupement's view that the Merfa serves a regulatory function that can in no circumstances be performed by the other commissions. It is possible that per transaction interchange fees as, for instance, the CIP and CIR perform a function of regulating externalities, as the Groupement argued until this case began (680) (681). One of the errors vitiating [the Groupement's consultant]'s economic study is precisely the absence of the CIP and CIR in his models.

497. This analysis is confirmed by the statements made by the new entrants. S2P, for example, states: "The CIR is a mechanism for balancing commissions between banks with a pool of ATMs and banks which issue bank cards" (682) According to GE Money Bank, Covefi and Axa "the CSR paid by card-issuing banks already compensates for the lack of a contribution to "interbanking" by their ATMs" (683) "The CSR is paid per withdrawal by the card issuer to the acquirer (ATM owner). For this reason, we believe that the Merfa duplicates the CSR" (684); "the closer a bank's number of ATMs is to zero, the greater its liability for the CSR. The Merfa's nominator makes the same penalising calculation as the CSR, as it measures the bank's share of the total number of ATMs and uses it for 50% of the total nominator. Ultimately, it pursues the same balancing goal as the CSR, a charge introduced in order to reestablish a balance between the number of cardholders and investments in ATMs" (685).

498. An appropriate modulation of the CIP and CIR rates, in particular with reference to the type of merchant and the areas to be equipped, so as to facilitate, where necessary, the penetration of the card in certain sectors and areas, like in other card payment systems (see recital 490), could help to regulate the system if necessary, and with far fewer penalising effects, as the commissions are paid per transaction and do not penalise a business model.

499. Not only do the measures duplicate the CIP and CIR, they also duplicate one another (Merfa, membership fee per card, additional fee and dormant members wake-up mechanism). Thus, the Merfa penalises what is deemed to be excessive issuing compared with acquiring, and issuance is also penalised by the membership fee based on the number of cards issued, the additional membership fee and the dormant members wake-up mechanism.

500. As regards the other economic justifications cited by the Groupement (the measures help to keep pace with innovation, cope with the establishment of a European payments area and meet new security requirements), the Groupement does not show in what way the measures are indispensable, i.e. the least restrictive of all possible means of meeting the various objectives.

501. It follows from the above that the Groupement does not show that the measures in question are indispensable: the objectives pursued by the measures in question can be achieved by other measures already in existence (appropriate modulation of the interchange fees); they also duplicate one another and the interchange fees. The measures in question therefore do not satisfy the third condition laid down in Article 81(3) of the Treaty.

11.4. The fourth test (the measures in question must not make it possible to eliminate competition in respect of a substantial part of the products in question)

502. If only one of the tests in Article 81(3) is not met, this is enough to render the exemption laid down therein inapplicable. In this case, since the first three conditions are not met, it is not necessary to examine whether the fourth is. This does not rule out the possibility that the measures may make it possible to eliminate competition in respect of a substantial part of the products in question.

11.5. Conclusion on the applicability of Article 81(3) of the EC Treaty

503. It follows from the above that the anticompetitive decisions by an association of undertakings at issue satisfy none of the first three tests of exemptibility laid down in Article 81(3) of the EC Treaty. The Groupement's decision is therefore contrary to the provisions of Article 81(1) of the EC Treaty and automatically void pursuant to Article 81(2) of the EC Treaty.

12. LEGITIMATE INTEREST

504. The Commission has a legitimate interest (686) in pointing out the infringement, in ordering that it is put to an end and in prohibiting any repetition of it, particularly as the infringement is still ongoing (687) and could even produce further effects688 if the Groupement decided to lift the mere suspension of the measures, as it stated that it envisages doing (689).

13. CONCLUSION

505. Article 7 of Regulation No 1-2003 empowers the Commission, where it finds that there is an infringement of Article 81 of the EC Treaty, to require the undertakings concerned to bring the infringement to an end. In the light of the above, the Commission notes that the pricing measures notified by the Groupement which are tied to card issuance (Merfa, membership fee per card, additional fee and dormant members wake-up mechanism) are contrary to Article 81 of the EC Treaty. Consequently, the Commission orders that the Groupement withdraws these measures and refrains from entering into any agreement or concerted practice or taking any decision of an association of undertakings having a similar object or effect.

Has adopted this decision:

Article 1

The pricing measures adopted by the Groupement des Cartes Bancaires "CB" (hereafter "the Groupement") by decision of its Board of Directors on 8 and 29 November 2002, that is to say, the mechanism for regulating the acquiring function ("Merfa"), the membership fee per card and the additional membership fee, and the dormant members wake-up mechanism, are contrary to Article 81 of the EC Treaty.

Article 2

The Groupement shall bring to an end the infringement mentioned in Article 1 by withdrawing the notified pricing measures mentioned in this Article, in so far as it has not already done so.

The Groupement refrains, in the future, from adopting any measure or behaviour having an identical or similar object or effect.

Article 3

This decision is addressed to:

G.I.E. Groupement des Cartes Bancaires " CB " 31, rue de Berri - Immeuble Monceau 75008 Paris - France.

Notes :

1 OJ L 1, 4.1.2003, p. 1. Regulation as amended by Regulation (EC) No 411-2004 (OJ L 68, 6.3.2004, p. 1).

2 OJ L 123, 27.4.2004, p. 18.

3 Meeting of the Advisory Committee on Restrictive Practices and Dominant Positions of 5 October 2007.

4 Even though they were already members of the Groupement, these banks were referred to as "new entrants" when the measures were being prepared and adopted (see for instance notification of the measures, paragraph 37, p.8, file p.9).

5 OJ C 80, 8.4.2003, p. 13. Comments were submitted by Visa International, Citibank International and Association française des usagers des banques (AFUB).

6 OJ L 123, 27.4.2004, p.18.

7 BNPP, Natexis Banques Populaires, Caisse Nationale des Caisses d'Épargne et de Prévoyance (hereinafter CNCEP), Crédit Mutuel, Crédit Lyonnais, La Poste, Crédit Agricole and Société Générale.

8 Cedicam (a subsidiary grouping together Crédit Agricole's electronic banking activities) and CT6 (which groups together the electronic banking activities of Crédit Industriel et Commercial).

9 Article 34(1) of Regulation No 1-2003.

10 File p. 25775.

11 See the Groupement's reply of 14 September 2004 to the Commission's request for information of 30 July 2004, file p. 15710.

12 The amendments related to Articles 7, 10, 12, 13 and 23. See the Groupement's reply of 9 March 2005 to the Commission's request for information of 11 February 2005 (reply to question 2 and Annex 2), file pp. 23154 and 23228-23244.

13 "Follow-up to the observations submitted on 19 October 2006", consisting of two memos: the first, dated 10 November 2006, produced by [the Groupement's consultant], and the second, dated 9 November 2006, produced by [the Groupement's consultant] (Annexes 18 and 19 to the Groupement's observations of 19 October 2006 in reply to the Statement of Objections of 18 July 2006, file pp. 27418-27499).

14 See the Groupement's reply to the Commission's request for information of 3 March 2003, file p. 1240.

15 Figures available on the Groupement's website, www.cartes-bancaires.com, in June 2006, file p. 25720. The notification sent by the Groupement in December 2002 refers to about 170 members, file p. 4.

16 See the Groupement's reply of 7 November 2003, question 3.2.1- Issuance activity, file p. 2438.

17 Document entitled "2005 Facts and Figures", March 2006, available on the Groupement's Internet site at www.cartes-bancaires.com, file p. 25757.

18 Figures for 2005, loc. cit., file p. 25757.

19 Figures for 2005, loc. cit., file p. 25757.

20 Figures for 2005, op. cit., file p. 25757.

21 Figures for 2004, RBR Report "Payment Cards in Western Europe 2006", chapter on France, p. 32, file p. 26165.

22 Figures for 2004, loc. cit., chapter on France, pp. 35 and 37, file pp. 26167-26168.

23 American Express's reply of 26 March 2004 to the Commission's request for information of 2 March 2004, file p. 4013.

24 Figures for 2004: 1.86 million, RBR Report "Payment Cards Western Europe 2006", chapter on international card organisations, p. 8, file p. 25778.

25 Figures for 2004, RBR Report "Payment Cards Western Europe 2006", chapter on international card organisations, p. 9, file p. 25779.

26 Figures for 2004, RBR Report "Payment Cards Western Europe 2006", chapter on France, p. 9, file p. 25777.

27 Figures for 2004, RBR Report "Payment Cards Western Europe 2006", chapter on France, pp. 35 and 37, file pp. 26167-26168.

28 See November 2003 version of the Groupement's Rules of Procedure and Formation Agreement, file pp. 68 et seq. and 89 et seq. See also March 2005 versions following the amendments proposed and adopted at the Combined General Meeting of the Groupement held on 11 March 2005, the Groupement's reply of 16 June 2005 to the Commission's request for information of 8 June 2005 (file pp. 24109 et seq.; 24130 et seq.; 24135 et seq.; 24159 et seq.; and 24178 et seq.).

29 In addition to the Articles of the Groupement's Formation Agreement and Rules of Procedure cited above, see the letter from the Groupement dated 23 February 2005 in reply to the Commission's request for information of 11 February 2005, file p. 23086.

30 See also Article 11.2 of the Formation Agreement, which states that "The Board of Directors shall consist of 11 main members", file pp. 95-96.

31 Such an amendment would require a decision by the Extraordinary General Meeting (Article 24.2 of the Formation Agreement, file p. 104) adopted by a three-quarters majority of the members' votes (Article 23.2 of the Formation Agreement, file p. 104) and the main members hold the majority of votes in this meeting.

32 Letter from the Groupement dated 23 February 2005 in reply to the Commission's request for information of 11 February 2005, file p. 23086.

33 It approves the accounts, appoints or dismisses the CEO, amends the Formation Agreement and may dissolve the Groupement (Article 24 of the Formation Agreement, file p. 104).

34 See previous section.

35 According to the letter of the Groupement of 16 June 2005 in reply to the Commission's request for information of 8 June 2005, p.2, file p. 24107 : " the adoption of the Merfa gave rise to the introduction of Article 13 of the Rules of Procedure [on 20 December 2002] ".

36 " Le Conseil de Direction peut mettre en place des mécanismes régulateurs afin d'inciter les Membres à remplir une fonction d'acquiring proportionnée à leur fonction d'émission " (Article 13 du Règlement intérieur, cote 76).

37 "The Board of Directors sets the financial conditions for payment and withdrawal transactions, in which the members ... agree to exchange between them the transactions received from their own clients. It lays down inter alia the rules on: - the merchant interchange fee (CIP), - ATM interchange fee (CSR)" (Article 14 of the Rules of Procedure, file p. 76).

38 Article 13.4 of the Formation Agreement, file pp. 99 and 24121.

39 Article 13.2 of the Formation Agreement, file pp. 98 and 24119.

40 Article 7.1 of the Formation Agreement, file p. 91.

41 Article 13.4 of the Formation Agreement, file pp. 99 and 24121.

42 Article 7.3 of the Formation Agreement, file p. 91.

43 Article 10.4 of the Formation Agreement: "The issuing of any CB card is subject to the prior agreement of the Board of Directors for the purpose of verifying compliance with the CB regulations".

44 "The Board of Directors lays down the conditions under which the member ... issuing CB cards guarantees the payment transactions initiated using his cards to the member ... that has accepted the transactions presented for encashment" (Article 15 of the Rules of Procedure, file p. 76).

45 See Articles 9, 10, 11 and 18 of the Rules of Procedure, file pp. 73-75 and 77.

46 Decisions referred to in Article 13.4 of the Formation Agreement, file p. 99.

47 Decisions referred to in Article 13.3 of the Formation Agreement, file p. 98.

48 For the decisions referred to in Article 13.2 of the Formation Agreement, file p. 98.

49 See Articles 13.2, 13.3 and 13.4 of the Formation Agreement, file pp. 98 and 99.

50 See section 7 below.

51 Pursuant to Article 35 of the Rules of Procedure, file p. 85.

52 See the column "foreign withdrawals" (retraits déplacés) in Annex 1 to the Groupement's reply of 10 March 2005 to the Commission's request for information of 11 February 2005, file pp. 23457-23464, and the table in recital 205 below (setting out the data in Annex 3 to the above-mentioned reply from the Groupement, file pp. 23479 et seq.).

53 Decision statement. COM meeting on 11 October 2002. File p. 13378.

54 "The policy decisions adopted on 11 October have been put into effect by the Board of Directors at its meetings of 8 and 29 November." File on preparations for the COM meeting on 7 January 2003, file p. 4871.

55 See Annex IX to the Groupement's letter dated 7 November 2003 (file pp. 3072-3088), which reproduces the notification of the CIP made on 20 December 1990 (Case 32746).

56 See Annex XI to the Groupement's letter of 7 November 2003 (file pp. 3091-3119), which reproduces the notification of the CIR made on 11 December 1995 (Case 35859). Strictly speaking (as the Groupement states further on in its notification of the CIR, file pp. 3095-3097), only the cash advance and the CSR "counter-commission" are paid by the issuing bank in respect of each interbank withdrawal transaction. The CSC "counter-commission" is, on the other hand, paid by the acquiring bank operating the ATM in respect of each interbank withdrawal transaction. The cash advance is paid in respect of each withdrawal transaction irrespective of the identity of the issuing bank, whereas the CSR and CSC "counter commissions" are paid only by the banks that are liable for them.

57 According to the Groupement, the CIP and the CIR "are intended to remunerate the interbank services which the banks provide to each other for each payment or withdrawal transaction", file p. 20738. As regards more specifically the CIP, see recital 50. As far as the CIR is concerned, as stated in recital 52 below, the cash advance component remunerates the service which the bank operating the ATM renders to the issuing bank by dispensing banknotes to the holders of cards issued by the latter.

58 For the regulating function of the CIP, see footnote 56. For the regulating function of the CIR, see recital 56.

59 See section 4.1 of Annex IX to the Groupement's letter of 7 November 2003, which reproduces the notification of the CIP, cited above: "The CIP is intended to remunerate the services provided by the cardholder's bank to the merchant's bank when any transaction takes place using a bank card ... these services have been broken down into three categories according to their type: - processing of the transaction; - implementation of collective security measures; - payment guarantee", file pp. 3075-3078.

60 As [...] the Groupement said in his introduction to the COM meeting on 9 November 2001, "it must be stressed that ... the CIP incorporates a 'natural' stabiliser", file p. 14263.

61 See the Groupement's reply of 23 February 2005 to the Commission's request for information of 11 February 2005, p. 2 and paragraph 5, file pp. 23086 and 23087.

62 See section 4.1 of Annex IX to the Groupement's letter of 7 November 2003, file pp. 3075-3078.

63 See Annex XI to the Groupement's letter of 7 November 2003, file pp. 3091-3119.

64 See the Groupement's letter of 7 November 2003, answer to question 5 (file p. 2443).

65 See section 4.1 of Annex XI to the Groupement's letter of 7 November 2003: "the cash advance ... introduces financial compensation for the services which members of the Groupement render to each other by dispensing banknotes to the other members' cardholders; it is therefore incurred by the issuing bank for the benefit of the bank operating the ATM" (file p. 3095).

66 Letter from the Groupement dated 23 February 2005, page 2, paragraph 2 (file p. 23086).

67 According to Annex XI, cited above, the servicing rate is equal to the number of serviced withdrawals divided by the total number of serviced withdrawals and the number of foreign withdrawals. The servicing rate is determined at the end of each calendar quarter (file p. 3096).

68 See the Groupement's reply of 7 November 2003 to the Commission's request for information of 24 October 2003 (file pp. 2470-2638).

69 See Annex XI to the Groupement's letter of 7 November 2003 (file pp. 3096 and 3097).

70 See Annex XI to the Groupement's letter of 7 November 2003, in particular file p. 3097.

71 The CSC was notified to the Commission on 11 December 1995.

72 See Annex XI to the Groupement's letter of 7 November 2003, file p. 3110.

73 See Annex XI to the Groupement's letter of 7 November 2003: "the CIR is based on a self-regulating mechanism. The amount of the CIR, like that of all interchange fees, including the Merchant Interchange Fee or CIP notified to the Commission on 20 December 1990, represents a balance between the divergent aims of the different members of the system, each of which pursues its own commercial strategy. The system therefore incorporates its own regulating mechanism which is born of the conflict of interests between issuers and ATM operators, with most of the members of the Groupement generally combining these two roles" (file p. 3110); "the cash advance introduces financial compensation; the CSR and the CSC, for their part, constitute the indispensable weighting mechanism without which it would not be possible for each member of the Groupement to pursue the commercial strategy of its choice (for example by specialising in the operation of ATMs or, on the contrary, deciding not to install any ATMs) while preserving the overall balance of the withdrawal function in the system" (file p. 3105; e(emphasis in the original).

74 See CCF's reply of 16 February 2005 to the request for information of 2 February 2005, file pp. 22422 and 22423.

75 See the Groupement's reply of 7 November 2003 to the Commission's request for information of 24 October 2003, file pp. 3003 et seq.

76 Budget item CS 2002, file p. 12448.

77 Budget item CS 2002, file p. 11954.

78 Memo entitled "Pricing/ATMs" dated 26 March 2001. File p. 12201.

79 Memo entitled "Pricing/ATMs". File p. 12202.

80 Summary report on the meeting of [...] on 29 August 2001. File p. 10749.

81 Memo dated 30 August 2001 from Crédit Agricole to Cedicam on the subject of "Groupement des Cartes Bancaires: Steering Committee meeting of 29 August 2001". File p. 11099.

82 Memo, author unknown, found in the office of the Groupement's [...], file p. 25839.

83 "GIE" means "Groupement d'Intérêt Economique" and designates the Groupement des Cartes Bancaires "CB".

84 Document referred to above AW 11 (GCB2), file p. 25839.

85 "Conversation with [...]", memo by [...] of Société Générale dated 19 October 2001, file p. 5736.

86 Document referred to above, file p. 5738.

87 Record of Steering Committee meeting, 9 November 2001. File pp. 14263 et seq.

88 Record of Steering Committee meeting, 9 November 2001. File pp. 14263 et seq.

89 Record of Steering Committee meeting, 9 November 2001, file p. 14273.

90 Letter from [consultancy firm] to [a] Groupement [representative] dated 15 November 2001. File p. 20921.

91 Confidential and personal working document, "CS 2002 project. Steering Committee", 18 December 2001. File p. 25842.

92 Document referred to above, file p. 25843.

93 Document referred to above, file p. 25848.

94 Document referred to above, file p. 25859.

95 Document referred to above, file p. 25849.

96 The page of the document referred to above (file p. 25859) containing this sentence is identical to that of page 10 (file p. 13806) of the [consultancy firm] document entitled " Projet CS 2002 Point DG " (file p. 13797 onwards) at the bottom of which figure la mention " Contrôle de Gestion CB ". This presentation for the COM was therefore carried out on the basis of data gathered from all of its members by the Groupement.

97 Document referred to above, file p. 25859.

98 Document referred to above, file p. 25860.

99 Document referred to above, file p. 25860.

100 [consultancy firm] document, undated: "The CB system founder banks must now face up to a threefold challenge". File p. 14207.

101 Document LS 103 (GCB2) referred to above, p. 8, file p. 14215.

102 Charges at 15 December 2001. [Consultancy firm] document referred to above, file p. 14216.

103 [Consultancy firm] document: "The CB system founder banks must now face up to a threefold challenge". File p. 14215.

104 Banque Accord's reply of 20 March 2003 to the Commission's request for information pursuant to Article 11 of Council Regulation No 17, p. 1, file p. 963.

105 Record of the COM meeting on 9 January 2002. File p. 4857.

106 Memo entitled "Distribution measures under consideration by GCB" (Cedicam), file p. 11390.

107 "The GIE is currently studying, on a confidential basis, changes to its structures ... Hence a number of studies within a very narrow circle (three persons at the GIE ...), with [La Poste] and myself acting as 'political commissars' for all the banks" (handwritten notes taken by [...] Société Générale [...], file p. 5710).

108 Memo referred to above, FMO2 (Cedicam), file p. 11389.

109 Internal BNP-Paribas memo entitled: "New economy of the CB system - Interim report", file p. 12018.

110 Cedicam document, file p. 5004 (author and date unknown; drafted shortly before 25 February 2002). See also handwritten notes taken by [...] Société Générale [...]. File p. 5592.

111 Record of the COM meeting on 26 February 2002. SDW3 (LP), file p. 10744.

112 Documents entitled :

- "CS 2002 project. Principles for setting up the SIC and charging for activities", 11 March 2002, file p. 13707.

- "CS 2002 project. Aspects of the organisation plan for the industrial and commercial structure", 11 March 2002, file p. 14146.

- "CS 2002 project. Characteristics of the simulation", file p. 13691.

113 Document referred to above, 18 March 2002, file p. 13693.

114 Document referred to above, 18 March 2002, file p. 13700.

115 Document referred to above, 18 March 2002, file p. 13706.

116 Internal Caisses d'Épargne (CNCEP) memo, file p. 13367.

117 Document entitled "CS 2002 project. Characteristics of the simulation. Société Générale", 22 March 2002. File p. 5541.

118 Document entitled "CS 2002 project. Simulations Crédit Agricole", 22 March 2002. File p. 6296.

119 Document entitled "CS 2002 project. Large retailers' CB card issuing strategy and analysis of the impact. BNP Paribas". April 2002. File p. 11955. This working document, drafted by [consultancy firm], enabled BNP Paribas to conclude that, over a two-year period, the effect of the measures would be favourable to it since they would reduce its potential loss of revenues by [...]%. File pp. 11971 and 11972.

120 Document dated 20 March 2002, with no indication of the author but marked "Amended [...]". Forwarded by letter from [...] ([adviser] SG) dated 1 March 2005, file pp. 23099 and 23100.

121 File p. 5713.

122 Handwritten notes taken by [...] Société Générale and a briefing note for the meeting between [...] and [...] 3 April, file p. 5707.

123 "- Founding principles of the CS 2002 project (cf. memo presented at the meeting of 28 March)

- Levels of charges

- Contours SIC/GIE

- Principles for apportioning the capital in the shareholder agreement

- Tax aspects

- Remit/MOU - Schedule - Budget

A dossier dealing with these topics will be forwarded to the members of the COM on 11 April", file p. 5709.

124 This is attested to by a number of documents. For example: "The GIE CB has supplied us with its spreadsheets, enabling us to carry out a number of simulations ... the model still needs to be further developed and the figures are not yet definitive ... It would appear that the way the model is currently designed penalises growth in the cardholder business more heavily than growth in the merchant and ATM businesses.

The reduction in the profitability of new activities which it causes can therefore be regarded as an 'insurance' against the financial impacts of large swings in market shares due to the arrival of new entrants" (Internal Caisses d'Épargne (CNCE) memo. Subject: "Electronic Banking Steering Committee - CS 2002 project." 4 April 2002. File p. 7054).

An internal La Poste email dated 12 April 2002 refers to the "simulations that were presented to us, and the spreadsheets that were made available to us for our own simulations" (internal La Poste email , file p. 13278).

125 Internal memo CNCEP of 4 April 2002 with the subject " Comité d'Orientation Monétique - Projet " CS 2002 " ", file p. 7050.

126 Handwritten (Société Générale) memo on the Steering Committee meeting on 12 April 2002. File p. 5706.

127 Ibid.

128 Item 2 on the agenda, file p. 11986.

129 "Electronic Banking Steering Committee - Meeting of 12 April 2002 - CS 2002 project". File p. 12013.

130 "Electronic Banking Steering Committee - Meeting of 12 April 2002 - Agenda". File pp. 14039 and 14040.

131 The document entitled "Schedule of work on the CS 2002 project" refers to the decision to abandon the idea of setting up an SIC as a "decision to give priority to the non-structural aspects of the project": "Mid-May 2002: In view of the difficulties of all kinds that would be involved in the implementation, within a reasonable timescale, of the structural reform contemplated, the Groupement decided to give priority to the nonstructural aspects of the project", file p. 10636.

132 "CS 2002 project: Large retailers' CB bank card-issuing strategy, 22 May 2002. CB logo. Confidential." SP 5 (BNP-Paribas), file p. 11937.

133 For example:

- According to an internal Caisses d'Épargne memo dated 27 May 2002:

"[T]he work done since 12 April, with the aid of [the Groupement's consultants], has come to the conclusion that the proposed arrangement is unacceptable from the standpoint of competition law, whatever the structure envisaged" (file p. 12782).

- According to an internal BNP-Paribas document also dated 27 May 2002:

"CB has carried out a 'heartrending revision' of the project following the veto by its lawyers, who consider the initial arrangement unacceptable from the standpoint of competition law (arrangement regarded as not in line with the general objectives of contributing to economic progress and improving the service rendered to customers).

There is no longer any question of setting up a company, or of introducing the planned charges for new activities. Instead, CB is proposing a more traditional reorganisation arrangement" ("CB reform", internal BNP Paribas document dated 27 May 2002. File p. 12046).

134 "It is unrealistic to think that the simple solution envisaged up to now (turning the GIE into a profit making company + tax on new cards) will attain the desired objective over a long enough period, even if it means facing after several years possible intervention by the competition authorities. Its anticompetitive nature is so blatant that such intervention would probably not be long in coming. The same cannot be said of the slightly more complex solution that we are proposing and that we could flesh out in the weeks and months ahead" (Email from [a consultant for the Groupement]. EEIG measures, file p. 5689).

"In the same way the JV could decide to tax some activities which do not raise the overall value of the network ... Of course this would raise antitrust scrutiny because this might be seen as a way to restrict competition in the card issuance market. But I guess antitrust problems would be less severe if Groupement Cartes Bancaires remains a nonprofit JV... and these taxes are off-set by subsidisation of other activities ... which is likely to push prices down ... such an efficiency driven tax/subsidy system would likely benefit the big 8 banks - our client" (document referred to above, file p. 5690).

135 Internal CNCE document entitled "Electronic Banking Steering Committee - 'CS 2002' project. Meeting on 29 May 2002", file p. 12783.

136 Internal Caisses d'Épargne (CNCEP) memo, file p. 12785.

137 See also the internal Caisses d'Épargne (CNCEP) email dated 27 May 2002 in which it is explained that "[A] new set of proposals will be developed at the meeting on 29 May 2002. It is a proposal pursuing the same objective, but highly simplified, based on the existing structures and no longer requiring a company with share capital to be set up, thereby avoiding the knock-on tax problems". It is also explained in this email that: "Caisses d'Épargne, like La Poste, are the only banks among the founder members that receive twice as much CIP as they pay and, therefore, could be very heavily penalised by this change, in the same way as 'new entrants'", and "the calculation rule is to be further developed so as not to penalise those two banks" (file p. 12783).

138 Presentation bearing the CB logo. "CS 2002 project. Progress report". 22 May 2002. File p. 11948.

139 Internal Crédit Lyonnais email on the preparations for the Steering Committee meeting on 28 May 2002. File p. 10567.

140 According to the record of the meeting:

"The ensuing discussions led to the following policy decisions:

1. Production of a general overview document on the solution contemplated ...

2. Detailed analysis of the mechanism envisaged with the colleagues in the banks by means of a set of simulations proving its soundness and demonstrating its effects over time for each of the members and for the GIE as a whole through intelligent variation of the different parameters. This techno-economic analysis should be used as a basis for discussion at the next meeting.

3. Confirmation of the preceding policy decision on the creation of special purpose subsidiaries, for example for approvals or the network" (file pp. 12037, 12038, and 12108).

141 According to the "schedule of work" for the CS 2002 project: "June-July 2002: Work will continue with the banks in order to determine in the greatest possible detail the practical arrangements for introducing the measures contemplated. Meetings will remain bilateral until 12 July.A working session to finalise the arrangements with all the representatives appointed by the members of the Board of Directors is scheduled for 26 July 2002" (file p. 10636)

142 File p. 25914.

143 Annex to document LS12 (GCB2), referred to above, file p. 25915.

144 Email from [the Groupement's consultant] to a [representative of] the Groupement [...], 14 June 2002: "The modified CIP has become: the mechanism for regulating the acquiring function". File p. 25912.

145 Document "CS 2002. Email with annexed note (CS 2002 progress report memo)" dated 14 June 2002 from [...] to [Groupement representatives] (LS10 (GCB), file p. 13899).

146 File p. 13900.

147 "The initial simulations by DAJB must be followed up so as to measure the mechanism's impact on all CB members: the traditional banks, retailers and private/regional banks, etc." ("Progress report memo", 14 June 2002, file p. 13900).

148 For example, according to handwritten notes by [...] (BNPP): "See Cetelem - Should be safe Scenario moderately dissuasive for large retailers Good protection v online banks, foreign banks" (handwritten notes taken by [...], 14 June 2002, SP41bis (BNPP), file p. 12166.

149 Document JPL 22 (GCB1). Memo for the CEOs Steering Committee meeting on 18 June 2002, file p. 25828.

150 See, for example, internal Crédit Lyonnais memo on the "CB Steering Committee meeting on 25 June 2002", file p. 10392; internal email dated 21 June 2002 prepared by [...] Société Générale (printed - not sent) marked "Confidential - subject: CS 2002", file p. 5601; memo to [...] BNP Paribas, subject: "Proposed reform of the Cartes Bancaires system", 21 June 2002, file p. 11915.

151 Internal Crédit Lyonnais memo on the "CB Steering Committee meeting on 25 June 2002", file pp. 10265 and 10392.

152 Memo from [...] to [...]. Copy to [...]. Subject: Proposed reform of the Cartes Bancaires system, 21 June 2002. SP2 (BNPP), file pp. 11915-11917.

153 Email from [...] to [...] on "CS 2002 / your meeting of 25 June". Date: 21 June 2002. EMR26 (SG), file p. 5601.

154 File p. 12036.

155 File p. 12036.

156 Internal Crédit Lyonnais memo on the Electronic Banking Steering Committee meeting. 27 June 2002. File p. 10419.

157 Communication dated 18 July 2002. "Electronic Banking Steering Committee - 'CS 2002' project. Meeting on 19 July 2002." MPI-20 (CNCEP2), file p. 7069.

158 Internal Caisses d'Épargne email. "Subject: CS 2002 - record of the meeting on 12 July 2002." 12 July 2002 or thereafter. File p. 13365. The three principles, according to this document, are:

- the creation of a mechanism for regulating the acquiring function (Merfa); ...

- a new method for setting membership fees: a progressive system based on the volume of cards issued

- changes to the formulae for allocating voting and ownership rights ... ".

159 See recital 285 below for an explanation of the SIREN concept as compared with the SIRET concept.

160 File p. 25838.

161 Internal La Poste email. 11 July 2002. File p. 16343.

162 Document "CS 2002 Banks meeting - CB of 12 July." File p. 14042.

163 Internal La Poste communication entitled "Summary of the CS 2002 meeting of 15/07/02 at the GIE CB". File p. 10598.

164 Internal La Poste document referred to above. File p. 10597.

165 Document "CS 2002 Banks meeting - CB of 12 July." File p. 14042.

166 "CB is to organise a meeting on the risks of these mechanisms being circumvented by the targeted issuers: risk in relation to cross-border issuance" (Internal La Poste communication, file p. 10598).

167 Document "CS 2002 Banks meeting - CB of 12 July." File p. 14042.

168 "We should receive simulations today and the dossier for the CEOs meeting scheduled for the 19th on Wednesday 17 July". Internal CNCEP email "Subject: CS 2002 - record of the meeting on 12 July 2002". File p. 13365.

169 Internal La Poste communication of 15 July 2002 : "Subject: Summary of the CS 2002 meeting of 15/07/02 at the GIE CB". File p. 10597.

170 Internal CNCEP communication: "Electronic Banking Steering Committee - 'CS 2002' project. Meeting on 19 July 2002." File pp. 7074 and 7070.

171 Document SB15 (CNCE), file p.12814.

172 Memo dated 24 July 2002 on the "Meeting of 17 July 2002 at the GIE on the risks of circumvention of the Merfa by certain banks". File p. 5731.

173 Memo dated 24 July 2002 on the "Meeting of 17 July 2002 at the GIE on the risks of circumvention of the Merfa by certain banks". File p. 5731.

174 Document EMR23 (SG). Electronic Banking Steering Committee meeting on 19 July 2002. Decision statement, file p. 12436.

175 "The GIE has a remit to resume the work and introduce a factor that will contribute to inertia, the pool of ATMs" (internal Société Générale record, file p. 5680).

176 Internal BNP Paribas email dated 23 July 2002 (file p. 12066). This email explains that "On the CEOs meeting at the GIE Cartes Bancaires, I have spoken on the phone to ..., [...] of the Groupement. The upshot is:

- Following an outburst by ... ([...] Crédit Agricole), who regarded the proposed measure as insufficient, the project presented was not endorsed.

- CB was asked to incorporate the numbers of ATMs into the calculation of the equilibrium ratio". 177 According to the minutes of the meeting:

"The Chair informed the Board of Directors that the work on CS 2002 had not yet come to fruition and that the CEOs Steering Committee had asked for the dossier to be supplemented.

The Chair pointed out that the costs of ATM operators needed to be studied for international systems in order to defend withdrawal acquirers vis-à-vis issuers.

The Board of Directors took note of this information" (minutes of the meeting of the Board of Directors held on 26 July 2002; file p. 10894).

The CS 2002 project was originally included in the agenda for the Board meeting on 26 July 2002 in order to have the measures formally adopted.

The motion initially prepared was, however, replaced by another one according to which the Board took note that the work had not come to fruition ("Minutes of the meeting of the Board of Directors held on 26/07/02", file pp. 10958-10968).

This change was prompted by the need to incorporate ATMs into the calculation of the Merfa:

" 4. CS 2002. The item was withdrawn from the agenda in view of the decision taken by the Steering Committee on 19 July 2002 to include the problem of ATMs in the discussion" (record of the meeting of the Cartes Bancaires Board of Directors on 26 July 2002, file p. 10164).

178 "Schedule of work", file p. 10635.

179 "Interbank challenges of electronic banking", internal BNP memo dated 13 August 2002. File p. 12203.

180 Internal La Poste memo, 16 September 2002. Subject: "Summary of the CS 2002 meeting on 13 September 2002 at the GIE", file p. 4705.

181 Formula determining: (1) contributions to the Groupement's operating costs; (2)voting rights within the Board of Directors; and (3) allocations in the event of liquidation, winding-up or the distribution of assets or the acquiring of holdings by the Groupement, file p. 4707.

182 "I. Introduction: The context

... There are many factors at work in the evolution of the CB environment ...

II. The objectives pursued and the measures proposed

The value of a payment system is greater than the sum of the investments made by each of those taking part: there is here a network effect, recognised by economists under the name of 'positive externalities' ... acquiring generates more positive externalities than issuance ...

II.1 The first objective is to take greater account of members' investment and effective contribution to the CB system and its development ...

II.2 The second objective is to adjust the CB system membership fees ...

II.3 The third objective of the measures presented is to create for members that are card issuers an incentive to develop their acquiring activities so as to continue to contribute to the increase in the value of the system ...

III. Conclusion:

(1) The mechanisms described strengthen interbanking and favour acquiring activity by ensuring a better return on the investments made by the CB members.

(2) Increasing the system's added value requires both an effort to capture sectors in which the growth of acquiring will be difficult and development programmes costly and adaptation of the pool of ATMs to the new security environment without compromising the scope of the withdrawal service.

(3) The new arrangement therefore constitutes a package that is economically justified in order to develop a product that everyone recognises as reliable, rapid and secure. The Board of Directors is invited to state its views on all these measures having due regard to the economic and legal opinions that have been presented to it". File pp. 4708-4712.

183 "Large retailers' CB bank card-issuing strategy. Confidential. Groupement [...] copy." 17 September 2002. File pp. 12397, 12398 and 12404.

184 "CS 2002. General overview. Summary". Author unknown. 25 September 2002. See, in particular, file pp. 13293 and 13298.

185 "CS 2002. General overview. Summary". Author unknown. 25 September 2002. Document referred to above.

186 Table prepared by the Groupement ("GIE CB: Audit") with the number of cards, number of ATMs, number of SIRENs, number of contracts and contract/SIREN ratio. The table includes a calculation of the Merfa for the main members. 18 September 2002, file p. 4774.

187 As can be seen from an internal Société Générale email dated 2 October 2002:

"... - A big problem: CA's pool of ATMs, in which there are many 'shoebox' type machines" ... Idea: find a way of differentiating between machines (the real ones and the others) using a weighting (1 'other' = 1/3, 1/2, etc. of a real one?) in order to take account of the problem. But Caisse Nationale du Crédit Agricole rather embarrassed by this sensitive problem;

The simulations seem not to give rise to any problem for La Poste". File p. 5696.

188 "Positions of the other banks

BNP: pushing for the interbank counting of transactions 'on us' and for refusing to approve ATMs that do not satisfy criterion A above.

SG: pro Merfa. Positive impact on the banks' revenues. Worth having.

CA : pushing for the introduction of criteria A and B in the approval list for ATMs.

CL: pro Merfa. No apparent reservation.

La Poste & CE: pro Merfa. Measure not challenged by these two banks.

CM: no view expressed at the meeting".

Preparations for the meeting of Cartes Bancaires [...] on 11 October 2002, file p. 5079.

189 Communication referred to above: preparations for the meeting of Cartes Bancaires CEOs on 11 October 2002, file p. 5077.

190 Communication referred to above, file p. 5077.

191 Memo dated 9 October 2002 headed "CS 2002 project. Meeting on 11 October 2002". File p. 7097.

192 Internal Société Générale memo dated 9 October 2002 on "CS 2002/CB Steering Committee of 11 October 2002". File p. 5557.

193 Internal communication (Natexis Banques Populaires) dated 14 October 2002 on the memo dated 8 October 2002 in preparation for the COM meeting on 11 October 2002. Annex 1: "Impact of the Merfa on the main members' incomes. Assumptions". File p. 5082.

194 Document entitled "List of decisions taken by the COM on 11 October 2002" from [...] to [...], CP03 (GCB), file p. 13939.

195 [...] handwritten notes, JPL7 (GCB1), file p. 13585.

196 Internal BNP Paribas email dated 22 October 2002. "Subject: Plan to overhaul the CB system", file p. 12207.

197 Extract from the minutes of the Board of Directors meeting on 8 November 2002, Annex 2 to the notification made by the Groupement on 10 December 2002, file p. 45. At the meeting, in reply to a question about the list of "approved ATMs", "[a Groupement representative] answered that the list was finalised, subject to clarifications requested by certain members, which would not affect the principles on which it was based" (minutes of the Board meeting on 8 November 2002, distributed by the Groupement, file p. 12974).

198 According to an internal Crédit Lyonnais email dated 12 November 2002, "The discussions that have been held on this occasion were covered by an important confidentiality obligation, broken on the eve of the Board of Directors meeting by certain members". File p. 10320.

199 "In accordance with the decisions taken by the Board of Directors on 8 November 2002, please find enclosed an information memo addressed to your secondary members. It is important that this memo be brought as quickly as possible to the attention of the members attached to you". Memo dated 13 November 2002 prepared by the Groupement, file p. 10970.

200 Board of Directors meeting on 29 November 2002. Discussion of Annex 2: Dormant members wake-up mechanism. 29 November 2002, file p. 13016.

201 "Ahead of the Board meeting on 20 December GIE CB is proposing as agreed an arrangement for cases like that of Groupama Banque (not mentioned in the dossier): members that joined GIE CB before the new system was introduced and have not developed a significant CB activity will be treated as if they had joined the GIE on 1 January 2003". Internal Société Générale email dated 18 December 2002. Subject : "REFORM CB". File pp. 5662 and 12613.

202 Internal SG email - "Valorisation CB - Case of Egg", file p. 5646.

203 Internal BNP Paribas email. Subject: "Egg" 5 November 2002. File p. 6363.

204 Internal SG email - "Optimisation of CB - Case of Egg", file p. 5646.

205 Internal Crédit Lyonnais email dated 12 November 2002, file p. 10320.

206 Internal Société Générale record on "GIE CB [...] Committee meeting on 7 January 2003", file p. 12427.

207 Internal CNCEP email on the record of the GIE CB meeting of 7 January 2003, file p. 7083.

208 See paragraph 19, page 6 (file p. 7) of the notification of 10 December 2002.

209 Paragraphs 24-29, pages 7 and 8 (file pp. 8 and 9) and annex 2 of the notification (file pp. 45 et seq.).

The Merfa formula as notified by the Groupement is the following (file pp. 52-53) :

" 1. Critère d'éligibilité au Merfa

Seront définis, pour chaque banque [i], deux ratios établis annuellement et définissant sa contribution aux activités d'acquiring:

• Primo, un ratio de contribution à l'activité retrait [R(i)] calculé par la formule:

R (i) = (Nb. De DAB de la banque [i] / Nb. Total de DAB)

• Secundo, un ratio de contribution à l'activité d'acquiring auprès dès accepteurs [ S(i)] calculé par la formule:

S(i) = (Nb. De SIREN actifs de la banque [i] /Nb. Total de SIREN actifs)

A partir de ces deux ratios, on calculera un indice global de contribution à l'acquiring [A(i)] en faisant la moyenne des deux indices précédents:

A(i) = 1/2 *(R (i) +S(i))

Enfin, on établira un indice d'utilisation du Système "CB" en émission [E(i)] par la formule:

E(i) = (nb. De cartes "CB" actives de la banque [i] /Nb. Total de cartes "CB" actives) et on calculera T(i) =A(i) / E(i)

Sera éligible au Merfa au titre d'un exercice donné toute banque qui satisfait à la condition suivante:

T(i) < 1/2

2. Détermination de l'assiette et du montant du prélèvement

Pour toute banque [i] qui satisfait à cette condition, on définira l'indice de prélèvement [P (i )] qui lui est applicable:

P(i) = 1-2 * T (i), qi varie entre 1 et 0 lorsque T (i) varie entre 0 et 1/2 .

Les montants prélevés seront égaux à:

• P(i)*€ 11 hors taxes par carte de retrait et de paiement active; et

• P(i)*€ 3 hors taxes par carte de retrait active. "

210 Point 25 of the notification, file p. 8.

211 The Groupement's reply of 24 March 2003 to the Commission's request for information of 3 March 2003, paragraph 1.1, file p. 1240.

212 The relative activity of a given bank, as both an acquirer and an issuer, is the ratio between its own activity and the activity of all the banks belonging to the CB system (see paragraph 27 of the notification, file p. 9).

213 Active SIRENs: SIRENs corresponding to outlets with which the member in question has at least one current CB affiliation contract and which have accepted at least one payment transaction during the course of the year.

214 Active card: a card that has been used at least once during the course of the year.

215 See Annex 3 to Annex 2 of the notification, file pp. 52 et seq.

216 Paragraph 29 of the notification, file p. 9.

217 See the Groupement's reply of 24 March 2003 to the Commission's request for information of 3 March 2003, answer to question 12 ("Section 1.(iii): effects of the Merfa"), file p. 1244.

218 Paragraphs 30-35, pages 8 and 9 (file pp. 9 and 10) and Annex 2 of the notification (file p. 49).

219 According to the notification, The tripling of card issuances by a member during years 4, 5 and 6 after it joined suggests that the institution in question deliberately scaled back its business in years 1, 2 and 3 in order to limit the membership fee based on its issuances during those years. The introduction of an additional membership fee is therefore intended to deal with this situation" (end of paragraph 32, page 9, file p. 10).

220 Annex 2 to the notification (file p. 49) and Groupement's reply to the request for information of November 2003, point 2 (page 4, file p. 2437).

221 Paragraphs 34 and 35 (page 9, file p. 10), and Annex 2 of the notification (file p. 50).

222 Paragraph 34 of the notification (page 9, file p. 10).

223 The arrangement was temporary and was not to apply after 2005.

224 See the Groupement's reply of 24 March 2003 to the Commission's request for information of 3 March 2003, answer to question 24. File p. 1249.

225 Memo to CB members containing "Explanations and simulations helping to clarify the implementation of the rules that have been adopted", p. 10/19, file p. 1346.

226 Abovementioned reply from the Groupement of 24 March 2003, answer to question 19, file p. 1248.

227 See paragraphs 36-39 (pages 9 and 10, file pp. 10 and 11) of the notification and point 19.2 of the Groupement's reply of 24 March 2003 to the Commission's request for information of 3 March 2003, file p. 1248.

228 The transactions taken into account for the calculation of the average number of transactions are payments and cash withdrawals made by each member's CB cardholders and transactions presented for settlement by each member's affiliated merchants, file p. 48.

229 Internal Société Générale email, file p. 5655.

230 Groupement's reply of 10 March 2005 to question 4 of the Commission's request for information of 11 February 2005, page 2 and Annex 3 (file pp. 23154-23155 and 23246-23256, and file pp. 23311-23312 and 23401-23411).

231 As indicated in section 4.3 above, the Board of Directors is alone empowered to adopt measures concerning the costs of the system; the General Meeting was not consulted about the adoption of the Merfa and the new membership fees.

232 File on preparations for the COM meeting on 7 January 2003. 30 December 2002. File p. 12073. See also the "Memo to CB members" attached to the communication from the Groupement to the Board of Directors of 17 January 2003 (pages 1 and 9), file pp. 10356 and 10364.

233 "Lastly, as for those members that joined in 2002, they will be given a choice between being considered new members or 'awakened' dormant members". Implementation of the CS2002 project. COM note 7 January 2003. File p. 13476. See also the "memo to CB members" attached to the communication from the Groupement to the members of the Board of Directors of 17 January 2003, point 3 of Annex 2.2 (p. 10 of the document), file p. 10365.

234 Main points of the [Groupement's consultant] report. Email from [a Groupement representative] to [Groupement's consultant], 25 March 2003, file pp. 25976 et seq.

235 Main points of the [Groupement's consultant] report. Email from [a Groupement representative] to [Groupement's consultant], 25 March 2003, file pp. 25977 and 25978.

236 Letter from the Groupement to the Commission dated 10 June 2004, file p. 25775.

237 See the Groupement's reply of 14 September 2004 to the Commission's request for information of 30 July 2004, file p. 15710.

238 Motion of the Groupement's Board of Directors of 28 July 1995, notified to the Commission on 4 September 1995 (Case IV/35.700), as amended by the motion of the Groupement's Board of Directors of 22 September 2000 attached as Annex XIII to the Groupement's reply of 24 October 2003 to the Commission's request for information of 7 November 2003, file p. 3124.

239 See notification, paragraphs 53-58, pages 17-18, file pp. 18-19; additional notification of 23 December 2002 (reply to the Commission's request for information of 17 December 2002), paragraph 25, page 7, file p. 233.

240 See paragraph 19 (p.7) of the Groupement's reply of 19 October 2006 to the 2006 Statement of Objections, file p.26605 ; et paragraph 74 (p.26) Groupement's reply of 8 November 2004 to the first Statement of Objections of July 2004, file p. 20751.

241 See the Visa International Decision of 24 July 2002, recital 47 (OJ L 318, 22.11.2002, p. 17).

242 See the Visa International Decisions of 9 August 2001 (OJ L 293, 10.11.2001, p. 24) and 24 July 2002 (OJ L 318, 22.11.2002, p. 17), recitals 39 and 48.

243 Witness the sharp fall in the number of "domestic bank cards", which are not accepted for payments abroad. Whereas in 1998 this type of card outnumbered all others, by the end of 2002 domestic bank cards accounted for just 9% of all CB cards in circulation (Source: Retail Banking Research, "Payment Cards in Europe 2004", France, p. 7, file p 26164).

244 Whereas in 2002 payment transactions made with CB cards and with cheques were equivalent in number, accounting for 31.15% and 32.22% respectively of the total number of mass payment transactions, the value of payments by CB card came to 3.42% of the total value of mass payments, compared with 43.47% for cheques, or 12 times more (Additional notification by the Groupement of 3 February 2003, Annexes I and II, Bank of France statistics on mass payments, file pp. 266, 271 and 272).

245 See the above-mentioned Visa International Decisions of 9 August 2001 and 24 July 2002, recitals 38 and 48 respectively.

246 See section 3.2.2 above.

247 Latest figures for 2005, document entitled "2005 Facts and Figures", March 2006, available on the Groupement's Internet site www.cartes-bancaires.com, file p. 25757.

248 See RBR Report "Payment Cards in Europe 2004", chapter on France, pp. 18 and 36, file pp. 25726 and 25734. The number of merchants and ATMs accepting Aurore cards are 2002 figures and are therefore compared to the number of merchants and ATMs accepting CB cards during that same year, i.e. 738 000 and 39 000 respectively.

249 At end-2002, Franfinance's cards were accepted in 30 000 sales outlets, those of Cofinaga in 25 000, those of Banque Accord in 2 700, those of S2P in Carrefour hypermarkets, and those of Finaref in 210 sales outlets. RBR Report "Payment Cards in Europe 2004", chapter on France, pp. 18-20, file pp. 25726- 25728.

250 Figures for 2004, RBR Report "Payment Cards Western Europe 2006", chapter on international card organisations, page 10, file p. 25780.

251 In 2004 there were hardly more than 30 000 Diners Club cards, and the number of Diners cards has not increased much over the past years, RBR Reports "Payment Cards Western Europe 2006" and "Payment Cards in Europe 2004", chapters on France, pp. 9 and 29, file pp. 25777, 25731, and 26166.

252 RBR Report "Payment Cards in Europe 2004", chapter on France, p. 33: "The average payment value on cards issued in France is dwarfed by that made on the cards of Amex, Diners and consumer credit organisations. The average value of a payment with a CB card was euro 46 in 2002 [against 222 with private label cards according to Figure 20]", file p. 27613.

253 See RBR Report "Payment Cards in Europe 2004", chapter on France, p. 32, file p. 25733.

254 See section 1.4.1 of the Groupement's reply of 19 October 2006 to the 2006 Statement of Objections.

255 See section 1.4.2 of the abovementioned reply.

256 See section 1.4.3 of the abovementioned reply, and in particular paragraphs 80 and 81.

257 See section 1.3 of the Groupement's reply of 19 October 2006 to the 2006 Statement of Objections.

258 See paragraph 34 of the Groupement's reply of 19 October 2006 to the 2006 Statement of Objections, file p. 26608.

259 See recitals 43 and 45 of the Visa International - Multilateral Interchange Fee Decision of 24 July 2002 (Case COMP/29.373); OJ L 318, 22.11.2002, p. 17.

260 See recital 65 of the abovementioned Decision.

261 Grüner+Jahr/Financial Times/JV Decision of 20 April 1999 (Case IV/M.1455), recital 15, OJ C 247, 31.8.1999, p. 5; Recoletos/Unidesa Decision of 1 February 1999 (Case IV/M.1401), recital 16, OJ C 73, 17.3.1999, p. 8; CEP/Groupe de la Cité Decision of 29 November 1995 (Case IV/M.665), recital 9, OJ C 338, 16.12.1995, p. 3.

262 Microsoft Decision of 24 March 2004 (Case COMP/C-3/37.792), recital 449. Available on the Internet at: http://ec.europa.eu/comm/competition/antitrust/cases/decisions/37792/en.pdf

263 Commission Notice on the definition of the relevant market for the purposes of Community competition law, OJ C 372, 9.12.1997, p. 5.

264 Point 66 of the Groupement's notification, page 21, file p. 22.

265 See footnote 238 above.

266 Paragraph 59 of the Groupement's notification dated 10 December 2002, page 18, file p. 19.

267 Opinion of Advocate General Ph. Léger delivered on 10 July 2001 in Case C-309-99 Wouters [2002] ECR I p. 1577, recital 62.

268 According to the case law of the Court and the administrative practice of the Commission, an association of undertakings in the sense of Article 81(1)EC is generally characterized by the existence of a standing body or other statutory arrangements for coordination between the members (see in particular the judgement of the Court in the " Wouters " case of 19 February 2002, Case C-309-99, recital 64 , and the Commission Decision in the case " Exposition européenne de la machine-outil " of 13 March 1969, OJ- L 69 of 20 March 1969, pp. 13 to 20). (See also legal commentary : " Kommentar zum deutschen und europäischen Kartellrecht, Band 2, Europäischen Kartellrecht " by Langen and Bunte, paragraphs 62 to 64, pp. 65 et 66).

According to the opinion of Advocate General Ph. Léger in the Wouters case, recitals 61 and 62 : " As a general rule, an association consists of undertakings of the same general type and makes itself responsible for representing and defending their common interests vis-à-vis other economic operators, government bodies and the public in general.

[...]

Article 85(1) [now 81(1)] covers not only direct methods of coordinating conduct between undertakings (agreements and concerted practices) but also institutionalised forms of cooperation, that is to say, situations in which economic operators act through a collective structure or a common body [decision of an association of undertakings]".

269 See in particular Article 7.1 of the Groupement's Formation Agreement, file pp. 24113 and 24139.

270 See in particular Article 3 of the Groupement's Formation Agreement, file pp. 24111 et 24137.

271 The Groupement has claimed to the Commission that "The Groupement constitutes, therefore, an association of undertakings in the sense of Article 81(1)EC [...] the notified measures constitute "decisions of an association of undertakings" in the sense of Article 81(1)EC" [" Le Groupement est donc une association d'entreprises au sens de l'article 81 paragraphe 1 du traité CE [...] les mesures notifiées constituent des " décisions d'une association d'entreprises " au sens de l'article 81 §1 du traité CE "] and that "it is a decision of an undertaking which was adopted on 8 November 2002, concerning the measures notified to the Commission" [" c'est une décision d'entreprise qui a été adoptée le 8 novembre 2002, portant sur les mesures notifiées à la Commission "] (See, respectively, the complementary notification of 23 December 2002 (reply to the Commission's request for information of 17 December 2002), paragraphs 1 and 4, page 2, file p. 228, and the title of Section 3.2.4. of the reply of the Groupement dated 8 November 2004 to the first Statement of Objections of the Commission of 7 July 2004, file p. 20759.

272 "[...] In this case membership of the association [the Groupement] entails, by virtue of the document constituting the association, the adhesion of its members to the decisions adopted by the managing bodies of the Groupement."; "[...] the Court considers [...] that the infringement was committed by the association of undertakings constituted by the Groupement". Joined cases T-39-92 and T-40-92 Groupement des cartes bancaires "CB" and Europay International SA v. Commission, [1994] ECR, p. II- 49, recitals 76 and 139.

273 Joined Cases 29-83 and 30-83 Compagnie royale asturienne des mines and Rheinzink v Commission [1984] ECR 1679, recital 26.

274 Commission guidelines on the application of Article 81(3) of the Treaty, OJ C 101, 27.4.2004, recital 22.

275 See recital 85 above.

276 See recital 90 above and footnotes 133 and 134.

277 See section 2.1 of the Groupement's reply of 19 October 2006 to the Commission's Statement of Objections of 18 July 2006, file p. 26620 onwards.

278 See Section 2.3 of the Groupement's reply of 19 October 2006 to the Commission's Statement of Objections of 18 July 2006, file p. 26623 onwards.

279 See paragraph 112 of the Groupement's reply, file p. 26624.

280 See section 2.5 of the Groupement's reply of 19 October 2006 to the Commission's Statement of Objections of 18 July 2006, file p. 26625.

281 Paragraph 25 of the notification, p. 7 (file p. 8).

282 Paragraphs 24 and 35 of the notification, p. 9, file p.10.

283 Due to the formula of the " dormant members wake up mechanism " (see recitals 147 to 149 above), les membres qui en sont potentiellement redevables peuvent y échapper en limitant leur activité d'émission durant les années 2003, 2004 et 2005. En revanche, ils ne peuvent pas y échapper en augmentant, notamment, leur activité d'acquiring (nombre de DAB et de SIREN) durant les années 2000, 2001 et 2002 - puisque ces années sont déjà écoulées lorsque cette mesure est portée à leur connaissance -, ni en augmentant leur activité d'acquiring durant les années suivantes - puisque celles-ci ne sont pas prises en compte dans la formule du droit de réveil des dormants.

284 See the above footnote (n° 283).

285 Groupement's reply of 24 March 2003 to the request for information of 3 March 2003, question 1.2 (file p. 1240).

286 Payment and withdrawal acquirings measured, as for the Merfa, by the number of active ATMs and SIRENs. See the Groupement's reply of 16 March 2005 to the Commission's request for information of 11 February 2005, file pp. 23715 to 23719.

287 [Groupement consultant], "Additional economic study on the externalities in the CB system and the effects of the mechanism for regulating the acquiring function", 26 October 2004, IV-2, pp. 26 and 29 (file pp. 20980 and 20983).

288 Citigroup's comments of 23 May 2003 following the publication in the OJ of the agreement notified by the Groupement on 10 December 2002 (file p. 2072).

289 Abovementioned comments by Citigroup Inc. (file p. 2072).

290 Covefi's reply of 19 July 2005 to the request for information of 29 June 2005 (file p. 24922).

291 Banque Accord's reply of 19 July 2005 to the request for information of 29 June 2005 (file p. 24930).

292 CNCE's comments of 8 November 2004 in reply to the Statement of Objections of July 2004, footnote 24 (in paragraph 56), p. 17 (file p. 19978).

293 See the replies of GE Money Bank, Banque Casino, Covefi, Banque Accord, Cofidis, Capital One, Citibank, Finaref, Banque AGF, Egg and S2P constituting Annex 3 to the 2006 Statement of Objections: replies to the Commission's requests for information of 26, 28 and 28 February and 3 March 2003 and of 29 June 2005 (file pp. 25712, 24660, 2287, 25679, 1004, 24929, 6630, 1175, 2061, 1773, 1889, 6701 and 25638 to 25640).

294 Cofidis's reply of 20 March 2003 to the Commission's request for information of 26 February 2003 (file p. 6630).

295 See footnote 283 above and the formula of the " dormant members wake-up mechanism " (recitals 147 to 151 above).

296 See the notification of the Groupement, paragraph 29 (file p. 9) and the table (set out in footnote 406 below) sent by the Groupement in Annex 5 to its reply of 24 March 2003, which show that the main members receive nearly all the amounts paid under the Merfa (file p. 1321).

297 See the Groupement's reply of 24 March 2003 to the Commission's request for information of 3 March 2003 (section 1(iii): effects of the Merfa (file p. 5)). " 12.1. The members who receive the Merfa are free to use the amounts they receive as they choose" (file p. 1244).

298 (Number of SIRENs of member / total number of SIRENs) + (Number of ATMs of member / total number of ATMs), see recital 142 above.

299 Number of active CB cards of member/total number of active CB cards in the system, see recital 142 above.

300 En mettant en rapport, d'une part, la part du membre dans l'activité d'émission de l'ensemble du système durant les années 2003, 2004 et 2005 avec, d'autre part, sa part dans l'activité (acquistion et émission) du système durant les années 2000, 2001 et 2002, afin de sanctionner ceux dont un tel rapport entre deux ratio serait supérieur à 3 (voir le formule du droit de réveil des dormants aux considérants 0 à 0), le Groupement impose à chaque membre de ne pas s'écarter de l'équilibre prévalent parmi les chefs de file, dont rien ne garantit qu'il soit le meilleur pour le système.

301 Submission by GE Money Bank at the hearing of 13 November 2006, file pp. 27415 and 27416.

302 Submission by GE Money Bank at the hearing of 13 November 2006, file p. 27406.

303 Banque Accord's presentation to the hearing of 13 November 2006.

304 See recital 144 above.

305 Introductory remarks by [a] Groupement [representative], when opening the COM meeting of 9 November 2001 (file p. 14263).

306 Annex XI to the Groupement's reply dated 7 November 2003, which reproduces the notification of the CIR made on 11 December 1995 (file pp. 3104-3105). In addition to this passage, the Groupement justifies the CIR in this document on the grounds of its regulatory function between issues and acquirers with the following remarks: "It is therefore necessary to introduce a financial compensation mechanism, without which the members of the system would not agree to ensure interbanking cooperation by dispensing banknotes to the other members' cardholders. ... the system's withdrawal function can operate only if the number of cardholders and the number of ATMs from which they make withdrawals grow in a generally balanced manner and if the imbalances observed at certain banks are corrected by way of financial compensation". "The goal is to allow the balanced development of the various components of the CB system in connection specifically with automated cash withdrawals" (file pp. 3104-3105).

307 Banque Accord's reply of 19 July 2005 to the Commission's request for information of 29 June 2005 (file p. 24930).

308 Reply of GE Capital Bank of 20 March 2003 to the request for information of 28 February 2003 (file p. 6328) and reply of GE Money Bank of 21 July 2005 to question B-5 of the request for information of 29 June 2005 (file p. 24659).

309 See the Groupement's reply of 19 October 2006 to the Statement of Objections of 18 July 2006, section 2.5.4.4, paragraphs 213 to 220, file pp. 26643 to 26645.

310 The present decision does not take a position on the lawfulness of the CIP/CIR interchange fees, nor on the role that they play or the real effects that they have as these CIP/CIR interchange fees currently stand.

311 Annex XI to the Groupement's reply of 7 November 2003 containing the notification of the CIR of 11 December 1995 (Case 35859), (file pp. 3094-3095).

312 Provided this share is bigger than the member's share of active ATMs in the number of active ATMs in the CB system and its share of active SIRENs in the number of active SIRENs in the CB system for the years 2000-02. See formula for calculating the wake-up fee, section 7.4.

313 (a) According to the CB memorandum entitled "CS 2002 project: Progress report", the CS 2002 project was set up in order to hinder competition from new entrants while ensuring the "preservation of the founders' electronic banking income".

(b) In a memo on the meeting of the COM on 19 July 2002, a CNCEP representative stated that the proposal was aimed "at reacting to various threats, in particular the fear of the large-scale entry of partners who would capture shares of the card issuance market and cut prices".

(c) According to an internal CNCEP email dated 24 May 2002, the second "defensive measure" successively proposed by the Groupement (the first one having been the SIC, now called into question) involved "penalising, without changing the present legal structure, new entrants by introducing new 'facially' objective criteria".

(d) In an internal Société Générale email, it is explained that the Merfa consists in "making new entrants pay on the basis of their issuance activity" (see recital 101(c)).

(e) The Merfa was also regarded as "a penalty" by a representative of the Groupement in a memo of 14 June 2002 reporting on the progress of the CS 2002 project.

(f) On the subject of the new dormant members wake-up mechanism, a representative of the Société Générale also referred to nature as a penalty (see recital 125(b) above).

"The new membership fee ... is supplemented by a complex, but essential, mechanism penalising any existing dormant members that start to issue massively.

We have not found any simpler alternative that can target the wake-up of the dormant banks."

(CS 2002 project: Progress report, 22 May 2002. CB logo. Confidential (file p. 11945). Internal CNCEP memo of 18 July 2002 "Electronic Banking Steering Committee - 'CS 2002' project. Meeting of 19 July 2002" (file pp. 7069 to 13379). Internal CNCEP email "Subject: CS 2002" (file, p. 13389). Groupement email of 14 June 2002 "Subject: CS 2002. Memorandum attached. CS 2002 progress report" (file p. 13899); Internal Société Générale memo on "CS 2002/CB Steering Committee of 11 October 2002", dated 9 October 2002 (file pp. 5557 and 12382)).

314 According to a general overview of the CS 2002 project found on the premises of Cedicam, the "founder members" (main members of the Groupement) were not to be penalised: "The introduction of the proposed measures and the distribution of the charges paid should not have an impact on the founder members as compared with the current situation". According to a CB presentation on the progress report on the CS 2002 project dated 22 May 2002, the sole beneficiaries of the CS 2002 project would be the founders: "Founders as sole beneficiaries of the CS 2002 reform". (Memo on "CS 2002. General overview". Version of 25 September 2002 (file pp.11290 and 12418); CS 2002 project. Project report, 22 May 2002. CB logo. Confidential (file p. 11945)).

315 Internal memo, Subject: Electronic Banking Steering Committee - CS 2002 project. Meeting of 29 May 2002, dated 27 May 2002 (file p. 12783).

316 See inter alia sections 10.2.2.1.A.1 and 10.2.2.1.A.3., recitals 265 to 278, 282 to 286, 291 to 295 and 299 below, and section 10.2.1.1.A.1, recitals 201 to 217 above.

317 - An internal CNCEP e-mail states that the intended purpose of the reform is to limit the loss of income to the founders: "[T]he aim of the model is to see that the system enables the income losses caused by the arrival of new entrants to be reduced"; "The loss of net banking income for the founder members, in the event of a large-scale entry of new players, would be reduced by one third by these mechanisms (loss of €93 million instead of €140 million, assuming the entry of four retailers issuing 4 million cards over a period of two years). In conclusion, these new measures appear to contribute to the objectives pursued".

- In a memo circulated among the parties to the agreement, the safeguarding of the incomes of the main members sitting on the COM was mentioned by the Groupement as being the purpose of the reform: "The introduction of the proposed measures and the distribution of the charges paid should not have an impact on the founder members as compared with the current situation".

- In the fax sent by a [representative of] the Groupement to the CNCEP, it is written that the Merfa reduces the loss of income: "This picture highlights the real issues: if the new issuers reach the figure of 5 million cards in five years (10% of all cards, which is not an unreasonable assumption), the loss or shortfall suffered by the main members is in the order of one billion euros over five years! The introduction of the Merfa reduces it to about €300 million (or 30%)!".

- A presentation by the Groupement on the characteristics of the simulation of the CS 2002 project and a CNCEP internal memo state that the system helps reduce income losses and limit the fall in the GDP of current members caused by the arrival of new entrants (see recitals 79 and 80 above).

(CNCEP internal email "this evening's GCB meeting", sent on 25 March 2002 (file pp. 7074 and 13367); Memo on "CS 2002 project. General overview", 1 October 2002 (file pp. 11290 and 12418); Fax on "CS2002", 8 October 2002 (file p. 13290)).

318 BNP Paribas document entitled "Interbank challenges of electronic banking", 13 August 2002 (file p. 12203).

319 CS 2002 project. Steering Committee. 18 December 2001. Working Paper (file p. 8109).

320 See the notification of the Groupement, paragraph 29 (file p. 9) and the table sent by the Groupement in Annex 5 to its reply of 24 March 2003, which show that the main members receive nearly all the amounts paid under the Merfa (file p. 1321).

321 Commission Decision of 24 July 2002 in Industrial and medical gases, OJ L 84, 1.4.2003, p. 1, recital 357: "Price being the main instrument of competition, the various collusive arrangements and mechanisms adopted by the suppliers were all ultimately aimed at an inflation of the price (or at least to limit its decline) to their benefit and above the level which would be determined by conditions of free competition".

322 "(b) Annual cardholder fees collected by the founder banks:

The founders' card issuance revenues are affected by three factors:

- the direct loss of customers (from which others do not suffer). It is assumed here that card issuance by new entrants is to the detriment of the founders;

- price cutting on new cards ...;

- the contagion effect on the number in circulation, which will result in a drop in annual fees on the portfolio, amounting to 5% of cards per year (5 % in the first year, 10 % in the second, etc.), the reduction being the same as for newly issued cards)".

(Natexis Banques Populaires internal memo of 14 October 2002 on CS 2002 which contains an annex drawn up by the Groupement on the Merfa's impact on the main members' income. Scenarios (file p. 5082)).

323 In addition to the examples already referred to (see recital 72 in fine), other elements prove the existence of such an objective:

(a) In a handwritten note dating from September 2001 (see recital 63 above), obtained during the inspections on the Groupement's premises, reference is made to the introduction of a fee per card issued in order to prevent a fall in prices: "to avoid slashing prices, selling prices have to be pegged: one (or more) issuance fees therefore have to be introduced".

(b) Crédit Lyonnais refers to the initial objective of reducing the cost to the founder members of any large-scale issuing of bank cards at prices significantly undercutting their own (see recital 101 (a)).

(c) Other representatives of large main member banks referred to the wish to "hinder dumping by new entrants" (see, for example, recitals 92and 126(b)) or to "safeguard the cardholder fee".

(d) In a CNCEP internal memo, it is written that:

"An impact study on these measures shows that the new entry fees and the mechanism for regulating the acquiring function will reduce by 25% the gains expected by the large retailers through card issuance.

This reduction in the expected gains would reduce the discount effect on the prices of these new entrants, and consequently the contamination of our own prices.

The loss of net banking income for the founder members, in the event of a large-scale entry of new players, would be reduced by one third by these mechanisms (loss of €93 million instead of 140, assuming the entry of four retailers issuing 4 million cards over a period of two years).

If the fear of the large-scale entry of the new partners were to prove unfounded, these new mechanisms would have no financial impact on the founder members. In conclusion, these new measures appear to contribute to the objectives pursued."

(Memo of 25 September 2001 (file p. 5205) "Conversation with [...]", report by [...] Société Générale dated 19 October 2001 (file p. 5735); CNCEP internal memo of 18 July 2002 "Electronic Banking Steering Committee - 'CS 2002' project. Meeting of 19 July 2002". Document dated 18 July 2002 (file pp. 7074 and 13384)).

324 The wish to influence the price of cards is clear from other documents obtained during the inspections.

For example:

(a) The intention to influence the level of bank card prices is manifest from the general overview found on the premises of Caisses d'Épargne:

"Assuming the case of a large-scale issuing of cards, the cost per card would be of the order of €10 for the first three years.

If this has a similar impact on the price reference, the gain for the banking profession is substantial (taking as a basis 45 million cards).

e.g.: The banking arm of a large retailer:

If it wishes to convert 2 000 000 proprietary cards into CB cards, then it needs:

- approximately 900 machines if it is not to pay the CSR;

- approximately 1 800 machines if it is not to pay the Merfa. If the qualifying criteria are strict, the investment might be €50 000 per machine, or €90 million in total,

- at the outset, it can therefore be considered that the bank will not pay any CSR but that it will pay a partial Merfa of the order of 40% of the maximum, resulting in an additional cost per card of around €4, which is important for the price reference for cardholder fees".

(b) BNP Paribas notes in a memo that the Merfa and the entry fee would have effects on card price setting by pure issuers and new entrants: "An initial simulation gives, for a 'pure' issuer, an average annual levy per card issued of €10 (on the basis of 50 transactions per card per year). That would represent 20% of the margin on a credit card (annual cardholder fee at €15; margin calculated after risk cost), and would have a favourable influence on the setting of annual cardholder fees for such cards. "As regards entry and dormant members wake-up fees, the same memo also states that these fees will increase the cost of card issuance for new entrants by about €5, which, coming on top of the additional cost due to the Merfa, will result in an additional cost of €15 per card during the first year, which the author finds very significant.

(c) According to Natexis Banques Populaires, the Merfa helps reduce the fall in card prices: (Memo on "CS 2002 project. General overview". Version of 25 September 2002 (file p. 13302). Internal BNP Paribas memo, Subject: "Proposed reform of the Cartes Bancaires system". 21 June 2002; (file pp. 11915 and 11916). Natexis Banques Populaires internal memo dated 14 October 2002 on CS 2002. Memo containing an annex drawn up by the Groupement on the Merfa's impact on the main members' income (file p. 5084).

325 See section 2.3 of the Groupement's reply of 19 October 2006 to the Commission's Statement of Objections of 18 July 2006, file pp. 26623 onwards.

326 See the next footnote.

327 - In the case Groupement des cartes bancaires "CB" et Europay International SA v. Commission (Judgment of 23 February 1994 in joined cases T-39-92 and T-40-92 ; [1994] ECR p.II-00049), the Court of First Instance took the view that " [...] In this case membership of the association [the Groupement] entails, by virtue of the document constituting the association, the adhesion of its members to the decisions adopted by the managing bodies of the Groupement", without requiring it to be shown that each of the members specifically and unanimously adhered to the decisions themselves;

- The same is true in numerous other judgments of the Courts, where they took the view that the adoption of decisions (or of agreements) by the competent bodies of a professional organization " in the [alleged] interest and on behalf of " its members was sufficient to characterize these decisions as decisions of associations of undertakings in the sense of Article 81(1) EC, without requiring it to be shown that there was any unanimity on the part of the members in favor of the decisions themselves (see in particular the " Wouters " judgment of the Court on 19 February 2002, case C-309-99, recital 64 and the judgment of the Court of First Instance in the " Viande bovine " case on 13 December 2006, joined cases T-217-03 and T-245-03, recital 50);

- In its judgment in the case of " Verband der Sachversicherer e.V. against Commission ", even the nonbinding recommendation of a professional organization of insurance undertakings was considered as a decision of an association of undertakings in the sense of Article 81(1)EC " regardless of what its precise legal status may be" after the Court had noted that " the statutes of the association state that it is empowered to coordinate the activities of its members " (judgment of 27 January 1987, case 45-85, ECR p. 405, recitals 30 to 32. See similar conclusions in the judgments of the Court in case 71-74 " Frubo " of 15 May 1975, recitals 29 to 31, ECR p.563; and recitals 96 to 110-82 of the judgment in the case " NV IAZ International Belgium vs. Commission " of 8 November 1983, recital 20, [1983] ECR p.3369).

328 See paragraphs 94 and 95 of the Groupement's reply of 19 October 2006 to the Commission's Statement of Objections of 18 July 2006, file pp. 26623 onwards.

329 According to Article 2 of Regulation No 1-2003 entitled " Burden of proof " " In any national or Community proceedings for the application of Articles 81 and 82 of the Treaty, the burden of proving an infringement of Article 81(1) or of Article 82 of the Treaty shall rest on the party or the authority alleging the infringement. The undertaking or association of undertakings claiming the benefit of Article 81(3) of the Treaty shall bear the burden of proving that the conditions of that paragraph are fulfilled " and, according to paragraph 5 of the preamble to the same Regulation : " It should be for the undertaking or association of undertakings invoking the benefit of a defence against a finding of an infringement to demonstrate to the required legal standard that the conditions for applying such defence are satisfied. ". Case law is consistent on this point :

" In any event, it should not be overlooked that whenever an exemption under Article 85(3) of the Treaty is sought, it is incumbent on the applicant undertaking to prove that it satisfies each of the four conditions laid down therein " ; " where an exemption is being applied for under Article 85 (3) [now 81 (3)] it is in the first place for the undertakings concerned to present to the Commission the evidence intended to establish the economic justification for an exemption " (See joined Cases 43-82 and 63-82 VBVB and VBBB v. Commission, recital 52, [1984] ECR p.19 ; Case 42-84 Remia v. Commission , recital 45, [1985] ECR p.2545 ; Case T-66-89 Publishers Association v. Commission, recitals 69 and 74, [1992] ECR p.II-1995 ; Case T-17-93 Matra Hachette v. Commission, recital 104, [1994] ECR p.II-595 ; Case T- 34-92 Fiatagri and New Holland Ford v. Commission, recital 99, [1994] ECR p.II-905 ; and Case T-35-92 John Deere v. Commission, recital 105, [1994] ECR p.II-957 ; Case T-29-92 SPO v. Commission, recital 262, [1995] ECR p.II-289 ; joined Cases T-213-95 and T-18-96 SCK and FNK v. Commission, recital 206, [1997] ECR p.II-1739 ; etc.).

330 See sections 11.1.1. to 11.1.4 below.

331 Joined Cases 56 and 58-64 Consten and Grundig v Commission [1966] ECR 299; see also Case C-277-87 Sandoz prodotti farmaceutici v Commission [1990] ECR I-45 and Case 219-95 P Ferriere Nord v Commission [1997] ECR I-4411, recitals 14 and 15.

332 Joined Cases T-374-94, T-375-94, T-384-94 and T-388-94 European Night Services and Others v Commission [1998] ECR II-3141, recital 137.

333 Thus, in case C-7-95 John Deere v. Commission ([1998] ECR p. I-3111), cited above, recital 78, the Court of Justice held that "the fact that the Commission was unable to establish the existence of an actual anti-competitive effect had no bearing on the outcome of the case".

334 Commission Decision 92-157-EEC of 17 February 1992 relating to a proceeding pursuant to Article 85 of the EEC Treaty (IV/31.370 and 31.446 - UK Agricultural Registration Exchange), OJ L 68, 13 March 1992, recital 51. The decision concerns an agreement notified to the Commission on 4 January 1988 the operation of which, following the dispatch of a statement of objections on 11 November 1988, was suspended on 24 November 1988. On 12 March 1990, five members of the Agricultural Engineers Association notified to the Commission a new agreement, undertaking not to implement the new system before they received the Commission's response to their notification.

335 Judgment in John Deere, cited above, recital 113.

336 Paragraph 29 of the notification, p. 8 (file p. 9).

337 See the Groupement's reply of 24 March 2003 to the Commission's request for information of 3 March 2003, answer to question 12 ("Section 1(iii): effects of the Merfa") (file p. 1244).

338 See recitals 71(c), 87, 120, 121 and 126 above.

339 Paragraph 47 of the notification, p. 11 (file p.12).

340 See recital 160 above. The measures relating to matters other than charging, namely the new formula for calculating voting rights and rights to the Groupement's assets remained in force until 11 March 2005 (see recital 161).

341 See the notification of the Groupement, paragraph 29 (file, p. 9) and the table (set out in footnote 406 below) sent by the Groupement in Annex 5 to its reply of 24 March 2003, which show that the main members receive nearly all the amounts paid under the Merfa (file p. 1321).

342 According to the Groupement, [more than 65% of the] institutions out of the [total number of institutions] subject to the Merfa in 2003 paid the maximum amount of €11 (file p. 20800).

343 Le droit complémentaire est calculé sur la base du nombre de cartes que le nouvel entrant " aurait dû émettre au cours de la 1° période pour que le stock en fin de 2° période ne dépasse pas 3 fois le stock en fin de période initiale ", soit le tiers du nombre de cartes émises " en excès " durant la 2ième période. Voir paragraphe 145 ci-dessus et Annexe 2 de la notification du Groupement, cote 49.

344 Bank Average cardholder fee over the period 2000-2005 for a standard immediate debit international card

Banque Accord €25

AXA Banque €26

Banque AGF €30.1

GE Money Bank €23

Covefi €22.62

S2P €25

Cofidis €24

Barclays €41.65

ING €39.37

Average €28.53

Source: Reply to the Commission's request for information of 22 July 2005 (file pp. 25768, 24934, 24882, 24740, 24735, 24674, 24944, 24689 and 24980).

345 Average price of around €33.7 for international immediate debit cards.

Bank Average cardholder fee over the period 2000-2005 for a standard immediate debit international card

Natexis Banques Populaires €37.5

Société Générale €31.29

BNP Paribas €33.3

La Poste €30.83

Crédit Agricole €41.26

Crédit Lyonnais €33.19

Caisses d'Épargne €30.66

CIC €31.53

Average €33.7

Source: Reply to the Commission's request for information of 26 July 2005 (file pp. 24669, 24840, 24953, 24648, 24711, 25804, 24888, 24890 and 24715).

346 €28.5 + €23 in additional costs.

347 €28.5 + €11 by way of the Merfa.

348 <table>

* if the additional membership fee is due. This fee being of 12 € per card for a third of all cards issued in excess ", it is of an average of € 4 (12/3) par card issued during the second period (4th to 6th year from membership).

349 See paragraph 272 of the Groupement's remarks of 19 October 2006 in reply to the Statement of Objections of 18 July 2006, referring to the list of "CB members liable for the Merfa in 2003" annexed to the Groupement's letter of 6 April 2004 (file p. 4042) and page 66 of the Groupement's presentation to the hearing of 13 November 2006.

350 See footnote 342 above.

351 Banque Accord (Auchan), Banque Chabrières (Intermarché), Banque AGF, Axa Crédit SA and Axa Banque (Banque Directe), Egg, Capital One, GE Capital Banque, Covefi (Monabanque), Cofidis, la Société de Consommation de la CAMIF C2C and Barclays.

352 E.g. Banque Accord, Axa Banque and Egg (see recitals 136 and 280 above).

353 Pass S2P (Carrefour), Banque Casino, Edel (the bank of the Leclerc group) and Réveillon (the Cora group's bank).

354 See recitals 320(c) and (g) and 300(e).

355 See recital 300(b) and (e) above.

356 See recitals 63, 100 and 136 above.

357 It emerged, for instance, during the preparation of the measures that the Leclerc group's Edel bank was uninterested in cards and settled for a system discouraging hasty initiatives (see recital 280 below).

358 Based on the 2000-2005 average fee. See recital 342(b) above.

359 List of "CB members liable for the Merfa in 2003" referred to in footnote 349 above (file p. 4042).

360 Memo of 13 November 2002 for the attention of the members of the Board of Directors (file p 10974).

361 Study by [the Groupement's consultant] dated 26 October 2004, page 23 (file p. 20977).

362 Study by [the Groupement's consultant] dated 26 October 2004, page 30 (file p. 20984).

363 See S2P's reply of 26 November 2003 to the Commission's request for information of 27 October 2003: "geographical coverage [by ATMs] is already assured and the best locations are already taken ... We do not therefore intend to install any new ATMs or to take over those closed by the traditional banks at sites which permit neither their profitability nor their security" (file pp. 25636 and 25637). To quote Banque Accord: "any additional ATMs we might install would not be profitable and what is more would reduce the profitability of those already in place" (reply of 30 August 20.05 to the Commission's request for information of 29 June 2005) and "Developing unprofitable ATMs would mean increasing considerably our operating costs and ultimately our customer charges" (reply of 10 November 2003 to the Commission's request for information of 28 October 2003 (file p. 3158)). According to the reply from Groupama Banque of 11 December 2003 to the Commission's request for information of 24 November 2003: "in view of the rural nature of Groupama's branches, its ATMs cannot be installed in busy locations, which results in loss-making operation" (file p. 3558).

364 GE Money Bank stated that: "Opting for an alternative that makes sense in business terms, like issuing more cards, for example, is immediately penalised by an increase in the Merfa. This increases the costs associated with the products offered, affects the bank's profitability and can lead to an increase in the cost of the card service. ... We would like to make it clear that the suspension ... leaves the firms liable to the Merfa in a position of some legal uncertainty. The suspension does not mean that the sums will never be due under any circumstances. ... the banks should setg aside reserves for the payment of the Merfa. That is what GE Money Bank has done. It goes without saying that the reserves tie up funds which could otherwise be invested in growing the business." Submission by GE Money Bank at the hearing of 13 November 2006

According to Groupama Banque: "the new rules on the membership fee ... and the so-called Merfa mechanism seriously add to Groupama Banque's forecast operating costs ... thus, if the projection is limited to the beginning of 2007 (but the period of application of the Merfa is unlimited), we estimate the total loss at €23 million. This figure ... therefore cancels out the profitability initially forecast" (file p. 3558).

Cofidis: "The new charges introduced by the Merfa are liable to jeopardise the supply of a CB card by Cofidis. This is because the cost of issuing CB cards would become prohibitive, and if customers were to be charged for all the costs borne by Cofidis there would be an increase in the cost of the card to the customer, who would switch from the Cofidis card to the cards offered by our competitors not subject to the Merfa" (file p. 6732).

According to Egg:

"the implications of these new charges are considerable.

These new charges increase the bank's operating costs either directly if they are not passed on to customers, or indirectly if they are.

... if these charges were to become effective, Egg France would be obliged to pass them on in whole or in part to its customers. ... Egg will have to increase the fee for the Egg Card and/or the cost of credit" (file p. 3429).

(See also the statements by Covefi (file p. 21692) and Axa Banque (file p. 24393).

365 Banque Accord took the view that the new charges would prevent new players from making attractive offers to customers and that they would oblige it to:

"- abolish or charge for services which are at present free of charge with Banque Accord, but charged for at traditional French banks: for example, withdrawals in Europe

- increase the price of the card. At present our charge is €25 (as against an average of €40 in France) and €15 for the cardholder's spouse. Six months free of charge are offered to our customers

- reduce, or possibly abolish, our fidelity premium paid on each payment transaction (0.10)".

"Our banking offer, which Banque Accord wants to be at a discount price, innovative and different from that of the traditional banks, would become standard. The main members of GIE Cartes Bancaires would thus achieve one of the objectives sought by introducing the Merfa" (file pp. 618 and 1006; see also file pp. 24926 and 21946, according to which Banque Accord also considered reducing or abolishing the cash back offered on its Visa cards).

Capital One likewise envisaged the possibility of withdrawing its offer of revolving credit on its cards, a move which would reduce the choice of services offered to French cardholders (file p. 1176).

366 According to BEUC and UFC-Que Choisir: "The new charging system will make it particularly difficult for new entrants to maintain their current offers, which for the most part oscillate around the new 'production cost' imposed by the GCB. This 'production cost' is far higher than the actual technical cost" (file p. 21612). These consumer bodies stress "the negative impact of the new charging system of Groupement des Cartes Bancaires on French consumers, including the risk of foreclosure for new entrants or at least of a restriction of competition and of the choice offered to consumers, in a market characterised by a generalised upward trend in bank card fees and the spread of foreign withdrawals" (file p. 21616). They also take the view that "the ousting of new entrants from the market is probable in the light of the drastic reduction in their margins" (file p. 21613).

367 See recital 147.

368 Article 3 of Decree No 73-314 of 14 March 1973 creating a national identification system and a register of enterprises and their establishments (Journal Officiel de la République française, 21 March 1973).

369 See the Groupement's reply of 16 March 2005 to the Commission's request for information of 11 February 2005 (file pp. 23686 and 23709).

370 Memo "CS 2002/CB Steering Committee of 11 October 2002" Société Générale internal memo of 9 October 2002 (file pp. 5558, 5565, 12383 and 12676).

371 "Proposed reform of the Cartes Bancaires system, 21 June 2002 (file p. 11916).

372 S2P's reply of 2 April 2003 to the Commission's request for information of 27 February 2003, question 1578 (file pp. 1578 and 6733).

373 Banque Accord's reply of 20 March 2003 to the Commission's request for information of 27 February 2003 (file p. 1004).

374 Banque Accord's reply of 20 March 2003 to the Commission's request for information of 27 February 2003 (file p. 1004).

375 See recital 287 above.

376 Summary of the CS 2002 meeting of 15 July 2002 at the GIE CB, 15 July 2002 (file p. 10598).

377 "CS 2002. General overview. Summary". Author and addressee unknown. Version of 25 September 2002. CNCEP (file p. 13302).

378 Memo on the preparatory dossier for the CEOs meeting of 11 October 2002 (file p. 5077).

379 Internal Caisses d'Épargne memo dated 9 October 2002 on the CS 2002 project. Meeting on 11 October 2002 (file p. 7097).

380 See the Groupement's reply of 23 February 2005 to the Commission's request for information of 11 February 2005 (file p. 23081).

381 See the Groupement's reply of 23 February 2005 to the Commission's request for information of 11 February 2005 (file p. 23077).

382 Cofidis's reply of 18 March 2003 to the Commission's request for information of 26 February 2003 (file p. 591).

383 €15 if, at the end of the sixth year, their stock is more than three times the number of cards they had at the end of the third year, otherwise €11.

384 (a) Banque Covefi stated in March 2003 that "we will have to recover the costs from our customers, via the price of services or restrictions on use".

This passing on of the costs to cardholders was again mentioned in November 2003:

"The cost of the Merfa will have to be passed on by increasing cardholder's fee and the associated services" (file p. 3214);

and, in December 2004, after the measures were suspended:

"The setting up of the Merfa and related measures leads to an appreciable increase in our costs since the impact is of the order of 10% of our current net result and will grow as more and more accounts are opened. These additional costs therefore significantly limit our room for manœuvre when it comes to continuing to offer attractive bank card fees to our customers.

This increase in costs also has the pernicious drawback of being handed over to the largest institutions with which we compete on the private market.

Since our activities are centred exclusively on private individuals, we cannot find compensation in the acquiring function and therefore the only way of recouping these additional costs is the fee we charge our customers" (file p. 21692).

Covefi also explained in 2005 that, if the suspension of the Merfa were lifted, it would pass on the resulting costs to cardholders: This would have the effect of increasing the price of its cards: "At the moment we do not have any estimates concerning the issuance of CB bank cards in the event of the Merfa arrangement being implemented as it stands.

However, the cost of the Merfa is certain to be passed on to the cardholder, resulting in an approximate annual cost of €10 per cardholder per year.

The fee payable by the cardholder would grow closer to or even catch up with the cost of the bank cards of the traditional banks that have adopted the Merfa within the GIE Cartes Bancaires" (file p. 24921).

(b) Increasing the price of cards was one of the options envisaged by Banque Accord in March 2003 following the adoption of the Merfa:

"We will change fewer proprietary cards to CB cards, since we will have to:

- abolish or charge for services which are at present free of charge with Banque Accord, but charged for at traditional French banks: for example, withdrawals in Europe

- increase the price of the card. At present our charge is €25 (as against an average of €40 in France) and €15 for the cardholder's spouse. Six months free of charge are offered to our customers

- reduce, or possibly abolish, our fidelity premium paid on each payment transaction (0.10)" (file p. 1006).

(c) Cofidis, too, stated in March 2003 that the Merfa would have an impact on the price of its cards:

"At all events, the Merfa arrangement will have repercussions for Cofidis's CB activity by compromising its profitability, but also for the cost of the services offered to customers by preventing CB cards from being offered free and by requiring a charge to be imposed for all the services offered with the CB card"

(file p. 6632).

(d) GE Capital Bank declared in March 2003 that the new charges risked being passed on to the final consumer via an increase in the price of the bank card or other services linked to the card: "These cost increases risk having an impact on the final consumer too. Since GE Capital Bank cannot market its cards at a loss, it might have to increase the amount of the annual fees charged to customers" (file p. 6329).

It confirmed in November 2003 that, following the adoption of the Merfa, it was considering increasing the price of its bank cards or of the services linked to them:

"GE Capital Bank is considering introducing several measures to enable it to ensure the profitability linked to the issuance of its CB cards

- The first year would no longer be free of charge to cardholders, as is currently the case for GE Capital Bank CB cardholders.

- An increase in the costs linked to ATM withdrawals.

- Withdrawing one of the services included in its CB cards" (file p. 6332).

(e) Egg - which has now withdrawn from the French market and sold its business to Banque Accord - stated that, if the new charges were to become effective, it would be obliged to pass them on, in whole or in part, to its customers and hence to increase the price of cards and/or the cost of credit:

"In fact, if these charges were to become effective, Egg France would be obliged to pass them on in whole or in part to its customers.

Reducing the quality of services is of course out of the question, but probably Egg will have to increase the fee for the Egg Card and/or the cost of credit" (file p. 3429).

(f) The Carrefour group's S2P bank, which chose to cut back its card issuance plans to escape the Merfa, saw the increase in the price of cards as one effect of the new costs:

"If we were liable for them, the new costs would seriously undermine the economics and business model of this activity, especially if we achieve significant commercial success. They would inevitably result in an increase in the costs of the service and hence in the price of the cards offered to customers, and beyond a given threshold the costs would make commercial supply uncompetitive compared with existing banks and would thus prevent the launch of any activity of this nature by a new actor" (file p. 25703). "If these costs were implemented and if our bank were obliged to pay them, our only solution, so as not to suffer any operating loss by our bank in this activity, would be to pass them on to our customers through their annual fee" (file p. 3362).

385 See recital 146, second indent.

386 See recital 150 above.

387 Reproduced below, by way of illustration, are a few statements by new entrants concerning the inhibiting effect on issuing plans which the measures would have if they were implemented. For instance:

(a) Axa Banque has stated that, if the measures at issue were applicable, it would have to "reconsider its bank card issuing strategy. The level of charges linked to the Merfa could not fail to strongly penalise the economic equilibrium surrounding the marketing of bank cards" (file p. 23393).

(b) S2P, which set its sights on issuing one million cards (file p. 25635), stated that if the suspension of the Merfa was lifted, it would limit its large-scale card issuance programme in order not to be subject to it: "If the Merfa were to be maintained, we would limit our programme to 500 000 cards, a figure which, it should be said, constitutes our untaxed ceiling" (file 25636).

(c) In the opinion of Banque Accord, the application of the Merfa would have a major impact on its CB bank card issuing plan: "The new charging system will induce us to modify, or even abolish, the current offering, which was economically viable before the decision was announced" (file p. 1007).

d) According to Cofidis:

"The new charges introduced by the Merfa are liable to jeopardise the supply of a CB card by Cofidis" (file p. 6632).

" ... [I]t is probable that these new charges would have dissuaded Cofidis from becoming a member of the Groupement. These charges represent a substantial financial burden which undermines the profitability of our CB card issuing programme" (file p. 6629);

"[A]t the end of 2003, our plan was to issue 20 550 CB cards. In mid-October, the number of CB cards issued was 7 000. We estimate that the number will be 8 000 between now and the end of this year 2003" (file p. 3199).

e) Banque Casino also states that: "The Merfa makes it, moreover, economically extremely difficult on the face of it to transform existing proprietary cards into bank cards ... The additional costs linked to the Merfa will unquestionably induce the Casino group bank to reconsider its CB plans. In particular, any programme for the issuing of substantial volumes of CB cards henceforth seems to us to be economically difficult to implement" (file p. 2288).

388 BNP Paribas internal email of 5 November 2002 (file p. 6363).

389 Internal Caisses d'Épargne e-mail on the record of the GIE CB meeting of 7 January 2003 (file p. 7083).

390 List of "CB members liable for the Merfa in 2003" sent to the Commission by the Groupement on 6 April 2004, file pp. 4042 and 7704.

391 Internal Société Générale memo, "CS2002/CB Steering Committee meeting on 11 October 2002", dated 9 October 2002, file p. 5562.

392 Internal CNCEP memo, "Electronic Banking Steering Committee - 'CS 2002' project. Meeting on 19 July 2002". Dated 18 July 2002. File pp. 7074 and 13384.

393 "CS 2002 project. Characteristics of the simulation", 18 March 2002. File pp. 13700 and 13706.

394 Memo on the preparatory dossier for the CEOs meeting of 11 October 2002. File pp. 5077 and 11483.

395 Comments by BEUC/UFC-Que Choisir, 25 November 2004, p. 8, file p. 21191.

396 See recital 170 of this Decision, and footnote 238.

397 For the restrictions on Visa and MasterCard cards issued outside France by a bank that is not a member of the Groupement, see recital 22.

398 Letter from Citibank to the Commission, 23 May 2003, file p. 2071.

399 Visa's submission of 2 May 2003, file p. 2031.

400 Paragraph 47 of the notification, file p. 12.

401 Ibid.

402 For example: Banque Accord, Covefi, GE Capital Bank, Cofidis, Banque Directe, Banque Bipop, Banque du Groupe Casino, Capital One. Reply to the Commission's request for information of 26/27 February 2003, file pp. 1006 (Accord), 568 (Covefi), 6321 (GE Capital Bank), 6630 (Cofidis), 1207 (Banque Directe), 1194 (Banque Bipop), and 2288 (Banque du Groupe Casino). The Merfa is calculated at the end of each year and to that end CB members must declare their data on active cards, active SIRENs and active approved ATMs as at 31 December of each year.

403 List of "CB members liable for the Merfa in 2003" sent to the Commission by the Groupement on 6 April 2004, file pp. 4042 and 7704.

404 See the Groupement's reply of 14 September 2004 to the Commission's request for information of 30 July 2004. File p. 15709.

405 List of "CB members liable for the Merfa in 2003" sent to the Commission by the Groupement on 6 April 2004, file pp. 4042 and 7704.

406 According to the Groupement's estimates the main members would benefit most from the Merfa: they would receive over [vertually all] % of the revenue. Annexe 5a to the Groupement's reply of 24 March 2003 to the Commission's request for information of 3 March 2003: "Groups of banks which would not have been subject to the Merfa if it had been applicable in 2002 and amounts they would have received", file p. 1321:

Code Name Euros

[...] [name of a main member bank of the Groupement] group 68 008

[...] [name of a main member bank of the Groupement] group 77 555

[...] [name of a main member bank of the Groupement] group 93 048

[...] [name of a main member bank of the Groupement] group 264 417

[...] [name of a main member bank of the Groupement] group 57 301

[...] [name of a main member bank of the Groupement] group 22 071

[...] [name of a main member bank of the Groupement] group 70.338

[...] [name of a main member bank of the Groupement] group 24 473

[...] [name of a main member bank of the Groupement] group 132 962

[...] [name of a main member bank of the Groupement] group 200 937

[...] [name of a main member bank of the Groupement] group 111 522

The total for all of the banking groups that would not have been liable for the Merfa if it had applied in 2002 and who would have received revenue from it is €1 051 259, including €1 022 632 for the eleven main members (€1 000 561 if we exclude CCF).

407 Banque Accord's reply to the Commission's request for information of 29 June 2005, file p. 24925.

408 Banque Accord's reply of 20 March 2003 to the Commission's request for information of 26 February 2003, file pp. 618 and 1006.

409 Banque Accord's reply of 10 November 2003 to the Commission's request for information of 28 October 2003, question 2, p. 5, file p. 3157.

410 Ibid.

411 GE Capital Bank's reply of 24 November 2003 to the Commission's request for information of 28 October 2003, file p. 6332.

412 GE Money Bank's reply of 21 July 2005 to the Commission's request for information of 29 June 2005, question A2, p. 2, file p. 24475.

413 "GE Capital Bank considers that the Merfa undermines its profitability and restricts its capacity to issue CB cards to its customers". GE Capital Bank's reply of 20 March 2003 to the request for information of 26 February 2003, file p. 6328. See also the reply of 12 December 2003 to the request for information of 27 October 2003, file p. 3645.

414 GE Money Bank's reply of 21 July 2005 to the Commission request for information of 2 June 2005, file p. 24475.

415 S2P's reply of 26 July 2005 to the Commission's request for information of 29 June 2005, file p. 25635.

416 S2P's reply of 26 July 2005 to the Commission's request for information of 29 June 2005, file p. 25635.

417 Cofidis's reply of 20 March 2003 to the request for information of 27 February 2003, file p. 6632.

418 Cofidis's reply of 12 November 2003 to the request for information of 27 October 2003, file p. 6684.

419 Cofidis had planned to issue 20 550 cards in 2003, but it issued only 7 094. In 2004 Cofidis had planned to issue 45 125 cards, but it issued only 5 936. Cofidis's reply of 14 September 2005 to the Commission's request for information of 29 June 2005, questions A1 and A2, file p. 25722.

420 Banque Casino's reply of 6 October 2003 to the Commission's request for information of 26 February 2003, file p. 2288.

421 Banque Casino's reply of 6 October 2003 to the Commission's request for information of 26 February 2003, file p. 2288.

422 The Groupement's remarks of 19 October 2006 in reply to the Statement of Objections of 18 July 2006, paragraph 300, file p 26662.

423 Paragraph 292 of the remarks just referred to, file p. 26660.

424 Paragraphs 292 and 298 of the remarks just referred to., file p. 26660 and 26661.

425 Paragraph 296 of the remarks just referred to, file p. 26661.

426 Paragraph 295 of the remarks just referred to, file p. 26660.

427 Paragraph 297 of the remarks just referred to, file 26661.

428 Recital 157 above.

429 List of "CB members liable for the Merfa in 2003" attached to a letter from the Groupement dated 6 April 2004, file p. 4042. The nine banks are Banque AGF, AXA Banque, Banque Accord, Capital One, Covefi, Barclays Bank, Egg, Finaref and GE Capital.

430 Such as the Carrefour group's bank S2P, and Banque Casino; see recitals 63 and 101.

431 Replies given by Banque Accord, GE Money Bank, Cofidis and Casino to the requests for information of 26 February 2003; replies given by Banque Accord, GE Money Bank, Cofidis and Egg to the requests for information of 28 October 2003; and replies given by Banque Accord, GE Money Bank, S2P, Capital One, Cofidis and Axa Banque to the request for information of 29 June 2005 (see Section 10.2.2.1.C, recital 300; Section 10.2.2.2.A(1), recital 320; and Section 10.2.2.2.B(1), recitals 346 and 347).

432 Notably Banque Accord, GE Money Bank, Cofidis, Casino, Egg, Capital One, Axa Banque and S2P.

433 Barclays' reply of 11 July 2005 to the Commission's request for information of 29 June 2005, file pp. 24385 and 24384.

434 In its profit-and-loss account Bipop described the bank card service as "isolated" (anecdotique). It issued only 1 000 cards, to the rare customers "who expressly asked for it" (out of 12 000 active customers' accounts; reply of 21 October 2005 given by Milaris on behalf of the former Bipop, to which Milaris had succeeded, in answer to the Commission requests for information of 29 June and 22 July 2005, file p. 25015).

435 "Offering means of payment is an indispensable necessity for us, and the need will grow as the number of our customers grows. But we will have to recover the costs from our customers, via the price of services or restrictions on use": answer given by Covefi on 19 March 2003 to the Commission's request for information of 26 February 2003, file p. 25679. "The card is an essential part of the service offered by Covefi, and the results of passing on the cost to cardholders have not been evaluated so far ... The cost of the Merfa will have to be passed on by increasing the cardholder's fee and the associated services": answer given on 19 November 2003 to the Commission's request for information of 27 October 2003, file p. 3218.

436 Covefi's reply of 19 July 2005 to the Commission's request for information of 29 June 2005, file pp. 24461 and 24462.

437 AGF's reply of 13 October 2005 to the Commission's request for information of 29 June 2005, file pp. 25009 and 25010.

438 Recital 346 above.

439 That is to say that in 2003 and 2004 Banque Accord issued only 48% of the cards it had planned to issue in each of the two years.

440 Banque Accord's reply to the Commission's request for information of 29 June 2005, question A2, file p. 24925.

441 Banque Accord's reply of 10 November 2003 to the Commission's request for information of 28 October 2003, file p. 3157.

442 See in particular the quotations from the Groupement, BNP Paribas, Caisses d'Épargne and Société Générale in recitals 92, 101(b), 110(a) and 126 (b).

443 On 15 December 2001 the charges applied by BNP, Société Générale and Crédit Lyonnais comprised an annual cardholder fee of €40-41 for a Visa card. [Groupement's consultant] document: "The CB system founder banks must now face up to a threefold challenge". File p. 14215.

444 Charges at 15 December 2001. [Groupement's consultant] document, "The CB system founder banks must now face up to a threefold challenge". File p. 14216.

445 Banque Accord's reply of 20 March 2003 to the request for information of 27 February 2003, file p. 994.

446 "CS 2002 project: Large retailers' CB bank card-issuing strategy", 22 May 2002. File p. 11936.

447 Natexis Banques Populaires, internal communication, 14 October 2002, CS 2002, Annex 1: "Impact of the Merfa on the main members' incomes. Assumptions". File p. 5082. Other documents give similar figures. A presentation dated 18 December 2001 refers to "a fee reduced to €15 per card" ([Groupement's consultant] document, "The CB system founder banks must now face up to a threefold challenge", file p. 14220; see also file p. 8126. Similarly, BNP Paribas, internal memo on the proposed reform of the Cartes Bancaires system, 21 June 2002, file p. 11915.

448 "CS 2002 project. Situation report for CEO. Working document. Confidential and personal". File p. 13818.

449 Axa Banque's reply of 11 October 2005 to the Commission's request for information of 22 July 2005, file p. 24934.

450 Banque AGF's reply of 9 September 2005 to the Commission request for information of 22 July 2005, file p. 24882.

451 Banque Covefi's reply of 9 September 2005 to the Commission request for information of 22 July 2005, file p. 24882.

452 GE Money Bank's reply of 9 September 2005 to the Commission's request for information of 22 July 2005, file p. 24735.

453 Banque Cofidis's reply of 15 September 2005 to the Commission's request for information of 22 July 2005, file p. 24944.

454 S2P's reply of 6 September 2005 to the Commission's request for information of 22 July 2005, file p. 24674.

455 Banque Accord's reply to the Commission's request for information of 22 July 2005, file pp. 25768 to 25771.

456 ING's reply of 5 October 2005 to the Commission's request for information of 22 July 2005, file pp. 24979 and 24980.

457 The vast majority of CB cards are international cards. According to [Groupement's consultant] (study dated March 2003, page 35, file p. 1293), 91.7% of CB cardholders have international cards. Banks issue more immediate debit cards than deferred debit cards in France: see RBR Report "Payment Cards in Europe 2004", chapter on France, file pp. 25729 and 25723 to 25725.

458 Reply by Crédit Agricole SA dated 8 September 2005 and by FNCA dated 9 September 2005 to the Commission's request for information of 26 July 2005, file p. 24863.

459 Crédit Lyonnais's reply of 12 October 2005 to the Commission's request for information of 26 July 2005, file p. 25804.

460 Crédit Agricole's reply of 21 September 2005 to the Commission's request for information of 26 July 2005, file p. 24927.

461 Crédit Agricole's reply of 21 September 2005 to the Commission's request for information of 26 July 2005, file p. 24975.

462 Société Générale's reply of 9 September 2005 to the Commission request for information of 26 July 2005, file p. 24840.

463 CNCEP's reply of 9 September 2005 to the Commission's request for information of 26 July 2005, file p. 24890.

464 BNP Paribas's reply of 13 September 2005 to the Commission's request for information of 26 July 2005, file p. 24953.

465 La Poste's reply of 5 September 2005 to the Commission's request for information of 26 July 2005, file p. 24648.

466 CIC's reply of 8 September 2005 to the Commission's request for information of 26 July 2005, file p. 24890.

467 Reply by Natexis Banques Populaires of 6 September 2005 to the Commission's request for information of 26 July 2005. It is not indicated whether the information supplied refers to immediate debit or deferred debit cards. File p. 24669.

468 Paragraph 286 of the Groupement's remarks of 19 October 2006 in reply to the Statement of Objections of 18 July 2006, file p. 26658.

469 Paragraph 283 of the Groupement's remarks just referred to, file p. 26657.

470 Groupement's presentation to the hearing of 13 November 2006, p. 68, file p. 27275.

471 Paragraph 288 of the Groupement's remarks already referred to, file p. 26658.

472 Title of section 3 of the Groupement's remarks already referred to, file p. 26655.

473 See in particular paragraphs 272-273 and 288 of the Groupement's remarks already referred to, file pp. 26655 and 26658.

474 Banque Chabrières, Banque Accord (paragraph 272 of the Groupement's remarks already referred to) and Axa Banque (p. 66 of the Groupement's oral statement at the hearing). Recital 276 above.

475 Paragraph 272 of the Groupement's remarks already referred to.

476 Paragraphs 148 and 274 of the Groupement's remarks already referred to.

477 Paragraphs 211, 288 and 289 of the Groupement's remarks already referred to.

478 Paragraphs 266-269 of the Groupement's remarks already referred to.

479 Paragraph 263 of the Groupement's remarks already referred to.

480 Paragraph 286 of the Groupement's remarks already referred to.

481 Paragraphs 255, 257 and 290 of the Groupement's remarks already referred to.

482 Paragraph 255 of the Groupement's remarks already referred to.

483 Paragraph 255 of the Groupement's remarks already referred to.

484 Paragraph 255 of the Groupement's remarks already referred to.

485 See in particular recitals 63, 66, 120 and 126 above.

486 See section 6.3.2.7, recitals 68 et seq.

487 This was the case of Barclays, Fortis, Caixa Bank and Banque AGF; see recitals 335 and 349 .

488 See recitals 338 and 354.

489 See recitals 71, 110(c) and 126(b).

490 That is to say a new entrant who issues cards without acquiring merchants or managing ATMs.

491 And the indications in section 10.2.2.1.A(1)(a), "The amount of the additional cost", where the table is to be found (recitals 265 to 277).

492 See above, footnote 342.

493 Such as Banque Chabrières, Banque Accord and Axa Banque. Recital 274.

494 See in particular section 2.1 of the Groupement's remarks already referred to.

495 Section 10.2.2.1.B(1) and (2) and C; section 10.2.2.2.A(1) and (2) and B.

496 Paragraph 284 of the Groupement's remarks already referred to: "the new entrants' replies to the request of 22 July 2005 show that ... Fortis Bank, CaixaBank Banque AGF and Barclays Bank increased their prices by comparison with 2002." Axa also increased its prices very considerably between 2004 and 2005: recital 351 above.

497 Recital 157 above.

498 Commission letters of 1 and 8 September 2006 replying to the Groupement's letter of 25 August 2006.

499 See footnotes 478, 479 and 480 above.

500 Paragraph 286 of the Groupement's remarks already referred to.

501 See in particular recitals 71, 79, 110(a), 126(a), 241, 242, 298 and 302(c) .

502 Cofidis's reply of 20 March 2003 to the request for information of 27 February 2003, file p. 6732. See footnote 364.

503 Recitals 300(d), 320(b) and (e), and 347.

504 See section 10.2.2.2. A(1), recital 320, and section 10.2.2.2. B(1), recitals 345 and 346.

505 See recitals 333 and 334.

506 Recital 316.

507 The sums paid by Chabrières, Axa Banque and Banque Accord still represent an extra cost of 29%, 28% and 19% respectively. Recital 276.

508 See footnote 496.

509 See recitals 319 and 346.

510 For example by cutting back or scrapping loyalty schemes, cutting back or scrapping services previously offered free of charge, reviewing marketing and communication plans, etc. (see recitals 280, 319(c) and 298(b) and footnote 364.

511 Recitals 255 and 257 of the remarks just referred to.

512 Recital 347.

513 The Groupement's reply of 14 September 2004 to the Commission's request for information of 30 July 2004, answer to question 3. File p. 15710.

514 The Groupement's reply to the Commission's request for information of 30 July 2004, file p. 15710.

515 Statement by GE Money Bank quoted in recital 347 and footnote 364.

516 Commission's requests for information of 29 June and 22 July 2005, file pp. 24296 to 24334 and 24480 to 24513.

517 Banque Accord's reply of 10 November 2003 to the Commission's request for information of 28 October 2003, file pp. 3154 and 3157.

518 GE Money Bank's reply of 21 July 2005 to the Commission's request for information of 29 June 2005, question A2, p. 2, file p. 24475. See also GE Money Bank's presentation at the hearing of 13 November 2006.

519 S2P's reply of 26 July 2005 to the Commission's request for information of 29 June 2005, file p. 25635.

520 GE Money Bank's presentation at the hearing of 13 November 2006.

521 GE Money Bank's reply of 21 July 2005 to the Commission request for information of 2 June 2005, file p. 24653.

522 Banque Accord's reply to the Commission's request for information of 29 June 2005, file p. 24925.

523 Banque Accord's reply to the Commission's request for information of 29 June 2005, file p. 24926.

524 S2P's reply of 13 July 2005 to the Commission's request for information of 29 June 2005, file p. 25636.

525 Cofidis's reply of 14 September 2005 to the Commission's request for information of 29 June 2005, file p. 24937.

526 [...], lawyer, speaking at the hearing of 13 November 2006.

527 Barclays' reply of 8 September 2005 to the Commission's request for information of 22 July 2005, file p. 24690.

528 Axa Banque's reply of 11 October 2005 to the Commission's request for information of 22 July 2005, file p. 24934.

529 Banque AGF's reply of 9 September 2005 to the Commission request for information of 22 July 2005, file p. 24882.

530 Citibank's reply of 11 October 2005 to the Commission's request for information of 22 July 2005. Citibank launched its two cards in 2004, file p. 25001.

531 GE Money Bank's reply of 9 September 2005 to the Commission's request for information of 22 July 2005, file p. 24740.

532 Banque Accord's reply to the Commission's request for information of 22 July 2005, file pp. 25768 to 25771.

533 Banque Covefi's reply of 9 September 2005 to the Commission's request for information of 22 July 2005, file p. 24735.

534 Cofidis's reply of 15 September 2005 to the Commission's request for information of 22 July 2005, file p. 24944.

535 ING's reply of 5 October 2005 to the Commission's request for information of 22 July 2005, file pp. 24979 and 24980.

536 That is to say, a price of Euro 15/20 per card ; see recitals 70 c), 87, 101, 120 and 126 above.

537 La Poste's reply of 5 September 2005 to the Commission's request for information of 22 July 2005, file p. 24648.

538 Reply by Crédit Agricole dated 8 September 2005 and by Fédération Nationale du Crédit Agricole dated 9 September 2005 to the Commission's request for information of 26 July 2005, file p. 24711.

539 Crédit Lyonnais's reply of 12 October 2005 to the Commission's request for information of 26 July 2005, file p. 25004.

540 Crédit Agricole's reply of 21 September 2005 to the Commission's request for information of 26 July 2005, file p. 24975.

541 Crédit Agricole's reply of 21 September 2005 to the Commission's request for information of 26 July 2005, file p. 24972.

542 Reply by Crédit Industriel et Commercial of 8 September 2005 to the Commission's request for information of 22 July 2005, file p. 24715.

543 BNP Paribas's reply of 9 September 2005 to the Commission's request for information of 26 July 2005, file p. 24953.

544 Reply by Natexis Banques Populaires of 6 September 2005 to the Commission's request for information of 26 July 2005, file p. 24669.

545 Société Générale's reply of 9 September 2005 to the request for information of 26 July 2005, file p. 24840.

546 CNCEP's reply of 9 September 2005 to the Commission's request for information of 22 July 2005, file p. 24890.

547 Statement by GE Money Bank quoted in recital 347 and footnote 364.

548 Statement by S2P quoted in recital 320(c).

549 Statement by Banque Accord quoted in footnotes 384 (b) and 365.

550 Recital 320, quoting Banque Accord's reply of 19 March 2003 to the Commission's request for information of 26 February 2003, file p. 3157.

551 Recital 320, citing GE Money Bank' reply of 21 July 2005 to the Commission's request for information of 29 June 2005, file p. 24475.

552 Paragraph 191 of the Groupement's remarks of 19 October 2006 replying to the Statement of Objections of 18 July 2006remarks.

553 Notification, paragraphs 42-46 and 81-88, file pp. 12 and 28-31.

554 Paragraph 82 of the notification, file p. 28.

555 Paragraphs 23, 82 and 84 of the notification, file pp. 8, 28 and 30.

556 See in particular paragraphs 94 and 252 of the Groupement's remarks of 19 October 2006 replying to the Statement of Objections of 18 July 2006.

557 Court of First Instance in Case T-112-99 Métropole and Others v Commission [2001] ECR II-2459, recital 106, and Court of Justice in Case 42-84 Remia and Others v Commission [1985] ECR 2545, recital 20. See also point 13 of the Commission Notice on restrictions directly related and necessary to concentrations (OJ C 56, 5.3.2005, p. 24).

558 Métropole, cited above, recital 109.

559 Rule adopted by the Board of Directors of the Groupement in 1995, cited in footnote 238.

560 Though transactions performed with Visa and MasterCard cards issued by a bank outside the Groupement can be processed by the latter's CB network, such transactions remain marginal and subject to conditions. Recital 22.

561 Court of First Instance in Case T-86-95 Compagnie Générale Maritime and Others v Commission [2002] ECR II-1011, recital 138.

562 Joined Cases T-25-95 and others Cimenteries CBR v. Commission [2000] ECR, p.II-491.

563 Case C-306-96 Javico [1998] ECR I-1983, recitals 16-17.

564 Rule dated 28 July 1995, amended on 22 September 2000: "Any banking institution located outside France issuing cards that are used primarily in the CB system must undertake to comply with all the CB rules by becoming a member of Groupement des Cartes Bancaires CB in the same way as any banking institution located inside France".

565 Paragraphs 47 to 49, pages 11 to 13, of the Groupement's notification of 11 December 2002, file pp. 12 to 14. According to paragraph 47 of the notification: "The notified provisions are liable to affect trade between Member States of the Community and/or of EFTA".

566 Article 2 of Council Regulation (EC) No 1-2003 of 16 December 2002 on the implementation of the rules on competition laid down in Articles 81 and 82 of the Treaty (OJ L 1, 4.1.2003, p. 1).

567 See, in particular, the judgment in Joined Cases T-213-95 and T-18-96 SCK and FNK v Commission [1997] ECR II-1739.

568 See, for example, the judgment in Compagnie générale maritime and Others v Commission, cited in footnote561, recital 349.

569 See paragraphs 91 to 100 of the Groupement's notification of 10 December 2002, pages 31 to 33 (file pp. 32-34) and section 4 of its remarks of 19 October 2006 on the first Statement of Objections of 18 July 2004.

570 The first study was supplied with the Groupement's reply of 28 March 2003 to the Commission's request for information dated 3 March 2003: [Groupement's consultant], "Externalities in the CB system": March 2003 (file pp. 13953 et seq.). The second was attached to the Groupement's reply of 8 November 2004 to the first Statement of Objections of July 2004: [Groupement's consultant], "Additional economic study on the externalities in the CB system and the effects of the mechanism for regulating the acquiring function", 26 October 2004 (file pp. 20955 et seq.). In addition, in response to the Commission's criticism that the studies were marred by methodological errors and questionable data, the Groupement provided further economic notes (by [Groupement's consultants], October and November 2006: Annexes 13, 14, 18 and 19 of the Groupement's comments of 19 October 2006 in reply to the Statement of Objections of 18 July 2006).

571 Groupement's notification of 10 December 2002, paragraphs 92 and 99, pp. 31 and 32 (file pp. 32 and 33).

572 Paragraph 25 of the Groupement's notification dated 10 December 2002, p. 7 (file p. 8).

573 See the Groupement's reply of 24 March 2003 to the Commission's request for information of 3 March 2003 (file p. 1242).

574 Paragraph 82 of the Groupement's notification dated 10 December 2002, p. 27 (file p. 28).

575 See the Groupement's reply of 24 March 2003 to the Commission's request for information of 3 March 2003 (file p. 1241).

576 Paragraph 23 of the Groupement's notification of 10 December 2002 (file p. 8).

577 Paragraph 170 of the Groupement's reply of 8 November 2004 to the Statement of Objections of July 2004.

578 Groupement's notification of 10 December 2002, paragraph 95 (file p. 32); point (ii) of paragraph 25 (file p. 8); paragraph 31 (file p. 9); paragraph 83 (file p. 29); paragraph 85 (file, p. 30).

579 It was only after the Commission had pointed out in its Statement of Objections of 18 July 2006 that the arguments in support of the two sources of free riding were used together to justify a single measure that the Groupement, in paragraphs 120 and 121 of its reply of 19 October 2006 to that Statement of Objections, began to link specific sources of free riding to specific groups of measures: on the one hand, it referred to free riding targeting "investments and the value of the system" to justify reforming the membership fee and introducing an additional membership fee and dormant members wake-up fee, on the other, to free riding targeting "the positive externalities generated by acquiring" to justify the Merfa (paragraph 120 of the Groupement's reply).

Nonetheless, the Groupement has maintained the confusion between the two sources of free riding by referring to the development of acquiring (said to generate positive externalities) and the regulatory (or balancing) function of the system to justify measures other than the Merfa, as, for example, in (a) [Groupement's consultant]'s memo on the reform of the Groupement's membership fee, III-1, p. 3, file p. 1311; (b) the Groupement's notification of 10 December 2002 (paragraph 31, file p. 9 and paragraphs 85 and 87, file p. 30).

580 Paragraph 95 of the Groupement's notification of 10 December 2002 (file p. 32).

581 Paragraphs 25 (file p. 8), 31 (file p. 9) and 85 (file p.30) of the notification.

582 Paragraphs 31(file p. 9) and 83 (file p. 29) of the notification.

583 Paragraphs 20 (file p. 8) and 87 (file p. 30) of the notification.

584 Paragraph 85 of the notification (file p. 30).

585 Groupement's reply of 8 November 2004 to the Statement of Objections of July 2004, paragraph 252 (file p. 20808).

586 Paragraph 137 of the Groupement's remarks of 19 October 2006 in reply to the Statement of Objections of 2006.

587 Paragraph 141 of the remarks cited above.

588 "Memo on the reform of the fee for membership of Groupement des Cartes Bancaires 'CB' drafted by [the consultant] for Groupement des Cartes Bancaires "CB", March 2003 (file pp. 1310 and 1313).

589 Memo cited above (file p. 1313). According to the same document (same page in file) "the abovementioned estimate puts at €4 billion the total cost of developing the CB system borne both by the GIE Cartes Bancaires 'CB' system and by the member banks directly" (emphasis added).

590 See Commission's request for information of 27 October 2003 (file p. 2335) and the replies from Crédit Mutuel (27 November 2003, file pp. 3423 et seq.), Caisse Nationale d'Épargne et de Prévoyance (27 November 2003, file pp. 3542 et seq.), CCF (1 December 2003, file p. 7662), Groupe Banques Populaires (26 November 2003, file pp. 3380 et seq.), CIC (27 November 2003, file pp. 3384 et seq.) and Crédit Lyonnais (28 November 2003, file pp. 3432 et seq.).

591 Annex XI to the Groupement's reply dated 7 November 2003, which reproduces the notification of the CIR made on 11 December 1995 (Case 35859) (file p. 3112).

592 Paragraph 141 of the Groupement's remarks of 19 October 2006 in reply to the Commission's Statement of Objections of 18 July 2006.

593 "This price difference between CB cards and the proprietary cards they replace suffices to show the minimum value derived from using the system"; paragraphs 143 to 145 of the Groupement's remarks of 19 October 2006 in reply to the Statement of Objections of 2006.

594 Paragraph 145 of the remarks cited above.

595 Paragraph 146 of the remarks cited above.

596 See paragraphs 214 and 151 of the remarks cited above.

597 See paragraphs 149 to 155 of the remarks cited above.

598 See paragraph 394 above, citing Annex XI to the Groupement's reply dated 7 November 2003, which reproduces the notification of the CIR made on 11 December 1995 (Case 35859) (file p. 3112).

599 See document dated 29 August 2001 entitled "Admission of new members: points for discussion" (file p. 14005).

600 Memo on the management of the CS 2002 project, dated 5 December 2001 (file p. 13257).

601 See memo on the management of the CS 2002 project, dated 5 December 2001 (file p. 13257), and document obtained on the premises of the Groupement originating from [consultancy firm] (file p. 14226).

602 See abovementioned documents (file pp. 13257 and 14226).

603 To the tune of €44.85 million between 1989 and 2003 in the case of Banque Accord and the Banque Accord group, see reply to the Commission's request for information of 28 October 2003 (file p. 6791).

604 See document obtained on the Groupement's premises (file p. 13822) comparing the activities performed by a number of card payment systems in Europe.

<table>

605 "Externalities in the CB network", study by [the consultant] for Groupement des Cartes Bancaires, March 2003, section II-3.b., footnote 17 and page 18 (file pp. 1276 and 1277).

606 Memo cited above (file p. 1313).

607 See recital 392 above.

608 See section 5 above.

609 Paragraphs 157 and 158 of the Groupement's remarks of 19 October 2006 in reply to the Commission's Statement of Objections of 18 July 2006.

610 Paragraph 157 of the remarks cited above.

611 Paragraph 159 of the remarks cited above.

612 Paragraph 166 of the Groupement's remarks cited above.

613 Groupement's remarks in reply to the first Statement of Objections, paragraph 252 (file p. 20808).

614 Paragraph 168 of the remarks cited above.

615 Paragraph 167 of the remarks cited above.

616 Paragraph 167 of the remarks cited above.

617 Paragraph 169 of the remarks cited above.

618 See paragraphs 171 to 177 of the remarks cited above.

619 "Study of network externalities in the CB system" by [the consultant] for Groupement des Cartes Bancaires 'CB', March 2003 (file pp. 1258 to 1308).

According to this study, the conclusion that acquiring generates more positive externalities is based on the following observations:

Issuance has grown continuously, unlike merchant acquiring (measured in the number of SIREN numbers belonging to merchants affiliated to the CB system as a proportion of the total number of SIREN numbers), which has expanded little. The number of ATMs per capita is lower in France than in Japan, the United States or the euro zone, the growth in the number of CB ATMs is slowing down in France and issuing is more profitable than acquiring.

Issuing CB cards with additional functions would create value for the cardholder, but no externalities on the acquiring side. It would not increase the number of holders, but would cause payment flows to switch from proprietary cards to CB cards; these flows would be limited and would bring little additional income for acquiring banks.

The fact that cardholders agree to pay 30% more for international CB cards than for national ones shows that the subjective value attached by cardholders to the creation of new payment and withdrawal points is greater than merely the additional volume of payments and withdrawals generated by the new acceptance points.

A system like the Merfa, determined on the basis of the number of access points (ATMs and SIRENs), rather than the volume and number of transactions like the CIR and CIP, is important in order to strengthen the CB system feature that distinguishes the feature from its competitors: its "ubiquity".

A cut in the CIP would not give rise to any new affiliations, but would instead exacerbate the already keen competition for merchants already acquired who generate large volumes. In particular, [the Groupement's consultant] argues that it would not give rise to new affiliations in the categories of merchant still outside the system, carrying little attraction for potential acquiring banks on account of the low transaction volumes they generated. By taking equal account of each merchant, whatever the number and volume of transactions generated by that merchant, the Merfa would encourage the acquiring of as yet non-affiliated merchants. The same reasoning could be applied to a reduction in the CIR, which would merely give rise to congestion in the more profitable areas and not to the installation of ATMs in rural areas.

620 [The Groupement's consultant], "Additional economic study on the externalities in the CB system and the effects of the mechanism for regulating the acquiring function", 26 October 2004 (file pp. 20955 et seq.). According to this second report, additional quantitative analyses confirm the existence of significant positive externalities from acquiring to issuing, but no positive externalities generated by issuing can be empirically determined.

Any mechanism designed to take account of this phenomenon will necessarily involve a charge on the issuing activity of certain members only, as is the case with the Merfa formula. The choice of the SIREN criterion (as opposed to the SIRET criterion or criteria based on volumes) will increase the incentive to seek new merchant affiliations; likewise, the choice of the number of ATMs as opposed to volumes will provide an incentive to install ATMs at new locations. The payment and withdrawal acquiring effort needed to avoid paying the Merfa is achievable: it would suffice to gain just a part of gross flows of new SIRENs still to be affiliated and new ATMs to be installed, and banks attached to large retailers have a large "affiliation reserve" within their own businesses.

The Merfa has no negative impact on issuing, because it is low compared with the investments by issuing banks, because the real cost of the Merfa for banks paying a fee under it is far lower than the maximum amount of the Merfa, and because most of the issuing banks are not subject to it and will continue to bring competitive pressure to bear on issuing to the benefit of consumers.

621 "Memo on the reform of the fee for membership of Groupement des Cartes Bancaires 'CB' drafted by [the consultant] for Groupement des Cartes Bancaires 'CB', March 2003 (file pp. 1309-1314).

622 Paragraph 332 of the Groupement's remarks of 19 October 2006 in reply to the Commission's Statement of Objections of 18 July 2006.

623 Paragraph 20 of the Groupement's notification dated 10 December 2002, page 6 (file p. 7).

624 See the Groupement's reply of 24 March 2003 to the Commission's request for information of 3 March 2003 (file p. 1242).

625 Paragraphs 23, 82 and 84 of the notification of 10 December 2002 (file pp. 8, 28 and 30).

626 "Externalities in the CB system", study by [the consultant] for Groupement des Cartes Bancaires "CB" in March 2003, Chapter IV, "Issuance generates few positive externalities", pp. 29-33 (file pp. 13982- 13986), and Chapter V, "Payment and withdrawal acquiring generates significant positive externalities", pp. 33-39 (file pp. 13986-13992).

627 An outlier is a data point that is located far from the rest of the data of a data sample.

628 The number of cards per adult is lower in France than in most other European countries (France is in ninth place out of the EU 15 for payment cards and last but two for debit cards): see "Payment Cards Western Europe 2006" by Retail Banking Research Ltd., International Survey and Analysis, International Overview, p. 5 (file p. 25038).

629 See Annex 1 to the Statement of Objections of 2006.

630 For example: "The French market already appears to have an overcapacity of ATMs", Egg's reply of 26 March 2003 to the Commission's request for information of 26 February 2003 (file p. 6702); "The French ATM market is already showing signs of saturation", Egg's reply of 27 November 2003 to the Commission's request for information of 27 October 2003 (file p. 3430); "We do not believe it makes economic sense [to install additional ATMs] considering that the level of penetration of ATMs in France is among the highest in Europe", Capital One's reply of 10 November 2003 to the Commission's request for information of 27 October 2003 (file p. 3163); "as regards installation of ATMs, the total number of ATMs already installed in France is particularly high, to the extent that many of them are unprofitable today", Cofidis's reply of 20 March 2003 to the Commission's request for information of 26 February 2003 (file p. 6630); "The GIE CB rule therefore ends up destroying value ... including, with respect to the market itself, by generating a surplus of supply on a market which is very competitive and already almost saturated", Groupama's reply of 8 December 2003 to the Commission's request for information of 24 November 2003 (file p. 3558).

631 Memo on charging principles for the CS 2002 project, file p. 13648.

632 Internal Société Générale email, file p. 12688. France is one of the countries with the highest number of point-of-sale terminals per inhabitant, while the number of transactions per inhabitant is nearly twice the European (EU 15) average. Source: ECB Blue Book 2005, table 5, p. 12, file p. 25056.

633 CB Magazine, No 18, first quarter 2003, file pp. 22314-22316. Commission's calculation, based on data in the magazine.

634 For example:

- the model proposed by [the Groupement's consultant] does not include variables for the CIP and CIR interchange fees, although they regulate the externalities generated by issuing and acquiring activities;

- The existence of externalities on both the issuance and the acquiring sides suggests that the variables measuring these activities should be correlated. This problem of the multicollinearity of the issuance and acquiring variables does not appear in the model used by [the Groupement's consultant]. This would suggest either that there is no externality or that the model is not suitable.

- On the basis of a brief review of the data, it is possible to contest the supposed linearity of the model and therefore its specification (i.e. [the Groupement's consultant] used the estimation method known as ordinary least squares, or OLS, based on the linearity needed for a robust estimation).

- The data base used by [the Groupement's consultant] is questionable because it includes outlying values used in [the Groupement's consultant]'s calculations. If he had eliminated these outlying values, the calculations would have produced the opposite result, namely that the effects on the acquiring side were not significant.

635 First, by introducing lagged explanatory variables into [the Groupement's consultant]'s original regression and then using raw data instead of measuring them in growth rates as [the Groupement's consultant] did.

636 Comments by Visa following publication in the Official Journal of the European Union of 3 April 2003 of a notice on the notification of the measures by the Groupement (file pp. 2027 et seq.)

637 Ibid.

638 In 2002, 4 681 million payments by credit and debit cards in the UK compared with 4 096 million in France for an average transaction value of €57.3 in the UK (debit cards) and €46.4 (debit and credit cards) in France. In 2003, 5 186 million payments by credit and debit cards in the UK compared with 4 342 million in France for an average transaction value of €56.0 in the UK (debit cards) and €47.0 (debit and credit cards) in France. ECB Blue Book: Payment and securities settlement systems in the European Union, Addendum incorporating 2003 figures, August 2005, tables 11 and 6, pp. 17 and 12.

639 ECB Blue Book: Payment and securities settlement systems in the European Union, September 2003, August 2005 and March 2006 (file pp. 25050 to 25053).

640 One of the comments on the [Groupement's consultant] report prepared by Groupement [representative] states: "it would be good if the report were to specify that the work on demonstrating positive externalities began in March 2002 ... The over-frequent references to recent press cuttings suggest that the report was not written until 2003 ... But competition authorities expect economic studies to precede policy decisions (not to mention legal evidence!) ... It should have been done before the Board of Directors took a decision on the Merfa and the membership fees" (file pp. 14438 to 14439).

641 Dated 25 February 2002 (file pp. 12463 et seq.), 3 July 2002 (file pp. 13343 et seq.), 22 July 2002 (file, pp. 12049 et seq.), 2 October 2002 (file, pp. 10235 et seq.) and November 2002 (file, pp. 14433 et seq.).

642 For example, in the July 2002 study, [the Groupement's consultant] proposed that banks with more than 100 000 cardholders be exempted from the CSR (withdrawal service commission). This criterion was not approved by the Electronic Banking Steering Committee, and was subsequently removed. The same applies to the first 100 000 cards issued and covered by the fixed membership fee being free of charge: this disappears after the study of July 2002.

643 See [the Groupement's consultant]'s "Note on the Report of the CET on Groupement des Cartes Bancaires" and [the Groupement's consultant]'s "Analysis of the economic studies produced by the Commission in response to the [the Groupement's consultant] studies" (the first can be found in Annex 18 to the Groupement's comments of 2006 in reply to the Statement of Objections of that year, while [the Groupement's consultant]'s analysis can be found in Annexes 13, 14 and 19 to the same comments).

644 [The Groupement's consultant]'s note cited above, in fine, pp. 4 and 5.

645 Presentation by [the Groupement's consultant] at the hearing on 13 November 2006.

646 [The Groupement's consultant]'s analysis cited above.

647 The objections concerned (1) the lagged variables, (2) problems of multicollinearity with the lagged values chosen by the Commission and (3) the use of raw data instead of the growth rates used in [the Groupement's consultant] model. The Commission contests the soundness of the objections to its economic study and regards the analyses by [the Groupement's consultants] as biased. Even if their studies are sound, which the Commission contests, they do not correct the methodological errors contained in [the Groupement's consultant]'s study.

648 [The Groupement's consultant], "Additional economic study on the externalities in the CB system and the effects of the mechanism for regulating the acquiring function", 26 October 2004, III-4, p.19 (file p. 20973 to 20976).

649 See Annex XI to the Groupement's reply dated 7 November 2003 (file p. 3110), which reproduces the notification of the CIR made on 11 December 1995 (Case 35859).

650 See the statements of Visa and Banque Accord at recitals 442 and 462.

651 Banque Accord's presentation to the hearing of 13 November 2006.

652 See the replies by GE Money Bank, Banque Casino, Covefi, Banque Accord, Cofidis, Capital One, Citibank, Finaref, Banque AGF, Egg and S2P to the Commission's requests for information dated 26, 28 and 28 February and 3 March 2003, and 29 June 2005 (file pp. 25712, 24660, 2287, 25679, 1004, 24929, 6630, 1175, 2061, 1773, 1889, 6701 and 25638 to 25640).

653 See paragraph 25 of the notification (file p. 8) and the Groupement's reply of 24 March 2003 to the Commission's request for information of 3 March 2003 (file p. 1241), and the Groupement's website at http://www.cartes-bancaires.com/FR/telechargements/chiffresCB2004FR.pdf (file pp. 24638 to 24641).

654 As indicated above, it is doubtful whether the Merfa actually encourages banks primarily engaged in issuing to develop the activities of acquiring ATMs and merchants as:

- such an incentive to develop the management of ATMs is already provided by the CSR (see recitals 53 and 492);

- the encouraging of the development of merchant and ATM acquiring is contradicted by the CIP and CSC commissions (see section 10.1.2.1 on "The existence of a restriction of competition by object", "The function of the Merfa is in contradiction with the functions of the interchange fees and of the other measures", recitals 223 to 231); and

- the Merfa has more as its object and effect the discouragement of the issuing of cards rather than the encouragement of acquiring (see section 10 above).

655 Cofidis's reply of 20 March 2003 to the Commission's request for information of 26 February 2003 (file p. 6630).

656 See the Groupement's reply of 24 March 2003 to the Commission's request for information of 3 March 2003 (file p. 1244).

657 According to the Groupement, the Merfa "is confined to taking into account [the fact that acquiring generates more positive externalities than issuance] without imposing a strategy on any of the Groupement's members; each member is free to give priority to its issuance business provided it pays the compensation" (paragraph 84 of the notification, page 29 (file p. 30)).

658 See Annexes IX and XI to the Groupement's reply of 7 November 2003 reproducing the notification of the CIP made on 20 December 1990 and the notification of the CIR made on 11 December 1995 (Cases 32746 and 35859 respectively) (file pp. 3072 et seq., especially pp. 3076 and 3077, and 3091 et seq.).

659 UFC-Que Choisir. Study on the cost of banking services, July 2004, p. 51 (file p. 21246).

660 See recital 236 and recital 230 above.

661 Annex XI to the Groupement's reply dated 7 November 2003, which reproduces the notification of the CIR made on 11 December 1995 (Case 35859).

662 See paragraph 20, page 6 (file p. 7) of the notification of 10 December 2002.

663 The Groupement merely states that it receives the income from the membership fees (see paragraph 326 of the Groupement's reply of 19 October 2006 to the Statement of Objections of 18 July 2006).

664 Additional economic study on the externalities in the CB system and the effects of the Merfa, October 2004, III-5, page 22 (file p. 20976).

665 Submissions by BEUC and UFC-Que Choisir of 25 November 2004, pages 6-8 (file pp. 21189 to 21191).

666 GE Money Bank's presentation to the hearing of 13 November 2006.

667 Replies by Cofidis and Egg to the request for information of February 2003, file pp. 6630 and 6702; replies by Groupama Banque, Capital One and Axa Banque to the request for information of October 2003, file pp. 3557, 3167 and 6558.

668 Paragraph 347 of the Groupement's remarks of 19 October 2006 in reply to the Statement of Objections of 18 July 2006.

669 See recital 464 above.

670 See the Visa and MasterCard websites, where they publish their cross-border multilateral interchange fees (file pp. 25047 and 25048): http://www.visaeurope.com/acceptingvisa/interchange.html http://www.mastercardintl.com/corporate/mif_information.html

671 RBR 2004 report cited above, file pp. 25737 to 25742.

672 CS 2002 project working document. Steering Committee, 18 December 2001. GCB2-AW12 (file p. 13774).

673 "Study of network externalities in the CB system" by [the consultant] for Groupement des Cartes Bancaires CB, March 2003, page 18, file p. 1276.

674 Annex XI to the Groupement's letter dated 7 November 2003, which reproduces the notification of the CIR made on 11 December 1995 (Case 35859) (file p. 3094 to 3095).

675 See recital 229 above.

676 Paragraph 217 of the Groupement's reply of 19 October 2006. See also Annex 1, Chapter VI-1 of the Groupement's reply of 24 March 2003 to the Commission's request for information of 3 March 2003 (file pp. 1298 and 1299).

677 Paragraph 23 of the notification (file p. 8).

678 See Annex IX to the Groupement's reply dated 7 November 2003 (file pp. 3076 to 3077), which reproduces the notification of the CIP made on 20 December 1990 (Case 32746).

679 According to the Groupement's website (file p. 24643): "Consumer expectations ... : Where would you like to be able to use your bank card more often? (Sofres survey, September 2004, 931 holders of CB cards) 67% mentioned the health professions, 43% public services, 38% home deliveries, 34% taxis, 29% automatic cash machines, 24% markets. ... The results: The card is increasingly accepted compared with 2003: at doctors +16%, dentists +30%, veterinary services +20%, medical laboratories +9%, retirement homes +29% hospitals +16% taxis+19%" (see: http://www.cartes-bancaires.com/FR/telechargements/chiffresCB2004FR.pdf).

680 See recitals 230 and 493 above.

681 The present decision does not take a position on the lawfulness of the CIP/CIR interchange fees, nor on the role that they play or the real effects that they have as these CIP/CIR interchange fees currently stand, as stated in footnote 310 above.

682 S2P's reply of 13 July 2005 to the Commission's request for information of 29 June 2005 (file pp. 24409 and 25633).

683 GE Money Bank's reply of 21 July 2005 to the Commission's request for information of 29 June 2005, p. 5, file p. 24660.

684 Covefi's reply of 19 July 2005 to the Commission's request for information of 29 June 2005 (file p. 24463).

685 Axa Banque's reply of 11 July 2005 to the Commission's request for information of 29 June 2005 (file p. 24394).

686 Il convient à cet égard de remarquer que si, selon la jurisprudence de la Cour et du Tribunal de première instance des Commuautés européennes (affaires GVL et Sumitomo), la Commission doit particulièrement veiller à démontrer un intérêt légitime à adopter une décision constatant une infraction lorsque l'entreprise en cause a déjà mis fin à l'infraction avant l'adoption de la décision contestée, en l'espèce l'infraction continue. (Case 7-82 GVL v. Commission of 2 March 1983 [1983] ECR p.483, recitals 24 to 28 ; Joined Cases T-22-02 and T-23-02 Sumitomo Chemicals Co. Ltd and others v. Commission of 6 October 2005 [2005] ECR II p. 4065, recitals 130 to 139).

687 The measures have not been annulled by the Groupement, but merely suspended ; see recitals 259 and 344 above.

688 See Section 10.2.2.2. B., recitals 345 to 357, where it is explained that the measures continue to have effects after their suspension.

689 See recital 344 and footnote 514 above.