Asda Stores Ld & Ors v Mastercard Incorporated & Ors

The Defendants are collectively referred to in this judgment as MasterCard; it is not necessary to distinguish between them for the purposes of the issues which I have to decide. MasterCard sets the level of fees to be charged for its credit and debit card transactions. Such fees are known as multilateral interchange fees or “MIFs”. The MIFs are (indirectly) paid by the merchants with whom cardholders use their cards to purchase goods or services. Although MasterCard sets the MIFs, it does not receive them; the MIFs are paid to the institutions which issue the payment cards to cardholders. The issuers are typically, but by no means exclusively, banks. The Claimants are well known high street retailers who collectively claim to have paid a total of about £437 million by way of MIFs in the relevant period (broadly speaking since 2006). The central issue in these cases is whether the MIFs set by MasterCard for credit and debit card transactions by consumers in the relevant period, and in the relevant territories (UK, Ireland and cross border transactions between EEA countries), are tortiously anti-competitive in breach of UK, Irish and EU competition law; and if so by how much.

There are twelve separate actions, two of which have been brought to an end by settlement since the commencement of the hearing. The trial before me was to determine a number of issues which were selected and framed to determine all issues of liability and some issues affecting quantum (if relevant), in what was referred to as “the Phase 1 trial”. Other questions relevant to quantum have been left for consideration (if appropriate) at a subsequent hearing (“the Phase 2 trial”).

In this section I address the background under the following headings:

The MasterCard scheme

The MasterCard payment card scheme, like that of Visa, is often described as a “fourparty” scheme, although it involves five parties. It is to be distinguished from “threeparty” schemes which involve three parties.

A three-party scheme is represented diagrammatically as follows:

By contrast, a “four-party” scheme involves five parties in the following structure:

The four parties in grey in the diagram are each necessarily involved in the flow of payments arising from a transaction: the issuer issues the payment card to the cardholder; the cardholder “pays” for the goods or services provided by the merchant by card; the acquirer provides the “acquiring” service to the merchant which allows it to receive card payments, and pays the merchant for the goods and services, less a discount. The acquirer obtains payment from the issuer, who charges the cardholder for the transaction. MasterCard charges a fee to both issuers and acquirers for their ability to participate in the scheme.

The financial dynamics for those participating in the scheme are as follows:

Interchange fees can in theory be agreed bilaterally between issuers and acquirers. In practice this is not how the interchange fee is determined. Under the Scheme Rules (Rule 8.3), MasterCard sets the interchange fees which are to apply compulsorily in default of bilateral agreements. These are the multilateral interchange fees or “MIFs”. In practice there are no material bilateral agreements, and so the MIF always applies. This is not surprising: in a putative bilateral negotiation between an issuer and an acquirer the issuer has no incentive to accept less than the default MIF and the acquirer no incentive to offer more.

Not only have the MIFs set by MasterCard varied over the period covered by the claims, but at any given time MasterCard has had a wide variety of different MIFs for different categories of transaction. There are distinctions between territories and between credit and debit card transactions, so that the broadest relevant categories are the card-type/territory groups that define the MIFs in issue in this claim: i.e., EEA debit, EEA credit, UK debit, UK credit, Irish debit and Irish credit MIFs. But within such groups (in the territories in issue and elsewhere) MasterCard also differentiated MIFs according to a variety of different factors, such as whether the transaction was face-to-face, online, or made by mail or telephone order; whether the card and/or the payment terminal had chip and pin functionality; how the payment was verified; whether cards were premium cards; and by the category of merchant.

MIFs may be expressed as a percentage of transaction value, sometimes referred to as “ad valorem”, as a fixed sum per transaction, or as a combination of both. Very broadly speaking, MasterCard’s debit card MIFs have tended to be set on a fixed or combined basis, and its credit card MIFs have been purely ad valorem.

The payment flows between the four parties can be illustrated as follows, assuming a transaction at £100, a MIF of 1% and an acquirer’s margin of 0.1%. The dotted line represents the notional payment by the cardholder for the goods or services with the black lines representing the actual payments.

A payment system of this kind is known as a two sided platform, with merchants on one side and cardholders on the other. The platform is two sided because the more users there are on one side the more attractive the platform is to the other side. The more consumers with a MasterCard payment card, the more attractive it is for merchants to accept MasterCard cards; the more merchants who accept such cards, the more attractive it is for consumers to carry them.

In a number of countries, including the UK, Amex not only operates a pure threeparty scheme which issues cards directly to cardholders, but also licenses third-party issuers (for example Lloyds Bank) to issue Amex cards. This part of Amex’s operation, known as “Amex GNS” (for “Global Network Services”) blurs the distinction between three-party and four-party schemes. MasterCard and Visa compete for issuers’ business with Amex GNS, but not in respect of the “pure” Amex business.

It can be seen that in order for acquirers to operate profitably they must include within their MSC charged to merchants the amount of the MIF payable to issuers, and the scheme fee payable to MasterCard, together with such profit element as they are able to negotiate with the merchant in question. The expression acquirer’s margin was used sometimes to reflect this profit element, and sometimes to represent the difference between the MSC and the interchange fee so as to include scheme fees. The evidence established that the acquirer’s margin on relevant transactions, in either sense, was a small proportion of the MSC; the MIF, which was to be passed on to the issuers, typically comprised over 90% of the MSC. In the agreements between the Claimants and their acquirers in evidence in this case, a common charging structure for the MSC was for it to be expressed simply as interchange plus scheme fees plus acquirer’s margin, in which it was only necessary to quantify the latter as the element of the agreement between the merchant and the acquirer which was bespoke.

The MIF therefore comprises a floor on the price which the merchants have to pay for accepting cards: in normal circumstances no acquirer will charge a merchant less than the MIF because to do so would necessarily create a loss. It is this aspect of the MIF which forms the central complaint of the Claimants in these proceedings. By creating such a floor when setting default MIFs, MasterCard is said to have unlawfully restricted competition on the market between acquirers and merchants, with the result that the merchants have had to pay too much to accept MasterCard’s credit and debit cards.

The MIF is not, however, what MasterCard has received; it is what has been paid to the issuers. Whilst there was once a time when the issuers could be closely identified with MasterCard because the latter was owned and/or controlled by issuing banks, as I explain below that is not the case for the majority of the period with which these claims are concerned. There is no complaint about the licencing fees charged by MasterCard to acquirers which forms an element of the MSC paid by merchants.

The Scheme Rules contain what was referred to as the Honour All Cards Rule or “HACR” (Rule 5.10.1), which requires that a merchant which accepts MasterCard cards must accept all such cards. A merchant cannot, for example, accept MasterCard credit cards issued by one bank but not another, or standard but not premium cards.

MasterCard used to impose a no discrimination rule, which prohibited merchants from imposing a surcharge when a customer used a MasterCard card to pay. The rule was abolished at the beginning of 2005, and was to some degree overridden by UK domestic legislation before then. Merchants have therefore been free to surcharge their customers for the use of cards throughout the claim period. Although some merchants do, none of the Claimants has chosen to do so.

Types of card and card user

There are various different types of payment cards within the MasterCard scheme. The two main categories of card are “pay now” cards, comprising debitand, more recently, prepaid cards; and “pay later”cards, comprising creditand charge cards. The parties referred to “pay now” cards generally as debit cards, and “pay later” cards generally as credit cards. I shall use the same shorthand.

Within these categories there are a number of subdivisions:

The card industry distinguishes between two types of credit card user: “revolvers” and “transactors”. Transactors pay off their bill in full at the end of every month, and so do not make use of a credit facility above and beyond the interest-free period. Revolvers, on the other hand, roll over some or all of their debt at the end of the month, and thus pay interest. Premium cards and reward cards are generally aimed at transactors, who by definition have no real need of extended credit: the rewards are an incentive for them to use credit cards despite that fact.

The legal framework for the claims

The claim is concerned with the MIFs charged for both credit and debit card transactions in three geographical markets, namely:

The period of the claim is different as between Claimants and territories because it is affected by the local law limitation regime and the date of commencement of proceedings.

In relation to the UK domestic MIFs:

In relation to the Irish domestic MIFs, the claim runs from 23 May 2006 or 5 October 2006 until 5 January 2007 (because the intra EEA MIF applied as the default MIF for domestic Irish transactions for that period) and thereafter from 20 January 2009 for all Claimants. There is no claim for the period between 5 January 2007 and 20 January 2009 for limitation reasons.

In relation to the intra EEA MIFs:

The claim is for infringement of:

It was not suggested that there are any differences between Article 101 TFEU and Article 53 EEA Agreement which are material to the issues I have to decide. Moreover it was common ground that for the reasons I explain below, the dispute in relation to the UK and Irish MIFs turned on the principles embodied in Article 101 and its application. I shall therefore use infringement of Article 101 as a shorthand to represent the infringement of the various provisions which is alleged.

The wording of Article 101 TFEU has remained unchanged throughout the claim period, although the Article number and name of the applicable treaty have changed: prior to December 2009 the wording of Article 101 TFEU was contained in Article 81 of the European Community Treaty.

Articles 101(1) and 101(2) have direct effect in the English courts and so confer individual rights which are enforceable as between private parties in civil proceedings under s.2 European Communities Act 1972. An agreement may be held to infringe Article 101(1) by the Commission, by national competition authorities (“NCAs”), and by national courts in the Member States. In the UK the NCA was formerly the Office of Fair Trading (“OFT”), and is now the Competition and Markets Authority (“CMA”).

Article 101(1) provides for a general prohibition on restrictive arrangements between undertakings as follows:

“1.

The following shall be prohibited as incompatible with the internal market: all agreements between undertakings, decisions by associations of undertakings and 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 internal market, and in particular those which:

(a)

directly or indirectly fix purchase or selling prices or any other trading conditions;

(b)

limit or control production, markets, technical development, or investment;

(c)

share markets or sources of supply;

(d)

apply dissimilar conditions to equivalent transactions with other trading parties, thereby placing them at a competitive disadvantage;

(e)

make the conclusion of contracts subject to acceptance by the other parties of supplementary obligations which, by their nature or according to commercial usage, have no connection with the subject of such contracts.”

In order for an arrangement to fall within that prohibition, three elements must be established:

The first requirement identifies three alternatives; it is fulfilled by an agreement between two or more undertakings or a decision by an association of undertakings or a concerted practice. There was ultimately no real argument that the first requirement is fulfilled in this case. The MIF is set as a default MIF in the Scheme Rules, which comprise part of the contractual terms between MasterCard and each issuer and acquirer. The payment of the MIF is made pursuant to the terms of the agreements between MasterCard and its licensees. Moreover the payment of the MIF at the levels set by MasterCard was a concerted practice. This was not in issue in anything other than a formal sense: the Claimants were also alleging that MasterCard constituted an association of undertakings; in the course of the trial it was agreed between the parties that the Claimants would not pursue such an allegation in return for MasterCard’s undertaking that it would advance no argument against the proposition that there was a relevant agreement or concerted practice. This was not a formal concession, but MasterCard’s position was plainly realistic: the setting of the MIF was pursuant to an agreement between undertakings and was a concerted practice. For simplicity I shall refer to the three concepts in the first requirement as “an agreement”, as did the parties.

Nor was there any dispute about the third requirement (effect on trade) if the second requirement were fulfilled.

The words “object or effect” in the second requirement are disjunctive: if an agreement has the object of restricting competition, there is no need to prove effect; conversely an agreement with restrictive effect fulfils the requirement even if that were not its object. The European Court of Justice, now the Court of Justice for the European Union (which I shall simply refer to throughout this judgment as the CJEU) has explained that the “distinction between ‘infringements by object’ and ‘infringements by effect’ arises from the fact that certain forms of collusion between undertakings can be regarded, by their very nature, as being injurious to the proper function of normal competition”: Case C-209/07 Beef Industry Development Society [2008] ECR I-8637, paragraph 16.

The Claimants in the present cases do not allege any restriction by object. Accordingly the central battleground on Article 101(1) in this case was whether the MIFs had the effect of the prevention, restriction or distortion of competition to an appreciable extent. I shall use the expression “restriction of competition” as a shorthand to represent that broader concept: the issue under Article 101(1) in this case is simply whether the MIFs were a restriction on competition. It is common ground that the burden of proof lies on the Claimants to establish a restriction of competition which infringes Article 101(1).

For an agreement to be restrictive by effect it must affect actual or potential competition to such an extent that on the relevant market negative effects on prices, output, innovation or the variety or quality of goods and services can be expected with a reasonable degree of probability. Such negative effects must be appreciable: see paragraph 24 of the Commission’s Guidelines on the application of Article [101(3)] of the Treaty, OJ C 101/97 (“the Article 101(3) Guidelines”). I shall adopt the shorthand used by the parties of “anti-competitive effect” to express this concept.

In order to determine whether an agreement has anti-competitive effect, there are two elements in the analysis whose existence was common ground but whose application was controversial in this case:

As to the relevant market, MasterCard initially argued that the relevant market in a two sided platform such as the MasterCard card system included all aspects of the markets in which cards were used, and in particular included an examination of the issuing market between issuers and cardholders, and the inter-system market between MasterCard and Visa and Amex, as well as the market between issuers and acquirers and between acquirers and merchants. MasterCard’s argument to the same effect was successful before the Competition Appeal Tribunal: see particularly paragraphs 130 to 137 of its judgment. The Claimants, on the other hand, argued that the relevant market was confined to the market in acquiring services alone; all that was required to be proved was a restriction in the market between acquirers and merchants; and that provided that such a restriction was made out, the effect of the MIFs on the issuing and inter-system market, and any benefits there created, were matters for consideration within the framework of Article 101(3), not 101(1). This was the conclusion reached by the Commission in the MasterCard Commission Decision: see paragraph 316. It is not necessary for me to address these rival arguments, because in the course of final speeches MasterCard conceded that I was bound by a recent decision of the General Court dated 30 June 2016 in Groupement des Cartes Bancaires (CB) v Commission Case T-491/07 RENV to hold that the Claimants were right on this point and that the relevant market was solely the acquiring market. This authority was not drawn to the attention of the CAT before it published its decision.

It was also common ground that the geographical identification of the relevant market was to be defined by national boundaries: it required a separate consideration of the effect of UK domestic MIFs on the UK acquiring market, that of Irish domestic MIFs on the Irish acquiring market and that of intra EEA MIFs on the intra EEA acquiring market.

What remained controversial between the parties was the appropriate counterfactual hypothesis against which to test whether the MIFs were a restriction on competition in the acquiring market compared with the position which would have obtained in such hypothetical counterfactual circumstances. This is what is referred to as the “restriction counterfactual”.

A restriction counterfactual must be such as would be likely to occur in the absence of the impugned agreement or practice and must be realistic: CJEU MasterCard Judgment, paragraphs 161-169. In this respect, it is permissible, where appropriate, to take account of the likely developments that would occur on the market in the absence of the agreement in question: ibid, paragraph 166.

There is also a principle, called the “ancillary restraint” doctrine, which may take an agreement or practice outside the scope of the prohibition in Article 101(1). It arises where there is a main operation which is neutral or pro-competitive but has as one of its constituent parts what would be a restraint on the autonomy of the parties if considered in isolation. The ancillary restraint will not offend the prohibition in Article 101(1) if it is directly related and objectively necessary to the implementation of the main operation and proportionate to its objectives: see the Article 101(3) Guidelines paragraph 29. The requirement of objective necessity means that it must be shown that in the absence of the ancillary restraint the main operation would be impossible to carry out; it is not sufficient if it would merely be more difficult or less profitable. As explained in the CJEU MasterCard Judgment:

“89.

……if a given operation or activity is not covered by the prohibition rule laid down in Article [101(1)], owing to its neutrality or positive effect in terms of competition, a restriction of the commercial autonomy of one or more of the participants in that operation or activity is not covered by that prohibition rule either, if that restriction is objectively necessary to the implementation of that operation or that activity and proportionate to the objectives of one or the other…

90.

Where it is not possible to dissociate such a restriction from the main operation or activity without jeopardising its existence and aims, it is necessary to examine the compatibility of that restriction with Article [101] in conjunction with the compatibility of the main operation or activity to which it is ancillary, even though, taken in isolation, such a restriction may appear on the face of it to be covered by the prohibition rule in Article [101(1)].

91.

Where it is a matter of determining whether an anticompetitive restriction can escape the prohibition laid down in Article [101(1)] because it is ancillary to a main operation that is not anti-competitive in nature, it is necessary to inquire whether that operation would be impossible to carry out in the absence of the restriction in question. Contrary to what the appellants claim, the fact that that operation is simply more difficult to implement or even less profitable without the restriction concerned cannot be deemed to give that restriction the ‘objective necessity’ required in order for it to be classified as ancillary. Such an interpretation would effectively extend that concept to restrictions which are not strictly indispensable to the implementation of the main operation. Such an outcome would undermine the effectiveness of the prohibition laid down in Article [101(1)].”

It was common ground that the burden of proof lies on MasterCard to bring itself within the ancillary restraint doctrine. This is, however, an evidential burden rather than a legal one; the burden is on the Claimants to establish a restriction of competition which infringes Article 101(1): see Racecourse Association v Office of Fair Trading [2005] CAT 29 at paragraphs 131-133.

Application of the ancillary restraint doctrine also requires the court to consider counterfactual circumstances in which the ancillary restraint does not exist, in order to determine whether it is objectively necessary for the existence of the main operation compared with the potential existence of the main operation in the counterfactual circumstances. This is referred to as the “ancillary restraint” counterfactual. However the test for choosing the ancillary restraint counterfactual is slightly different from that which determines choice of the restriction counterfactual. The same counterfactual is not necessarily appropriate for conceptually distinct issues: CJEU MasterCard Judgment paragraphs 163, 108-111.The ancillary restraint counterfactual can be one which might arise in the absence of the restraint, although it must be realistic: ibid at 111.

Thus the test for choosing the counterfactual for the purposes of the ancillary restraint doctrine provides a lower threshold for the regulator or other person complaining of a restraint than the test for choosing the counterfactual for the purposes of establishing a restriction of competition within the prohibition of Article 101(1). For the ancillary restraint doctrine it is enough for a person complaining of infringement to point to one or more counterfactuals which might arise, by comparison with which the ancillary restraint is not necessary for the survival of the main operation. By contrast, in order for the prohibition in Article 101(1) to bite on the restriction, it must be a restriction on competition by comparison with a restriction counterfactual which likely would arise. In both cases the counterfactual must also be realistic.

Article 101(2) provides that any agreement or decision prohibited in Article 101(1) “shall be automatically void”. In this connection:

Article 101(3) TFEU provides for what is commonly referred to as “exemption” from the prohibition in Article 101(1) TFEU in the following terms:

“The provisions of paragraph 1 may, however, be declared inapplicable in the case of:

-

any agreement or category of agreements between

undertakings,

-

any decision or category of decisions by associations of undertakings,

-

any concerted practice or category of concerted practices,

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:

(a)

impose on the undertakings concerned restrictions which are not indispensable to the attainment of these objectives;

(b)

afford such undertakings the possibility of eliminating competition in respect of a substantial part of the products in question.”

Thus, the exemption applies where the agreement, decision or practice meets the four cumulative conditions, namely:

It is for the undertaking wishing to claim the benefit of exemption to demonstrate that the agreement is exempt under Article 101(3).

An agreement may be exempt by either of the following means:

Until 1 May 2004, under the old enforcement regime pursuant to Regulation 17/62 (OJ Series I Volume 1959-1962 p87-93), it was the case that only the Commission could rule on whether or not an agreement benefited from exemption. Moreover, in order to obtain such an exemption from the Commission, an agreement had to be formally notified by the parties to the Commission. In the absence of an applicable block exemption regulation, an agreement which infringed Article 81(1) (as Article 101(1) then was) was automatically void by virtue of that infringement, unless and until the Commission had made a formal exemption decision. The exemption took the form of a decision addressed to the parties to a specific agreement or decision. It applied only for a certain period of time specified by the Commission in its decision.

On 1 May 2004, Regulation 1/2003 (OJ 2003 L1/1) and a new implementing Regulation 773/2004 (OJ 2004 L123/18) came into force, which introduced a number of policy reforms to “modernise” the enforcement of EU competition law. In particular, the new regime aimed to decentralise the enforcement of EU competition law by devolving authority to NCAs and/or national courts. The NCAs and national courts are charged with applying EU competition law consistently and uniformly across the Member States and in close cooperation with the Commission: see Articles 15 and 16 of Regulation 1/2003.

As a result of such modernisation:

In any national court proceedings, the burden of proving an infringement of Article 101(1) to the requisite legal standard rests on the party or authority alleging the infringement. Once that is established the burden shifts: it is the undertaking or association of undertakings claiming the benefit of Article 101(3) which bears the burden of proving that the exemption conditions have been satisfied: see recital 5 and Article 2 of Regulation 1/2003.English competition law

Sections 2 and 9 of the Competition Act 1998 (“the CA98”), which came into force on

The CA 98 contains certain provisions which are designed to give primacy to EU law and to reduce the possibility of conflicting solutions being reached under the two regimes:

Regulation 1/2003 makes further provision to ensure consistency between UK and EU competition law as applied by the national courts, including that:

As to administrative regulation, the CMA, formerly the OFT, has powers of investigation and enforcement similar to those of the Commission, which it can exercise in relation to Article 101 TFEU, the CA 98 Chapter I prohibition or a combination of the two: see sections 25-44 of the CA 98 and Article 5 of Regulation 1/2003.

The relevant law in Ireland is contained in section 4 of the Competition Act 2002 which applies within Ireland and can be taken for present purposes to be materially equivalent to English law applicable within the UK.

Accordingly it was common ground that the dispute in relation to the UK and Irish MIFs turned on the principles embodied in Art 101 and its application; and that for the purposes of the issues which I have to decide, English and Irish law can be taken to be identical to EU law. Since much of the jurisprudence is in the European Court, it is convenient to refer to the provisions of Article 101 TFEU rather than the equivalent domestic provisions. I shall therefore use infringement of Article 101 as a shorthand to represent all the alleged infringement, including that under English and Irish domestic legislation.

Regulatory History

MasterCard’s MIFs and Visa’s MIFs have been the subject matter of regulatory and court decisions in Europe which form important context to the issues in the case.

The Commission was first notified of Visa’s MIFs and other rules by Visa International’s predecessor, Ibanco Ltd, in 1977. Having sent a comfort letter, the Commission reopened its investigation in 1985. The investigation at first focused on certain Visa rules other than the MIF. In a decision of 9 August 2001 [2001] OJ L 293/24 the Commission held that none of those rules restricted competition within the meaning of what is now Article 101.

On 24 July 2002, the Commission adopted a decision in relation to Visa’s EEA MIFs [2002] OJ L 318/17. It found that they restricted competition within the meaning of what is now Article 101(1), but that the setting of those MIFs could be exempted pursuant towhatis now Article 101(3) for a period of five years, pursuant to undertakings from Visa that it would:

The Commission made clear in its decision that over the five-year exemption period, it would re-examine the practical impact of the modified Visa scheme on the market and in particular its effect on merchant fees.

The 2002 Visa Decision had no immediate effect on MasterCard MIFs. MasterCard was not bound by it and did not respond by voluntarily reducing its MIFs.

In the UK, the OFT investigated MasterCard’s MIFs and on 5 September 2005 issued a Decision against MasterCard (the First, Third and former Fourth Defendants) and the UK licensees of the First and Third Defendants. It held that MasterCard’s UK MIFs infringed the EU and UK competition rules in that they restricted competition and did not satisfy the criteria for exemption. In November 2005, MasterCard appealed against the OFT’s infringement decision. In defending that appeal, the OFT changed the factual basis on which it had held that there had been an infringement of the relevant competition provisions. In the circumstances, and following submissions from Visa who supported MasterCard’s appeal, the OFT decided to withdraw its infringement decision on procedural grounds. The Decision was formally set aside by the Competition Appeal Tribunal on 10 July 2006: see [2006] CAT 14. The OFT continued to investigate MasterCard’s UK MIFs (and Visa’s equivalent MIFs), albeit focussing on the (then) current MIFs rather than the historic arrangements. In practice, the investigation was put on hold shortly thereafter pending the outcome of the Commission’s investigation into MasterCard’s EEA MIFs and MasterCard’s subsequent appeals (see below); and, soon after the CJEU’s Judgment in September 2014, the CMA announced that, in view of the proposed EU regulation on interchange fees (see below), it would not be progressing its investigations.

Immediately upon expiry of the Visa exemption period in 2007, the Commission commenced a fresh investigation into Visa’s EEA MIFs.

Meanwhile the Commission had been investigating MasterCard’s EEA MIFs and in a decision of 19 December 2007 (“the MasterCard Commission Decision”) it found that, in breach of Article 81 EC, MasterCard’s cross border EEA MIFs for the period since 22 May 1992 had restricted competition and were not exempt.

MasterCard reacted to the Commission Decision by appealing, and in the meantime reduced its EEA MIFs to zero. Visa did not respond by reducing its own EEA MIFs but continued to engage with the Commission to negotiate acceptable commitments in relation to them. Neither Visa nor MasterCard reacted by reducing their UK MIFs. In July 2009 MasterCard increased its EEA MIFs from zero to 0.3% for credit cards and 0.2% for debit cards, following discussions with the Commission.

Visa’s negotiation strategy was successful. On 8 December 2010, the Commission adopted a decision accepting commitments offered by Visa in relation to its EEA debit card MIFs. The Decision:

This 2010 Visa debit card Decision marked a departure from its 2002 Decision in the methodology it adopted when considering the lawful level of a MIF. The 2002 Decision was based on an issuer costs based methodology. The 2010 Decision adopted the concept of a Merchant Indifference Test and announced that the Commission no longer looked at an issuer cost-based justification for MIFs. These two different methodologies play an important part in the Article 101(3) issues in this case and are discussed more fully below.

MasterCard’s appeal against the Commission Decision was dismissed by the General Court on 24 May 2012: [2012] 5 CMLR 5 (GC) (“the MasterCard General Court Judgment”). MasterCard appealed to the CJEU.

Prior to the determination of MasterCard’s appeal, on 26 February 2014 the Commission adopted a decision accepting further commitments offered by Visa in relation to its EEA credit MIFs. The Commission:

On 11 November 2014 MasterCard’s appeal from the General Court was dismissed by the CJEU: [2014] 5 CMLR 23 (ECJ) (“the MasterCard CJEU Judgment”).

In July 2013, the Commission had published a proposal for a regulation capping interchange fees across Europe. The proposal was based at least in part on the Commission’s frustration that “[i]n spite of the [MasterCard] General Court Judgment…international and national card schemes operating in the EU currently do not seem willing pro-actively to adjust their practices to comply with the European and national competition rules”: (Proposal for a Regulation of the European Parliament and of the Council on interchange fees for card-based payment transactions, COM(2013) 550 final, page 6).

Regulation (EU) 2015/751 on interchange fees for card-based payment transactions [2015] OJ L 123/1 (“the IF Regulation”) was adopted on 29 April 2015 and came into force on 8 June 2015. Articles 3 and 4 cap interchange fees for debit cards at 0.2% (or €0.05), and for credit cards at 0.3%. Member States may impose lower caps for domestic transactions, but the UK has not done so. Ireland has imposed a lower debit card interchange fee of 0.1%. The caps do not apply to three-party schemes, except insofar as they license third parties to act as issuers and/or acquirers: Article 1(4)–(5).

All parties before me submitted that the IF Regulation’s maximum caps are irrelevant to the determination of the present dispute.

This court is bound by legal principles established by the CJEU, both in applying Article 101 (by reason of s.3 European Communities Act 1972) and in applying the equivalent provisions of the CA 98 to the UK MIFs (by reason of s.60 of CA 98). It is agreed that the same applies in respect of the Irish MIFs. Decisions of the General Court are binding insofar as consistent with CJEU decisions.

So far as concerns the MasterCard Commission Decision:

It follows that:

Section 60(3) CA 98, by which this court must “have regard to any relevant decision or statement by the Commission”, applies not only to the MasterCard Commission Decision, but also to the Article 101(3) Guidelines published by the Commission. They are sometimes referenced to decisions of the CJEU, and in such cases obviously derive their authority from those decisions to the extent they are an accurate summary. In other respects they are to be had regard to and given weight, although I bear in mind that they are only guidelines and by their own terms recognise that they require flexible rather than mechanical application:

“6.

The standards set forth in the present guidelines must be applied in light of the circumstances specific to each case. This excludes a mechanical application. Each case must be assessed on its own facts and the guidelines must be applied reasonably and flexibly.”

Mr Lowenstein QC urged me to approach the issues by starting with the MasterCard Commission Decision, and applying it to the EEA MIFs for the majority of the claim period (which were not the subject of the Decision), and to the UK and Irish MIFs for the claim period, unless I could identify material differences which justified drawing a distinction. This process was characterised as “read across”. This suggested approach reflected the way the claims had been framed in the Statements of Case, with the Claimants relying on the MasterCard Commission Decision and MasterCard identifying respects which made its application to the current dispute inappropriate. This in turn infected the framing of the Phase 1 issues and of some of the issues on which the experts were asked to express their views.

I do not consider that this is a helpful way to address the issues which I have to decide, for a number of reasons. First, I am not bound by the Commission’s findings of fact and although it is sometimes possible to discern the evidence before the Commission which informed its conclusions, that is by no means generally the case. There is a logical flaw in the suggestion that this court should follow another tribunal’s findings of fact unless it can identify a specific and material difference in evidence when this court is not in a position to identify the extent of the evidence before that tribunal. There was, for example, a lively debate on whether by reference to the memorandum referred to at recitals 626ff and in Annex 7 the Commission had considered UK MIFs. It remains unclear exactly what aspect of this evidence the Commission took into account or how, without an understanding of which it is impossible to assess the validity of any “read across”.

Secondly, the period with which I am concerned, (2006 onwards) is very different from that being addressed by the Commission (1992-2007) with only a small period of overlap. The state of the markets was not identical and the data which I have had to work on is time specific and was not addressed by the Commission. It is also the case that economic theory has developed over the relevant period in a way which was simply not considered by the Commission. Moreover the link between MasterCard and the issuing and acquiring banks has also changed. For most of the period considered by the Commission the licensees could be closely identified with MasterCard in terms of economic interest. All acquirers were issuers, and MasterCard was owned by the issuers and acquirers; it was in substance the issuing and acquiring banks which were setting the MIFs. The terms of the Commission Decision suggest that this was a commercial reality which it regarded as important, and which is reflected in its language throughout: see, by way of example, the language of paragraph 13 referring to “MasterCard’s member banks” setting the MIF. Following an Initial Public Offering on 25 May 2006 (“the IPO”), the banks’ shareholding was reduced to a substantial minority shareholding (41%) with no voting rights, but a right to appoint three members of the Global Board of directors (eight being appointed by public investors and a charitable foundation). Bank equity in MasterCard declined to 15% in June 2010 and thereafter the banks had no right to appoint any directors. Since 2010 the banks’ equity interest has continued to decline so that today MasterCard licensees hold less than 2% of the economic interest in MasterCard, with no voting rights.

Thirdly, the EEA MIFs being considered by the Commission fell to be considered in a very different market context from the domestic MIFs which I have to consider. To give some examples, the UK credit card market sits in a culture of free banking, compared with say France where typically cardholders pay a substantial annual fee; the number of card payments and average spend per cardholder per inhabitant in the UK in 2014 (167 transactions totalling €9,996) is more than double that for the average in Europe (79 transactions totalling €4,055), and the position is more pronounced if one compares the UK market over the claim period with Europe prior to 2007; the same is true of the number of cards per person, with UK inhabitants averaging over 2.5 cards compared with the European average of 1.5; this comparatively higher degree of penetration of cards in the UK is potentially relevant to competition between issuers and schemes in circumstances where cardholders can choose between cards they hold in making payment; credit cards accounted for around 11% of overall consumer transactions in the UK in 2014 compared with 4% in the rest of Europe; the UK has a disproportionately high share of online card sales compared with the rest of Europe: some 25% of all European e-commerce sales. These are but a few examples. It would be an unduly difficult and burdensome exercise first to try to identify all differences between the markets over a period of almost 25 years and then seek to attribute weight or significance, or the lack of it, to each difference.

Since I am not bound by the Commission’s findings of fact, nor (save for a brief period in relation to the EEA MIFs) its application of law to the facts found by it, it is unhelpful to start by an analysis of two different bodies of evidence and to compare them; and doubly so when the evidential basis for the Commission’s conclusions cannot be determined with certainty. I must decide the case on the evidence before me, which is largely directed to a different period of time and a different market.

That is not to say that I should ignore the Commission’s findings or reasoning save where they are binding. I have had regard to them and taken them into account, giving due weight to its expertise.

The CAT decision

Apart from these proceedings, there are a number of other actions which have been commenced in England bringing claims against both MasterCard and Visa on the grounds that their MIFs infringe competition law. Two sets of proceedings deserve mention.

Sainsbury’s Supermarkets Ltd (“Sainsbury’s”) brought proceedings in the Chancery Division against MasterCard claiming damages for infringement of Article 101 and Chapter I CA 98 in respect of the MasterCard’s UK MIFs since 19 December 2006 (19 December 2007 for Scotland). The action was transferred to the Competition Appeal Tribunal (“CAT”) and was heard by Barling J sitting with an experienced economist and an experienced competition QC. The hearing took place between January and March 2016. Judgment was delivered on 14 July 2016, in the course of the Phase 1 trial in the present actions: [2016] CAT 11 (“the CAT Judgment”). I shall have to refer to parts of the CAT Judgment in greater detail in the course of this judgment, but its essential conclusions, so far as potentially relevant to these actions, were as follows:

In the course of its detailed judgment, the CAT had to consider and resolve many of the same issues as those I have to decide, both of fact and law. Insofar as it made findings of law, I am not bound by its decision, but I naturally accord it considerable respect, coming from a body chaired by a judge of coordinate jurisdiction and which has very considerable competition law experience. However I have had the benefit of argument from a different legal team on the Claimants’ side, and of argument on both sides addressed specifically to the reasoning expressed in the CAT Judgment. This has inevitably resulted in different formulations and refinements of the legal arguments from those which were presented to the CAT, and in one instance an important authority being brought to my attention which was not before the CAT.

There is also, of course, a very substantial overlap between the factual issues decided by the CAT and those I have to decide. Here too I am not bound by the findings, although the parties agreed I should take them into account and give them such weight as I thought appropriate. It is important to keep in mind in this context that the evidence before me was not the same as that before the CAT in important respects. For example, the CAT Bilaterals counterfactual, which is at the heart of the CAT’s conclusion, was a construct of the Tribunal itself; it had not been addressed in the witness statements or experts reports of either party and had not been put to factual witnesses. By contrast the parties put before me detailed factual and expert evidence on the point, tailored specifically to the findings and reasoning in the CAT Judgment, which was all to the effect that such bilaterals were unrealistic. Moreover there was no identity between expert evidence in the two trials: Dr Niels gave evidence for MasterCard in both cases but different experts gave evidence on behalf of the respective claimants. They were not expressing the same views. For example, in the CAT proceedings Sainsbury’s and its expert accepted that a MIF at some positive level was lawful; whereas the Claimants before me and their expert contended that any MIF above zero was unlawful. Nor was there anything like identity in the factual evidence put before the CAT and this court, either documentary or oral. The CAT had documentary material which was not in evidence before me and vice versa. The CAT heard from four Sainsbury’s witnesses whose evidence I did not have; whereas I heard from a variety of Claimants’ witnesses whose evidence was not before the CAT. Some MasterCard witnesses were common to both sets of proceedings but some were not. The experience of having arguments and evidence tested in the Sainsbury’s proceedings inevitably led to fuller or more focused evidence before me on some points, both factual and expert; for example Dr Niels had the opportunity to consider over time, and address in writing, points which he had faced in cross examination in the CAT without forewarning. Even where the evidence was materially similar, I must make my own assessment of the witnesses and the other evidence before me; it would be an abdication of judicial responsibility simply to accept findings of fact made by the CAT.

Accordingly my approach to the CAT Judgment has been to use its reasoning on issues of both fact and law as a cross check for my own provisional conclusions, paying particular attention to the evidence on which findings of fact are based and how closely such evidence matches that before me.

The other proceedings which I should mention are the actions brought against Visa by the Claimants in these actions, and also by Sainsbury’s and Marks & Spencer Plc, alleging infringement in relation to Visa’s MIFs. The phase 1 trial in those proceedings commenced shortly after the conclusion of the hearing before me and is being heard by Phillips J. Similar issues arise, although the evidence and argument is not identical to that before me.

It is common ground that I am not in a position to make any findings about infringement by Visa: Visa was not represented in these actions and the lawfulness of Visa’s MIFs was not an issue raised by the parties on the pleadings or addressed in the evidence. Nevertheless Mr Lowenstein submitted that I should assume that Visa’s

Thirdly Mr Hoskins submitted that the evidence of both experts was that competition in the acquiring market is not affected by the level of the MIF. This was a mischaracterisation of the evidence. Mr Dryden’s evidence was not that there would be no material difference in competition on the acquiring market, but merely that the level of the MIF did not affect the ability of acquirers to compete with each other on the acquirer’s margin. His evidence was that the MIF did restrict competition by creating a floor, a restriction which is not removed or mitigated by the ability to compete on margin. Dr Niels stated that “Even if the MIF has the effect of raising the price in the acquiring market, this does not restrict acquirer’s ability to compete on the level of acquiring margin. Acquiring competition would be the same whether the MIF is set at say zero or at 1%.” This reasoning is fallacious because it conflates competition on “the level of the acquiring margin” with “acquiring competition” as a shorthand for competition on the acquiring market. It ignores the restriction on an acquirer from setting a MSC below the MIF. It is equivalent to saying that a manufacturers’ pricing agreement which inflates the price charged to wholesalers is not a restriction of competition on the retail market because wholesalers are free to compete on the margin of profit they charge in the retail price to customers, a proposition whose falsity is self-evident.

Fourthly, Mr Hoskins submitted that a prohibition on ex post pricing was not a likely counterfactual, relying on the evidence of Mr Douglas and others that it was not one of the commercial options considered in the event of the MIFs being ruled to be unlawful. I have already explained why this is not a bar to the zero MIF/no MIF counterfactual.

The Claimants rely upon the decision of the General Court in Metropole Television (M6) v EC Commission (2001) T-112/99 [2001] 5 CMLR 33, and in particular paragraph 109. What was said was in the context of the ancillary restraint doctrine, but the Claimants argue that if an analysis of competition is impermissible for the ancillary restraint doctrine, it must equally be impermissible for the restriction of competition analysis. I will consider the argument first in the context of the ancillary restraint doctrine and then in the context of the restriction counterfactual.

In Metropole, the General Court was concerned with an appeal by a partnership of French television companies from a negative clearance decision by the Commission in which they argued that the Commission had wrongly given clearance for only three years when it should have given clearance for the full duration of the relevant agreement as falling within the ancillary restraint doctrine and therefore outside what was then Article 81(1). The provision under scrutiny gave the partnership preferential broadcasting rights in respect of certain programmes produced by the parties to the agreement. They argued that it was objectively necessary to have such a restriction because the object of the partnership was to break into the market for the broadcasting of satellite pay-TV programmes and services in digital mode. The General Court upheld the Commission’s decision that the competition faced by the partnership in breaking into the new market was irrelevant in determining whether the restriction was objectively necessary for the operation of the partnership: the ancillary restraint doctrine was a relatively narrow and abstract one in which the inquiry was directed simply to whether the restraint was necessary to implement the main operation, not whether it was necessary to make the main operation a commercial success. At paragraphs 107 to 109 the General Court said:

“107 As regards the objective necessity of a restriction ….. it would be wrong, when classifying ancillary restrictions, to interpret the requirement for objective necessity as implying a need to weigh the pro and anti-competitive effects of an agreement. Such an analysis can take place only in the specific framework of Article 85(3) of the Treaty.

MasterCard submitted that this did not reflect the jurisprudence of the CJEU, in decisions both before and since, by which I am bound.

In Remia BV & others v The Commission (1985) Case 42/84, the Court was concerned with two agreements for the transfer of businesses manufacturing pickles and sauces which contained non-competition clauses intended to protect the purchasers from competition from the vendors on the same market immediately after the transfers. The Commission had refused the transferees’ application for exemption under what was then Article 85(3). In the context of the ancillary restraint doctrine the CJEU said:

“17 It should be stated at the outset that the Commission has rightly submitted - and the applicants have not contradicted it on that point - that the fact that non-competition clauses are included in an agreement for the sale of an undertaking is not of itself sufficient to remove such clauses from the scope of Article 85 (1) of the Treaty.

18

In order to determine whether or not such clauses come within the prohibition in Article 85 (1), it is necessary to examine what would be the state of competition if those clauses did not exist.

19

If that were the case, and should the vendor and the purchaser remain competitors after the transfer, it is clear that the agreement for the transfer of the undertaking could not be given effect. The vendor, with his particularly detailed knowledge of the transferred undertaking, would still be in a position to win back his former customers immediately after the transfer and thereby drive the undertaking out of business. Against that background non-competition clauses incorporated in an agreement for the transfer of an undertaking in principle have the merit of ensuring that the transfer has the effect intended. By virtue of that very fact they contribute to the promotion of competition because they lead to an increase in the number of undertakings in the market in question.

20

Nevertheless, in order to have that beneficial effect on competition, such clauses must be necessary to the transfer of the undertaking concerned and their duration and scope must be strictly limited to that purpose. The Commission was therefore right in holding that where those conditions are satisfied such clauses are free of the prohibition laid down in Article 85 (1).”

(my emphasis)

In in Gottrup-Klim Grovvareforeni v Dansk Landbrugs Grovvareselskab AmbA (DLG) (1992) C-250/92, the Court was concerned with a provision in a farming cooperative purchasing agreement that its members could not belong to a competing cooperative. In addressing whether such a provision was objectively necessary for the purposes of the main operation of a cooperative collective, the CJEU said:

“28 In the second set of questions, the national court seeks to ascertain whether a provision in the statutes of a cooperative purchasing association, the effect of which is to forbid its members to participate in other forms of organized cooperation which are in direct competition with it, is caught by the prohibition in Article 85(1) of the Treaty.

29

The plaintiffs in the main proceedings claim that the object or effect of such an amendment to the statutes is to restrict competition, inasmuch as the objective pursued was to put an end to ‘B’ members purchasing through LAG in competition with DLG, and thus to acquire a dominant position on the markets concerned.

30

A cooperative purchasing association is a voluntary association of persons established in order to pursue common commercial objectives.

31

The compatibility of the statutes of such an association with the Community rules on competition cannot be assessed in the abstract. It will depend on the particular clauses in the statutes and the economic conditions prevailing on the markets concerned.

32

In a market where product prices vary according to the volume of orders, the activities of cooperative purchasing associations may, depending on the size of their membership, constitute a significant counterweight to the contractual power of large producers and make way for more effective competition.

33

Where some members of two competing cooperative purchasing associations belong to both at the same time, the result is to make each association less capable of pursuing its objectives for the benefit of the rest of its members, especially where the members concerned, as in the case in point, are themselves cooperative associations with a large number of individual members.

34

It follows that such dual membership would jeopardize both the proper functioning of the cooperative and its contractual power in relation to producers. Prohibition of dual membership does not, therefore, necessarily constitute a restriction of competition within the meaning of Article 85(1) of the Treaty and may even have beneficial effects on competition.

35

Nevertheless, a provision in the statutes of a cooperative purchasing association, restricting the opportunity for members to join other types of competing cooperatives and thus discouraging them from obtaining supplies elsewhere, may have adverse effects on competition. So, in order to escape the prohibition laid down in Article 85(1) of the Treaty, the restrictions imposed on members by the statutes of cooperative purchasing associations must be limited to what is necessary to ensure that the cooperative functions properly and maintains its contractual power in relation to producers.” (my emphasis)

The Metropole judgment in 2001 made no reference in the relevant section to the Gottrup-Klim judgment, but did refer to Remia at paragraph 110, immediately after the paragraph relied on by the Claimants, in the following terms:

“110 Thus, in the judgment in REMIA v. E.C. COMM1SSION, the Court of Justice held that a non-competition clause was objectively necessary for a successful transfer of undertakings, inasmuch as, without such a clause, “and should the vendor and the purchaser remain competitors after the transfer”, it is clear that the agreement for the transfer of the undertaking could not be given effect. The vendor, with his particularly detailed knowledge of the transferred undertaking, would still be in a position to win back his former customers immediately after the transfer and thereby drive the undertaking out of business.”

This is an accurate summary of Remia, but does not support the statement in the previous paragraph or the conclusion of the General Court at paragraph 120. On the contrary, Remia is inconsistent with its reasoning at paragraph 109 and conclusion at paragraph 120, because in Remia the CJEU held that competitive effects on the purchaser, in the absence of the restriction, were the very thing which had to be taken into account when deciding whether the ancillary restraint on the seller of a noncompetition clause was necessary.

In the MasterCard General Court Judgment in 2012, the General Court made no reference to Remia or Gottrup-Klim but relied on Metropole in the following paragraphs:

“89 As [Metropole] shows, examination of the objective necessity of a restriction is a relatively abstract exercise. Only those restrictions which are necessary in order for the main operation to be able to function in any event may be regarded as falling within the scope of the theory of ancillary restrictions. Thus, considerations relating to the indispensable nature of the restriction in the light of the competitive situation on the relevant market are not part of an analysis of the ancillary nature of the restriction (see, to that effect, [Metropole] at [121]).

90 Accordingly, the fact that the absence of the MIF may have adverse consequences for the functioning of the MasterCard system does not, in itself, mean that the MIF must be regarded as being objectively necessary, if it is apparent from an examination of the MasterCard system in its economic and legal context that it is still capable of functioning without it.”

In the MasterCard CJEU Judgment, the CJEU did not refer to Metropole, but cited both Remia and Gottrup-Klim when defining the ancillary restraint doctrine at paragraph 89, quoted earlier.

At paragraph 111 the CJEU stated that in the context of objective necessity the question was whether the counterfactual was realistic and “enabled the MasterCard system to be economically viable”. Economic viability necessarily requires consideration of the competitive context in which the activity is carried out. Indeed competitive pressures are likely to be one of the most important aspects of any economic effect on the main operation in the counterfactual world.

It is noteworthy that when the CJEU referred to what the General Court had said at paragraphs 89 and 90 of the latter’s judgment, it carefully did not endorse the reliance on Metropole or the principle the General Court derived from it. At paragraph 94 it chose to quote and endorse only parts of paragraphs 89 and 90 of the General Court Judgment, and in particular it carefully omitted the parts which purported to rely on and apply the Metropole principle for which the Claimants contend:

“94 In ruling, in [89] of the judgment under appeal, that:

“[o]nly those restrictions which are necessary in order for the main operation to be able to function in any event may be regarded as falling within the scope of the theory of ancillary restrictions,” and in concluding, in [90] of the judgment under appeal, that:

“the fact that the absence of the MIF may have adverse consequences for the functioning of the MasterCard system does not, in itself mean that the MIF must be regarded as being objectively necessary, if it is apparent from an examination of the MasterCard system in its economic and legal context that it is still capable of functioning without it,” the General Court did not, therefore, err in law.”

In my judgement the CJEU jurisprudence clearly establishes that the court can and must take account of competition facing the main operation when considering whether the ancillary restraint is objectively necessary to the main operation, and the passage in Metropole relied on by the Claimants is out of line with that jurisprudence. In particular:

This conclusion seems to me to be consistent with the purpose of the ancillary restraint doctrine and its rationale within the framework of the control of anticompetitive activity laid down in Article 101. Where an operation has a neutral or positive competitive effect, it is not objectionable; therefore it ought not to be rendered unlawful by reason of some constituent element being a restriction if taken in isolation from the greater operation of which it forms a necessary part. If it is a necessary part of a desirable whole, the restraint “escapes” the application of Article 101(1): it is simply outside the scope of the type of restraints which are offensive to EU competition law. The doctrine is aimed at permitting the main operation where it would be unable to function if the constituent part were removed. That involves choosing a counterfactual which is “realistic”. If the main operation would be unable to survive in the realistic counterfactual world as a result of competition, the policy of the doctrine is engaged. There is no logic in drawing a distinction between non survival which arises from something inherent in the structure of the main operation itself and non-survival arising from external pressures on the main operation in the realistic counterfactual world. It is clear from paragraph 111 of the MasterCard CJEU Judgment that the inquiry has to address whether the operation is “economically viable”. To include economic consideration of all kinds other than competition is illogical, and moreover likely to be impractical in many cases. Further, such exclusion is inconsistent with the concept of the counterfactual being “realistic”. An ancillary restraint counterfactual which ignores what would in fact happen when the counterfactual assumptions are made can hardly be regarded as realistic.

Mr Lowenstein’s explanation of the rationale for the restrictive approach for which he was contending was that at the Article 101(1) stage, all one was concerned with was whether the restraint was necessary for the main operation to function at all, because if it is argued that the restriction is to benefit competition with third parties, that is an argument about the pro-competitive benefit of the restriction, which falls to be analysed under Article 101(3). This echoed what was said in paragraphs 107 and 108 of Metropole which immediately precede the passage relied on. But the approach dictated by Remia and Gottrup-Klim does not cut across the scheme of Articles 101(1) and 101(3) and does not require any weighing up of pro- and anti-competitive effects which is the function of Article 101(3). It requires an analysis of whether the main operation requires the ancillary restraint to be in place in order to be able to operate in the realistic counterfactual world. It starts from the premise that the main operation is desirable in competition terms because it has positive or neutral competitive effects. If that is so, the restrictive part falls outside the scope of Article 101 altogether if it is a necessary part of the desirable main operation. In deciding whether it is necessary, one examines what would happen realistically without it. If competition would kill off the main operation without the restriction being in place, the restriction is necessary for the main operation. That does not involve weighing pro and anticompetitive benefits, or making a value judgement in relation to such benefits; it is a binary question whether the main operation would survive at all in the realistic counterfactual world, taking account of such competitive pressures as would exist in that counterfactual world.

Mr Lowenstein also relied on the decision of the CJEU in Pronuptia de Paris v Schillgalis (Case 161/84) [1986] ECR 353; [1986] 1 CMLR 414, which was decided after Remia and before Gottrup-Klim. The case concerned restrictions in franchising arrangements for selling wedding dresses. At paragraphs [14] to [15] the Court identified the legitimate pro-competitive aspects of a franchise distribution agreement in permitting a franchisor to take advantage of his business experience and acumen in building up his know-how and commercial methods, which meant that it was not by itself a restriction on competition. At paragraphs [16] to [22] the Court identified restrictions in the franchising agreement which were justifiable as objectively necessary for the franchisor’s business and fell outside the scope of Article 101(1); these included restrictions on the franchisee from opening a shop in an area where he would be in competition with another franchisee. They were outside Article 101(1) because they were objectively necessary for the franchisor’s business in preserving the legitimate protection of his know-how. At paragraphs [23] to [24], on the other hand, the Court addressed a clause which required the franchisee to sell the franchised goods only at the named location, the vice of which was that the franchisee was prevented from doing so at a second location; and because this was a term for all franchisees it involved the franchisor determining the geographical division of the market for the franchised goods. As the Court put it: “The juxtaposition of clauses of this type results in a kind of market partitioning between the franchisor and the franchisees or among the franchisees and thus restricts competition within the network.” It was for that reason that the clause was held to be a restriction of competition falling within Article 101(1).

In the context of that clause, the Court went onto say in the passage relied on by Mr Lowenstein at [24]:

“It is of course possible that a prospective franchisee would not take the risk of becoming part of the chain, investing his own money, paying a relatively high entry fee and undertaking to pay a substantial annual royalty, unless he could hope, thanks to a degree of protection against competition on the part of the franchisor and other franchisees, that his business would be profitable. That consideration, however, is relevant only to an examination of the agreement in the light of the conditions laid down in Article 85(3).” (his emphasis)

This does not assist the Claimants or cast doubt on the analysis or conclusions I have reached above. The passage comes at the end of an analysis of what is necessary for the franchisors’ business and in particular necessary for the existence of such a franchise business which taken as a whole is neutral in competition terms. As the terms of the preceding paragraphs make clear, the passage is addressed to the question whether the franchisor could escape the operation of Article 101(1) by invoking the benefits to a franchisee of such a restriction, so as to justify the restriction from the point of view of the franchisor’s operation as one which attracted franchisees. In the language of the ancillary restraint doctrine, the “main operation” being considered was that of the franchisor, not the franchisee. In that context, the competition faced by the franchisee was a consideration which fell within Article 101(3) rather than 101(1) because it was a franchisor asserting that the restriction conferred a benefit to it, the franchisor, in attracting franchisees; but that benefit could not take the franchisor outside the scope of Article 101(1) because the franchisor could not suggest that it was objectively necessary for his operation; it was therefore a benefit (to the franchisor) which fell to be weighed in the Article 101(3) exercise. The Court was not concerned with whether the term was objectively necessary for the survival of a franchisee, and the passage relied on was simply not addressed to an issue of whether competition faced by the franchisee made the restriction objectively necessary for the franchisee’s operation. It tells one nothing about the relevance or otherwise of market competition to that question, or to the issue which arises before me.

Competition from Visa (and Amex) is an equally relevant factor if the death spiral argument is being considered not for the purposes of the ancillary restraint doctrine, but for the purposes of whether there is a restriction of competition which fulfils Article 101(1).

In Société Technique Minière (Case 56/65) the CJEU stated that when determining whether there was a restriction of competition within what was then Article 85(1) “the competition must be understood within the actual context in which it would occur in the absence of the agreement in dispute” and that this requires the purpose (or effect) of the agreement to be considered “in the economic context in which it is to be applied”. The economic context must take account of the economic effect of competitive activity.

The CJEU made clear in the MasterCard CJEU Judgment that in order to determine whether the impugned provision has an appreciable adverse effect on the relevant market compared with the counterfactual, it is necessary to take account of likely developments on all markets which might have an effect on the relevant market: see paragraph 177. In this case the relevant market on which to assess whether there is a restriction is the acquiring market, but in order to do so it is necessary to take account of likely developments on the issuing and inter-system market because the two sided nature of the platform results in interaction between the markets: see the MasterCard CJEU Judgment at paragraphs 177-179.

Accordingly the effect of competition from Visa in the inter-system market is relevant to the assessment of the impact on the acquiring market in the restriction counterfactual world. If the likely development on the inter-system market would be the collapse of the MasterCard scheme, so as to eliminate the MasterCard acquiring market altogether, that is a relevant consideration. Accordingly the death spiral argument is available for the restriction of competition issue just as it is for the ancillary restraint issue.

Mr Lowenstein submitted that if Visa’s MIFs were unlawfully anti-competitive, the counterfactual would have to assume Visa MIFs which were lawful. As he put it “two wrongs cannot make a right”. Although this appeared to be in issue for most of the hearing, Mr Hoskins ultimately conceded the principle, and accepted that if the court were to find Visa’s MIFs to be unlawful, the counterfactual world would involve taking MIFs which Visa could lawfully set, not those which it actually set.

It was also common ground that because the inquiry is what one assumes in a hypothetical counterfactual world, it is sufficient for the Claimants’ argument that Visa’s MIFs be unlawful without regard to complications of whether or when a regulatory body would have so declared them if MasterCard had reduced its MIFs to zero. If competition is to be taken into account in the counterfactual world it can only be lawful competition. Just as the counterfactual hypothesis is of MasterCard voluntarily adopting a zero MIF, or in the alternative counterfactual a putatively lawful MIF, so too the counterfactual is of Visa voluntarily adopting a lawful MIF, or in the alternative counterfactual a putatively lawful MIF, not that it is forced to by regulatory action.

Mr Lowenstein in turn confirmed that he did not seek to prove that Visa’s MIFs were unlawful and was not inviting the court to make a finding that Visa’s MIFs were unlawful. Nevertheless, he submitted, this court could and should proceed upon the assumption that they were unlawful to the same extent as MasterCard’s MIFs because the schemes were materially identical. MasterCard challenged the logic or validity of any such assumption.

I have concluded that if the premise is correct, namely that the schemes are materially identical, Mr Lowenstein is correct that I must assume in the counterfactual world that Visa’s MIFs would be constrained to the same extent as MasterCard’s and there would be no differential which could support the death spiral argument. My reasons, which do not entirely coincide with those advanced by Mr Lowenstein, are as follows.

Compared with the counterfactual under consideration of a MasterCard zero MIF, the MasterCard MIFs are a restriction of competition within Article 101(1) (as I have found under issue (2)) unless (a) saved by the death spiral argument or (b) exempt under Article 101(3). One can ignore for present purposes whether the MasterCard MIFs would be saved by exemption under Article 101(3) because if they are exempt under Article 101(3), the death spiral argument now under consideration is irrelevant. So the counterfactual of a MasterCard zero MIF assumes that the materially identical Visa MIFs are a restriction of competition unless saved by an equivalent death spiral argument to the effect that they would have been necessary for the Visa scheme to survive in the face of competition from MasterCard’s actual MIFs as set by MasterCard.

Such an argument faces two obstacles. First the circularity of reasoning betrays its falsity. Each scheme is only saved from being a restriction of competition if the other is lawful, but the other cannot be lawful if they are materially identical. The lawfulness of one cannot be saved solely by reason of the existence of the other when what has to be assumed in a counterfactual world is that the other is a hypothetically lawful scheme.

Secondly the reasoning is internally inconsistent. It starts by assuming a counterfactual in which the MasterCard MIF is set at zero; but it only holds good if the MasterCard MIFs are at a positive level above Visa’s because it is only in such circumstances that there could be a Visa death spiral argument which would render the Visa restriction of competition lawful.

Put another way, MasterCard’s argument involves asserting that it escapes Article 101 in relation to a practice which in fact restricts competition compared with a zero MIF counterfactual by arguing that it would face elimination by reason of competition from a competing scheme which is itself unlawful unless it would be saved because it would be eliminated by competition from MasterCard. If the schemes are identical the circularity of reasoning illustrates that the argument is self-serving. Moreover it is internally inconsistent because the counterfactual assumes a zero MasterCard MIF but

I have expressed the point in terms of the restriction of competition argument, but it applies logically and in the same way in the context of the ancillary restraint counterfactual. The ancillary restraint counterfactual is a MasterCard zero MIF. MasterCard seeks to justify the MIFs as necessary for the survival of the MasterCard scheme because of competition from Visa’s MIFs. But if the schemes are materially identical, Visa’s MIFs, which are also a restraint, could only be justified as an ancillary restraint if necessary for the survival of the Visa scheme, which could only be established in the face of MasterCard’s MIFs being as set, rather than, as required by the counterfactual world in which the argument arises, a zero MasterCard MIF. In a zero MasterCard MIF world, Visa could not justify positive MIFs as necessary to its survival by reference to competition from MasterCard.

This is obviously a sensible result in the context of regulation of competition. In the course of argument I put the example of a manufacturers’ price cartel which anticompetitively raised prices for consumers, and asked whether such a cartel could be justified as lawful on the grounds that it was the only way the manufacturers could stay in business in the face of a similar cartel of other manufacturers. Mr Hoskins accepted that such an argument would not be available, but on the grounds that cartels were restrictions by object, which were unlawful per se irrespective of effect, which would not apply to a consideration of agreements said to be anti-competitive by effect only. This does not strike me as a material distinction. If one posits a “cartel” agreement which on its particular facts had an anti-competitive effect, there is no reason why it should be saved by a rival “cartel” agreement which was anticompetitive by effect any more than if either or both were anti-competitive by object. If it were established that the terms of the anti-competitive agreements were materially identical, it would be a surprising result if each escaped unlawfulness by reference to competition from the other. Indeed it would be a result which is inconsistent with the policy behind Article 101.

The analysis also holds good for the alternative counterfactual of a putatively lawful MIF. Suppose, for example, that a MasterCard MIF of 0.5% would be exemptible under Article 101(3). If the schemes are materially identical, the assumption in that counterfactual world would be of a Visa MIF of 0.5% because any higher Visa MIF would be unlawful. Wherever the level of a lawful MasterCard MIF, the lawful Visa MIF in that counterfactual would be the same. There would be a level playing field in which case the death spiral argument would obviously not be available.

The CAT came to a different conclusion and accepted MasterCard’s argument in the following terms:

“159.

In our judgment, MasterCard is right and Sainsbury’s wrong as to the manner in which the counterfactual hypothesis is to be constructed:

(1)

The agreement whose anti-competitive effect we are testing is the agreement described in paragraph 84(3)(iii) above, namely the agreement between MasterCard and its licensees to have a default UK MIF.

(2)

We stress that we are testing the anti-competitive effect of this agreement. It would be wholly wrong for us to enter upon this inquiry presuming the default UK MIF to be anti-competitive. The whole point of the counterfactual exercise we are undertaking is to provide an analytical framework whereby the effect of an agreement can be tested by hypothesising its absence.

(3)

That being the case, it would be wrong in principle to make any presumption as to the constraints on rivals to MasterCard, like Visa, that is not rooted or based on fact.

160.

We reject, therefore, any suggestion that Visa’s Interchange Fees should be presumed, in the counterfactual world, to be the same as MasterCard’s.”

To my mind this approach does not dispose of the issue. The Tribunal did not abstain from making presumptions; it made a presumption about Visa’s MIFs in the counterfactual, namely that they would be as set. That must involve one of two assumptions: either it involves assuming that the lawfulness of Visa’s MIFs is irrelevant to the argument, a proposition which I have rejected and which ultimately was not pursued by MasterCard; or it assumes that Visa’s MIFs as set were lawful, which for the reasons I have endeavoured to explain is not a sound assumption if the premise of Mr Lowenstein’s argument is correct, namely that there is no material distinction between the schemes.

I return, therefore, to the critical question whether the premise of Mr

There was no focus on the detail of the Visa scheme in the evidence or argument before me. If in the many trial bundles there is material identifying the detailed working of the Visa scheme, such as its scheme rules, its card make-up, functionality and features, its issuer and acceptance network, its detailed individual MIFs, its scheme fees etc, none of it was drawn to my attention or relied on. As I clarified with the parties at the outset, and the parties agreed, material would only be treated as put in evidence at the trial if it was referred to in the course of the hearing or in the written evidence the Court was asked to read.

Mr Lowenstein put the features of the Visa scheme which were sufficient to make it materially identical at a very high level of generality. In the Claimants’ written opening the point was put in this way:

“213.

The Court does not have to make a finding that Visa’s MIF would be constrained (or how it would be constrained); it need only work on the counterfactual assumptions that (a) the MasterCard and Visa schemes are very similar (b) they are direct competitors with approximately the same market share and (c) that, therefore, if one scheme is unlawful it is more likely than not that the other will be too.”

When pressed in the course of oral submissions on what features of the Visa scheme he relied on, Mr Lowenstein submitted that the identity which was material for these purposes was that the Visa scheme was a four party scheme with an honour all cards rule for credit and debit cards in which Visa set a MIF to be paid to issuing banks, and retailers were charged a MSC by acquirers for which the MIF was a floor.

In taking this approach Mr Lowenstein argued that the only material identity which needed to be assumed was that which was relevant to the question of a restriction of competition at the Article 101(1) stage, which was where the death spiral argument arose for consideration; and that the burden of proof lay on MasterCard to point to material differences in the schemes, a burden which it had not sought to discharge.

I cannot accept that this is the right approach in either respect. In order for the premise to hold good, the features of the two schemes must be materially identical. In this context what is material is whether and to what extent the Visa MIFs as set constituted an unlawful restriction of competition infringing Article 101; or, put another way, what MIFs could lawfully have been set by Visa without infringing Article 101. That requires a consideration of all the features of the scheme which might affect the lawfulness of Visa MIFs, not just those which arise under Article 101(1) issues; it includes those which are relevant to Article 101(3) issues as they would apply to the Visa scheme. Unless the features of both schemes are identical in all respects which would affect the outcome of that legal inquiry, they are not materially identical. If, for example, there were features of the Visa scheme which would impact on the putatively lawful level of MIF determined as exemptible under Article 101(3), which differed from the MasterCard scheme, the schemes would not be materially identical for the purposes of the present inquiry. If Visa credit card MIFs were a restriction of competition at the levels set but would have been lawfully exemptible at say 0.7%, to take a hypothetical example, that is the lawful level of Visa MIF which would have to be taken as that with which MasterCard would have been competing, in a MasterCard zero MIF counterfactual, for the purposes of analysing the death spiral argument; or with which a putatively lawful MasterCard MIF, which on this hypothesis might be might be lower and different, would have been competing in the putatively lawful counterfactual. That would be the Visa MIF with which to populate what Mr Lowenstein called the counterfactual world of lawfulness. It is not sufficient, as Mr Lowenstein argued, to posit material identity between the schemes only in respect of aspects which are relevant to the restriction of competition issue under Article 101(1) and ignoring aspects which may affect a lawful level which is exemptible under Article 101(3).

As to the burden of proof on the material identity of the schemes, it is ultimately the Claimants who would fail on the relevant issue if the burden were not discharged. It is true that insofar as the issue arises for the purposes of the ancillary restraint argument, that is an argument on which MasterCard bears the evidential burden of proof. However the issue also arises for the purposes of the restriction of competition argument, on which the Claimants bear the burden of proof; and because MasterCard potentially succeeds on the Article 101(1) issue if the death spiral argument works for either ancillary restraint (making the MIFs necessary for the main operation) or restriction of competition (making the counterfactual one in which there is no acquiring market), it is the Claimants who must ultimately establish material identity between the schemes for the purposes of the death spiral argument. Moreover, even for the purposes of the ancillary restraint argument, the evidential burden would fall on the Claimants to establish the unlawfulness of Visa’s MIFs and so therefore, in the light of the way the argument has been put, would the evidential burden on material identity of the schemes. This is because the default position for the realistic counterfactual is the real world, and the court treats conduct as lawful unless and until it is shown to be unlawful. If the Claimants had not asserted the unlawfulness of Visa’s MIFs, on the grounds of material identity, MasterCard would not have needed to address the question and would not have failed in discharging its evidential burden on the ancillary restraint issue merely by failing to adduce evidence of Visa lawfulness.

In these circumstances I do not feel able to hold that the schemes were materially identical. The features which Mr Lowenstein selected as all that were necessary to constitute material identity are by no means all those which might affect the lawful level of Visa MIFs, including in particular those which might have a bearing on Article 101(3) issues. Indeed they did not purport to be, because Mr Lowenstein submitted that the only relevant identity was for Article 101(1) purposes. I might have been more attracted to his forensic point that MasterCard had not sought to identify material differences, irrespective of the burden of proof, if he had been advocating material identity for Article 101(3) purposes.

The Claimants also referred to a few strands of evidence as positively showing material identity, again in the context of their argument that the only identity which was material was that relevant to Article 101(1) issues. In my judgement none of these strands, singly or cumulatively, came close to establishing material identity for Article 101(3) purposes. In particular:

In response to questioning on this issue by the court, Mr Lowenstein did in the course of his final speech submit briefly, and in the alternative, that in any event there was material identity for Article 101(3) purposes because “if the schemes are materially identical for Article 101(1) purposes it is very hard to see how that would be different [for 101(3) purposes]” and “the Article 101(3) issues do not relate to matters which are scheme specific, for example issuer pass-through and the number of always-card against switching transactions will be the same [for both schemes]”.

I am afraid that again I find myself unable to accept the submission. As is apparent from what I say later in this judgment when addressing exempt and exemptible levels of the MIF, the nature and extent of issuer pass-through is an important element in the Article 101(3) analysis. Mr Lowenstein asserted that issuer pass-through to cardholders must be the same for both schemes, but I do not understand why that should be so. The extent of pass-through of the interchange fee to cardholders is likely to be affected by the rewards and benefits attached to particular cards in particular schemes and be scheme specific. MasterCard operated premium credit cards during the period, branded as World or World Signia or World Elite, for which much higher MIFs were charged than for other credit cards (for example the UK MIF for chip and pin credit cards transactions in 2010 was 0.8% for standard cards, 1.3% for World cards and 1.5% for World Signia cards). Whilst issuers had some latitude as to exactly what benefits to offer with such premium cards, MasterCard set the rules concerning what sort of customers could hold such cards, the minimum spend on the cards and the percentage of the spend which had to

In this context Mr Lowenstein relied on a remark in opening by Mr Hoskins, in response to a question of mine, that the data which would be applicable to the exempt or exemptible level of a Visa MIF would be the same as for MasterCard because the MIF data was based on merchant data, not MasterCard or Visa data. It would not be right, however, to treat this as a concession on a point which was not even raised by Mr Lowenstein until final submissions. Mr Hoskins no doubt took that stance because Dr Niels did not make any adjustment to his adjusted MIT MIF to take account of anything less than 100% issuer pass-through, whereas Mr Lowenstein’s case (which I have accepted when considering the Article 101(3) analysis in the later section of my judgment) was that issuer pass-through was less than 100%.

Mr Lowenstein also submitted that a conclusion that MasterCard’s MIFs were capable of being saved by reference to Visa MIFs which might themselves subsequently be held to be unlawful, would entail a risk of inconsistent judgments dictated by an accident of case management, from which the Court should draw back; and one which was all the more abhorrent if Visa were then able to justify the lawfulness of its MIFs before Phillips J on the hypothesis of competition from MasterCard MIFs which had only been held to be lawful on the hypothesis that Visa’s MIFs were lawful.

The Court is always reluctant to reach a conclusion which may give rise to a risk of inconsistent findings. But it is a commonplace in complex commercial litigation that the court will sometimes reach conclusions on the same point as has been or will be dealt with by another court. Sometimes the evidence is not the same and/or the parties are not the same. There is sometimes an unavoidable risk of inconsistent judgments.

The risk of inconsistency in this case does not point in one direction any more than the other. If I were to assume that Visa’s MIFs were unlawful and Phillips J were to find that they were not unlawful, the inconsistency would work against MasterCard.

The risk of inconsistent judgments in this case does not arise from an “accident” of case management. None of the parties to either set of proceedings supported the idea of their being heard together. That did not put these Claimants at an insuperable disadvantage in the current context. They shied away from asking me to make any findings about the lawfulness of Visa’s MIFs and instead sought to make good their case by advancing what they contended were logical assumptions which the court was bound to make. In this trial they could have adduced and relied on the necessary evidence in relation to the Visa scheme and invited me to make findings about the lawful level of Visa MIFs by reference to that evidence, at least so as to establish material identity. That need not have involved the same evidence as that which would be before Phillips J, and no findings of mine would be binding on Visa or the claimants in those proceedings; but nevertheless there was an opportunity for the Claimants to address the issue in evidence and argument before me. The risk of inconsistency to which Mr Lowenstein refers arises not so much out of case management but from the Claimants’ approach to how to prove their case and meet MasterCard’s defences before me.

It may be that Mr Lowenstein’s reluctance to ask me to make any findings about the lawfulness of the Visa MIFs, prompted perhaps by my pressing him during the course of the hearing as to whether I was being asked to make such findings, arose from a concern that it might involve pre-judging the Visa trial and reaching a conclusion on the basis of different evidence and argument. If so, that would itself be a recognition that there are aspects of the evidence about the Visa scheme in that trial which are relevant to the lawfulness of the Visa MIFs and which I cannot assume would produce the same answers for the MasterCard scheme. That is inconsistent with the schemes being materially identical for present purposes.

I should record that in any event I had some difficulty with Mr Lowenstein’s disavowal that he was seeking any finding that the Visa MIFs were unlawful. The argument is dependent on making a finding, not an assumption, that the schemes were materially identical. If they were materially identical, my finding that MasterCard’s MIFs were unlawful would inevitably amount to a finding that the materially identical Visa Scheme’s MIFs were equally unlawful. Mr Lowenstein’s disavowal may be explained by his submission that I had to assume at this stage of the analysis a restriction of competition by MasterCard and all he was asking me to do was assume a similar restriction of competition by Visa; but for the reasons I have tried to explain, that is not sufficient to treat the Visa MIFs in the counterfactual as constrained to the same level as MasterCard’s.

Mr Lowenstein argued in the alternative that if the Visa Scheme were not materially identical, the death spiral argument must still fail. He submitted that there would have to be some aspect of the Visa scheme which prevented its MIFs, taken by themselves, from restricting competition which is missing from the MasterCard scheme; yet if that were the case, MasterCard could have adopted that aspect as part of its own scheme, and so MasterCard’s MIFs will fail the proportionality requirement of objective necessity, as they could be rendered non-restrictive like Visa’s by similar means.

There are three answers to this argument:

Accordingly, in agreement with the conclusion of the CAT, but for different reasons, the death spiral argument falls to be addressed on the hypothesis that in the counterfactual world Visa’s MIFs would have been set at the level which in fact occurred.

The evidence strongly supported the conclusion that if MasterCard had applied a zero MIF throughout the claim period, both Visa’s MIFs and Amex’s merchant fees would have remained at or near to their actual levels. It is the Visa MIFs which are here relevant because although MasterCard contended (correctly in my view) that its demise would have been hastened by competition from Amex, it is not suggested that competition from Amex alone would have caused such extinction unless the competition from Visa was sufficient to do so. The CAT reached a similar conclusion in relation to Visa’s MIFs at paragraph 116. The evidence which supports that conclusion, including that of past behaviour of both MasterCard and Visa in the face of MIF changes by the other, was set out at length in MasterCard’s written closing submissions; it is not necessary to recite it in this judgment.

The real battleground on this issue was whether, in those circumstances, the MasterCard scheme would have survived. A few points of principle need to be addressed before considering the evidence:

It is instructive to look first at the sums of money involved for issuers in a differential between MasterCard zero MIFs and Visa actual MIFs. The CAT took Visa’s weighted average credit card MIFs over its claim period to be 0.8%. In fact Visa’s average was higher than this throughout almost all the claim period, between 2006 and 2014, with the average being brought down by the lower rates in 2015-2016 under the IF Regulation. I had evidence of a figure for 2012 of about 0.84%. If one were to take 0.84% as illustrative, it would reflect a difference of around £441 million per annum to all issuers in 2007, rising annually over the period (£616 million, for example, in 2011). For debit cards, taking the evidence before me of a weighted average for Visa of 0.3% in 2012, the difference would be an annual figure of £226 million for all issuers. These are annual figures; an issuer would look prospectively at the cumulative effect on revenue over the whole period in deciding whether to switch to Visa.

Looking at it from the perspective of individual issuers, there can be little doubt that the economic effect on the larger issuers would give rise to a very large incentive to switch. In 2007 HSBC’s gross dollar value of UK MasterCard transactions was about $44 billion for credit cards and $72 billion for debit cards. At 0.84% and 0.3% respectively, that would represent about $370 million and $216 million respectively in lost interchange fees if MasterCard’s MIFs were zero compared with Visa. That is for a single year. The effect over time would run into several billions of dollars. Even a relatively small issuer would suffer a significant loss of revenue. Conister Bank Ltd, for example, one of the smallest issuers in 2007, with gross dollar value of $20.9 million for all cards, would suffer an annual loss of over $100,000 if one assumed for these purposes a weighted average for Visa across both types of card of 0.5%. Again the cumulative effect of an annual figure of this size over the claim period would be substantial.

The cost to issuers of switching would involve (a) the administrative costs of replacing customers’ cards and (b) any additional costs in acquiring and running a processing platform for Visa cards. As to the former, switching might take the form of immediate replacement of all MasterCard cards amongst customers; or what was described as “forward flow”, namely switching at the natural expiry of the cards in the system, which would occur over a period of two to three years. The costs of forward flowing to Visa cards would likely not be significantly greater than those of reissuing MasterCard cards for those issuers already using Visa’s processing platform. Most MasterCard issuers are likely to fall into this last category of already having Visa processing platforms: so far as credit cards are concerned, it was Mr Douglas’ unchallenged evidence that most issuers issued both Visa and MasterCard credit cards during the claim period; very few were exclusive to one or the other, and most were already familiar with Visa cards and operationally set up to issue them to take advantage of the best offer from each for credit products. Moreover all major UK issuers except Barclaycard process credit transactions on outsourced processing platforms such as Tsys and First Data, which would make switching between schemes straightforward from the issuer’s perspective. As to debit cards, the majority of retail bank issuers issuing MasterCard credit cards are likely already to have been issuing Visa debit cards to its customers from 2010 or so, because of Visa’s dominance of the debit card market following the collapse of Maestro. The switching costs to issuers, particularly if by forward flowing, are therefore unlikely to have formed a serious deterrent to switching if the price differential justified it. The collapse of Maestro between 2009 and 2011, which I consider further below, shows that issuers were prepared to bear the costs of switching, whether the switch was on that occasion motivated by a price differential or differences in functionality or both.

The evidence of the MasterCard witnesses, in particular Mr Titarelli, Mr Lane, Mr Douglas, and Mr Willaert, was that if the UK MIF had been reduced to zero, MasterCard would have lost its entire UK business. Although they were speaking from their experience of dealing with issuers rather than of being issuers (with the exception of Mr Douglas who had himself worked in UK retail banking before joining MasterCard in 2005) their experience made them well placed to judge issuers’ behaviour. Their evidence was convincing on this point. By contrast, the Claimants did not seek to adduce evidence on the point from any issuer or any witness with relevant experience of issuers.

The experts approached it from an economic perspective. Both experts believed that a material differential between Visa and MasterCard MIFs could lead to very substantial switching, sufficient to give rise to market exit.

Dr Niels opined that “If MasterCard’s domestic MIFs were materially lower than Visa’s, MasterCard would be likely to lose all or most of its market share, as issuers and/or consumers would switch towards Visa (and Amex), as happened in the UK debit card market.” Mr Dryden was more qualified, stating in his second report that “…I note that if the Court was to consider that the prohibition on ex-post pricing counterfactual should be assessed assuming Visa is unconstrained, then I consider that it is likely that this could have a substantial commercial impact on MasterCard, and cannot exclude the possibility that it might be driven from the market over some period of time.” When questioned by the Court, Mr Dryden accepted there was a “very high probability” of a “very substantial shift” in these circumstances and it was a “possibility” that MasterCard would be driven out of the market.

Mr Dryden returned to the topic in his sixth report when addressing Dr Niels’ conclusion that the differential of 0.3% posited by the CAT Bilaterals counterfactual would be sufficient for collapse, saying that “while Dr Niels may be correct in his conclusion…I do not think there is sufficient information to know that he is correct.” He identified four issues on which he thought there was insufficient evidence, which are potentially relevant for the differential in the zero MIF counterfactual. The first was that “it is not clear why some issuers (especially those with lower spending customer bases) might not keep some of their customers (the lower spenders) on MasterCard at least for a period of time.” Mr Dryden does not suggest that this point would prevent collapse, merely that it might delay it “for a period of time”. Given the counterfactual assumes a substantial differential between the MasterCard MIF and the Visa MIF over the whole period of the claim (around 10 years), it is clear that collapse would occur over this period. Mr Willaert convincingly explained, based on his experience of actual migrations between credit card schemes, that the cost of flipping lower spenders would not prevent this from happening. Banks did not in practice migrate only part of a portfolio: if they decide to switch schemes they migrate all their customers.

Mr Dryden’s second point was that “there would be issuers with very low spending customer bases who might fall far enough below the average spend not to [switch] at least for a period of time.” Again, Mr Dryden did not suggest that this would prevent a collapse; merely that it might delay it. The size of the sums involved indicate that for even the smaller issuers the migration would make financial sense. If there were a rump of one or two issuers, merchants would be unlikely to continue to accept such cards: this is the “spiral” element in the death spiral argument. In any event, if all that remained were one or two issuers issuing a very small number of cards, it is not realistic to think that the scheme in its current recognisable form would have been maintained by MasterCard for the benefit of them or their cardholders. If anything were to be salvaged of MasterCard’s UK business it would have to be some different model.

Mr Dryden’s third point was that that “various aspects of product differentiation (i.e., the fact that the MasterCard and Visa products are not identical) could be a relevant factor that could lead (at least some) issuers to still issue cards of a scheme with lower interchange fees (at least [to] some cardholders.” This is unspecific and speculative, and as Mr Dryden fairly accepted in his fifth report, the existence and materiality of such distinctions was outside his expertise.

Fourth, Mr Dryden suggested that brand loyalty on the part of some customers to MasterCard would mean that banks would continue to offer MasterCard cards to those customers. This theoretical possibility was ruled out as a practical or realistic one by Mr Willaert’s evidence that whilst issuers will obviously take into account cardholders’ preferences for brand and particular product features, if they stand to make a lot more money from one card this will inevitably dictate their decision and override cardholder preferences for brand/product features. On the basis of his experience from working with issuers and having gone through many migrations from Visa to MasterCard, the number of customers who might wish to stay with a particular scheme out of brand loyalty would be “very, very limited”. This limited number of such customers would not prevent a complete migration from one scheme to the other because of the expense and work involved in maintaining a scheme just for those few customers.

In cross examination Mr Dryden raised a further point, suggesting that issuers would wish to maintain a second four-party scheme in the market so as “to maintain a degree of leverage in their dealing with the other card scheme”. This was speculation outside the scope of his expertise, and did not address the countervailing considerations which would affect issuers, including that this would involve collective action by issuers and that the alternative four party scheme would have to be “maintained” by the issuers receiving no MIF. It is not supported by the evidence of those with experience of issuer behaviour.

I conclude, therefore, that in the zero MIF counterfactual world with Visa MIFs unconstrained, the MasterCard scheme would not have survived in the UK in a materially and recognisably similar form. On that hypothesis:

It seems that the CAT was of the same opinion: it is because a zero MIF world would have threatened the very existence of the MasterCard scheme that they concluded that bilaterals would have emerged to prevent a Visa monopoly: see paragraph 174 and the paragraphs dealing with the CAT Bilaterals counterfactual summarised above.

I have reached this conclusion without the need to address the effect of competition from Amex, but it is clear that such additional competition would have contributed further to the inevitable demise.

I have also reached this conclusion without taking into account the fate of the Maestro scheme in the UK, which formed the subject matter of a good deal of disputed evidence and argument at the trial. MasterCard submitted that the demise was mainly due to the interchange fee differential between the Maestro MIF (which was not then set by MasterCard) and the Visa Debit MIF, and relied upon the dramatic loss of market share as demonstrating that even a modest difference in interchange fees leads to extensive switching. The Claimants, on the other hand, submitted that the switch was primarily driven by differences in functionality and factors other than the price differential, and accordingly provided no relevant guidance in the context of the current issue.

The differential between Maestro and Visa Debit increased to 4.3 pence per transaction with effect from 1 January 2007. This is equivalent to about 0.09%. Mr Willaert explained that this differential was worth around £70 million per annum to Maestro issuers. In April 2007 MasterCard launched its own Debit Card at an ad valorem rate which was equivalent to a little more than Visa’s debit fixed rate; the issuers were resistant to the ad valorem element and it was relaunched a few months later with a fixed price MIF very similar to Visa’s.

In the last quarter of 2007 both HSBC and RBS announced their decision to transfer their debit business from Maestro to Visa Debit. Between them they represented about 90% of Maestro’s business. By 2009 MasterCard had gained control over Maestro pricing, and the MIFs for both Maestro and Debit MasterCard were set at a level which was competitive with Visa Debit, but by that stage the switching had taken hold and its effects were being felt. MasterCard’s share of the UK debit card market declined from 2008 dropping to only to 3% by 2011, whilst Visa’s share of the UK debit card market increased to 97%, as shown by the graph below.

Mr Willaert’s evidence was that without the competitively priced Debit MasterCard alternative, it was “inevitable” that MasterCard market share would have fallen to zero. Mr Dryden agreed that if it had not been priced competitively, then MasterCard would not even have retained 3% of the market. Mr Dryden also said “it would seem to me entirely plausible that [Visa’s higher interchange fees] was a very significant contributor, perhaps the dominant contributor to [this] outcome”; and agreed in cross examination that given that the value of switching was £70 million in one year, it was no surprise that Maestro's market share collapsed thereafter because it was economic sense for the issuers to switch at that level of differential.

Each of the factual witnesses from MasterCard who addressed this topic gave evidence that the switching from Maestro to Visa Debit was primarily due to the interchange fee differential:

There was extensive cross examination of these MasterCard witnesses by reference to contemporaneous documents, to the effect that other factors were responsible for the switch, including in particular a document headed “UK Debit strategy” dated January 14 2008 and “slide 33” from a set of slides dated 31 August 2009.

The 2008 document stated that the HSBC/RBS switch was the result of a combination of factors. It identified the fee differential as one of two economic factors affecting the decision, the other being the possibility of incentives given by Visa to the banks in the impending Visa IPO. As a matter of “Product offer”, Visa Debit was more attractive because it was lower risk as an established product compared with Debit MasterCard; and the lower international acceptance of Maestro outside Europe, particularly in the US put it at a competitive disadvantage in the UK market. There was also identified a regulatory motive for the banks, who were both prominent participants in S2 Card Services which set the Maestro MIF, in joining Visa debit as an established product rather than switching to another MasterCard product with a higher interchange fee.

Slide 33 was headed “Reasons for the portfolio losses in Debit”. Its headline summary was “Multiple factors influenced HSBC and RBS’s decisions to migrate their debit portfolios from [MasterCard], notably overall economics, acceptance and marketing.”

As Mr Willaert correctly observed, it is important to distinguish between (a) reasons why the banks chose to abandon Maestro and (b) reasons why they chose Visa Debit over Debit MasterCard; but the distinction is not a clean one because MasterCard had launched MasterCard Debit to sit alongside Maestro compared with a single Visa offering. Focusing on those which were relevant to a switch away from Maestro, rather than being specific to the Debit MasterCard as a potential replacement, under the three epithets identified as “notable” in the slide 33 headline, the factors were:

In relation to acceptance, Mr Willaert gave three reasons why wider overseas acceptance of Visa Debit than Maestro was not a major factor namely (a) 98% of the debit volume at that time was domestic (b) Maestro was accepted overseas “where it mattered” (for example in large merchants in New York) and (c) if lower global acceptance were significant, it would be expected that Maestro would have suffered in other European markets, but that has not been the case; in Germany and the Netherlands, for example, Maestro is still one of the main debit cards issued by banks. This last point is a powerful one: where there is no price differential in Netherlands and Germany, Maestro has held its market share, suggesting that the price differential was the single most important factor in the UK. The detail of this evidence was not challenged, but I bear in mind that Mr Willaert was not at MasterCard at the time of the switch and that the acceptance deficiencies of Maestro were recorded as one of the relatively few contributing factors in both the contemporaneous documents identified above, suggesting that they were regarded as a significant factor. This impression is reinforced by a later UK Debit Strategy document of 24 October 2011 in which overseas acceptance was described as “the main weakness” of Maestro versus Visa debit and Debit MasterCard (although the document recorded that there had been some decline in acceptance over the intervening years based on a belief that Maestro was being phased out). Limited international acceptance was, in my judgement a significant factor in the decision to switch away from Maestro.

The documents suggest that other factors were also considered significant, including the inferior marketing of Maestro, the influence of other Visa financial incentives to the banks, and the downside of the MasterCard strategy of seeking to maintain both the Maestro and MasterCard brand, with a recognised difficulty in communicating a dual brand strategy to consumers.

In the CAT, the evidence relating to the Maestro decline included the same documents, and evidence from some of the same witnesses as before me. The CAT considered the evidence at paragraphs 251 to 258 and reached the conclusion at paragraph 259 that the interchange fee differential was not the main cause of Maestro’s decline but merely one of a number of factors which led to switching. Its reasoning is contained in the following passages:

“258.

We find that:

(1)

Unsurprisingly, sophisticated Issuing Banks like HSBC and RBS do not look only at one factor. The Interchange Fee they are likely to be paid will, obviously, be a significant factor – but, at the end of the day, it is just one consideration amongst many.

(2)

In this case, the significance of the Interchange Fee differential between MasterCard’s offering and that of Visa was complicated by two additional factors:

(i)

MasterCard’s offering was based on two products (Maestro and Debit MasterCard), with different Interchange Fee rates. As we have noted, the difference between Maestro’s Interchange Fee and that of Visa was significant (see paragraph 250 above). But – as the MIF rates in paragraph 206 above demonstrate – the MIF for Debit MasterCard was significantly higher than that for Maestro, and so correspondingly closer to the Visa debit MIF. Although we appreciate that Table 8 above sets out the “blended” MIF for all relevant cards over the entire claim period (and not just for the time when MasterCard was engaged in the procurement process with HSBC and RBS), the figures are nevertheless instructive. The blended MIFs paid by Sainsbury’s over the claim period were:

(a)

0.36% for all Debit MasterCard transactions.

(b)

0.19% for all Maestro transactions.

(c)

0.26% for all Visa debit card transactions.

Slide 33 shows that MasterCard’s offering to HSBC and RBS involved a potentially higher MIF for Debit

MasterCard transactions than offered in the case of Maestro transactions. The problem for MasterCard was that the “mixed” ad valorem plus flat fee rate it was proposing was meeting with significant Merchant push-back, doubtless because Merchants disliked the ad valorem element.

(ii)

The Interchange Fee was not the only financial consideration in favour of Visa Debit: it is clear that Visa put forward other financial incentives to tempt RBS and HSBC to transfer their portfolios.

(3)

Slide 33 demonstrates a whole range of other factors that MasterCard considered would have influenced RBS and HSBC to go to Visa rather than stay with MasterCard.

(4)

Cross-examination also established that Maestro (though not Debit MasterCard) suffered a number of shortcomings compared with Visa Debit. Mr Douglas, Mr Perez and Mr Willaert all accepted that Maestro had a limited international acceptance, particularly in the USA. ¹⁷⁶ MasterCard had approximately 23 million locations worldwide, whilst Maestro had just 10 million locations. ¹⁷⁷ Despite these witnesses’ explanations that international spending represented a relatively small share of spending by cardholders, and the suggestion that Maestro found acceptance “where it mattered” (i.e. in airports and major cities), we nevertheless consider that this was a major contributory factor which led to the decision of HSBC and RBS to reject Maestro in favour of Visa Debit.

(5)

Maestro also suffered from a number of other less significant shortcomings compared with Visa Debit. For example, Mr Douglas conceded that Maestro was less suitable for use in online transactions (a mode of distribution which was beginning to gain importance in 2007). The documentary evidence also indicates that the Maestro card was less able to process recurring payments, which would have reduced its convenience for some Cardholders.

Conclusion on the Maestro example

259.

The evidence relating to Maestro was described by Mr Hoskins in opening as the "prize evidence" in the case. We find the contention that it was the Interchange Fee differential between Visa and Maestro that was the main cause of the collapse of the Maestro market in the UK to be unsubstantiated by the facts, and we reject it. Helpful although we found the evidence of MasterCard's witnesses in general, we do not accept their evidence on this point, preferring the analysis in MasterCard's own contemporaneous documents. We find that MasterCard has established no more than that Maestro's lower Interchange Fee was one of a number of factors which led to the decision of HSBC and RBS to reject MasterCard's bid in the procurement processes. Taking the evidence as a whole, we consider that the combined Maestro/Debit MasterCard offering was unattractive to these banks compared with the simplicity of the well-established Visa Debit offering.”

The Tribunal referred at paragraph 258(2)(i) to the average interchange fees which Sainsbury’s had paid for different types of debit cards over the claim period: an average of 0.36% for Debit MasterCard, 0.19% for all Maestro transactions and 0.26% for Visa Debit. While the Tribunal noted these were averages over the entire claim period (2006-2015), it describes them as “instructive” and appears to have treated them as showing that banks had chosen Visa Debit over Debit MasterCard despite Debit MasterCard having higher interchange fees. If so, the evidence before me shows that the Tribunal was mistaken. The average interchange fee figures used by the Tribunal were produced by the Tribunal itself after the trial and were not, therefore, the subject of evidence or submissions. As a result, the Tribunal appears to have been unaware that, as explained by Mr Willaert in his evidence to me, in July/August 2007 the average interchange fee for Debit MasterCard was set at an almost identical level to that for Visa Debit, and as can be seen from a MasterCard 2011 analysis of the debit card market, the average interchange fee for Debit MasterCard remained close to the same level as Visa Debit until January 2013. It was only after January 2013 that Debit MasterCard began to offer materially higher rates than Visa Debit. It is this which results in the average interchange fees quoted by the Tribunal being higher for Debit MasterCard than Visa Debit. Looking at a single average for the entire claim period, therefore, concealed the fact that Debit MasterCard and Visa Debit had comparable rates from 2007 to 2013. It was not, therefore, the case, that the HSBC/RBS decision to move to Visa Debit rather than to Debit MasterCard was despite Visa Debit having lower interchange fees. When the decision to migrate was made at the end of 2007 (and for the following five years), Visa Debit and Debit MasterCard had close to identical rates. Therefore, all this shows is that, having decided to migrate from Maestro (with its low interchange fees), faced with two cards with almost identical interchange fees, RBS and HSBC chose to migrate to the established Visa Debit product rather than Debit MasterCard for reasons other than a MIF differential. This does not provide any indication that the lower interchange fees offered by Maestro were not an important factor in the decision to move from Maestro.

My conclusion on the evidence before me is that Maestro’s uncompetitive interchange fee pricing was a very significant factor in the issuers’ decision to switch away, quite possibly the single most influential one, whilst not being the only significant one. This is in line with Mr Dryden’s evidence that it was “entirely plausible that [the interchange fee differential] was a very significant contributor, perhaps the dominant contributor to [this] outcome.” He also said that given that the value of switching to the issuers as a whole was about £70 million, it was no surprise that Maestro’s market share collapsed at that level of differential.

This is similar to the conclusion reached by the CAT but perhaps with a difference of emphasis which accords the fee differential more influence in the outcome.

The Maestro experience lends support to my conclusion that for a zero MIF counterfactual, wholesale switching would have taken place so as to lead to the collapse of the MasterCard scheme. In the counterfactual I am considering, the differential would be much greater than the Maestro differential: about 9 times greater for the credit card MIF and three times greater for the debit card MIF. The issuer loss would be very much greater than the £70 million per annum which Mr Dryden thought plausibly explained the Maestro switching. The Maestro experience supports the conclusion that the much larger differential would be fatal.

Although I did not have equivalent data for issuers’ revenues for Ireland, the other considerations in the above analysis hold good for Ireland as for the UK and my conclusions are the same (for the period from 20 January 2009 which is in issue). They are supported by the evidence of Mr Douglas who addressed the issue in relation to Ireland and explained how the Irish domestic debit card scheme, Laser, suffered from the higher interchange fees of Visa Debit. I accept his evidence that unless MasterCard had had MIFs at a competitive level with Visa in Ireland it would have expected to lose its entire business. It would have done so in a world in which the MasterCard MIF was zero and Visa unconstrained.

I conclude, therefore, that in the zero MIF counterfactual world with Visa MIFs unconstrained, the MasterCard scheme would not have survived in the UK or Ireland in a materially and recognisably similar form. On that counterfactual hypothesis:

There are potentially two remaining questions: (1) what is the putatively lawful (lower) level of MIF; and (2) would the MasterCard scheme have collapsed in a counterfactual world with such maximum lawful MIFs and Visa MIFs unconstrained.

I have answered the first question in my analysis of the Article 101(3) issues later in this judgment, in which I conclude that the MIFs as set were below the exempt and exemptible level, save for the EEA debit card MIF for the earliest part of the claim period prior to June 2008. In other words there is no putatively lawful level capped below the level at which MasterCard MIFs were actually set. It follows that there is no putatively lawful lower level of MIF than that actually set with which to populate the alternative counterfactual and the death spiral argument does not arise in that context.

In case I am wrong in that conclusion, or in my rejection of the CAT Bilaterals counterfactual, I propose to answer the second question by identifying the level of interchange fee differential which marks the dividing line between extinction and survival for the MasterCard scheme.

This is obviously not an exercise which permits mathematical precision, but my assessment is that a differential of 0.2%, would be sufficient to cause issuers to switch to Visa so as to lead to the collapse of the MasterCard scheme for UK and Ireland, for the following reasons:

This conclusion is not necessarily inconsistent with the CAT Judgment. The CAT reached the conclusion that if bilateral agreements at an interchange fee of 0.5% emerged, the differential between this weighted average and Visa’s MIFs expressed as a weighted average would not be sufficient to lead to a MasterCard scheme collapse: see CAT Judgment at paragraph 261. The CAT took Visa’s weighted average to be 0.8% although that would be an underestimate because for the whole period other than 2015-2016 Visa’ MIF was higher than 0.8% with the average being brought down by the lower rates set in 2015-2016 under the IF Regulation; for example it was about 0.84% for 2012. But in any event the CAT’s reasoning was not simply that the difference was between 0.5% and 0.8% and that such a difference of 0.3% was insufficient for wholesale switching. On the contrary the conclusion was based on

This reasoning can only apply in the Bilaterals counterfactual being considered by the Tribunal, involving innovative pricing structures in response to the abandonment by MasterCard of its process of setting a MIF and the emergence of bilateral agreements. That is not the counterfactual I am considering, having rejected the CAT Bilaterals counterfactual. In the counterfactual I am considering MasterCard would still be setting the MIF in just the same way as now, albeit at a lower putatively lawful level, and there would be no restructuring of the market in accordance with different and innovative pricing structures.

The CAT also supported its conclusion by holding that an interchange fee at 0.5% would be enough to enable issuers to cover their costs and make a profit, so that it would not have led them to “drift away from the scheme”: paragraph 261(7)(iii). However this does not address the effect of the differential: as Mr Douglas confirmed, if issuers would make more profit from the rival scheme, that is what causes the switching.

The rationale of Article 101(3) is expressed in paragraph 33 of the Article 101(3) Guidelines in the following terms:

“33.

The aim of the Community competition rules is to protect competition on the market as a means of enhancing consumer welfare and of ensuring an efficient allocation of resources. Agreements that restrict competition may at the same time have pro-competitive effects by way of efficiency gains. Efficiencies may create additional value by lowering the cost of producing an output, improving the quality of the product or creating a new product. When the pro-competitive effects of an agreement outweigh its anti-competitive effects the agreement is on balance pro-competitive and compatible with the objectives of the Community competition rules. The net effect of such agreements is to promote the very essence of the competitive process, namely to win customers by offering better products or better prices than those offered by rivals. This analytical framework is reflected in Article 81(1) and Article 81(3). The latter provision expressly acknowledges that restrictive agreements may generate objective economic benefits so as to outweigh the negative effects of the restriction of competition.”

In order to qualify for exemption under Article 101(3), an anti-competitive restriction must meet four cumulative conditions. There is an issue as to three of them in their application to the MasterCard MIFs, which are that to qualify for exemption, any given MIF must:

A number of points of principle arise.

The relevant benefits must be causally linked to the MIF; it is not sufficient to identify benefits which result from the use of cards generally or the MasterCard scheme generally. Only benefits specifically produced by the MIF can be taken into account. This is because it is the particular restrictive agreement itself which requires justification for exemption under Article 101(3); and if it has not escaped Article 101(1) altogether as a necessary ancillary restraint, it is the pro-competitive benefits of that restriction which fall to be analysed and weighed against its anti-competitive restrictive effect. This is well established: see the Article 101(3) Guidelines paragraphs 50-51, the MasterCard General Court Judgment at paragraph 207 and the MasterCard CJEU Judgment at paragraph 232.

The causal link must also be sufficiently direct to be capable of proof. The Article 101(3) Guidelines explain:

“54.

The causal link between the agreement and the claimed efficiencies must normally also be direct. Claims based on indirect effects are as a general rule too uncertain and too remote to be taken into account. A direct causal link exists for instance where a technology transfer agreement allows the licensees to produce new or improved products or a distribution agreement allows products to be distributed at lower cost or valuable services to be produced. An example of indirect effect would be a case where it is claimed that a restrictive agreement allows the undertakings concerned to increase their profits, enabling them to invest more in research and development to the ultimate benefit of consumers. While there may be a link between profitability and research and development, this link is generally not sufficiently direct to be taken into account in the context of Article 81(3).”

Economists draw a distinction between “Total User Surplus (TUS)”, sometimes also referred to as consumer welfare, and “Total Welfare (TW)”. The Claimants contended that TUS was the relevant measure in this case; MasterCard contended that TW was the relevant measure. The essential distinction for the purposes of this case is that TUS benefits would be limited to those conferred on consumers, namely cardholders and merchants; whereas TW benefits would potentially include those conferred on issuers, acquirers and MasterCard itself, including their profits, as well as any wider benefits to others from the MasterCard scheme.

In Case T-86/95 Compagnie Générale Maritime Belge v Commission paragraph 343, the CJEU held that regard should be had to the advantages arising from the agreement in question, not only for the relevant market, but also, in appropriate cases, for every other market on which the agreement in question might have beneficial effects, and even, in a more general sense, for any technical or economic progress not specifically linked to the relevant market. Other case law of the CJEU has made clear that the objectives of Article 101 extend to the public interest, the interests of individual undertakings, and the economic well-being of the European Union: see in particular Hoechst AG v Commission Case C-46/87 paragraph 25; TeliaSonera Case C-52/09 at paragraph 22; Roquette Frères SA Case C-94/00 at paragraph 42; and TMobile Netherlands, Case C-8/08 at paragraph 38.

The wording of Article 101(3) recognises that the relevant benefits may be wider than those enjoyed by consumers. It provides for the exemption of a restriction “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”, which necessarily implies that “the resulting benefit” is a wider concept than the consumer benefit which must constitute a fair share of it, i.e. a part of a greater whole.

The Article 101(3) Guidelines support this approach: see in particular paragraphs 13, 33, and 84-85.

In the MasterCard CJEU judgment, the Court held that in the case of a two-sided market such as the MasterCard scheme, the relevant consumers are not only those subject to the restriction, but the other group of consumers associated with the system: see paragraph 237.

In the CAT Judgment, the Tribunal made observations at paragraph 289 with which I agree:

“(1)

We consider that an “exemptible” MIF must, as a starting point, reflect the fact that it represents a price in three markets: it is the price paid by Merchants/Acquiring Banks in the acquiring market; it is the price received by Issuing Banks in the issuing market; and it is a price that affects inter-scheme competition, in that the relative Interchange Fees between schemes (e.g. as between MasterCard and Visa) will be a factor (but no more than that) bearing on Issuing Banks as to what card types they issue.

(2)

In other words, the MIF acts as a “pivot” or the “see-saw” between various markets. For this reason, a MIF that is calibrated by reference to one market only, ignoring the others, is fundamentally inappropriate. This approach is consistent with the Court of Justice’s dicta at paragraphs 237-242 of its MasterCard judgment. …”

The benefits requirement is not, therefore, confined to benefits arising solely on the acquiring market. Subject to the important proviso that the benefits must have a direct causal link with the MasterCard MIFs (not just the scheme), it may take into account (i) benefits to all consumers on both sides of the market, merchants and cardholders; and (ii) benefits conferred on those using the Visa system insofar as the MasterCard MIF can be said to be a direct cause of such benefits through inter-system competition; and (iii) wider benefits by way of technical and economic progress.

It is not the Claimants’ case that TUS is alwaysthe full extent of the relevant measure for Art 101(3) purposes. They submitted that the measure depends on the relevant benefits in the particular case, and that in this case TUS is the relevant measure because:

The Claimants submitted that both aspects are clear from the MasterCard Commission Decision and General Court and CJEU Judgments. As to the exclusion of benefits conferred on those who are party to the restrictive agreement, this is made clear by the terms of the Commission Decision at paragraph 693:

“[A]ny transfer which results from a restriction of competition must be more than compensated by benefits for somebody else other than the undertakings participating in the restrictive arrangements. The increase of sales volumes in MasterCard’s scheme is clearly to the advantage of MasterCard’s member banks [ie qua issuers, as they then were]. An increase of system output only contributes to appreciable objective advantages if parties other than the organisation’s member banks benefit from it.”

This is also made clear in the MasterCard General Court Judgment at paragraphs 206 and 221 approving this aspect of the Commission Decision and supported by the CJEU authorities there cited: Case T-65/98 Van Den Bergh Foods v Commission

In my judgement the Claimants’ submission on this point is sound. MasterCard is correct in its general approach that consumer welfare does not form the totality of welfare benefits which are potentially relevant, as the wording of Article 101(3) makes clear, and as indeed the Claimants concede: see the authorities considered above. However benefits to the parties to the restrictive agreement are excluded. This follows as a matter of principle as well as being supported by the authority cited. It is difficult to see why the profits made by parties to an anti-competitive agreement should be included within the pool of relevant pro-competitive benefits which fall to be taken into account in determining whether they outweigh the anti-competitive effects on those adversely affected by the restriction. If the agreement in question is an anti-competitive restriction, the profits thereby made by the parties cannot easily be categorised as pro-competitive benefits or “efficiencies” in order to undertake the balancing exercise which forms the rationale behind Article 101(3). Moreover, if such profits were relevant benefits, it would have the surprising result that the greater the profit from the restrictive agreement, the easier it would be to establish exemption on the grounds that such profits were benefits which outweighed the anti-competitive effects. Of course what is done with revenues received by the parties to the restrictive agreement is capable of conferring relevant benefits on others which fall to be taken into account; but any benefit retained in the hands of a party to the anti-competitive restriction is not a relevant benefit for Article 101(3) purposes.

As to the second aspect, the Commission, General Court and CJEU rejected MasterCard’s reliance on the benefits of the system output of the scheme as a whole as sufficient to establish exemption: see the Commission Decision at paragraphs 674, 688-693, the General Court Judgment at paragraphs 207, 220, 221, 226ff and the CJEU Judgment at paragraph 234. There is no room for consideration in this case of some more general social benefits beyond those to the MasterCard merchants and cardholders. This was the system output argument rejected in the Commission proceedings, and generally the problems of identification, and of the need for a direct causal link to the MIF, have no more been overcome before me in relation to any of the MIFs with which I am concerned than they were in those proceedings in respect of the intra EEA MIFs. For example, the wider social costs of cash, such as tax evasion or the cost of printing money, which are reduced by card use, are to a significant extent charged by banks to users through bank charges. There is no element which can be quantified as being a direct result of the MIF.

Accordingly the relevant benefits to be considered in this case are those conferred on merchants and cardholders only.

Paragraph 85 of the Article 101(3) Guidelines states:

“The concept of ‘fair share’ implies that the pass-on of benefits must at least compensate consumers for any actual or likely negative impact caused to them by the restriction of competition found under Article 81(1). In line with the overall objective of Article 81 to prevent anti-competitive agreements, the net effect of the agreement must at least be neutral from the point of view of those consumers directly or likely affected by the agreement. If such consumers are worse off following the agreement, the second condition of Article 81(3) is not fulfilled. The positive effects of an agreement must be balanced against and compensate for its negative effects on consumers. When that is the case consumers are not harmed by the agreement. Moreover, society as a whole benefits where the efficiencies lead either to fewer resources being used to produce the output consumed or to the production of more valuable products and thus to a more efficient allocation of resources.”

This invites a comparison between (1) the disadvantages to the MasterCard merchants of paying the MIF and (2) the benefits conferred by the MIF on all those who are relevant consumers for the purposes of the benefits requirement. The former must be balanced by the merchants getting a fair share of the latter. This necessarily involves identifying (3) what benefits the MasterCard merchants are getting from the MIF and (4) what share is to be regarded as a “fair” share.

As to (4), a fair share for the merchants must not leave them worse off as a result of the restrictive agreement; in other words the fair share requirement will not be met unless the MasterCard merchants obtain greater benefits from the MIF than the anticompetitive disadvantage it imposes on them: “the net effect of the agreement must at least be neutral from the point of view of those consumers directly or likely affected by the agreement. If such consumers are worse off following the agreement, the second condition of Article 81(3) is not fulfilled”: paragraph 85 sup.

This is a minimum threshold for meeting the fair share requirement, but it is not necessarily the totality of the inquiry. The burden borne by merchants in paying the MIF must generate for them a fair share of all the relevant benefits conferred on others as well as themselves. If merchants receive benefits which match the restrictive burden on them, in theory they may still be receiving less than their fair share of the overall benefits. That is why the inquiry at the benefits requirement stage encompasses relevant benefits to others, and the inquiry at the fair share stage is as to a fair share of those overall benefits.

There are two classes of beneficiary of the MIF other than merchants who fall to be considered. The first is the issuers. Insofar as issuers receive profits as a result of the MIF, that may affect whether the merchants are getting a fair share of the benefits generated by the MIF. If the MIF creates an unfair degree of profit for issuers, merchants will not be getting their fair share of the benefits conferred by the payment of the MIF, even if they are receiving benefits which match the burden imposed by their paying the MIF.

The second category of beneficiary is the cardholders. Cardholder benefits fall into two categories:

These cardholder benefits are unlikely to affect the fair share question when it is kept in mind that it is only benefits directly caused by the MIF which are relevant. Insofar as they are benefits to cardholders as a result of the effect of the MIF on system output, the effect similarly benefits merchants by increasing the merchant benefits. Insofar as they are benefits to cardholders in the way of lower costs or rewards (eg airline flight points): if there is no issuer pass through of the MIF in the form of lower costs or rewards, they are not benefits to cardholders directly attributable to the MIF; and insofar as there is issuer pass through of the MIF to confer such benefits on cardholders, it would not be such as to render the merchants’ benefits from the MIF an unfair share of the overall benefits provided that the threshold requirement is satisfied and merchant benefits at least matched the MIF: if the merchants are paying no more than the value of the benefit they receive from the MIF, I would not regard it as unfair that cardholders may also benefit to some extent from the MIF. There is nothing in the evidence to suggest that in those circumstances any additional benefit to cardholders would be disproportionately unfair.

I should record that MasterCard submitted that the case law leaves open the issue of whether a global approach can instead be taken, assessing all benefits (enjoyed by cardholders and merchants) against all detriments. MasterCard’s submission was that if it were necessary to decide the issue (which it did not invite the court to do), there is no requirement to establish that the benefits of the MIF to merchants exceed the disadvantage to merchants; it is sufficient that the benefits to all relevant consumers exceed the disadvantages to merchants. This was said to be expressly stated in paragraph 241 of the MasterCard CJEU and to follow as a matter of logic that if it were necessary to prove that the benefits to merchants exceeded the disadvantages to them, there would be no need to consider any other relevant market. Had I been asked to decide this point I would have rejected it. Paragraph 85 of the Article 101(3) Guidelines makes clear that it is a minimum condition of fulfilment of the fair share requirement that the net effect on the consumers who are subject to the restriction must be positive. The passage referred to in the MasterCard CJEU Judgment is addressed to the benefit requirement, not the fair share requirement. The argument from logic does not hold good, because as I have explained, net compensatory effect on the consumers affected by the restriction is a necessary condition for fulfilment of the fair share requirement, but not, as a matter of principle, a sufficient condition. As a matter of principle, if merchants were to bear an unfair proportion of the cost compared with others on whom the restriction conferred relevant benefits, they would not be getting a fair share of the benefits. As a matter of logic, therefore, it is, or at least may be, necessary to quantify the benefits to others who are beneficiaries of relevant benefits in accordance with the benefits requirement. However for the reasons I have identified it is not necessary to do so in the particular circumstances of this case.

Accordingly in the particular circumstances of this case I regard the only relevant questions under the fair share requirement to be (1) whether the benefits to MasterCard merchants from a MIF at any given level (if any) match the cost to those merchants of the MIF at that level and (2) whether the MIF at that level generates unduly high profits for issuers so that merchants are not getting a fair share of the benefits generated by a MIF which matches the value of the benefit to them conferred by the MIF.

When assessing the benefits conferred on merchants by the MIF it is the position of MasterCard merchants as a whole which must be considered, not individual merchants or groups of merchants. It is not material that the benefits produced by a restrictive agreement do not entirely compensate each particular consumer if the average consumer does enjoy that compensation and the agreement therefore produces a beneficial or neutral effect on the market generally: see Case T-131/99 Shaw and Falla v Commission at paragraph 163 and the Article 101(3) Guidelines at paragraph 87. Mr Lowenstein argued that to be Article 101(3) compliant it was necessary to set MIFs for different size categories of merchant, who had different cost/benefit ratios from use of cards. Larger merchants, for example, would have lower costs of cash through economies of scale and so gain a smaller benefit from a customer using a card in place of cash. This approach is not only contrary to principle and authority but impossible of application. Mr Lowenstein chose the merchant size categories which were adopted by the Commission’s 2015 Survey (see below), but there was no logic or principle to such a choice. I can see no principled basis for subcategorising merchants so as to require differentiated MIFs for each category stopping short of examining the position of every single individual merchant.

Fulfilment of the indispensability requirement must be “realistic and attainable”, not “hypothetical or theoretical”. The Article 101(3) Guidelines state:

“75.

The first test contained in the third condition of Article 81(3) requires that the efficiencies be specific to the agreement in question in the sense that there are no other economically practicable and less restrictive means of achieving the efficiencies. In making this latter assessment the market conditions and business realities facing the parties to the agreement must be taken into account. Undertakings invoking the benefit of Article 81(3) are not required to consider hypothetical or theoretical alternatives. The Commission will not second guess the business judgment of the parties. It will only intervene where it is reasonably clear that there are realistic and attainable alternatives. The parties must only explain and demonstrate why such seemingly realistic and significantly less restrictive alternatives to the agreement would be significantly less efficient.”

…

80.

The assessment of indispensability is made within the actual context in which the agreement operates and must in particular take account of the structure of the market, the economic risks related to the agreement, and the incentives facing the parties…”

There are two aspects to the indispensibilty requirement. First, if a MIF at a particular level satisfies the fair share requirement, it will only be indispensable if the benefits cannot be provided by a realistic alternative to a MIF. My findings in relation to counterfactuals are dispositive of this aspect of the indispensability requirement. Insofar as there are relevant benefits conferred by the MIF, the MIF is indispensable to conferring those benefits because there is no other realistic counterfactual other than a zero or lower MIF. There is no other realistic counterfactual in which something other than a MIF could confer the relevant benefits if it be established that the MIF confers them.

The second aspect of indispensability arises in relation to issuer profits. If a MIF generates an equivalent level of benefits for merchants, it does not follow that it is necessarily indispensable to conferring those benefits. If the MIF creates an unfair degree of profit for issuers, it will not be indispensable to the benefits conferred on merchants by the payment of the MIF, even if merchants are receiving benefits which match the burden imposed by their paying the MIF.

Accordingly the exercise required in this case in determining what level of MIF, if any, qualifies for exemption involves:

There was a dispute between the parties as to the burden of proof and approach to the evidence. This is important because there are evidential difficulties which infect the Article 101(3) issues. The data upon which the issues turn was in significant respects lacking or incomplete. A number of issues had to be addressed at a level of generality which reflected not only the paucity of data but also the nature of the exercise and the constraints of the legal process. Time and resources did not allow for the analysis of every permutation of the many interconnected issues. To take one example, the parties addressed their arguments, and the experts their evidence, with a view to determining the level of an exemptible MIF as two single figures in each territory, one for credit and one for debit. Although the data used was taken from various points of time over the claim period, it was often treated as if it reflected the position throughout, and there was no attempt to identify different exemptible/exempt MIFs for different parts of the period. Equally, although in practice there were many MIFs for both credit and debit cards at any given time, broken down, for example by transaction type, there was no attempt to identify more than a single MIF as a blended weighted average of such MIFs when examining both the MIFs as set by MasterCard and those levels which would be exemptible. It was a recurrent theme of Mr Lowenstein’s submissions on many aspects of the Article 101(3) arguments that MasterCard had not adduced any or any sufficient evidence to address a point with the result that it should be determined in the Claimants’ favour.

There are two aspects to the issue of exemption:

MasterCard is correct in contending that the Claimants bear the burden of proof in relation to the exemptible level of the MIF. The issue of exemptibility is only relevant to the quantum of the assessment of damages. If the MIFs as set are not exempt, it follows that any exemptible MIF must be lower. In those circumstances the Claimants would have established the necessary ingredients of tortious liability, namely breach of duty and the suffering of some damage (subject to arguments in relation to pass through to customers which were rejected by the CAT and with which I am not concerned in this phase 1 trial). Whilst exemption of the MIFs as set is a liability issue, the exemptibility of a lower MIF is relevant solely to quantification of damages. It is trite law that a claimant bears the burden of proving the extent of its loss.

Mr Lowenstein argued that once it was established that the MIF as set was unlawful, the Claimants had established their loss at the full amount of the MIF element of the MSCs which they had paid, and that if MasterCard sought to challenge that fact by asserting that it could and would have charged a lawful lower MIF, it was for MasterCard to prove such assertion as the party putting it forward. He also relied upon Article 101(2) which has the effect of rendering the MIFs void if they are not exempt under Article 101(3). He submitted that when it came to compensation this made the starting point that the Claimants are entitled to the full amount of the MIFs (indirectly) paid through the MSC.

I cannot accept that this is the right approach. The Claimants do not establish that the extent of their loss is the full amount of the MIF merely by establishing that the MIFs as set were unlawful. The extent of their loss is measured by the extent of MasterCard’s tortiously unlawful activity, as required by the principles of causation. It is for the Claimants to establish the extent of their loss by reference to the extent of the unlawfulness. Put another way, their loss is to be measured not by the amount of the MIF they were required to pay, but the amount of the MIF they were unlawfully required to pay.

The position is no different from that of any claimant tortiously induced to pay a price for goods or services, for example by tortious misrepresentation. He does not prove his loss at the total amount paid for the goods or services merely by establishing the tortious inducement and the fact of payment; he must prove that he was induced to pay more for the goods or services than he otherwise would in the absence of the tort, usually by proving he has paid more than the market value, the difference being the measure of his loss: see for example McGregor on Damages 19th edn. at paragraph 47-055. If he makes no attempt to do this he has not proved his loss. In this case, the market value of the services for which the Claimants have had to pay by indirectly bearing the MIF can be treated as that which MasterCard could lawfully have charged by setting an exemptible MIF. It is for the Claimants to establish as their measure of loss the difference between this market value and their actual payment.

Article 101(2) does not affect this analysis. Its effect is to avoid the relevant restrictive agreement. If there were a claim for restitutionary relief by a party to such an agreement, Article 101(2) might well be the starting point. The Claimants, however, were not party to the agreement in the Scheme Rules by which the MIF was set, and do not make any restitutionary claim. The infringement of Article 101 is a tort, and it is this tort which founds the Claimants’ claim for damages. The voiding effect of Article 101(2) cannot affect the requirement for the Claimants to prove their loss in a tortious claim for damages. Article 101(2) is not the “starting point” for, or indeed relevant to, such quantification.

This analysis accords with the way the Claimants pleaded their claims in their Statements of Case. Taking the Record Shop 2 Ltd Particulars of Claim as representative of all actions, as the parties did (despite that being the action which settled before trial), paragraph 37 claimed the “Overcharge” as damages, which was defined as the amount of the MSC “over and above the amount which would otherwise have prevailed had there been no such infringements”. Paragraphs 41 and

I do not here intend to express any view on whether the measure must take account of what MasterCard would lawfully have charged or what it could lawfully have charged, if it be different, which is not within the issues in the phase 1 trial. If the correct question were what MasterCard would have charged it is still constrained by what it could lawfully have charged and that is issue 2 which I have to decide. It remains an essential element of the Claimants’ quantification of their loss.

Mr Lowenstein submitted that this analysis would result in the peculiar position that the burden of proof would be on different parties on the same issues which were being tried at the same time. It would lie on MasterCard on each of the criteria when seeking to prove exemption but on the Claimants when seeking to prove exemptibility. Each party would be carrying the burden of proof on the issues at the same time. This is indeed the result on many of the issues which arise under Article 101(3). However the principled allocation of the burden of proof should not be discarded merely because of apparent difficulties or anomalies in its application. It is not unheard of for one party to bear the burden of proof on a particular issue for one purpose and the other for another purpose. If the burden of proof is determinative of the issue, the consequence is that each party fails where it bears the burden.

Whether a restriction produces relevant pro-competitive effects which outweigh its anti-competitive restrictive effects involves examination of four elements in relation to the benefits. The Article 101(3) Guidelines identify them (as “efficiencies”) in paragraph 51:

“51.

All efficiency claims must therefore be substantiated so that the following can be verified:

(a)The nature of the claimed efficiencies;

(b)The link between the agreement and the efficiencies;

(c)The likelihood and magnitude of each claimed

efficiency; and

(d)

How and when each claimed efficiency would be achieved.”

(emphasis in original)

Before the Commission, MasterCard had sought to rely on the benefits of the scheme as a whole, or “system output”, in claiming exemption (exemptibility at any lower level of MIF was not in issue). In rejecting this argument the Commission said at paragraph 686 that “robust empirical evidence” was required and referred at paragraph 695 to “some form of convincing empirical evidence” and at paragraph 690 to a “detailed, robust and compelling analysis that relies in its assumptions and deductions on empirical data and facts”.

Mr Lowenstein referred to these passages to support a submission that there was a particularly rigorous evidential burden on MasterCard in all aspects of the Article 101(3) issues. This to impose greater weight on the language used than it will bear. The passages are directed to a party in breach of Article 101(1) seeking to establish exemption, not to a lawful level of exemptibility for a damages claim, where different considerations are at play, as I consider below. The language used was in contradistinction to what was characterised as mere “assertion” on the part of MasterCard, an argument without any evidential basis at all. In the context of exemption the requirement for “substantiation” is no more than a requirement for evidence, and the suggestion that it needs to be empirical and convincing means no more than that it must be based on evidence, not speculation, and be sufficient to convince the court to the requisite standard of proof which is the balance of probabilities. If the epithet “robust” is intended to add more and connote an enhanced standard of proof, it is difficult to discern any legal basis for such an approach, and especially so when it comes to exemptibility which was not being considered by the Commission.

When it comes to the burden on the Claimants to prove their loss, however, they are entitled to invoke the long established principles that the court takes a pragmatic approach. In particular:

However, where the court is compelled to use a broad brush in the absence of precision in the evidence of the harm suffered by a claimant, it should err on the side of under-compensation so as (a) to reflect the uncertainty as to the loss actually suffered and (b) to give the defendant the benefit of any doubts in the calculation: SPE International Ltd v Professional Preparation Contractors (UK) Ltd[2002] EWHC 881 (Ch), per Rimer J at paragraph 87, approved by the Court of Appeal in Blayney (t/a Aardvrk Jewelry) v Clogau St David’s Gold Mines Ltd[2002] EWCA Civ 1007, [2003] F.S.R. 19, per Sir Andrew Morritt V.-C. at paragraphs 31-34.

MasterCard submitted that merchants who accept MasterCard credit and debit cards enjoy the following appreciable objective advantages, some of which apply to both credit and debit cards, others to credit cards only:

The Claimants accepted that merchants enjoyed a number of these benefits from the use of cards. However they challenged the existence or relevance of four categories, namely business stealing, online sales, guaranteed payment and increased earlier sales as a result of the availability of credit. They also disputed that any benefits to merchants were directly caused by the MIF. It is convenient to address the causation question before returning to address particular disputed merchant benefits.

MasterCard submitted that these were all benefits to merchants which were at least to some extent the result of the charging of a positive MIF, for the following reasons:

The Claimants objected that this was a rehash of the “system output argument” which had been advanced by MasterCard in the Commission proceedings and rejected by both the Commission and the General Court; and that it fell into the error of confusing benefits conferred by the scheme with those caused by the MIFs, it being only the latter which were relevant.

In my judgement these are not sound objections. MasterCard’s argument that charging positive MIFs led to an increase in the use of cards and therefore an increase in the amount of the benefits enjoyed by merchants as a result of the use of cards is made good on the evidence before me. So too is its case that because cardholders received benefits from issuers which were funded by the MIF, the benefits to merchants of card use are to some extent directly caused by the MIF. That does not mean that all the benefits enjoyed by merchants are directly attributable to the level of MIFs charged by MasterCard. It does, however, mean that a MIF at some positive level is directly causative of some benefits to merchants. That is the starting point for the Article 101(3) process. There then remains to be addressed the difficult quantification exercise involved in valuing those merchant benefits which are directly attributable to the MIF. This raises difficult questions, which I address below, but none of them were issues which MasterCard sought to address before the Commission, or which the Commission, General Court or CJEU needed to address. The unsuccessful argument advanced by MasterCard before the Commission was that that the MIF contributed to an increase in the use of cards (“system output”) and that that was in itself sufficient to prove exemption for the MIFs at the levels set. There is nothing in the Commission decision or General Court Judgment which precludes the conclusion on the evidence before me that a positive MIF contributes to some extent to merchant benefits.

Mr Lowenstein argued, in reliance on evidence from Mr Dryden and Rochet & Tirole 2008 (see below) that maximising system output can be positively deleterious rather than beneficial. The theory, in a nutshell, is that payment card systems can exploit merchants’ fears of losing business, so as to cause them to be willing to pay higher interchange fees than are justified by the benefits received by taking a card payment in place of cash payment. This was characterised as the “must take cards” phenomenon. The view that this creates system “inefficiencies” assumes (a) that business stealing is an irrelevant merchant benefit, and (b) that merchant benefits are to be measured by reference only to the difference between cash and card transactions, neither of which are correct assumptions for the reasons which I explain below.

I turn now to consider the merchant benefits alleged by MasterCard whose existence or relevance was in dispute.

It was common ground that the avoided cost of cash sales was a relevant benefit to merchants. This is the basis for the MIT MIF methodology which Mr Dryden and the Claimants adopted. However it makes no allowance for the saved costs of other forms of purchase. Merchants receive a benefit from MasterCard card acceptance by virtue of the reduction in the number of transactions which would otherwise take place with three party scheme cards, predominantly Amex but also Diners Club. The merchant service fee for these cards is considerably greater than the MIF element of the MSC for MasterCard cards.

The competitive advantage of accepting MasterCard cards may be expressed in two ways which are opposite sides of the same coin. Merchants who accept either debit or credit cards (or both) benefit from a competitive advantage over their rivals because they gain sales from rivals who do not accept cards; and they avoid the loss of sales to card-accepting rivals which would occur if they did not accept cards.

This was the view of Dr Niels and was supported by evidence which emanated from a number of the Claimant retailers. Mr Dryden said in his report that he regarded it as plausible that merchants who accepted cards gained incremental sales from other merchants who did not, and in cross examination accepted this as not merely plausible but “likely”. In my judgement these are real and valuable benefits to merchants who pay the MIF. Not all merchants took credit cards during the claim period. For example the supermarket chains Aldi and Lidl, with whom a number of the Claimants were competing, did not do so for much of the claim period. When Aldi chose to accept cards from 11 October 2014, its decision was explained as a means of winning more business from its rivals. I have little doubt that taking cards conferred a competitive advantage on merchants which translated into incremental sales.

However Mr Dryden opined, and the Claimants submitted, that such business stealing was not a relevant benefit for the purposes of Article 101(3) because the relevant pool of merchants whose benefit fell to be considered was all merchants, not just merchants who took MasterCard cards; business stealing is a zero sum game amongst all merchants because insofar as business is stolen by card taking merchants from those not taking cards, any advantage for the former category is matched by an equivalent disadvantage to the latter.

I am unable to accept this argument in the context of either the benefit requirement or the fair share requirement. What are being considered for the purposes of the benefit requirement are “efficiencies” i.e. benefits from the standpoint of what contributes to greater competition. The MIF is borne only by those who take MasterCard cards. Such cards promote competition amongst all merchants, whether or not they take those cards. They do so by being attractive to those providing goods and services because they assist in promoting sales to customers. They are in this respect no different from the many commercial tools by which providers of goods and services seek to make themselves more competitive by comparison with their rivals: they may compete on price, on quality of service, by providing car parking, helpful staff, attractive retail layouts, by advertising or promotions and in numerous other different ways. Accepting cards is just another one of these competitive tools available to merchants.

The competitive benefit is not negated by the fact that the competitive advantage of taking such cards is enjoyed by the merchants who take them at the expense of those who do not; indeed it is the very fact that the benefit is gained at the expense of a competitor which makes it competitive. The fallacy in the Claimants’ approach is to treat business stealing as neutral in terms of pro-competitive or anti-competitive effect. It is not. It reflects increased competition which benefits some at the expense of others. That is the very nature of competition.

The pro-competitive advantage of taking cards amongst all merchants results in a gain for those who win out in such competition, namely the merchants who take MasterCard cards. It is they alone who pay for the MIF and, to the extent that the MIF is causative of the benefits, it is they who enjoy the competitive benefits of the MIF. It is therefore legitimate and sound in principle to take account of the value to those merchants of business stealing as one of the benefits which may contribute to a justifiable level of MIF for those merchants to bear.

This approach is supported by paragraph 84 of the Article 101(3) Guidelines which provides:

“The concept of “consumers” encompasses all direct or indirect users of the products covered by the agreement, including producers that use the products as an input, wholesalers, retailers and final consumers, i.e. natural persons who are acting for purposes which can be regarded as outside their trade or profession. In other words, consumers within the meaning of Article 101(3) are the customers of the parties to the agreement and subsequent purchasers.” (Emphasis added.)

This focuses on those consumers who use the products covered by the agreement i.e. MasterCard cards. It focusses on those who suffer from the anti-competitive effects of the restrictive agreement, whether directly or indirectly. Merchants who choose not to take MasterCard cards are entirely unaffected by the restrictive effect of the MIF on merchants (or acquirers) who do.

This is not to ignore the position of potential consumers in any consideration of benefits or restrictive effects. It is of course legitimate to take account of those merchants who do not take MasterCard cards if that is the result of the restrictive agreement, i.e. if the burden of the MIF outweighs the benefits of accepting cards. However the exercise is still to evaluate the benefits which such merchants would obtain from taking cards. It is the merchant benefit of accepting MasterCard cards which falls to be considered, because that is the benefit enjoyed or potentially enjoyed by “consumers” of the subject matter of the restrictive agreement.

This approach is also supported by paragraph 236 of the MasterCard CJEU Judgment where the Court said:

“…it is the beneficial nature of the effect on all consumers in the relevant markets that must be taken into consideration…” (emphasis added).

It is common ground that there are three relevant markets in the present case, the inter-system market for payment cards, the issuing market and the acquiring market. There is no relevant market of all merchants. Merchants who do not accept payment cards do not fall within the inter-system market, the issuing market or the acquiring market. They may potentially be participants in the acquiring market, but only as potential MasterCard accepting merchants, for which purpose it is only the benefits as a potential card accepting merchant which are relevant.

This is also the correct approach when considering the fair share requirement. The fair share requirement is that those adversely affected by the restriction which offends Article 101(1) should get a fair share of benefits flowing from it. That focuses on the position of merchants who bear the MasterCard MIF, not those who do not. The latter are not subject to the anticompetitive effects of the MIF and Article 101(3) is not concerned to accord them a fair share of anything.

Once it is recognised that it is only relevant to look at merchants who accept payment cards, it is obvious that incremental sales which those merchants obtain or retain at the expense of merchants who do not accept such cards are a relevant benefit.

It is well known that online sales account for an increasing and substantial proportion of merchants’ retail sales. As Mr Dryden confirmed, merchants offer online sales because they expect it to be profitable and may see it as a way of taking business from competitors. Card use undoubtedly facilitates such sales. It is not feasible to pay online by cash and even “secure payment methods” such as PayPal are often backed by cards rather than cash.

In The Cat Judgment, the Tribunal identified the guarantee of payment as a particular (and obvious) benefit in relation to fraud or default at paragraph 216(3)(ii):

“The payment guarantee is also important in cases of fraud and Cardholder default. Fraud is a problem in the case of both credit and debit cards. Where a card is only ostensibly valid (i.e. is a “clone” or forgery) or where the card is valid, but has been stolen from the Cardholder, the Cardholder typically does not pay. As between the Issuing Bank and the Merchant, this loss is typically borne by the Issuing Bank, to the obvious benefit of the Merchant.”

I agree. In relation to the fraud guarantee, the Claimants suggested in crossexamination of some MasterCard’s witnesses of fact that, even absent the MIF, MasterCard would have taken other steps to deploy measures to combat fraud, such as education programmes, or by including requirements in the Scheme Rules for merchants to adopt procedures in the use of cards to minimise fraud. This did not address the central benefit to merchants, however, which is the guarantee of payment by the issuing bank despite the fraudulent use of a card presented to the merchant. While Mr Willaert accepted that MasterCard would have sought to deploy anti-fraud technology even if interchange revenue had not been available, it was never put to Mr Willaert that MasterCard or issuing banks would or could continue to honour the fraud guarantee. Indeed, Mr Willaert, while accepting that anti-fraud technology could still be deployed, maintained his evidence that in a hypothetical world without interchange, MasterCard would have to make changes to allow the Scheme to remain competitive such as changing the scheme rule providing for a payment guarantee.

Mr Dryden accepted in cross examination that the availability of credit is likely to result in an aggregate increase in sales to merchants. This reflects both impulse buying, and buying on credit more generally where the consumer could not afford to pay immediately and would not save up to make the same purchase. Again the evidence from a number of the Claimant retailers and their internal documentation supported this conclusion that the use of credit cards was regarded as leading to an overall increase in sales. Average transaction values for credit cards are greater than for debit cards, suggesting that people spend more if they have access to credit. I agree with the conclusions of the CAT on this point expressed at paragraph 219(2)(ii) of their Judgment:

“(ii)

Some of the main benefits to Merchants, which we consider they would perceive and value, are…

(a)

Merchants benefit from customers being able to purchase goods and pay next month at no cost – which is the facility which the interest-free period provides. Moreover, Merchants do not have to bear the costs of setting up their own credit facility for their customers. ….

(b)

For the Merchant, the benefit is not only that the Merchant receives payment now, whereas the Cardholder pays later, but also, there are some transactions that either would not take place at all absent the provision of a creditfree period or which would take place, but in a manner less advantageous to the Merchant. By way of example, we have seen that the average transaction value in a Sainsbury’s store using a MasterCard credit card is nearly £6 higher than what is spent using a Maestro debit card, and there is a similar differential between Visa credit and Visa debit cards.”

Apart from generating sales which would not otherwise have occurred, the evidence suggested that merchants also perceived advantages from the availability of credit in bringing forward purchases which would only have occurred at a later date. A sale today has a higher net present value to a merchant than a sale at some point in the future; this is in substance the time value of the price over the period between the sale and the expiry of credit.

MasterCard also submitted that there was an additional benefit to merchants from impulse sales on cards, both credit and debit, even where credit was not required, because some customers do not carry around sufficient quantities of cash and would not go to the inconvenience of getting the cash, for example, from an ATM. There was no satisfactory evidence that this was a significant phenomenon. If consumers use a card in place of cash just because they do not have cash in their pocket, that begs the question whether they would have returned with cash later to make the same purchase in the absence of possession of a card. I am not satisfied that this translates into a real or significant benefit to merchants in the form of increased sales.

The MIF directly contributes to some extent to benefits to merchants in the form of:

Two alternative methodologies were canvassed as potential ways in which to quantify the exemptible level of the MIF. The first is the Merchant Indifference Test (“MIT”), also known as the tourist test. It takes its name from the premise that the only benefit to merchants of taking cards which can justify an interchange fee is the amount which would make the merchant indifferent to whether the customer paid in cash or by card. To avoid taking account of the advantages of repeat trade, the customer is assumed to be a tourist. The premise is that the merchant’s incremental cost of processing a cash payment is greater than that of processing a card payment, and it is the difference between the two which makes him indifferent to the method of payment if it has to be paid by way of a MIF. As described by Mr Dryden, in its most basic form the MIT is applied as follows:

“F.5 A MIT-compliant level of MSCis equal to the difference between a merchant’s incremental costs of processing a cash payment, on the one hand, and a card payment, on the other. Such costs are sometimes referred to generally as “costs of payment” or “incremental cost”.

F.6 A MIT-compliant level of MIFis then the MIT-compliant level of MSC less the elements of the MSC other than the MIF (that is, the scheme fee paid to MasterCard and the acquirer’s own mark-up, collectively referred to as the “acquirer margin”)”.

The second approach canvassed is a benefit-cost balancing approach. This involves trying to identify the costs of producing the benefits to merchants which the MIF confers, based on the costs incurred by issuers in creating those benefits or at least some of them.

Regulators have in the past used both methods and continue to do so. Regulators in America and Australia, for example, use a cost based methodology. The Commission used to do so. It now applies a MIT MIF methodology.

Interchange fees, and these methodologies, have been the subject matter of a considerable body of published academic work by economists, to which the experts and the parties made extensive reference. It is convenient to start with a brief overview of four of the important developments in that academic work.

Professor William Baxter’s 1983 paper “Bank Interchange of Transactional Paper: Legal and Economic Perspectives” (“Baxter 1983”) provided a theoretical explanation of, and justification for, interchange fees, in the context of a broader historical survey of payment systems. As Mr Dryden summarises it, the paper:

“provided the insight that buyers’ incentives to use payment cards may not be aligned with merchants’ interests. Merchants may benefit when a buyer pays with a card instead of cash. Yet it is the buyer who makes the choice of which payment instrument to use for a given transaction. If the merchant does not charge the buyer different prices for the goods or services purchased depending on the choice of payment instrument, the buyer will not take into account the merchant’s benefits of the various payment alternatives. The buyer will only take into account his or her own costs and benefits, which (except by coincidence) will not be the same as the merchants’. The result is that the buyer may fail to choose the payment instrument that is optimal for the merchant and buyer jointly.

On the basis of this insight, Baxter concluded that the proportion of purchases paid for by card could fall below the level at which the joint benefits of merchants and buyers would be maximised. This provides an economic rationale for an interchange fee (including a MIF): a subsidy, directed from the merchant to the cardholder, paid in the form of a MIF via the Acquirer and Issuer, could in principle address this problem by aligning buyers’ incentives when choosing between payment instruments with merchants’ benefits of accepting each method of payment.”

It can be seen that as the origin of the MIT MIF methodology this makes certain basic assumptions. It assumes that the only benefits to merchants which can justify a MIF are the avoided transactional costs of a cash purchase. It also assumes that the MIF has complete pass through via acquirers and issuers to the cardholder, so that it is the passed though MIF which incentivises the cardholder to use his card to an extent which aligns the benefit of card use to the cardholder with that to the merchant.

A paper by Professors Jean-Charles Rochet and Jean Tirole of the Toulouse School of Economics first published in 2008, although republished again in the years to 2011 (“Rochet & Tirole 2008”), addressed the justification for the MIT MIF as a “tourist test”. The tourist test was a response to what was described as the “must take cards” argument advanced by regulators and retailers, namely that merchants could not afford to refuse to accept cards because it put them at a competitive disadvantage with other retailers; and so were forced to pay a higher fee than that which reflected the cash/card alignment benefit of customers paying by card. Rochet and Tirole identified the tourist test as a valid proxy for establishing a MIF by eliminating any such consideration because it assumed a customer who was already in the store and who had not therefore been attracted to the merchant by the fact that it took cards. Rochet & Tirole concluded that the tourist test was in theory an exact measure of TUS if the “must take cards” factor was eliminated in this way.

It is immediately apparent that the MIT MIF methodology which this encapsulates is inadequate for present purposes. The exercise required by Article 101(3) does have to take account of the competitive advantages of taking cards; payment by card offers real attractions to the merchants’ potential customer base which makes merchants who accept cards more attractive, as the “must take cards” argument recognises. This competitive advantage is a benefit which falls to be taken into account for the purposes of the Article 101(3) exercise, but is left entirely out of account by the tourist test.

Professor Rochet produced a further paper in 2010, co-authored by Professor Julian Wright of the National University of Singapore (“Rochet & Wright 2010”) which focused specifically on credit cards. It drew the distinction between use of credit cards for “ordinary purchases” where the customer might as happily use a debit card or cash, and credit purchases where credit was necessary or a factor in the sale taking place, such as impulse purchases, or purchases for which the customer does not have the cash or funds immediately available. The authors expressed the view that taking credit cards allowed merchants to make sales they would not otherwise have made and this was the major reason why they accepted such cards and were prepared to pay higher interchange fees for doing so than for debit cards. The authors made a number of other points amongst which are these:

Professor Tirole produced a further paper in 2011 (“Tirole 2011”). In that paper he emphasised that the tourist test failed to take into account the benefits which accepting cards conferred on merchants in avoiding missed sales in two different ways. First it ignored the competitive advantage of accepting cards, whereas the fact that a merchant accepted cards was a significant benefit in increasing sales. Secondly, where credit cards were used, the merchant received advantages from sales which would not otherwise have taken place. In both these respects the relevant comparison for the benefit conferred by cards was with no sale taking place, not a cash sale; this had been underexplored in the economic literature and was not taken account of in the tourist test. The conclusion was that an interchange fee which reflected the benefits to merchants of card use lay somewhere between that derived from the tourist test, and what merchants were in fact prepared to pay because of the “must take cards” considerations of maintaining the competitive position vis a vis their rivals in attracting customers to shop with them. The paper also considered the tourist test on its own terms, and concluded that for various reasons it was probably a conservative test for a socially desirable interchange fee if the measure was total welfare: amongst other things it did not take account of the beneficial impact on social welfare of the long term impact of industry profit on “entry, innovation and end-user welfare”, or of the social disadvantages of cash use such as tax evasion.

In 2015, Professors Marc Rysman and Julian Wright produced a further paper based on research funded by Visa (“Rysman & Wright 2015”). Amongst other things

The MIT methodology depends upon data establishing and categorising merchant costs in order to analyse the benefits to them of card use. In 2015 the Commission published its Survey on merchants’ cost of processing cash and card payments, Final results, (the “Commission Survey”). This was based on a survey of merchants’ costs of processing payments by cash and cards undertaken by Deloitte on the Commission’s behalf (the “Deloitte Study”). The Commission quantified what the MIT MIF would be based on that data. Although the underlying data is not all available, the Commission Survey is sufficiently detailed to provide the best data available for calculating merchant benefits from card use and therefore the best starting point for quantification of any exempt or exemptible MIF. It is supplemented by the data contained in the survey responses of these particular Claimants, which have been disclosed in the proceedings, and by data from these Claimants in response to a Request for Further Information served by Visa in the Visa proceedings.

The Commission originally intended to collect information from a large representative sample of European merchants. However it soon discovered that it would not be possible to carry out the survey on such a large scale because much of the detail required to apply the MIT test was not readily available to merchants. Accordingly the study targeted 500 merchants in the EEA area, who were all large merchants. The Commission complemented this with a less detailed survey targeted at a wider and more representative group of merchants. The response rate for this wider survey was poor (about 1,300 responses out of almost 600,000 merchants approached) and the Commission concluded that it could not extrapolate any reliable results from this wider survey. Accordingly it based its conclusions as to the appropriate MIT MIF on the survey from the large merchants. The study took almost 7 years to complete, having commenced in 2008. In the event only 254 of the large merchants who were targeted completed the survey to the required standard.

In line with the basic MIT methodology the Commission estimated the cost function for cash and cards, comprising fixed and variable costs. The Commission excluded fixed costs. Whether a cost is fixed or variable depends upon the timeframe being considered for the analysis. The Commission considered short term (single transaction), medium term (3-4 years) and long term. The first two were based on the so called “arithmetic” analysis which used the merchant’s own categorisation in their survey responses as to what costs they deemed to be fixed or variable. The long term analysis by the Commission used a so called “econometric” analysis which does not rely on the merchant’s own judgements as to the fixed versus variable cost split.

The Commission’s results were as follows:

The approach advocated by Mr Dryden and the Claimants was to apply a strict MIT MIF approach, applying the tourist test. Mr Dryden’s approach was based on his opinion that “MIFs are capable of generating one specific benefit, namely aligning buyers’ incentives in choosing between different payment instruments with merchants’ costs of accepting those different payment instruments”. On the basis of this premise, he then created his own framework for analysis which involved identifying two necessary conditions which had to be fulfilled if a MIF was to satisfy the criterion for exemption:

This approach is unsatisfactory in seeking to determine whether the MIFs as set by MasterCard were exempt, or what level of MIF, if any, would be exemptible. In treating the only relevant merchant benefit as the alignment benefit of cash switching to cards it suffers from two fundamental flaws:

MasterCard advocated the approach adopted by Dr Niels, which was to start with the Commission’s MIT MIF but to make a number of adjustments to reflect what were said to be imperfections and errors in using it as a guide to an exemptible level of MIF; and to use a cost based “benefit-cost balancing” approach as a cross check.

There are serious flaws in an issuer cost based approach, just as there are in a pure MIT MIF approach. In particular:

In my judgement, the best available approach to quantifying an exempt or exemptible level of MIF is to use the MIT methodology and the Commission Survey results as the starting point; to adjust the Commission Survey results to take account of other relevant benefits to merchants, and of points made by the experts; this gives the best approximation of the value of benefits to merchants from card use; to assess the extent to which that quantification of benefits to merchants from card use is attributable to a MIF rather than simply the scheme as a whole; and last to consider whether and to what extent an issuer cost approach provides a useful cross check on the adjusted MIT MIF results.

Dr Niels’ view was that the long term figure was the most appropriate. Mr Dryden accepted that the appropriate approach was to look at the data for “the longer term” but did not accept that the medium term figures were inappropriate. His view was that the medium term figures captured the necessary variable costs and that the assumptions underlying the econometric method, on which the long term figures were based, “may not hold”.

In my judgement there are two reasons why the long term (econometric approach) is preferable to the medium term arithmetic approach “slightly” favoured by Mr Dryden. It is more likely to capture all costs which vary over time; and it avoids reliance on subjective classification by merchants of costs as fixed or variable (which is both difficult and subject to potential bias).

As to the first, the Commission recognised that the categorisation of costs as fixed or variable was critical to its calculations of the MIT MIF because in addressing the costs of cash saved by card transactions they included only variable costs and excluded fixed costs. Costs which are treated as fixed costs over a short period become variable over a longer period. For example if a cash register requires replacement once a year, the costs of doing so are fixed costs if the time horizon taken is less than a year but a variable cost if a period of longer than a year is taken. The proportion of merchants’ fixed costs is higher for cash than for cards; accordingly if the exercise does not account for the full range of costs which vary over time, it will systematically underestimate the appropriate MIT MIF.

In principle all costs of accepting cash should be taken into account in determining the MIT MIF, even if they recur only in the longer term than every 3 to 4 years. There are some costs which will only be variable if looked at over a period of greater than 3 to 4 years. Costs of cash acceptance include both goods (registers, counterfeit machines etc) and services (e.g. cash transportation). The goods may need replacing less often than 3 to 4 years. Similarly, contracts for services may be for longer periods than 3 to 4 years. A medium term approach would categorise such costs as fixed when they are in reality part of the costs of accepting cash which are saved by taking cards.

As to the second (subjective classification):

This conclusion is in line with what Dr Niels said was the common approach adopted by regulators, which was to use a long term horizon.

I have already explained that as a matter of law Article 101(3) is concerned with a fair share for all relevant consumers as a group, not individual consumers. The present exercise is therefore to determine the exempt/exemptible MIF for an average merchant.

This also accords with the economic literature. Rochet and Tirole 2008 makes it clear that the MIT should be based on the average merchant:

“In this case the average merchant (among those who accept cards) is ex post indifferent as to the means of payment chosen by the consumer. This implies that unless all retailers are identical, some would want to reject cards ex post. Efficiency cannot require that the tourist test be met by all participating merchants because cardholders must internalize the welfare of the average merchant and not of the marginal one, who values card payments less than the average merchant. Capping merchant discounts at the convenience benefit of the most reluctant merchants provides the cardholder with an incentive for under consumption of card payments.” (page 26)

“Third, merchants are heterogeneous, and an IF that properly guides cardholders’ decisions must reflect the average, not the marginal merchant benefit. This implies that the merchants who benefit least from the card, say the large retailers, are likely to fail the tourist test at the social optimum.” (page 29)

This last passage reflects the fact that because of economies of scale the costs of large merchants will be lower than those of the average merchant.

The Commission’s 2015 Survey was based on a survey of large merchants only. The Survey explains that merchants were divided into eight size classes according to turnover:

However, Deloitte only obtained useful data for size classes 6, 7 and 8 (all but 8 respondents were in these categories). As a result the Commission’s MIT MIF will understate the MIT MIF for an average merchant because economies of scale mean that the incremental cost of cash saved by smaller merchants will be higher than that for larger merchants. This is well illustrated by the median marginal cost of cash for sizes 6 and 7 being about 50% higher than that for size 8.

There are limitations in the data available which make it difficult to make any precise adjustment for this. There is none available for sizes 1 to 5, either in terms of their costs or as to what proportion they form of the total card accepting merchant cohort. Dr Niels’ solution was to take a cost of cash based on sizes 6 and 7, on the basis that the costs for the size 6 and 7 merchants were much higher than for the size 8 merchants, and it could be assumed that as one went down the class sizes the merchant costs would continue to increase. The statistical data for putting classes 6 and 7 as a proxy for the average merchant’s costs was simply not available. Nevertheless I am satisfied that Dr Niels’ approach is the most reasonable available one in the circumstances, because the cost differences amongst the merchants who were surveyed show a dramatic increase in costs as the size of merchant reduces. Figure 11 in the Commission’s Survey illustrated that of the sampled group (sizes 6, 7 and 8):

I have already concluded that merchants receive other benefits from the use of MasterCard cards, of which the MIF is to some extent directly causative. Save for the avoided cost of cash sales, none are captured in the MIT MIF methodology used by the Commission 2015 Survey. They include the following:

How are these to be quantified? Dr Niels has made an adjustment to the Commission’s MIT MIF figures to take account of credit functionality for the purpose of assessing an adjusted credit card MIT MIF. Once adjusted in this way it is, of course, no longer a “MIT MIF” because the MIT test is a tourist test which looks only at the avoided cost of cash. I shall for convenience call it an adjusted MIT MIF.

Dr Niels identified the proportion of transactions which depend on credit based on the percentage of cardholders who do not pay off their cards at the end of the interest-free period. These are cardholders who are categorised as “revolvers”. For these consumers he has used the cost of Amex as a comparator, on the basis that the customers require the deferred payment element of credit and the alternative would be the use of a more expensive three party scheme card, namely Amex. For the proportion of transactions which do not use the extended credit facility, where cardholders are “transactors” who pay off the balance at the end of the month without incurring interest, Dr Niels uses cash as a comparator.

For transactors, the use of Amex as a comparator was justified on the following basis. In accordance with Rochet & Wright 2010, Dr Niels took as the starting point for his adjustment in relation to credit functionality, the comparator of merchants providing credit themselves, through store cards. He then took as a proxy for the cost to merchants of offering store credit themselves, the costs of accepting Amex. Dr Niels’ analysis of the cost of store credit to a number of the Claimants who offer store credit shows that it is cheaper for merchants to accept credit cards, including Amex, than offer store credit themselves. Because Amex is cheaper than providing store credit, the use of Amex as a proxy will produce a lower adjusted MIT MIF. The use of Amex as a comparator also captures some of the cost saving benefit that merchants obtain from customers using MasterCard/Visa rather than Amex.

I accept Dr Niels’ reasoning and conclusions on this issue. However, it is important to note that it will still produce a level that is too low because:

Mr Dryden’s argument against including online sales as a comparator in the calculation of the MIF was his conclusion that the only relevant benefit to merchants was avoiding the costs of cash, and cash is not a realistic alternative for online sales. I cannot accept this reasoning. Even accepting the MIT MIF on its own terms, it ignores the fact that consumers have a choice to make purchases online or in a shop.

Dr Niels accounts for the benefit of online sales in his calculations in the following way. For the relevant proportion of online sales, for both credit and debit cards he changed the comparator from cash to Amex and PayPal, appropriately weighted by an estimate of the online market shares of each in the relevant market.

It was contended by the Claimants that the use of PayPal as a comparator is inappropriate, because 65% of PayPal accounts are in any case linked to a payment card. The point was that for these accounts, switching from PayPal to payment by a credit card is unlikely, as the customer will in any case obtain the rewards and other incentives available from using a payment card. These points do not invalidate Dr Niels’ methodology. They do not touch on the 35% of PayPal transactions which are not linked to a payment card. The 65% of PayPal transactions linked to a card would likely include a proportion of accounts linked to debit cards. Credit cards also have some advantages from the cardholder’s perspective over payment made using PayPal, including the availability of protection under the Consumer Credit Act 1974, s.75 so that customers buying online benefit from direct credit card use rather than via PayPal. The use of PayPal as a comparator actually results in a lower MIT MIF calculation than would be the case if Dr Niels had instead used only Amex as a comparator because Amex is more expensive to merchants than PayPal.

Again, I accept Dr Niels’ reasoning and conclusions on this issue.

In calculating a MIT MIF the experts have had to use a figure for average transaction values in order to correlate costs with a MIF expressed as a percentage of transaction value.

Mr Dryden uses an ATV of €25 - which is the average 2012 European transaction value for cash and card transactions as reported in the 2015 Commission Survey. Using the same 2012 average exchange rate used by Mr Dryden this equates to £20.30. This ATV for all transactions (both cash and card) is referred to as the “retail ATV”. Mr Dryden uses the retail ATV for Europe as a whole in order to calculate the MIT MIF for the UK, Ireland and cross-border transactions.

Dr Niels uses separate ATVs, specific to the relevant type of card (credit/debit) and specific to the relevant territory (UK/Ireland/EEA). This card specific ATV is referred to as the “card ATV”. For example, the equivalent UK debit card ATV is £46 and the UK credit card ATV is £60.

I prefer Dr Niels’ approach on this issue for the following reasons:

In the Commission Survey the marginal costs of both cash and card payments are driven by only a few of the costs categories. This does not include “front office costs” because in the Survey’s words “front office time costs are rather similar for cash and cards”. For these purposes “front office costs” means the cost of the staff time involved in processing the payment part of the transaction. The Commission’s analysis took account of the front office costs of processing card transactions and cash transactions for face-to-face transactions, but not self-service tills or online transactions. These are substantial omissions – the data for one of the Claimants, Asda, for example, shows that around of transactions in store are at self-service tills and of transactions by value are online. For other merchants online sales will be a much higher proportion. The average cost of processing self-service transactions or online transactions will be lower than the cost of processing face-to-face transactions precisely because there is no front office cost involved in processing the transaction. Consequently, by looking just at face-to-face transactions, the Commission has overstated the average cost of processing transactions.

If there were the same overstatement in the average cost of processing both cash and card transactions, this would not matter, since under the MIT MIF, we are concerned with the difference in the cost. However, the overstatement will be much greater in relation to card transactions, since online transactions will largely be card transactions (or an alternative like PayPal), since cash is not a realistic option for most online transactions and there are no front office costs for online transactions. A similar issue arises in relation to self-service tills i.e. it is likely that a higher proportion of card transactions take place at self-service tills (which again result in no front office costs being incurred) than face to face.

The front office costs as a whole make a difference of 3 basis points. Unfortunately the detailed data necessary to make the appropriate adjustment for front office costs savings due to use of cards is not available. One of the Claimants in its response to the Commission identified that self-service checkouts account for the following approximate transaction volumes – cash 9%, debit card 14%, credit card 12%. According to the UK Cards Association, 9% of debit card transactions and 24% of credit card transactions are online. Using those figures as a guide, I conclude that an adjustment of 1 basis point is required to account for the difference in front office costs from the use of cards.

The acquirer margin forms part of the calculation of the MIT MIF in the following way. The comparison between the costs of cards and other methods of payment is intended to arrive at a figure which values the benefits which the merchant receives. A MIF which equates to that figure would cause the merchant to pay more than the value of such benefits because the merchant pays the MSC, which comprises both the MIF and the margin of the acquiring bank. The experts and the Commission have therefore deducted an estimated acquirer margin from their calculated MIT MSC, in order to arrive at the compliant level of MIF. It follows that the larger the assumed acquirer margin, the smaller the adjusted MIT MIF would be.

The Commission Survey found that for the size 6 to 8 merchants which participated in the survey the acquirer margin was 0.06% of the transaction value. In seeking to derive a MIT MIF for the average merchant what is required is an average acquirer margin for all merchants. This would likely be larger than the acquirer margin for size 6 to 8 merchants because the largest merchants are likely to be better able to negotiate down the margin charged to them by their acquiring banks.

Dr Niels used the figure of 25% of the total MSC which is a whole market figure he took from the Commission’s July 2013 Impact Assessment accompanying the proposal for the Interchange Fee Regulation. Mr Dryden by contrast used the Commission’s figure of 0.06% of the transaction value.

Dr Niels’ approach results in a larger acquirer margin and therefore a lower figure for the MIT MIF than would have been the case had he followed Mr Dryden’s approach. MasterCard was content that I should adopt such approach and I will do so.

Applying these conclusions to adjust the MIT MIF to reflect the value to merchants from use of cards, the appropriate per transaction values are those set out at tables 6.5 & 6.6 (UK), 6.8 & 6.9 (Ireland) and 6.11 & 6.12 (intra EEA) in Dr Niels’ first report, using medium PayPal costs and adjusted downwards by two basis points to reflect the fact that he has adjusted for front office costs by adding a full 3 basis points whereas I have concluded that a 1 basis point adjustment is appropriate:

These figures do not take account of the benefit to merchants of the competitive advantage of taking cards over merchants who do not, i.e. “business stealing”. There is little empirical data available to me to enable the value of this benefit to be accurately quantified. However it is clear that the competitive advantage of taking cards is a real and substantial benefit to merchants, which for the reasons I have explained is a relevant benefit. It would therefore be wrong to ignore it in any quantification exercise required by Article 101(3).

The “must take cards” attitude adopted by most larger retailers suggests that they put a significant value on the competitive advantage over those who do not take cards. The statistics set out in the introductory section of this judgment illustrate that acceptance of cards by merchants is widespread. There were during the claim period a few major retailers in the UK who did not accept cards (e.g. Aldi and Lidl) but most retailers are prepared to pay the interchange fees which are set. I do not confuse a willingness to pay the MIF with an acceptance that the charge is reasonable; nor do I conclude that such willingness determines the issue whether the charge is anticompetitive; nor do I equate the amount which merchants are willing to pay with the value to merchants of taking cards. However I cannot accept Mr Lowenstein’s submission that willingness to pay has nothing to do with benefits because it is “ransom pricing”. The so called “ransom” is a price paid for a perceived benefit, and the perceived benefit is the competitive advantage which a card taking merchant has over one who does not take cards; or, which is the other side of the coin, the disadvantage he will suffer by not taking cards against merchants who do. When Aldi chose to accept cards from 11 October 2014, its decision was explained as a means of winning more business from its rivals. A number of internal documents disclosed by Claimants indicate that they regarded taking cards as conferring a competitive advantage which translated into increased sales. Mr Dryden’s evidence was that “merchants would stand to lose a large amount of business if they did not adopt MasterCard at all”.

How is this benefit to be quantified? The authors of Rochet & Wright 2010 expressed the view not only that taking credit cards allowed merchants to make sales they would not otherwise have made, but also that this was the major reason why they accepted such cards and were prepared to pay higher interchange fees for doing so than for debit cards:

“Thus offering credit allows an individual merchants (sic) to make sales that they otherwise would not make. The ability to make these sales is, we think, the major reason explaining why merchants accept credit cards and indeed are willing to pay higher fees to do so compared to the fees paid to accept debit cards, and why prior to the widespread use of credit cards, store credit was much more widely used than today (Evans and Schmalensee, 2005, pp 48-51)”

In other words the authors recognised that the credit functionality and business stealing aspects of accepting credit cards were a benefit to merchants which they valued as more significant than the other benefits (such as avoided costs of cash) and justified the differential interchange fee over debit cards. The difference between the weighted average MIF for MasterCard credit cards in the UK and that for debit cards for the claim period was of the order of 0.6%. Dr Niels’ adjustment for the credit functionality of credit cards makes a difference of only 0.18%, which suggests that business stealing justifies the remainder.

I am satisfied that the value to merchants of the competitive advantage over rivals in accepting credit cards increases the adjusted MIT MIF values I have arrived at by at least 0.4%. This is a rough and ready approach but one which is required by the exigencies of the available data, and in which if anything I have erred in favour of the Claimants.

For debit cards, the position is not quite the same. The competitive advantage of accepting debit cards is a real one: customers perceive security and convenience benefits in carrying and using debit cards rather than cash, and thereby obtain consumer protection remedies for defective goods and services. Online sales are facilitated by debit card use. However accepting debit cards is unlikely to confer the same competitive advantage value to merchants as does accepting credit cards; for a debit card purchase the customer can make the same purchase in cash at a rival store, whereas the availability of credit for credit cards makes the difference for transactors between making a purchase and not, and for revolvers gives an advantage over cash by virtue of the interest free period (which is not captured in Dr Niels’ adjustment for credit functionality).

In relation to debit cards I would increase the adjusted MIT MIF figures identified above by at least 0.2% to reflect the value of the competitive advantage of taking cards. Again this is a rough and ready approach but one which is required by the exigencies of the available data, and in which if anything I have erred in favour of the Claimants.

Accordingly the value to merchants of accepting cardsunder the adjusted MIT MIF methodology is at least

The level of pass through of the MIF by issuers to cardholders is relevant to three aspects of the exemption/exemptibility quantification exercise:

Issuers potentially pass through the MIF to cardholders by spending it to cover the cost of providing cards to customers and administering their use. These costs include:

Quantification of pass through is difficult to estimate. It cannot be an exact exercise because the MIF is not an isolated pot of money treated as such by issuers; they derive other revenues from their card business, for example interest income and card fees. Neither expert in his reports sought to quantify the extent of issuer pass through other than in adjectival terms: “very high” per Dr Niels, and “a significant proportion” per Mr Dryden.

In his report Mr Dryden drew attention to three sources of potentially helpful data:

Dr Niels based his opinion that pass through was very high on economic theory and the competitive nature of the issuing market: as a matter of economic theory, in a perfectly competitive market suppliers set prices equal to marginal costs; and a single cost change to an industry which is highly competitive results in all or almost all of the cost change being passed on to the consumer; there are a large number of issuers in the UK and in general the UK market is regarded as highly competitive, particularly for credit cards; the FCA November 2015 report concluded that there was strong competition in the credit card market, with competition focussing on rewards and introductory promotional offers; over half of consumers claimed that they shopped around when choosing their credit card with 66% of those using price comparison websites; the competition intensified from the late 1990s with monoline issuers and retailers entering the issuing market; the combination of a large number of competitors, high levels of switching and issuers competing strongly on rewards suggests a highly competitive market in which the economic theory supports high pass through. In cross examination Dr Niels described the level as “very high” but he couldn’t “pinpoint a number, whether it was 80%, 90% or 100%”. He drew attention to publicly available information showing that four issuers, RBS, Capital One, Tesco and M & S Bank had reduced or removed their reward offerings in late 2015 and 2016 following the cap imposed by the European IF Regulation. He was cross examined on one of these (Tesco) which indicated that the reduction in rewards did not approach the level of even 80%; but in my view these examples provided no firm empirical basis to support or undermine his estimate. They are not a representative sample and do not reflect the end point in any reduction of benefits. His estimate of 80%-100% was not based on any identified data or calculations.

Mr Dryden did not attempt any estimate of the relevant percentage. He made a further point in relation to pass through, which was that it was not enough for the MIF to be passed through by issuers to cardholders in the abstract; any pass through must be of a nature that could cause cardholders to respond and they must at least have some degree of responsiveness to that pass through. In other words the pass through of the MIF must be such as to incentivise card use. This is sound in principle, because a MIF which does not translate into card use cannot have a direct causative effect on merchant benefits arising out of card use. Mr Dryden and Dr Niels accepted that pass through for credit cards, which were stand-alone products, met this hurdle but Mr Dryden opined that for debit cards it “may not”. He identified three reasons why the responsiveness of debit cards to changes in the interchange fee may be less than that of credit cards:

Dr Niels disagreed. He drew attention to the fact that some debit cards involve reward schemes which are readily responsive to changes in interchange fee. The MasterCard witnesses also gave as an example of pass through money spent by issuers on innovation in relation to contactless payments.

There is some force in the points made by Mr Dryden, which lead me to conclude that the extent of effective issuer pass through for debit cards is less than that for credit cards.

Mr Dryden’s analysis involved as his Necessary Condition 1E that if pass through is less than 100%, it is necessary that the number of “switching transactions”, i.e. purchases which would switch from being made from cash to card must be sufficiently large relative to the number of “always card” transactions (i.e. transactions that would always be made with a card). Mr Lowenstein devoted a number of submissions towards seeking to establish that this condition was not fulfilled. In my judgement this analysis does not assist the debate. Mr Dryden’s framework assumes that the only relevant benefit to merchants is the alignment benefit of not taking cash. For reasons I have explained, this is a false premise for the Article 101(3) exercise. The error may be illustrated by one of Mr Lowenstein’s arguments in this context: he submitted that on the basis of the evidence most credit cards transactions did not involve substitution for a cash payment: that 40% of transactions using credit cards were by revolvers, and so by definition were transactions which did not involve switching from cash because those who needed credit could not have paid in cash; and that for the other 60% who did not need credit, the transactors, the vast majority would otherwise have paid with a debit card rather than cash. This illustrates both the fundamental flaws in Mr Dryden’s framework of Necessary Conditions. It simply ignores the merchant benefits from the extension of credit which credit cards provide, and all the other relevant merchant benefits from the use of both credit and debit cards which I have identified. Moreover it ignores Amex as a competitor; whilst a transaction which is dependent on credit will not switch from MasterCard to cash, the evidence established that if MasterCard could not set interchange fees for premium cards at a competitive level for issuers, MasterCard would have lost at least a significant proportion of its premium business to Amex, and possibly all of it. This is a considerable part of MasterCard’s credit card business in the UK, up to 45% of which was estimated by Project Porsche to be in “the Amex sweet spot”. I was invited by Mr Lowenstein not to make any finding as to the extent of the loss of business to Amex because the evidence on the issue was directed to the death spiral argument in respect of which Amex competition was not determinative, and because it would be an issue in Phase 2. I shall accept the invitation because the point I am here addressing is merely illustrative of the flaws in the principle underlying Mr Dryden’s framework.

The mere fact that banks make profits does not, of course, mean that issuer pass through of the MIF to cardholders is less than 100%: If the MIF represents only 20% of card related income, it may all be passed through to cardholders and the profit generated by other revenues, for example interest charged. It would be a non sequitur to assume that if the MIF were reduced the income could be made up elsewhere, such as in card fees, for example.

I have not had substantial evidence of issuer profits or issuer finances which might assist on this question. Neither side called any witness with substantial contemporaneous experience as an issuer. The documentary evidence was exiguous. There was a MasterCard document based largely on 2006 data which estimated that MIF revenues typically represented about 30% of issuers’ credit card income for EEA transactions, and that profits were about 33%. This was not however a sound guide to the position in the UK, as Mr Douglas explained. It estimated that card fees accounted for 51% of income, whereas in the UK the figure for that point of time was probably closer to 10-15%: there are generally no card fees for debit cards and only some for some credit cards. Mr Sidenius, who accepted that he did not have detailed knowledge of actual issuer profitability, estimated that an issuer would typically have made a profit of low 20s% to 40% on its UK credit revolving cards prior to the global financial crisis but the figure would be lower since then. Mr Douglas gave an estimate of 10% to 40% for issuers’ margins which he estimated would vary from issuer to issuer. Both suggested that interchange fee might account for about 20% of an issuer’s UK card related income. The FCA November 2015 report suggests that issuer profit on credit cards is not significantly driven by interchange revenue, but rather by interest payable by revolvers on credit balances.

Doing the best I can, my conclusions are that the degree of issuer pass through is likely to be at least 75% for credit cards and at least 40% for debit cards. These percentages are intended to reflect the causative effect of the MIF in incentivising card use. The level of profit for issuers is not such as to violate the fair share or indispensability requirements. A profit at levels which varied across the market between 10% and 40% is not indicative of an unreasonable return on capital for a bank and would not of itself represent such a degree of profit as to cause the fair share or indispensability requirement to be breached if the threshold test of merchant benefit neutrality were fulfilled.

Accordingly the value of merchant benefits caused by the MIF using the adjusted MIT MIF methodology, rounded to the nearest basis point is

This exercise was only undertaken by Dr Niels for the UK MIF because data was not available for the Irish or EEA MIFs.

For credit cards Dr Niels took particular heads of costs which are incurred by issuers in conferring benefits on merchants without generating revenue for issuers, namely (1) transaction processing costs; (2) costs of the payment guarantee, i.e. credit and fraud write offs and the costs involved in mitigating those risks; and (3) the cost of interest free credit. This was the method adopted by MasterCard as part of its process in setting the MIF for credit cards and was used by the Commission in its Visa 2002 Decision. The costs used by Dr Niels were based on two studies carried out for MasterCard by EDC in 2005 and 2008. These are likely to underestimate the costs for the period shortly after that because the global financial crisis meant that credit write-off peaked in 2010, but there is no reason to treat them as unrepresentative across the whole of the claim period.

The three heads of costs add up to 2.11% of total transaction value in the 2005 study and 2.41% in the 2008 study. Dr Niels then adjusted these to reflect the fact that the costs are incurred in conferring benefits on cardholders as well as merchants. The benefit to merchants from these costs is only a proportion of the costs. He takes a figure of “at least “25% or 50%”, which he describes as “conservative”; applied to the 2008 study figures this would produce a range of 0.75% to 1.31%.

This exercise is of limited value as a cross check. An issuer cost based approach is not supported in any of the academic literature as having any sound theoretical basis. Rysman & Wright 2015 describe it as “not supported by any economic theory” and Tirole as “unfortunately bears little relationship with the theoretically correct level”. It does not capture all costs and costs are not to be equated with value to merchants. Moreover, and critically, the figures of 25% to 50% were not underpinned by any calculations or data analysis and seemed no more than figures plucked out of the air.

For debit cards the methodology was different. EDC conducted a full end to end study of the costs to issuers of a debit card transaction in 2006, which produced an ad valorem estimate of 0.34%. This too is of limited assistance because it fails to distinguish between the cost to issuers and the benefit to merchants, and fails to take any account of the benefit conferred on cardholders by the costs incurred by issuers.

The most that can be said of the figures produced by Dr Niels’ issuer cost based approach is that they suggest that the exempt/exemptible MIF levels calculated by the adjusted MIT MIF methodology are not wholly out of line with an approximately appropriate proportion of the costs involved in conferring those benefits on both merchants and cardholders.

It is appropriate to use slightly different figures for exemption and exemptibility, to reflect the fact that the burden of proof and approach to the evidence is different. The figures arrived at above are those which are appropriate when considering exemption, because where in doubt I have erred in favour of the Claimants. When considering exemptibility, rather than exemption, where the evidential approach and burden of proof involve resolving doubts in MasterCard’s favour because it is a damages issue, the figures require upward revision: the extent of issuer pass through and the value of the competitive advantage over rivals may in each case be greater than the adjustments I have used in arriving at these figures. I would allow a 10% margin in MasterCard’s favour to take account of these factors.

Accordingly I conclude that below the following levels the MIFs as set would be exempt under Article 101(3):

Applying a 10% margin and rounding to the nearest basis point, the following levels of MIF would be exemptible under Article 101(3) for the purposes of calculating any damages claim:

These compare with the MIFs as set by MasterCard, expressed as a single weighted average for each year varying as follows:

It follows that the MIFs as set were below the exempt and exemptible level, save for the EEA debit card MIF for the earliest part of the claim period prior to June 2008.

In this phase 1 trial, I have not been asked to address the extent to which the MIF borne by merchants is passed on to customers, both cardholders and non-cardholders, in the form of retail pricing. It was assumed in argument before me that that is a question which affects only quantum of damages recoverable by the Claimants. However it is not self-evident to me that it is not also relevant to the Phase 1 issues I have been addressing. If the merchants pass some of the cost of the MIF on to cardholders, the fair share condition would be focusing on a different level of benefit/burden to both merchants and cardholders. Because these matters were not argued, and may arise in consumer claims against MasterCard, I express no view upon them.

The issues in Phase 1 will be answered in accordance with this Judgment.