Monthly Archives: December 2017

Market dynamics in the counterfactual: more competitive, not just cheaper

The judgment of Phillips J in Sainsbury’s v Visa [2017] EWHC 3047 (Comm) demonstrates the importance to claimants in competition damages cases of identifying a counterfactual which not only involves lower prices but also involves higher levels of competition.

Sainsbury’s case

Visa’s payment card scheme required ‘acquirers’ (who process card payments on behalf on merchants) to pay an ‘interchange fee’ to the issuer of a payment card whenever a payment was made. All acquirers were required to accept all cards issued in the scheme (the so-called honour all cards rule or ‘HACR’). All issuers were required to remit to the acquirer the whole of the payment made by the customer, less the applicable interchange fee (this was called the ‘settlement at par’ rule). Acquirers passed on all of the interchange fees to merchants, as part of the merchant service charge which also included an element of profit margin for acquirers.

Visa set a default interchange fee (the multilateral interchange fee or ‘MIF’), though acquirers and issuers were free to negotiate different fees bilaterally. However, no acquirer had an incentive to agree to pay more, and no issuer had an incentive to agree to accept less than the MIF. The settlement at par rule prevented issuers from, in effect, forcing a higher interchange fee on acquirers by remitting customer payments at a discount. The HACR prevented acquirers with market power from forcing lower interchange fees on issuers by refusing to accept cards unless bilateral interchange fees were agreed.

The combined effect of these rules was to eliminate any competition as to the level of interchange fees. This was Sainsbury’s case and Visa accepted that these arrangements constituted a restriction of competition ‘in absolute terms’: [103-104].

Sainsbury’s proposed counterfactual, which the court accepted, was one in which there was no MIF set by Visa, but the settlement at par rule and HACR remained in force: [98]. This, in effect, amounted to a MIF of zero and the same dynamics between issuers and acquirers as existed in the factual would have prevented bilateral interchange fees from being agreed in the counterfactual: [126-129]. It followed that on Sainsbury’s counterfactual, the interchange fees paid would have been lower but not because of any re-introduction of competition in the setting of such fees. The suite of rules which had operated to eliminate competition in the factual would operate in the same way and with the same result in the counterfactual, just at different prices: [161]. Sainsbury’s claim therefore failed as it had not established that its loss was caused by a reduction in competition.

What went wrong?

Sainsbury’s counterfactual retained two key elements from the factual scheme: the settlement at par rule and the HACR. All parties agreed that the scheme would be unworkable if it incorporated the HACR but did not require settlement at par: [99]. Such a scheme would be equivalent to allowing issuers to set interchange fees unilaterally, by settling payments at a discount. Assuming acquirers continued to pass interchange fees on to merchants in full, this would result in merchants ceasing to accept Visa, as continued participation in the scheme would mean accepting all card payments regardless of the fee charged.

The parties do not appear, however, to have explored the possibility of a truly bilateral system in which neither the settlement at par rule nor the HACR applied. In a counterfactual from which both of these rules were absent, issuers and acquirers would have been forced to negotiate terms of settlement bilaterally. Issuers would have wished to agree a higher interchange fee (or a larger discount from par) but would have been prevented from demanding too high a fee because of the risk that acquirers (in order to retain merchant business) would cease to accept a certain issuer’s cards.

This counterfactual involves a radical departure from the Visa scheme in the factual, but the market dynamic which would result is familiar: this is how competition works in so-called three-party schemes, such as American Express. The issuer faces competing incentives: higher fees make for greater profits per transaction, but too high fees reduce card acceptance by merchants and reduce transaction volume, ultimately reducing the appeal of the card to customers.

The outcome in this counterfactual could be a patch-work of differing fees charged by different issuers with corresponding variances in merchant acceptance. There are over 50 issuers and around 30 acquirers in the UK[1], which would give rise to a large (but in principle manageable) number of bilateral negotiations assuming the same number of issuers existed in the counterfactual. There may be good reason, however, to think that there would be fewer issuers in the counterfactual: If the setting of MIFs inhibited competition and raised issuer profits, it is likely also to have encouraged more issuers to enter the market than could have been sustained in a competitive scenario. Lower interchange fees, lower profits, the requirement to negotiate deals with all (or almost all) acquirers and the need to reassure prospective customers that the card would be widely accepted would all act to restrain the number of viable issuers in the counterfactual.

Merchants now indicate whether they accept Amex or Diners. Before the emergence of four-party schemes in the UK, merchants indicated whether they accepted Barclaycard, a card issued only by the bank of the same name. In the counterfactual described here, merchants would be required to indicate which of the major banks’ and independent issuers’ cards they accepted.

Of payment cards and dog races: monopoly and monopsony in price setting

Competition within the Visa scheme was inhibited by the fact that fees were set centrally for all participants. The scheme was controlled by participating issuing banks, so central price setting tended to result in higher fees. If merchants (or acquirers who did not also issue cards) had controlled the scheme, the result might have been that low or even negative interchange fees were set, but the result would have been no more competitive. The mischief against competition was the setting of prices centrally; control over the scheme determined which party stood to gain.

In Bookmaker’s Afternoon Greyhound Services [2009] LLR 584, cited by Phillips J at [91], the boot was on the other foot. Whereas in the Visa scheme, prices were set by or on behalf of the party receiving payment, in the BAGS case, prices were set by the paying party, which was controlled by the leading bookmakers, and which had a monopsony on buying live television footage from racecourses. When a group of racecourses jointly agreed to sell their footage exclusively through a newly formed distributor, prices for footage rose and BAGS claimed that the racecourses had acted anti-competitively. On the contrary, the court found, the market power of BAGS had been reduced and competition had been increased. As Phillips J pointed out in Sainsbury’s, the BAGS case shows the pitfalls of a facile analysis which equates price decreases with increases in competitive intensity and vice-versa.

It ought to be possible for the payment card market to operate competitively given the large number of retailers, banks and other issuers. Where collective price-setting is interposed between the parties on each side of the transaction, there is an obvious possibility of distortions to competition. It was common ground in Sainsbury’s that the Visa scheme as implemented had eliminated competition in the setting of interchange fees: [103-104]. A claimant seeking damages for the setting of payment card interchange fees should ask the court to consider a simple counterfactual, in which these distortions are eliminated: What would the result have been if the parties on each side had decided whether to transact with one-another and on what terms, without the central setting of prices or a compulsion to transact?

[1] In Arcadia v MasterCard [2017] EWHC 93 (Comm), at [103] Popplewell J found that there were 55 MasterCard issuers in 2015 in the UK; it is assumed that there were a similar number of Visa issuers.

Leave a comment

Filed under Abuse, Agreements, Damages, Economics

Intel Corporation Inc v European Commission

In its recent judgment in Intel, the Grand Chamber shed valuable light on the “qualified effects test” for jurisdiction and on the room for loyalty rebates to be compatible with competition law.


Intel designed computer processors and sold them to original equipment manufacturers (“OEMs”) to use in central processing units (“CPUs”). One of its competitors, Advanced Micro Devices Inc (“AMD”), complained to the Commission that Intel was abusing its dominant position by offering loyalty rebates to its OEMs if they purchased all or most of their processors from Intel.  The Commission agreed and imposed a €1.05 billion fine. The General Court dismissed Intel’s appeal.

On appeal, the Grand Chamber of the CJEU rejected Intel’s complaints about jurisdiction and procedural irregularities but allowed its appeal on the assessment of the rebates as abusive.  That rendered three other grounds unnecessary to consider.

There are two key points of interest arising from the judgment:

  1. Arrangements that are intended to form part of a grander anti-competitive scheme may fall within CJEU jurisdiction, even though they are relatively removed from the EEA, under the “qualified effects” route to jurisdiction.
  2. Loyalty rebates are not automatically anti-competitive; in particular, they can be saved if the undertaking can show that they could not have the effect of foreclosing an as efficient operator from the market.

The qualified effects test for jurisdiction

The Court considered two tests for jurisdiction.

  1. The “implementation test”:  Were the anticompetitive practices implemented in the EEA?
  2. The “qualified effects test”: Would the practices have foreseeable, immediate and substantial effects in the EEA?

The General Court had found that it had jurisdiction over Intel’s agreements with Lenovo (a Chinese OEM) on both tests. Intel unsuccessfully challenged the latter as a valid route to jurisdiction, and the Court’s application of both.

The Court confirmed that the “qualified effects test” is a valid route to jurisdiction.  Although the test had previously been accepted by the General Court in Gencor v Commission (T‑102/96, EU:T:1999:65) at §92, this is the first time it has been recognised by the CJEU.  It explained that it pursues the objective of “preventing conduct which, while not adopted within the EU, has anticompetitive effects liable to have an impact on the EU market”. If EU competition law were confined to the places where agreements were reached or concerted practices engaged in, it would “give undertakings an easy means of evading” Articles 101 and 102 (§§41-45).

How then is that test to be applied?  The Court provided some guidance. The question is whether “it is foreseeable that the conduct in question will have an immediate and substantial effect in the European Union”. In answering that question it is necessary to examine the undertaking’s conduct as a whole.  On the particular facts (§§51-57):

  1. The agreements with Lenovo in China had a “foreseeable” impact on competition, taking account of their “probable effects”.
  2. They had an “immediate” effect, because they formed part of an overall strategy to ensure that no Lenovo notebook equipped with an AMD CPU would be available on the market.
  3. They had a “substantial” effect on the EEA market, having regard to the whole of the conduct.

This last point is the most interesting one. Even though agreements with Lenovo for CPUs for delivery in China would by themselves have had a negligible effect, they formed part of conduct that would have a substantial effect. The Court refused to examine them in isolation on the basis that such an approach would “lead to an artificial fragmentation of comprehensive anticompetitive conduct”.

Loyalty rebates

The main substantive implications of this case arise from the findings that loyalty rebates are not always be abusive: it will depend on their scope and effect.

The purpose of Article 102 is to promote, not inhibit, competition.  So it does not protect undertakings which are not as efficient as the dominant undertaking.  Rather, it prevents illegitimate competition that pushes equally efficient undertakings out of the market.  That includes forcing purchasers to meet their requirements from the dominant undertaking.  It also includes inviting purchasers to undertake a contractual obligation to do so.  By extension, it might include incentivising purchasers to do so through loyalty rebates.  But, the Court has now made clear, that latter category is not inherently abusive.

In order to determine whether it is, it is necessary:

  1. First, to consider all the circumstances, including the level and duration of the rebates, the market shares concerned, and the needs of customers. Most importantly, the Commission must consider the capability of the rebates to foreclose an “as efficient competitor” (the “AEC test”).  That is, could the rebates force such a competitor to sell below cost price?
  2. Second, even if the rebates do have an exclusionary effect, they might still be redeemed if that effect is counterbalanced by efficiency advantages (§§139-140).

There are three key points of interest.

First, it appears that the general rule remains that loyalty rebates are abusive, unless the undertaking can produce evidence to the contrary.  The Court recounted that loyalty rebates have an anti-competitive effect, and “clarified” its case-law to say that undertakings can displace that presumption by showing that they could not have that effect in the particular case.  That brings Article 102 in line with the position under Article 101.

Second, the Court made clear that the AEC test applies generally to assessing whether conduct is an abuse of dominant position. Article 102 is not calculated to come to the aid of less efficient undertakings. Accordingly, to determine whether the practice is illegitimate, it is necessary to determine the effect it would have on a competitor who is as efficient as the dominant undertaking. That principle had been applied to attracting purchasers and excluding competitors by predatory pricing in Post Danmark v Konkurrencerådet (C‑209/10) and AKZO Chemie BV v Commission (C-62/86). In a victory for consistency, it is now clear that it applies more generally, including to attracting purchasers and excluding competitors by loyalty pricing schemes.

Third, in the context of that AEC test the Court said the inquiry was as to the “capability” of the rebates to foreclose an as efficient competitor (§§138, 141), even though Intel’s objection was that the General Court had failed to consider the “likelihood” of the rebates having that effect (§§113-114).  That is a harder task for an undertaking seeking to avoid breaching the Article 102 prohibition.  However, it is also consistent with other cases of actions with an anti-competitive object (which are less easily excused, only if they could not have that effect) rather than those with an anti-competitive effect (which are, for obvious reasons, excused if they are not likely to have that effect).

Leave a comment

Filed under Abuse, Conflicts, Economics