The case of Ohio v. Amex is the final phase of a long legal battle that started back in 2010 where Amex was accused of infringing section 1 of the Sherman Act for imposing its anti-steering provisions on merchants accepting Amex. While not delivering on all fronts, the case of Amex is an important one in the context of market definition where it provides for another example where a single market may be defined for two-separate but related customer groups of a two-sided platform. Beyond this, the legal analysis does not add much to what has been discussed by the CJEU in Cartes Bancaires– and certainly does not imply any immunization of two-sided platforms from competition law scrutiny.
Background to the case
The anti-steering provisions are contractual obligations that prohibit merchants that accept Amex from convincing customers to pay with different credit cards at their stores. The reason for merchants to try dissuading customers from paying with Amex lies in the business model of its scheme. For every transaction processed by Amex, merchants pay Amex a certain percentage of the total transaction sum. Part of the fees paid by merchants are then used to finance various rewards for Amex card holders in order to convince customers to sign up to Amex and to use their card as much as possible. In turn this increase in customer signup and use of Amex by cardholders is supposed to increase merchant acceptance of Amex which again will help finance better awards and keep on fueling the working of these indirect network effects between merchants and cardholders.
The fees charged by Amex from merchants were, however, higher than competing credit card schemes that rely on different business models. Accordingly, while merchants were interested in attracting Amex cardholders, they also had an incentive to dissuade them from using Amex at the point of sale to avoid the high transaction fees. Therefore, in order to prevent such practices, Amex included anti-steering provision in its contracts with merchants.
As the anti-steering provisions did not constitute a per se infringement they had to be assessed based on a three-step analysis in light of their potential anti-competitive effects and efficiencies in the identified relevant market. Although the structure of the analysis was not by any means unusual, the two-sided nature of the Amex scheme proved to pose quite a challenge, particularly with regard to the market definition.
Credit card schemes, generally speaking, are two-sided platforms that interact with two customer groups and meet the demands of these groups by facilitating interaction between the two. In practice, this interaction eventually translates into enabling monetary transactions between merchants and consumers. Accordingly, demand-side substitutability can be assessed with regard to more than one customer group, meaning that there may be more than one relevant market in such cases. Essentially, this situation comes down to defining a single relevant market for the platform of separate relevant markets with regard to the customer groups participating on the platform. The choice between these possibilities has significant consequences to the subsequent legal analysis of market power, anti-competitive effects or efficiencies (for more on this topic post on Funda).
In the case of Amex the choice between one or two relevant markets was debated in all three instances. The District Court found, based on pervious Judgment in the case of Visa/MasterCard, that the credit card market should be treated as two separate markets- one for cardholders and one for merchants [p. 171-175]. The Second Circuit reversed this finding and concluded that the credit card market in this case is to be treated as one single relevant market [p.196-200]. According to the Second Circuit this deviation from the previous findings is justified by the fact that the concerned anti-competitive practices in the case of Amex and that of Visa were different in nature. The case of Visa dealt with horizontal restraints among the credit card schemes whereas the case of Amex deals with a vertical restraint regulating the relation between Amex and the merchants that accept it. The Supreme Court, relying greatly on academic literature, confirmed the findings of the Second Circuit that the relevant market should comprise of both the cardholder and the merchant side. According to the Court, Amex is a two-sided transaction platform, which provides a single product and thus require that both sides of the platform be considered part of one relevant market.
In Europe the credit card cases also required to look into the market definition in order to assess the anti-competitive effects of the practices in each case. In Cartes Bancaires and MasterCard the Commission found that the credit card markets consisted of two separate but related relevant markets for merchants and cardholder banks (referred to as issuing market and acquiring markets). Similarly, the recent damages claims in the UK concerning Sainsbury’s and the anti-competitive practices of Visa/MasterCard equally followed this market definition of two separate but related markets.
Does this different market definition make sense?
The difference in market definitions may be explained in light of the arguments of the Second Circuit relating to the specific facts of the case. Namely, there is a difference in the type of anti-competitive restraint (horizontal v. vertical) and the level in the supply/distribution chain in which the restraint took place (merchant level v. bank level).
Accordingly, the focus of the previous credit card cases was on the relation between acquiring and issuing banks, and their choice for a certain scheme. The US case of Visa/MasterCard dealt with a prohibition imposed on issuers of Visa/MasterCard to issue cards of competing schemes Amex and Discover. The CJEU and UK cases concerned agreements that determined the level of MIFs between acquiring and issuing banks participating in the Visa/MasterCard schemes. In these cases the parities to the anti-competitive practices and effects were the (issuing and acquiring) banks and competition among the credit card schemes with regard to these parties was distorted. Thus the ‘two-sides’ of the credit card platform were in those cases the banks that adopted the concerned credit card schemes.
In the case of Amex, the anti-steering obligations do not relate to anti-competitive practices with regard to issuing or acquiring services for the Amex scheme. Instead, the restraint interferes directly with the relation between merchants and cardholders, limiting the possibility of price competition among credit card schemes for merchants to spill over to competition among schemes for cardholders.
Accordingly the ‘two sides’ of the platform in this case and their choice for adopting a certain scheme are the cardholders and the merchants and not the issuing and acquiring banks. Although both scenarios concern a two-sided platform situation, the dependency relation between the sides of the platforms in these cases is different. Banks can namely choose to be both issuing and acquiring banks (or must do so in the case of Amex) when choosing to adopt a credit card scheme. Thus, meeting the demand of such parties for credit card services is not entirely dependent on getting the other side of the platform ‘on board’, because each bank can be on both sides of the platform providing both issuing and acquiring services. In contrast, when the two sides of the credit card platform are the merchants and cardholders, such dual role is evidently not possible. Accordingly in this relation, ‘getting both sides on board’ is necessary in order to meet the demand of both parties for the credit card services, namely the possibility to transact. Consequently, demand interdependency between acquiring and issuing banks is weaker than that between merchants and cardholders which is likely also observable in differences in the intensity of indirect network effects in both scenarios. The difference in the demand interdependency between the sides of the platform in each case may constitute a solid reason to define the relevant market in two different ways.
This demand-interdependency between the sides of the platform is in essence the logic between the division among ‘transaction-’ and ‘non- transaction markets,’ as the presence of a transaction possibility indicates a higher degree of demand-interdependency between the two sides of the platform. For this same reason the market definitions for two-sided newspapers have resulted in two separate relevant markets – demand-interdependency between readers and advertisers is quite weak from the perspective of readers. Similarly the definition of a single market for more one customer group has also been done in the case of two (or multi) sided online platforms because the demand interdependency between such customer groups is significant. The lack of the ‘transaction’ aspect was not considered an impediment to defining the relevant market as the platform market in such cases (e.g. Google shopping, travel booking platforms, real estate advertising platforms). Thus, overall defining a single market in the case of Amex seems to make sense regardless of one agreeing with the notion of transaction and non-transaction two-sided markets.
Does the difference in market definition matter?
The short answer to this questions is likely yes– but it does not have to be so. The reason it matters is that in many cases the market definition will determine the scope of the legal analysis; whatever happens outside of the relevant market receives usually limited or no attention at all. Accordingly, if two markets are defined there is a risk that the analysis will only focus on one of the two, which will likely lead to erroneous findings. This was namely the reason why the CJEU case in Cartes Bancaires annulled the judgement of the General Court [paras. 74-86]. However, if both sides of the two-sided platform and the indirect network effects between the sides are accounted for with regard to both anti-competitive effects and efficiencies then the outcome may sometimes be similar despite the different market definitions. The errors in the competition law analysis occur essentially when only one side of a two-sided platform is addressed.
In this regard, the definition of a single relevant market in Amex circumvented this potential risk and allowed for a wider analysis of the pros and cons of the anti-steering rules, albeit not in a very thorough or convincing manner.
Anti-competitive effects and efficiency arguments
When addressing the existence of anti-competitive effects, the Supreme Court did not find an infringement as it concluded that the plaintiffs did not provide sufficient evidence to carry out their burden of proof. According to the Court, the observable increases in merchant fees by Amex did not demonstrate the existence of supra-competitive pricing. The increase in fees was a result of the skewed pricing structure behind the business model of Amex. The Court relied on the fact that other credit card schemes also increased their merchants’ fees during the same period and there was no evidence of a reduction of transaction output following such increases. Furthermore, the Court found that there was no proof that the anti-steering provision stifled competition between credit card companies with regard to merchant fees. On the contrary, the Court found that the business model of Amex introduced inter-brand competition in the credit card market.
Unlike the market definition, the findings of the Court in this regard are a bit hard to agree with. Admittedly, the raising merchants’ fees to finance rewards for cardholders and increase the adoption of Amex may indeed be a form of legitimate business practices in light of the two-sided nature of the card scheme which often result in a highly skewed pricing structure. Furthermore, anti-steering provisions may benefit competition by allowing another credit card business model to compete in the market and gain traction among merchants and cardholders. In a sense, such anti-steering provisions resemble the MFN clauses in online booking platforms – although they are not inherently anti-competitive and may be pro-competitive at the launching phase of a two-sided platforms, their use by a prominent undertaking in a concentrated market may interfere greatly with (price) competition among such platforms. If merchants cannot steer customers to other cards there is less incentive for credit card companies to reduce such fees and in fact such provisions may even provide incentives to raise their fees. This, in turn, may however result in many of the merchants passing on such surcharges to all consumers, regardless of the payment method.
In the EU the balance between these potential pro- and anti-competitive effects of the anti-steering provision has been settled in the sector specific regulation concerning credit/payment cards. According to Art. 11 of Regulation 2015/751, anti-steering provisions are absolutely prohibited. The potential efficiency gains of such provisions are thus considered to be outweighed by their anti-competitive effects. Considering that in the process of creating this regulation, the two-sided nature of card payment schemes was taken into account together with the market structure in area of payment systems it is unfortunate to see how the anti-steering provisions of Amex escaped competition law scrutiny without a proper substantive analysis.
That being said – the shortcomings in this case regarding anti-competitive effects are no reason to believe that the two-sided nature of a market can ever immunize undertakings from competition law scrutiny; nor should it ever be the case. The only consequence of such ‘two-sidedness’ that can be drawn for both EU and US case law is that the assessment of both efficiencies and anti-competitive effects must take account of this nature and its relevance to the case. The only ‘novelty’ introduced is that the balancing test between the anti-competitive effects and efficiencies must be observed in the context of a two-sided market rather than only on one side of such market. This is can hardly be considered to mean that two-sided platforms are immunized from competition law scrutiny as proved by the CJEU in Cartes Bancaires.
The Amex case is not as revolutionary as it was build up to be, but is important nonetheless as it confirms, once again, that at times two-sided platforms may entail the definition of a single relevant market. Unfortunately, the substantive analysis of anti-competitive effects is basically missing- the dissenting opinion does contain some aspects, however, at the same time it also seems to decline the relevance of the two-sided nature of such payment schemes. Thus, it is still up to future cases to provide guidance with regard to such an analysis when dealing with two-sided platforms/ markets.