This paper argues that the recent Supreme Court decision in American Express v Ohio is misguided. It is available here.

Platform competition creates challenges for antitrust, but does not warrant the upheaval of the antitrust laws that the Supreme Court’s majority opinion prescribed. Instead, the traditional rule-of-reason approach is much better suited to deal with such cases.

The paper is structured as follows:

The paper begins by providing an overview of the distinctive features of platforms and platform competition, as reflected in the platform economics’ literature.

There is no universally accepted definition of a two-sided platform, since multi-sidedness is a matter of degree. The economic literature identifies various types of platforms, such as: (a) transaction platforms, i.e. platforms that provide instrumental value by facilitating transactions between the two sides of a market; and (b) media platforms, where the two-sides of a platform comprise consumers of content and advertisers. It is sufficient here to describe a two-sided platform as a firm that (a) deals with two distinct customer groups such that (b) each group’s demand for the platform’s services depends not only on the price charged, but also on the extent of the other group’s participation on the platform. The latter point reflects the existence of indirect network effects between the two sides of a platform, which are a hallmark of two-sidedness.

There are some more general economic attributes of two-sidedness that apply to all varieties of platforms. These include:

  • Pricing is more complicated in a platform. In the context of multisided markets, one needs to deal with two price concepts: price level and price distribution (also known as price “structure”). Price level is simply the total price, and as such is sometimes referred to as the “two-sided price”. Price distribution concerns how the price level is charged between the various sides of a platform. Even holding the price level constant, adjustments to the price distribution can affect the relative participation levels on the two platform sides. However, in some cases a platform is able to accomplish this only by imposing a restriction that prevents one side from “passing through” its platform-usage price to the user on the other side, e.g. through anti-steering restraints.
  • A price change on one side of the market will produce feedback effects on the other side of the market that either magnify or counteract the initial demand response, ultimately affecting participation on both sides of the market.
  • It follows from the above that platforms attempt to induce a participation-balance on all sides of the platform that will best stimulate its overall appeal to all prospective users. Given feedback effects, pricing is governed by the ‘see-saw’ principle, i.e. a platform will optimally set prices that elicit opposite effects on participation levels across its two sides. This ultimately explains why many two-sided markets seem to have skewed price distributions, with one side paying much more than the other. For example, Google Search is free to consumers, with all revenues being obtained from advertisers.
  • Platform competition has specific characteristics that distinguish it from other competition in other markets. Such characteristics include the extent of multi-homing (i.e. whether consumers may use two or more platforms simultaneously), steering (i.e. “efforts to induce users to start using one platform instead of another), and within platform competition (i.e. in addition to competition between platforms, there may be competition between sellers on one or more sides of a platform).

Section II then explains how the distinctive features of platforms may influence the analysis of traditional anticompetitive practices.

Platforms present some distinct risks of anticompetitive conduct. A transaction platform directly mediates the dealings between its two sides, and thus has a clear mechanism by which it can influence the terms of within platform competition (e.g. by promoting collusion, as in the Apple Books cases). Platforms can also benefit the products they offer in one side of the market, in a manner akin to a vertical restraint – e.g. as in the Microsoft cases. Significantly, due to feedback effects, the adverse impact of exclusionary practices like tying or exclusive dealing can be magnified.

In evaluating a restraint imposed by a platform, the most important economic question is whether it is reasonably necessary to maintain adequate participation on both sides of the market. If one side would lose interest absent the restraint, then the restraint is presumptively procompetitive, since the platform cannot mediate transactions without dual participation. For example, Uber’s business model is an instance where restraints on seller-side competition likely enhances the volume of trade. If drivers could set their own prices, there might be aggressive price competition for riders to the point where a typical trip gives the driver only a tiny profit. This might lead many (perhaps most) drivers to determine that they cannot earn a satisfactory rate of pay and to abandon the platform. This justification for Uber’s pricing is intuitively similar to minimum resale price maintenance (RPM) clause, wherein an “upstream” manufacturer sets a lower limit on the retail prices that its downstream distributers can charge consumers. As with RPMs, this is a practice which will only be procompetitive in certain context. Likewise, exclusive dealing may help a transaction platform to solve its initial “chicken-and-egg” problem by coaxing sellers to offer their product exclusively over the new platform, thereby spurring demand on the other side since these sellers cannot be reached over other platforms.

Section III addresses the economic effects of an important category of steering restraints.

Steering restraints which inhibit parties’ ability to induce users to opt for one platform over another will only arise in two-sided markets, and are thus comparatively novel. In terms of their animating structure, steering clauses often take the form of “most-favoured nations” (MFN) provisions, which have long been the subject of more general discussion in antitrust. Although they can take many different forms, the general function of an MFN is to provide some assurance to its beneficiary that the MFN grantor will not offer better terms or treatment in its dealings with third parties, where the relevant third parties are usually competitors of the MFN’s beneficiary. Such provisions may be motivated by various considerations, both pro- and anticompetitive.

In the platform context specifically, MFN clauses are frequently intended to curb steering efforts in some way.  The restraint at issue in AmEx provides a useful example of this. Most merchants accept a range of payment cards, and their preference is for consumers to use cards that charge them lower “swipe fees.” As such, merchants have an interest in steering buyers toward low-fee cards. A natural way to do this is to apply a surcharge to purchases made with the high-fee card, or, alternatively, to offer a discount for using the low-fee alternative. However, the platform may attempt to restrain such efforts by requiring merchants to agree not to attempt to steer buyers toward other cards. For example, AmEx charges higher merchant fees than competing credit cards like Visa, while also offering cardholders larger per-transaction rewards (e.g. miles) to cardholders. The higher fees charged by Amex create an incentive for merchants to offer better retail prices to buyers who use alternative payment methods. To curb this, Amex insists on a restraint that precludes merchants from steering buyers toward rival payment platforms, or even from telling consumers about the fee disparity.

Numerous economics articles attempt to shed light on the dynamics and effects of steering restraints, most of which find that welfare effects are ambiguous. In this paper, the author argues that steering clauses can give rise to antitrust concerns. A steering restraint that merely constrains inter-platform switching does not enhance welfare simply by virtue of maintaining greater participation on its own platform, which is just one of several platform competitors. On the contrary, such conduct may soften platform competition without providing any cognizable procompetitive benefit. In such cases, the steering restraint is anticompetitive and should be condemned.

In particular, the kind of steering restraint employed by Amex may undermine competitors’ incentive to behave competitively. The usual strategic impetus for a price cut is to steal sales from one’s rival. Amex’s steering restraint undermines this process and softens inter-platform competition. Lastly, Amex’s steering restraint’s competitive footprint extends beyond the AmEx user base. By raising merchants’ costs, it leads them to charge higher prices to everyone for their goods and services. This injures all consumers, including those who do not pay with cards. Furthermore, in every instance in which a merchant is injured by the steering restraint, there must also be an injured cardholder. A merchant would only discount the price for a non-Amex purchase by up to the incremental amount by which its Amex fee exceeds that of the alternative card; but, importantly, the buyer would accept this discount only if it exceeds his incremental valuation of the Amex reward relative to that provided by the alternative card. In other words, both sides of the market lose out as a result of the anti-steering provision.

Finally, Section IV turns to the broad question of law that was at issue in AmEx.

In a holding facially about market definition, the Supreme Court held that that “both sides matter,” and that it would be inappropriate myopically to focus on only one side of the payment cards’ market. However, no party to the dispute challenged this proposition. The question at issue was merely whether the first step of the rule-of-reason analysis required a balancing of effects on two sides of the market, or whether this exercise should be pursued at a later stage. In its judgement, the Supreme Court significantly deviated from prior approaches to the rule-of-reason. Most difficult cases evaluated under the rule of reason involve potential countervailing pro- and anticompetitive effects. To address this, the courts developed a multi-stage burden-shifting framework precisely. Under this framework, a plaintiff can fulfil its initial burden without having to show that the defendant’s conduct is anticompetitive overall. Instead, a plaintiff has only to demonstrate that the conduct is prima facie anticompetitive, which is why this is merely the first stage among several.

It is indeed true that a platform’s conduct may have countervailing effects within the platform’s two sides, and that this requires courts to take the market’s two-sidedness into account. However, it does not follow that the appropriate way to deal with this is to require a plaintiff to “net out” all such considerations when making its prima facie case. That is hardly the only way to address platform commerce; it is just the most arduous way, since the defendant would be better placed to prove those efficiencies. It is also a substantial deviation from precedent. The Supreme Court engaged in a needless reconstruction of new legal principles when the old ones would do just fine. It is true that platform economics has important implications for antitrust policy and practice; but such considerations can already be accounted for, more practicably and more reliably, within the rule of reason’s existing structure.



This is an interesting, if demanding article. Its main goal is clearly to criticise the AmEx decision, which means that the analysis is not really on the effects of anticompetitive practices in platform contexts. This is particularly visible in the discussion of steering provisions, where there is barely any discussion of potential pro-competitive effects, even as they are acknowledged. The consequences of this are that the argument is somewhat hard to follow and the title of the article is a bit misleading – this review had to reconstruct the argument of the article more than I usually do.

Nonetheless, it is a piece well worth reading. I really enjoyed the initial discussion on the characteristics of platforms and of how they might affect antitrust analysis, and the discussion of the potential anticompetitive impact of anti-steering provisions.

Author Socials A weekly email with competition/antitrust updates. All opinions are mine

What do you think?

Note: Your email address will not be published