The Yale Law Journal


Market Definition and Anticompetitive Effects in Ohio v. American Express

18 Jan 2021

abstract. Ohio v. American Express held that because the credit-card market is two-sided—involving a service that is consumed simultaneously by merchants and cardholders—evidence of higher transaction fees paid by merchants was insufficient to carry the government’s prima facie burden of proving anticompetitive effects in the relevant market. This Essay addresses whether the determination that a market is two-sided can appropriately guide the evaluation of relevant competitive effects. In American Express itself, the Court’s analysis reflects a sensitive consideration of competitive context and the nature of the challenged practices more than its categorical determination that the relevant market is two-sided. By the same token, competitors on one-side of a two-sided market may constrain anticompetitive behavior just as two-sided competitors do; it might often be a mistake, therefore, to ignore such competitors on the basis of market definition. Accordingly, antitrust litigants and courts should use caution before giving a platform’s strong cross-network effects—the basis for the Court’s two-sided market determination in American Express—undue emphasis in evaluating the market impact of challenged practices.


The Supreme Court’s decision in Ohio v. American Express Co.,1 which addresses the proper antitrust treatment of two-sided platforms, may be one of the most significant market-definition decisions since the Court first articulated a cross-elasticity-of-demand test for market delineation in Times-Picayune Publishing Co. v. United States,2 sixty-five years earlier.3 At issue in American Express was whether the government, by proffering evidence that merchants paid elevated fees to American Express (AmEx), had carried its burden of showing anticompetitive effects. The Court held that it had not, basing that determination on the holding that the proper definition of “credit-card market” must include “both sides of the platform—merchants and cardholders.”4 Because the market includes both sides, proof of higher prices charged to purchasers on one side of the platform failed to establish a prima faciecase of competitive harm. That result holds significant implications for antitrust litigation: it suggests that if a defendant succeeds in demonstrating that a market is two-sided, the plaintiff may face the burden of establishing not just harm to buyers on one side of the market, but net harm, taking effects on all groups of consumers into account.By the same token, a plaintiff might seek to establish a defendant’s market power by limiting relevant competitors to those that, like the defendant, compete on both sides of a two-sided market.

Given the current antitrust focus on a variety of businesses that implicate one or more two-sided platforms—including Google (with its ad tech stack) and Apple (with its App Store), to name only two—the reach of American Express is likely to be heavily litigated in the years ahead. In each case, litigants will have to consider the relative advantage of establishing a two-sided market in such cases. On the one hand, as in American Express, defendants may gain a litigation advantage if plaintiffs are required to establish net harm in a two-sided market. On the other hand, plaintiffs may gain an advantage by embracing a two-sided market definition to exclude entities that participate in only one side of the market.5

This Essay addresses whether the Court’s determination that the credit-card market is two-sided was necessary to the outcome in American Express. It is not immediately clear why the question whether a relevant market is one-sided or multi-sided should control the plaintiff’s prima facie burden of demonstrating anticompetitive effects. The Court’s conclusion that the market for credit-card services is two-sided served as shorthand for its conclusion that the plaintiffs could not establish that the challenged practice harmed competition and consumers simply by showing that it resulted in higher merchant fees, without accounting for the impact on cardholders at the front end. But the determination that the market is two-sided does not provide a satisfying answer to the question why, in general, the presence of significant indirect network effects—which the Court cited as the basis for its two-sided market definition—demands a showing of net harm.

Despite the broad and categorical tenor of parts of the Court’s opinion, American Express is best understood in the context of the vertical restraints at issue and the market context. The Court’s market-definition conclusion reflects the (in my view sound) determination that to accept proof of higher merchant fees as prima facie evidence of harm to competition from the challenged restraints would be to favor the interests of merchants over those of cardholders, without an adequate competition-policy justification. But the Court’s market-definition conclusion is best treated as dicta—the Court’s discussion of anticompetitive effects does not depend on it. Moreover, even in the context of two-sided platforms with strong indirect network effects, evidence of market effects on a single side of a two-sided platform may provide a sufficient basis for antitrust concern. What is required is careful consideration of competitive context and the nature of the challenged practice, not an approach that allocates burdens based on a categorical determination that the market is one-sided or two-sided.

In Part I of what follows, I briefly describe the course of the litigation in American Express and the Court’s two-sided market definition determination. In Part II, I describe and comment on the critique set forth in Justice Breyer’s dissent and the key questions it raises regarding the Court’s reasoning. In Part III, I explain my view that the Court’s holding was not, after all, based on its determination that the credit-card market was two-sided, but instead rested on a sensitive, contextual evaluation of the likely effects of the challenged restraints. In Part IV, I propose that in cases—unlike American Express—that rest on indirect proof of market effects based on evidence of market power and likely harm, evidence of substantial market power on one side of a two-sided platform will often suffice for purposes of meeting a plaintiff’s prima facie burden under the rule of reason.

I. the court’s two-sided market analysis

American Express involved a government challenge to “antisteering” provisions in contracts between American Express and merchants that accept AmEx cards.6 Those policies restrict merchants from encouraging customers, at the point of purchase, to use another credit card, rather than AmEx. The government’s assertion was that because merchants are unable to steer customers to credit cards that might charge lower fees, merchants were less able to put pressure on American Express to keep its merchant fees low; as a result, its fees increased.7

Vertical arrangements like the ones between American Express and its merchants are analyzed under the “rule of reason,” which demands—as a first step—that the plaintiff demonstrate proof of anticompetitive harm.8 In general, a plaintiff can do that either by showing that the defendant has substantial market power and has imposed a restriction that interferes with market competition or by establishing actual anticompetitive effects, such as an increase in price or reduction in output relative to a competitive benchmark.9 In American Express, by the time the case reached the Supreme Court, the government was not pursuing any claim—apart from evidence of increased prices—that American Express had substantial market power.10 Instead, the government was relying on proof of actual anticompetitive effects—namely, an increase in the merchant fees charged by American Express.11

The principal attack on that showing, in both the Second Circuit and in the Supreme Court, was that higher merchant fees were insufficient to show anticompetitive effects because the merchant fees represent only one side of a two-sided transaction.12 When a consumer uses a credit card, the consumer often receives benefits in the form of points or cash back—in effect, a negative price for using the card.13 To measure the price of a given credit-card transaction properly, the argument goes, one needs to consider the net price—that is, the sum of the price paid by the merchant and the (potentially negative) price incurred by the cardholder. Suppose, for example, that the cardholder receives a 1 percent cash-back payment for use of a particular credit card; suppose further that the merchant pays a fee of 2.5 percent for each transaction. It would be strange to assert that the price charged for the transaction in that example (a 1.5 percent net fee) is higher than if the merchant paid a 2-percent fee and the cardholder received no cash back (a net 2-percent fee). Treating the merchant fee in isolation seems to miss that basic point.

The district court’s response was that any benefits to cardholders are “potential pro-competitive justifications for the challenged restraints” rather than something for which the government was required to account to make a prima facie showing of anticompetitive harm.14 That difference is critical because of the allocation of burdens of proof: if the government, to establish antitrust harm, is required to show that the net price of credit-card transactions increased, the government would bear the initial burden of demonstrating the absence of sufficient offsetting benefit to cardholders—a difficult thing to do, requiring construction of a but-for world without the challenged provisions. By contrast, if the government carried its prima facie burden of establishing harm by showing that merchant fees increased, then AmEx would bear the burden of showing that the restrictions on merchants served a legitimate competitive objective, and plaintiffs would have the opportunity to show that any such objective could be achieved in a way that is less restrictive of competition.15

The Supreme Court, affirming the Second Circuit, determined that the burden of showing an increase in net price properly rested with the government, framing its conclusion around the question of market definition. The Court had noted that credit-card companies operate two-sided platforms, which “offer[] different products or services to two different groups who both depend on the platform to intermediate between them.”16 “Due to indirect network effects”—that is, the fact that the value of a platform to purchasers on one side of the platform increases with the addition of purchasers on the other side of the platform—”two-sided platforms cannot raise prices on one side without risking a feedback loop of declining demand.”17 A platform sets prices on either side based on relative price elasticity of purchasers on each side, not based on the cost of services provided. “Price increases on one side of the platform likewise do not suggest anticompetitive effects without some evidence that they have increased the overall cost of the platform’s services.”18 “Thus, courts must include both sides of the platform—merchants and cardholders—when defining the credit-card market.”19

The Court was quick to cut back on the apparent breadth of this holding, however, noting that “it is not always necessary to consider both sides of a two-sided platform,” giving the example of a newspaper.20 Though newspapers that sell ads “arguably operate a two-sided platform because the value of an advertisement increases as more people read the newspaper . . . newspaper readers are largely indifferent to the amount of advertising that a newspaper contains.”21 Because indirect network effects are weak, “the market for newspaper advertising behaves much like a one-sided market and should be analyzed as such.”22

Two-sided transaction platforms, however, are different because they “facilitate a single, simultaneous transaction.”23 “[T]he network can sell its service only if a merchant and cardholder both simultaneously choose to use the network.”24 As a result, the indirect network effects are more pronounced, and the “network must find the balance of pricing that encourages the greatest number of” transactions.25 Furthermore, “[e]valuating both sides of a two-sided transaction platform is also necessary to accurately assess competition” since “[o]nly other two-sided platforms can compete with a two-sided platform for transactions.”26 Given that market definition, evidence that the antisteering provisions increased merchant fees could not establish anticompetitive effects because increased merchant fees fail to “demonstrate anticompetitive effects on the two-sided credit-card market as a whole.”27

II. the dissent’s critique

The conclusion that increased merchant fees are insufficient to establish harm because the market is two-sided did not satisfy the dissent. The dissent considered matters from the point of view of merchants: merchants purchase credit-card services, and they want credit-card networks to compete for their business; steering encourages credit-card networks to compete.28 To be sure, the fees that credit-card networks charge merchants will affect the fees credit-card networks charge (or the benefits they pay) to cardholders, but why should that change the analysis? A credit-card network has to optimize its pricing, but it could optimize that pricing subject to the recognition that merchants have the right, as a matter of antitrust law, to discourage the use of a card that is over-priced from the merchants’ point of view.29 The dissent would have found it sufficient that the steering provisions impeded competition, leading to higher fees, in the sale of credit-card services to merchants; it would have left to AmEx the (difficult) burden of showing that any benefits to cardholders outweighed the harm to merchants.30

The majority’s conclusion that the market is two-sided is no response to the dissent’s objection. Characterizing the market as two-sided does not supply a satisfying answer to the question whether restraints that are proven to impede competition on one side of the platform should be subject to condemnation for that reason in the absence of proof—which the defendant has the burden ’to provide—that the benefit on the other side outweighs that harm. Putting the burden of establishing net harm on the plaintiff will, as a practical matter, insulate many practices from challenge, because proving what prices would be charged to one set of consumers in a but-for world is hard enough, let alone consumers on both sides of the platform.31 It is a fair criticism that the majority did not offer a satisfying reason that all two-sided platforms that display strong indirect network effects—or even all two-sided transaction platforms—should be protected by this net-harm requirement, irrespective of the nature of the challenged practice.

Furthermore, the Court did not explain why resolving the market-definition issue was necessary to the conclusion that a showing of net harm was required. The dissent contended that it was enough to show actual effects—higher merchant fees.32 The majority responded that defining the relevant market was necessary because the government was challenging a vertical restraint, and “[v]ertical restraints often pose no risk to competition unless the entity imposing them has market power, which cannot be evaluated unless the Court first defines the relevant market.”33 But the Court’s decision did not, in any obvious way, rest on a determination that American Express lacked market power. Furthermore, if the absence of market power was sufficient, as a matter of law, to foreclose liability for the challenged vertical restraint, any evidence of increased prices would have been beside the point.34

III. the court’s analysis made its two-sided market definition largely beside the point

Lower courts will also have to ponder the significance of the fact that, in explaining why the government’s evidence was insufficient, the majority did not rely on its market-definition holding, but instead addressed competitive dynamics in a way that left the two-sided nature of the market largely to one side. At the outset, the Court found that the government had failed to prove that Amex’s increased merchant fees reflected any ability to charge a supracompetitive price, given the “increases in the value of its services and the cost of its transactions,” and rejected the evidence that “Amex’s antisteering provisions are the cause of any increase in merchant fees.”35 It found that, even if merchant fees increased and cardholder rewards did not increase, there was no evidence that Amex either restricted output or charged merchants more than its competitors.36 But that conclusion, again, has nothing much to do with market definition—all agreed that output was appropriately measured by transaction volume; whether the market was one-sided or two-sided did not influence this analysis.

The Court concluded by noting signs of robust competition in the credit-card market: higher output; broader adoption, including for lower-income households; and fierce competition among networks for merchants and cardholders. And the Court noted that “there is nothing inherently anticompetitive” about antisteering provisions, which help AmEx to deliver on “welcome acceptance,” that is, its “promise of a frictionless transaction.”37 “A lack of welcome acceptance . . . makes a cardholder less likely to use Amex at all other merchants” and “undermines the investments that Amex has made to encourage increased cardholder spending, which discourages investments in rewards and ultimately harms both cardholders and merchants.”38 And “antisteering provisions do not prevent Visa, MasterCard, or Discover from competing against Amex by offering lower merchant fees or promoting their broader merchant acceptance.”39

The majority’s emphasis on the vertical nature of the restraints at issue—placed in competitive context—provides a strong justification for the result it reached. AmEx competes with Visa, MasterCard, and Discover. A merchant is, in a sense, a distributor of American Express’s services—it makes those services available to cardholders by accepting AmEx, which is not so different from a department store selling a brand of shoes to its customers.40 AmEx requires merchants that choose to accept its card to refrain from damaging its brand by communicating to AmEx cardholders that they prefer a different credit card. This preserves AmEx’s incentives to provide excellent services and benefits to its cardholders and motivates its competitors to find innovative ways to compete. Absent those restrictions, merchants would be free to undermine the very cardholder-value proposition that makes acceptance of AmEx cards attractive and valuable.41 These are of course some of the very arguments that have been traditionally offered to justify rule-of-reason treatment of vertical restraints more generally.42

Furthermore, although some commentators have disputed this,43 cardholders and merchants make a joint purchase decision to use a credit card and pay for credit-card services jointly. To allow a plaintiff to establish harm by showing only that the merchants pay higher fees risks privileging the interests of merchants over the interests of end-consumers in a way that seems in tension with the fundamental antitrust concern for consumer welfare. It may well be true—as the dissent suggested44—that, in the United States, merchant fees are higher than merchant fees in markets that have imposed regulation. But that may be the result of robust competition: the best strategy for encouraging widespread use of a credit card may be to provide benefits to cardholders paid for with higher merchant fees. The dissent cites no evidence proffered by the government that this is not the case. And, by the same token, it therefore stands to reason that if the antisteering provisions were banned, there is no assurance that the credit-card market would become more competitive—merchants might be better off, but cardholders would be correspondingly worse off.45 And if the high-merchant-fee/high-benefit model is indeed the outcome of robust competition, it is hard to justify the claim that a low-merchant-fee/low-benefit model would better promote consumer welfare.46

IV. market definition and anticompetitive effects

What does all that have to do with market definition? There is a doctrinal answer: As noted above, under the rule of reason, the plaintiff bears the burden of proving anticompetitive effects within the relevant market to establish its prima facie case; if the market is for sale of credit-card services to merchants, a plaintiff can meet its burden by showing an impact on merchants, and the impact on cardholders comes into the equation, at best, when the defendant seeks to establish a procompetitive justification for its practice. In fact, as the dissent pointed out,47 the benefit to cardholders might be out-of-bounds entirely—ordinarily, only benefits within the relevant market are considered. But the claim that benefits to cardholders from a challenged practice should be disregarded based on market definition is no more satisfying than the argument that the benefits to cardholders should be included for the same reason.

In many cases involving vertical restraints, a plaintiff will proceed by indirect proof: that is, by establishing that the defendant has market power in a relevant market and that the challenged restraint is of a type that will significantly restrain competition in that market. In making out such a case, market definition is appropriately treated as very important—often determinative—because, in that context, the market-definition exercise will identify relevant competitors. This is a necessary step in understanding the relative market dominance (or lack thereof) of the defendant and the likely competitive impact of a restraint. In that context, however, ignoring competitors that operate on only one side of a two-sided market would generally be a mistake, even if there are strong indirect network effects. Using market power to exclude competitors through exclusive dealing agreements, for example, should be treated as unlikely to be justifiable on the ground that preserving the ability to charge supracompetitive prices on one side of a two-sided platform (because no competitors can enter) is necessary to provide benefits to consumers on the other side.48 At the same time, the two-sided nature of the platform should mean that the possibility that the harm to competition on one side of the platform is outweighed by the benefit to competition and consumers on the other should not be categorically ruled out. But the burden of establishing those countervailing benefits is appropriately placed on the defendant in that circumstance.

What I have said about the burdens of proof in an indirect case may appear to be in tension with my view that American Express was correctly decided. I start, however, from the premise that, in a case where a defendant is not shown to have substantial market power, courts need to exercise caution to avoid interfering with vertical arrangements that may intensify rather than suppress competition. The antisteering provisions did not exclude any credit-card company from any store; it did not restrict any credit-card company from pursuing additional cardholders; it did not prevent any credit-card company from competing to encourage cardholder usage. To be sure, the government argued that the provisions took away one particular competitive strategy, namely, offering merchants lower rates in exchange for merchants’ discouraging the use of competitors’ cards. But the most obvious impact of privileging that strategy would be to shift promotional investment from cardholders to merchants, not to reduce the real cost of the transactional services that the credit-card industry provides.

On the other hand, for purposes of identifying market power, a dominant share on one side of a two-sided platform should suffice. AmEx generally issues AmEx cards, deals with merchants (both by contracting for services and paying the amounts charged on AmEx cards), and processes transactions. It did not have a dominant share in any of these services.49 Suppose, however, that a credit-card network (call it Acme) were organized differently, contracting with merchants and paying their charges, relying on a half-dozen competing platforms to process transactions, and relying on banks to issue cards and bill and collect from cardholders. Suppose further that Acme has a dominant share among merchants—say, sixty percent of transactions. In such a case, vertical restraints imposed on merchants should be subject to scrutiny for potential exclusionary impact on competing credit-card networks seeking to deal with those merchants, and evidence that the challenged restraints reinforce Acme’s power (preserving its ability to charge supracompetitive prices) on the merchant side should suffice to make out a prima facie case. In that situation, competition on the cardholder side will not prevent Acme from reaping profits by preserving its dominant position: for example, to the extent Acme uses its market power to raise rivals’ costs of dealing with merchants,50 it will be able to reap profits on the merchant side that other competitors cannot match, and it will not have to use all of those profits to compete in the competitive market for card issuance.

Some might object that, in the example just given, Acme is not really without power on both sides of the market: because it controls sixty percent of the transactions, it effectively has a sixty percent share of the two-sided market, even if it chooses to contract out certain transaction-processing and cardholder-side functions. But that goes to prove the point: the fact that a market is two-sided does not mean that market power on one side is insufficient to raise concern, because that power may allow the defendant to raise prices and suppress competition on both sides. Furthermore, the Acme example illustrates why proof that a practice may restrict competition on one side of the platform in the presence of market power may also suffice: even assuming that competition on one side of the platform eliminates any possibility of charging supracompetitive profits from the consumers on that side, it does not protect consumers on the other side of the platform that are subject to a seller’s dominant share.

At the same time, it would be equally wrong to ignore competitors that operate on one side of a two-sided platform when those competitors constrain the exercise of market power by the platform operator. For example, in their 2018 article (which pre-dated the Supreme Court’s decision in American Express), Jonathan Baker and Fiona Scott Morton criticize platform “most favored nations (MFN) provisions,” such as those imposed on hotels by online travel agents like Expedia, which restrict the ability of vendors to charge lower prices on other websites than they charge customers of, say, Expedia.51 Baker and Scott Morton propose that such terms may harm competition by deterring entry by competing platforms, which may be impeded in their ability to offer vendors lower commissions in exchange for those vendors’ charging lower prices. The Court’s decision in American Express might suggest that the relevant market for purposes of evaluating such an MFN is the market for online travel agents (a two-sided transaction platform). But, in selling travel services, Expedia competes not just with other two-sided platforms but also with vendors’ own websites and other aggregators or resellers, all of which can be discovered by consumers through online search tools. This competition from other vendors makes a singular focus on the impact of the challenged provisions on other two-sided platforms inappropriate. Because a travel site must compete for sales with all such outlets, legal rules and standards should be constructed to avoid condemning vertical restraints that may intensify such interbrand competition.


In evaluating a challenge to alleged exclusionary conduct affecting a two-sided platform, the conclusion that a market is two-sided does not contribute much to a clear understanding of the relevant effects that should be the focus of antitrust concern. Sometimes, practices that affect competition on one side of a two-sided transaction platform may be challenged as anticompetitive without requiring, at the threshold, an exhaustive investigation of effects on the other side. Rather, the proper antitrust question should be whether competitive conditions and the nature of the restraint are such that the failure to place the burden of proving net harm on the plaintiff in the first instance is likely to lead to false condemnation of competitive conduct. The Supreme Court correctly recognized that American Express was such a case. By the same token, excluding a competitor from the relevant market (whether that exclusion benefits the defendant or the plaintiff) simply because it competes on one side of the platform but not both will generally be unjustified. It should be the task of enforcers, plaintiffs, antitrust lawyers, and courts to conduct a similar contextual inquiry when confronted with additional platform cases in the future.

Aaron M. Panner is a partner at Kellogg, Hansen, Todd, Figel & Frederick, PLLC, where he practices antitrust law and appellate litigation. I and my firm represented American Express in Ohio v. American Express Co.; the views herein are mine alone.