The Role Of Market Definition in Assessing Anticompetitive Harm in Ohio v. American Express

This article shows that the Supreme Court reached the right outcome in Ohio et al. v. American Express. The District Court had found that American Express was a two-sided transaction platform that provided joint services simultaneously to cardholders and merchants. But it then chose, by adopting a single-sided merchant services market, to analyze the effect of the anti-steering provisions at issue solely on one side of these simultaneous transactions. That decision appears to have prevented the lower court from seeing that the plaintiffs’ evidence of anticompetitive harm to merchants was weak. The District Court also decided that case law prevented it from considering the effect of the conduct on the other half of the transactions even at the second-stage of the rule-of-reason. The Supreme Court did not discuss the limitations of the plaintiffs’ theory and evidence at length but simply and properly found that the plaintiffs had failed to prove antitrust injury to platform competition for transactions. This article also shows that criticisms of the Supreme Court decision seem to be based on the rejection or misunderstanding of the economics literature on multi-sided platforms on which the District Court, the Appeals Court, and the Supreme Court all relied.

PDF – Click here for the full article.

David S. Evans, Richard Schmalensee1


In this essay, we argue that the Supreme Court reached the right outcome in Ohio et al. v. American Express.2 We explain that the single-sided market definition adopted by the District Court, despite its finding that American Express (“Amex”) was a two-sided transaction platform as described in the economics literature, effectively prevented it from seeing how weak the plaintiffs’ case was.3 The Supreme Court did not discuss the limitations of the plaintiffs’ theory and evidence at length but simply and, we believe, properly found that the plaintiffs had failed to prove antitrust injury. Many criticisms of the Supreme Court decision seem to be based on the rejection or misunderstanding of the economics literature on multi-sided platforms on which the District Court, the Appeals Court, and the Supreme Court all relied.


The U.S. Department of Justice, together with the State of Ohio and other states, brought the case against American Express in 2010. Before addressing the merits of the plaintiffs’ lawsuit, the District Court provided some background on the payment card business and the use of anti-steering provisions as well as other issues.

When a purchase is made with a general-purpose credit or charge (“GPCC”) card, the merchant pays, to a third party, a fraction of what it charges the buyer. Most of that payment, the merchant fee, goes to the firm that issued the card: American Express in the case of American Express cards, and a bank in the case of Visa or Mastercard cards. Historically, Amex has had a “spend-centric” business model: it has focused on attracting consumers who are likely to spend heavily, in part by offering more generous rewards for using its cards than Visa or Mastercard issuers had typically done. It has financed its reward programs by charging higher merchant fees than Visa or MasterCard. Visa and Mastercard issuers, in contrast, had “lend-centric” business models: they did not focus on attracting heavy spenders and made much of their money by lending to cardholders.

Despite the higher Amex merchant fees, and even though most Amex cardholders also carried one or more Mastercards and Visa cards, the Amex card was accepted at around 6.4 million U.S. merchant locations. But it was not accepted at around 3 million U.S. merchant locations that had chosen to accept Visa and MasterCard.4

Since the 1950s, Amex’s contracts with merchants that had chosen to accept its card generally prohibited merchants from using both price and non-price forms of what has come to be called “steering.”5 Steering via price would involve the merchant imposing special surcharges on purchases made with Amex cards rather than other GPCC cards.6 Non-price steering would involve the merchant trying in other ways — by pleading hardship, disparaging Amex, or posting “We Prefer Visa” signs — to persuade customers who carried and perhaps had presented their Amex cards instead to use a means of payment that was less expensive for the merchant. If they did so, customers would generally give up some standing in the Amex rewards program and possibly use a less preferred card from their standpoint. There were similar anti-steering provisions in Visa and MasterCard merchant contracts.

Beginning in the late 1980s, MasterCard and Visa mounted campaigns aimed at persuading consumers that their cards were more useful than Amex cards and persuading merchants to steer consumers toward their cards using non-price methods. These two campaigns together were effective: between 1990 and 1995, Amex’s share of GPCC volume declined from 25 percent to 20 percent. Amex responded by strengthening and enforcing the anti-steering provisions in its merchant contracts.

In 2010, the U.S. Department of Justice (“DOJ”) and several states charged that the restrictions on non-price steering in the merchant contracts of Amex, Visa, and MasterCard were unreasonable restraints of trade and thus violations of Section 1 of the Sherman Act. Independent class-action cases brought by groups of merchants challenged the restrictions on surcharging.7 The plaintiffs’ case thus had nothing to do with direct restrictions on price competition at the merchant level, and certainly nothing to do with discounts for cash or debit cards, despite some commentators’ claims to the contrary.8

An important theory of harm, in the case that was brought initially, was that the restrictions on non-price steering by the three leading systems unreasonably limited the ability of smaller GPCC systems, like Discover, to compete by charging low merchant fees. Visa and Mastercard, which accounted for 68.3 percent of GPCC volume in 2013,9 agreed to drop their restrictions on non-price steering, thus greatly weakening the plausibility of that theory as applied to Amex alone.10 The plaintiffs nonetheless persisted, as did Amex, and a seven-week trial ensued during the summer of 2014.


The District Court issued a 97-page decision in February 2015.

A. District Court Findings on Amex as a Platform

At trial, experts for both sides described Amex as a two-sided platform. The court agreed, citing a number of works from the relevant economics literature. The judge noted the existence of indirect network effects between merchants and consumers on the two sides of that platform. This ready acceptance of the economic literature on multi-sided platforms is in marked contrast to the strenuous attacks on that literature that has appeared in later commentary on this case,11 some of which we discuss below.

The judge went on to describe Amex as a two-sided transactions platform:12

…the two sides of the platform are brought together to consummate a single, simultaneous transaction, and the products provided by the platform are consumed in fixed proportions by the consumer and the merchant.

That finding became a key predicate of the Supreme Court’s decision.

Having found that Amex was a two-sided platform, the judge faced a critical choice that has been discussed at some length in the academic literature: whether to define a single market linking both sides of the platform, or to carry out the analysis working with two closely coupled markets.13 He chose to describe the GPCC business as consisting of two markets, one involving Amex and merchants and the other involving Amex and consumers. He then decided to limit consideration to the merchant side of the business in the first step of the rule-of-reason analysis and thus to consider initially whether Amex’s policies in that market had unreasonably restrained competition.14 As we will discuss below, he also decided that he could not consider any pro-competitive benefits from the consumer market in the second step of the rule-of-reason analysis.

This single-sided market approach basically precluded the court from considering the implications of its own finding that Amex was a two-sided transaction platform. It had to view the facts of the case through a lens that distorted the business reality the court itself had emphasized.

B. The Choice of Market Definition and the Rule-of-Reason to Two-Sided Platforms.

In principle, the conclusions of an economic analysis of the effects of a challenged practice by a two-sided transactions platform should be the same whether based on consideration of a single platform market or two closely coupled markets corresponding to each side. Unfortunately, the conclusions of a legal analysis under the three-step structure of rule-of-reason analysis in U.S. courts can depend critically on this choice of market definition. In particular, the single platform market definition allows consideration of all the relevant evidence and accounts for the business realities surrounding platform competition, while the side-specific platform market definitions suppress this evidence and distort business reality.

Under the rule-of-reason, plaintiffs have the initial burden of showing that challenged conduct harmed competition. If they do so, the defense has an opportunity to demonstrate pro-competitive benefits. In principle, if both sides meet their burdens, the finder of fact must balance pro- and anti-competitive effects. As a practical matter, however, if plaintiffs succeed at the first step, defendants have a very difficult task.

The American Express case illustrates why. First, it isn’t clear that the court could consider the other side-specific market in the second-stage of the rule-of-reason inquiry. The trial court judged noted that pro-competitive benefits on the consumer side, in “a separate, though intertwined antitrust market,” could not be used to offset anti-competitive effects on the merchant side.15 Second, after finding that a practice is anti-competitive in the first stage, courts seldom give much weight to pro-competitive benefits in the second stage. In this case, the judge essentially ignored the tight linkage between the two markets he had defined: Amex’s pro-competitive justifications for its conduct are not discussed until the last 14 pages of the 97-page District Court opinion.16

When a challenged practice clearly has effects on both sides of a two-sided transactions platform, as in this case, to exclude either side of the platform in the first step of the analysis is to bias the result.17 After all, the output (transactions consummated by both sides), the price of that output (paid by both sides), and the profits earned (contributed by both sides) necessarily depends on both sides. Once a court has found that a business is a two-sided transaction platform, it makes no economic sense to ignore the consequences of the challenged conduct for half of the parties to the joint transaction. And in the case of transactions platforms, the most natural way to take into account the impact of the challenged conduct on both sides of the same transaction is to define a single market for the service of connecting the two sides.18 Doing otherwise means, as a practical matter, the court ignoring pro-competitive benefits for the other interlinked side or putting little weight on this evidence.

C. What the District Court Missed from Relying on a Single-Sided Platform Market

In this case, it is instructive to suppose that the trial court had decided to take a serious look at the consumer side of the platform in the first step of the analysis and that Amex had fully availed itself of this opportunity. Amex could have made a good argument for the facial reasonability of its anti-steering provisions in light of general business practice. As one commentator asked, rhetorically, when the DOJ complaint was initially filed,19

[T]he larger question is whether … American Express, or any firm, could possibly violate the Sherman Act by telling agents that are distributing its services as well as the services of its competitors that once the customer has expressed a clear preference to use its service rather than a competing offering, the agent must accept the consumer’s preference.

In addition, if the District Court had been able to look at the platform as whole, Amex might have been able to make its free-riding argument more persuasive. After all, why would a merchant decide to accept the Amex card and then to try to persuade customers not to use it rather than simply not accepting the card unless accepting the Amex card generated incremental business? That incremental business must have resulted from investments by Amex, on which some merchants who wanted to engage in non-price steering wished to ride free.20

Amex could have gone on to note that it is common for two-sided platforms to restrict the behavior of participants on one side in order to benefit those on the other side. For instance, OpenTable terminates the accounts of diners who are no-shows four times in a 12-month period.21 This rule is an inconvenience to diners but clearly benefits restaurants. Similarly, Amex’s restrictions on non-price steering by merchants clearly benefitted its cardholders: it freed them from being hassled to give up rewards in order to lower merchants’ costs. Or from just being hassled when they’d like to pay and get out of the store. These restrictions enabled Amex to offer a more attractive product by ensuring what it called “welcome acceptance.”22

Experience abroad provides additional support for the direct consumer benefit from rules that restrict merchant steering. Australia and the United Kingdom both prohibited card networks, including American Express, from forbidding merchants from imposing surcharges which is the leading price-based steering method.23 Both found that, of the merchants who surcharged, some did so opportunistically.24 The surcharges sometimes greatly exceeded the fees merchants paid. Most troubling, some online merchants imposed these fees at the end of the check-out process as an extra fee—a practice known as “drip pricing.” Having persuaded the consumer to go through the purchase process and enter their payment details, the merchant anticipates that the surcharge at the end won’t dissuade them for completing the purchase.25

Australia revised its regulations to limit the surcharges so they could not exceed merchant fees.26 The United Kingdom has prohibited merchant surcharging altogether.27 Puerto Rico and ten U.S. states have passed legislation that prohibit merchants from imposing surcharges.28 Under the single-sided market approach, a court could not consider the possibility that American Express was prohibiting surcharges to protect its cardholders from opportunistic behavior, nor could it consider any consumer benefits that this protection provides.

To address the charge that its restrictions on non-price steering nonetheless constituted an unreasonable restraint of trade that reduced consumer welfare, Amex would stress that payment systems compete for transactions, which requires them to cater to both merchants and consumers. It would point to the many merchants that had elected not to accept the Amex card because of its high merchant fees as evidence that price competition is alive and well in the GPCC card business. It would remind the court that it did not restrict merchants’ ability to offer discounts for cash, checks, or debit cards and that their ability to charge more when more expensive payment systems were used — to surcharge — was not an issue in this case. The only competition that was suppressed by the Amex restrictions at issue was merchant jawboning aimed at the customers of a firm with a 26.4 percent share of GPCC transactions’ volume.

All of these arguments go to the heart of the question that should have been before the court at the first stage of the rule-of-reason — did the practice restrict competition among two-sided transaction platforms? — but couldn’t be considered under the single-sided definition adopted by the District Court in this or similar cases.

D. The District Court’s Evidence on Antitrust Injury in the Single-Sided Merchant Market

Plaintiffs stressed evidence that Amex had market power and that the anti-steering provisions restricted one form of non-price competition which some merchants testified that they would have employed but for those restrictions. And, as noted above, the District Court was somehow persuaded that the Amex restrictions on non-price steering had, by themselves, made it “nearly impossible” for Discover or other systems to compete on the basis of low merchant discounts.29

The rest of plaintiffs’ evidence relied on by the District Court seems to add little economic substance to this.30 Plaintiffs stressed that Amex had increased its merchant discounts substantially over the 2005-2010 period with only a slight decline in merchant acceptance, though these increases were in response to earlier increases by MasterCard and Visa. Amex was selling a differentiated product in a concentrated market, and generally offering higher consumer rewards, so price differences and price changes are hardly symptoms of competitive breakdown. Plaintiffs were unable to persuade the trial judge that Amex charged supra-competitive prices, or that it earned supra-competitive profits, or that its merchant fees were above those of Visa and MasterCard. So, in the end, plaintiffs did not provide any quantitative evidence showing a causal link between Amex’s more stringent enforcement of its ban on non-price steering and any change in market competition.

Perhaps a trial court that considered both sides of the platform in this fashion in the first step of the rule-of-reason analysis would nevertheless have found that Amex’s restrictions on non-price steering constituted an unreasonable restraint of trade and that consumer welfare was on balance reduced by it. But we think that looking at the interlinked consumers and merchants together, in a single platform market, would more likely have revealed just how weak the DOJ’s case was and would have led to a decision for Amex by the District Court. In the actual world, the District Court found that Amex had violated Section 1.31


In 2016 a three-judge panel of the Court of Appeals for the Second Circuit unanimously reversed the District Court, largely on the grounds that the correct product market definition was GPCC transactions:

The District Court’s definition of the relevant market in this case is fatal to its conclusion that Amex violated §1.32.32

It held that by looking only at services to merchants, plaintiffs had not established antitrust injury in the relevant antitrust market, the market for transactions. The Second Circuit then declined to have the full court reconsider the panel’s decision.

The State plaintiffs appealed, and the Supreme Court granted certiorari in 2017. It issued its decision in June 2018.33 In that decision, the majority began, almost exactly as the District Court and the Court of Appeals had done, by defining two-sided platforms and indirect network effects, citing much of the relevant economic literature, and concluding that Amex is a two-sided platform:

As the name implies, a two-sided platform offers different products or services to two different groups who both depend on the platform to intermediate between them.34

This description of two-sided platforms has been criticized as being over-inclusive, notably by Justice Breyer in dissent,35 and some have argued that it would enable almost any business to claim special treatment because it is two-sided.36 Like the District Court and the Court of Appeals, however, the Supreme Court majority cited relevant studies in the economics literature with more systematic, and less-inclusive, definitions. The literature cited by the Supreme Court includes a 2008 paper of ours, for instance, that offers a less-inclusive definition that is widely accepted in the economics literature:37

Two-sided platforms serve two distinct groups of customers who need each other in some way, and … provide a common (real or virtual) meeting place … to facilitate interactions between members of the two distinct customer groups.”

This context and the lower court decisions demonstrate that the majority was not departing from the now-voluminous economics literature on this point. That literature is entirely consistent with the proposition that, even though multi-sided platforms are increasingly important, many businesses, large and small, are not multi-sided.

Like the District Court and the Court of Appeals below, the majority went on to find that Amex operates a transaction platform, with indirect network effects running in both directions, providing a joint product simultaneously to two parties engaged in a transaction.38 And like the Court of Appeals, it found that given this finding it was appropriate to define a single relevant market for GPCC transactions, rather than two different markets for merchant services and consumer services.39 As with the Court of Appeals, it also found that the plaintiffs’ evidence on antitrust injury was completely wanting.

Given that market definition, the Supreme Court found that evidence on merchant fees, on which “plaintiffs stake their entire case,” was “unpersuasive” absent more.40 The majority noted the lack of evidence of supra-competitive pricing of transactions and cited the District Court’s finding that there was no reliable evidence on Amex’s transactions’ prices or profit margins, nor conclusive evidence about whether Amex charged more than its competitors.41 The Court also took note of evidence that Amex’s price had increased from 2005 to 2010, but it found, properly, that an increase in price by a single firm, not found to have engaged in supra-competitive pricing, during a period of output growth did not establish an anti-competitive effect. It went on to point to evidence of vigorous competition among networks. It affirmed the judgement of the Court of Appeals.

It is worth repeating that this outcome was largely dictated by the findings of the District Court. There wasn’t any dispute among the parties that Amex was a two-sided platform, the District Court found that it was a transaction platform, and the District Court didn’t cite credible evidence that would establish that the anti-steering provisions had caused antitrust injury. The Supreme Court decision was not like Athena, full born from the head of Zeus. It was the logical outcome of the District Court’s findings but for the trial judge’s decision to ignore the consumer side of the two-sided transaction platform.


The Supreme Court’s conclusion that the facts in American Express should be viewed through the lens of a single market for transactions, which we have endorsed above, seems to have attracted the most hostile commentary. In dissent, Justice Breyer argued at length that it is simply wrong because it aggregates complements — services to merchants and services to consumers — rather than substitutes.42 The majority’s market definition has been described by prominent commentators as “incoherent” and “economic nonsense.”43 In contrast, the economics literature generally indicates that a single-market lens may be more appropriate for use in some cases involving two-sided platforms, depending on fact patterns and analytical convenience.44

The American Express majority has not erred in treating complements as if they were substitutes for purposes of market definition. Rather, it has defined a market for the product, GPCC transactions, that is produced by the card systems by combining merchant-side and consumer-side complements in production. Antitrust markets of this sort are hardly novel. Left shoes and right shoes are plainly not substitutes in consumption. Rather, they are complements in production. They are combined to produce the product of interest to both suppliers and demanders: pairs of shoes. Similarly, engines and brakes are complements in production that are combined, along with other complementary inputs, to produce automobiles, potentially a relevant antitrust product market.

Justice Breyer said in the oral argument that the two sides were just like “nuts and bolts” and in his dissent like “tires and gasoline.”45 Professor Carlton, who has made the same point in earlier writing, has another analogy:46

Steel and rubber are used to make a golf club, but it would make no sense to claim that steel and rubber are in one market.

But there is nothing obviously wrong with defining a market for golf clubs, which, as Professor Carlton notes, are produced by combining the two complementary inputs he mentions along with other inputs. These comparisons, and the complement point, seem to willfully ignore a voluminous literature on two-sided platforms that, since the early 2000s, has recognized that the two-sides aren’t just ordinary complements.

The definition of “transaction market” adopted by the District Court and quoted above emphasizes fixity of proportions as well as simultaneity.47 Fixity of proportions is central to the examples in the preceding paragraph and other similar examples, as well as to the production of GPCC transactions. It is hardly irrelevant, as the Supreme Court majority said, that “…credit cards determine their market share by measuring the volume of transactions they have sold.”48 The plaintiffs used those same shares which of course are exactly the same from both sides of the two-sided transaction platform.

The Supreme Court found that the plaintiffs “have not carried their burden to prove anticompetitive effects in the relevant market.”49 Justice Breyer in dissent argued that as a legal matter, market definition was unnecessary if “proof of actual detrimental effects” on competition were at hand.50 We have argued above that the evidence for “actual detrimental effects” presented to the District Court was weak when considered on one-side of the platform and incomplete by refusing to consider the other side of the platform.

Apart from the facts in this case, we believe that as an economic matter, it is essential to consider market definition and, more fundamentally, market power in Sherman Act rule-of-reason cases, even if market boundaries are often blurry and market power often eludes quantification. The use of “direct evidence” to prove anti-competitive effects in American Express illustrates why.

In discussing proof of actual detrimental effects, for instance, Justice Breyer noted that American Express raised its merchant fees 20 times in five years without losing appreciable market share.51 Since Visa and MasterCard were also raising their merchant fees over the same period, it seems at least plausible that the JCB card, which issued cards in several U.S. states until 2018 but had a trivial share of GPCC card volume,52 also raised its merchant fees. Is that fact, taken alone, à la Justice Breyer without the market context, proof that the JCB card’s conduct had actual detrimental effects on competition?

Justice Breyer also pointed to testimony from numerous merchants that they would have engaged in steering but for Amex’s anti-steering restrictions.53 Suppose the JCB card’s merchant agreements also had anti-steering provisions to which some merchants objected. Would that constitute evidence that those agreements had had anti-competitive effects sufficient for a Sherman Act Section 1 violation? If the JCB card had market power, perhaps. But without more than the quantum of market power that comes from selling a differentiated product, a firm’s unilateral conduct simply cannot have any appreciable impact on competition in a relevant antitrust market. JCB’s hypothetical anti-steering provisions may be a restraint of trade in the literal sense, but without market power they simply could not be an unreasonable restraint of trade, the requirement for an antitrust offense. Thus, if Ohio v. American Express imposes a new requirement to consider market-level effects when attempting to prove anti-competitive effects from direct evidence, as some commentators have argued,54 we do not think this is a bad development.

If nothing else, Ohio v. American Express stands for the proposition that the now well-established economics of multi-sided platforms cannot be ignored in antitrust litigation. The Supreme Court and the Court of Appeals followed the District Court and found that Amex was a two-sided transactions platform, and all three decisions cited some of the voluminous relevant economics literature in support of those findings. Aside from how to treat market definition it doesn’t appear that the basic economics was controversial at all.

Nonetheless, Justice Breyer complains that “The phrase ‘two-sided transactions market’ is not one of antitrust art …”55 This seems to be correct but, in light of the history of antitrust law and policy, irrelevant. At the time the Sherman Act was enacted, and for quite some time after, modern microeconomics and industrial organization theory, including game theory, hadn’t even been developed. “Barriers to entry” was not a term of antitrust art from 1890 until sometime after the concept emerged in the economics literature in the 1950s, and the hypothetical monopolist (or SSNIP) approach to market definition was unheard of in antitrust litigation from 1890 until the publication of the 1982 merger guidelines. Over the decades, antitrust lawyers and courts have proven able to incorporate new developments in economics in pursuit of more economically rational antitrust outcomes.

Despite the volume of economics literature on multi-sided platforms that has been produced over nearly two decades, Professor Hovenkamp argues that multi-sided platform theory may be something of a fad, the implication being that courts should curb their enthusiasm for it.56 The analytical value of multi-sided platform theory is not seriously disputed among economists, however, and economic research on multi-sided platforms shows no sign of slowing after nearly two decades.

Finally, some have argued that taking the multi-sided platform literature seriously will dramatically weaken antitrust enforcement.57 Taking the correct economics into account may complicate at least some cases. But the argument for weakened, rather than more accurate enforcement is an argument that courts will be persistently confused by defendants, despite plaintiffs’ best efforts at adducing relevant economic evidence of harm to competition. In the end the case against American Express failed because the plaintiffs didn’t have any credible evidence of harm to competition. Vertical restraints can harm platform competition, and when they do, plaintiffs should be able to demonstrate that with quantitative and qualitative evidence.

In light of the substantial and growing economic importance of multi-sided platforms, it is hard to see a responsible alternative to taking seriously the economic literature that helps understand their unique characteristics. Professor Jean Tirole, Nobel Prize-winning economist and co-author of pioneering work on two-sided platforms (which he calls two-sided markets) has described essential elements of the necessary, if difficult, path forward:58

Regulators, then, will need to refrain from mechanically applying traditional principles of competition policy. When it comes to multi-sided platforms, these principles simply are not applicable in many cases. New guidelines for adapting competition policy to two-sided markets would require that both sides of the market be considered together, rather than analyzed independently, as competition authorities still sometimes do. This will require care and a new analytical approach. But this is better than misapplying traditional principles or simply treating these sectors as legal no-go zones for competition authorities.

Sound antitrust policy has always focused on market-specific competitive realities rather than just applying abstract theory. In markets with multi-sided platforms, new learning has made it clear that competitive realities often differ fundamentally from those in ordinary single-sided markets. New tools may well be necessary to apply traditional principles appropriately in markets with multi-sided platforms, but there is no reason to abandon those principles.

1 Evans is Chairman, Global Economics Group, Boston MA; and Executive Director, Jevons Institute for Competition Law and Economics, and Visiting Professor, University College London, London, UK. Schmalensee is Dean Emeritus and Howard W. Johnson Professor of Management Emeritus at the Massachusetts Institute of Technology (“MIT”) Sloan School of Management and Professor of Economics Emeritus at the MIT Department of Economics.

2 Ohio v. American Express Co. 138 S.Ct. 2274 (2018). The U.S. Department of Justice was the lead plaintiff in the cases below. It decided not to join to the States in seeking certiorari, so the States are the plaintiffs in the Supreme Court case. After cert was granted, the Justice Department ended up filing a brief for plaintiffs and participating in oral arguments.

3 The argument that the plaintiffs’ case was fatally flawed is consistent with the discussion provided by Huang, Thu & Joshua H. Soven (2018) “More Old News than New News in American Express” Antitrust Magazine (Fall 2018).

4 U.S. v. American Exp. Co. 88 F.Supp. 143, 204 (2015).

5 U.S. v. American Exp. Co. 88 F.Supp. 143, 161 (2015). American Express was launched in 1958. At the time Diners Club was the dominant GPCC system in the US. See Evans, David & Richard Schmalensee (2005) Paying With Plastic, 2nd Edition, The MIT Press, at pp. 57-59.

6 There have never been restrictions on giving discounts from list price for the use of cash or other means of payment. The Amex contracts’ restrictions on steering via price did not bar surcharging relative to cash, checks, or debit cards, only relative to other GPCC cards.

7 The merchants’ suit against MasterCard and Visa was brought in 2005 and was finally settled in January 2019. Both networks now permit surcharging with disclosure requirements and limits on the charges: see Visa, “Visa Core Rules and Visa Product and Service Rules” at pp. 338-368, available at download/about-visa/visa-rules-public.pdf; MasterCard, “MasterCard Rules” at pp. 262-269, available at In 2008, another group of merchants challenged all of Amex’s anti-steering provisions. Individual merchant cases and a putative merchant class action challenging Amex’s provisions were consolidated in 2011. In January 2019, the trial court ordered both parties to proceed to trial using the two-sided market definition in the Supreme Court Amex decision: Memorandum and Order, In Re: American Express Anti-Steering Rules Antitrust Litigation, 11-MD-2221 (NGG) (RER) January 14, 2019. On April 12, 2019, the individual merchant cases were dismissed with prejudice pursuant to a joint stipulation between the parties that settled the litigation on undisclosed terms. There is a pending motion to dismiss and compel arbitration of the class action. It seems that all of Amex’s anti-steering restrictions are still in force: American Express, “American Express US Merchant Reference Guide” at Section 3.2, available at Complicating this picture, ten states and Puerto Rico had anti-surcharge laws in effect in 2016. Several of these have been voided as impermissible restrictions on commercial speech, and the validity of others is being litigated. See National Conference of State Legislatures, “Credit and Debit Card Surcharge Statutes” October 13, 2016, available at

8 See, for instance, Hovenkamp, Herbert (2019) “Platforms and the Rule of Reason: The American Express Case,” Columbia Business Law Review (forthcoming) at pp. 1, 35, 43, 45, 77-78.

9 U.S. v. American Exp. Co. 88 F.Supp. 143, 188 (2015).

10 Nevertheless, the District Court found that Amex’s restrictions on non-price steering, the only restrictions at issue in the case, by themselves “…render it nearly impossible for a firm to enter the relevant market by offering merchants a low-cost alternative to the existing networks.” The testimony from Discover on which this finding apparently rested, however, referred to a period in which MasterCard and Visa also had restrictions barring both price and non-price steering. Id. at 213-214.

11 See Evans, David S. & Richard Schmalensee, “Two-Sided Red Herrings,” CPI Antitrust Chronicle, October 2018.

12 U.S. v. American Exp. Co. 88 F.Supp. 143, 155 (2015).

13 A useful overview of that discussion is provided by Wright, Joshua D. & John M. Yun (2019) “Burdens and Balancing in Multisided Markets: The First Principles Approach of Ohio v. American Express,” Review of Industrial Organization 54(4), pp. 717-740. See also Affeldt, Pauline, Lapo Filistrucchi, Damien Geradin & Eric Van Damme (2014) “Market Definition in Two-Sided Markets: Theory and Practice,” Journal of Competition Law and Economics 10(2) pp. 293-339 and, for a different view, Katz, Michael & Jonathan Sallet (2018) “Multisided Platforms and Antitrust Enforcement,” Yale Law Review 127(7), pp. 2142-2175.

14 After excluding debit cards and other forms of payment from the relevant market, the trial judge found Amex had a 26.4 percent share of GPCC transactions volume. Despite this relatively small share he found that Amex had sufficient market power to affect competition and proceeded to analyze the effects of the challenged conduct. U.S. v. American Exp. Co. 88 F.Supp. 143, 207 (2015). We take market power as given for the analysis below even though one could quarrel with the court’s finding. For example, the court found that Amex’s cardholders’ loyalty was “…critical to the court’s finding of market power…” even though that loyalty was, at least in large part, purchased by Amex through its generous rewards program. The Court of Appeals disagreed, holding that there is no reason to intervene because of market power that depends on rewards and prestige: U.S. v. American Exp. Co. 838 F.3d. 179, 204 (2016).

15 U.S. v. American Exp. Co. 88 F.Supp. 143, 229 (2015). It seems unsettled whether under U.S. case law it is possible to consider the benefits from a related market. Several eminent law professors who filed an amicus brief in support of the plaintiffs before the Supreme Court said it was not appropriate to do so. The U.S. Department of Justice seemed sufficiently uncertain about this that they advocated that the Court find that those benefits could be considered. See “Brief of 28 Professors of Antitrust Law as Amici Curiae Supporting Petitioners” State of Ohio, et. al., v. American Express Company, et. al, (2017) No. 16-1454 (SCOTUS); “Brief for the United States in Opposition” State of Ohio, et. al., v. American Express Company, et. al, (2017) No. 16-1454 (SCOTUS).

16 Justice Breyer, in dissent, would have allowed the benefits from the interlinked market to be considered in the second step of the rule-of-reason. He then noted the likely futility of that defense: “A Sherman Act §1 defendant can rarely, if ever, show that a pro-competitive benefit in the market for one product offsets an anticompetitive harm in the market for another.” Ohio v. American Express Co. 138 S.Ct. 2274, 2302 (2018) (Breyer, J., dissenting).

17 We have argued elsewhere that there is no reason to presume that this bias always works against the defendant since the anti-competitive harm could be felt on the side that is ignored. See Evans, David & Richard Schmalensee (2018) “Applying the Rule of Reason to Two-Sided Platform Businesses,” University of Miami Business Law Review 26(2) pp. 1-15.

18 For a more detailed discussion of this point, see Id.

19 Brown, Thomas P. (2010) “U.S. v. American Express, et al.—Failing To Make Something Out Of Nothing” Lydian Journal, available at

20 On this point, see the discussion by the Court of Appeals, U.S. v. American Exp. Co. 838 F.3d. 179, 204 (2016).

21 OpenTable, “what is your no-show policy?” available at

22 U.S. v. American Exp. Co. 88 F.Supp. 143, 156 (2015). This point was accepted by the Supreme Court: Ohio v. American Express Co. 138 S.Ct. 2274, 2289-90 (2018).

23 Reserve Bank of Australia, “A Guide to the Card Payments System Reforms” September 2010, available at

24 Reserve Bank of Australia, “Review of Card Payments Regulation” May 2016, available at; Office of Fair Trading, “Payment surcharges: Response to the Which? Super-complaint” July 2012, available at

25 See Amelia Fletcher, “Drip pricing: UK experience” Presentation to the FTC, May 21, 2012, available at economics-drip-pricing/afletcher.pdf.

26 Reserve Bank of Australia, “Review of Card Payments Regulation” May 2016, available at

27 GOV.UK, “Card surcharge ban means no more nasty surprises for shoppers” January 13, 2018, available at

28 See supra note 7.

29 See supra note 10.

30 On what follows in this paragraph, see Ohio v. American Express Co. 138 S.Ct. 2274, 2288 (2018). The District Court also noted that in the absence of surcharging or discounting, the cost of Amex’s merchant fees is paid by all consumers at merchants that accept Amex cards, even those consumers that don’t use Amex cards. It argued that this could result in a regressive subsidy from poor consumers who use cash to rich consumers who use Amex cards: U.S. v. American Exp. Co. 88 F.Supp. 143, 216-7 (2015). The claim that GPCC cards result in cash users subsidizing card users (and poor people subsidizing rich people) is often made by commentators on American Express as well as by the plaintiffs. The point isn’t as obvious as it may seem at first blush. Merchants incur significant costs from handling cash, after all, and cash-intensive and card-intensive users seem likely to tend to patronize different merchants. The unpublished paper by Schuh, Stavins & Oz, which is often cited in support of the cash/poor subsidy point demonstrates the fragility of the evidence on this point. See, Schuh, Scott, Oz Shy & Joanna Stavins “Who Gains and Who Loses from Credit Card Payments? Theory and Calibrations” Presentation at the Joint ECB/OeNB conference on The Future of Retail Payments: Opportunities and Challenges, May 12-13, 2011 at Slide 18, available at They find no cross-subsidy if the merchant pass-through rate is 50 percent or less, which is within the range of pass-through rates found in the literature, and report results based on 100 percent pass-through. For an overview of evidence on pass-through rates see Chang, Howard, David Evans & Steven Joyce (2015) “The Impact of the U.S. Debit-Card Interchange Fee Regulation on Consumer Welfare” Journal of Competition Law and Economics 11(1), pp. 23-67.

31 U.S. v. American Exp. Co. 88 F.Supp. 143, 238-9 (2015).

32 U.S. v. American Exp. Co. 838 F.3d. 179, 196 (2016).

33 Ohio v. American Express Co. 138 S.Ct. 2274 (2018).

34 Id. at 2280.

35 Id. at 2298-2300 (Breyer, J., dissenting).

36 See, e.g. Sagers, Chris (2018) “Ohio v. American Express: Clarence Thomas Sets Sail on a Sea of Doubt, and, Mirabile Dictu, It’s Still a Bad Idea,” Pro Market, June 27, 2018, available at

37 Evans, David & Richard Schmalensee (2008) “Markets With Two-Sided Platforms,” 1 Issues in Competition Law and Policy 667, emphasis added. See also Evans & Schmalensee, supra note 11, and the studies cited in note 13, supra.

38 Ohio v. American Express Co. 138 S.Ct. 2274, 2280 (2018).

39 Id. at 2287.

40 Id. at 2287.

41 Id. at 2288.

42 Id. at 2297-2301 (Breyer, J., dissenting).

43 These descriptions are from Tim Wu (2019) “The American Express Opinion, Tech Platforms & The Rule of Reason” The Journal of Antitrust Enforcement (forthcoming) and Hovenkamp, supra note 8, respectively.

44 See the articles cited in note 13, supra. Justice Breyer cites us (Evans & Schmalensee, supra note 37) for the proposition that in some cases it is appropriate to ignore linkages between the two sides of a platform (Ohio v. American Express Co. 138 S.Ct. 2274, 2300 (2018) (Breyer, J., dissenting)). We still agree with that proposition, but American Express is not one of those cases.

45 Transcript of Oral Arguments at 22, Ohio v. American Express Co., 138 S.Ct. 2274 (No. 16- 1454); Ohio v. American Express Co. 138 S.Ct. 2274, 2298 (2018) (Breyer, J., dissenting).

46 Carlton, Dennis (2019) “The Anticompetitive Effects of Vertical Most-Favored-Nations Restraints and the Error of AmexColumbia Business Law Review 2019(1) pp. 93-106 at pp. 93, 105.

47 Note 12, supra.

48 Ohio v. American Express Co. 138 S.Ct. 2274, 2286 (2018). The District Court found that transactions’ volume was the best indicator of market share: U.S. v. American Exp. Co. 88 F.Supp. 143, 189 (2015).

49 Ohio v. American Express Co. 138 S.Ct. 2274, 2287 (2018).

50 Id. at 2296 (Breyer, J., dissenting), citing Indiana Federation of Dentists, 476 U.S., at 460-61, 106 S.Ct. 2009.

51 Id. at 2293 (Breyer, J., dissenting).

52 See Wikipedia, “JCB Co., Ltd.” available at,_Ltd..

53 Ohio v. American Express Co. 138 S.Ct. 2274, 2296 (2018) (Breyer, j., dissenting).

54 See, e.g. Kully, David & Joseph Vardner (2018) “Vertical Restraints after Amex: Quietly Imposing New Burdens on Section 1 Plaintiffs” Antitrust 33(1), pp. 31-36.

55 Id. at 2298 (Breyer, J., dissenting).

56 Professor Hovenkamp, supra note 8, argues that the economic theory of multi-sided platforms may follow the trajectory of contestability theory and recede into relative obscurity. We think this very unlikely: contestability theory rested on very strong assumptions and was controversial from its inception, while the economic theory of multi-sided markets is much more robust and has been almost universally accepted among economists.

57 Compare, Wu, supra note 43, and Evans, David & Richard Schmalensee (2018) “Ignoring Two-Sided Business Reality can also Hurt Plaintiffs” CPI Antitrust Chronicle (April 2018). Professor Wu’s argument is distinct from that advanced by Kully & Vardner, supra note 54, which has nothing to do with two-sidedness.

58 Jean Tirole “Regulating the Disruptors” LiveMint, January 1, 2019, available at Professor Tirole’s pioneering work on multi-sided platforms was cited in all three American Express decisions.