Ohio v. American Express Co., 585 U.S. ___ (2018)

NOTICE: This opinion is subject to formal revision before publication in the preliminary print of the United States Reports. Readers are requested to notify the Reporter of Decisions, Supreme Court of the United States, Washington, D. C. 20543, of any typographical or other formal errors, in order that corrections may be made before the preliminary print goes to press.

SUPREME COURT OF THE UNITED STATES

_________________

No. 16–1454

_________________

OHIO, et al., PETITIONERS v. AMERICAN EXPRESS COMPANY, et al.

on writ of certiorari to the united states court of appeals for the second circuit

[June 25, 2018]

Justice Thomas delivered the opinion of the Court.

American Express Company and American Express Travel Related Services Company (collectively, Amex) provide credit-card services to both merchants and cardholders. When a cardholder buys something from a merchant who accepts Amex credit cards, Amex processes the transaction through its network, promptly pays the merchant, and subtracts a fee. If a merchant wants to accept Amex credit cards—and attract Amex cardholders to its business—Amex requires the merchant to agree to an antisteering contractual provision. The antisteering provision prohibits merchants from discouraging customers from using their Amex card after they have already entered the store and are about to buy something, thereby avoiding Amex’s fee. In this case, we must decide whether Amex’s antisteering provisions violate federal antitrust law. We conclude they do not.

I

A

Credit cards have become a primary way that consumers in the United States purchase goods and services. When a cardholder uses a credit card to buy something from a merchant, the transaction is facilitated by a credit-card network. The network provides separate but inter- related services to both cardholders and merchants. For cardholders, the network extends them credit, which allows them to make purchases without cash and to defer payment until later. Cardholders also can receive rewards based on the amount of money they spend, such as airline miles, points for travel, or cash back. For merchants, the network allows them to avoid the cost of processing transactions and offers them quick, guaranteed payment. This saves merchants the trouble and risk of extending credit to customers, and it increases the number and value of sales that they can make.

By providing these services to cardholders and merchants, credit-card companies bring these parties together, and therefore operate what economists call 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. See Evans & Schmalensee, Markets With Two-Sided Platforms, 1 Issues in Competition L. & Pol’y 667 (2008) (Evans & Schmalensee); Evans & Noel, Defining Antitrust Markets When Firms Operate Two-Sided Platforms, 2005 Colum. Bus. L. Rev. 667, 668 (Evans & Noel); Filistrucchi, Geradin, Van Damme, & Affeldt, Market Definition in Two-Sided Markets: Theory and Practice, 10 J. Competition L. & Econ. 293, 296 (2014) (Filistrucchi). For credit cards, that interaction is a transaction. Thus, credit-card networks are a special type of two-sided platform known as a “transaction” platform. See id., at 301, 304, 307; Evans & Noel 676–678. The key feature of transaction platforms is that they cannot make a sale to one side of the platform without simultaneously making a sale to the other. See Klein, Lerner, Murphy, & Plache, Competition in Two-Sided Markets: The Antitrust Economics of Payment Card Interchange Fees, 73 Antitrust L. J. 571, 580, 583 (2006) (Klein). For example, no credit-card transaction can occur unless both the merchant and the cardholder simultaneously agree to use the same credit-card network. See Filistrucchi 301.

Two-sided platforms differ from traditional markets in important ways. Most relevant here, two-sided platforms often exhibit what economists call “indirect network effects.” Evans & Schmalensee 667. Indirect network effects exist where the value of the two-sided platform to one group of participants depends on how many members of a different group participate. D. Evans & R. Schmalensee, Matchmakers: The New Economics of Multisided Platforms 25 (2016). In other words, the value of the services that a two-sided platform provides increases as the number of participants on both sides of the platform increases. A credit card, for example, is more valuable to cardholders when more merchants accept it, and is more valuable to merchants when more cardholders use it. See Evans & Noel 686–687; Klein 580, 584. To ensure sufficient participation, two-sided platforms must be sensitive to the prices that they charge each side. See Evans & Schmalensee 675; Evans & Noel 680; Muris, Payment Card Regulation and the (Mis)Application of the Economics of Two-Sided Markets, 2005 Colum. Bus. L. Rev. 515, 532–533 (Muris); Rochet & Tirole, Platform Competition in Two-Sided Markets, 1 J. Eur. Econ. Assn. 990, 1013 (2003). Raising the price on side A risks losing participation on that side, which decreases the value of the platform to side B. If participants on side B leave due to this loss in value, then the platform has even less value to side A—risking a feedback loop of declining demand. See Evans & Schmalensee 675; Evans & Noel 680–681. Two-sided platforms therefore must take these indirect network effects into account before making a change in price on either side. See Evans & Schmalensee 675; Evans & Noel 680–681.[1]

Sometimes indirect network effects require two-sided platforms to charge one side much more than the other. See Evans & Schmalensee 667, 675, 681, 690–691; Evans & Noel 668, 691; Klein 585; Filistrucchi 300. For two-sided platforms, “ ‘the [relative] price structure matters, and platforms must design it so as to bring both sides on board.’ ” Evans & Schmalensee 669 (quoting Rochet & Tirole, Two-Sided Markets: A Progress Report, 37 RAND J. Econ. 645, 646 (2006)). The optimal price might require charging the side with more elastic demand a below-cost (or even negative) price. See Muris 519, 550; Klein 579; Evans & Schmalensee 675; Evans & Noel 681. With credit cards, for example, networks often charge cardholders a lower fee than merchants because cardholders are more price sensitive.[2] See Muris 522; Klein 573–574, 585, 595. In fact, the network might well lose money on the cardholder side by offering rewards such as cash back, airline miles, or gift cards. See Klein 587; Evans & Schmalensee 672. The network can do this because increasing the number of cardholders increases the value of accepting the card to merchants and, thus, increases the number of merchants who accept it. Muris 522; Evans & Schmalensee 692. Networks can then charge those merchants a fee for every transaction (typically a percentage of the purchase price). Striking the optimal balance of the prices charged on each side of the platform is essential for two-sided platforms to maximize the value of their services and to compete with their rivals.

B

Amex, Visa, MasterCard, and Discover are the four dominant participants in the credit-card market. Visa, which is by far the largest, has 45% of the market as measured by transaction volume.[3] Amex and MasterCard trail with 26.4% and 23.3%, respectively, while Discover has just 5.3% of the market.

Visa and MasterCard have significant structural advantages over Amex. Visa and MasterCard began as bank cooperatives and thus almost every bank that offers credit cards is in the Visa or MasterCard network. This makes it very likely that the average consumer carries, and the average merchant accepts, Visa or MasterCard. As a result, the vast majority of Amex cardholders have a Visa or MasterCard, but only a small number of Visa and Master-Card cardholders have an Amex. Indeed, Visa and MasterCard account for more than 432 million cards in circulation in the United States, while Amex has only 53 million. And while 3.4 million merchants at 6.4 million locations accept Amex, nearly three million more locations accept Visa, MasterCard, and Discover.[4]

Amex competes with Visa and MasterCard by using a different business model. While Visa and MasterCard earn half of their revenue by collecting interest from their cardholders, Amex does not. Amex instead earns most of its revenue from merchant fees. Amex’s business model thus focuses on cardholder spending rather than card- holder lending. To encourage cardholder spending, Amex provides better rewards than other networks. Due to its superior rewards, Amex tends to attract cardholders who are wealthier and spend more money. Merchants place a higher value on these cardholders, and Amex uses this advantage to recruit merchants.

Amex’s business model has significantly influenced the credit-card market. To compete for the valuable cardholders that Amex attracts, both Visa and MasterCard have introduced premium cards that, like Amex, charge merchants higher fees and offer cardholders better rewards. To maintain their lower merchant fees, Visa and MasterCard have created a sliding scale for their various cards—charging merchants less for low-reward cards and more for high-reward cards. This differs from Amex’s strategy, which is to charge merchants the same fee no matter the rewards that its card offers. Another way that Amex has influenced the credit-card market is by making banking and card-payment services available to low-income individuals, who otherwise could not qualify for a credit card and could not afford the fees that traditional banks charge. See 2 Record 3835–3837, 4527–4529. In sum, Amex’s business model has stimulated competitive innovations in the credit-card market, increasing the volume of transactions and improving the quality of the services.

Despite these improvements, Amex’s business model sometimes causes friction with merchants. To maintain the loyalty of its cardholders, Amex must continually invest in its rewards program. But, to fund those investments, Amex must charge merchants higher fees than its rivals. Even though Amex’s investments benefit merchants by encouraging cardholders to spend more money, merchants would prefer not to pay the higher fees. One way that merchants try to avoid them, while still enticing Amex’s cardholders to shop at their stores, is by dissuading cardholders from using Amex at the point of sale. This practice is known as “steering.”

Amex has prohibited steering since the 1950s by placing antisteering provisions in its contracts with merchants. These antisteering provisions prohibit merchants from implying a preference for non-Amex cards; dissuading customers from using Amex cards; persuading customers to use other cards; imposing any special restrictions, conditions, disadvantages, or fees on Amex cards; or promoting other cards more than Amex. The antisteering provisions do not, however, prevent merchants from steering customers toward debit cards, checks, or cash.

C

In October 2010, the United States and several States (collectively, plaintiffs) sued Amex, claiming that its antisteering provisions violate §1 of the Sherman Act, 26Stat. 209, as amended, 15 U. S. C. §1.[5] After a 7-week trial, the District Court agreed that Amex’s antisteering provisions violate §1. United States v. American Express Co., 88 F. Supp. 3d 143, 151–152 (EDNY 2015). It found that the credit-card market should be treated as two separate markets—one for merchants and one for cardholders. See id., at 171–175. Evaluating the effects on the merchant side of the market, the District Court found that Amex’s antisteering provisions are anticompetitive because they result in higher merchant fees. See id., at 195–224.

The Court of Appeals for the Second Circuit reversed. United States v. American Express Co., 838 F. 3d 179, 184 (2016). It concluded that the credit-card market is one market, not two. Id., at 196–200. Evaluating the credit-card market as a whole, the Second Circuit concluded that Amex’s antisteering provisions were not anticompetitive and did not violate §1. See id., at 200–206.

We granted certiorari, 583 U. S. ___ (2017), and now affirm.

II

Section 1 of the Sherman Act prohibits “[e]very contract, combination in the form of trust or otherwise, or conspir- acy, in restraint of trade or commerce among the several States.” 15 U. S. C. §1. This Court has long recognized that, “[i]n view of the common law and the law in this country” when the Sherman Act was passed, the phrase “restraint of trade” is best read to mean “undue restraint.” Standard Oil Co. of N. J. v. United States, 221 U. S. 1, 59–60 (1911). This Court’s precedents have thus understood §1 “to outlaw only unreasonable restraints.” State Oil Co. v. Khan, 522 U. S. 3, 10 (1997) (emphasis added).

Restraints can be unreasonable in one of two ways. A small group of restraints are unreasonable per se because they “ ‘ “always or almost always tend to restrict competition and decrease output.” ’ ” Business Electronics Corp. v. Sharp Electronics Corp., 485 U. S. 717, 723 (1988). Typically only “horizontal” restraints—restraints “imposed by agreement between competitors”—qualify as unreasonable per se. Id., at 730. Restraints that are not unreasonable per se are judged under the “rule of reason.” Id., at 723. The rule of reason requires courts to conduct a fact-specific assessment of “market power and market structure . . . to assess the [restraint]’s actual effect” on competition. Copperweld Corp. v. Independence Tube Corp., 467 U. S. 752, 768 (1984). The goal is to “distinguis[h] between restraints with anticompetitive effect that are harmful to the consumer and restraints stimulating competition that are in the consumer’s best interest.” Leegin Creative Leather Products, Inc. v. PSKS, Inc., 551 U. S. 877, 886 (2007).

In this case, both sides correctly acknowledge that Amex’s antisteering provisions are vertical restraints—i.e., restraints “imposed by agreement between firms at different levels of distribution.” Business Electronics, supra, at 730. The parties also correctly acknowledge that, like nearly every other vertical restraint, the anti- steering provisions should be assessed under the rule of reason. See Leegin, supra, at 882; State Oil, supra, at 19; Business Electronics, supra, at 726; Continental T. V., Inc. v. GTE Sylvania Inc., 433 U. S. 36, 57 (1977).

To determine whether a restraint violates the rule of reason, the parties agree that a three-step, burden-shifting framework applies. Under this framework, the plaintiff has the initial burden to prove that the challenged restraint has a substantial anticompetitive effect that harms consumers in the relevant market. See 1 J. Kalinowski, Antitrust Laws and Trade Regulation §12.02[1] (2d ed. 2017) (Kalinowski); P. Areeda & H. Hovenkamp, Fundamentals of Antitrust Law §15.02[B] (4th ed. 2017) (Areeda & Hovenkamp); Capital Imaging Assoc., P. C. v. Mohawk Valley Medical Associates, Inc., 996 F. 2d 537, 543 (CA2 1993). If the plaintiff carries its burden, then the burden shifts to the defendant to show a procompetitive rationale for the restraint. See 1 Kalinowski §12.02[1]; Areeda & Hovenkamp §15.02[B]; Capital Imaging Assoc., supra, at 543. If the defendant makes this showing, then the burden shifts back to the plaintiff to demonstrate that the procompetitive efficiencies could be reasonably achieved through less anticompetitive means. See 1 Kalinowski §12.02[1]; Capital Imaging Assoc., supra, at 543.

Here, the parties ask us to decide whether the plaintiffs have carried their initial burden of proving that Amex’s antisteering provisions have an anticompetitive effect. The plaintiffs can make this showing directly or indirectly. Direct evidence of anticompetitive effects would be “ ‘proof of actual detrimental effects [on competition],’ ” FTC v. Indiana Federation of Dentists, 476 U. S. 447, 460 (1986), such as reduced output, increased prices, or decreased quality in the relevant market, see 1 Kalinowski §12.02[2]; Craftsman Limousine, Inc. v. Ford Motor Co., 491 F. 3d 381, 390 (CA8 2007); Virginia Atlantic Airways Ltd. v. British Airways PLC, 257 F. 3d 256, 264 (CA2 2001). Indirect evidence would be proof of market power plus some evidence that the challenged restraint harms competition. See 1 Kalinowski §12.02[2]; Tops Markets, Inc. v. Quality Markets, Inc., 142 F. 3d 90, 97 (CA2 1998); Spanish Broadcasting System of Fla. v. Clear Channel Communications, Inc., 376 F. 3d 1065, 1073 (CA11 2004).

Here, the plaintiffs rely exclusively on direct evidence to prove that Amex’s antisteering provisions have caused anticompetitive effects in the credit-card market.[6] To assess this evidence, we must first define the relevant market. Once defined, it becomes clear that the plaintiffs’ evidence is insufficient to carry their burden.

A

Because “[l]egal presumptions that rest on formalistic distinctions rather than actual market realities are generally disfavored in antitrust law,” Eastman Kodak Co. v. Image Technical Services, Inc., 504 U. S. 451, 466–467 (1992), courts usually cannot properly apply the rule of reason without an accurate definition of the relevant market.[7] “Without a definition of [the] market there is no way to measure [the defendant’s] ability to lessen or destroy competition.” Walker Process Equipment, Inc. v. Food Machinery & Chemical Corp., 382 U. S. 172, 177 (1965); accord, 2 Kalinowski §24.01[4][a]. Thus, the relevant market is defined as “the area of effective competition.” Ibid. Typically this is the “arena within which significant substitution in consumption or production occurs.” Areeda & Hovenkamp §5.02; accord, 2 Kalinowski §24.02[1]; United States v. Grinnell Corp., 384 U. S. 563, 571 (1966). But courts should “combin[e]” different products or services into “a single market” when “that combination reflects commercial realities.” Id., at 572; see also Brown Shoe Co. v. United States, 370 U. S. 294, 336–337 (1962) (pointing out that “the definition of the relevant market” must “ ‘correspond to the commercial realities’ of the industry”).

As explained, credit-card networks are two-sided platforms. Due to indirect network effects, two-sided platforms cannot raise prices on one side without risking a feedback loop of declining demand. See Evans & Schmalensee 674–675; Evans & Noel 680–681. And the fact that two-sided platforms charge one side a price that is below or above cost reflects differences in the two sides’ demand elasticity, not market power or anticompetitive pricing. See Klein 574, 595, 598, 626. 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. See id., at 575, 594, 626. Thus, courts must include both sides of the platform—merchants and cardholders—when defining the credit-card market.

To be sure, it is not always necessary to consider both sides of a two-sided platform. A market should be treated as one sided when the impacts of indirect network effects and relative pricing in that market are minor. See Filistrucchi 321–322. Newspapers that sell advertisements, for example, arguably operate a two-sided platform because the value of an advertisement increases as more people read the newspaper. Id., at 297, 315; Klein 579. But in the newspaper-advertisement market, the indirect networks effects operate in only one direction; newspaper readers are largely indifferent to the amount of advertising that a newspaper contains. See Filistrucchi 321, 323, and n. 99; Klein 583. Because of these weak indirect network effects, the market for newspaper advertising behaves much like a one-sided market and should be analyzed as such. See Filistrucchi 321; Times-Picayune Publishing Co. v. United States, 345 U. S. 594, 610 (1953).

But two-sided transaction platforms, like the credit-card market, are different. These platforms facilitate a single, simultaneous transaction between participants. For credit cards, the network can sell its services only if a mer- chant and cardholder both simultaneously choose to use the network. Thus, whenever a credit-card network sells one transaction’s worth of card-acceptance services to a merchant it also must sell one transaction’s worth of card-payment services to a cardholder. It cannot sell transaction services to either cardholders or merchants individually. See Klein 583 (“Because cardholders and merchants jointly consume a single product, payment card transactions, their consumption of payment card transactions must be directly proportional”). To optimize sales, the network must find the balance of pricing that encourages the greatest number of matches between cardholders and merchants.

Because they cannot make a sale unless both sides of the platform simultaneously agree to use their services, two-sided transaction platforms exhibit more pronounced indirect network effects and interconnected pricing and demand. Transaction platforms are thus better understood as “suppl[ying] only one product”—transactions. Klein 580. In the credit-card market, these transactions “are jointly consumed by a cardholder, who uses the payment card to make a transaction, and a merchant, who accepts the payment card as a method of payment.” Ibid. Tellingly, credit cards determine their market share by measuring the volume of transactions they have sold.[8]

Evaluating both sides of a two-sided transaction platform is also necessary to accurately assess competition. Only other two-sided platforms can compete with a two-sided platform for transactions. See Filistrucchi 301. A credit-card company that processed transactions for merchants, but that had no cardholders willing to use its card, could not compete with Amex. See ibid. Only a company that had both cardholders and merchants willing to use its network could sell transactions and compete in the credit-card market. Similarly, if a merchant accepts the four major credit cards, but a cardholder only uses Visa or Amex, only those two cards can compete for the particular transaction. Thus, competition cannot be accurately assessed by looking at only one side of the platform in isolation.[9]

For all these reasons, “[i]n two-sided transaction markets, only one market should be defined.” Id., at 302; see also Evans & Noel 671 (“[F]ocusing on one dimension of . . . competition tends to distort the competition that actually exists among [two-sided platforms]”). Any other analysis would lead to “ ‘ “mistaken inferences” ’ ” of the kind that could “ ‘ “chill the very conduct the antitrust laws are designed to protect.” ’ ” Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U. S. 209, 226 (1993); see also Matsushita Elec. Industrial Co. v. Zenith Radio Corp., 475 U. S. 574, 594 (1986) (“ ‘[W]e must be concerned lest a rule or precedent that authorizes a search for a particular type of undesirable pricing behavior end up by discouraging legitimate price competition’ ”); Leegin, 551 U. S., at 895 (noting that courts should avoid “increas[ing] the total cost of the antitrust system by prohibiting procompetitive conduct the antitrust laws should encourage”). Accordingly, we will analyze the two-sided market for credit-card transactions as a whole to determine whether the plaintiffs have shown that Amex’s antisteering provisions have anticompetitive effects.

B

The plaintiffs have not carried their burden to prove anticompetitive effects in the relevant market. The plaintiffs stake their entire case on proving that Amex’s agreements increase merchant fees. We find this argument unpersuasive.

As an initial matter, the plaintiffs’ argument about merchant fees wrongly focuses on only one side of the two-sided credit-card market. As explained, the credit-card market must be defined to include both merchants and cardholders. Focusing on merchant fees alone misses the mark because the product that credit-card companies sell is transactions, not services to merchants, and the competitive effects of a restraint on transactions cannot be judged by looking at merchants alone. Evidence of a price increase on one side of a two-sided transaction platform cannot by itself demonstrate an anticompetitive exercise of market power. To demonstrate anticompetitive effects on the two-sided credit-card market as a whole, the plaintiffs must prove that Amex’s antisteering provisions increased the cost of credit-card transactions above a competitive level, reduced the number of credit-card transactions, or otherwise stifled competition in the credit-card market. See 1 Kalinowski §12.02[2]; Craftsman Limousine, Inc., 491 F. 3d, at 390; Virginia Atlantic Airways Ltd., 257 F. 3d, at 264. They failed to do so.

1

The plaintiffs did not offer any evidence that the price of credit-card transactions was higher than the price one would expect to find in a competitive market. As the District Court found, the plaintiffs failed to offer any reliable measure of Amex’s transaction price or profit margins. 88 F. Supp. 3d, at 198, 215. And the evidence about whether Amex charges more than its competitors was ultimately inconclusive. Id., at 199, 202, 215.

Amex’s increased merchant fees reflect increases in the value of its services and the cost of its transactions, not an ability to charge above a competitive price. Amex began raising its merchant fees in 2005 after Visa and MasterCard raised their fees in the early 2000s. Id., at 195, 199–200. As explained, Amex has historically charged higher merchant fees than these competitors because it delivers wealthier cardholders who spend more money. Id., at 200–201. Amex’s higher merchant fees are based on a careful study of how much additional value its cardholders offer merchants. See id., at 192–193. On the other side of the market, Amex uses its higher merchant fees to offer its cardholders a more robust rewards program, which is necessary to maintain cardholder loyalty and encourage the level of spending that makes Amex valuable to merchants. Id., at 160, 191–195. That Amex allocates prices between merchants and cardholders differently from Visa and MasterCard is simply not evidence that it wields market power to achieve anticompetitive ends. See Evans & Noel 670–671; Klein 574–575, 594–595, 598, 626.

In addition, the evidence that does exist cuts against the plaintiffs’ view that Amex’s antisteering provisions are the cause of any increases in merchant fees. Visa and Master- Card’s merchant fees have continued to increase, even at merchant locations where Amex is not accepted and, thus, Amex’s antisteering provisions do not apply. See 88 F. Supp. 3d, at 222. This suggests that the cause of increased merchant fees is not Amex’s antisteering provisions, but rather increased competition for cardholders and a corresponding marketwide adjustment in the relative price charged to merchants. See Klein 575, 609.

2

The plaintiffs did offer evidence that Amex increased the percentage of the purchase price that it charges merchants by an average of 0.09% between 2005 and 2010 and that this increase was not entirely spent on cardholder rewards. See 88 F. Supp. 3d, at 195–197, 215. The plaintiffs believe that this evidence shows that the price of Amex’s transactions increased.

Even assuming the plaintiffs are correct, this evidence does not prove that Amex’s antisteering provisions gave it the power to charge anticompetitive prices. “Market power is the ability to raise price profitably by restricting output.” Areeda & Hovenkamp §5.01 (emphasis added); accord, Kodak, 504 U. S., at 464; Business Electronics, 485 U. S., at 723. This Court will “not infer competitive injury from price and output data absent some evidence that tends to prove that output was restricted or prices were above a competitive level.” Brooke Group Ltd., 509 U. S., at 237. There is no such evidence in this case. The output of credit-card transactions grew dramatically from 2008 to 2013, increasing 30%. See 838 F. 3d, at 206. “Where . . . output is expanding at the same time prices are increasing, rising prices are equally consistent with growing product demand.” Brooke Group Ltd., supra, at 237. And, as previously explained, the plaintiffs did not show that Amex charged more than its competitors.

3

The plaintiffs also failed to prove that Amex’s antisteering provisions have stifled competition among credit-card companies. To the contrary, while these agreements have been in place, the credit-card market experienced expanding output and improved quality. Amex’s business model spurred Visa and MasterCard to offer new premium card categories with higher rewards. And it has increased the availability of card services, including free banking and card-payment services for low-income customers who otherwise would not be served. Indeed, between 1970 and 2001, the percentage of households with credit cards more than quadrupled, and the proportion of households in the bottom-income quintile with credit cards grew from just 2% to over 38%. See D. Evans & R. Schmalensee, Paying With Plastic: The Digital Revolution in Buying and Borrowing 88–89 (2d ed. 2005) (Paying With Plastic).

Nor have Amex’s antisteering provisions ended competition between credit-card networks with respect to merchant fees. Instead, fierce competition between networks has constrained Amex’s ability to raise these fees and has, at times, forced Amex to lower them. For instance, when Amex raised its merchant prices between 2005 and 2010, some merchants chose to leave its network. 88 F. Supp. 3d, at 197. And when its remaining merchants complained, Amex stopped raising its merchant prices. Id., at 198. In another instance in the late 1980s and early 1990s, competition forced Amex to offer lower merchant fees to “everyday spend” merchants—supermarkets, gas stations, pharmacies, and the like—to persuade them to accept Amex. See id., at 160–161, 202.

In addition, Amex’s competitors have exploited its higher merchant fees to their advantage. By charging lower merchant fees, Visa, MasterCard, and Discover have achieved broader merchant acceptance—approximately 3 million more locations than Amex. Id., at 204. This broader merchant acceptance is a major advantage for these networks and a significant challenge for Amex, since consumers prefer cards that will be accepted everywhere. Ibid. And to compete even further with Amex, Visa and MasterCard charge different merchant fees for different types of cards to maintain their comparatively lower merchant fees and broader acceptance. Over the long run, this competition has created a trend of declining merchant fees in the credit-card market. In fact, since the first credit card was introduced in the 1950s, merchant fees—including Amex’s merchant fees—have decreased by more than half. See id., at 202–203; Paying With Plastic 54, 126, 152.

Lastly, there is nothing inherently anticompetitive about Amex’s antisteering provisions. These agreements actually stem negative externalities in the credit-card market and promote interbrand competition. When merchants steer cardholders away from Amex at the point of sale, it undermines the cardholder’s expectation of “welcome acceptance”—the promise of a frictionless transaction. 88 F. Supp. 3d, at 156. A lack of welcome acceptance at one merchant makes a cardholder less likely to use Amex at all other merchants. This externality endangers the viability of the entire Amex network. And it undermines the investments that Amex has made to encourage increased cardholder spending, which discourages investments in rewards and ultimately harms both cardholders and merchants. Cf. Leegin, 551 U. S., at 890–891 (recognizing that vertical restraints can prevent retailers from free riding and thus increase the availability of “tangible or intangible services or promotional efforts” that enhance competition and consumer welfare). Perhaps most importantly, antisteering provisions do not prevent Visa, MasterCard, or Discover from competing against Amex by offering lower merchant fees or promoting their broader merchant acceptance.[10]

In sum, the plaintiffs have not satisfied the first step of the rule of reason. They have not carried their burden of proving that Amex’s antisteering provisions have anticompetitive effects. Amex’s business model has spurred robust interbrand competition and has increased the quality and quantity of credit-card transactions. And it is “[t]he promotion of interbrand competition,” after all, that “is . . . ‘the primary purpose of the antitrust laws.’ ” Id., at 890.

*  *  *

Because Amex’s antisteering provisions do not unreasonably restrain trade, we affirm the judgment of the Court of Appeals.

It is so ordered.

SUPREME COURT OF THE UNITED STATES

_________________

No. 16–1454

_________________

OHIO, et al., PETITIONERS v. AMERICAN EXPRESS COMPANY, et al.

on writ of certiorari to the united states court of appeals for the second circuit

[June 25, 2018]

Justice Breyer, with whom Justice Ginsburg, Justice Sotomayor, and Justice Kagan join, dissenting.

For more than 120 years, the American economy has prospered by charting a middle path between pure lassez-faire and state capitalism, governed by an antitrust law “dedicated to the principle that markets, not individual firms and certainly not political power, produce the optimal mixture of goods and services.” 1 P. Areeda & H. Hovenkamp, Antitrust Law ¶100b, p. 4 (4th ed. 2013) (Areeda & Hovenkamp). By means of a strong antitrust law, the United States has sought to avoid the danger of monopoly capitalism. Long gone, we hope, are the days when the great trusts presided unfettered by competition over the American economy.

This lawsuit is emblematic of the American approach. Many governments around the world have responded to concerns about the high fees that credit-card companies often charge merchants by regulating such fees directly. See GAO, Credit and Debit Cards: Federal Entities Are Taking Actions to Limit Their Interchange Fees, but Additional Revenue Collection Cost Savings May Exist 31–35 (GAO–08–558, 2008). The United States has not followed that approach. The Government instead filed this lawsuit, which seeks to restore market competition over credit-card merchant fees by eliminating a contract-ual barrier with anticompetitive effects. The majority rejects that effort. But because the challenged contractual term clearly has serious anticompetitive effects, I dissent.

I

I agree with the majority and the parties that this case is properly evaluated under the three-step “rule of reason” that governs many antitrust lawsuits. Ante, at 9–10. Under that approach, a court looks first at the agreement or restraint at issue to assess whether it has had, or is likely to have, anticompetitive effects. FTC v. Indiana Federation of Dentists, 476 U. S. 447, 459 (1986). In doing so, the court normally asks whether the restraint may tend to impede competition and, if so, whether those who have entered into that restraint have sufficient economic or commercial power for the agreement to make a negative difference. See id., at 459–461. Sometimes, but not always, a court will try to determine the appropriate market (the market that the agreement affects) and determine whether those entering into that agreement have the power to raise prices above the competitive level in that market. See ibid.

It is important here to understand that in cases under §1 of the Sherman Act (unlike in cases challenging a merger under §7 of the Clayton Act, 15 U. S. C. §18), it may well be unnecessary to undertake a sometimes complex, market power inquiry:

“Since the purpose [in a Sherman Act §1 case] of the inquiries into . . . market power is [simply] to determine whether an arrangement has the potential for genuine adverse effects on competition, ‘proof of actual detrimental effects, such as a reduction in output,’ can obviate the need for an inquiry into market power, which is but a ‘surrogate for detrimental effects.’ ” Indiana Federation of Dentists, supra, at 460–461 (quoting 7 P. Areeda, Antitrust Law ¶1511, p. 429 (3d ed. 1986)).

Second (as treatise writers summarize the case law), if an antitrust plaintiff meets the initial burden of showing that an agreement will likely have anticompetitive effects, normally the “burden shifts to the defendant to show that the restraint in fact serves a legitimate objective.” 7 Areeda & Hovenkamp ¶1504b, at 415; see California Dental Assn. v. FTC, 526 U. S. 756, 771 (1999); id., at 788 (Breyer, J., dissenting).

Third, if the defendant successfully bears this burden, the antitrust plaintiff may still carry the day by showing that it is possible to meet the legitimate objective in less restrictive ways, or, perhaps by showing that the legitimate objective does not outweigh the harm that competition will suffer, i.e., that the agreement “on balance” remains unreasonable. 7 Areeda & Hovenkamp ¶1507a, at 442.

Like the Court of Appeals and the parties, the majority addresses only the first step of that three-step framework. Ante, at 10.

II

A

This case concerns the credit-card business. As the majority explains, ante, at 2, that business involves the selling of two different but related card services. First, when a shopper uses a credit card to buy something from a participating merchant, the credit-card company pays the merchant the amount of money that the merchant’s customer has charged to his card and charges the merchant a fee, say 5%, for that speedy-payment service. I shall refer to that kind of transaction as a merchant-related card service. Second, the credit-card company then sends a bill to the merchant’s customer, the shopper who holds the card; and the shopper pays the card company the sum that merchant charged the shopper for the goods or services he or she bought. The cardholder also often pays the card company a fee, such as an annual fee for the card or an interest charge for delayed payment. I shall call that kind of transaction a shopper-related card service. The credit-card company can earn revenue from the sale (directly or indirectly) of each of these services: (1) speedy payment for merchants, and (2) credit for shoppers. (I say “indirectly” to reflect the fact that card companies often create or use networks of banks as part of the process—but I have found nothing here suggesting that that fact makes a significant difference to my analysis.)

Sales of the two basic card services are related. A shopper can pay for a purchase with a particular credit card only if the merchant has signed up for merchant-related card services with the company that issued the credit card that the shopper wishes to use. A firm in the credit-card business is therefore unlikely to make money unless quite a few merchants agree to accept that firm’s card and quite a few shoppers agree to carry and use it. In general, the more merchants that sign up with a particular card company, the more useful that card is likely to prove to shoppers and so the more shoppers will sign up; so too, the more shoppers that carry a particular card, the more useful that card is likely to prove to merchants (as it obviously helps them obtain the shoppers’ business) and so the more merchants will sign up. Moreover, as a rough rule of thumb (and assuming constant charges), the larger the networks of paying merchants and paying shoppers that a card firm maintains, the larger the revenues that the firm will likely receive, since more payments will be processed using its cards. Thus, it is not surprising that a card company may offer shoppers incentives (say, points redeemable for merchandise or travel) for using its card or that a firm might want merchants to accept its card exclusively.

B

This case focuses upon a practice called “steering.” American Express has historically charged higher merchant fees than its competitors. App. to Pet. for Cert. 173a–176a. Hence, fewer merchants accept American Express’ cards than its competitors’. Id., at 184a–187a. But, perhaps because American Express cardholders are, on average, wealthier, higher-spending, or more loyal to American Express than other cardholders, vast numbers of merchants still accept American Express cards. See id., at 156a, 176a–177a, 184a–187a. Those who do, however, would (in order to avoid the higher American Express fee) often prefer that their customers use a different card to charge a purchase. Thus, the merchant has a monetary incentive to “steer” the customer towards the use of a different card. A merchant might tell the customer, for example, “American Express costs us more,” or “please use Visa if you can,” or “free shipping if you use Discover.” See id., at 100a–102a.

Steering makes a difference, because without it, the shopper does not care whether the merchant pays more to American Express than it would pay to a different card company—the shopper pays the same price either way. But if steering works, then American Express will find it more difficult to charge more than its competitors for merchant-related services, because merchants will respond by steering their customers, encouraging them to use other cards. Thus, American Express dislikes steering; the merchants like it; and the shoppers may benefit from it, whether because merchants will offer them incentives to use less expensive cards or in the form of lower retail prices overall. See id., at 92a, 97a–104a.

In response to its competitors’ efforts to convince merchants to steer shoppers to use less expensive cards, American Express tried to stop, or at least to limit, steering by placing antisteering provisions in most of its contracts with merchants. It called those provisions “nondiscrimination provisions.” They prohibited steering of the forms I have described above (and others as well). See id., at 95a–96a, 100a–101a. After placing them in its agreements, American Express found it could maintain, or even raise, its higher merchant prices without losing too many transactions to other firms. Id., at 195a–198a. These agreements—the “nondiscrimination provisions”—led to this lawsuit.

C

In 2010 the United States and 17 States brought this antitrust case against American Express. They claimed that the “nondiscrimination provisions” in its contracts with merchants created an unreasonable restraint of trade. (Initially Visa and MasterCard were also defendants, but they entered into consent judgments, dropping similar provisions from their contracts with merchants). After a 7-week bench trial, the District Court entered judgment for the Government, setting forth its findings of fact and conclusions of law in a 97-page opinion. 88 F. Supp. 3d 143 (EDNY 2015).

Because the majority devotes little attention to the District Court’s detailed factual findings, I will summarize some of the more significant ones here. Among other things, the District Court found that beginning in 2005 and during the next five years, American Express raised the prices it charged merchants on 20 separate occasions. See id., at 195–196. In doing so, American Express did not take account of the possibility that large merchants would respond to the price increases by encouraging shoppers to use a different credit card because the nondiscrimination provisions prohibited any such steering. Id., at 215. The District Court pointed to merchants’ testimony stating that, had it not been for those provisions, the large merchants would have responded to the price increases by encouraging customers to use other, less-expensive cards. Ibid.

The District Court also found that even though American Express raised its merchant prices 20 times in this 5-year period, it did not lose the business of any large merchant. Id., at 197. Nor did American Express increase benefits (or cut credit-card prices) to American Express cardholders in tandem with the merchant price increases. Id., at 196. Even had there been no direct evidence of injury to competition, American Express’ ability to raise merchant prices without losing any meaningful market share, in the District Court’s view, showed that American Express possessed power in the relevant market. See id., at 195.

The District Court also found that, in the absence of the provisions, prices to merchants would likely have been lower. Ibid. It wrote that in the late 1990’s, Discover, one of American Express’ competitors, had tried to develop a business model that involved charging lower prices to merchants than the other companies charged. Id., at 213. Discover then invited each “merchant to save money by shifting volume to Discover,” while simultaneously offering merchants additional discounts “if they would steer customers to Discover.” Ibid. The court determined that these efforts failed because of American Express’ (and the other card companies’) “nondiscrimination provisions.” These provisions, the court found, “denied merchants the ability to express a preference for Discover or to employ any other tool by which they might steer share to Discover’s lower-priced network.” Id., at 214. Because the provisions eliminated any advantage that lower prices might produce, Discover “abandoned its low-price business model” and raised its merchant fees to match those of its competitors. Ibid. This series of events, the court concluded was “emblematic of the harm done to the competitive process” by the “nondiscrimination provisions.” Ibid.

The District Court added that it found no offsetting pro-competitive benefit to shoppers. Id., at 225–238. Indeed, it found no offsetting benefit of any kind. See ibid.

American Express appealed, and the U. S. Court of Appeals for the Second Circuit held in its favor. 838 F. 3d 179 (2016). The Court of Appeals did not reject any fact found by the District Court as “clearly erroneous.” See Fed. Rule Civ. Proc. 52(a)(6). Rather, it concluded that the District Court had erred in step 1 of its rule-of-reason analysis by failing to account for what the Second Circuit called the credit-card business’s “two-sided market” (or “two-sided platform”). 838 F. 3d, at 185–186, 196–200.

III

The majority, like the Court of Appeals, reaches only step 1 in its “rule of reason” analysis. Ante, at 10. To repeat, that step consists of determining whether the challenged “nondiscrimination provisions” have had, or are likely to have, anticompetitive effects. See Indiana Federation of Dentists, 476 U. S., at 459. Do those provisions tend to impede competition? And if so, does American Express, which imposed that restraint as a condition of doing business with its merchant customers, have sufficient economic or commercial power for the provision to make a negative difference? See id., at 460–461.

A

Here the District Court found that the challenged provisions have had significant anticompetitive effects. In particular, it found that the provisions have limited or prevented price competition among credit-card firms for the business of merchants. 88 F. Supp. 3d, at 209. That conclusion makes sense: In the provisions, American Express required the merchants to agree not to encourage customers to use American Express’ competitors’ credit cards, even cards from those competitors, such as Discover, that intended to charge the merchants lower prices. See id., at 214. By doing so, American Express has “disrupt[ed] the normal price-setting mechanism” in the market. Id., at 209. As a result of the provisions, the District Court found, American Express was able to raise merchant prices repeatedly without any significant loss of business, because merchants were unable to respond to such price increases by encouraging shoppers to pay with other cards. Id., at 215. The provisions also meant that competitors like Discover had little incentive to lower their merchant prices, because doing so did not lead to any additional market share. Id., at 214. The provisions thereby “suppress[ed] [American Express’] . . . competitors’ incentives to offer lower prices . . . resulting in higher profit-maximizing prices across the network services market.” Id., at 209. Consumers throughout the economy paid higher retail prices as a result, and they were denied the opportunity to accept incentives that merchants might otherwise have offered to use less-expensive cards. Id., at 216, 220. I should think that, considering step 1 alone, there is little more that need be said.

The majority, like the Court of Appeals, says that the District Court should have looked not only at the market for the card companies’ merchant-related services but also at the market for the card companies’ shopper-related services, and that it should have combined them, treating them as a single market. Ante, at 14–15; 838 F. 3d, at 197. But I am not aware of any support for that view in antitrust law. Indeed, this Court has held to the contrary.

In Times-Picayune Publishing Co. v. United States, 345 U. S. 594, 610 (1953), the Court held that an antitrust court should begin its definition of a relevant market by focusing narrowly on the good or service directly affected by a challenged restraint. The Government in that case claimed that a newspaper’s advertising policy violated the Sherman Act’s “rule of reason.” See ibid. In support of that argument, the Government pointed out, and the District Court had held, that the newspaper dominated the market for the sales of newspapers to readers in New Orleans, where it was the sole morning daily newspaper. Ibid. But this Court reversed. We explained that “every newspaper is a dual trader in separate though interdependent markets; it sells the paper’s news and advertising content to its readers; in effect that readership is in turn sold to the buyers of advertising space.” Ibid. We then added:

“This case concerns solely one of those markets. The Publishing Company stands accused not of tying sales to its readers but only to buyers of general and classified space in its papers. For this reason, dominance in the advertising market, not in readership, must be decisive in gauging the legality of the Company’s unit plan.” Ibid.

Here, American Express stands accused not of limiting or harming competition for shopper-related card services, but only of merchant-related card services, because the challenged contract provisions appear only in American Express’ contracts with merchants. That is why the District Court was correct in considering, at step 1, simply whether the agreement had diminished competition in merchant-related services.

B

The District Court did refer to market definition, and the majority does the same. Ante, at 11–15. And I recognize that properly defining a market is often a complex business. Once a court has identified the good or service directly restrained, as Times-Picayune Publishing Co. requires, it will sometimes add to the relevant market what economists call “substitutes”: other goods or services that are reasonably substitutable for that good or service. See, e.g., United States v. E. I. du Pont de Nemours & Co., 351 U. S. 377, 395–396 (1956) (explaining that cellophane market includes other, substitutable flexible wrapping materials as well). The reason that substitutes are included in the relevant market is that they restrain a firm’s ability to profitably raise prices, because customers will switch to the substitutes rather than pay the higher prices. See 2B Areeda & Hovenkamp ¶561, at 378.

But while the market includes substitutes, it does not include what economists call complements: goods or services that are used together with the restrained product, but that cannot be substituted for that product. See id., ¶565a, at 429; Eastman Kodak Co. v. Image Technical Services, Inc., 504 U. S. 451, 463 (1992). An example of complements is gasoline and tires. A driver needs both gasoline and tires to drive, but they are not substitutes for each other, and so the sale price of tires does not check the ability of a gasoline firm (say a gasoline monopolist) to raise the price of gasoline above competitive levels. As a treatise on the subject states: “Grouping complementary goods into the same market” is “economic nonsense,” and would “undermin[e] the rationale for the policy against monopolization or collusion in the first place.” 2B Areeda & Hovenkamp ¶565a, at 431.

Here, the relationship between merchant-related card services and shopper-related card services is primarily that of complements, not substitutes. Like gasoline and tires, both must be purchased for either to have value. Merchants upset about a price increase for merchant-related services cannot avoid that price increase by becoming cardholders, in the way that, say, a buyer of newspaper advertising can switch to television advertising or direct mail in response to a newspaper’s advertising price increase. The two categories of services serve fundamentally different purposes. And so, also like gasoline and tires, it is difficult to see any way in which the price of shopper-related services could act as a check on the card firm’s sale price of merchant-related services. If anything, a lower price of shopper-related card services is likely to cause more shoppers to use the card, and increased shopper popularity should make it easier for a card firm to raise prices to merchants, not harder, as would be the case if the services were substitutes. Thus, unless there is something unusual about this case—a possibility I discuss below, see infra, at 13–20—there is no justification for treating shopper-related services and merchant-related services as if they were part of a single market, at least not at step 1 of the “rule of reason.”

C

Regardless, a discussion of market definition was legally unnecessary at step 1. That is because the District Court found strong direct evidence of anticompetitive effects flowing from the challenged restraint. 88 F. Supp. 3d, at 207–224. As I said, supra, at 7, this evidence included Discover’s efforts to break into the credit-card business by charging lower prices for merchant-related services, only to find that the “nondiscrimination provisions,” by preventing merchants from encouraging shoppers to use Discover cards, meant that lower merchant prices did not result in any additional transactions using Discover credit cards. 88 F. Supp. 3d, at 213–214. The direct evidence also included the fact that American Express raised its merchant prices 20 times in five years without losing any appreciable market share. Id., at 195–198, 208–212. It also included the testimony of numerous merchants that they would have steered shoppers away from American Express cards in response to merchant price increases (thereby checking the ability of American Express to raise prices) had it not been for the nondiscrimination provisions. See id., at 221–222. It included the factual finding that American Express “did not even account for the possibility that [large] merchants would respond to its price increases by attempting to shift share to a competitor’s network” because the nondiscrimination provisions prohibited steering. Id., at 215. It included the District Court’s ultimate finding of fact, not overturned by the Court of Appeals, that the challenged provisions “were integral to” American Express’ “[price] increases and thereby caused merchants to pay higher prices.” Ibid.

As I explained above, this Court has stated that “[s]ince the purpose of the inquiries into market definition and market power is to determine whether an arrangement has the potential for genuine adverse effects on competition, proof of actual detrimental effects . . . can obviate the need for” those inquiries. Indiana Federation of Dentists, 476 U. S., at 460–461 (internal quotation marks omitted). That statement is fully applicable here. Doubts about the District Court’s market-definition analysis are beside the point in the face of the District Court’s findings of actual anticompetitive harm.

The majority disagrees that market definition is irrelevant. See ante, at 11–12, and n. 7. The majority explains that market definition is necessary because the nondiscrimination provisions are “vertical restraints” 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 determines the relevant market.” Ante, at 11, n. 7. The majority thus, in a footnote, seems categorically to exempt vertical restraints from the ordinary “rule of reason” analysis that has applied to them since the Sherman Act’s enactment in 1890. The majority’s only support for this novel exemption is Leegin Creative Leather Products, Inc. v. PSKS, Inc., 551 U. S. 877 (2007). But Leegin held that the “rule of reason” applied to the vertical restraint at issue in that case. See id., at 898–899. It said nothing to suggest that vertical restraints are not subject to the usual “rule of reason” analysis. See also infra, at 24.

One critical point that the majority’s argument ignores is that proof of actual adverse effects on competition is, a fortiori, proof of market power. Without such power, the restraints could not have brought about the anticompetitive effects that the plaintiff proved. See Indiana Federation of Dentists, supra, at 460 (“[T]he purpose of the inquiries into market definition and market power is to determine whether an arrangement has the potential for genuine adverse effects on competition” (emphasis added)). The District Court’s findings of actual anticompetitive harm from the nondiscrimination provisions thus showed that, whatever the relevant market might be, American Express had enough power in that market to cause that harm. There is no reason to require a separate showing of market definition and market power under such circumstances. And so the majority’s extensive discussion of market definition is legally unnecessary.

D

The majority’s discussion of market definition is also wrong. Without raising any objection in general with the longstanding approach I describe above, supra, at 10–11, the majority agrees with the Court of Appeals that the market for American Express’ card services is special because it is a “two-sided transaction platform.” Ante, at 2–5, 12–15. The majority explains that credit-card firms connect two distinct groups of customers: First, merchants who accept credit cards, and second, shoppers who use the cards. Ante, at 2; accord, 838 F. 3d, at 186. The majority adds that “no credit-card transaction can occur unless both the merchant and the cardholder simultaneously agree to use to the same credit-card network.” Ante, at 3. And it explains that the credit-card market involves “indirect network effects,” by which it means that shoppers want a card that many merchants will accept and merchants want to accept those cards that many customers have and use. Ibid. From this, the majority concludes that “courts must include both sides of the platform—merchants and cardholders—when defining the credit-card market.” Ante, at 12; accord, 838 F. 3d, at 197.

1

Missing from the majority’s analysis is any explanation as to why, given the purposes that market definition serves in antitrust law, the fact that a credit-card firm can be said to operate a “two-sided transaction platform” means that its merchant-related and shopper-related services should be combined into a single market. The phrase “two-sided transaction platform” is not one of antitrust art—I can find no case from this Court using those words. The majority defines the phrase as covering a business that “offers different products or services to two different groups who both depend on the platform to intermediate between them,” where the business “cannot make a sale to one side of the platform without simultaneously making a sale to the other” side of the platform. Ante, at 2. I take from that definition that there are four relevant features of such businesses on the majority’s account: they (1) offer different products or services, (2) to different groups of customers, (3) whom the “platform” connects, (4) in simultaneous transactions. See ibid.

What is it about businesses with those four features that the majority thinks justifies a special market-definition approach for them? It cannot be the first two features—that the company sells different products to different groups of customers. Companies that sell multiple products to multiple types of customers are commonplace. A firm might mine for gold, which it refines and sells both to dentists in the form of fillings and to investors in the form of ingots. Or, a firm might drill for both oil and natural gas. Or a firm might make both ignition switches inserted into auto bodies and tires used for cars. I have already explained that, ordinarily, antitrust law will not group the two nonsubstitutable products together for step 1 purposes. Supra, at 10–11.

Neither should it normally matter whether a company sells related, or complementary, products, i.e., products which must both be purchased to have any function, such as ignition switches and tires, or cameras and film. It is well established that an antitrust court in such cases looks at the product where the attacked restraint has an anticompetitive effect. Supra, at 9; see Eastman Kodak, 504 U. S., at 463. The court does not combine the customers for the separate, nonsubstitutable goods and see if “overall” the restraint has a negative effect. See ibid.; 2B Areeda & Hovenkamp ¶565a. That is because, as I have explained, the complementary relationship between the products is irrelevant to the purposes of market-definition. See supra, at 10–11.

The majority disputes my characterization of merchant-related and shopper-related services as “complements.” See ante, at 14, n. 8. The majority relies on an academic article which devotes one sentence to the question, saying that “a two-sided market [is] different from markets for complementary products [e.g., tires and gas], in which both products are bought by the same buyers, who, in their buying decisions, can therefore be expected to take into account both prices.” Filistrucchi, Geradin, Van Damme, & Affeldt, Market Definition in Two-Sided Markets: Theory and Practice, 10 J. Competition L. & Econ. 293, 297 (2014) (Filistrucchi). I agree that two-sided platforms—at least as some academics define them, but see infra, at 19–20—may be distinct from some types of complements in the respect the majority mentions (even though the services resemble complements because they must be used together for either to have value). But the distinction the majority mentions has nothing to do with the relevant question. The relevant question is whether merchant-related and shopper-related services are substitutes, one for the other, so that customers can respond to a price increase for one service by switching to the other service. As I have explained, the two types of services are not substitutes in this way. Supra, at 11–12. And so the question remains, just as before: What is it about the economic relationship between merchant-related and shopper-related services that would justify the majority’s novel approach to market definition?

What about the last two features—that the company connects the two groups of customers to each other, in simultaneous transactions? That, too, is commonplace. Consider a farmers’ market. It brings local farmers and local shoppers together, and transactions will occur only if a farmer and a shopper simultaneously agree to engage in one. Should courts abandon their ordinary step 1 inquiry if several competing farmers’ markets in a city agree that only certain kinds of farmers can participate, or if a farmers’ market charges a higher fee than its competitors do and prohibits participating farmers from raising their prices to cover it? Why? If farmers’ markets are special, what about travel agents that connect airlines and passengers? What about internet retailers, who, in addition to selling their own goods, allow (for a fee) other goods-producers to sell over their networks? Each of those businesses seems to meet the majority’s four-prong definition.

Apparently as its justification for applying a special market-definition rule to “two-sided transaction platforms,” the majority explains that such platforms “often exhibit” what it calls “indirect network effects.” Ante, at 3. By this, the majority means that sales of merchant-related card services and (different) shopper-related card services are interconnected, in that increased merchant-buyers mean increased shopper-buyers (the more stores in the card’s network, the more customers likely to use the card), and vice versa. See ibid. But this, too, is commonplace. Consider, again, a farmers’ market. The more farmers that participate (within physical and esthetic limits), the more customers the market will likely attract, and vice versa. So too with travel agents: the more airlines whose tickets a travel agent sells, the more potential passengers will likely use that travel agent, and the more potential passengers that use the travel agent, the easier it will likely be to convince airlines to sell through the travel agent. And so forth. Nothing in antitrust law, to my knowledge, suggests that a court, when presented with an agreement that restricts competition in any one of the markets my examples suggest, should abandon traditional market-definition approaches and include in the relevant market services that are complements, not substitutes, of the restrained good. See supra, at 10–11.

2

To justify special treatment for “two-sided transaction platforms,” the majority relies on the Court’s decision in United States v. Grinnell Corp., 384 U. S. 563, 571–572 (1966). In Grinnell, the Court treated as a single market several different “central station services,” including burglar alarm services and fire alarm services. Id., at 571. It did so even though, for consumers, “burglar alarm services are not interchangeable with fire alarm services.” Id., at 572. But that is because, for producers, the services were indeed interchangeable: A company that offered one could easily offer the other, because they all involve “a single basic service—the protection of property through use of a central service station.” Ibid. Thus, the “commercial realit[y]” that the Grinnell Court relied on, ibid., was that the services being grouped were what economists call “producer substitutes.” See 2B Areeda & Hovenkamp ¶561, at 378. And the law is clear that “two products produced interchangeably from the same production facilities are presumptively in the same market,” even if they are not “close substitutes for each other on the demand side.” Ibid. That is because a firm that produces one such product can, in response to a price increase in the other, easily shift its production and thereby limit its compet- itor’s power to impose the higher price. See id., ¶561a, at 379.

Unlike the various types of central station services at issue in Grinnell Corp., however, the shopper-related and merchant-related services that American Express provides are not “producer substitutes” any more than they are traditional substitutes. For producers as for consumers, the services are instead complements. Credit card companies must sell them together for them to be useful. As a result, the credit-card companies cannot respond to, say, merchant-related price increases by shifting production away from shopper-related services to merchant-related services. The relevant “commercial realities” in this case are thus completely different from those in Grinnell Corp. (The majority also cites Brown Shoe Co. v. United States, 370 U. S. 294, 336–337 (1962), for this point, but the “commercial realities” considered in that case were that “shoe stores in the outskirts of cities compete effectively with stores in central downtown areas,” and thus are part of the same market. Id., at 338–339. Here, merchant-related services do not, as I have said, compete with shopper-related services, and so Brown Shoe Co. does not support the majority’s position.) Thus, our precedent provides no support for the majority’s special approach to defining markets involving “two-sided transaction platforms.”

3

What about the academic articles the majority cites? The first thing to note is that the majority defines “two-sided transaction platforms” much more broadly than the economists do. As the economists who coined the term explain, if a “two-sided market” meant simply that a firm connects two different groups of customers via a platform, then “pretty much any market would be two-sided, since buyers and sellers need to be brought together for markets to exist and gains from trade to be realized.” Rochet & Tirole, Two-Sided Markets: A Progress Report, 37 RAND J. Econ. 645, 646 (2006). The defining feature of a “two-sided market,” according to these economists, is that “the platform can affect the volume of transactions by charging more to one side of the market and reducing the price paid by the other side by an equal amount.” Id., at 664–665; accord, Filistrucchi 299. That requirement appears nowhere in the majority’s definition. By failing to limit its definition to platforms that economists would recognize as “two sided” in the relevant respect, the majority carves out a much broader exception to the ordinary antitrust rules than the academic articles it relies on could possibly support.

Even as limited to the narrower definition that economists use, however, the academic articles the majority cites do not support the majority’s flat rule that firms operating “two-sided transaction platforms” should always be treated as part of a single market for all antitrust purposes. Ante, at 13–15. Rather, the academics explain that for market-definition purposes, “[i]n some cases, the fact that a business can be thought of as two-sided may be irrelevant,” including because “nothing in the analysis of the practices [at issue] really hinges on the linkages between the demands of participating groups.” Evans & Schmalensee, Markets With Two-Sided Platforms, 1 Issues in Competition L. & Pol’y 667, 689 (2008). “In other cases, the fact that a business is two-sided will prove important both by identifying the real dimensions of competition and focusing on sources of constraints.” Ibid. That flexible approach, however, is precisely the one the District Court followed in this case, by considering the effects of “[t]he two-sided nature of the . . . card industry” throughout its analysis. 88 F. Supp. 3d, at 155.

Neither the majority nor the academic articles it cites offer any explanation for why the features of a “two-sided transaction platform” justify always treating it as a single antitrust market, rather than accounting for its economic features in other ways, as the District Court did. The article that the majority repeatedly quotes as saying that “ ‘[i]n two-sided transaction markets, only one market should be defined,’ ” ante, at 14–15 (quoting Filistrucchi 302), justifies that conclusion only for purposes of assessing the effects of a merger. In such a case, the article explains, “[e]veryone would probably agree that a payment card company such as American Express is either in the relevant market on both sides or on neither side . . . . The analysis of a merger between two payment card platforms should thus consider . . . both sides of the market.” Id., at 301. In a merger case this makes sense, but is also meaningless, because, whether there is one market or two, a reviewing court will consider both sides, because it must examine the effects of the merger in each affected market and submarket. See Brown Shoe Co., 370 U. S., at 325. As for a nonmerger case, the article offers only United States v. Grinnell as a justification, see Filistrucchi 303, and as I have already explained, supra, at 16–18, Grinnell does not support this proposition.

E

Put all of those substantial problems with the majority’s reasoning aside, though. Even if the majority were right to say that market definition was relevant, and even if the majority were right to further say that the District Court should have defined the market in this case to include shopper-related services as well as merchant-related services, that still would not justify the majority in affirming the Court of Appeals. That is because, as the majority is forced to admit, the plaintiffs made the factual showing that the majority thinks is required. See ante, at 17.

Recall why it is that the majority says that market definition matters: because if the relevant market includes both merchant-related services and card-related services, then the plaintiffs had the burden to show that as a result of the nondiscrimination provisions, “the price of credit-card transactions”—considering both fees charged to merchants and rewards paid to cardholders—“was higher than the price one would expect to find in a competitive market.” Ante, at 16. This mirrors the Court of Appeals’ holding that the Government had to show that the “nondiscrimination provisions” had “made all [American Express] customers on both sides of the platform—i.e., both merchants and cardholders—worse off overall.” 838 F. 3d, at 205.

The problem with this reasoning, aside from it being wrong, is that the majority admits that the plaintiffs did show this: they “offer[ed] evidence” that American Express “increased the percentage of the purchase price that it charges merchants . . . and that this increase was not entirely spent on cardholder rewards.” Ante, 17 (citing 88 F. Supp. 3d, at 195–197, 215). Indeed, the plaintiffs did not merely “offer evidence” of this—they persuaded the District Court, which made an unchallenged factual finding that the merchant price increases that resulted from the nondiscrimination provisions “were not wholly offset by additional rewards expenditures or otherwise passed through to cardholders, and resulted in a higher net price.” Id., at 215 (emphasis added).

In the face of this problem, the majority retreats to saying that even net price increases do not matter after all, absent a showing of lower output, because if output is increasing, “ ‘rising prices are equally consistent with growing product demand.’ ” Ante, at 18 (quoting Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U. S. 209, 237 (1993)). This argument, unlike the price argument, has nothing to do with the credit-card market being a “two-sided transaction platform,” so if this is the basis for the majority’s holding, then nearly all of the opinion is dicta. The argument is also wrong. It is true as an economic matter that a firm exercises market power by restricting output in order to raise prices. But the relevant restriction of output is as compared with a hypothetical world in which the restraint was not present and prices were lower. The fact that credit-card use in general has grown over the last decade, as the majority says, see ante, at 17–18, says nothing about whether such use would have grown more or less without the nondiscrimination provisions. And because the relevant question is a comparison between reality and a hypothetical state of affairs, to require actual proof of reduced output is often to require the impossible—tantamount to saying that the Sherman Act does not apply at all.

In any event, there are features of the credit-card market that may tend to limit the usual relationship between price and output. In particular, merchants generally spread the costs of credit-card acceptance across all their customers (whatever payment method they may use), while the benefits of card use go only to the cardholders. See, e.g., 88 F. Supp. 3d, at 216; Brief for John M. Connor et al. as Amici Curiae 34–35. Thus, higher credit-card merchant fees may have only a limited effect on credit-card transaction volume, even as they disrupt the marketplace by extracting anticompetitive profits.

IV

A

For the reasons I have stated, the Second Circuit was wrong to lump together the two different services sold, at step 1. But I recognize that the Court of Appeals has not yet considered whether the relationship between the two services might make a difference at steps 2 and 3. That is to say, American Express might wish to argue that the nondiscrimination provisions, while anticompetitive in respect to merchant-related services, nonetheless have an adequate offsetting procompetitive benefit in respect to its shopper-related services. I believe that American Express should have an opportunity to ask the Court of Appeals to consider that matter.

American Express might face an uphill battle. 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. In United States v. Topco Associates, Inc., 405 U. S. 596, 611 (1972), this Court wrote:

“If a decision is to be made to sacrifice competition in one portion of the economy for greater competition in another portion, this . . . is a decision that must be made by Congress and not by private forces or by the courts. Private forces are too keenly aware of their own interests in making such decisions and courts are ill-equipped and ill-situated for such decisionmaking.”

American Express, pointing to vertical price-fixing cases like our decision in Leegin, argues that comparing competition-related pros and cons is more common than I have just suggested. See 551 U. S., at 889–892. But Leegin held only that vertical price fixing is subject to the “rule of reason” instead of being per se unlawful; the “rule of reason” still applies to vertical agreements just as it applies to horizontal agreements. See id., at 898–899.

Moreover, the procompetitive justifications for vertical price-fixing agreements are not apparently applicable to the distinct types of restraints at issue in this case. A vertically imposed price-fixing agreement typically involves a manufacturer controlling the terms of sale for its own product. A television-set manufacturer, for example, will insist that its dealers not cut prices for the manufacturer’s own televisions below a particular level. Why might a manufacturer want its dealers to refrain from price competition in the manufacturer’s own products? Perhaps because, for example, the manufacturer wants to encourage the dealers to develop the market for the manufacturer’s brand, thereby increasing interbrand competition for the same ultimate product, namely a television set. This type of reasoning does not appear to apply to American Express’ nondiscrimination provisions, which seek to control the terms on which merchants accept other brands’ cards, not merely American Express’ own.

Regardless, I would not now hold that an agreement such as the one before us can never be justified by procompetitive benefits of some kind. But the Court of Appeals would properly consider procompetitive justifications not at step 1, but at steps 2 and 3 of the “rule of reason” inquiry. American Express would need to show just how this particular anticompetitive merchant-related agreement has procompetitive benefits in the shopper-related market. In doing so, American Express would need to overcome the District Court’s factual findings that the agreement had no such effects. See 88 F. Supp. 3d, at 224–238.

B

The majority charts a different path. Notwithstanding its purported acceptance of the three-step, burden-shifting framework I have described, ante, at 9–10, the majority addresses American Express’ procompetitive justifications now, at step 1 of the analysis, see ante, at 18–20. And in doing so, the majority inexplicably ignores the District Court’s factual findings on the subject.

The majority reasons that the challenged nondiscrimination provisions “stem negative externalities in the credit-card market and promote interbrand competition.” Ante, at 19. The “negative externality” the majority has in mind is this: If one merchant persuades a shopper not to use his American Express card at that merchant’s store, that shopper becomes less likely to use his American Express card at other merchants’ stores. Ibid. The majority worries that this “endangers the viability of the entire [American Express] network,” ibid., but if so that is simply a consequence of American Express’ merchant fees being higher than a competitive market will support. “The antitrust laws were enacted for ‘the protection of competition, not competitors.’ ” Atlantic Richfield Co. v. USA Petroleum Co., 495 U. S. 328, 338 (1990). If American Express’ merchant fees are so high that merchants successfully induce their customers to use other cards, American Express can remedy that problem by lowering those fees or by spending more on cardholder rewards so that cardholders decline such requests. What it may not do is demand contractual protection from price competition.

In any event, the majority ignores the fact that the District Court, in addition to saying what I have just said, also rejected this argument on independent factual grounds. It explained that American Express “presented no expert testimony, financial analysis, or other direct evidence establishing that without its [nondiscrimination provisions] it will, in fact, be unable to adapt its business to a more competitive market.” 88 F. Supp. 3d, at 231. It further explained that the testimony that was provided on the topic “was notably inconsistent,” with some of American Express’ witnesses saying only that invalidation of the provisions “would require American Express to adapt its current business model.” Ibid. After an extensive discussion of the record, the District Court found that “American Express possesses the flexibility and expertise necessary to adapt its business model to suit a market in which it is required to compete on both the cardholder and merchant sides of the [credit-card] platform.” Id., at 231–232. The majority evidently rejects these factual findings, even though no one has challenged them as clearly erroneous.

Similarly, the majority refers to the nondiscrimination provisions as preventing “free riding” on American Express’ “investments in rewards” for cardholders. Ante, at 19–20; see also ante, at 7 (describing steering in terms suggestive of free riding). But as the District Court explained, “[p]lainly . . . investments tied to card use (such as Membership Rewards points, purchase protection, and the like) are not subject to free-riding, since the network does not incur any cost if the cardholder is successfully steered away from using his or her American Express card.” 88 F. Supp. 3d, at 237. This, I should think, is an unassailable conclusion: American Express pays rewards to cardholders only for transactions in which cardholders use their American Express cards, so if a steering effort succeeds, no rewards are paid. As for concerns about free riding on American Express’ fixed expenses, including its investments in its brand, the District Court acknowledged that free-riding was in theory possible, but explained that American Express “ma[de] no effort to identify the fixed expenses to which its experts referred or to explain how they are subject to free riding.” Ibid.; see also id., at 238 (American Express’ own data showed “that the network’s ability to confer a credentialing benefit trails that of its competitors, casting doubt on whether there is in fact any particular benefit associated with accepting [American Express] that is subject to free riding”). The majority does not even acknowledge, much less reject, these factual findings, despite coming to the contrary conclusion.

Finally, the majority reasons that the nondiscrimination provisions “do not prevent Visa, MasterCard, or Discover from competing against [American Express] by offering lower merchant fees or promoting their broader merchant acceptance.” Ante, at 20. But again, the District Court’s factual findings were to the contrary. As I laid out above, the District Court found that the nondiscrimination provisions in fact did prevent Discover from pursuing a low-merchant-fee business model, by “den[ying] merchants the ability to express a preference for Discover or to employ any other tool by which they might steer share to Discover’s lower-priced network.” 88 F. Supp. 3d, at 214; see supra, at 7. The majority’s statements that the nondiscrimination provisions are procompetitive are directly contradicted by this and other factual findings.

*  *  *

For the reasons I have explained, the majority’s decision in this case is contrary to basic principles of antitrust law, and it ignores and contradicts the District Court’s detailed factual findings, which were based on an extensive trial record. I respectfully dissent.