TY - JOUR
AU1 - Manne, Geoffrey, A
AU2 - Wu,, Tim
AB - This issue’s Contemporary Critique, discusses the US Supreme Court decision in Ohio v. American Express (Amex). The ruling has sparked much interest (and controversy) as to its treatment of digital markets and the platform economy. In his commentary, Geoffrey A. Manne supports the Supreme Court's decision, and offers insightful analysis of why, to his mind, the decision properly identifies the competitive process of platforms. Tim Wu's critique, on the other hand, raises concerns as to the treatment of evidence and theory in this case and the legacy of the ruling, which may undermine the efficacy of future antitrust enforcement. In defence of the Supreme Court’s ‘single market’ definition in Ohio v American Express Geoffrey A Manne Geoffrey A Manne Manne, President and founder of the International Center for Law and Economics (ICLE) Manne, President and founder of the International Center for Law and Economics (ICLE) Geoffrey A. Manne, President and founder of the International Center for Law and Economics (ICLE). Email: gmanne@laweconcenter.org The Supreme Court’s decision in Ohio, et al v Am Express Co, et al (‘Amex’)1 is uniquely important for the antitrust analysis of firms in the modern platform economy. Although it is undoubtedly not the last word on the subject, the case represents the Court’s first comprehensive effort to address the thorny and previously indeterminate question of how courts should define the relevant market and assess competitive effects in antitrust cases involving two-sided or multi-sided platforms (‘two-sided markets’). In this article, I evaluate the Supreme Court’s approach to this question, discuss the economic and legal underpinnings of how it approached market definition and effects analysis, and demonstrate why the primary criticisms of the Court’s decision are misguided. While the Court’s approach has been roundly decried by some, its conclusion—that both sides of a two-sided market must be considered in defining the relevant market and evaluating the existence and consequences of a firm’s exercise of market power—is, indeed, the proper one. I. THE IMPORTANCE OF MARKET DEFINITION IN ANTITRUST ANALYSIS Brief background of market definition in the case The district court in Amex held that American Express (‘Amex’) offers its services in two separate, but ‘deeply interrelated,’ product markets: one for the issuance of Amex cards to consumers and the other for the acceptance of Amex cards by merchants.2 Having found that Amex offers its services in two markets, the district court further held that the assessment of the competitive effects of Amex’s conduct must be limited to each separate market, in separate analyses.3 The Second Circuit reversed, holding that Amex operates in a single ‘highly interdependent’ two-sided market combining both the services Amex provides to merchants as well as the services it provides to cardholders, and requiring a unified competitive effects analysis incorporating both sides of this market.4 The Supreme Court affirmed the Second Circuit: ‘[C]ourts must include both sides of the platform — merchants and cardholders — when defining the credit-card market.’5 Market definition as the crucial identifier of the relevant competitive process Why ‘market definition’ and not ‘effects analysis’? The Court’s decision to frame the case in terms of market definition is important. The question presented—was the Government’s proof of anticompetitive effects sufficient to establish its prima facie case and shift the burden to Amex?6—suggested that the case was about effects analysis more broadly: whether the identification of certain anticompetitive effects is properly indicative of the impermissible exercise of market power, and how any offsetting, procompetitive benefits should be weighed against them. But the Court quite accurately notes that assessing competitive effects entails ‘a fact-specific assessment of ‘market power and market structure … to assess the [restraint]’s actual effect’ on competition’.7 And assessing market power and market structure typically requires first defining the market: ‘Without a definition of [the] market there is no way to measure [the defendant’s] ability to lessen or destroy competition.’8 Defining which transactions are actually affected by alleged anticompetitive behaviour is necessary for properly identifying anticompetitive effects. Indeed, because the exercise of monopoly power entails a reduction in market output (and a corresponding increase in market price),9 a competitive effects analysis entails (explicitly or implicitly10) a market definition exercise in order to identify what the ‘output’ is that is being evaluated and to assess the effect of a platform’s conduct on it. As the Court in Amex stated: ‘Market power is the ability to raise price profitably by restricting output.’11 But crucial, I believe, to the Court’s focus on market definition is that ‘[a]lleging the relevant market in an antitrust case does not merely identify the portion of the economy most directly affected by the challenged conduct; it identifies the competitive process alleged to be harmed.’12 The key question in Amex concerns not (just) what products are properly in the relevant market, but also what the nature of those products is, how they are priced and on what terms they are sold, what levers Amex can use to increase its profits, and what competitive constraints affect its ability to do so.13 The answers to all of these, and thus the understanding of the competitive process at issue in Amex, turn critically on how the market is defined. As the Second Circuit held: ‘The proper market definition thus can be determined ‘only after a factual inquiry into the “commercial realities” faced by consumers.’ ‘The basic principle is that the relevant market definition must encompass the realities of competition.’14 Market definition is about more than demand-side substitution The classic product market definition test asks whether consumers would purchase substitute products in response to a small price increase of the target product, and, if so, it then deems these products in the same market. Some critics have rightly pointed out that the expansion of the market in Amex to include both sides of the two-sided platform does not take this character: obviously, card services provided to users are not in any way, or at any price, a substitute for the services provided to merchants.15 While these critics do sometimes acknowledge that cross-market effects in platform markets are significant,16 they object that it is inappropriate to include both sides of a two-sided market in the market definition exercise because ‘a relevant product market includes only products or services “that have reasonable interchangeability” of use …’.17 This objection is misplaced, however. It is correct, as Justice Breyer notes in his dissent in Amex, that ‘[t]he reason that substitutes are included in the relevant market is that they restrain a firm’s ability to profitably raise prices ….’18 But things other than product substitutes constrain market power as well; market definition and effects analysis entail much more than assessing merely consumer demand elasticity. The assessment of market power also entails a geographic market definition, for example. While the US geographic market definition process has historically focused primarily on demand-side substitutability,19 the exercise is also about which sellers or manufacturers provide or could provide competing products. Indeed, ‘[p]roper geographic and product market definition requires understanding how demand and supply would respond to potential price increases that stem from anticompetitive activity — not just a demand-side response.’20 Similarly, the product market under the 2010 Horizontal Merger Guidelines (2010 HMGs) includes ‘[f]irms that are not current producers in a relevant market, but that would very likely provide rapid supply responses ….’21 Indeed, even in the absence of substitutability, if products are sufficiently interrelated that ‘the sale of one product affects the prices of another product sold by the same company, the two products should be placed in the same candidate market’.22 The point is that the market definition exercise always turns on matters other than just consumers’ demand-side response. And in order to properly evaluate whether observed output (or price) effects actually connote the impermissible exercise of market power sufficient to meet a plaintiff’s initial burden, the relevant constraints must be taken into account in defining the scope of the relevant market. The Court’s inclusion of non-substitute products in the market definition in Amex was perfectly appropriate. II. THE CENTRALITY OF MARKET DEFINITION IN THE COURT’S AMEX DECISION Assessing output, price, and non-price product characteristics in platform markets requires a two-sided market definition The Court in Amex concluded that both sides of Amex’s platform—and transaction platforms more generally—must be considered in the relevant market in order to capture the actual competitive process at issue: Due to indirect network effects, two-sided platforms cannot raise prices on one side without risking a feedback loop of declining demand …. Price increases on one side of the platform [] do not suggest anticompetitive effects without some evidence that they have increased the overall cost of the platform’s services.23 By including both sides of a two-sided platform market in the relevant market, the Court’s rule ensures that the price, quality, and output of the relevant product are properly considered. This is especially important because, while output of a two-sided platform may often be measured looking at only one side (particularly for two-sided transactional platforms where it is presumed that a transaction on one side also entails a transaction on the other side),24 price and non-price product characteristics generally will not be properly measured by doing so. First, the Court’s approach ensures that the nature of the product at issue properly reflects the two-sidedness of the platform that produces it. The benefit of incorporating both sides of the market from the point of view of output is in properly defining the product. Thus, in the case of Amex, the product at issue is not ‘payment acquisition services provided to merchants’, but ‘platform services facilitating transactions between cardholders and merchants’.25 Once properly defined, viewing one side of the market may be sufficient for counting these, but the two-sided market definition ensures that what is being counted is the correct product on offer. Secondly, defining the market to include both sides ensures that the full price and quality attributes of the product are considered. Where a single, two-sided product is at issue, the price may be spread across users on both sides of the market. Moreover, non-price product characteristics will necessarily differ between different sets of users. In fact, it is often the case that product quality is adjusted to users on one side of a platform in order to offer them an effectively negative price. Given the differential incidence of price and quality across a platform, it is impossible to capture the competitive dynamics and to measure the competitive effects by viewing only the partial price on one side. The economic sense of the ‘interrelated markets’ test in Amex Arguably, by including non-substitute, interrelated markets in the same analysis, the Court added another—and crucial—dimension to the market definition exercise in cases involving multi-sided platforms. The Court seems to hold that, in addition to traditional product and geographic market analyses, courts must also perform an interrelated markets analysis in order to ensure that the market power and market structure assessments incorporate effects occurring as a necessary consequence of cross-platform interrelatedness. ‘Evaluating both sides of a two-sided transaction platform is also necessary to accurately assess competition.’26 As the multi-sided markets literature makes clear, multi-sided platforms are defined by the interrelatedness between their various sides, and market definition (and competitive effects) analysis must entail an assessment of all sides of a platform.27 For platforms, the structure and interrelatedness of the relative prices (and other terms—like the ‘anti-steering’ terms at issue in Amex—that, in effect, determine the quality of the service) is what matters, not the specific prices charged to users on a given side of the market.28 ‘[T]here is no meaningful economic relationship between benefits and costs on each side of the market considered alone …, [and] any analysis of social welfare must account for the pricing level, the pricing structure, and the feasible alternatives for getting all sides on board.’29 Failing to account for cross-platform product characteristics and the interplay of price and quality across a platform will almost assuredly result in inaccurate antitrust decision-making. Absent consideration of both sides of a platform, the analysis will arbitrarily include and exclude various sets of users and transactions, and incorrectly assess the extent and consequences of market power.30 Indeed, evidence of a price effect on only one side of a two-sided platform can be consistent with either neutral, anticompetitive, or procompetitive conduct.31 Only when output is defined to incorporate the two-sidedness of the product, and where price and quality are assessed on both sides of a sufficiently interrelated two-sided platform, is it even possible to distinguish between procompetitive and anticompetitive effects. ‘Any other analysis would lead to “mistaken inferences” of the kind that could ‘chill the very conduct the antitrust laws are designed to protect.’32 III. ADDRESSING THE CRITICISMS OF THE AMEX DECISION ‘Some’ harm is not the same thing as ‘competitively relevant’ harm A number of scholars believe that the Court erred in considering both sides of the two-sided market in Amex—not only as a function of market definition, but also as a function of an effects analysis.33 For these commentators, an effect that makes consumers worse off on one side of a multi-sided market is sufficient to meet the plaintiff’s prima facie burden34 of showing antitrust harm—irrespective of the effects on other side(s) of the platform.35 This equating of some harm to competitively relevant harm is misplaced, however, particularly for multi-sided platforms. For conduct to be actionable it must create or maintain monopoly power and permit its exercise. But, plainly, not all conduct that causes harm has this characteristic. Consider, for example, the bedrock antitrust principle which states that harm to certain competitors is not the same thing as harm to competition. One necessary implication of this principle is that harm to those competitors’ customers is also not necessarily harm to competition because, for some customers (say, with brand loyalty to a particular firm), the loss of one competitor would also harm them—even if competition itself were not affected. The same is true for price discrimination that does not result from the exercise of market power, for mergers that differentially affect marginal and inframarginal consumers, and for temporary disruptions of competition insufficient to constitute the ‘durable’ exercise of market power, among others. Because this distinction between harm to some consumers and anticompetitive harm is so crucial to ferreting out conduct that creates or maintains monopoly power, antitrust law imposes the burden of proof on the plaintiff to first demonstrate that the conduct at issue has an actual anticompetitive effect—not merely that it results in some harm. In the platform context, understanding whether there is harm to competition at all requires an assessment of the effects of conduct on all sides of the platform. ‘[N]o economic basis exists for establishing a presumption that ‘harm’ on one side of a two-sided platform is sufficient to demonstrate that market output has been restricted, or that consumer welfare has otherwise been harmed.’36 In fact, ‘[s]eparating the two markets allows legitimate competitive activities in the market for general purposes to be penalized no matter how output-enhancing such activities may be.’37 ‘Separate markets’ analysis is insufficient to incorporate cross-platform effects Nevertheless, some critics assert that courts can and should ‘account for’ cross-platform effects in the effects analysis, but not by inextricably intertwining both sides of a two-sided market in the definition of the relevant market.38 In other words, these critics would consider Amex’s ‘payment acquisition services to merchants’ as a separate product, assess output in terms of the number of merchants who use Amex’s platform, and assess price as the direct price charged by Amex to merchants for these services. Consumer ‘feedback’ would be relevant, but only to the extent that it affects the number of merchants who use the platform. This perspective runs into pitfalls of its own making, however. Consider the hypothetical ‘Dine Out’ restaurant reservation app imagined by Katz and Sallet. The app attracts customers to its platform by providing rewards and charges restaurants a fee when customers book through the app.39 As Katz and Sallet point out: Suppose one defined a single, two-sided market comprising [reservation] services sold to both restaurants and consumers. Although both scenarios below correspond to a net price of $1 per consummated reservation, merchant and consumer welfare can be significantly different depending on whether: (a) restaurants pay a $2 fee and consumers receive incentive payments equal to $1, or (b) restaurants pay a $10 fee and consumers receive payments equal to $9. Holding the number of participating restaurants and consumers fixed, restaurants are clearly worse off under scenario (b), while consumers are better off. Because the net price is $1 in each case, the net-price approach would consider consumers’ gains to fully offset the restaurants’ losses…. To understand output and welfare effects, one must also examine the individual components of the net price.40 The problem they point to—that consideration of net price effects is insufficient to capture overall welfare effects—is a problem only because the separate markets approach assumes away the relevant mechanism of potential harm. That is, ‘[h]olding the number of participating restaurants and consumers fixed’ in their example means simply assuming that there are no welfare effects, where such effects are precisely a function of the number of transactions between restaurants and consumers enabled by the platform’s pricing structure. Indeed, to the extent that Katz and Sallet would ‘consider’ cross-platform effects, they are forced to drop the all-else-equal assumption and assess whether output would change because of the challenged conduct: In looking at the individual components, it can be necessary to consider their interaction. For example, one can examine whether restaurants benefitted from the higher fees that resulted from Dine Out’s challenged practices, as those higher fees induced Dine Out to pay larger rewards payments to diners. Notice, however, that a restaurant would benefit from the larger rewards only if they induced consumers to patronize the restaurant to a greater degree.41 Yet they continue to assert that netting out the effects on both sides of the platform will lead to incorrect results: [B]ecause the benefits to the restaurant due to higher rewards would be mediated by consumer behavior, there is no guarantee that the benefits to the restaurants would equal the change in rewards paid to consumers, contrary to the assumption of the net-price approach. For example, it is possible that the rewards merely shift diners’ choice of reservation mechanism to Dine Out without changing the restaurants at which they ultimately dine. If this were the case, then the increase in rewards paid to diners would not offset harm suffered by the restaurants at all.42 The problem, of course, is that this simply shows that the initial market definition was incorrect. According to this hypothetical, other reservation mechanisms must be included in the relevant market—and, apparently, Dine Out will not be able to sustain the hypothetical price increase. If Dine Out’s fees are too high, merchants will simply refuse to use the service. And they can do so ‘without changing the restaurants at which [consumers] ultimately dine’ because, by the terms of the hypothetical, consumers are already using other mechanisms to make reservations without any effect on their choice of restaurant. This suggests that Dine Out does not have market power in the first place, not that measuring net effects is inappropriate. The single market approach avoids this problem by, first, correctly identifying the relevant market and, secondly, by assessing welfare effects based on transactions between restaurants and diners. If, indeed, the fee increase results in a decrease in transactions (because restaurants drop the service and comparable alternatives aren’t available), the greater rewards offered to consumers will not offset the higher fees charged to restaurants because neither will be incurred in the first place. Similarly, to the extent that the rewards do increase the number of transactions, the net effect will be the appropriate measure of welfare, even if the gains are distributed differently across the sides of the platform. The separate markets framework applied to Amex In a separate markets analysis in Amex that ‘accounts for’ cross-market effects but evaluates them on only one side of the platform, the determination might (accurately) be made that cardholders’ preferences constitute a significant constraint on Amex, but, viewing only the effect on services offered by Amex to merchants, this would not necessarily alter the conclusion that the challenged conduct was harmful. According to Katz and Sallet, accounting for feedback effects in a separate markets framework means ‘consider[ing] price changes on one side of the platform while holding prices on the other side constant and examining whether there are significant, plausible feedback effects’.43 But this approach excludes consideration of feedback effects that are not automatic, but that arise from a change in pricing structure by the platform in response to lost revenue on one side of the market. And yet that is the sine qua non of platforms: the imposition of pricing and usage terms on two sides of a market simultaneously in order to balance the participants and maximize revenue. Consider the market dynamics in the absence of the anti-steering provisions at issue in Amex. In such a world, Amex would not reliably earn revenue from the service charges it imposes on merchants: either these charges would be lower (as the Petitioners contend), or they would be irrelevant (because merchants would ‘steer’ consumers to other payment mechanisms and no charges would be incurred). In either case, there would be fewer Amex card transactions, and Amex would earn less revenue. In response, Amex would not likely ‘hold[] prices on the other side constant’. Given the expected volume of transactions and merchant-fee revenue, Amex would ‘rebalance’ its pricing structure, presumably by reducing the benefits offered to consumers. Fewer consumers would pay for an Amex card given the reduced benefits, and thus Amex’s revenue from the annual fee it charges to cardholders—which also helps to fund its benefits—would also be reduced, leading Amex to offer even fewer benefits, fewer consumers to carry and use the card, and so on. Moving from this world to one in which anti-steering provisions are imposed would, presumably, have the reverse effect: increased revenue derived from merchant fees. Crucially, this ability to increase revenue from the imposition of higher merchant fees need not result from the exercise of market power. In fact, merchants willingly accept Amex cards despite their higher fees because they benefit from the cardholder subsidy provided by Amex in the form of cardholder rewards.44 In the absence of Amex’s anti-steering provisions, however, each merchant’s incentive is to accept Amex cards in the abstract, but to convince consumers to pay with a different, lower cost card at the point of sale when doing so does not mean losing the sale. If the merchant cannot induce a switch, the merchant is in no worse position and it (willingly) accepts payment via Amex. If it can induce an Amex cardholder to pay with a different card with a lower merchant fee, so much the better. Each individual merchant thus benefits from the Amex ‘system’, which increases the overall volume of sales, but also from strategically undermining the system. No single merchant bears a significant cost of doing so, and yet merchants and consumers overall lose out if the sum total of merchants’ strategic conduct reduces the size of the cardholder subsidy and/or the number of Amex cardholders.45 Thus, it is simply incorrect that: Amex’s fundamental argument is that it must prevent steering away from its cards in order to keep its rewards high and support its business model. That argument depends on using an exercise of market power to raise prices and direct the benefits elsewhere, and imposes a restraint that ensures that those benefits cannot redound to merchants’ benefit through the competitive process.46 Rather, Amex’s ‘fundamental argument’ is that its pricing structure increases transaction volume (benefitting both merchants and cardholders), and that that the lure of that benefit—not the exercise of market power—is what induces merchants to accept Amex’s higher merchant fees. The restraint ensures that those benefits are not whittled away by individual merchants trying to capture more of the benefit for themselves; it does not prevent merchants from realizing the benefits of a competitive system in which Amex’s prestige and cardholder rewards induce cardholders to engage in more and/or higher value transactions. The separate markets analysis of this market fails to capture the competitive process at issue, which requires consideration of how Amex’s pricing structure affects the incentives and conduct of actors on both sides of the platform simultaneously. IV. THE PNB PROBLEM In theory at least, an analysis of offsetting procompetitive efficiencies could accommodate the requisite complexity and incorporate the effects of conduct on both sides of a two-sided market. But why would that make any sense? If, because of the interrelatedness of the sides of the market, it is improper to consider competitive effects on only one side in isolation—a clear consequence of the economic literature on multi-sided markets—why would it make sense to limit the market definition to only one side in order to facilitate the plaintiff making out its prima facie case, only to have the entire analysis upended by the defendant’s demonstration of countervailing effects elsewhere? Not only is such an approach certain to increase the risk of false positives, it is also analytically incoherent (as well as a likely waste of judicial resources). But aside from the analytical problems with this approach, there is another consideration that may have impelled the Court to incorporate both sides of a two-sided market in the market definition exercise and not in steps two and three of the burden-shifting framework: the possible constraints on the analysis of ‘out-of-market’, procompetitive effects imposed by the Court’s decisions in United States v Philadelphia Nat’l Bank (‘PNB’) and United States v Topco Assocs, Inc (‘Topco’).47 The district court in Amex held that the case involved ‘two separate yet complementary product markets’,48 and, citing Topco and PNB, asserted that, ‘[a]s a general matter …, a restraint that causes anticompetitive harm in one market may not be justified by greater competition in a different market’.49 Justice Breyer echoes this conclusion in his dissent. Although some scholars assert that PNB and Topco apply to preclude the consideration of offsetting, ‘out-of-market’ efficiencies, it is not, in fact, clear that the PNB limitation applies in Sherman Act cases: as a matter of precedent, PNB applies only to mergers evaluated under section 7 of the Clayton Act; it is not controlling in the context of Sherman Act cases. And to the extent that Topco might be relied upon as a basis for arguing that the Supreme Court has extended the holding in PNB to the Sherman Act, it is a thin, weak reed. The Court in Topco cited PNB in dictum, not for a doctrinal proposition relating to the operation of the rule of reason, but for a general, conceptual point about the asserted difficulty of courts adjudicating between conflicting economic rights (‘The fact is that courts are of limited utility in examining difficult economic problems.’50) and the use of per se analysis to sidestep this problem. Our inability to weigh, in any meaningful sense, destruction of competition in one sector of the economy against promotion of competition in another sector is one important reason we have formulated per se rules. * * * Antitrust laws in general, and the Sherman Act in particular, are the Magna Carta of free enterprise. They are as important to the preservation of economic freedom and our free enterprise system as the Bill of Rights is to the protection of our fundamental personal freedoms. And the freedom guaranteed each and every business, no matter how small, is the freedom to compete — to assert with vigor, imagination, devotion, and ingenuity whatever economic muscle it can muster. Implicit in such freedom is the notion that it cannot be foreclosed with respect to one sector of the economy because certain private citizens or groups believe that such foreclosure might promote greater competition in a more important sector of the economy.51 Nowhere does the Court in Topco suggest that it is inappropriate within a rule-of-reason analysis to weigh out-of-market efficiencies against in-market effects. Somewhat more resolutely, Justice Breyer, also citing Topco, concludes in his dissent that a burden-shifting analysis would not incorporate consideration of both sides of the market: ‘A Sherman Act §1 defendant can rarely, if ever, show that a procompetitive benefit in the market for one product offsets an anticompetitive harm in the market for another.’52 None of these is a strong basis upon which to assert that the Court has held that it is inappropriate for courts to weigh out-of-market efficiencies in step three of a rule-of-reason analysis. But it is certainly a risk, and, at the very least, the Court may have been reluctant to weigh in on the matter either way in Amex. More troubling, however, are the implications of the logic of PNB as applied to platforms. Katz and Sallet, for example, contend that the logic of PNB should preclude considering the full extent of platform markets both in steps two and three and in step one of the rule-of-reason burden-shifting analysis: [A]ntitrust analysis has consistently rejected, and should continue to reject, the notion that harm to competition can be justified on the grounds that it also confers benefits to another group of users. Regardless of whether one defines a single, multisided market or a set of closely linked one-sided markets, courts should continue to apply separate-effects analysis. Stated another way, the doctrinal principles in [PNB] and Illinois Brick Co. counsel against balancing harms and benefits across distinct user groups, regardless of how the markets are labeled.53 Unfortunately, this is where the rubber meets the road. The net effect is that proponents of assessing the ‘other’ side of a two-sided market only through a balancing of procompetitive efficiencies in steps two and three of the rule-of-reason analysis are, in fact, arguing against any consideration of those effects at all. Indeed, the scholars who signed the Amicus Brief of 28 Antitrust Law Professors in Amex reject the view—and counsel that the Court should, as well—‘that a “relevant market” for antitrust purposes must (or even can) include “both sides” of a two-sided platform’, while simultaneously denying ‘that competitive harms and benefits must be (or even can be) ‘netted’ across relevant markets in rule-of-reason cases’.54 In the end, this suggests some disingenuousness on the part of scholars who contend that it is the incorporation of both sides of a two-sided market into a single relevant market that is the problem in Amex, not that the Court ‘accounted for’ cross-platform effects. V. CONCLUSION The only way to properly identify the competitive process at issue and to assess the competitive effects of platform conduct intermediating two interconnected sets of market actors—particularly when it imposes constraints or costs unevenly between them—is through a more nuanced market definition analysis. Whether that is best characterized as adding a new market definition test (an ‘interrelated markets’ test) to the traditional product and geographic market tests, or as enhancing the product market test to incorporate two-sidedness, it is not possible to reliably account for tightly interrelated two-sided markets through a separate markets analysis. The Court in Amex appears—sensibly—to have reached this same conclusion. Footnotes 1 (2018) 138 S Ct 2274 (Slip Opinion) [hereinafter, ‘Amex’ and ‘Amex Slip Op.’]. 2 United States, et al v Am Express Co, et al (ED NY 2015) 88 F Supp 3d 143, 151. 3 ibid. 4 United States, et al v Am Express Co, et al (2nd Cir 2016) 838 F 3d 179, 185. (2nd Cir. 2016). See also ibid 197–98 (‘The District Court erred in excluding the market for cardholders from its relevant market definition …. Separating the two markets here — analyzing the effect of Amex’s vertical restraints on the market for network services while ignoring their effect on the market for general purpose cards — ignores the two markets’ interdependence.’). 5 Amex Slip Op, 12. See also Amex Slip Op, Syllabus, 2 (‘[C]redit-card networks are best understood as supplying only one product — the transaction — that is jointly consumed by a cardholder and a merchant. Accordingly, the two-sided market for credit-card transactions should be analyzed as a whole.’). (emphasis added). 6 Petition for Writ of Certiorari in Ohio, et al v Am Express Co, et al (2017) 2017 WL 2472075, *i. 7 Amex Slip Op., 8–9 (quoting Copperweld Corp v Independence Tube Corp (1984) 467 US 752, 768). 8 Amex Slip Op., 11 (quoting Walker Process Equip, Inc v Food Machinery & Chem Corp (1965) 382 US 172, 177). 9 See eg Michael E Levine, ‘Price Discrimination Without Market Power’ (2002) 19 Yale J Reg 1, 17 (‘Without the ability to restrict industry output, there is no market power.’). 10 Even a GUPPI or similar upward pricing pressure exercise often implicitly incorporates market definition assumptions. See Jan M Rybnicek and Laura C Onken, A Hedgehog in Fox’s Clothing? The Misapplication of GUPPI Analysis (2016) 23 Geo Mason L Rev 1187, 1197 (‘If the GUPPI is dependent on diversion ratios, and those diversion ratios are based on market shares, then the GUPPI adds little value beyond older tools that evaluate market structure and the degree of increased concentration.’). 11 Amex Slip Op, 17 (quoting Phillip E. Areeda and Herbert Hovenkamp, Antitrust Law (4th edn, Wolters Kluwer 2017) §5.01) (emphasis added). See also Brief for Amici Curiae Antitrust Law & Economics Scholars in Support of Respondents in Ohio et al v Amex Co [hereinafter, ‘Amicus Brief of Law & Econ Scholars’] <https://laweconcenter.org/wp-content/uploads/2018/01/20180123160215215_16-1454-State-of-Ohio-v-American-Express-Brief-for-Amici-Curiae-Antitrust-Law-in-Support-of-Respondents.pdf> accessed 12 February 2019. 12 Greg Werden, ‘Why (Ever) Define Markets? An Answer to Professor Kaplow’ (2013) 78 Antitrust LJ 729, 741 (2013) (emphasis added). 13 See Amex Slip Op., 14 (‘Evaluating both sides of a two-sided transaction platform is also necessary to accurately assess competition.’). 14 US v Am Express Co (n 4) 196–97 (citations omitted). 15 eg Brief of 25 Professors of Antitrust Law as Amici Curiae Supporting Petitioners in Ohio et al v Amex Co, 4–5 [hereinafter, ‘Amicus Brief of 25 Antitrust Law Professors’] (‘[T]he very different services that payment card companies offer to merchants and cardholders respectively are not substitute products, and so do not belong in the same relevant market from the standpoint of antitrust law and economics.’). See also Brief of 28 Professors of Antitrust Law as Amici Curiae Supporting Petitioners in Ohio et al v Amex Co, 17–18 [hereinafter, ‘Amicus Brief of 28 Antitrust Law Professors’]. Michael Katz and Jonathan Sallet argue that defining separate markets for each side of a two-sided advertising platform is ‘consistent with the principle that two products are in the same antitrust market only if they are sufficiently close substitutes: reading a newspaper clearly is not a substitute for purchasing advertising’. Michael Katz and Jonathan Sallet, ‘Multisided Platforms and Antitrust Enforcement’ (2018) 127 Yale LJ 2142, 2154. But they also apply this to transactional platforms, and contend that it applies generally: ‘Because users on different sides of a platform have different economic interests, it is inappropriate to view platform competition as being for a single-product offered at a single (i.e., net, two-sided) price.’) ibid 2170. 16 See Katz and Sallet, ibid 2170 (‘[In order] to reach sound conclusions about market power, competition, and consumer welfare, any significant linkages and feedback mechanisms among the different sides must be taken into account.’). 17 Amicus Brief of 25 Antitrust Law Professors, 16 (quoting United States v EI DuPont de Nemours & Co (1956) 351 US 377, 394); See also Katz and Sallet (n 15) 2161–62 (2018) (‘We ultimately argue [] that platforms are better viewed as operating in multiple separate, yet deeply interrelated, markets …. Crucially, when applied appropriately, this approach gives careful consideration to any significant linkages between the markets on the different sides of a platform that might be present.’), and ibid 2142 (‘We … reject[] the view that anticompetitive conduct harming users on one side of a platform can be justified so long as that harm funds benefits for users on another side.’). 18 Amex Slip Op., 11 (Breyer J, dissenting). 19 The 2010 Horizontal Merger Guidelines, for example, assert that ‘market definition focuses solely on demand substitution factors’. US Dep’t of Justice and FTC, Horizontal Merger Guidelines (2010) § 4. Importantly, however, they do go on to note that ‘[t]he responsive actions of suppliers are also important in competitive analysis. They are considered in these Guidelines in the sections addressing the identification of market participants, the measurement of market shares, the analysis of competitive effects, and entry.’ ibid. 20 Kenneth G Elzinga and Vandy M Howell, ‘Geographic Market Definition in the Merger Guidelines: A Retrospective Analysis’ (2018) 53 Rev Indus Org 453. See also US Dep’t of Justice and FTC, ibid, § 4.2 (‘The arena of competition affected by [a] merger may be geographically bounded if geography limits … some suppliers’ willingness or ability to serve some customers.’). 21 US Dep’t of Justice and FTC (n 19) § 5.1. 22 Dissenting Statement of Commissioner J Thomas Rosch, In the Matter of Laboratory Corporation of America and Laboratory Corporation of America Holdings, FTC Docket No. 9345, at 1 (citing 2010 HMGs, fn 4). 23 Amex Slip Op., 12. 24 See Benjamin Klein and others, ‘Competition in Two-Sided Markets: The Antitrust Economics of Payment Card Interchange Fees’ (2006) 73 Antitrust LJ 571, 583 (‘Because cardholders and merchants jointly consume a single product, payment card transactions, their consumption of payment card transactions must be directly proportional.’). 25 Amex Slip Op., 13 (‘These platforms facilitate a single, simultaneous transaction between participants …. Transaction platforms are thus better understood as “suppl[ying] only one product” — transactions.’) (quoting Klein and others, ibid 580). 26 ibid 14. 27 See eg James D Ratliff and Daniel L Rubinfeld, ‘Is There a Market for Organic Search Engine Results and Can Their Manipulation Give Rise to Antitrust Liability?’ (2014) 10 J Comp L Econ 517. 28 See Jean-Charles Rochet and Jean Tirole, ‘Two-Sided Markets: A Progress Report’ (2006) 37 Rand J Econ 645, 646. 29 David S Evans, ‘The Antitrust Economics of Multi-Sided Platform Markets’ (2003) 20 Yale J Reg 325, 355–56. See also Jean-Charles Rochet and Jean Tirole, ‘Platform Competition in Two-Sided Markets’ (2003) 1 J Eur Econ Assn 990, 1018. 30 See eg Michal S Gal and Daniel L Rubinfeld, ‘The Hidden Cost of Free Goods’ (2016) 80 Antitrust LJ 521, 557 (discussing the problematic French Competition Tribunal decision in Bottin Cartographes v Google Inc, where ‘[d]isregarding the product’s two-sided market, and its cross-network effects, the court possibly prevented a welfare-increasing business strategy’). 31 See eg Brief for Amici Curiae Prof David S Evans and Prof Richard Schmalensee in Support of Respondents in Ohio et al v Amex Co, 21 Judgment of the 15th Chamber of the Paris Commercial Tribunal dated 31 January 2012, Bottin Cartographes v Google Inc. & Google France, available at http://www.legalis.net/spip.php?page=jurisprudence-decision&id_article=3327. (‘The first stage of the rule of reason analysis involves determining whether the conduct is anticompetitive. The economic literature on two-sided platforms shows that there is no basis for presuming one could, as a general matter, know the answer to that question without considering both sides of the platform.’). 32 Amex Slip Op., 14 (quoting Brooke Group Ltd v Brown & Williamson Tobacco Corp (1993) 509 US 209, 226 (internal quotation marks omitted)). 33 See eg Amicus Brief of 28 Antitrust Law Professors, 3. 34 See Amex Slip Op, 9–10 for a basic description of the three-step, rule-of-reason, burden-shifting framework. 35 See Amicus Brief of 28 Antitrust Law Professors, 22 (‘[T]his Court held, in a case where the defendant operated a two-sided platform, that each side represented a “separate … market” and that injuring competition in the restrained market alone was sufficient to violate the Sherman Act.’). 36 Amicus Brief of Law & Econ Scholars, 19. See also Amex Slip Op., 16 (‘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.’). 37 US v Am Express Co (n 4) 198. 38 Katz and Sallet (n 15) 2161 (‘[I]t is essential to account for any significant feedback effects and possible changes in prices on both sides of a platform when assessing whether a particular firm has substantial market power.’). 39 ibid 2146. 40 ibid 2167 (emphasis added). 41 ibid 2168. 42 ibid. 43 ibid 2159 (emphasis added). 44 See Amex Slip Op., 16 (‘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.’). 45 See Todd J Zywicki, Geoffrey A Manne and Kristian Stout, ‘Behavioral Economics Goes to Court: The Fundamental Flaws in the Behavioral Law & Economics Arguments Against No-Surcharge Laws’ (2017) 82 Mo L Rev 769, 787–88 (‘Whatever the reduction in the network’s value caused by this [strategic] conduct, each merchant would bear only a fraction of that cost but enjoy the full benefit of [its strategic conduct]. So, each establishment in the network would have the individual incentive to cheat to the collective harm of everyone.’). 46 Amicus Brief of 28 Antitrust Law Professors, 35. 47 (1963) 374 US 321 [hereinafter, ‘PNB’]; (1972) 405 US 596 [hereinafter ‘Topco’]. 48 US v Am Express Co (n 2) 171. 49 ibid 247 (citing Topco, 610 and PNB, 370). 50 Topco, 609. 51 ibid 610–11 (citing PNB, 371) (emphasis added). 52 Amex Slip Op., 24 (Breyer J, dissenting). 53 Katz and Sallet (n 15) 2171. 54 Amicus Brief of 28 Antitrust Law Professors, 3. See also ibid 33 (courts should not consider ‘out-of-market benefits’), and ibid 12 (balancing must occur ‘within the relevant market’ (emphasis in original)). The American Express opinion, the rule of reason, and tech platforms Tim Wu Tim Wu Julius Silver Professor of Law, Science, and Technology, Columbia University Law School Julius Silver Professor of Law, Science, and Technology, Columbia University Law School *Tim Wu, Julius Silver Professor of Law, Science, and Technology, Columbia University Law School. Email: wu@pobox.com. I wish to thank Nicole Fleming for research assistance. This commentary on the US Supreme Court’s American Express opinion is limited to two points (readers unfamiliar with the decision can read the summary in the margin).1 First, in my view, the American Express opinions in both the Supreme Court and Second Circuit exemplify an unfortunate trend in the antitrust law, one best described as the tendency to elevate theory over evidence. There is some irony in this trend: there was a time, at its height during the 1970s and 1980s, that critics charged the antitrust with too much pro-government bias and an indifference to realities of economy. That movement yielded the well-known turn towards a more evidence-based, ‘rule of reason’ approach. But today, in decisions like American Express, matters have come full circle, so much that courts in cases like American Express are once again willing to disregard evidence of anticompetitive harm in favour of abstract theory—in favour of the defendant. As such, Justice Thomas’s opinion in American Express can be seen as a mirror image of the per se rules that once prevailed throughout antitrust. If those per se cases suggested an automatic victory for the plaintiff without evidence of competitive harm, the American Express courts have granted dismissal for the defendant despite good evidence of such harm. And as a result, American Express is an example of a retreat from the evidentiary focus supposedly motivating the Court’s shift towards a rule-of-reason analysis. That is the first point. The second concerns the future impact of the decision. The Supreme Court’s opinion does have one great merit as compared to the Second Circuit’s: it is narrow, indeed far narrower than some have suggested. In particular, claims that the decision ‘immunizes’ the major tech platforms from antitrust scrutiny are incorrect.2 It seems clear that firms like Google, Facebook, and Twitter are not covered by the American Express opinion, as explained here. It is true that an opinion creating rules for all two-sided platforms would have fundamentally changed much of antitrust law, by reaching so much of American commerce. For the concept of a two-sided platform is open-ended enough to conceivably describe businesses as diverse as malls, sports leagues, real estate agents, stock exchanges, and most tech platforms.3 The Second Circuit’s alarmingly open-ended ruling in favour of American Express shows the need for careful treatment in this area.4 But the Court emphasized that credit card companies are so-called ‘transaction platforms,’ a subset of two-sided platforms.5 The opinion goes on to define transaction platforms as those that cannot provide a service to one side of the market independently—those that, by necessity, facilitate a ‘simultaneous’ interaction between the two sides.6 As such, according to the Court, transaction platforms only compete with other transaction platforms.7 That limitation makes American Express inapplicable to most major tech platforms as well as most of what are traditionally called ‘two-sided’ platforms. Consider, for example, two firms of much interest in the present discourse: Facebook and Google. While both firms, roughly speaking, have business models based on bringing together different groups—advertisers and users—they do not exist for the sole purpose of facilitating simultaneous transactions between the two. Users do plenty of things on Google or Facebook outside of transactions with advertisers, such as interacting with other platform users, conducting searches, and so on. For that matter, the American Express opinion would not cover a nightclub that offers a subsidized ‘ladies’ night’—the textbook example of a two-sided platform8—given that the nightclub does not facilitate a ‘simultaneous’ transaction between the male and female customers, and offers more than just the facilitation of such transactions. Of course, there are major tech platforms whose existence seems predicated on facilitating simultaneous transactions and little more. An obvious example is ride-sharing companies such as Uber, Lyft, and their brethren. However, that said, it is not entirely clear that American Express offers a defendant something ultimately all that different than what the defendant already might have presented as a pro-competitive justification under the rule-of-reason analysis. This further suggests that American Express is a case of limited long-term import. A final point relates to the slightly unpredictable impact of the American Express ruling. By suggesting that the only competitors to two-sided transaction platforms are other two-sided transaction platforms, the logic of the decision may sometimes yield very narrow market definitions. Consider, for example, a hypothetical merger of ride-sharing services Uber and Lyft, and assume, for a moment, that both are ‘transaction platforms’. If their only competitors are other two-sided ride-sharing operations, then taxis and other forms of competition might be excluded, yielding higher market shares and making any potential merger more likely to violate the Clayton Act. All this goes to underscore the essential folly of the decision. By mixing in unnecessarily complex economic principles into the decision, the courts are creating an antitrust law in which generalist courts, wary of the rule-of-reason’s balancing, might cherry-pick a new economic theory for each case so as to yield the result preferred. That has the impact of dragging the law further and further away from anything approaching the competitive realities of the industry in question. I. RULE-OF-REASON ANALYSIS IN AMERICAN EXPRESS The American Express opinion should be understood in the broader context of the last several decades and the ruse of ‘rule-of-reason’ antitrust. As any student of antitrust knows, there was a time when American law broadly took various forms of restrictive agreements as categorically, or per se anticompetitive. Such agreements, including many forms of vertical agreement, were categorically condemned without any inquiry into their purpose or effects.9 But over the last forty years, antitrust doctrine has retreated from per se rules and moved towards the ‘rule-of-reason’ analysis for most categories of alleged anticompetitive conduct.10 The reasons for this shift are relevant to the American Express decision. The attack by courts on per se rules came from a desire that antitrust law analysis be more deeply grounded in evidence. As adopted by the courts, and in the hands of more thoughtful critics, the movement away from per se rules was not premised on a blanket tolerance of potentially restrictive vertical agreements. Rather, courts sought proof—and in particular, proof of harm to competition. Take, for example, the problem of vertical pricing restraints, also known as retail (or resale) price maintenance. Such restraints might serve the salutary purpose of promoting peace among retailers or protecting a value-added retail model.11 But these benign motives are not the only possible explanation. Courts and commentators are also willing to consider that what is facially described as merely a means for ensuring peace among retailers may turn out, on further investigation, to an anticompetitive scheme. For instance, the restraints could be the means by which the retailers create their own pricing cartel.12 The premise of a rule-of-reason analysis is to find, as opposed to assume an answer, or so Justice Kennedy’s opinion in Leegin13 suggests. Leegin and similar cases promised antitrust law that was smarter than per se rules were—antitrust law that targeted only proven harms to the competitive process. American Express violates that promise—not explicitly, but in its method. After inviting the government to present evidence of harm under a rule-of-reason framework, the Court effectively moved the goalposts at the point of appeal, demanding something far more challenging to prove. In other words, it suggested that direct evidence of harm to competition was not enough. In that undertaking, the Court committed the same mistakes that the vertical per se rules did at their worst, by choosing to ignore evidence in favour of theory. We can show this more explicitly by considering the steps in a rule-of-reason analysis and how they interacted with the American Express litigation. The prima facie rule-of-reason case under section 1 of the Sherman Act asks the plaintiff to show two elements: proof of an agreement in restraint of competition and proof of an anticompetitive effect.14 In the burden-shifting analysis, the defendant may then offer procompetitive rationales.15 The fact-finder—that is, the trial court—then determines whether the anticompetitive effect on competition was demonstrated. The anti-steering provisions are a facially restrictive agreement designed—successfully—to insulate American Express against competition from other credit card providers at the point of sale. This was no secret. American Express sought to prevent merchants from steering consumers towards less expensive alternatives, and in that manner, prevent the company from losing market share to cards with lower merchant fees. For procompetitive rationales, American Express justified the restrictions by arguing that its business model depended on both higher merchant fees and more benefits for its cardholders.16 At the American Express trial, the government presented significant evidence tending to support the premise that the anti-steering rules did inhibit price competition between credit card companies.17 For instance, as Justice Breyer noted in his dissent, the anti-steering provisions stifled the attempt of a competing credit card company to innovate in favour of lower prices.18 Discover tried a business model during the 1990s that offered much lower fees to merchants.19 But the anti-steering provisions of American Express and other credit card companies meant that merchants could not communicate to customers a preference for Discover.20 Over time, Discover ultimately gave up on price competition and decided to raise its merchant fees as well.21 This outcome tends to suggest a failure of the competitive process along with an industry-wide lack of price discipline. Importantly, American Express’s practices suggested that it felt little or no competitive pressure to lower its prices. The company was able to raise its prices 20 times during the disputed five-year period without losing the business of any large merchants, suggesting further that its anti-steering rules affected competition.22 This outcome is not surprising, given the intent of the restrictions. Nonetheless, if the Second Circuit or Supreme Court believed that American Express should have won based on the evidence, the appropriate remedy within the rule-of-reason framework would have been to credit American Express’s procompetitive justifications. The courts could have found the district court too dismissive of the importance of the anti-steering rules to its distinctive business model. The appellate courts could have concluded that the anticompetitive anti-steering rule was, despite harming competitive, nonetheless justified because it made possible the company’s particular model of prestigious, reward-intensive card. Yet they did not, as we have seen, and instead opted for a far more confusing and theoretically complex approach.23 Why take a more complicated approach? While one can only speculate, one reason is that the Courts may have been wary of the ‘balancing’ called for by the rule of reason after a procompetitive rationale is credited.24 Literature suggests that courts are uncomfortable with such balancing to the point that they almost never do it.25 Sometimes courts avoid the challenge of balancing by using the ‘less restrictive alternative’ test: comparing the action to a hypothesized alternative and asking whether the alternative action is ‘less restrictive’ and hence less harmful.26 Perhaps, then, the strange path taken in the American Express case was a reflection of that anxiety. Instead of balancing, the Court disposed of the case by adjusting the market definition for transactional platforms to include ‘both sides’ of the market.27 This creates an incoherent market. Market definitions are determined by evaluating available substitutions,28 yet somehow the credit card market includes services that could not possibly be substitutes. (How can cardholder and merchant services be interchangeable?) Perhaps more charitably, it redefined the product as the entire transaction, yielding a market for transaction facilitators, and then searched for evidence—on appeal—that competition between networks had been damaged. The fact that this demand regarding the market definition was imposed on appeal is important. Had the government known it needed to show harm to competition with that market definition, it might have done so. Even taking into account a better ability to reach consumers, the government could argue that the anti-steering rules were still anticompetitive. But none of this was developed at trial. Instead, the Court’s approach allowed it to make its own de novo findings. At bottom, the approach announced the Court is unprecedented, procedurally indefensible, unnecessarily complex, and ultimately incoherent. The per se rules were at least honest about what they were doing. Their mirror-image counterpart in American Express takes the less-appealing approach of using complex economic theory to create near-impossible burdens of proof—burdens particularly hard to meet when they emerge on appeal. At worst, they offer a highly jazzed-up way of getting around some awfully damaging facts. The danger inherent in the American Express decision’s deviation from the rule of reason lies in its arbitrariness. As everyone knows, it is nearly always possible to manipulate market definition to justify deciding a case in a particular direction.29 The existence of direct evidence of harm to competition serves as one of the checks on that possibility. The promise of cases decided on the basis of good, direct evidence is at the core of the rule of reason, as discussed earlier. But if a court can ignore direct evidence of harm to competition by demanding something else be proven, the whole analysis is—once again—open to the whims of the judiciary. Any case can be decided by adopting an economic theory tailored to the facts at hand. All this said, the Court’s deviation from a typical rule-of-reason analysis also limits its relevance as a decision. The five justices in the American Express majority certainly did not overrule landmark cases that describe the rule of reason and how it is meant to function, such as Board of Trade of the City of Chicago v United States,30National Society of Professional Engineers v United States,31Broadcast Music, Inc v Columbia Broadcasting System, Inc (BMI),32California Dental Association v Federal Trade Commission,33 and Leegin.34 To the contrary, the American Express decision claims to be following the rule-of-reason framework.35 Accordingly, the case ultimately falls in the category of what a Supreme Court clerk would call ‘fact-bound error’. II. ‘TRANSACTION PLATFORMS’ IN AMERICAN EXPRESS As already mentioned, the American Express holding is much narrower than it might have been. The Second Circuit and some amici suggested an approach that could have had sweeping implications by potentially establishing a new form of market definition analysis for any case involving a two-sided platform.36 And the concept of a two-sided platform is expansive. Consider that businesses at issue in many landmark antitrust cases could be thought of as two-sided platforms. An oil refinery like that in Standard Oil Co of New Jersey v United States brings together crude oil producers and gasoline buyers;37 a rights association like that in Broad Music, Inc v Columbia Broad Sys, Inc (BMI) brings together composers and media outlets;38 a sports league like that in National Collegiate Athletic Association v Board of Regents of the Univ of Oklahoma brings together sports fans and sponsors;39 an operating system like that in United States v Microsoft Corp brings together application makers and computer users.40 Hence, a decision applicable to any two-sided platform might have shaken American antitrust law to its core. Instead, the American Express holding is limited to ‘transaction platforms’, explicitly excluding ‘non-transaction platforms’ from its analysis.41 This narrow approach may have reflected discomfort from members of the majority as well as an unwillingness to explicitly overrule prior cases.42 The term ‘transaction platform’ is new to antitrust,43 but the Court defines it as a two-sided market in which the platform ‘facilitate[s] a single, simultaneous transaction between participants’.44 This means platforms that, by their nature or design, offer only services of simultaneously conducting a transaction between parties on each side of the market—platforms that ‘cannot sell transaction services to either [one side or the other] individually’.45 In such a setting, the Court reasoned, platforms ‘exhibit more pronounced indirect network effects and interconnected pricing and demand’.46 The limitation is still abstract but suggests that the American Express decision would not apply to most two-sided markets, and in particular, would not apply to most major tech platforms. Accordingly, some of the claims about the case’s relevance to the tech field are overstated. As described above, an enormous range of firms operate what might be called two-sided platforms, but only a small subset are transaction platforms. To take a few important examples, firms like Google, Facebook, and Twitter are not covered by the American Express opinion. Their business models depend on attracting users and ultimately reselling their audiences to advertisers.47 Users do plenty of things on each of these platforms that does not involve making a transaction with an advertiser. Indeed, the lure of Facebook and Twitter is actually interactions with other users, while Google, in its original conception, primarily offered a means of finding content on the web. To that end, by the Court’s own implication, Google, Facebook, and Twitter are covered by the rule of Times-Picayune Publishing Co v United States48 and not American Express. The Court makes this point clearest in a footnote distinguishing ‘non-transaction platforms’ (and also implicitly recognizes the existence of attention markets49): ‘Non-transaction platforms, by contrast, often do compete with companies that do not operate on both sides of their platform. A newspaper that sells advertising, for example, might have to compete with a television network, even though the two do not meaningfully compete for viewers.’50 Consistent with this, Times Picayune had established earlier that the market definition for a newspaper should focus on the entity’s dominance in advertising, not readership.51 The implication is clear for firms like Google, Facebook, and Twitter, which do compete for digital advertising but do not directly compete for users in the relevant attention markets: American Express only applies to firms that are both transactional and in competition on both sides of the platform. A firm like Amazon might seem at first like a closer case, given that the e-commerce giant clearly facilitates transactions between its users and sellers. That said, it seems clear that the Court in American Express could not have intended for every retail operation to be treated as a ‘transaction platform’ in the meaning of the opinion. Stores that bring together buyers and sellers can face competition from stores that exclusively carry their own goods. For example, Best Buy, which sells many brands of computers, still competes with the Apple store, which sells computers only of its own brand. There is far more to the business of Amazon and other retailers than the simultaneous facilitation of a transaction. Indeed, credit card companies are often the entities that complete the retail transaction, suggesting that they are distinct from the retailers themselves. Among platforms of much antitrust discussion nowadays, the American Express decision would seem to be most relevant to Uber and Lyft. Such ride-sharing firms primarily owe their existence to the market for facilitation of transactions between drivers and riders. So, if Uber were accused of anticompetitive practices, American Express offers it an opportunity to argue that even in the face of direct evidence of harm, a court must take into account the goal (if any) of making Uber’s services more attractive to its passengers. Imagine that Uber sought to hurt its rivals in the competition for drivers by trying to sabotage Lyft’s recruitment of new drivers. (A real-world example: In a ride-sharing version of the ‘ding-dong ditch’, Uber employees reportedly ordered and canceled thousands of Lyft rides to suggest to Lyft drivers that Lyft was unreliable.52) After American Express, Uber may be able to suggest that its conduct was ultimately in the interest of making its product more appealing to consumers. Uber customers benefit from lower wait times if Uber has more drivers available. It seems hard to argue that sabotaging a rival might accomplish that, but stranger arguments have been made in federal court. As this illustration suggests, what American Express offers Uber in terms of a defence is not really all that different from the rule of reason’s procompetitive rationales. This further indicates that much of the mess created by the American Express case could have been minimized: if the appellate courts were convinced that the anti-steering provisions actually benefited competition among cards as between consumers, they should have found ‘clearly erroneous’ the district court’s assessment of American Express’s procompetitive justifications. Just as hard cases make bad law, so does a fear of the rule of reason’s balancing. The Uber–Lyft discussion also helps demonstrate some of the unpredictable implications of the American Express decision. The Court suggests that two-sided transaction platforms really only compete with each other and not with one-sided businesses. That point suggests that, in some cases, the Court’s analysis might yield much narrower market definitions that might ordinarily be the case. It might suggest that, in a hypothetical merger of Uber and Lyft, traditional competitors like taxis or black cars are not actually competitors to the ride-sharing firms, yielding a much narrower market definition and a higher likelihood of presumptive illegality. All this underscores the inherent incoherence of the analysis. III. AFTERWORD At a public hearing of the Federal Trade Commission in 2018, panel members were asked whether the Microsoft case53 would have reached a different outcome if it had been decided after American Express. Back in 2001, the D.C. Circuit found that Microsoft had maintained a monopoly in the operating system market in violation of section 2 of the Sherman Act.54 At the time, there was fear that Microsoft would kill the then-up-and-coming challenger Netscape, monopolize the browser market, and use that point of control to dominate the coming age of the web.55 Factually, though Microsoft dealt with a two-sided platform,56 this alone would not be enough to assume that American Express would cause a different result. As I have explained here, the American Express holding’s self-imposed limits suggest it is not directly relevant to two-sided platforms generally—only transactional platforms. The operating system at issue in Microsoft is a two-sided platform, but not one of a transactional nature. But of course, in another way, American Express could have procedural consequences. In cases like this, the Court is creating opportunities for appellate courts to work around the Justice Department and a district court that have turned up clear and unmistakable evidence of harm to competition. Any halfway-decent lawyer can find an economic theory that could account for all but the most blatantly anticompetitive acts. At that point, a court—if allowed to convert a procompetitive justification into a burden of proof—gains a new way to dismiss any case. For example, in Microsoft, a hostile appellate court might have demanded proof not just that Netscape was excluded (and hence that competition was compromised), but to prove the impossible: that the entire ecosystem was damaged as a consequence, or something along those lines. American Express suggests that a judge can keep demanding more proof, in concentric lines, until the government’s lawsuit collapses. Here we have the opinion's most pernicious attribute. At bottom, it offers appellate courts the comfort of the Supreme Court’s support in finding novel ways to throw out antitrust cases with strong evidence of anticompetitive effects. Footnotes 1 The case concerned the legality of American Express’s use of ‘anti-steering provisions’. The provisions are contractual rules meant to prevent merchants who accept American Express from, nonetheless, steering consumers towards cheaper options (ie those with lower merchant fees). The Justice Department and seventeen states sued American Express based on the premise that the rules inhibit competition and prevent cards with lower fees, like Discovery, from gaining a competitive advantage by lowering their prices. Though the government won at the trial level, see United States v Am Express Co (EDNY 2015) 88 F Supp 3d 143, American Express prevailed before the Second Circuit and the Supreme Court. Ohio, et al v Am Express Co, et al (2018) 138 S Ct 2274. 2 To the degree that any writings with my byline may suggest otherwise, I remind readers that the author does not control his headlines. 3 Shopping malls bring together shoppers and stores. Geoffrey G Parker and Marshall W Van Alstyne, ‘Two-Sided Network Effects: A Theory of Information Product Design’ (2005) 51 Mgmt Sci 1494, 1495 table 1. Stock exchanges bring together equity purchasers and listed companies: ibid. Residential real-estate agents bring together home buyers and home sellers; Id. Travel reservation websites like Expedia, Travelocity, and Orbitz bring together travelers and hotels, airlines, and rental car services; and Id. Sports leagues bring together fans and advertisers. Helmut Max Dietl and others, ‘A Contest Model of a Professional Sports League with Two-Sided Markets’ (2012) 232 J Econ Stat 336, 337. 4 See nn 36–40 and accompanying text. 5 ibid 2280 (‘[C]redit-card networks are a special type of two-sided platform known as a “transaction” platform.’) 6 ibid 2280–81 (defining a ‘transaction’ platform and noting that ‘no credit-card transaction can occur unless both the merchant and the cardholder simultaneously agree to use the same credit-card network’). 7 ibid 2287 (‘Only other two-sided platforms can compete with a two-sided platform for transactions.’) 8 See eg Parker and Van Alstyne (n 3) 1495 table 1 (listing ‘ladies’ nights’ as an example of a two-sided market, as the bars or restaurants act as an intermediary that brings together men’s admissions and women’s admissions); Julian Wright, ‘One-Sided Logic in Two-Sided Markets’ (2004) 3 Rev Network Econ 44, 45–50 (illustrating fallacies that arise from applying the logic of one-sided markets to two-sided markets, using ‘ladies’ nights’ at heterosexual nightclubs as an example of the latter). 9 See Mark A Lemley and Christopher R Leslie, ‘Categorical Analysis in Antitrust Jurisprudence’ (2008) 93 Iowa L Rev. 1207, 1213–14, 1219–20. 10 See eg Leegin Creative Leather Prods Inc v PSKS Inc (2007) 551 US 877, 885 (2007) (‘The rule of reason is the accepted standard for testing whether a practice restrains trade in violation of §1.’); Texaco Inc v Dagher (2006) 547 US 1, 5 (‘[T]his Court presumptively applies rule of reason analysis … .’); State Oil Co v. Khan (1997) 522 US 3, 10 (As a consequence, most antitrust claims are analyzed under a “rule of reason,” according to which the finder of fact must decide whether the questioned practice imposes an unreasonable restraint on competition, taking into account a variety of factors, including specific information about the relevant business, its condition before and after the restraint was imposed, and the restraint's history, nature, and effect); Continental TV, Inc v GTE Sylvania Inc (1977) 433 US 36, 49 (‘Since the early years of this century a judicial gloss on this statutory language has established the “rule of reason” as the prevailing standard of analysis.’); Michael Carrier, ‘The Rule of Reason: An Empirical Update for the 21st Century’ (2009) 16 Geo Mason L Rev 827, 828 (‘The rule of reason is even more crucial today than it was a decade ago. The Supreme Court has increasingly overturned per se rules of illegality.’) (footnote omitted). 11 See Robert H Bork, The Antitrust Paradox (1st edn, Basic Books 1978) 280–89. 12 See Leegin (n 10) 893 (describing the risk of vertical price restraints being used to organize cartels at the retail level); see also Monsanto Co v Spray-Rite Service Corp (1983) 465 US 752, 765–66 (describing behaviour that would amount to a retail cartel). 13 In holding that vertical price restraints should be judged under the rule of reason, Justice Kennedy emphasized the importance of evidence in determining the effects of the restraints. See Leegin (n 10) 889–90 (explaining that vertical price restraints may be procompetitive in some cases but anticompetitive in others, and discussing various kinds of proof a court might use to reach at either of these conclusions). 14 See Phillip E Areeda and Herbert Hovenkamp, Antitrust Law (4th edn, 2017) ¶1502. 15 ibid. 16 US v Am Express Co (n 1) 224–38. 17 ibid 238. 18 Ohio v Am Express Co (n 1) 2293–94 (Breyer J, dissenting). 19 US v Am Express Co (n 1) 213–14. 20 ibid. 21 ibid. 22 ibid 195–97. 23 Because the Court found that the government had not proven anticompetitive harm per the rule of reason’s first step, it did not reach the second step of crediting the defendant’s procompetitive justifications. See Ohio v Am Express Co (n 1) 2290. 24 Judges are often not comfortable with the task of balancing harms and benefits in antitrust cases. C Scott Hemphill, ‘Less Restrictive Alternatives in Antitrust Law’ (2016) 116 Colum L Rev 927, 947–51. As Hemphill suggests, they are generalists who see antitrust cases comparatively rarely; they have a limited fact-finding capacity, which is hampered by the adversarial nature of antitrust litigation; and they can sometimes feel as though the values at stake are incommensurable. ibid. 25 For instance, a study of 222 rule-of-reason cases (decided by judges as opposed to juries) over a decade showed courts disposing of 97% of cases at the first stage of the burden-shifting framework, on the grounds that there was no anticompetitive effect. See eg Carrier (n 10) 828–29. Judges balanced in only 2% of the cases. ibid. 26 Hemphill (n 24) 952–55 (describing how judges turn to the less restrictive alternative test due to anxiety about balancing). 27 Ohio v Am Express Co (n 1) 2286. 28 United States v EI du Pont de Nemours & Co (Cellophane) (1956) 351 US 377, 395 (holding that the relevant market must include all products ‘reasonably interchangeable by consumers for the same purposes’). 29 See eg Jones v Metzger Dairies, Inc (5th Cir 1964) 334 F 2d 919, 922 (‘[T]he term “relevant market” is a rather evanescent term which can be skillfully manipulated somewhat in the manner of an accordion.’); Louis Altman and Mark Pollack, Callmann on Unfair Competition, Trademarks, and Monopolies (4th edn) § 4:31. 30 (1918) 246 US 231. 31 (1978) 435 US 679. 32 (1979) 441 US 1. 33 (1999) 526 US 756. 34 Leegin (n 10). 35 Ohio v Am Express Co (n 1) 2284(stating that ‘like nearly every other vertical restraint, the antisteering provisions should be assessed under the rule of reason’). 36 The Second Circuit opinion addressed two-sided platforms more generally. See United States v Am Express Co (2nd Cir 2016) 838 F 3d 179, 185–86, 196–200. 37 (1911) 221 US 1, 32–33. 38 (1979) 441 US 1, 4–5. 39 (1984) 468 US 85, 88–89. 40 (DC Cir 2001) 253 F 3d 34, 51–54. 41 Ohio v Am Express Co (n 1) 2285–87, 2287 fn 8. 42 For instance, the Court had previously ruled that for analyzing a two-sided platform in the case of a newspaper, dominance in one side of the market—not both—is key. See Times-Picayune Publ’g Co v United States (1953) 345 US 594, 610 (holding that for a newspaper, a ‘dual trader’ connecting advertisers and readers, ‘dominance in the advertising market, not in readership, must be decisive in gauging the legality’ of the conduct). In contrast, in American Express, the Court used a newspaper to illustrate the concept of a non-transaction platform. See nn 50–51 and accompanying text. 43 A Westlaw search for federal and state cases that used both the terms ‘transaction platform’ and ‘antitrust’ returns nothing prior to the June 2018 American Express decision. 44 Ohio v Am Express Co (n 1) 2286. 45 ibid. 46 ibid. 47 See Tim Wu, ‘Blind Spot: The Attention Economy and the Law’ Antitrust LJ (2018, forthcoming). 48 Times-Picayune (n 42). 49 See generally Tim Wu, The Attention Merchants: The Epic Scramble to Get Inside Our Heads (2016). 50 Ohio v Am Express Co (n 1) 2287 fn 8. 51 Times-Picayune (n 42) 610. 52 See Erica Fink, Uber’s Dirty Tricks Quantified: Rival Counts 5,560 Canceled Rides (CNN, 12 August 2014) <https://money.cnn.com/2014/08/11/technology/uber-fake-ride-requests-lyft/> accessed 13 February 2019, (reporting on allegations by Lyft that Uber employees cancelled 5560 Lyft rides between October 2013 and August 2014); see also Laurie Segall, Uber Rival Accuses Car Service of Dirty Tactics (CNN, 24 January 2014) <https://money.cnn.com/2014/01/24/technology/social/uber-gett/> accessed 13 February 2019, (describing similar allegations that Uber employees in New York cancelled rides with Gett, a rival ride-sharing application). 53 Microsoft Corp (n 40). 54 ibid 64, 71, 74, 78. Early on, it seemed like the remedy would be for Microsoft to be broken into two companies: one for the Windows operating system and one for other products. See United States v Microsoft Corp (DDC 2000) 97 F Supp 2d 59, 64–74. Ultimately though, the company was allowed to stay whole, albeit with some limitations on its behaviour. See United States v Microsoft Corp (DDC, 22 April 2009) No Civ A 98-1232, 2009 WL 1348218 (modifying the judgment for a second time). 55 Richard Blumenthal and Tim Wu, ‘What the Microsoft Antitrust Case Taught Us’ New York Times (New york, NY, 18 May 2018) <https://www.nytimes.com/2018/05/18/opinion/microsoft-antitrust-case.html> accessed 13 February 2019. 56 See Microsoft Corp (n 40) 51–54. © The Author(s) 2019. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Ohio v American Express
JF - Journal of Antitrust Enforcement
DO - 10.1093/jaenfo/jnz003
DA - 2019-03-01
UR - https://www.deepdyve.com/lp/oxford-university-press/ohio-v-american-express-wJ0UHQjtdO
SP - 104
VL - 7
IS - 1
DP - DeepDyve
ER -