chevron-down Created with Sketch Beta.

The Antitrust Source

The Antitrust Source | June 2021

Missing the Forest for the Trees: The Application of Amex in United States v. Sabre

Lindsey Edwards and Jonathan M Jacobson


  • In the two and a half years since the Supreme Court decided Ohio v. American Express Co., the scope and effect of the Court’s holding that certain markets must be treated as twosided has been the subject of much debate within the antitrust bar.
  • In United States v. Sabre Corp., Judge Stark held that the U.S. Department of Justice failed to identify a proper relevant market because, under Amex, “[a]s a matter of antitrust law, Sabre, a two-sided transaction platform, only competes with other two-sided platforms, but Farelogix only operates on the airline side of Sabre’s platform."
  • The authors argue that Sabre/Farelogix court’s application of Amex was inappropriate given the facts at hand. According to the authors, the finding that two-sided platforms only compete with other two-sided platforms for transactions was based on factual analysis of the credit card industry. Extending that finding as a broad principle of antitrust law that requires ignoring evidence of competition between merging parties where only one party is a two-sided platform misses the forest for the trees.
Missing the Forest for the Trees: The Application of Amex in United States v. Sabre
Roksana Bashyrova via Getty Images

Jump to:

In the two and a half years since the Supreme Court decided Ohio v. American Express Co. (Amex), the scope and effect of the Court’s holding that certain markets must be treated as two-sided has been the subject of much debate within the antitrust bar. Commentators have questioned whether the Court’s holding applies only narrowly to operators in the credit card industry (or other “transaction platforms” engaging in vertical restraints), or whether it functions more broadly as a shield for big tech platforms by imposing a higher burden on plaintiffs challenging conduct in two-sided markets. 

To date, in cases involving industries other than the credit card industry, district courts have largely rejected defendants’ arguments that they are two-sided platforms and that Amex thus requires the court to consider effects on both sides of the market. Judge Leonard Stark of the U.S. District Court for the District of Delaware, on the other hand, issued an opinion on April 7, 2020 extending Amex’s two-sided market paradigm to a merger case in the airline travel industry. In United States v. Sabre Corp. (Sabre/Farelogix), Judge Stark held that the U.S. Department of Justice failed to identify a proper relevant market because, under Amex, “[a]s a matter of antitrust law, Sabre, a two-sided transaction platform, only competes with other two-sided platforms, but Farelogix only operates on the airline side of Sabre’s platform.” Although the decision was later vacated as moot by the Third Circuit after the parties abandoned the deal, the appeals court still considered it a “precedent” and it may continue to be cited in the future. We join several others in regarding the district court’s decision as misguided—it is inconsistent with longstanding Supreme Court merger precedents and basic economic analysis.

The Evolution of Market Definition in Merger Cases and Application to Two-Sided Markets

Defining the relevant market has been a central part of merger cases for at least 70 years. Even prior to the passage of the Celler-Kefauver Act of 1950, which amended the Clayton Act to prohibit mergers that harm competition “in any line of commerce” and arguably rendered market definition statutorily necessary, the Supreme Court understood market definition to be fundamental to the analysis of antitrust harm. Later, in 1974, in Marine Bancorp., the Court confirmed that: “Determination of the relevant product and geographic markets is ‘a necessary predicate’ to deciding whether a merger contravenes the Clayton Act.” Methods for determining the relevant market have evolved over time as the economic literature has developed, but the goal of the exercise has, since at least 1982, been the same: to identify parts of the merging parties’ businesses where the parties constrain each other’s ability to raise price or reduce output or innovation.

Courts initially struggled to delineate a specific rule or procedure for determining the relevant market in which to analyze effects of a potential merger. That changed in the 1950s, as the economic literature surrounding cross-elasticity of demand developed and the Supreme Court began applying those economic principles to define the relevant market. Analyzing cross-elasticity of demand addresses the question of whether consumers are able to substitute products, which is an important determination in merger analysis. If merging parties are the only two companies that supply substitutable goods or services, they will have an economic incentive to raise prices post-merger because customers will have nowhere to turn if they do so. In Times-Picayune Publishing Co. v. United States, the Court explained how cross-elasticity of demand could be leveraged to determine substitutability for purposes of defining the relevant market: “The circle must be drawn narrowly to exclude any other product to which, within reasonable variations in price, only a limited number of buyers will turn; in technical terms, products whose ‘cross-elasticities of demand’ are small.”

The language of Section 7 prohibits mergers that are likely to substantially lessen competition in any line of commerce, meaning plaintiffs need only prove harm in one relevant market to prevail. But plaintiffs may of course allege and prove harm in multiple relevant markets. This is a common practice, particularly when mergers involve a company that acts as a two-sided platform. For example, plaintiffs challenging newspaper mergers almost always allege a relevant market consisting of the newspapers’ advertising customers and a second relevant market consisting of their readers. Typically, courts have rejected defenses that adverse effects in one market can be justified by beneficial effects in another. But as markets evolve in light of new technologies and new ways for firms and consumers to interact and new constraints appear outside traditional brick-and-mortar markets such that incumbents cannot in fact exercise market power, that traditional thinking may be changing. The decision in Amex can be viewed in that light.

Two-sided platforms differ from one-sided companies in that the platform is facilitating interactions between two different sets of customers and is thus serving interlinked demand. Economists, most prominently Jean-Charles Rochet and Jean Tirole as well as David Evans and Richard Schmalensee, began to publish works on two-sided platforms and the network effects associated with serving interrelated customers in the early 2000s. What emerged was a rich economic literature on how profit maximization is different for two-sided platforms because of the interdependencies between the customers on each side. The profit-maximizing price for two-sided platforms depends “in a complex way on the price sensitivity of demand on both sides, the nature and intensity of the indirect network effects between the two sides, and the marginal costs that result from changing output of each side.”

Evans and Schmalensee explain that “profit-maximizing prices may entail below-cost pricing to one set of customers over the long run and, as a matter of fact, many two-sided platforms charge one side prices that are below marginal cost and are in some cases negative.” That is particularly the case when a platform exhibits indirect network effects—that is, where the value of the platform to one side depends on the level of participation on the other side. In that scenario, “[t]he effect of an increase in price on one side results in a decrease in demand on the first side because of the direct effect of the price elasticity of demand; then, demand on both sides decreases as a result of the indirect effects from the externalities.” Consequently, two-sided platforms with strong indirect network effects must account for this “feedback loop” of declining demand in determining the optimal price for each side of the platform.

Whether one or multiple relevant markets are appropriate in a case involving a two-sided platform depends on the nature of the claim, as well as the platform’s business and how interrelated the consumers on each side of the platform are. For example, the Amex Court identified newspaper companies as an example of a two-sided platform that should be analyzed as one-sided. This is because readers’ demand does not depend on the number of advertisements in the newspaper; consequently, in an antitrust case, it would not be necessary to include both newspaper readers and advertisers in the relevant market. Put another way, a newspaper’s decision to increase prices to advertisers and the subsequent decline in demand for advertising space would have little to no impact on demand on the reader side of the platform, so omitting that side of the two-sided platform when defining a relevant market would have little to no impact on the competitive analysis. On the other hand, the Amex Court found that defining a relevant market including both credit-card holders and merchants does make economic sense in an antitrust case because the value of a credit card to a cardholder depends in large part on the number of establishments at which the card can be used and, in many cases, on the rewards funded by charges to merchants.

Arriving at the correct market definition is thus a fact-intensive exercise that requires close examination of the economic realities that govern the merging parties’ conduct and the resulting consumer response.

Unpacking the Amex and Sabre/Farelogix Decisions

The Amex Decision

The Supreme Court’s Amex decision settled a near decade-long dispute regarding anti-steering contractual provisions between Amex and merchants that accept Amex credit cards. These provisions prohibited merchants from discouraging customers seeking to make a purchase in their store from using their Amex credit card. The United States, joined by several states, sued American Express, alleging that the anti-steering provisions violated Section 1 of the Sherman Act.

The district court decided that for purposes of evaluating the alleged harm, the credit card market consists of two separate relevant markets—the merchant market and the cardholder market—and that the anti-steering provisions violated Section 1 because they resulted in higher merchant fees, harming competition on the merchant side of the market. The Second Circuit reversed, holding that, in part because the merchant harm of higher fees was associated with a benefit to cardholders in terms of increased benefits and rewards, the relevant market for analyzing the alleged harm consisted of both the merchant and cardholder sides of the credit card market; the court also concluded that the anti-steering provisions did not have an anticompetitive effect in the credit card market as a whole. The Supreme Court affirmed, holding that plaintiffs failed to carry their burden of proving that the anti-steering provisions had an anticompetitive effect in the relevant market, which must be defined to include both merchants and cardholders.

The Supreme Court’s relevant market definition rested upon the finding that credit card companies are two-sided transaction platforms which “facilitate a single, simultaneous transaction between participants.” Because transaction platforms like credit card companies “cannot make a sale unless both sides of the platform simultaneously agree to use their services,” they “exhibit more pronounced indirect network effects and interconnected pricing and demand” and are “better understood as ‘suppl[ying] only one product’—transactions.” Citing economic literature on two-sided markets, the Court explained that when indirect network effects are strong—that is, the value of the platform on one side depends on the level of participation on the other side—a two-sided platform must consider those effects prior to changing price on either side of the platform. Consequently, “courts must include both sides of the platform—merchants and cardholders—when defining the credit-card market.”

The Court concluded that the direct evidence of increased merchant fees offered by plaintiffs was insufficient to prove anticompetitive effects in the relevant market. Focusing on the merchants alone ignored the benefits to cardholders that are funded by merchant fees. The Court deemed focusing on merchant fees alone the fatal flaw in plaintiffs’ case, since doing so “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.” The critical metric was output—did the arrangements increase or decrease the number of transactions? Measuring output was a way to capture the interests of both sides.

The Sabre/Farelogix Decision

Nearly two years later, the district court in Sabre/Farelogix issued the first decision applying Amex in the context of a merger challenge. The DOJ sued to enjoin Sabre Corporation’s acquisition of Farelogix, Inc. Sabre is a two-sided global distribution system (GDS) platform that connects airlines and other travel suppliers to travel agencies, allowing travel suppliers to distribute their content more broadly and travel agencies to create and manage bookings for their customers. Farelogix supplies information technology solutions to airlines, including its “core” product FLX OC, which allows airlines to distribute their content directly to travel consumers as well as through travel agencies. Airlines utilizing FLX OC have the ability to distribute their content directly to travel agencies without using a GDS as an intermediary (referred to as “direct connect” or “GDS bypass”), or to distribute content through a GDS (referred to as “GDS passthrough”).

Farelogix’s FLX OC is a new distribution capability application programming interface (NDC API) solution that “enables airlines to communicate offers and orders between the airline’s [passenger services system] and third parties.” NDC APIs such as FLX OC

shift[] two key functions away from the GDS. First, unlike legacy GDS technology, NDC allows an airline to create its own offer instead of relying on the GDS for offer creation. . . . Second, NDC enables airlines to decouple order management from the GDS bundle, allowing airlines to reduce their distribution costs.

Both Sabre’s and Farelogix’s offerings allow airlines to distribute content to travel agencies. The primary difference in the parties’ business models is that while Sabre’s GDS operates as a two-sided platform—in that travel agencies which contract with Sabre are provided access to offers from airlines which contract with Sabre to distribute their content—Farelogix only contracts with airlines and provides a product which allows the airlines to distribute content directly to travel agencies without a GDS or to distribute more complex and personalized offers through a GDS. Farelogix does things differently than Sabre, but its product offering constrains or at least potentially constrains Sabre’s market power.

Citing the parties’ ordinary course documents, as well as ordinary course documents and testimony from airline executives of three of the top U.S. airlines, the court found that Sabre and Farelogix consider each other competitors and that some of the services they offer airlines overlap with each other:

  • “Sabre considers Farelogix a competitor in developing NDC technology for direct connects.”
  • “Farelogix identified Sabre as a ‘key competitor’ in order delivery and offer management.”
  • “Sabre and Farelogix also competed to provide an NDC direct connect platform to [Redacted]. Sabre viewed Farelogix as its ‘main competitor’ for the [Redacted] opportunity.”
  • “AA has described Farelogix’s direct connect technology as ‘providing a low cost substitute for GDSs.’” “AA’s VP of sales and distribution strategy has described Farelogix as ‘the GDSs’ leading competitor/agitator for years.’”
  • “[A] United executive testified that Farelogix offers United its only alternative way to reach U.S. travel agencies, other than going through a GDS.” “In 2017, United’s director of ­distribution advised United’s leadership that partnering with Farelogix for NDC distribution would ‘improve[ ] United’s position’ in upcoming contract negotiations with Sabre, Amadeus, and Travelport. He added that, even in the most recent GDS negotiations, ‘[t]he existence of the direct connect alternative powered by Farelogix gave me the ability to command a lower price from the GDSs’ and otherwise obtain ‘better terms.’” “United’s director of distribution wrote to colleagues that ‘[i]f a GDS owns Farelogix, they may . . . remove a major threat that is out there in the industry that helps apply pressure to GDSs when we negotiate. Without that alternative in the market, we lose leverage.’ He described the acquisition as the ‘stuff of nightmares’ and ‘the worst case scenario coming true.’”
  • “A Delta executive testified that Farelogix provides airlines with an alternative to the GDS for distributing content to travel agencies.” “Delta’s managing director of distribution strategy testified that having Farelogix as a GDS alternative improved Delta’s bargaining position with the GDSs.”

The court noted four differences between Sabre’s and Farelogix’s offerings: (1) Sabre’s GDS provides services to both airlines and travel agencies, while FLX OC is “only an input for airlines”; (2) Sabre’s GDS creates and delivers offers, while FLX OC only delivers offers; (3) Farelogix’s prices are significantly lower than Sabre’s, “[i]n part because Farelogix provides fewer services to airlines”; and (4) airlines pay Sabre on a per-segment basis for flights booked through its GDS, while they pay a subscription fee to Farelogix in addition to fees on a per-ticket basis for flights booked through FLX OC.

The DOJ alleged that the proposed acquisition would substantially lessen competition in the one-sided “booking services” market. Relying entirely upon Amex and a Second Circuit holding that Sabre is a two-sided transaction platform in a case alleging Sabre had engaged in unlawful vertical and horizontal restraints in violation of Section 1 and Section 2 of the Sherman Act, the court held that the DOJ could not prevail on its claim because “[a]s a matter of antitrust law, Sabre, a two-sided transaction platform, only competes with other two-sided platforms[.]” The court rejected the DOJ’s assertion that “‘a firm can compete in more than one market [a]nd the booking services portion of the Sabre GDS competes in a one-sided booking service market’” on the grounds that “Amex establishes that two-sided transaction platforms such as the Sabre GDS supply ‘only one product:’ the transactions that link both sides of the market.”

Analyzing the Sabre/Farelogix Court’s Application of Amex

The Sabre/Farelogix court’s application of Amex focused entirely on the truism identified in Amex that “[o]nly other two-sided platforms can compete with a two-sided platform for transactions” as a blanket prohibition on conducting the necessary factual inquiry required to assess the relevant market proposed by the DOJ. But two-sided transactions were not the issue; the parties did not compete for transactions. Moreover, Sabre could not and did not explain how harm to airlines would be offset or overcome by benefits to travel agencies. Judge Stark evidently believed that the Amex dictum precluded recognition of the competition he had found between the two simply because Sabre competed on both sides while Farelogix did not. But nothing in the Amex decision called for any such result.

In Amex, the Court carefully considered the facts of the credit card industry to determine the appropriate market in which to consider the alleged harm of the vertical restraints at issue. Because “the relevant market is defined as ‘the area of effective competition,’” the Court needed to first identify the product or service Amex supplied and who its competitors were for that product or service. As the Court astutely pointed out, “A credit-card company that processed transactions for merchants, but that had no cardholders willing to use its card, could not compete with Amex.” The Court consequently concluded that “competition [could not] be accurately assessed by looking at only one side of the platform in isolation.”

In Sabre/Farelogix, the court detailed the ways in which Sabre and Farelogix compete for opportunities to distribute content for airlines, only to turn a blind eye to that evidence and conclude that the parties do not compete in a relevant market as a matter of antitrust law “[d]ue to the combination of the Supreme Court’s 2018 Amex decision, holding that ‘[o]nly other two-sided platforms can compete with a two-sided platform for transactions,’ and the Second Circuit’s 2019 finding that the Sabre GDS is a two-sided transaction platform.” This application of Amex, where the DOJ alleged and put forth evidence of a relevant market consisting of only one side of Sabre’s business, is inconsistent with well-settled merger precedents.

Courts, including the Supreme Court, have long understood the exercise of defining the relevant market in the context of merger analysis to be geared toward identifying the segments of the merging parties’ businesses for which demand-side substitutability exists. Determining that the analysis for mergers involving two-sided platforms must now be broadened to include segments where demand-side substitution does not occur––in instances where the plaintiff is alleging harm on only one side of the market (where demand-side substitution does occur)––would be a substantial departure from longstanding practice.

Section 7 of the Clayton Act prohibits transactions that may substantially lessen competition in any line of commerce. Plaintiffs seeking to challenge a transaction bear the burden of defining and proving a relevant market in which competition will be harmed. Market definition is a fact-intensive inquiry that may well vary among cases involving the same industry or even the same company where the facts or challenged conduct differ. Consider two-sided transaction platform Uber: the relevant market for an exclusionary conduct case brought by a ride-sharing competitor against Uber for engaging in predatory pricing could differ from the relevant market(s) in which the DOJ analyzed Uber’s acquisition of Postmates, which competed with Uber for food-delivery services. Likewise, the relevant market(s) in which the DOJ analyzed Uber’s acquisition of Postmates appears to have differed from the relevant market in which the FTC analyzed Aurora’s acquisition of Uber’s automated driving division.

The Supreme Court’s holding in Amex that two-sided transaction platforms only face competition from other two-sided platforms for transactions is a truism that does not preclude a finding that a two-sided platform faces meaningful competition on only one side of the platform in at least some respects. Evans and Schmalensee acknowledge this in their 2008 article Markets with Two-Sided Platforms (which the Amex majority opinion cites 12 times). Evans and Schmalensee explain that the fact that a business is two-sided may be irrelevant when “nothing in the analysis of the practices really hinges on the linkages between the demands of the participating groups.” Such a scenario is certainly conceivable in the context of a merger review—and particularly one involving one company that is a two-sided platform and one that is not. In that context, a plaintiff may allege and ultimately prove that multiple relevant markets exist. Beyond a market for transactions facilitated by the two-sided platform, a market may also exist for a product or service sold on one side of the market. Such a finding would not contradict Amex because the parties did not compete against each other for transactions and because there was no suggestion that increased prices to airlines would be mitigated by lower prices or better terms to agencies. Two-sidedness is most appropriately recognized in market definition where the alleged harm to one side can provide benefits to the other. When that is not the case, focusing only on the side where the harm has occurred makes sense.

The district court in Sabre/Farelogix erred in our view by concluding that the DOJ did not define a proper relevant market as a matter of antitrust law. Merger precedents require examining the factual record to determine whether the plaintiff has sufficiently defined the relevant market. The court purported to conduct this analysis in the alternative, ultimately rejecting the DOJ’s proposed relevant market as a factual matter. However, the discussion appears to continue the improper ­application of Amex discussed above by concluding that the “DOJ improperly excludes the services Sabre’s GDS provides to travel agencies, inconsistent with Amex.” The court further criticizes the DOJ for “contend[ing] that a single technical functionality of Sabre’s unitary GDS transaction platform, which is sold in a two-sided market, can be extracted from the GDS and placed in competition against a slice of FLX OC, which is sold in a different, one-sided market’” and for urging the court to “evaluate the one ‘submarket’ of ‘booking services.’”

The court’s determination that the “DOJ failed to show that the Sabre GDS ‘booking services’ is a distinct product that generates separate demand” and thus its proposed product market “is at odds with the ‘commercial realities of the industry’” misses the mark. Times-Picayune makes clear that the relevant inquiry is whether the booking services provided by Farelogix to airlines are a reasonable substitute to the booking services Sabre provides to airlines. A cursory review of the findings of fact in the district court decision suggests that at least the merging parties and their airline customers believe the answer to be yes. Indeed, the court noted in the conclusion of the decision that it was more persuaded by the DOJ than by the defendants on the questions that get to the heart of market definition—namely, “whether Sabre and Farelogix compete, whether Sabre understands GDS bypass is a threat, whether Sabre stands to lose revenue even from the expansion of GDS passthrough, and Sabre’s motivation for its proposed acquisition of Farelogix.” The court concluded, however, that the DOJ’s failure to identify harm in a relevant product market was dispositive.


Courts, antitrust enforcers, and practitioners have long understood the practice of defining the relevant market in merger cases to be a question of fact. The Sabre/Farelogix court’s application of Amex was inappropriate given the facts at hand. The finding that two-sided platforms only compete with other two-sided platforms for transactions was based on factual analysis of the credit card industry. Extending that finding as a broad principle of antitrust law that requires ignoring evidence of competition between merging parties where only one party is a two-sided platform misses the forest for the trees—and is entirely inconsistent with precedents like Times-Picayune, which hold that market definition depends on the nature of the particular claim at issue. The Amex court determined that both sides of the platform must be analyzed in that case because doing so was “necessary to accurately assess competition.” In Sabre/Farelogix, on the other hand, the DOJ presented extensive evidence about the competition that would be lost should the merger be permitted; the defendants made no showing of how that harm would be offset by procompetitive effects that would flow to travel agencies.

Unlike the credit card industry, where the product companies are supplying is transactions, Sabre’s GDS platform is better understood as supplying different services on each side of the platform—offer creation on the airline side, and normalization and aggregation services on the travel agency side. The fact that Farelogix does not compete on the travel agency side of the platform does not preclude it from acting as a competitor on the airline side. Even the parties’ economic expert testified that “‘GDSs can face competition from one-sided competitors.’” Indeed, the record clearly demonstrated that Sabre, Farelogix, and their airline customers all agreed that Sabre and Farelogix compete. The court erred by turning a blind eye to that evidence in the name of following precedents based on an entirely different industry and set of facts that were inconsistent with the case at hand.

Factually concluding that the parties compete and then holding that they don’t because of ambiguous language in another case undermines the entire purpose of the antitrust laws. Courts should carefully consider the facts of the industry before them when examining a proposed merger to determine the relevant market and whether Amex’s two-sided market paradigm applies, lest they wish to reach an outcome which the Sabre/Farelogix court aptly described as “somewhat odd.”

An earlier version of this article was presented by Mr. Jacobson at the ABA Antitrust Law Section’s 2021 Spring Meeting. The authors’ firm represents a number of companies operating in markets that could be considered two-sided, and Ms. Edwards represented a non-party in the Sabre/Farelogix case.