The Impact of Amex and Its Progeny on Technology Platforms

Kacyn H. Fujii* February, 2022

Big Tech today faces unprecedented levels of antitrust scrutiny. Yet antitrust enforcement against Big Tech still faces a major obstacle: the Supreme Court’s 2018 decision in Ohio v. American Express. Popularly called Amex, the case imposed a higher initial burden on antitrust plaintiffs in cases involving two-sided markets. Two-sided markets connect two distinct, noncompeting groups of customers on a shared platform. These platforms have indirect network effects, meaning that one group of customers benefits when more of the second group of customers joins the platform. Two-sided markets are ubiquitous in the technology sector, encompassing social media, search engines, and online marketplaces.

Many have observed that the Amex Court’s reasoning drew on questionable economic principles, contrary to the typical approach in antitrust law. This Note examines and adds to these critiques through a novel analysis of lower-court cases post-Amex. This analysis reveals that Amex has resulted in inconsistencies and confusion in the lower courts, opening the door for technology defendants to manipulate Amex’s definition of two-sided markets for their own benefit. To resolve these inconsistencies, this Note proposes a two-part legislative solution to curb Amex’s reach.

Introduction

After years of growing concern about Big Tech’s influence over our markets, data, and society more generally, in July 2020 the House Judiciary Committee’s antitrust subcommittee held an unprecedented hearing that brought the leaders of Amazon, Facebook, Apple, and Google to Washington.¹ At the hearing, members on both sides of the aisle expressed concern over Big Tech’s power, especially during the COVID-19 pandemic.² By September, the Federal Trade Commission (FTC) had begun investigating Amazon and gearing up to file a possible antitrust suit against Facebook.³ In October, the Senate Commerce Committee unanimously voted to subpoena the CEOs of Facebook, Google, and Twitter.⁴ Only days later, the House Judiciary antitrust subcommittee released a landmark report that concluded a sixteen-month investigation into Amazon, Facebook, Google, and Apple.⁵ This report broke new ground by declaring that Google and Facebook had monopoly power and by indicating Congress’s support for major antitrust legislation for the first time in decades.⁶ The report also signaled that Congress would support stronger antitrust enforcement by federal and state enforcers.⁷ The Department of Justice, the FTC, and multiple state attorneys general subsequently brought lawsuits against Google and Facebook for anticompetitive behavior.⁸ Amid these lawsuits, Congress has continued to press forward with legislation that would make sweeping changes to antitrust law, such as a bill proposed by Senator Klobuchar in February 2021.⁹

Technology companies¹⁰ have never faced this level of scrutiny, but existing antitrust law has limited power over Big Tech companies. Courts interpret the Sherman Act using the consumer welfare standard, which relies on economic measures like price to determine violations of antitrust law.¹¹ Typically, monopolies raise prices, which directly harms consumers. Technology companies are unique in that they offer their products for free or for very low prices making it difficult to demonstrate harm to consumers.¹²

A 2018 Supreme Court opinion presents another major obstacle to enforcing antitrust laws against technology companies.¹³ Ohio v. American Express Company (Amex) created a framework for regulating two-sided markets (interchangeably referred to as two-sided platforms) under antitrust law.¹⁴ In a two-sided market, two distinct customer groups create benefits for each other through their shared interest in a particular product or service. For example, platforms like Uber rely on demand from each side of the market—drivers and riders—to succeed. Two-sided platforms are ubiquitous in the technology sector and the economy in general; prominent examples include credit- and payment-card systems, search engines, online marketplaces, social media, newspapers, airline- and restaurant-reservation systems, and ridesharing services.¹⁵ Under Amex, antitrust plaintiffs bear the additional burden of showing net anticompetitive harm on both sides of a two-sided market.16F16F¹⁶ In addition, Amex’s broad definition of two-sided markets may shield many Big Tech companies from antitrust liability.¹⁷

Many scholars have speculated about how Amex will affect Big Tech and antitrust law in general.F¹⁸ While scholars have critiqued Amex and analyzed some post-Amex cases,¹⁹ this Note is the first to examine Amex’s impact on Big Tech by pulling together several recent lower-court decisions. It also offers a novel legislative compromise that would limit Amex’s reach without overturning it outright. In doing so, this Note provides a politically divided Congress with another option to consider as they debate changing the antitrust laws. Part I overviews the evolution of antitrust law and explains how Amex has modified antitrust analysis for two-sided markets. Part II analyzes recent lower-court cases to illustrate Amex’s inconsistencies and demonstrate how Amex makes it even harder to rein in Big Tech. Part III responds by proposing legislation to limit Amex’s holding to a narrow set of two-sided markets that can be adjusted to meet new antitrust challenges.

I. Amex and Two-Sided Markets

Given its potential impact on the technology sector, Amex has been called the most consequential antitrust decision of the decade.²⁰ This Part presents background on the purpose and development of antitrust law prior to the Amex decision. Section I.A describes the evolution of antitrust law. Section I.B introduces current modes of antitrust analysis. Section I.C explains how Amex modified antitrust analysis for two-sided markets.

A. The Evolution of Antitrust Law

Antitrust law is governed primarily by the Sherman Act, which was passed in 1890 to promote fair competition in the economy.²¹ The Sherman Act’s broad language allowed courts to play a large role in its interpretation, effectively making it a common law statute.²² Thus the Sherman Act’s application has varied in accordance with changing public values and goals surrounding antitrust enforcement since its enactment.

The framers of the Sherman Act were concerned with a small number of firms having too much power.F²³ They also wanted to make sure that firms played fairly so that small businesses would have the chance to compete.²⁴To achieve these goals, courts first relied on structuralism, the idea that certain market structures can impede competition.²⁵ For example, when there are fewer firms in a market, it is easier for them to collude and engage in oligopolistic behavior like price fixing.²⁶ Until the 1960s, courts blocked mergers that they determined would result in too much market concentration.²⁷

In the 1970s, however, the Supreme Court replaced structuralism with the “consumer welfare” standard propounded by the Chicago School of economics and antitrust.²⁸ The Chicago School defines consumer welfare as allocative efficiency across both consumers and producers.²⁹ Under this standard, behavior that results in efficiency gains to either consumers or producers should be upheld under antitrust law.³⁰ This is based on the belief that firms will try to maximize profits and efficiency, which ultimately helps consumers through lower prices and better products.³¹ In other words, even if there is no direct benefit to consumers, the Chicago School subscribes to the view that greater efficiencies for companies will eventually help consumers.

The consumer welfare standard focuses on empirical harm to the market as measured by economic indicators, especially price.32³² Given this focus on economics, plaintiffs must also show actual anticompetitive harm instead of merely showing that the market structure typically leads to anticompetitive behavior.³³ Under the Chicago School standard, it is not enough to show that Google, for example, has monopoly power—plaintiffs must demonstrate that Google’s specific behavior resulted in anticompetitive harm to the market. Moreover, vertical agreements (agreements between firms at different levels of the production chain, like manufacturers and retailers) receive less scrutiny than horizontal agreements (agreements between competitors) under the Chicago School approach.34³⁴ This shift from structuralism to the Chicago School approach has thus led to lax antitrust enforcement and a greater reliance on economics in antitrust law.³⁵

B. Antitrust Modes of Analysis

The first section of the Sherman Act reads: “Every contract, combination in the form of trust or otherwise, or conspiracy, in restraint of trade or commerce among the several States, or with foreign nations, is declared to be illegal.”³⁶ Initially, the Supreme Court read this to prohibit all restraints of trade, defined as any activity that limited competition. However, all kinds of activities limit competition, such as agreements between small businesses to not compete with each other or to share information on employee salaries. Only one year after United States v. Trans-Missouri Freight Association, the Court began to recognize that some restraints of trade were reasonable,³⁷ giving rise to the “rule of reason.”³⁸

Yet the Court maintained that some restraints were so nakedly anticompetitive that a reasonableness inquiry would be unnecessary.³⁹ In United States v. Socony-Vacuum Co., the Court said, “a combination formed for the purpose and with the effect of raising, depressing, [or] fixing . . . price[s] . . . is illegal per se.”40⁴⁰ This language marked the beginning of “per se” antitrust analysis.

Today, courts still apply either per se or rule of reason analysis to determine whether the Sherman Act has been violated.⁴¹ In most cases, courts apply rule of reason analysis, in which courts must assess the economic impact of the allegedly anticompetitive behavior.⁴² First, courts must define the “relevant market,” which provides the frame of reference for judging anticompetitive harm.⁴³ Market definition, which requires parties and courts to conduct fact-intensive economic analysis, often has an outsized impact on antitrust analysis.⁴⁴ For example, even if Zingerman’s Deli has a large market share among delis in Ann Arbor, Michigan, it would not look like a monopolist if the relevant market is defined as every restaurant in the United States. Courts may only assess anticompetitive harm to competitors within that relevant market; indeed, it would make little sense to treat a shave ice shop in Hawaiʻi and a deli in Michigan as competitors.

Once the relevant market is defined, rule of reason analysis involves a “three-step, burden-shifting framework.”⁴⁵ First, the plaintiff carries the burden of showing that the restraint has a “substantial anticompetitive effect that harms consumers in the relevant market.”⁴⁶ If the plaintiff succeeds, the defendant then bears the burden of demonstrating that the restraint has a procompetitive effect, meaning any favorable effect on the market.⁴⁷ If the defendant meets this burden, the plaintiff must prove either that the anticompetitive effects outweigh the procompetitive effects or that the procompetitive effects could be “reasonably achieved” in a less anticompetitive manner.⁴⁸

In contrast, per se analysis avoids market definition or fact-intensive inquiries into economic harm.⁴⁹ Because of the Chicago School’s influence, per se analysis is relegated to a narrow category of cases where the restraint is “manifestly anticompetitive,”⁵⁰ as with price-fixing.⁵¹ For example, an agreement by San Francisco restaurants to fix prices of avocado toast at the same rate would be per se illegal, even without an inquiry into the relevant market.

C. The Amex Formulation

Within rule of reason analysis, Amex carved out a separate rule for two-sided platforms, which are “business[es] that depend[] on relationships between two different, noncompeting groups of transaction partners.”⁵² American Express (Amex) is a two-sided market because it operates a credit card transaction platform that connects merchants and customers. In Amex, the DOJ and several state attorneys general alleged that the “no-steering” clauses in Amex’s contracts with its partner merchants were anticompetitive.⁵³ These clauses said that merchants could not “steer,” or encourage, customers to use other credit cards despite Amex’s higher merchant fees.⁵⁴ The plaintiffs claimed that these provisions harmed competition since they restricted what merchants could do to offset or avoid Amex’s higher fees.⁵⁵ Since merchants were not allowed to give discounts for competitor credit cards such as Visa or Discover, customers had less of a price incentive to select these cards over Amex. Amex countered that its no-steering provisions and higher merchant fees were procompetitive because they allowed Amex to give large rewards to its cardholders on the other side of the two-sided market.56F56F⁵⁶

Though the district court found that the no-steering provisions resulted in anticompetitive harm,⁵⁷ the Second Circuit reversed, claiming that the plaintiffs did not adequately prove anticompetitive harm.⁵⁸ The Supreme Court agreed with the Second Circuit, holding that the plaintiffs did not meet their burden of showing that Amex’s higher merchant fees and no-steering clause had anticompetitive effects.⁵⁹

Writing for the majority, Justice Thomas explained that two-sided markets must be treated differently than other markets.⁶⁰ Two-sided markets, which bring together two different customer groups, are characterized by indirect network effects between the two sides of the platform.⁶¹ This means that each group benefits when a customer joins on the other side.62⁶² Credit card networks, for example, connect merchants and customers and are more valuable to merchants when more customers join the network.⁶³ Likewise, customers benefit when more merchants accept their credit card.⁶⁴ The majority held that because both sides of the platform are interdependent, courts need to consider both sides of a two-sided market when conducting rule of reason analysis.⁶⁵ In other words, the plaintiffs had to show that both cardholders and merchants were harmed by the no-steering provisions.

Next, the majority laid out some exceptions to this rule for two-sided markets. It noted that indirect network effects might be so negligible that only one side must be taken into account in rule of reason analysis.⁶⁶ Newspapers, which connect readers and advertisers, are the quintessential example of a platform with weak indirect network effects.⁶⁷ Although advertisers benefit from more people reading the newspaper, readers are indifferent as to how many advertisers the newspaper has.⁶⁸ Because of weak indirect network effects, cases like this require analysis of only one side of the market.

The Court further limited the two-sided market rule to platforms that “facilitate a single, simultaneous transaction between participants.”⁶⁹ It viewed the credit card market as a single two-sided market “‘suppl[ying] only one product’—transactions.”⁷⁰ For two-sided transaction markets like credit card networks, courts can consider the net effect of the restraint on this single market of transactions.⁷¹ Since the Amex plaintiffs did not prove net harm on both sides of the market, they failed to demonstrate anticompetitive effects.⁷²

In sum, Amex instructs courts that they must take both sides into account when the two-sided market involves simultaneous transactions, has non-negligible indirect network effects, and can be viewed as a single market dealing in transactions.

II. Problems with Amex

Amex has sparked vigorous debate among antitrust and economics scholars. Some praised the decision for taking contemporary economic theory of two-sided markets into account,⁷³ while others said it “devastates antitrust law.”⁷⁴ This Part outlines two major categories of concerns with the Amex decision. Section II.A discusses problems with how Amex modified antitrust analysis. Section II.B reveals problems with Amex’s definition of two-sided markets and, by extension, its applicability beyond payment platforms.

A. Amex’s Application to Antitrust Analysis

Amex requires collapsing two-sided markets into a single market and proving net anticompetitive harm by taking into account both sides of the market. For example, if Amex’s no-steering provision benefited customers more than it harmed merchants, then the provision would not be anticompetitive under Amex. This requirement, which increases the plaintiff’s burden in the first step of rule of reason analysis, has been controversial among scholars. This Section explains why collapsing the market and looking at the net effect creates problems. First, there is no consensus in the economic literature about the Court’s approach, and antitrust law pre-Amex was already equipped to handle two-sided markets. Second, the burden Amex places on plaintiffs is too high and is inconsistent with the goals of antitrust law. Third, United States v. Sabre’s holding on competition between two-sided and one-sided markets could further stifle competition if adopted by other courts.⁷⁵

1. Net-Effects Analysis Versus Separate-Effects Analysis

Modern antitrust law relies heavily on economics.⁷⁶ Yet Amex did not rely on settled economic principles. In fact, there is vigorous debate in the economic literature about how to analyze two-sided markets.⁷⁷ Currently, the two leading approaches outlined in the literature are net-effects analysis and separate-effects analysis.⁷⁸

Amex adopted net-effects analysis,⁷⁹ an approach advanced by Lapo Filistrucchi.⁸⁰ Under this approach, the consumer welfare standard treats consumers on both sides of the market equally.⁸¹ In Amex, this meant that the harm to merchants could be directly balanced against the benefit to cardholders.⁸² Thus, even if Amex’s behavior resulted in cognizable anticompetitive harm to merchants, there was no violation of the antitrust laws so long as there was an offsetting benefit to cardholders. Put another way, anticompetitive harm on one side of the platform would not be actionable if it were outweighed by the benefits to the other side.⁸³ Since net-effects analysis involves directly weighing harms and benefits to each customer group, this approach assumes that each customer group has equal importance when assessing anticompetitive harms.⁸⁴

By contrast, other scholars advocate for separate-effects analysis, which requires that each side of the platform be considered separately.85⁸⁵ Under separate-effects analysis, two-sided platforms are treated as multiple separate but interrelated markets and cannot be collapsed.⁸⁶ If applied to Amex, it would mean that plaintiffs would only need to show harm to either merchants or cardholders. In other words, separate-effects analysis allows each group of consumers on either side of the platform to benefit from competition under antitrust law.⁸⁷

Separate-effects analysis avoids the assumptions upon which net-effects analysis relies, namely, that each side of the market has equal weight and can be directly compared. Although these assumptions may hold for platforms with strong indirect network effects like payment systems, they should not be applied indiscriminately to two-sided markets for which these assumptions may not make sense. Collapsing both sides of the platform into one “obscure[s] the underlying economic forces” that drive this relationship⁸⁸ and renders “any coherent economic analysis of the relevant market impossible.”⁸⁹

Advocates for the net-effects approach claim that traditional rule of reason analysis, which uses the separate-effects approach,⁹⁰ cannot adequately account for anticompetitive harm since it does not explicitly take both sides of the platform into account.⁹¹ Yet traditional rule of reason analysis would have sufficiently captured anticompetitive harm even on the facts of Amex. Justice Breyer’s Amex dissent pointed to the harms that Discover suffered as a result of Amex’s no-steering provisions.⁹² Discover had tried to lower its merchant fees to encourage adoption, but Amex’s no-steering provisions prevented merchants from encouraging customers to use Discover over Amex.⁹³ Since there was no price difference between the two credit cards from the customer’s perspective, customers had little incentive to switch to Discover. When Discover’s lower merchant fees did not result in greater adoption, Discover abandoned this strategy and raised its merchant fees to match the other credit card companies.⁹⁴ As a result, merchants uniformly had to pay higher fees.⁹⁵ Higher merchant fees hurt the merchant directly unless the merchant decided to pass on these costs to its customers. Given that merchants and consumers could be directly harmed through higher prices, the harm from Amex’s no-steering provision would have been captured under traditional rule of reason analysis.

That wasn’t the only direct evidence of anticompetitive harm in Amex. Amex had also increased its merchant fees on twenty different occasions within five years without increasing its cardholder rewards.⁹⁶ Notably, Amex did not lose any major merchants after these fee increases.⁹⁷ Amex’s ability to keep these merchants after numerous fee increases shows that Amex’s no-steering provision allowed it to exercise market power in a way that resulted in market inefficiencies.⁹⁸ This direct evidence of anticompetitive harm would have been captured by the traditional rule of reason approach. Thus, Amex itself demonstrates that traditional rule of reason analysis, which relies on the separate-effects approach, can adequately capture anticompetitive behavior in two-sided markets.⁹⁹

Amex’s adoption of the net-effects approach also creates problems in the technology context. Many Big Tech companies have both one-sided and two-sided markets, which makes analyzing the market as one unit a challenging undertaking. Even if the assumptions of net-effects analysis apply to payment systems, extending this framework to complex technology platforms goes too far.

Amazon is one example of a company with both two-sided and one-sided platforms. Amazon Marketplace is a two-sided platform because it facilitates transactions between third-party sellers and customers; by contrast, Amazon’s business of selling directly to customers is one-sided. Interestingly, Amazon competes directly with third-party sellers as a participant on its own platform.¹⁰⁰ Collapsing this system into a single market and calculating the net anticompetitive harm ignores the company’s dual role as both platform and participant.¹⁰¹ Moreover, Amazon’s participation in its own marketplace may result in weaker indirect network effects for the third-party two-sided market.¹⁰² Since Amazon can lower its prices to undercut its competition, competing third-party sellers are less likely to see the benefits of more buyers joining the platform.¹⁰³ This interaction highlights the importance of capturing Amazon’s role as a participant in Amazon Marketplace.¹⁰⁴ Reducing this complicated relationship into a single market would fail to capture all of the important market forces at play.¹⁰⁵

Amazon is not an anomaly. Other technology companies follow Amazon’s model of participating in their own platforms. For example, Google promotes its own products over others in its search engine. If the user searches for a flight, Google can make Google Flights appear at the top of the list of search results,¹⁰⁶ while competitors like Expedia are demoted within the results.¹⁰⁷ Likewise, the Google search engine promotes its own restaurant listings over Yelp listings108¹⁰⁸ and places Google ads at the top of the results page.¹⁰⁹ Similar to the Amazon example, collapsing the Google ecosystem into a single market would ignore the anticompetitive effect it exerts as a participant on its own platform. Thus, even if net-effects analysis is correctly applied to payment systems, it should not apply to complex technology platforms.

2. Plaintiff’s Burden

Amex places too high a burden on plaintiffs.¹¹⁰In traditional rule of reason analysis, anticompetitive and procompetitive effects are weighed only after both parties have a chance to make their case.¹¹¹ Amex changed that for two-sided markets.¹¹² Under Amex, plaintiffs challenging two-sided markets must show net anticompetitive harm on both sides of the platform at the first step of the rule of reason analysis.¹¹³ This raises a few concerns.

First, there are practical challenges that make it difficult for plaintiffs to prove net anticompetitive harm on both sides of a two-sided platform.¹¹⁴ Generally, defendants, not plaintiffs, have the best understanding of their own platforms and pricing structures, as well as of how consumers on either side of the platform interact.¹¹⁵ Thus, Amex’s burden imposes a higher cost on plaintiffs than it does on defendants. This cost is even greater for complicated technology platforms that require specialized knowledge to understand their operations. In contrast, traditional rule of reason analysis requires that defendants bear the burden of proving procompetitive justifications because they best understand how the challenged conduct results in efficiencies.¹¹⁶ The analysis for two-sided platforms should follow a similar principle by placing the burden on defendants since they typically have the lowest cost of producing evidence.117F117F¹¹⁷

Second, weighing the anticompetitive harm to each side should not take place in the first step of rule of reason analysis. One of the goals of antitrust law is to promote competition.¹¹⁸ Each side of the platform should have the opportunity to receive protection from anticompetitive harm.¹¹⁹ Typically, procompetitive and anticompetitive effects are weighed in the final step of rule of reason analysis.¹²⁰ Under Amex, however, the burden of weighing anticompetitive effects for two-sided platforms is instead located at the first step.¹²¹ Since anticompetitive harm to one side could be outweighed up front by the procompetitive benefits to the other side, the higher initial burden makes it harder to protect each side from anticompetitive harm.

This high upfront burden on plaintiffs provides an incentive for defendants to claim their platforms are two-sided markets.¹²² Although they have been largely unsuccessful thus far, defendants in lower courts have already begun to claim that they are two-sided or multi-sided markets despite not meeting Amex’s definition.¹²³ For example, in In re National Collegiate Athletic Ass’n Athletic Grant-in-Aid Cap Litigation, the NCAA’s expert witness claimed that universities are multi-sided platforms because of their many constituencies, including student-athletes, alumni, coaches and staff, the institution, and the community.4F¹²⁴ The court rightfully rejected this claim because the NCAA failed to describe the product, price, or economic interactions between these different constituencies.¹²⁵ Similarly, the In re Delta Dental Antitrust Litigation defendant failed to demonstrate that the dental-insurance market was two-sided.¹²⁶

Finally, this burden on plaintiffs is even more concerning in the technology context since it is difficult to show net anticompetitive harm. Many technology companies like Google and Facebook offer their products to users for free. They would argue that any anticompetitive behavior on the seller side of the market is offset by these low prices. This would give these companies an extra defense.

3. Competition Between Two-Sided and One-Sided Markets

United States v. Sabre Corp. considered whether it is possible to define the relevant market as including both two-sided and one-sided competitors under Amex. Even though the case was vacated by the Third Circuit and is therefore not precedential, it still provides a useful roadmap for technology defendants in other circuits.¹²⁷

Sabre involved a merger between two companies, Sabre and Farelogix.¹²⁸ Sabre makes a global distribution system that connects travel agents and airlines.¹²⁹ Farelogix provides software systems only to airlines.¹³⁰ Despite recognizing that “Sabre [a]nd Farelogix [v]iew [e]ach [o]ther [a]s [c]ompetitors” and that “[t]he record reflects competition between” Sabre and Farelogix in software for airlines, the district court held that the two companies could not compete in the same relevant market.¹³¹ This was because Sabre was two-sided while Farelogix was not.¹³² To support its holding, the district court cited dicta from Amex, which noted that “[o]nly other two-sided platforms can compete with a two-sided platform for transactions.”¹³³

This decision marked the first time that Amex was applied in a horizontal merger context.134¹³⁴ This is notable because Amex was previously thought to only apply to vertical restraints under Section 1 of the Sherman Act.135F¹³⁵ Even more concerning is the court’s holding that two-sided platforms cannot compete with one-sided platforms in the same relevant market. This could lead to economically confusing and unintuitive results. In his submission to the House Judiciary Committee’s 2020 antitrust investigation, Herbert Hovenkamp observed that “[t]housands of traditional taxicab companies and drivers who have been injured by Uber, Inc., would be surprised to hear that Uber and taxicabs cannot be competitors.”¹³⁶ The Sabre decision was also made as a matter of law, meaning that factual questions about the markets remain “outside the reach of fact finding.”¹³⁷

In addition, Sabre based its decision on misinterpreted dicta from Amex. The Amex majority’s statement that only other two-sided platforms can compete with a two-sided transaction platform was dicta because neither the case nor the parties asked the court to decide this point.¹³⁸ Further, as Andrew Ewalt points out, Amex’s statement misreads an article written by Lapo Filistrucchi and cited by the Court.¹³⁹ While Filistrucchi’s article did say that two-sided platforms should be distinguished from other two-sided platforms, it did not say “that two-sided transaction platforms only compete with other two-sided platforms.”¹⁴⁰

Soon after Sabre was decided, the DOJ filed a motion to vacate the district court’s opinion, citing concerns about the case’s effects on competition involving technology platforms.¹⁴¹ While the Third Circuit did ultimately vacate the decision,¹⁴² there is still a blueprint for defendants to follow when presenting similar arguments in the future.¹⁴³ This could have wide-reaching effects on antitrust law, especially for technology platforms. Given the prevalence of two-sided platforms in the technology sector, it would be dangerous to say that two-sided platforms cannot be in the same relevant market as one-sided businesses.

B. Amex’s Characterization of Two-Sided Markets

Just as there is a lack of consensus on how to analyze two-sided markets, there is also no consensus on how to define them.¹⁴⁴ Instead of relying on an economist’s definition, the Amex majority formulated its own.145¹⁴⁵ Recognizing that there are many platforms that exhibit indirect network effects, the Amex Court defined two-sided markets as “transaction platforms” that deal in simultaneous transactions and have non-negligible indirect network effects.¹⁴⁶ While other scholars have critiqued Amex’s definition of two-sided markets, this Section contributes novel analysis of post-Amex cases to supplement these critiques and demonstrate how lower courts’ interpretations of Amex raises enforcement concerns for Big Tech platforms.

1. Simultaneous-Transaction Requirement

Amex limited its holding to platforms where the two customer groups interact through transactions that occur at the same time for both groups.¹⁴⁷ This simultaneous-transaction requirement suffers from a number of defects. The requirement is unsupported by antitrust precedents or the economic literature.¹⁴⁸ The Court justified the addition of this requirement by explaining that two-sided transaction platforms “exhibit more pronounced indirect network effects” and have “interconnected pricing and demand.”¹⁴⁹ Yet the majority opinion failed to explain why transaction platforms tend to have stronger indirect network effects than other two-sided platforms.¹⁵⁰

Some scholars contend that the simultaneous-transaction requirement prevents Amex from being applied too broadly.¹⁵¹ They view Amex’s additional requirement as a safeguard that ensures courts will not treat companies like Google and Facebook as two-sided platforms.¹⁵² But although Amex’s holding was narrower than the Second Circuit’s, which omitted the simultaneous-transaction requirement,¹⁵³ Amex’s definition of two-sided markets is still broader than that of most economists.¹⁵⁴ In fact, Justice Breyer’s dissent argued that the majority’s formulation was overbroad.155F¹⁵⁵ The dissent contended that each element of the majority’s “two-sided transaction platform” definition is in fact commonplace.¹⁵⁶ Many businesses—including farmers markets, travel agencies, and internet retailers—connect two groups of customers to each other in simultaneous transactions and have indirect network effects.¹⁵⁷ Thus, it is more than plausible that Amex’s definition of two-sided markets is broad enough to include Big Tech platforms.

In addition to being overly broad, the simultaneous-transaction requirement is also susceptible to framing. First, the simultaneity of the transaction can easily be manipulated. For example, when someone makes a search query in Google, there are no obvious simultaneous transactions between this user and an advertiser.¹⁵⁸ If Google charged its advertisers upfront, similar to newspaper or television advertisements, this would be an accurate assessment. But if Google charged the advertiser at the moment the user clicked on the search result, then this assessment would be inaccurate because such a scheme would satisfy the simultaneity requirement.¹⁵⁹

Uber is another example of the simultaneous-transaction requirement’s susceptibility to framing. Uber used to charge customers after their ride was complete, calculating the price based on the actual time and distance traveled.¹⁶⁰ Since the customer hailed the ride and paid at different times, Uber transactions would not meet the simultaneity requirement. Later, Uber began charging its customers when they requested a driver instead of after the ride.¹⁶¹ Under Amex, this seemingly small design change classifies this transaction as simultaneous since payment is made at the same time as the ride request. Thus, companies only have to change when money is exchanged in order to satisfy simultaneity. Google and Uber illustrate the ease with which technology companies can manipulate the simultaneity of their transactions without changing the competitive effect of their actions.

The simultaneity requirement can also produce unintuitive results. For example, Amazon might seem like a paradigmatic two-sided transaction platform,¹⁶² but purchases fulfilled by Amazon itself do not meet the simultaneity requirement. This is because Amazon’s suppliers sell products to Amazon long before the user purchases it.163¹⁶³ Thus, the sale spans two distinct transactions that do not occur simultaneously. By contrast, transactions with third-party Amazon Marketplace sellers meet the simultaneity requirement since the sellers and users transact directly and simultaneously on the Amazon platform.¹⁶⁴ Although there is little difference between buying a product from a third-party Marketplace seller and buying one directly from Amazon from the shopper’s perspective, the two transactions come out differently under Amex.

Another problem with Amex’s simultaneous-transaction requirement is that it is sometimes difficult to define the relevant transaction. Returning to the Uber example, the “transaction” took place when the customer requested and simultaneously paid for the ride. But the transaction could just as easily be framed as “occurring” upon completion of the ride; since the customer paid for transportation to a particular destination, the transaction is concluded only when she receives the service she paid for. This framing affects whether the transaction is considered simultaneous. If the transaction “occurs” upon completion, Uber’s upfront pricing scheme would not be simultaneous while the post-ride pricing scheme would be. There is little guidance from Amex about what constitutes a transaction,¹⁶⁵ creating uncertainty about whether a scheme like Uber’s would qualify. This could lead to an increase of technology defendants intentionally shifting what constitutes a “transaction” in their business.¹⁶⁶

Because of Amex’s lack of guidance, lower courts have treated the simultaneous-transaction requirement inconsistently, potentially opening the door to a broader application that would encompass Big Tech platforms. In another case involving Sabre, U.S. Airways, Inc. v. Sabre Holdings Corp., the Second Circuit considered a “global distribution system” that allows travel agents to book flights for their customers.¹⁶⁷ When an agent books a flight on Sabre’s platform, Sabre collects a booking fee from the airline and gives the agent an incentive payment once the agent has met the threshold number of bookings.¹⁶⁸ Sabre argued that its platform was a two-sided transaction platform under Amex, with travel agents on one side and airlines on the other.169¹⁶⁹ The court agreed, citing all of Amex’s elements, including indirect network effects, simultaneous transactions, and a single market of transactions.¹⁷⁰

While the court correctly referenced Amex’s formal elements, this outcome is nonetheless inconsistent with Amex because it misinterprets the simultaneity requirement. When the travel agent makes the booking, the airline pays Sabre, but the travel agency only receives an incentive payment after reaching a certain booking threshold.¹⁷¹ Since there is no transfer of money between airlines and agents until the threshold is met, this scheme does not actually satisfy the simultaneity requirement. U.S. Airways’ incorrect holding exemplifies the confusion of lower courts attempting to apply Amex’s simultaneous-transaction requirement.

Other lower courts seem to ignore the simultaneous-transaction requirement altogether. In Viamedia, Inc. v. Comcast Corp., the Seventh Circuit found the “Interconnect,” a clearinghouse for television providers to pool advertising resources, to be a two-sided market because of the indirect network effects between advertisers and retail customers.¹⁷² It explained that advertisers benefit when more television providers—and thus, more retail customers—participate in the market.173¹⁷³ Significantly, it did not mention the simultaneous-transaction requirement in its discussion. Coupled with its ambiguity, the simultaneous-transaction requirement’s uneven treatment by lower courts signals that some courts will apply Amex too broadly, especially in the technology context. Lower courts’ confusion thus far means that technology defendants can capitalize on this confusion for their own benefit.

Finally, Amex’s simultaneous-transaction requirement is problematic in that it serves as a poor proxy for platforms with strong indirect network effects, like payment systems.¹⁷⁴ The Court relied on the assumption that transaction platforms were generally more likely to have strong indirect network effects,¹⁷⁵ which is not necessarily true. Although payment systems are unique for their strong indirect network effects, there are nontransaction platforms with strong indirect network effects and transaction platforms with weaker indirect network effects.¹⁷⁶ For example, Amazon Marketplace is a transaction platform with weak indirect network effects.¹⁷⁷ Amazon shoppers are unlikely to care how many sellers there are on Amazon as long as they can buy the product they seek, making the platform closer to a newspaper than a payment system under Amex.178¹⁷⁸ In sum, the simultaneous-transaction requirement is problematic for several reasons: it is overbroad, ambiguous, and a poor proxy for the strength of indirect network effects.

2. Strength of Indirect Network Effects

Amex’s second requirement—that the platform has non-negligible indirect network effects—also suffers from a lack of clarity that could extend Amex to Big Tech. Although Amex used payment systems and newspapers as examples of strong and weak indirect network effects, respectively,¹⁷⁹ the Court did not provide guidance on where to draw the line for anything in between.¹⁸⁰ Even in the economic literature, there is no bright-line rule that establishes how strong a platform’s indirect network effects must be in order to qualify as two-sided.¹⁸¹ This may be why the Court turned to a seemingly more concrete test like the simultaneous-transaction requirement. This lack of a bright-line rule opens the door for defendants to falsely claim that their platforms have strong indirect network effects and should be treated as two-sided markets.

In some cases, defendants have argued that there are strong indirect network effects because consumers benefit from lower prices. In Delta Dental, the umbrella insurance company pointed to the mutual benefits that member companies and patients enjoy as evidence of strong indirect network effects.182F182F¹⁸² It argued that member companies benefited from having more patients, and patients likewise benefited from lower premiums when more companies joined Delta Dental.¹⁸³ Similarly, in Viamedia, the court determined that the indirect network effects of the “Interconnect” were strong enough for Comcast’s advertising clearinghouse to be considered two-sided.¹⁸⁴ It reasoned that the clearinghouse’s indirect network effects were strong because advertisers benefit from having more consumers on the platform, and consumers benefit because having more advertisers allows Comcast to subsidize prices.¹⁸⁵

Such reasoning contradicts well-established knowledge about two-sidedness. Drawing the line between strong and weak indirect network effects—and, by extension, determining two-sidedness—cannot depend solely on companies’ benevolence in passing discounts on to customers. For example, as Amex itself recognized, newspapers are the quintessential example of platforms with negligible indirect network effects.¹⁸⁶ Newspapers connect advertisers with readers, and readers are generally indifferent as to how many advertisers take out ads.¹⁸⁷ Amex would have rejected the claim that a newspaper has strong indirect network effects because having more advertisers enables the newspaper to lower its prices for subscribers.188¹⁸⁸ Similar to newspapers, the customers in Delta Dental and Viamedia would likely be indifferent to the number of insurers or advertisers on the platform, even if they were to get lower premiums or prices. Despite this, Viamedia implied that Comcast’s platform had strong enough indirect network effects to be two-sided, and only for other reasons did Delta Dental hold that the platform was one-sided.¹⁸⁹ This confusion in the lower courts demonstrates the problems that come with extending Amex beyond payment systems.

Despite being analytically incorrect, these arguments could provide a roadmap for future technology defendants because of confusion in the lower courts. As an example, Google could argue that its platform has strong indirect network effects because the more advertisers it accepts, the more easily it can keep consumer prices at zero. This argument ignores the fact that most Google users are indifferent to how many advertisers Google has on its platform, so long as their queries are correctly answered. Even though this reasoning is technically incorrect, confusion in the lower courts means that Big Tech defendants might eventually succeed with such an argument. As more lower courts begin to grapple with Amex, this is one area of potential inconsistency that deserves further scrutiny.

In conclusion, Amex’s definition of two-sided markets is overbroad, ambiguous, and susceptible to manipulation by technology companies and other antitrust defendants. These flaws open the possibility of lower courts’ extending Amex’s protections to Big Tech.

III. Solutions

The Court’s analysis in Amex is problematic, both because of how it modifies antitrust analysis and because of its potential applications beyond payment systems. In response, Part III proposes a two-part legislative solution to limit Amex’s application to future cases. Section III.A remedies issues raised by Amex’s tenuous formulation with a novel compromise solution: replacing the simultaneous-transaction requirement with narrow categories of two-sided markets. Section III.B proposes additional legislation to explicitly override Sabre, which extended Amex in mistaken and dangerous ways. Finally, Section III.C contends that legislative solutions are appropriate to remedy problems in judge-made antitrust law.

A. Replacing the Simultaneous-Transaction Requirement with Narrow Categories of Two-Sided Markets

This Section proposes legislation that would replace the simultaneous-transaction requirement with enumerated categories of two-sided markets. Even if Amex’s net-effects analysis is the better approach for some platforms like payment systems, it should only apply to narrow categories of platforms. Net-effects analysis assumes that one can directly weigh the harms and benefits to each customer group directly.¹⁹⁰ Recognizing that it might not be possible to directly weigh harms and benefits for all types of two-sided markets, the Amex majority limited its holding only to simultaneous-transaction platforms.¹⁹¹ However, the simultaneous-transaction requirement is overbroad and prone to manipulation.192¹⁹² Recognizing this requirement as a proxy for strong indirect network effects does not solve the problem because there is no clear line delineating strong and weak indirect network effects in antitrust law.¹⁹³

Because of its flaws, legislators, scholars, and practitioners alike have advocated for overturning or narrowing Amex.¹⁹⁴ However, only the Supreme Court has the authority to overturn its own precedent and it is unlikely that the Court would overturn a case decided as recently as 2018. Alternatively, lower courts could limit Amex’s reach by interpreting it narrowly. Courts could achieve this by noting that the simultaneous-transaction requirement is simply trying to capture markets like payment systems with strong indirect network effects. They could also reason that, given the unique nature of payment systems, Amex should be limited to its facts.¹⁹⁵

Yet curbing Amex in the lower courts brings its own set of challenges. As this Note demonstrates, lower courts have not applied Amex uniformly.¹⁹⁶ They have ignored aspects of the Amex formulation and, at times, even misapplied Amex outright.¹⁹⁷ A judicial approach would take years to play out in the lower courts, with many hiccups along the way. Meanwhile, our ever-increasing reliance on technology, accelerated by the COVID-19 pandemic, means that Big Tech companies will continue to solidify their positions of power in the global economy.¹⁹⁸

Given the challenges of limiting Amex judicially, the House Judiciary Committee called for legislative change in its recent report on Big Tech and antitrust.¹⁹⁹ The report recommended overriding Amex by crafting legislation that clarifies that “cases involving platforms do not require plaintiffs to establish harm to both sets of customers.”²⁰⁰ Even more recently, Senator Klobuchar proposed an expansive antitrust reform bill with similar language to override Amex.²⁰¹ These recent developments signal political desire to change antitrust laws. However, Senator Klobuchar’s bill would fundamentally change many aspects of antitrust law unrelated to platforms, and it remains unclear how much support such a wide-reaching bill will garner.²⁰²

In light of these potential political challenges, combined with continued disagreement on how to analyze two-sided markets, this Note proposes a compromise solution: legislatively replacing the simultaneous-transaction requirement with enumerated categories of two-sided markets. Accounting for the vigorous debate on whether to employ the net-effects versus separate-effects approaches, this solution would restrict net-effects analysis only to the platforms where it makes the most sense—those with strong indirect network effects. For example, legislators could statutorily define two-sided markets as platforms that belong to specifically enumerated categories, including payment systems. The list would initially include payment systems since these are well understood to have strong indirect network effects.²⁰³ As economists reach a consensus about the two-sidedness of other platforms, they too could be added to the list.

This solution avoids overbreadth by starting with a narrow statutory definition that can expand as economic understanding evolves. Such an approach would ensure that plaintiffs are not required to prove net anticompetitive harm for novel technology platforms that economists have not yet studied. Given Amex’s high upfront burden on plaintiffs,²⁰⁴ it is better to err on the side of being too narrow than being too broad.

A narrow approach would also solve Amex’s susceptibility to framing and confusion in the lower courts. Instead of relying on the simultaneous-transaction requirement, which can be easily manipulated by technology platforms,²⁰⁵ Amex would only apply to a set of clearly defined categories. This clarity is important, as it will reduce frivolous arguments by defendants like those in Delta Dental and the NCAA antitrust litigation.²⁰⁶ This solution also addresses concerns about judicial expertise.²⁰⁷ It would ensure that judges do not have to grapple with increasingly difficult economic questions related to complicated technology platforms. Instead, decisions about two-sidedness would shift to Congress, which has the ability to consult economic and technical experts throughout its deliberations. Since Congress and economists would do the difficult work of defining two-sidedness, this solution would relieve judges from their present obligations of finding the blurry line between strong and weak indirect network effects and applying the tenuous simultaneous-transaction requirement. Other countries have acknowledged these economic complexities by enacting laws tailored to regulating payment platforms.²⁰⁸ It is time for the United States to follow suit.

B. Clarifying the Relationship Between One- and Two-Sided Markets

Section III.A proposed a compromise solution to Amex that takes into account opposing economic views of two-sided markets. However, that solution only solves problems with Amex’s framework and does not address the issues that Sabre created. Sabre incorrectly extended Amex’s holding beyond vertical restraints and held that two-sided markets could not exist in the same relevant market as one-sided markets.²⁰⁹ In effect, Sabre’s holding meant that two-sided technology platforms could only be successfully challenged by other two-sided platforms under antitrust law. This mistaken and dangerous holding has spurred the House Judiciary antitrust subcommittee to advocate for a legislative clarification that would ensure Sabre does not repeat itself in the lower courts.²¹⁰ This Section joins them in advocating for a legislative override of Sabre to supplement the compromise solution in Section III.A.

From the economic and legal perspectives alike, Sabre’s holding is utterly incoherent.²¹¹ Sabre held that Farelogix and Sabre could not compete within the same relevant market because Sabre was a two-sided platform and Farelogix was not.²¹² The court based its holding on dicta from Amex that misinterpreted a single source—hardly representative of the economic literature at large.²¹³ Since antitrust law relies heavily on market definition,²¹⁴ the Sabre court’s finding that Sabre and Farelogix were not participants in the same relevant market insulated their merger from antitrust oversight.²¹⁵ Michael Katz and Douglas Melamed have observed that Sabre’s holding “runs directly counter to the purpose of defining a relevant market, which is to identify the sources of competition faced by the firm under consideration.”²¹⁶ Indeed, in Sabre itself, the court did not factor into its analysis that the parties considered themselves competitors.²¹⁷

Not only is Sabre inconsistent with antitrust doctrine, but it also risks providing far too much protection for technology defendants.²¹⁸ For example, under Sabre, Uber could not possibly be in the same relevant market as taxi companies since taxi companies only operate on one side of the market and Uber is considered two-sided.²¹⁹ Similarly, under Sabre, Amazon Marketplace could not be in the same relevant market as traditional retailers like Target. Unlike Amazon Marketplace, which is two-sided because it facilitates transactions between third-party sellers and customers through its platform, Target is one-sided because it deals directly with customers. Since rule of reason analysis only considers harm to competitors that are in the same relevant market, Sabre’s rationale would prevent taxis from suing Uber, or Target from suing Amazon Marketplace, for anticompetitive conduct. Yet we intuitively recognize Uber and taxis, and Target and Amazon, as competitors. If other courts adopt Sabre’s confused reasoning, two-sided technology platforms may escape antitrust liability to their one-sided competitors.

Given Sabre’s implications for technology platforms, it must be overridden. This is best accomplished through legislation. Specifically, this Note calls for statutory language that explicitly asserts that multi-sided platforms can compete in the same relevant market as one-sided firms. This legislation is necessary because even though the Third Circuit later vacated Sabre, it did not reverse it.²²⁰ So, while Sabre is not technically precedential, its reasoning has not been affirmatively rejected and thus provides defendants with a tempting blueprint to raise in other courts.²²¹

Further, clarifying that Sabre was decided incorrectly is best left to Congress rather than the courts. With the recent bipartisan support for antitrust reform and broad agreement regarding Sabre’s flaws, a legislative fix would likely be faster and clearer than going through the courts. Especially with the high burden that Amex places on plaintiffs,²²² awaiting judicial self-correction poses considerable risks. It would give more time for technology platforms to become even more dominant in our economy. Thus, a legislative clarification specifying that one-sided and two-sided markets can compete in the same relevant market, together with the compromise solution from Section III.A, will go a long way toward curbing Amex’s damage.

C. Legislative Solutions Are Consistent with the Intent of the Sherman Act

This Note’s proposed legislative solutions align with Sherman Act’s intent because they would bring the law more in line with antitrust’s original goals. Scholars may argue that legislative narrowing of Amex would take power away from the courts, in violation of Congress’s intent when drafting the Sherman Act. It is true that, with its broad statutory language, the Act is widely viewed as a congressional delegation to the courts.²²³ However, the Sherman Act was also intended to prevent monopolies from forming and to allow small businesses to compete in a fair market.²²⁴

Under the Chicago School, current antitrust law has strayed from both the language and the intent of the antitrust statutes.²²⁵ Antitrust law has become far less interventionist than the drafters of the Sherman Act would have intended.226²²⁶ Amex is a prime example of this. Before Amex, the rule of reason already discouraged antitrust enforcement through a high upfront burden on plaintiffs.²²⁷ Amex made that burden on plaintiffs even higher, furthering this trend towards non-interventionism.²²⁸ In addition, Amex did not follow antitrust law’s adherence to a cautious, fact-based, common law-like approach.²²⁹ Instead, it “prescribed broad, new principles” based on controversial and evolving economic scholarship.²³⁰ Thus, even if the Sherman Act was intended as a broad delegation to courts, a legislative solution that remedies Amex would only serve antitrust law’s original goals.

Further, Congress has previously intervened when courts strayed too far from its intent with antitrust laws. For example, Congress passed the McCarran–Ferguson Act in 1945 after the Supreme Court’s United States v. South-Eastern Underwriters Association decision.²³¹ South-Eastern Underwriters held that the Commerce Clause permitted federal regulation for the insurance industry.²³² Since the insurance industry was subject to federal laws, it was also subject to the Sherman Act.²³³ This decision “precipitated widespread controversy and dismay,”²³⁴ leading to the passage of the McCarran–Ferguson Act. The Act exempted the insurance industry from the antitrust statutes, including the Sherman Act.²³⁵

Another example is the Sports Broadcasting Act of 1961, which Congress passed after a lower court held the NFL’s attempt to pool the television broadcasting rights of its member teams illegal under the Sherman Act.²³⁶ In response, Congress passed the Sports Broadcasting Act, which established an antitrust exemption for professional sports leagues.²³⁷

In sum, courts’ interpretations of antitrust laws have departed from the Sherman Act’s original goals. Historically, Congress has pushed back against court decisions with which it disagreed through new legislation. Thus, if Congress deems Amex to be in conflict with the goals of antitrust law, a legislative solution is appropriate.

Conclusion

As this Note’s analysis of Amex’s progeny has demonstrated, Amex’s formulation is overbroad, prone to framing, and insufficiently based on current economic understanding. Inconsistent application of Amex by the lower courts opens the door for Big Tech to exploit lower courts’ confusion to their advantage. Given antitrust law’s limited power over Big Tech, Amex’s additional protections for technology companies must be curbed. This Note’s two-part legislative solution promotes clarity and consistency in the courts and contributes to the broader effort to rein in Big Tech’s growing power over our society.

^* J.D. Candidate, May 2022, University of Michigan Law School. Thank you to Professor Steven Cernak for his guidance, advice, and encouragement throughout the writing process. I also want to thank every member of MLR, but especially the Notes editors, for their invaluable input on this piece. Finally, thank you to my parents for supporting me in countless ways during this project and throughout law school.