Antitrust After American Express: Down a Competitive Effects Rabbit Hole

In 2018, the Supreme Court affirmed a decision throwing out a major antitrust case that the Department of Justice and seventeen States Attorneys General had brought against American Express. The case concerned contractual restrictions the company imposes on merchants accepting its credit cards that prevent them from steering shoppers to other credit cards. The Supreme Court concluded that the government had failed to prove that these restrictions eliminated competition on both sides of a “two-sided” market because its case focused on harm imposed on merchants without giving enough attention to what it meant for cardholders.

American Express accelerated a half-century trend in modern antitrust orthodoxy that has disarmed enforcers by imposing impossibly high burdens on them to analyze and predict the competitive conditions of complex markets in order to make out a case. It also ensured that conditions supporting lax antitrust enforcement in the analog era would persist in the digital one. The consequences have already played out in two major tech cases that the government saw tossed out by courts following the precedent of American Express: the review of the Sabre/Farelogix merger, and the monopolization case against Qualcomm.

The trio of cases exemplifies the failings of basing the antitrust laws on predictions of “actual competitive effects” using economic theories, and highlights the need for a different approach that can take on the challenges of regulating a complex modern economy.

The American Express Decision

The DOJ and the seventeen states joining its case had a simple theory for why American Express had violated antitrust laws prohibiting companies from entering into contracts that “unreasonably restrain trade.”

American Express charged a higher “swipe fee” to merchants than its competitors (Visa, Mastercard, Discovery) each time that a shopper used one of their credit cards to pay for something. Meanwhile, its contracts with merchants prohibited them from “steering” shoppers away from paying with an Amex in favor of an alternative credit card—for example, one charging a lower swipe fee. The higher swipe fees paid by merchants to American Express were then passed down to shoppers in the form of higher prices. And since a sufficiently large and high-spending number of shoppers uses its credit cards, American Express had sufficient “market power” to make it difficult for a merchant to simply stop accepting them as a form of payment.

The court hearing the case in the first instance agreed with the government’s claims. Relying on a massive fact record of documents, data, and testimony gathered in a five-year litigation and culminating in a seven-week trial, the court ruled in a 150-page decision that American Express had violated the antitrust laws. It concluded that the anti-steering restrictions “caused actual anticompetitive harm” in the market when they eliminated “price competition” by “denying merchants the opportunity to influence their customers’ payment decisions and thereby shift spending to less expensive cards.”¹ This, “in turn, results in higher costs to all consumers,” to whom the higher swipe fees are passed down in the form of higher retail prices.

But the tide turned when American Express appealed and got the decision overturned. The appeals court ruled that the trial court judge and the DOJ had erred in focusing the analysis on the competition between credit card networks to gain acceptance by merchants, and not doing enough to consider what was happening in their competition to attract cardholders. The court held that a proper analysis had to take into account both sides of this “two-sided” market.²

Specifically, the government had to show that the anti-steering provisions “made all Amex consumers on both sides of the platform—i.e., both merchants and cardholders—worse off overall”, by showing that they had the effect of increasing the “net price” (or “two-sided price”) across the entire transaction platform. “Without evidence of the net price affecting consumers on both sides of the platform,” the government failed to meet this burden and its case came up short.

The Supreme Court affirmed that ruling. Its decision mirrored the reasoning of the appeals court in concluding that the trial court erred when it failed to “include both sides of the platform—merchants and cardholders—when defining the credit-card market.”³ Considering the two-sided market “as a whole”, the Supreme Court found that the government had failed to prove anticompetitive effects because it had only put forth evidence of increases in merchant’s swipe fees, without due regard for what might have happened on the other side of the market with cardholders (the consumers).

In particular, the government had not shown whether cardholders benefited (for example, from obtaining card loyalty rewards) even while merchants may have suffered harm (for example, from higher swipe fees). With insufficient evidence that the price on a broader “transactions” market had been increased by the anti-steering provisions, the DOJ’s case failed because it did not “offer any evidence that the price of credit-card transactions was higher.”

Competitive effects run amok: American Express as antitrust gospel, not anathema

Though no doubt disappointing to the DOJ and other supporters of the case, its fate should not have come as a big surprise. Trial courts get overturned all the time, including in high-profile antitrust cases brought by the government, and often on similar grounds. American Express was simply the latest case to run up against a tried-and-true stonewall erected by a modern antitrust orthodoxy: that intervening in private markets should require strong proof that its actors are, in fact, harming competition.

This was not a preordained state for antitrust. As I discussed in my recent article, antitrust used to be guided by a set of objective, hard-and-fast rules that presumed particular market structures and conduct were harmful to competition, and deemed them unlawful absent a strong rebuttal by the defending parties. But for the last half-century, antitrust has shifted away from this so-called “economic structuralism” in favor of a complex, subjective framework that delves into the “actual competitive effects” of market activities before adjudging them unlawful.

It was seemingly a sensible and rationale transformation (though one also motivated by a neoliberal, pro-market ideology). But, as I discuss, it has turned antitrust into a precarious prediction game dependent on economic theories and models whose effectiveness in complex markets is highly questionable. And requiring proof of actual anticompetitive effects has become a litigation morass as antitrust plaintiffs are made to clear higher and higher evidentiary hurdles. The consequence has been an ever-more lax antitrust enforcement regime. This trend is most obvious in the online economy. Fast-changing and difficult-to-understand digital markets make it all but impossible to prove an antitrust claim by showing that a given merger, monopolist’s conduct, or agreement in restraint of trade is, in fact, harmful to competition.

Against the backdrop of an antitrust transformed into an Economism-based, predictive regime, the Supreme Court’s American Express decision is just another in a long line of cases increasing the burden of proof for antitrust plaintiffs. And so, as far as mainstream antitrust goes, there was nothing controversial in the Supreme Court’s framing the government’s obligation as one of proving that the anti-steering provision “has a substantial anticompetitive effect that harms consumers in the relevant market.”

What does make American Express unique, and the reason it has pushed the trajectory of antitrust even further into a competitive effects abyss, are the implications on the modern tech-based economy of the Supreme Court’s views on the proof that is required in cases involving two-sided markets.

Two-sided platforms are at the core of wide swaths of the online ecosystem, including retail (Amazon’s marketplace), social media (Facebook), online advertising (Google Ads), the internet of things (Apple’s HomePod), search (Microsoft’s Bing), and the gig economy (Uber), to name a few examples. The American Express decision has significantly raised the evidentiary bar for proving up an antitrust case in such markets. It will no longer be enough to show that a platform harmed competition on one side of the market—as difficult and burdensome as that task already is. Now “substantial anticompetitive effects” must be shown across both sides of the market, accounting for all the participants and users of a multi-sided platform in something akin to the “credit card transactions” market proposed in American Express.

But the logic underlying the American Express decision does not stop at multi-sided platforms. It is not difficult to imagine how creative defendants and laissez faire-inclined judges could spin a web of ever-increasing complexity in any case about a sprawling market with interconnections and interrelationships among different users, partners, and participants. This is a natural consequence of falling down the competitive effects rabbit hole. If it is not reined in, the competitive effects machinery tends towards entropy, especially in complex digital markets where a single player can be interacting with various segments of a broader digital ecosystem.

These are no longer just predictions, but lived realities. Since American Express came down, parties opposing government antitrust enforcement actions have taken that decision and run with it.

Antitrust in tech markets after American Express

In the two years since the American Express decision, courts have already relied on it to toss out two more major antitrust cases brought by the government, both involving tech markets.

Sabre/Farelogix

The first of these cases involved the DOJ’s effort to block a merger. Sabre was seeking to acquire Farelogix, its competitor in offering booking services to airlines. Sabre operates a two-sided transaction platform that connects airlines to travel agencies (or travelers) for the sale of tickets and other services. Farelogix provides IT solutions to airlines that are used to sell tickets to travel agencies (or travelers).

The DOJ concluded that the deal would harm competition. It believed that Farelogix acted as a competitive constraint on Sabre to the extent that it provided an alternative for airlines that rely on such third-party services to sell tickets to travel agencies and end customers. The evidence at trial—including company documents and testimony from airlines—showed that the two viewed each other as competitors and that some airlines were able to use this to seek lower commission fees from Sabre. The court hearing the case found that “it is logical to conclude that part of Sabre’s interest in acquiring Farelogix is to mitigate the risk” resulting from the fact that its technology enables airlines to bypass Sabre’s transaction platform.⁴

Nevertheless, the court ruled that the DOJ failed to meet its burden of proof to “show that this purchase will harm competition on both sides of the two-sided market” for travel services provided to airlines and travel agencies. Citing the American Express decision, the court said: “As 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.” Therefore, it was not enough to prove that the merger would harm competition on only the one side of the two-sided market that Farelogix is active on.

And so despite the extensive evidence of competition between the companies, the court had to conclude that, as a matter of law, “Sabre and Farelogix do not compete in a relevant market.” To succeed in blocking the merger, the DOJ would have had to “produce evidence that the anticompetitive impact of the merger on the airline side of the [transaction] platform would be so substantial that it would sufficiently reverberate throughout the [platform] to such an extent as to make the two-sided [transaction] platform market, overall, less competitive.”

Qualcomm

The second case that shows how American Express left its mark on antitrust is a monopolization (abuse of a dominant position) case brought by the Federal Trade Commission against Qualcomm. The case involved modem chips used in smart phones. Qualcomm made the chips, but it also held important patents for the technology. Rival chip makers licensed that technology from Qualcomm to produce their own competing chips.

The FTC alleged that Qualcomm had abused a dominant market position when it refused to sell its chips to smartphone manufacturers unless they also entered into a patent license (which required making a royalty payment) for any chips that they acquired from not only Qualcomm but also any of its rival chip makers. This practice, the FTC argued, imposed an anti-competitive surcharge on rivals’ chips which raised the barriers for competing with Qualcomm. This, in turn, hurt the phone manufacturers by inflating the price they paid for chips.

The court hearing the case in the first instance agreed, and ruled for the FTC. But an appeals court overturned the decision. On the main antitrust theory of the case, the appeals court reasoned that the FTC had failed to prove that Qualcomm’s “no license, no chip” policy harmed the “area of effective competition.”⁵ Although its evidence had shown how the policy could have increased costs for Qualcomm customers (phone makers) who buy the chips, it had not shown how the policy harmed competition by directly impacting Qualcomm competitors (rival chip makers). It pointed to the ruling in American Express that the DOJ in that case had failed to meet its burden of proof because it did not show how restrictions imposed on merchants “have anticompetitive effects that harm consumers” (italics my own).

The analogy to the Qualcomm case seems to have been that the FTC needed to connect all the dots—customers and competitors alike—in proving anticompetitive effects. Showing that the “all-in” (royalty plus sales) price charged to customers might have been inflated by Qualcomm’s licensing practices was not enough because it “falls outside the relevant antitrust markets” at issue.

Down the competitive effects rabbit hole

The American Express, Sabre/Farelogix and Qualcomm cases share three traits in common that show how the half-century transformation of antitrust into an Economism-driven, predictive framework is undermining enforcement, especially in tech markets.

First, the cases show how the government agencies bringing an antitrust case and the courts rendering the decisions in them must undertake a massive burden. They have to dissect the inner workings of a market and then make predictions or conjectures about actual competitive effects in the market that result from the conduct at issue. In American Express and Sabre/Farelogix, it was proving lower output and higher overall “net” (or “two-sided”) prices on multi-sided transaction platforms. In Qualcomm, it meant proving “an anticompetitive surcharge on rivals’ modem chip sales” by directly linking up proof of harm to customers with proof of hindering competitors.

In all three instances, the burden imposed by the courts for proving these so-called “actual anticompetitive effects” was simply too high for the government to meet. Qualcomm arguably went even further in raising the evidentiary bar for tech cases. The influential appeals court issuing that decision went so far as to declare that “novel business practices—especially in technology markets—should not be ‘conclusively presumed to be unreasonable and therefore illegal without elaborate inquiry as to the precise harm they have caused or the business excuse for their use’” (italics my own). Requiring “elaborate” and “precise” proof would seem to doom all but the slam-dunk government actions against tech.

Second, the trio of cases shows how proof of actual anticompetitive effects depends heavily on economic theory and models. The Supreme Court sets the pace in American Express by relying entirely on a string of academic articles by economists—citing nothing from the fact record of the case before it—to construct its “two-sided transaction platform” market and reach the critical conclusion that “[e]valuating both sides of a two-sided transaction platform is [] necessary to accurately assess competition.”

Sabre/Farelogix picks up the baton and runs with it, relying on that theory-based legal holding in American Express to ignore an exhaustive factual record of company documents, executive testimony, and third-party complaints showing close competition between the merging companies. Qualcomm then carries the baton across the finish line when it frames the case with a skepticism of “novel” theories of competitive harm by citing blanket assertions in two academic article about how antitrust cases of technology markets skew towards over-enforcement.⁶ When it comes to economic theory and a predictive antitrust that requires proof of actual anticompetitive effects, the tail wags the dog.

Third, these three cases rest on a critical assumption—arguably bordering on a blind faith—that economics is up to the task of proving actual competitive effects. Baked into the courts’ reasoning is that economics can be used to understand and predict complex market environments that change in real-time in often unexpected ways. Yet, as discussed in my recent article, it has yet to be empirically proven—or seriously tested—that economics can perform the sort of analyses and predictions that would justify its having become the foundational underpinning of the enforcement of the antitrust laws. If anything, real-world experience in competition law practice combined with general research on uncertainty and decision-making suggest that expert judgments are poor predictors in complex environments like those at issue in antitrust cases.

And as they push antitrust further down an Economism-driven path, the courts provide little guidance on how plaintiffs are to meet their super-sized burden for proving actual anticompetitive effects. In American Express and Sabre/Farelogix, the government’s case is thrown out because it failed to prove an increase in the “net” or “two-sided” prices on a multi-sided transaction platform. But such a thing exists only as a figment of a court’s imagination. It does not exist in the real world. No one pays it, and no one charges it. And it’s unclear how an antitrust plaintiff is to go about the precarious exercise of weighing benefits to one side of a market against the harms to another. In American Express, for example, would it mean weighing the swipe fees charged to merchants against the rewards points earned by shoppers? In the absence of any guidance, it can safely be assumed that economic theories and models are expected to conjure such “net” prices into existence.

The trio of cases, therefore, reflects and even propels a broader trend that has eviscerated antitrust enforcement—especially in tech—by erecting high barriers for plaintiffs to prove actual anticompetitive effects using dubious economic tools.

A modern antitrust in peril

With the Sabre/Farelogix and Qualcomm cases, the American Express decision has rounded out its influence on the three main pillars of US antitrust law: mergers, monopolization, and contracts in restraint of trade.

None of the three cases sets out groundbreaking new law. Their significance lies rather in accelerating a trend, half of a century in the making, among policymakers, academics, and judges to require antitrust plaintiffs to take on an ever-increasing burden of proof in using economic tools to show how market conduct harms competition. Each such case is an individual brick in a rising wall—reaching its tallest heights in tech markets that are especially difficult to understand and predict—that plaintiffs must scale to bring a successful antitrust case.

The consequence is not just an intellectual failing about humankind’s ability to make accurate predictions in unpredictable markets. It also means lax antitrust enforcement and the mass-consolidation of economic power across the economy.

The American Express case and its recent progeny are yet another warning sign about the need to rethink antitrust for the modern economy. As discussed in my recent article, one solution may be found in looking to social science research in decision-making and antitrust’s own history with economic structuralism to create a simple, formulaic and non-predictive approach to antitrust. At a minimum, the questions being asked in discussions about antitrust reforms need to dig much deeper to get at the foundational source of the problems.