Nearly three years ago, the Supreme Court decided the case of Ohio vs. American Express, which turned out to be one of the most significant—and controversial—antitrust decisions in recent memory. Here is what the Amex decision got right, what it got wrong, and how we can bridge the two.

In June 2018, the Supreme Court ruled on Ohio vs. American Express, one of its most important and controversial modern antitrust rulings. The decision held that all transaction platforms should be treated as a single two-sided market, a view that could substantially shape the analysis of all antitrust cases in the digital economy. After almost three years, it is important to take stock of this ruling and consider its strengths and weaknesses.

The so-called Amex case considered whether American Express violated antitrust laws when it used contractual provisions to prevent merchants such as super-markets or hotels from steering clients towards credit card brands that charged lower fees, such as Visa or Mastercard. At the heart of the judicial controversy was the definition of the “relevant market” in which American Express competed. Relevant markets are the lenses through which competition authorities and courts assess the existence of market power and the effects of specific conduct. This process of market definition is key to finding antitrust infringement, as different lenses highlight different aspects of the conduct and lead to different conclusions.

The DOJ and multiple states claimed that the analysis should focus on whether American Express’ high transaction fees harmed merchants, which would make this a one-sided market. American Express’ defense was that the analysis should consider not only the impact on merchants but on credit card transactions as a whole, including potential offsetting efficiencies to credit card holders that receive higher rewards per transaction (meaning, a multi-sided market).

In a 2015 decision, the US District Court for the Eastern District of New York affirmed the existence of a one-sided relevant market of services rendered to merchants and condemned American Express. However, the 2nd Circuit (in 2016) and later the Supreme Court (2018)—split along ideological lines—affirmed a two-sided relevant market and ruled in favor of the company. The ruling was also controversial for its apparent endorsement that antitrust injuries require decreased output—a mistake, as quality, innovation, etc. are definitely also relevant to market competition and to antitrust—but in this piece we focus on the issue of relevant market definition.

While credit card markets are important by themselves, Amex is a key ruling because of its impact on antitrust litigation as a whole, including cases involving Big Tech. Facebook, for example, recently asked courts to dismiss the antitrust lawsuit filed against it by the Federal Trade Commission because the agency supposedly failed to present “a plausible relevant market,” relying on Amex as a precedent. Indeed, Amex touched upon a sensitive topic and split the antitrust community: University of Chicago law professor Randy Picker argued that Justice Clarence Thomas’s majority opinion was “a bold leap into modern industrial organization economics,” while Columbia law professor Tim Wu asserted that the Court delivered a “weak, highly abstract decision” that was “a big blow to antitrust law”. October’s House Majority Report on Big Tech called on Congress to override the decision, as well as other lower court rulings that relied on it.

In a paper recently published in the Antitrust Law Journal, we explore the weaknesses of both the majority and the minority opinions in Amex and develop a middle ground that can help bridge these stark differences in academic and judicial views.

Justice Stephen Breyer’s minority opinion argued that credit-card markets are better understood as the combination of multiple one-sided markets (e.g., services rendered to merchants, cardholders, etc.). In doing so, it equated multi-sided markets to complementary goods that must be consumed together to have value, such as gasoline and tires.

This view, however, is problematic because it washes away network effects that create a much stronger interconnection between price structure and demand than what is normally found in a traditional gasoline and tire duo. As a result, the minority opinion’s limited focus on only one side of the market (i.e., the card services rendered to merchants) is faulty: as Picker rightly emphasizes, it ignores much of the modern literature in industrial organization that culminated with the Nobel Prize in Economics awarded to Prof. Jean Tirole in 2014. Understanding that American Express is running a multi-sided market is important because otherwise the company cannot argue in its defense that part of the higher merchant fees it charges on one side translates into higher cardholder rewards on the other side, an essential part of its business strategy.

Yet, this does not mean that the majority opinion is automatically correct. On the contrary, Wu’s critique that the decision was weakly worded and abstract is also accurate, as the majority opinion overstepped, used broader language than necessary, and appeared to argue that all two-sided transaction platforms should be treated as a single two-sided relevant market. This conclusion similarly ignores the evolution of economic and legal theory and is likely to lead to much confusion as lower courts try to apply the Amex precedent to other cases involving digital platforms—as shown by the fact that even the Supreme Court itself failed to apply Amex to a case involving Apple decided in its very next term, despite a clear applicability.

The Amex majority opinion also goes against international precedents involving the enforcement of antitrust laws to credit card markets: in both the EU and Brazil—sophisticated jurisdictions with more relevant enforcement in this industry than the US—authorities and courts have acknowledged that credit card networks are composed of multiple relevant markets.

American Express Building
American Express Building. Photo by Ken Lund, via Flickr [CC BY-SA 2.0]

There is, however, a way to bridge the gap between the minority and the majority opinions in Amex: recognizing that that many digital platforms are often composed of multiple markets that interact in a complex manner, thus providing better guidance to lower courts and to the antitrust community.

This multi-layered approach to market definition can be achieved through a four-step process framework, described below in a nutshell:

Step 1: Evaluate the structure of the platform

Authorities must evaluate the structure of the platforms and identify whether and where they enable/encourage intra-platform competition, inter-platform competition, or both. This means mapping out the different economic activities that compose or that are directly connected to the platform under analysis—merchants, cardholders, issuing and acquiring banks, and other payment networks in the example of payment cards, but also buyers, sellers, advertisers, and other platforms in the example of marketplaces.

Step 2: Define a preliminary relevant market (focusing on one-side)

It is important to start by properly understanding the competitive dynamics on the specific, usually one-sided competition space most affected by the conduct or merger under scrutiny. For example, in a merger between banks that issue credit cards, credit card issuing should be the preliminary relevant market that authorities must then scrutinize. At this stage, it is especially important to verify whether the platform itself is designed in a way that promotes/encourages intra-platform competition on that side of the market (i.e., multiple banks competing to issue Mastercard cards) and/or inter-platform competition (i.e., the same banks also issuing other cards, such as Visa). Strong intra-platform competition suggests a one-sided market, while strong inter-platform competition may indicate the presence of a single multi-sided relevant market.

“a more nuanced approach to market definition will increase predictability, give authorities and courts a harder edge to deal with competitive issues in complex industries, and lead to a sounder antitrust policy than the current interpretation of the Amex decision.”

Step 3: Consider potential multi-sided constraints and define one-sided or multi-sided markets. 

Once authorities identify the competitive constraints on one side of the market and define one (or, in case necessary, multiple) preliminary relevant markets, they must then assess whether constraints posed by the other side(s) of the platform are relevant enough to influence the strategic behavior of the players under scrutiny—something that justifies moving to the definition of a two-sided market.

In most cases, this decision for one- or two-sided markets will depend on the following aspects: (i) the structure of the platform and its internal rules (e.g., whether it allows for intra-platform competition); (ii) whether the players have any control over how the platform reaps the benefits of the value it generates to multiple users; and (iii) the strength of indirect network externalities (i.e., the interdependence between different sides).

This is the most important step, and source of much of the controversy in the literature around this topic. When network effects are intense and the firms under scrutiny have some control over variables that affect them, it is reasonable to assume that competitive strategies will consider more than one side of the platform, leading to the definition of a multi-sided market. Although regulators and parties can use some quantitative methods in assessing these factors, they may also consider qualitative evidence indicating the presence of strong indirect network effects, such as when:

I. The company controls or influences prices on both sides of the platform;
II. The company can prevent arbitrage between the different sides;
III. The company is exposed to inter-platform competition, such as when markets face intense multi-homing;
IV. The company obtains most profits from fees, rather than the buying/re-selling of goods/services;

If most of the four items above hold true, then regulators and parties have strong signs that the market should be considered multi-sided.

Step 4: Define the extension of the market.

If the evidence in Step 3 indicates the presence of a multi-sided market, a final question remains as to how to assess the extension of this market vis-à-vis other potential substitutes—for example, to what extent credit cards compete with debit cards.

On the other hand, if the evidence in Step 3 does not indicate the presence of strong network effects, authorities should then focus on the extension of this one-sided relevant market—for example, whether banks that issue credit cards located in Chicago compete with banks in Springfield/IL. Even if Step 3 rules out the presence of strong network effects, authorities may still consider the impact of weak indirect network effects in assessing the firm’s market power in one-sided markets.   

This multi-layered framework can provide a better rationalization for a lot of the antitrust litigation surrounding digital platforms. Acknowledging that multi-sided platforms can include multiple one-sided and two-sided relevant markets is a more accurate description of reality. For example, depending on the circumstances, payment cards networks can include relevant markets at the platform level (i.e., competition between Visa, Mastercard and Amex), at the issuing side (i.e., competition among issuing banks) and at the acquirer side (i.e., among acquirers). This structure is somewhat similar to online marketplaces and App Stores, which also include a two-sided market with competition at the platform level and multiple one-sided markets where sellers compete to sell substitute products. This clarification helps us understand, for example, why Apple could be accused and ultimately condemned for organizing a price-fixing scheme in the one-sided market of publishers of e-Books despite having 0 percent market-share in the two-sided market of e-Books platforms—another recent and controversial antitrust case.

Paradoxically, a more nuanced approach to market definition will increase predictability, give authorities and courts a harder edge to deal with competitive issues in complex industries, and lead to a sounder antitrust policy than the current interpretation of the Amex decision. While defining relevant markets is important for antitrust, it should not prevent a solid analysis of how business conduct impacts markets in reality.

Disclaimer: Both authors acted as external counsel to companies operating in the electronic payments sector in Brazil, including Alelo, Bradesco, Cielo and Elo. Filippo’s involvement ceased in 2015, Caio is still retained by these companies.