Platforms in the Spotlight at the FTC Hearings
The Spotlight on Platforms
Over a series of three days, from October 15 to 17, 2018, the Global Antitrust Institute (GAI) hosted the third session of the FTC Hearings on Competition and Consumer Protection in the 21st Century at the Antonin Scalia Law School, George Mason University. After introductory remarks by Scalia Law School Dean Henry Butler and Federal Trade Commissioner Rohit Chopra, the panels broadly focused upon three topics: (1) multisided platforms, (2) nascent/potential competition, and (3) labor policy. The overarching focus of the hearings was clearly multisided platforms, which accounted for seven of the twelve panels.
In the wake of the Supreme Court’s ruling in Ohio v. American Express, platforms are rightly the subject of significant attention and discussion ranging from the economic tools and methods to evaluate platforms to the correct legal framework for capturing and shaping that analysis. The economic research on platforms began in earnest fifteen years ago with pioneering work from Evans (2003) and Rochet and Tirole (2003).2 Much of the important economic learning over the past fifteen years—economic concepts such as cross-group effects, the interrelationship of demand, and the difference between transaction and non-transaction platforms—are becoming part of the antitrust lexicon. Much of the discussion at the hearings mirrored the conversation taking place among academic economists and lawyers, as well as practitioners and courts, focusing upon how to integrate that economic learning into antitrust institutions and doctrine.
While the dueling set of amicus briefs for the American Express case indicates there are certainly areas of disagreement, we highlight what we believe emerged throughout the hearings as areas of general consensus in how to assess platforms in an antitrust context.
Areas of Consensus
First, cross-group effects are important to understanding the competitive effect of conduct involving platforms.3 Both David Evans and Catherine Tucker, who presented an overview of platforms, emphasized the importance of cross-group effects and their role in driving the value that platforms offer. Platforms reduce the frictions between two or more potential trading groups, such as passengers and drivers on a ride sharing platform. The reduction of transaction costs and the internalization of externalities are the defining roles of platforms, and previously unrealized gains from trade are unlocked. Without material cross-group effects, there is little that “attracts” two or more groups to each other—which limits the role that a platform can play in unlocking that value.4
Cross-group effects are important regardless of whether the platform in question is a transaction or non-transaction platform.5 For instance, the panelists for the session “Understanding Exclusionary Conduct in Cases Involving Multi-Sided Platforms: Predatory Pricing, Vertical Restraints, and MFN,” agreed that a wholistic view of both sides must be taken when evaluating a multi-sided platform to see the platform’s true value to consumer welfare. One application of this is in regard to the relevant product market. Regardless of whether one or two relevant markets are defined, cross-group effects are critical to incorporate in market delineation exercises. For instance, in the panel “Defining Relevant Markets and Establishing Market Power in Cases Involving Multisided Platforms,” Joanna Tsai emphasized that the application of standard antitrust analytical techniques such as the hypothetical monopolistic test, when applied to platforms, would often result in relevant markets that are too narrow if not considering all sides.
Second, platforms are not a homogeneous group. Special rules for assessing platforms are likely to suppress important economic distinctions. The panel “Multisided Platforms in Action,” brought together experts from diverse backgrounds: two academics, two industry venture capitalists, and an economic analyst. The overarching contribution of that panel was that while many large technology companies are platforms, distinctions between the many different types of platforms must be made in order to understand their respective value propositions, the competitive constraints they impose on rivals, and the impact of changes in their business models on competition and consumers. Ben Thompson drew a distinction between platforms that facilitate connections between third parties and users, and aggregators that stand between the third party and the users. Steven Tadelis compared platforms and marketplaces with marketplaces enabling one-to-one transactions between buyers and sellers, and Roger McNamee viewed transaction platforms as starkly different from non-transaction platforms such as Facebook and Google. Echoing this theme, panelists in the session “Understanding Exclusionary Conduct in Cases Involving Multi-Sided Platforms: Predatory Pricing, Vertical Restraints, and MFN,” agreed that multi-sided platform cases are very fact-specific and one-size does not fit all.
Third, price effects on one side of platform cannot be completely understood without understanding the incentives to cater to all groups on the platform. Competitive effects analysis requires more than making simple inferences from the price on one side of the platform. For instance, platforms can profit maximize with below cost pricing to one group, e.g., social media users, in order to attract—via cross-group effects—another group, e.g., advertisers. This was a point emphasized by David Evans. For platforms, focusing solely on price is complicated by the fact that there are two prices that determine output for transactional platforms such as payment cards and services such as Uber. Even for non-transaction platforms, such as search engines and newspapers, the prices on both sides are interrelated due to cross-group effects. While one can conceptually construct a combined “platform price” or a “net price,” this approach has a number of shortcomings. Consequently, for markets involving platforms, output should be the touchstone for measuring changes in consumer welfare rather than price. Thus, there is little justification in discussing welfare effects on one side of platform while ignoring the effects on the other side—even if separate markets are defined. Thus, the primary emphasis should be on measuring changes to quantity rather than price wherever possible, precisely because price is a significantly noisier signal of consumer welfare in platforms relative to single-sided markets.
Fourth, complements and platforms are different. In his presentation, David Evans stated that complements are usually consumed by the same customer, while platforms serve different customers or groups. Complements are also often sold by different suppliers, whereas platforms supply services to both types of customers. Similar to Evans, in a recent article, Lapo Filistrucchi, while acknowledging that complements such as printers and ink have some features similar to two-sided platforms—including the interrelationship of prices between the two sides—finds complements and platforms to be quite different from an economic perspective.6 For instance, cross-group effects in platforms can be negative, which is not the case for complements. Additionally, two groups on a platform do not internalize the externality imposed on each other—a problem which a platform can solve. This is not true for complements since complementary goods are bought by the same consumer, with the consumer considering both prices when making decisions. Finally, there are two or more welfare considerations for platforms, in that there are two or more groups participating on a platform, while there is only one consumer of complementary products.
Fifth, not everything is a platform. The correct focus should always be the economic forces at work rather than the labels applied to the business model or specific firm. While there is some sense in which indirect network effects exist in the context of single-sided markets, e.g., when more manufacturers supply Walmart this increases the value to Walmart customers, it is not materially the same as for platform markets. This was a point of agreement in the first panel “The Current Understanding of Multi-Sided Platforms.” While there are elements of cross-group effects in a traditional retail setting, the difference lies in whether they are fundamental to the traditional retail business model. Katja Seim, David Evans, and Catherine Tucker all raised doubts as to whether or not that is true. For instance, Katja Seim mentioned that Walmart’s success is not due to cross-group effects but due to its superior logistics and other unrelated factors.
Sixth, multihoming and switching costs are important components in assessing the market power of platforms. These topics were addressed in Catherine Tucker’s opening presentation. While it is important to understand the nature of the multihoming and whether or not the platforms that are multihomed are actually economic substitutes or not, the point remains that multihoming is material to understanding a platform’s market power and the durability of that market power. The existence of multihoming and low switching costs further raises questions regarding the value of just using market shares as a proxy for market power.
Finally, platforms do not get an antitrust free pass. For instance, during the panel “Understanding Exclusionary Conduct in Cases Involving Multi-Sided Platforms: Issues Related to Vertically Integrated Platforms,” the participants offered differing views in regard to how to assess exclusionary conduct involving platforms. Nonetheless, there was agreement that exclusionary conduct should be closely examined. However, in turn, platforms should not be subject to different rules or special scrutiny. For instance, big data plays an important role in platform quality, but could potentially represent an area of competitive advantage for incumbents that entrants must overcome. However, it is not clear this is much different from other hurdles that entrants in many markets must overcome including economies of scale, patents, trade secrets, and brand loyalty.
To sum up, while the sessions on platforms certainly raised areas of disagreement and further work, a surprising amount of economic consensus emerged.
1 Joshua D. Wright is University Professor and Executive Director of the Global Antitrust Institute, Antonin Scalia Law School, George Mason University. John M. Yun is Associate Professor of Law and Director of Economic Education, Global Antitrust Institute, Antonin Scalia Law School, George Mason University.
2 The groundwork for the economic literature on platforms can be traced to the earlier work on the related topic of network effects.
3 Cross-group effects are also referred to as indirect network effects.
4 This is a point made by Leah Brannon in the panel titled “What Can U.S. v. Microsoft Teach About Antitrust and Multi-Sided Platforms?”. Specifically, courts need to consider both sides of a two-sided platform except when the indirect network effects are minor.
5 Examples of transaction platforms include payment card systems, ride sharing apps, and eBay. Transaction platforms enable a direct exchange between the two sides and, thus, the two sides share a common output. Examples of non-transaction platforms include newspapers, online search engines, and ad-supported social media. Non-transaction platforms are not necessarily facilitating a direct exchange and are generally monetized through advertising.
6 Lapo Filistrucchi (2018), “Two-Sided vs. Complementary Products,” CPI Antitrust Chronicle, pp. 1-7.x