MONOPOLIZING DIGITAL COMMERCE.

• Author: Hovenkamp, Herbert

TABLE OF CONTENTS INTRODUCTION 1680 I. THE MEANING OF "MONOPOLIZATION" 1684 II. THE EXPANDING DOMAIN OF [section] 2 1688 III. MONOPOLIZATION AND FIRM STRUCTURE 1696 A. The Tort Theory of Monopolization 1696 B. Structuralism and the Failed Effort to Kill It 1698 C. Unique Structural Presumptions for Digital Networks? 1707 D. Market Structure and Attempts 1709 E. Secondary Leverage and Abuse of Dominance 1712 1. Dominated Networks: Monopolists an Secondary Markets 1712 2. "Leveraging" Under U.S. Law: Recognition and Rejection 1714 3. Agreements and Secondary Market Harm 1717 4. The Case for Abuse of Dominance? 1718 5. Abuses of the "Abuse" Standard 1721 6. Competitive vs. Dominated Networks 1726 IV. EXCLUSIONARY PRACTICES ON DIGITAL PLATFORMS 1729 A. Vertical Integration: Refusal to Deal and "Self-Preferencing" 1730 1. "Sacrifice" 1735 2. Refusal to Deal, "Self-Preferencing," and Copying 1736 B. Exclusionary Mergers 1741 C. Anticompetitive Technology Design and Restraints on Innovation 1746 D. Exclusionary Patent Practices 1751 CONCLUSION 1755 INTRODUCTION

In a single sentence, section 2 of the Sherman Act condemns firms that "monopolize," "attempt to monopolize," or "combine or conspire" to monopolize--all without explanation. (1) Passed a quarter century later, section 1 of the Clayton Act offered a few helpful definitions. (2) It defined "antitrust laws," "commerce," and "person" so as to include corporations. (3) But the definition provision said nothing about the meaning of "monopolize." (4) That is, it failed to define the term that was most important and ultimately became most controversial. No other federal statute has used so few words to condemn acts that are as eclectic, diverse, and unspecified as those covered by section 2 of the Sherman Act. It should go without saying that nothing in the text of the Sherman Act, which was enacted in 1890, contemplates networks, digital markets or anticompetitive conduct that might affect them. (5) Of course, neither did it contemplate markets for automobiles, airplanes, or televisions. One reason for antitrust's durability has been that its broad language cuts across all technologies, past, present, and future.

As a result, criticisms that the antitrust statutes are out of date and not up to dealing with dominant digital firms today cannot be based on readings of the text. (6) The antitrust statutes, including section 2 of the Sherman Act, are literally broad enough to reach nearly every threat to competition that the dominant firms pose. Rather, decades of narrow construction have led to most of the problems.

One exception is situations in which a dominant firm's conduct threatens harm falling short of monopoly in a second market. Section 2's condemnation of "monopolizing" conduct cannot literally be construed to reach such behavior. While this problem is ubiquitous in the law of monopolization, it is particularly prominent in networks. In an increasingly networked economy, the operations and fates of more firms are linked together. Although networks are socially very valuable, they can also lead to a broader range of competitive harms, including situations in which monopoly is not realistically threatened in a second, or complimentary, market. Here, the United States would do better to adopt an "abuse of dominance" standard such as the one used in the European Union (EU) and other jurisdictions, but with limitations on potential overreach.

This Article considers the problem of monopolization when firms operate in more than one market, particularly on networks. Much of the focus is on the large digital platforms that have claimed so much public attention, namely, Amazon, Apple, Meta (Facebook), and Alphabet (Google). But the problem is not limited to them. The Sherman Act itself is formally indifferent to the market in which challenged conduct occurs. For example, the statements of the law are the same for all markets regarding refusal to deal, (7) predatory pricing, (8) or exclusive dealing. (9) Further, most of the law in those areas was developed in situations that did not involve either digital firms or networks.

The distinctive feature of a network is some linkage other than that which occurs between a single seller and a single buyer. A network may connect two or more buyers. In some cases a single owner controls the network. Often networks are "collaborative" in the sense that they are operated by multiple firms. Among these are "dominated" networks, such as Microsoft or Apple, in which one firm manages the network and issues interconnection rules or protocols that must be followed by others. Other networks, such as the phone system, are more collaborative and do not have a dominant firm. The various participants on a network operate as both competitors and as sellers of complements. For example, cellular phone sellers Apple and Nokia compete for sales, but they must also cooperate when the two are connecting a call.

To the extent that firms operate in networked markets, the need for interfirm cooperation is intensified. In particular, the current antitrust law of unilateral refusal to deal is not well designed to handle conduct that arises in multifirm networks. (10) That concern is hardly limited to the largest digital firms. Further, it is even more serious in firms that might be smaller overall but have larger market shares in their respective products. For example, for purposes of assessing competitive effects, the fact that Alphabet is a very large firm is not nearly as important as the fact that Google Search, one of its products, has a dominant market share. (11)

To the extent the digital platforms have high fixed costs and deal in information, some of the tools we use to assess exclusionary practices work poorly. This is particularly problematic for "cost based" theories of exclusion, such as those applied in the law of predatory pricing. (12) The high fixed cost problem is not universal, however. For example, Amazon deals heavily in ordinary tactile products that have conventional cost structures. By contrast, Meta's (Facebook's) content is almost exclusively digital, as is most of Alphabet's. A related feature of all of the platforms is that, to one degree or another, they are "two-sided," which typically means that customer engagement and revenue come from two different groups of transactions with different parties. (13) The manager of the platform often acts as a broker, or go-between, in addition to being a seller. (14)

Further, many of the injuries imposed by dominant firms in networks occur in complementary or vertically related markets in which monopoly is not seriously threatened. This poses particular problems for section 2 of the Sherman Act, which requires a realistic threat of dominance in the particular market where the injury is claimed. (15) For example, Apple's insistence that application sellers use its own store and pay Apple's commissions harms these sellers and consumers by denying them the benefit of a more competitive marketplace. (16) However, Apple is not realistically threatening to create a monopoly in the market currently occupied by, say, Epic Games, (17) where Apple has only a modest presence. (18) Other jurisdictions, such as the EU, whose law defines the violation as "[a]ny abuse... of a dominant position," (19) either do not face this problem or else face one that is seriously attenuated. (20) Of all the statutory reforms that antitrust law in the United States might take, switching to this "abuse" standard would be the most beneficial, particularly for networked markets, provided that the concerns are managed properly. It can also lead to harmful overuse. (21)

Finally, the digital marketplace has been highly productive, with an economic growth rate three or four times larger than that of the economy overall. (22) As a result, antitrust policy faces a problem that it has confronted in some form since its inception: how to control anticompetitive conduct without limiting innovation and technological progress unnecessarily.

None of these characteristics is limited to a few dominant digital platforms. Antitrust policy should apply to all similar situations as best it can. For example, eBay.com, Match.com, and Uber.com are all two-sided digital networks that have many of the same features as the larger platforms, (23) to say nothing of Microsoft, which is larger than three of the four firms currently under the antitrust microscope. (24) If legislation is to be passed under the antitrust laws, it should apply equally to similarly situated firms and circumstances.

1. THE MEANING OF "MONOPOLIZATION"

   Early case law under section 1 of the Sherman Act, which prohibited agreements in restraint of trade, (25) relied heavily on common law precedents defining restraint of trade as agreements to restrain market output or exclude competitors. (26) Few such precedents existed, however, for the law of single-firm monopolization. At common law, the term "monopoly" almost always referred to an exclusive grant or title from the government. (27) Aside from business torts and criminal law, there was no history of prohibiting purely unilateral conduct not authorized in a grant from the government. (28) The passage of the Sherman Act changed that perspective. Section 2 did not limit its reach to firms with exclusive government grants. In a 1905 state antitrust case, the Supreme Court observed that "the idea of monopoly is not now confined to a grant of privileges" but also includes a "condition produced by the acts of mere individuals." (29) The Supreme Judicial Court of Massachusetts elaborated a few years later:

   The earlier conception of a monopoly was a grant of an exclusive right from the sovereign power. This still defines with accuracy that which an inventor receives under the patent laws. But in a wider sense monopoly denotes a combination, organization or entity so extensive, exclusive and unified, that its tendency...