Faculty of law blogs / UNIVERSITY OF OXFORD

Fixing Platform Monopoly in the Google Search Case

Author(s)

Herbert Hovenkamp
Professor at the University of Pennsylvania

Posted

Time to read

8 Minutes

Alphabet (Google) is on trial, charged by the Department of Justice with monopolistic practices involving Google Search. A second trial scheduled for early 2024 involves its advertising business. In September 2023, the Federal Trade Commission filed a monopolization case against Amazon, following an earlier complaint against Meta (Facebook) in 2020.

Every complex antitrust case must begin by considering the remedy. Anticipating the appropriate fix is like having an exit strategy in battle. Court injunctions that prohibit a specific behavior or action are easier to obtain, but they may also accomplish less. ‘Structural’ relief, such as a breakup, requires proof of conduct that only a structural change can fix, as well as proof that the new structure will be better. The recent platform monopolization cases raise a recurring issue in antitrust law: creating the right remedy is often more difficult than establishing unlawful conduct. Most cases, like Amazon and Facebook, are intended to restore competitive behavior in a market that is presumed capable of sustaining competition. The market itself determines the number of competitors. Often, as in the 2001 Microsoft case, injunctions are the most effective remedy, and they have the best track record. A good injunction should remove unreasonable obstacles to user or supplier switching or multihoming, and that may be all that is needed.

In DOJ antitrust suits, the federal courts are empowered to ‘prevent and restrain’ antitrust violations. That is all the statute says. This gives the courts broad power to design a remedy—limited by the requirement that it must first find a violation. The Microsoft decision elaborated that an antitrust remedy should seek to [1] ‘unfetter a market from anticompetitive conduct,’ [2] ‘terminate the illegal monopoly, [3] deny to the defendant the fruits of its statutory violation, and [4] ensure that there remain no practices likely to result in monopolization in the future...’ That court then found that a ‘breakup,’ or structural decree, was not warranted. The parties settled on relief consisting heavily of injunctions that made it easier for Microsoft’s trading partners to switch to other firms.

Cases like Google Search suggest a more formidable problem: the possibility of natural monopoly, which some believe are genuine possibilities but others dispute. Search quality is driven mainly by two things: data scale and the quality of the search algorithm. Search engines with a large user base have an advantage over smaller ones. If data scale dominates continuously, search may be a ‘winner-take-all,’ or natural monopoly, market. On the other hand, if the quality of the algorithm is the driving force and the relative advantages of scale decline with size, then competition among multiple search engines may be possible.

A natural monopoly market has room for only one firm even if it does nothing anticompetitive. Platforms become natural monopolies when they are more attractive as they become larger and effective product differentiation is difficult to achieve. For example, while Facebook is a social network that benefits from larger size, its competitors, such as X (Twitter) and TikTok, have different features and appeal to different participants, enabling competition. That is also true of Amazon, which competes with Walmart and Target in e-commerce. Search is a different matter because effective product differentiation is difficult to maintain. In that case, a larger firm is likely to have advantages over a small one up to the point that the market is saturated.

The Google Search court may have to make a fact-finding on the question of whether data scale makes search a natural monopoly. Even judicial fact finding does not remove all doubt, however. Technology changes over time may affect natural monopoly status as well. For example, the telegraph was a hard-wired natural monopoly, eventually dominated by Western Union, until the telephone replaced it. Telephone’s landline monopoly, dominated by AT&T, was later replaced by competitively structured wireless. If any natural monopoly conclusion is less than ironclad, a remedy should restrict unreasonable impediments to switching or multihoming. Customers should be free to switch, and thus to avail themselves of any competition that the market makes available. It should also be clear, however, that removing such impediments will help preserve competition in a market that is capable of sustaining it. It will not create competition in a natural monopoly market.

In August 2023, the court approved but pared down the Google Search case. What remains is mainly a challenge to payments that Google pays other firms to make Google the ‘default’ search engine on their devices and browsers. If these payments are found to be unlawful, one obvious remedy is an injunction prohibiting them and leaving each device maker and browser to select its own default. A slightly broader remedy would give the initial choice of default to the customer, such as the user ‘choice screen’ that the European Commission ordered in 2018 for new Google Android devices. These alternatives produce loud objections from rival search engines that they are too little, too late. Google Search, with its ninety-plus percent market share, already has an established lead. Indeed, the effect of the EU choice screen has been nearly nonexistent: more than 97% of customers continue to select Google as their preferred mobile search engine. If a superior search technology should come along, however, a default or other impediment to switching could limit or delay its development.

Likelihood of natural monopoly also places important limitations on divestiture as a remedy. Forcing Alphabet to sell off Google Search simply transfers the monopoly to someone else but does nothing to reduce the monopoly power of Google Search itself. It might remove some ‘leveraging’ issues. For example, Google is currently the default search engine on Android phones, and separate ownership might reduce the incentive to do that. If Search favors YouTube, another Alphabet asset, separate ownership might eliminate that. More likely, however, a search result favors a particular site to the extent that a particular searcher’s own history favors it.

How about a breakup? Even if it were technically feasible, a structural breakup of a natural monopoly search engine would be a disaster. It would impose two or more suboptimal products on consumers. This is not a stable situation, and it would immediately deteriorate the quality of search for both of the resulting products. One alternative, famously suggested by Harold Demsetz and critiqued by Oliver Williamson, is to leave the natural monopoly intact but make firms bid for the right to be the monopolist, with price terms as the principal bid item. It’s hard to see how that could be made to work in the case of Google Search.

Other possible remedies stand a better chance of succeeding. One such possibility is an ‘interoperability’ decree that requires Alphabet to share, or license, its database to rivals. While that solution would not break up the data monopoly, it might create ‘last mile’ competition in consumer platforms and sales to advertisers. Rivals could use alternative search algorithms, competing for usage and advertisers while sharing a common database. By analogy, travel agents sell airline tickets out of a common inventory of flights. While the agents cannot control ticket prices and airline schedules, they compete with each other on commissions and related services.

Another promising remedy is to leave Google Search’s structure intact, but change its internal decision making processes to create competition within the platform. The history of antitrust includes many cases in which a single monopoly ‘platform’ was operated and sometimes even owned by numerous competing participants. The earliest was the Terminal Railroad case (1912), in which a ‘gateway’ railroad transfer terminal at the Mississippi River was a single holding company owned by several firms which included railroads, bridges, ferries, haulers, and loading and storage facilities. Another was the Chicago Board of Trade, an Illinois corporation whose 1,800 shareholders were also its traders. Yet another was Associated Press, a wire service for the sharing of news whose newspaper owners were members of a New York Corporation. More recently, the American Needle case involved a corporation that accepted and licensed out the intellectual property rights of its member owners, NFL football teams. While the structure and operations of these various organizations were diverse, in every case the structure served to maintain competition within a single organization while permitting it to function efficiently as a single unit.

In order to succeed as a competitive antitrust remedy, such a board must satisfy one legal and one economic condition. Legally, it must be constituted as separate actors in order to meet the multiple actor requirement of Section One of the Sherman Act. That creates more aggressive liability standards as well as weaker market power requirements for joint action. As a single entity, Google, Amazon, or other platforms are subject only to Section Two of the Sherman Act, whose requirements are very strict. Unilateral price setting is almost always legal unless it is predatory, but price setting by agreement faces much closer scrutiny and can be unlawful per se. A unilateral refusal to sell to someone or represent its product is rarely unlawful. On the other hand, a ‘concerted’ refusal to deal can be evaluated under a much more aggressive standard. That was the principal issue in the AmericanNeedle case, which held that the Corporation’s grant of an exclusive license to make NFL-logoed headgear should be treated as an agreement among its owners (the NFL teams) rather than as the unilateral conduct of a single firm. Each team had its own separate business whose interests were at stake.

A decision-making board comprising Google itself, representative users, advertisers, and others with a business interest in Google Search would be treated as an association of separate actors. This is different from, say, the CEO, two vice-presidents, and the head of accounting discussing the price of a new car. Each of these is a full-time employee of the manufacturer and with no independent business of its own.

The economic condition for such a board is sufficient diversity of actors to prevent anticompetitive collusion or exclusion. Too many of these organizations have a lopsided membership that favors certain groups and excludes others. For example, a local real estate organization is ordinarily composed of licensed brokers who represent both buying and selling agents, but not the clients themselves. They are tempted to fix commission rates because higher commissions benefit agents on both sides, while injuring nonmember home sellers and purchasers. To function competitively, these organizations should also include homeowners, home purchasers, and even financing institutions. As another example, the North Carolina Dental case involved a dentist-dominated association that controlled the practice of dentistry in that state. Six of its eight members were dentists, one was a dental hygienist, and one was designated as a ‘consumer.’ The hygienist and consumer representatives did not participate in fashioning an anticompetitive rule that only dentists could whiten teeth. Another example is the NCAA, which has 1,200 members but all of them are ‘employers’ who lacked the incentives to compensate student athletes properly, as the Supreme Court concluded in the Alston case.

Alternatively, physician hospital boards can behave far more competitively when their membership includes a full range of providers as well as patient interests. A hospital staff admissions board composed entirely of anesthesiologists would be tempted to engage in anticompetitive actions for the benefit of anesthesiologists. But if the decision makers also included surgeons, who sell a complementary product, patients, the hospital itself, and insurers, those anticompetitive incentives would be suppressed in favor of more competitive outcomes.

A competition-reinforcing board for an entity like Google Search should include searchers, advertisers, and other market participants who have an independent interest in search quality and product pricing. The individual members of such a board need not be shareholders, but they must be composed of individuals with final authority over the relevant elements of production, sales, distribution, and use. Remaining restraints could be evaluated under Section One of the Sherman Act.

Collaborative control organizations are not the best remedy for every structural antitrust problem. They can work when the social cost of breaking up a firm is very high, but alternative remedies such as injunctions seem too weak. It bears repetition that the remedies suggested here are intended to reinforce competition. They tend to produce higher output and lower prices. As such, they are good for customers, suppliers, including labor, and others who benefit from a more competitively behaving firm. They may not be good for competitors, who would prefer that their rivals charge higher prices or provide lower quality.

Herbert Hovenkamp is the James G. Dinan University Professor, Penn Law and the Wharton School, University of Pennsylvania.

Share

With the support of