Erik Hovenkamp and A. Douglas Melamed discuss what Judge Amit Mehta got right and wrong in his remedy decision in the Google Search antitrust case.
Judge Amit Mehta’s decision is careful, detailed, and thoughtful. With one exception regarding the scope of the prohibition on Google’s illegal conduct, Judge Mehta seems to have done a good job tailoring the remedy “to fit the wrong creating the occasion for the remedy.” In this article, we reiterate the purposes of antitrust remedies, review inconsistencies in Judge Mehta’s remedy terminating the unlawful agreements Google made with partners to establish its search engine as the default option, discuss the data-sharing remedies ordered by Judge Mehta, and explain why we think his decision not to order a breakup was correct.
The purposes of antitrust remedies
Antitrust is law enforcement. In a civil enforcement action like the Google Search case, equitable remedies serve two basic purposes: to terminate the illegal conduct and restore competition that was injured by that conduct.
The first remedial purpose required Judge Mehta to terminate the conduct that he found to be illegal in his 2024 ruling that found Google liable and to prevent its recurrence. For that, the remedy should both require the defendant to cease the illegal conduct and prohibit the defendant from engaging in different but substantially identical conduct.
Courts have described the second remedial purpose in different ways, referring, for example, to “cur[ing] the ill effects of the illegal conduct” and “deny[ing] the defendant the fruits” of the violation. In context, it is clear that these varying descriptions mean that the court should identify the harm to competition caused by the illegal conduct and design remedies intended to undo that harm.
In his 2024 ruling, Judge Mehta found that Google had maintained its monopoly in web search using agreements that make Google Search the default search engine on most browsers and Android devices. (For a short summary of that decision, see this ProMarket piece.) Judge Mehta found that those agreements impaired the opportunities of rival search engines to compete, denied rivals access to data needed to compete effectively, and reduced the incentives of actual and potential rivals to invest and innovate in internet search. Judge Mehta did not find that, absent the illegal agreements, the market structure would have been materially different or there would have been new entrants into the search market. But he did find that Google’s conduct was anticompetitive and “reasonably appears capable” of contributing to Google’s dominance in search.
As the DC Circuit explained in the seminal 2001 Microsoft case, the “[m]ere existence of an exclusionary act does not itself justify full feasible relief against the monopolist to create maximum competition.” Instead, remedies intended to restore competition should be focused on the specific harms to competition found to have been caused by the illegal agreements. Other courts, including the Supreme Court, have said that the “scope of the remedy must be proportional to the scope of the violation” and must be “tailored to the harm suffered.”
Termination of the unlawful conduct
The most basic and necessary function of antitrust remedies is to proscribe the conduct deemed illegal. While normally straightforward and uncontroversial, this proved to be the most problematic aspect of Judge Mehta’s remedy decisions.
The antitrust concerns raised by Google’s default agreements depend in large part on their breadth. Google’s default agreements covered 50% of all domestic search queries in the U.S., with an additional 20% covered by Google’s self-implemented Google Search default for users that downloaded its Chrome browser. As with other vertical restraints that limit how products are sold, such as tying or exclusive dealing, the breadth of the restraint determines the extent to which rivals are “foreclosed” from important trade opportunities. Judge Mehta found that Google’s agreements were illegal in large part because they substantially foreclose rivals from default search engine distribution, undercutting their ability to compete and chilling incentives for entry and investment. Termination of Google’s illegal conduct would thus require, at minimum, a significant restriction on the breadth of Google’s default agreements.
But Judge Mehta did not impose such a restriction. Instead, while he prohibited Google from entering into exclusive default agreements with web browsers, original equipment manufacturers of mobile phones using Google’s Android operating system, and wireless carriers, his decision permits Google to continue to offer substantial payments conditioned on default status for its search product. Judge Mehta restricts those payments only by limiting the duration of Google’s default agreements to one year and prohibiting payments conditioned on Google being the exclusive default search engine. Those restrictions leave Google free to pay for very broad defaults. For example, nothing in the decision would seem to bar Google from paying Apple to be the default option for, say, 95% of Safari users, even though this would differ only trivially from an “exclusive” default.
It would have been more effective for Judge Mehta simply to prohibit Google from making payments conditioned on its search engine being the default, or to limit the breadth of its paid defaults. But the remedy decision imposes no such limit on the aggregate scope of Google’s default agreements.
Some of Judge Mehta’s explanations for the narrowness of its restrictions on payments for default status seem inconsistent with his liability decision last year. For example, Judge Mehta suggested that Google should be allowed to retain its defaults because consumer surplus would fall if smaller search engines were able to obtain greater default distribution, given the higher quality of Google Search. But Judge Mehta’s liability decision was predicated on the conclusion that Google’s agreements impaired competition precisely because they cut off rivals from the default distribution channel and thereby undermined their ability to compete, improve, and grow—developments that would have enhanced competition and consumer surplus in the long run. That makes sense, because economic research shows that allowing small firms to employ modest vertical restraints on distribution can be essential to long-run competition in network industries. When such restraints are used only by the dominant firm, they can slow or even prevent the development of meaningful competition.
Judge Mehta also asserted that prohibiting Google from paying for default status would cut off lucrative revenue to its partners and thus result in weaker browser competition and higher smartphone prices. But Judge Mehta rejected those very arguments at the liability phase, in part because they were speculative and “the evidence is thin.” Moreover, although Judge Mehta did not restrict Google’s ability to offer revenue sharing not conditioned on default status, he did not consider whether this less restrictive payment scheme would have been sufficient to allay its concerns about downstream competition.
In light of emerging generative artificial intelligence products, Judge Mehta remarked that “allowing Google to continue making payments is more palatable now than when the liability phase concluded.” We agree that such a determination could potentially justify a decision not to ban Google’s defaults. But to justify setting aside the court’s own findings of anticompetitive harm—findings extracted from a lengthy trial and immense evidentiary record—this determination would have to be grounded in significant new evidence. Judge Mehta does not appear to cite any economic evidence for its assertion that Google’s agreements are no longer anticompetitive.
Since his decision, Judge Mehta has spoken publicly on the importance of being “cautious” when crafting antitrust remedies. While we share this concern, it is properly directed at competition-restoring remedies that go beyond terminating unlawful conduct. If “caution” means allowing a defendant’s illegal behavior to persist, it defeats the purpose of finding liability in the first place and undermines the law enforcement function of antitrust law.
Restorative remedies—data sharing
Judge Mehta found that Google’s default agreements diverted user search queries and the associated data rivals needed to compete effectively. Search engines rely on data to improve their algorithms, so the diversion of queries and data likely had the effect of reducing rivals’ quality. Some kind of data sharing is thus a conceptually appropriate remedy. Data-sharing remedies are especially important in light of the narrowness of the restrictions Judge Mehta imposed on payments for default status because those restrictions will probably do little to restore competition harmed by the diversion of queries and data from rival search engines.
The purpose of data-sharing remedies is not just to replace data diverted from rivals; it is to address the harm to competition caused by the weakening of rival search engines. Judge Mehta’s order requiring Google to share its search index with rivals is thus also an appropriate remedy, even though the default agreements did not literally divert search indexes from rivals. The search index is important to the quality of the search results, and its quality depends in part on the volume of queries received by Google and its rivals.
Data-sharing remedies raise a host of questions. These include the kinds of data that should be shared, how such remedies could be implemented, whether they would be effective and not unreasonably costly, how they might affect the incentives of both Google and its rivals, and what risks they create for privacy or the interests of third parties. Judge Mehta seems to have done a careful and thorough job of addressing those questions, modifying the proposed remedies to take them into account, and relying on a technical committee to address issues not suitable for judicial resolution.
Restorative remedies: divestiture
The DOJ asked Judge Mehta to require Google to divest its Chrome browser and to order a contingent future divestiture of Google’s Android operating system. There is nothing wrong in principle with these requests. It has long been clear that courts may require “any reasonable method of eliminating consequences of the illegal conduct.” Still, Judge Mehta rightly rejected these requests, for a number of independent reasons.
First, as noted above, the causal nexus between Google’s misconduct and its market dominance was not very strong. Judge Mehta found anticompetitive effects but no concrete impact on Google’s dominance. Those findings were enough for liability, but not for the requested divestitures. This follows the court’s opinion in Microsoft, where though an “edentulous test for causation” was sufficient for determining Microsoft’s liability in the case, Microsoft’s concerns about the sufficiency of the proof of harm to competition had “more purchase in connection with the appropriate remedy issue.”
In addition, there is little connection between the DOJ’s proposed divestitures and the illegal conduct or the harm Google’s agreements were found to have caused. Judge Mehta found Google’s exclusive default agreements to be illegal, but he did not question the lawfulness of Google’s decision to make its search product the default on Chrome. Forcing Google to divest Chrome would presumably mean that it could no longer be assured of default distribution on Chrome, but there is little reason to think that the new owner of Chrome would be less likely than other search access points to choose Google Search as the default. Restricting Google from paying for default status or prohibiting Google from making its search product the default on Chrome would be a more effective means of restoring queries and data to Google’s rivals than requiring Google to divest Chrome.
The requested divestitures would also be very costly and disruptive, and there is a good chance that they would not even succeed on their own terms. It is not uncommon for divestitures to fail. Even when they do not fail, they often destroy valuable synergies between the divested businesses and other assets of the selling party, especially when the divested businesses were internally developed and not themselves acquired from third parties. Judge Mehta found that Chrome “does not run as a standalone business,” “depends on Google for a host of administrative functions,” and “is deeply reliant on Google’s ‘hyperscale’ technical systems and infrastructure.” A forced divestiture of Chrome would end these synergies. While divestiture remedies can be appropriate in some cases, they should be required only if they are reasonably necessary to unwind an illegal acquisition or to remediate harm to competition that was found to have been caused by the illegal conduct. The proposed divestitures are neither.
In an exclusion case, a vertical divestiture (one that spins off a complementary product) is most likely to be necessary when a defendant’s conduct is unilateral and hard for a court to police. (For example, that was the situation in U.S. v. AT&T, the case that led to the breakup of AT&T.) But that clearly does not apply to the Google Search case, which centers entirely on contracts, not unilateral conduct.
Finally, the requested divestitures would be accompanied by substantial risks of unintended consequences. The buyers would likely be huge tech companies that would anticipate possible synergies between the divested assets and their own businesses, which will no doubt evolve over time. They might well, for example, have incentives similar to Google’s to engage in some kind of “self-preferencing” or otherwise to use the acquired businesses for anticompetitive, strategic purposes.
Judge Mehta’s decision not to break up Google has already provoked critical responses from antitrust populists. The critics want to punish Google, which is not a proper remedial purpose in a case like this, and to solve what they perceive as the problem of Google’s dominance over general internet search and other aspects of digital commerce. But antitrust remedies are intended to solve only the problems created by the illegal conduct. If a remedy sufficient for that purpose is not also sufficient to solve broader problems, it is because the conduct at issue in the case was not a major cause of those problems.
Authors’ Disclosures: The authors report no conflicts of interest. You can read our disclosure policy here.
Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.
Subscribe here for ProMarket‘s weekly newsletter, Special Interest, to stay up to date on ProMarket‘s coverage of the political economy and other content from the Stigler Center.
