At the heart of the United States Google Search case is the monopolizing effect of Google securing for its own search offering the status of default search engine on a web browser, such as Safari, Chrome, or Firefox. The authors review the behavioral economics and empirical evidence of this effect and suggest several conduct and structural remedies to open up the search market to competition.
The United States and several states have accused Google of unlawfully monopolizing the U.S. markets for general search services. The crux of the matter lies in Google’s deals with a variety of parties. These include original equipment manufacturers (OEMs) of Android devices, such as Samsung; its mobile ecosystem rival, Apple; web browsers that enable search engines such as Firefox; and cell phone carriers such as Verizon.
These arrangements effectively make Google the exclusive default search engine on Android and Apple phones; establish Chrome as the default browser on Android; and ensure Google’s default search engine status within Firefox.
The pivotal question in relation to these default arrangements relates to their impact on consumer choices and competition. In a nutshell, the plaintiffs argue that, by making Google the exclusive default search engine at all search access points, these agreements create a huge barrier for rival search engines and stifle competition and consumer choice. Google contends users can freely change their search default, or use an alternative search app or browser, and that its strong market position simply reflects that it offers a product superior to those of its rivals, in great part because it invests more in quality.
Examining these claims and counterclaims requires delving into behavioral economics. Do consumers stick with the default search engine and the most prominent browser, as the plaintiffs suggest? Or do they simply choose the product that best suits their preferences? Or is the truth somewhere in between?
There is extensive behavioral economics literature documenting substantial default effects in both laboratory and field experiments. For instance, defaulting people into organ donation schemes leads to 99% participation, compared to just 12% when people need to register. Similarly, when consumers are asked whether they would like additional coverage on their auto insurance, requiring them to opt in results in much lower levels of participation (20%) than setting this as the default and allowing them to opt out (75%). A variety of possible drivers of these default effects have been identified. For example, people may not realize they have a choice, they may be deterred from opting out due to the effort involved or due to loss aversion, or they may view the default as a recommendation.
In this article, we explore what the evidence presented in US v Google, which the U.S. District Court for the District of Columbia heard last fall, teaches us about the force of default effects in the online search market. We also discuss the remedies that may be suitable, in the presence of such default effects, if the court were to determine that Google has engaged in unlawful monopolization.
Evidence on the Value of Defaults
Google’s apparent payment to Apple of over $18 billion per year for default search engine status on the iPhone would seem to be a significant piece of evidence for the power of defaults. This comprises a large fraction—apparently 36%—of all search advertising revenues earned on Apple devices. It is hard to find an explanation for Google accepting to pay so much unless it believed that defaults play a critical role in shaping user choices.
In fact, evidence from the plaintiffs in this case shows Google’s managers’ awareness of the potency of defaults, prompting Google to adopt a “pay to play” strategy. In his testimony for the U.S. Department of Justice, Caltech professor Antonio Rangel highlighted internal documents from Google indicating that such choice frictions to switching from default search engines are indeed highly significant. MIT professor Michael Whinston, also a DOJ expert witness, drew the court’s attention to internal documents from Microsoft and Apple in which they, too, acknowledge Google’s default position as crucial.
Overall, then, the evidence underscores Google’s acknowledgment of the significance of default effects, shows the agreement of industry peers, and sheds light on the substantial impact defaults have on consumer behavior and choices.
Behavioral Impact of Defaults on Consumer Choices
Examining the impact of defaults on consumer choices also reveals compelling evidence, even though exact data are often redacted in the case documents. Apple Maps is a well-known example. After becoming the default application, Apple Maps became the leading map application on iPhones, despite initial users’ complaints about quality.
Similarly, Bing has higher market share on Windows desktops, where it serves as the default search engine for Microsoft’s Edge and Internet Explorer browsers, than it does on mobile devices with Google as the default. Bing’s search engine market share, when analyzed browser by browser, is below 5% on Chrome, Firefox, and Safari (where Google is the default) but exceeds 50% on Internet Explorer and nearly 80% on Edge (where Bing is the default).
Other instances are cited in the expert reports too, all of which emphasize the significance of defaults for consumer choices. For instance, the introduction in Russia of a choice screen, which asks mobile users to select a default search engine when setting up their phones, has led to a fall in Google’s market share from over 60% to around 40%. Likewise, the share of Google Search is 16 percentage points higher on Mac OS devices, where it is the default, than on Windows PC devices, where it is not.
There is further evidence, telling a similar story, in the EU Commission‘s 2018 Google Android decision. Table 10 in the decision compares use of Google Apps on Android and iOS devices, and shows a systematic effect of pre-installed defaults. For example, Google Maps is used by 59% of Android users (for whom it is the default maps option), but only 22% of Apple iOS users. Table 11, which compares search query market shares between 2014 and 2017 in the EU, across Google Android and Windows Mobile devices, finds an even more dramatic effect. In 2017, Google Search was used for 90-100% of search queries on Android phones (where it is the default) but just 10-20% on Windows Mobile devices.
While not all of this evidence is directly linked to Google’s default status for its search engine, the default effects in that context are likely to be strong. Whinston estimates that Google would lose as much as 33% of all search queries if rivals were assigned the default position instead.
In his testimony, Rangel described the thought process consumers would need to go through in order to choose an alternative search engine. Consumers may be unaware that there exist other search engines and that they can switch, especially if this option is not salient. Even if consumers are aware, they need to identify an alternative search engine that they would prefer. In this context, Google’s default status may be perceived as a recommendation, which consumers are inclined to accept. Finally, even if they decide to switch, the relevant steps may be far from straightforward, and we know that people tend to delay or avoid complex tasks.
Default Effects and Other Market Dynamics Precluding Competition
In his testimony, Whinston also discusses the interaction between these default advantages and another fundamental feature of search markets: economies of scale in data. With more data, a search engine can improve its algorithm and deliver better search results. While the extent of these scale economics is empirically unmeasured, conventional wisdom suggests that they eventually are diminishing as the search engine handles more queries. Independently of whether Google has a market share of 90% or 96%, it will be able to deliver high quality results. The same, however, is not true of its rivals. Whether Bing has a 2% or 8% market share significantly impacts its quality and its competitiveness.
Google’s monopolization of default search positions prevents its rivals from securing the critical search queries needed for a high-quality user experience. Even small reductions in Google’s market share matter, because the corresponding increases in rivals’ market share could allow for a positive feedback loop in which the higher quality from more data further increases rivals’ market share, which further increases their quality, and so on. The Apple Maps example shows how default status on a significant platform can transform a rival into a substantial competitor.
Additionally, it is currently unrealistic for the price of search to fall below zero. This further strengthens the role of default effects in stifling competition. New entrants frequently employ lower prices to gain market share in the face of popular incumbents. Their discounts act as a salient inducement for customers to try out the new service, despite its unknown quality. In internet search, because prices are already zero and can’t go negative, this entry strategy is unavailable, leaving consumers more likely to stick with the default option.
The Counterfactual to Google’s Default Status
To properly assess the impact of defaults, it is crucial to consider the relevant “counterfactual”—what would have happened “but for” the alleged anticompetitive conduct. In considering this, a more positive aspect of defaults is relevant. Defaults simplify choices for consumers, providing for a smooth consumer journey and a well-functioning device. For instance, consumers do not wish to choose their preferred search engine every time they open their browser.
As such, assuming a “but for” world without any defaults would not be accurate. On the other hand, it should equally not be assumed that the counterfactual would simply involve all consumers choosing Google’s browser and search engine anyway, and thus that the restrictive agreements between Google and its partners have no material effect.
First, although Google is popular and may be considered by many to be the highest quality search engine, it may not be the best fit for all consumers. For instance, some will prioritize privacy, reduced advertising, or environmental considerations. A counterfactual world in which consumers were encouraged to choose their own search default might thus imply a weaker market position for Google.
Second, Google’s current position reflects past restrictions. Identifying the appropriate counterfactual requires us to consider the likely consumer preferences in a world where these restrictions had never been imposed. Assessing this is complex, since it depends partly on how the OEMs and other third parties would have acted. However, rivals would likely have gained greater market share, improving their own access to data, and thus enabling them to enhance their own search results, making them more effective alternatives to Google. In his testimony, Whinston describes how Google responded to the European Commission’s Android decision, which entailed the use of choice screens on Android phones, by setting up a new quality initiative. This presumably reflected Google’s expectation that it would start to see greater competition from rivals.
In this scenario, entirely new search options might also have emerged, further weakening Google’s position. For example, Whinston discussed a proposed innovation that Samsung wished to introduce, through a partnership with Branch, a mobile software company, which would have enabled Samsung users to search inside Apps as well as on the web. Internal documents indicate that this was abandoned due to Google’s contractual restrictions. (This was corroborated in testimony by Branch’s founder, Alex Austin.)
Third, if these restrictions persist, they could stifle the development of more drastic innovations going forward. Google’s position could potentially be undermined by disruptive rivals using new business models or revolutionary new technology (such as generative AI), but these opportunities can only be seized if such rivals can attract customers.
A final argument that Google could potentially make is that scale economies are so extensive, that search is a natural monopoly and thus it is unrealistic to expect to see competition in the market at all; the best we might hope for is competition for the market. However, while there clearly are economies of scale, there is no evidence that these are so great as to imply a natural monopoly.
Potential Remedies
If the court determines that Google’s practices constitute unlawful monopolization, the persistent nature of default effects also raises challenges for implementing effective remedies. Below, we outline a number of potential conduct requirements. These would require monitoring and enforcement over time, as well as potential refinement as it becomes clear what works and what doesn’t. This task would ideally be carried out under specific regulation, as competition law processes are not typically well set up for such ongoing responsibility.
In the absence of such digital regulation, however, the court can impose remedies. These would ideally seek to restore the competition lost due to Google’s illegal monopolization, or at least open up the market to greater rivalry going forward. In doing so, the court should recognize that Google’s past conduct has reinforced its widespread brand recognition and consumer loyalty, especially given consumers’ tendency to stick with the status quo. It should also take into account the lack of data available to rival search engines, which would have enabled them to offer better search results.
At the same time, the court should ensure that imposed remedies do not negatively impact consumers. For instance, denying Google access to the data vital for high-quality search results, or imposing a low-quality search default, would probably reduce consumer welfare, at least over the short term.
A straightforward cease-and-desist order, like prohibiting Google from imposing pre-installed default status and exclusive positioning, would allow OEMs and browsers to select their own defaults or to present consumers an upfront choice. In this scenario, OEMs and browsers could still engage in non-conditional revenue-sharing arrangements with Google, and would have more flexibility, for instance to give exclusive default status to rival search engine providers. By fostering competition among search engines, OEMs may even be able to secure a larger share of search revenues, potentially reducing device costs for consumers. They would also be in a position to adopt innovative search options, such as the Samsung/Branch example above. Likewise, a novel, say AI-enhanced, search engine should be able to more quickly earn market share.
A key problem with this simple option is that it is unlikely to fully undo the harm arising from Google’s conduct: Google would continue to benefit from its installed base, algorithm, and brand. If OEMs and browsers auction off default positions, Google would likely win, given its current popularity, even if it can no longer require this contractually.
A second possible remedy might also prohibit Google from competing for defaults at all, exclusive or not. The downside of this approach would be that, if default status is awarded to an alternative search engine, some consumers may, by accepting the default, end up with a search engine that they would not have chosen if given an option.
However, this downside could be ameliorated with a third possible remedy, already adopted in the EU, namely mandatory choice screens. Google could be required to ensure that any device using the Google Android OS provides users an upfront choice screen allowing users to decide on their preferred search engine and browser at their first use of either service.
Choice screens are highly unlikely to provide a complete solution. Search engines are experience goods; that is goods whose quality can only be evaluated by using them. As a consequence, a choice screen, which usually presents little information other than a short advertising sentence, does not really enable consumers to reliably judge quality. In this context, Google’s strong brand recognition means that it may well be chosen by many consumers, even if this would not have been their choice in the counterfactual world where Google’s past conduct had never taken place.
Nonetheless, choice screens can likely help to enhance competition, alongside other remedies. Their impact will be heavily influenced by the way in which options are framed—the so-called “choice architecture.” Recent empirical analysis of two distinct choice screen remedies imposed on Google in different jurisdictions estimates a market share shift towards rival search engines ranging from just two percentage points up to seven percentage points. Ensuring that choice screens are as effective as possible will therefore require careful planning, testing and oversight. (We do recognize a difficulty with this solution: some consumers might choose a search engine they would find inferior to Google given full information. The Court has access to more data than we have to assess how big a problem this is.)
A fourth possible remedy would be to impose sharing of click and query data with rivals, again a remedy that has already been introduced in the EU by the Digital Markets Act. With access to such data, rivals would be able to train their own algorithms with a large sample of data and in that way compete with Google on a more equal footing. This remedy also protects consumers, since Google’s own access to data, and thus its own search quality, remains unimpaired. However, as pointed out by University of Maryland professor Douglas Oard, a further expert testifying for the DOJ, many types of user data are used to improve indexing of the web, such as spelling corrections and image responses. Sharing of click and query data will thus help even though it does not fully address rivals’ data disadvantages.
We have discussed the conduct requirements that, given the current state of our information, we believe would enhance competition and improve consumer welfare, but we recognize that they are unlikely to be a panacea and may also have some negative consequences. The court will have the difficult task of weighing the pros and cons.
Structural Remedies
Of course, a decision maker could conclude that the conduct requirements discussed above—even in combination—would not fully correct for Google’s past conduct, nor improve competition quickly or significantly enough. In this case, structural remedies should be considered. For example, a very significant component of a search engine’s fixed costs relates to the frequent webcrawling and indexing that must be done to keep the list of web content up to date. Such a resource is both valuable and highly costly to duplicate. It could be used by all search engines, and therefore is a good candidate for divestiture into an independent non-profit corporation, which would charge both Google and rival search engines for access. Alternatively, Google could commit to sharing its index with any interested search engine for a regulated low fee if, despite the ongoing regulatory intervention that would be necessary in this case, Google preferred that solution.
As some of us have discussed elsewhere, another example of a possible divestiture is the Google Android OS (including Google Play Services), the root source of Google’s market power over handset makers.
Overall, structural remedies such as these, perhaps alongside some of the conduct requirements discussed above, may well be more successful than conduct requirements alone at providing a decisive and enduring solution to competition in search.
Finally, while the remedies we describe here may not work immediately to generate a high quality search entrant, they can be viewed as opening a door, even if there is no one to walk through it right now. It is important for the incentive to innovate that the door be kept open. Without the access to end consumers that these reforms create, entrants and entrepreneurs have less incentive to invest.
Authors’ Affiliations and Disclosures:
Jacques Crémer: Professor of Economics, Toulouse School of Economics.
Amelia Fletcher: Professor of Competition Policy, Centre for Competition Policy and Norwich Business School, University of East Anglia, and Research Fellow at the Centre for Regulation in Europe (CERRE). Within the last three years, she has been a Non-Executive Director at the UK Competition and Markets Authority and has engaged in economic consulting in relation to antitrust litigation against Google but not in the context of search defaults.
Paul Heidhues: Professor of Behavioral and Competition Economics, Düsseldorf Institute for Competition Economics (DICE), Heinrich-Heine University of Düsseldorf. Within the last three years – in collaboration with E.CA economics – he engaged in competition consulting in the context of trucking and timber industries, and advised E.CA on work done by E.CA for Apple in the context of a competition case.
Gene Kimmelman: Senior Policy Fellow, Tobin Center for Economic Policy at Yale University and Research Fellow, Mossavar-Rahmani Center for Business and Government, Harvard Kennedy School. Within the last three years he has done antitrust and competition policy trainings for communications professionals, not involving tech sector companies.
Giorgio Monti: Professor of Competition Law, Tilburg Law and Economics Center, Tilburg University & Research Fellow, Centre for Regulation in Europe (CERRE). He is a member of the Supervisory Board for the Consumer Competition Claims Foundation in the Netherlands.
Rupprecht Podszun: Chair of Civil Law, German and European Competition Law at Heinrich Heine University of Düsseldorf, Director of the Institute of Antitrust.
Monika Schnitzer: Professor of Economics, Ludwig-Maximilians-University Munich & Chair, German Council of Economic Experts.
Fiona Scott Morton: Theodore Nierenberg Professor of Economics, Yale School of Management; Senior Fellow, Bruegel; and Research Associate, National Bureau of Economic Research. Within the last three years, she has provided economic consulting for corporate clients including Amazon and Microsoft on issues unrelated to Google’s search defaults, and for a number of government plaintiffs.
Articles represent the opinions of their writers, not necessarily those of ProMarket, the University of Chicago, the Booth School of Business, or its faculty.
