Recent antitrust interventions have put forward behaviorally informed theories of harm. However, they have adopted a deterministic model of behavior, missing the nuances that allow behavioral economics to provide a richer picture of people’s conduct. The recently concluded Google trial, grounded on the stickiness of defaults, is a good example. A more careful application of behavioral economics would have shown how Google’s purchase of default search engine status was a part of a broader monopolization plan. It would also show why the dominant remedy, forced choice, would have negligible effects.
The ascent of behavioral economics has led to its widespread application in business and government. In consumer finance, it informs strategies to enhance financial decision-making, and in food law it has helped craft more user-friendly food labels. While its application to antitrust law has taken longer, behavioral economics is now at the forefront of competition policy. It drives critical interventions aimed at regulating the complex challenges posed by Big Tech and the evolving digital economy.
A notable example is the recently concluded Google trial concerning its agreements with several search engine distributors (including Apple) to set Google Search as their users’ default search engine. In this trial, a critical theory is at play: default agreements are de facto exclusive dealing because people rarely change their devices’ default settings (Google has an about 90% market share of all online search in the United States). The de facto exclusive dealing theory—and related “tying” theories—proposes regulators force users to choose their default applications to foster competition. This approach has been embraced by European regulators, exemplified by the implementation of choice screens (i.e., pop-up windows that prompt people to choose an option within a set of alternatives) in seminal cases such as Microsoft and Android.
However, there is a deep paradox underlying these theories and their policy applications. One of the core principles of behavioral economics—and sound social science in general—is to observe people’s conduct in the real world to evaluate the validity of hypotheses and formal models. A critical examination of current interventions reveals they are based on untested assumptions and naive theories. They hinge on the belief that inertia is much more psychologically relevant than user satisfaction and that simply offering users a choice will meaningfully alter entrenched market dynamics. While regulators’ arguments that the status quo effect of default settings pose an anticompetitive obstacle are not as strong as they think, Google’s conduct still lessens competition in a way that violates antitrust law.
An empirical study of default status as a competitive harm
In two recent papers, I examine the impact of default status and forced choice remedies on competition. My research shows that while status quo effects exist, i.e. inertia can entrench the default setting, they are not as large as conventional wisdom assumes. Moreover, my work finds that when there is an application that most people think is the best, forced choice remedies merely lead most people to pick the option they normally use.
An application’s default position is only one of many variables that may influence the likelihood that consumers use it. Experiments are the gold-standard for identifying the causal effect of an intervention on an outcome. The great advantage of experiments comes from random assignment. For instance, a drug manufacturer may want to assess whether a drug makes people more productive. The company recruits 1,000 participants and gives the drug to 50% and a placebo to the other 50%. When researchers assign the treatment randomly, all the other variables that affect people’s productivity—observable and unobservable—should be balanced among the two groups (e.g., people’s IQ, caffeine intake, exercise habits). Then, the only difference between the average of the groups is the treatment itself, and if there is a change in productivity between the groups, the difference is attributable to the intervention.
We can follow the same procedure to assess the effect of an application’s default status and forced choice remedies. Ideally, we would gather a large group of people, ask (force) half to choose their application for a specific service (e.g., Bing, DuckDuckGo or Google Search for an internet search engine), and assign an application at random to the other half. Then, by comparing the market shares of each application among the forced choice and default conditions, we could assess whether the apps benefit from a status quo effect and how large such an effect is. Alternatively, we can create a fictitious scenario that closely resembles how people behave in the real world.
This is precisely what I’ve done. To test the stickiness hypothesis, I designed a survey-experiment and collected data on an online survey platform. The participants had to search online for the answers to two trivia games. They were incentivized to find the correct answers with a bonus. Half were asked to choose the application they wanted to use. The other half were assigned to an application by default. In the first experiment, there were two parts: one about popular culture and another about weather forecasts. The first tested status quo effects for search engines and the second for weather applications. The choice of search engines was due to their prominence in today’s antitrust landscape. The weather applications part was to have a contrast with a product for which people have a less strong preference. The second experiment focused on search engines exclusively. One of the advantages of my design is that we just need to assume that people wanted to minimize the time they spent taking the survey and maximize the probability of winning the bonus to have confidence in the results.
Biases are always comparisons. A status quo effect refers to the difference in probability that a consumer will use an application when it is the default choice versus when the consumer is compelled to choose. My analysis found status quo effects close to 50% for all search engines but Google and around 30% for weather applications. This means that default status increases an application’s probability of use in a short trivia game by 30-50%. Notably, Google was the only search engine that did not benefit from a status quo effect (i.e., most people selected it when they were forced to choose their preferred application and most people stuck to it when they were assigned to Google by default). The most important empirical takeaways for antitrust policy are that (i) default positions do influence people’s probability of using most applications, but not as much as conventional wisdom assumes, and that (ii) forcing people to choose their default search engine does not lower Google’s market share.
A common criticism of experimental research is its generalizability to the real world. To properly assess this concern, one must identify the key elements of the experimental task and hypothesize how it resembles people’s actual choice environments. The primary limit of my study design is the participants’ short interaction with the task (up to five questions per application). In fact, because a substantial part of the participants quickly switched to their preferred application despite the trivia game being such a low-stakes task, it is sensible to assume that the status quo effect would be even lower in the real world where more people would take the time to switch to their preferred application.
Nevertheless, my work uses observational data to complement the experimental part. It assesses the effect of a 2014 agreement between Firefox and Yahoo by which the latter became Firefox’s default search engine in the U.S. Using other developed and developing countries to build a synthetic counterfactual that shows how Yahoo’s market share would have evolved in the U.S. had the agreement not taken place, my analysis finds that Yahoo’s default status did increase its market share. The effect was close to 2% considering the whole desktop search market. Since Yahoo’s market share was close to 10% in desktop search, Yahoo benefited from a status quo effect close to 20%. However, less than a year after the agreement took place, its impact was not statistically different from zero.
Legal implications: defaults as synergistic anticompetitive conduct
A careful assessment of consumer behavior is not only of academic interest. So far, enforcers have assumed that the share of the market a default agreement forecloses equals the number of customers affected by the agreement (e.g., if Apple users equal 50% of weather apps users, a default agreement between Apple and AccuWeather would foreclose 50% of the market). However, since not everyone sticks to the defaults, a proper foreclosure analysis should discount the share of the market that is not affected by an application’s default status. I refer to this implication as the default multiplier. It has critical consequences for exclusive dealing claims, as I discuss below.
In the U.S., plaintiffs may bring an exclusive dealing case on the grounds of three main legal provisions. The first one is Section 1 of the Sherman Act, which bans anticompetitive agreements. Most courts assessing Section 1 claims require market foreclosure of 40-50% to establish an antitrust violation. If we assume default agreements trigger an exclusivity of 50%, in practice only agreements covering almost 100% of the market would meet the foreclosure bar. The second provision is Section 3 of the Clayton Act, which is the only one that refers to “exclusivity” contracts expressly. This standard is the strictest one; it relies on a qualitative assessment for which the foreclosed share of the market is only one of many factors to be considered. Yet many courts interpret the “exclusivity” prerequisite strictly, meaning that a contract that is only partially sticky would not meet the legal standard either. The third provision is Section 2 of the Sherman Act, which condemns monopolization. Section 2 only applies to defendants having monopoly power. This may appear as a substantial limitation. However, the most relevant cases that have dealt with defaults in the U.S. and the European Union have dealt with monopolists intending to preserve their market dominance through the use of defaults, which have been a mere part of a broader monopolization scheme.
The 2001 Microsoft case is a good monopolization example. In the late 1990s, Microsoft did not only preinstall its internet browser on its operating system and preset it as its users’ default. It also made Internet Explorer virtually undeletable—in a time when hard drives had much more limited capacity—and its code opacity impaired the development of competing applications. Notably, the DOJ’s settlement approved by the D.C. Circuit Court didn’t prohibit Microsoft from “commingling” Internet Explorer (IE) with its operating system. Instead, it required Microsoft to make its code accessible, which was aimed at supporting the development of rival applications within its operating system. This part of the court’s decision was crucial in facilitating the entry of new competitors into the internet browser market. The significant decline in IE’s market share is largely attributed to the emergence of competitors like Firefox and Chrome.
The Google case
The current trial against Google is another good example of using defaults as synergistic practices that strengthen a broader monopolization plan. Google pays billions of dollars to several tech firms to secure the default search engine status on their browsers (including Apple’s Safari). The part of the case that deals with Google Android devices is almost a copy of the Microsoft case (Google not only presets its search engine as people’s default but also makes a set of applications undeletable and forbids the preinstallation of competing applications, among other restrictions). However, the agreement between Google and Apple does not prevent the latter from preinstalling other search engines on its devices. This could suggest that the default status is important enough to be worth several billion a year (this is what one of the trial’s economic experts merely assumed based on anecdotal evidence and studies—some confidential—about status quo effects in other domains). But Google’s contracts not only limit its competitors’ distribution, they also deter Apple from entering search and advertising, thereby reinforcing its monopolistic hold. The default status—together with the contractual restraints imposed on the Android licensees—and Apple’s lessened incentives to enter the market (due to Google’s large payment) should provide strong indicia of a monopolization scheme that unreasonably restrains competition.
A rule of reason assessment would require Google to show an objective justification for its conduct. While most of Apple’s users get their preferred search engine by default (i.e., Google Search)—which means the predetermined setting minimizes people’s switching costs—forcing people to choose their preferred default would achieve the same efficient matching and break the potentially collusive alliance between Google and Apple at a negligible cost (i.e., just one click per user). Choice screens have been ineffective in leveling the field among search engines in Europe. However, invalidating the RSA and mandating Apple to display a choice screen would achieve an efficient matching precisely because people tend to choose Google when forced to choose their default. And Apple would no longer be deterred from competing with Google. This fuller picture shows much more clearly why the revenue-sharing agreement (RSA) between Google and Apple is anticompetitive. A careful analysis of the “de facto” exclusive dealing theory and the current regulatory efforts to enhance competition in digital markets reveals that conventional wisdom exaggerates the influence of default positions and overlooks people’s limited interest in engaging with choice screens. Behaviorally informed theories usually provide much less generalizable policy prescriptions than economic theories based on strict rationality assumptions. An assessment of the suppositions guiding current interventions is essential for guiding sound policymaking and law enforcement. When considering that defaults are not always sticky and that defaults may stick because they are what people prefer, it is evident that Google’s conduct goes beyond exploiting people’s inertia.
Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.