Steven C. Salop writes that only Google’s full divestiture of its Android operating system can avoid incentives on the part of Android and Google to preference Google’s apps, including its search engine, and stifle competition.
In my previous ProMarket article discussing Google Search remedies, I focused on conduct remedies. One provision focused on Google’s Android operating system, which Google uses to promote its search engine and other apps by requiring pre-installation from Android phone manufacturers. I made the point that if the court decides to mandate Google’s divestiture of Android, it still would be important also to mandate restrictions on the conduct of the newly independent AndroidCo. These would involve prohibitions of formal agreements or understandings with Google that involve tying, conditioning, discriminating, or other conduct favoring Google Search. It also would entail prohibitions on other exclusionary conduct that could foreclose or delay search rivals and competition. Absent such provisions, conduct that amounts to de facto re-integration by contract could sink the restoration of the competitive process.
The fact that similar conduct prohibitions would be needed—whether or not there is a divestiture—raised the question of whether it would be substantially easier for the court to monitor and enforce these restrictions when there is an independent AndroidCo created by a divestiture rather than Android remaining part of Alphabet. In that short article, I did not have the space to provide any potential explanations.
In this follow-up article, I explain why a divestiture is the superior approach and likely necessary for effective monitoring and enforcement of restrictions designed to prevent foreclosure conduct through Android that would disfavor competitors of Google Search and other Google apps that expose search access. I also discuss why a divestiture is superior to an intermediate remedy which would mandate structural separation of Android from Google, but with the Android subsidiary still being part of the Alphabet holding company
Inherent foreclosure incentives of a unified Alphabet
The basic explanation is straightforward. Absent divestiture, Android and Google Search would be two divisions within Alphabet corporation. The managers and executives of the Android division would have inherent and overarching incentives to take actions that maximize the profits of Alphabet, and thus the Google division as well as the Android divisions—and vice versa for Google division employees. While certain judicially mandated restrictions might limit certain actions, it is doubtful that the court can fully specify and detect the myriad ways in which the intent of the remedy can be evaded. And those managers and executives of the divisions would have the incentive (if not the duty) to seek out and exploit those loopholes.
This is not a novel observation. Judge Florence Pan faced this issue in the Bertelsmann book publishing case. In that matter, Penguin promised that its separate divisions would continue to compete vigorously against one another, even though the merger created inherent incentives not to do so. Judge Pan rejected these unenforceable promises. She quoted the AT&T case that said the argument is undermined by the presumption that “[c]ompanies with multiple divisions must be viewed as a single actor, and each division will act to pursue the common interests of the whole corporation.” Judge Pan similarly observed that Copperweld makes it clear that “a parent and its wholly owned subsidiary have a complete unity of interest” and that any such conduct that deviates from that unity would not be profit-maximizing. Bertelsman was a horizontal merger, but the same concerns apply to vertical mergers where the concern involves the incentives to engage in anticompetitive foreclosure.
Absent a divestiture, these inherent incentives are not extinguished. Alphabet can compensate executives and managers that engage in the desired foreclosure conduct. All the corporation really needs to do is pay some of the employees’ compensation on the basis of corporate profits rather than divisional profits. A manager of the Android division that owns Alphabet stock or is partly compensated with stock options would have direct financial incentives to take actions that benefit the Google Search division. Alphabet also can informally provide rewards to Android managers who favor Google Search relative to the treatment of managers that act neutrally in accord with the remedial order. If the corporation ends up informally awarding bonuses on this basis, how would the court’s remedy monitor ever figure out that this was the real cause?
It is also difficult for a court-appointed monitor to effectively detect and deter prohibited communications within a firm. Remedies may mandate “firewall provisions” to prevent such information exchanges. But it is much harder to prevent the firewalls from being breached in oral conversations, deleted chats, or indirect references. Concerns about recidivism are not limited to criminal defendants.
Less complex cases, such as ones like the Comcast/NBCU merger, where the only foreclosure lever is the availability and price of an input, are more easily remedied solely with behavioral restrictions, particularly if the same input is sold to numerous non-competitors. But in a case with many potential complex foreclosure levers, regulatory experience indicates that remedy design and monitoring would be plagued by the court’s asymmetric information. A regulated firm is far better informed than the court about how various actions might allow them to foreclose and exercise their market power. Moreover, the options for anticompetitive behavior may evolve over time as market conditions change or as the regulated firm causes them to change.
Preventing Android foreclosure conduct
Turning to specifics, the court in Google Search found that Google’s Android agreements with Android phone manufacturers, also called original equipment manufacturers (OEMs), qualified as exclusives. There are numerous ways in which Alphabet could use its continued control over Android to maintain its search monopoly. For example, absent restrictions, it could continue to require Android licensees to favor Google Search, including requiring pre-installation of Google’s Chrome browser, the Google search engine and widget, and other Google apps. Even if these contractual restrictions were prohibited, Android could charge a higher license fee to device makers that do not favor Google Search or those that promote a competing search competitor. Indeed, there is evidence Google did this in France. Android alternatively could use its Mobile Services Incentive Agreements (MADAs) to substitute per-device payments to OEMs that favor Google Search for revenue-sharing. These payments could be calibrated to expected search revenue, thereby achieving the same basic incentives as it did with revenue-sharing. Google also could punish device makers or app developers that favor Google’s Search competitors in other ways, for example, by delaying or withholding advice or information on code changes, by assigning the B-team to interact with them, or by simply winking to the teams to give such firms less than best efforts. The Microsoft case provides various examples of such behavior. Since app developers write Android’s code, Android also could design the Android operating system to favor Google Search or its other apps that provide search access. It also could provide earlier access to code revisions to the Google app developers.
The remedy therefore must include provisions to prevent any and all such conduct by Android that might favor Google Search or incentivize device makers and app developers to do so. The problem is that these prohibitions can be evaded in subtle ways by Android when it is part of Alphabet—or even by an independent AndroidCo that is compensated by Alphabet. For this reason, it is necessary to create a structure where the direct incentives are eliminated and where conduct to create the incentives or reward compliance with them can be observed.
Divestiture is the better remedy to deter foreclosure
These two goals are far better accomplished by a divestiture of Android. First, and most important, a divested AndroidCo does not have the inherent incentives to maintain the monopoly power of Google Search. Second, it is easier to monitor communications between managers of two independent firms than conversations within a single firm.
By contrast, a partially structural solution of making Android a separate subsidiaries of Alphabet would not solve these problems. The separate subsidiary would still be part of the same corporation and so would share its overarching market valuation. It would continue to have inherent incentives to maximize corporate profits and maintain Google’s search monopoly. The corporation also would control compensation and promotion, even if indirectly. Thus, mandating structurally separate subsidiaries does not nearly approximate a divestiture.
AndroidCo required conduct for restoring search competition
Android was used as an important foreclosure instrument for Google Search. For this reason, in addition to the prohibitions discussed above, the divested AndroidCo should be required to engage in affirmative conduct for some period of time to aid in the restoration of competition. AndroidCo specifically should be required to counter its previous conduct (when it was part of Alphabet) by pre-installing the search widgets and search apps of other search engines until the market achieves a sufficient level of self-correction. AndroidCo might also be prohibited from pre-installing the Google Search widget and Search app. These requirements follow directly from its previous anticompetitive conduct when it was part of the Alphabet corporation. A divestiture does not automatically cleanse Android of responsibility for its previous anticompetitive actions.
The Google Search court also might be concerned that an independent AndroidCo has the potential to engage in exclusionary conduct to maintain its mobile operating system monopoly power or use that power in conjunction with exclusionary conduct to obtain market power for itself in search or other applications. Such conduct can be addressed in the future with conventional antitrust enforcement by the Department of Justice and Federal Trade Commission. Alternatively the court’s order could contain general language regarding exclusionary conduct and tying if he wishes to maintain more direct judicial oversight.
Should Chrome also be divested?
It has been suggested that Google Search presiding judge Amit Mehta also might require divestiture of Chrome to the independent AndroidCo. This additional divestiture raises the basic same conceptual issues as above. My proposed conduct remedy would prohibit Google Search from being the default search engine on Chrome. But there also are concerns about technical incompatibility and degraded performance of rival search engines which might favor divestiture. At the same time a divestiture also would raise new concerns about the incentive for for Android exclusionary conduct targeting Chrome’s competitors. Thus, this divestiture issue deserves its own separate detailed analysis that is beyond the scope of this short article.
Author Disclosure: Steve Salop is Professor of Economics and Law Emeritus, Georgetown University Law Center and Senior Consultant, Charles River Associates. He consulted with Epic in its cases against Apple and Google.
Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.