Congress is currently considering two major bills that would regulate “self-preferencing” and related conduct by dominant digital platforms. Criticism of these bills is heavily influenced by “Single Monopoly Profit” theory, which purports to show that such conduct cannot be both anti-competitive and profitable. David J. Balan writes that the theory’s shortcomings limit its validity in general and its application to self-preferencing in particular.
Congress is currently considering two bills whose purpose is to address dominant tech platforms’ potentially anti-competitive practices. The American Innovation and Choice Online Act (AICOA) prohibits “self-preferencing,” such as when Google favors Google-affiliated search results over unaffiliated ones, or when Amazon favors its own products sold on the Amazon platform over those of its rivals. The Open App Markets Act (OAMA) prohibits self-preferencing in app marketplaces (Apple’s “App Store” and Google’s “Google Play”) and requires that consumers be allowed to “side-load” third-party apps onto their phones rather than have to purchase them through the app marketplaces.
AICOA and OAMA respond to the concern that these practices might be a way for a dominant platform owner to leverage its existing market power to gain additional market power over goods and services advertised or sold on its platform. A related concern is that the practices might be a way for a platform owner to thwart the entry of potential rivals that may have eventually displaced it from its dominant position. In what follows, I refer to these possible anti-competitive outcomes as “expanding” and “extending” market power.
These concerns seem intuitive. If a firm owns a platform and also operates on that platform, and if its current or potential future rivals need to operate on that platform (which they likely will if the platform is dominant), then it seems obvious that the platform owner has the ability and the incentive to engage in self-preferencing and related practices to disadvantage those rivals. To use a well-worn metaphor, it would seem straightforwardly problematic to allow the platform owner to be both a player and the referee in the same game.
However, the “Single Monopoly Profit” (SMP) theory–famously associated with the Chicago School of economics and developed during the second half of the 20th century–purports to demonstrate that no such problem exists. In its strongest, most orthodox form, the key result of SMP theory is that a firm with market power in one market (such as a dominant platform owner) can do no better than to simply enjoy, for as long as it lasts, the “single monopoly profit” associated with that market power. Any attempt to use practices such as self-preferencing to expand or extend that market power must backfire, reducing profits rather than increasing them.
The logic behind SMP theory—as it applies to self-preferencing conduct—can be described as having two prongs (see here for a more detailed discussion). The first prong observes that any practices chosen by a platform owner that deviate from those that a neutral owner (i.e., an owner who does not also operate on the platform) would have chosen must decrease the platform’s profits. The reason is that the neutral owner chooses practices that maximize profits, which means that any different practices must result in lower profits. Self-preferencing is by definition such a deviation, so self-preferencing must reduce the platform’s profits. This prong is really just a version of the “no free lunch” idea (rightly) beloved of economists, and it is valid.
However, the fact that self-preferencing causes the platform’s profits to decrease does not necessarily mean that the practice is unprofitable overall. Self-preferencing may result in a profit increase by expanding or extending the platform’s market power, and this can more than offset the decrease. This brings us to SMP’s second prong, which claims that this is not possible and that the profits lost from any such deviation must exceed the profits gained. Put another way, the second prong of SMP theory says that self-preferencing and related practices cannot be both anti-competitive and profitable.
Before proceeding further, it is worth pausing to note that the stakes here are high. Platform owners commonly employ practices such as self-preferencing, and they presumably do so because those practices increase total profits. If SMP theory really does rule out, as a matter of economic theory, the possibility that those profits are the result of an anti-competitive expansion or extension of market power, then there is no reason to be concerned about the practices. Moreover, if these practices cannot be anti-competitive, then the only remaining explanation for their existence is that they confer a pro-competitive efficiency benefit. In other words, a direct and remarkable implication of SMP theory is that, if it is valid and applicable, the efficiency of self-preferencing and related practices can be inferred from their mere existence, and therefore need not be demonstrated with evidence.
With these high stakes in mind, we return to the second prong of SMP theory, namely that self-preferencing cannot be both anti-competitive and profitable. It is undoubtedly possible to construct reasonable, internally consistent economic models in which this prong, and therefore the SMP theory overall, does hold (the first prong is not in dispute). However, there are strong reasons to doubt the theory’s validity and also its applicability to self-preferencing. First, a great deal of “post-Chicago” theoretical research over the last twenty-five years has demonstrated that there are many realistic circumstances in which the second prong of SMP theory does not hold and that conduct that disadvantages rivals can in fact be used profitably to anti-competitively expand or extend market power.
Second, I am not aware of any model that shows that the second prong of SMP theory holds for self-preferencing conduct specifically, and it is likely that it does not (the models in which SMP theory does hold, and the post-Chicago responses to those models, mostly deal with exclusive dealing or tying/bundling and are not directly applicable to self-preferencing).
To see why, begin by noting that while the self-preferencing lunch is not free, it can be very cheap. For example, suppose that the profit-maximizing presentation (i.e., the presentation that a neutral platform owner would choose) would be to list product “A” first and product “B” second, but that the reverse ordering is only slightly less profitable. This means that if the platform owner owns product B, the loss from self-preferencing (listing B before A) would be small. At the same time, the gain from self-preferencing could be large, much larger than the loss, if it results in “stealing” a substantial number of profitable sales that would have otherwise gone to A. I am aware of no SMP theory result that rules this out.
It should be clarified now that the orthodox version of SMP theory is something of a straw man, as few economists would defend the theory as literally true. However, many do embrace a more moderate but still strong version in which it is highly unlikely (though not impossible) for a dominant platform owner to profitably use practices such as self-preferencing to expand or extend its market power. The conclusion for these economists is that there remains little reason to be concerned about these practices and the amount of evidence required to demonstrate any claimed efficiency benefits from these practices is still low. However, the second prong of SMP theory fails in this version for the same reasons as mentioned above, though the reasons are of a slightly weaker form.
SMP theory, whether in its orthodox or its more moderate form, has had a deep influence on the debate over practices such as self-preferencing, including the debate over AICOA and OAMA. While not always called by its name, that influence is apparent in how readily some economists and lawyers dismiss as naïve and outdated the notion that “the player should not also be the referee” and in how readily efficiency claims are made and taken seriously with little evidence to support them. But as discussed above, twenty-five years of post-Chicago research has undermined SMP theory (specifically the second prong), and in any case, the theory largely does not apply to self-preferencing. SMP theory’s continued influence over these debates is therefore unmerited, and the old and much-maligned wisdom about players and referees has a great deal to recommend it.
The conclusion that SMP theory cannot be used to justify self-preferencing practices is an important consideration in the debate over AICOA and OAMA. Ruling out what has heretofore been considered an important defense of those practices tends to support the conclusion that they should be limited by law when employed by powerful dominant platforms.
Author’s note: I am grateful to George Deltas for his extremely valuable insights. I also thank Paul Stancil and other participants of the 2022 BYU “Tech Platforms and Online Retail in a New Age of Competition Law” conference for very helpful comments and discussion.
Read more about our disclosure policy here.