Why is there a widespread view that existing American antitrust law is ill-equipped to address dominant platforms that exclude or discriminate against rivals? How should we think about these cases economically? And what is the right way to address them in practice? Answering these questions is essential to help guide courts and policymakers as they struggle to apply antitrust policy to dominant platforms.
Antitrust is once again in the public spotlight, thanks mainly to growing concerns about dominant tech platforms like Google, Facebook, and Amazon.
Much of the controversy surrounds accusations that these firms exclude rivals from their platforms or otherwise discriminate against them. For example, Facebook has been accused of withholding APIs (which are necessary for apps to run on the Facebook ecosystem) from firms who compete with products offered by Facebook. Similarly, Apple, Google, and Amazon have all been accused of discriminating against smaller rivals that rely on their platforms to reach consumers (for example, by making it harder for consumers to find rivals’ products on the platform). Both Democratic and Republic legislators have proposed bills that would curb platform discrimination, though it is not yet clear whether any of them will pass.
But why are such efforts necessary at all? Why is there a widespread view that existing antitrust law is unlikely to do anything about these alleged platform abuses?
The answer begins with the way antitrust lumps different types of cases into different buckets, each with its own set of legal standards. Antitrust has a single bucket for all cases in which a monopolist unilaterally refuses to deal with rivals (usually by denying them access to an important technology or service) or otherwise deals with them only on discriminatory terms. This is the “refusal-to-deal” doctrine (along with its sibling “essential facilities” doctrine, which is similar).
The problem for prospective plaintiffs is that this doctrine has long been the most controversial area of antitrust. Judicial animosity toward the doctrine peaked in 2004, when it was substantially narrowed by the Supreme Court’s Trinko decision. Since then, it has become virtually impossible for plaintiffs to win cases in this area. This is why platforms do not face meaningful antitrust scrutiny under current law when they refuse to deal with (or discriminate against) competitors.
Why the controversy? There is a common concern among antitrust scholars that imposing an antitrust “duty to deal” with competitors will stifle incentives for investment in valuable new technologies. This is analogous to the rationale underlying the patent system: it may be sensible to let inventors exclude competitors from using their technologies in order to promote incentives for investment in innovation. If firms will be forced to share the fruits of their investments with competitors, they may decide not to make such investments in the first place.
I share the popular view that platform refusals to deal are sometimes anticompetitive. However, I do not think the solution is to breathe new life into the refusal-to-deal doctrine. The main problem is not so much its animating standard of liability (although it is indeed problematic), but rather its basic premise that we should apply the same standard in all cases involving a unilateral refusal to deal with rivals. As a result of that premise, the doctrine conjoins two very different lines of cases (described below) that have no business being evaluated under a common standard. If, instead, antitrust distinguished between these lines of cases, it could open the door to meritorious enforcement actions without jeopardizing incentives for investment. In fact, antitrust already draws an analogous distinction when evaluating other forms of unilateral conduct. These are some of the main arguments in my recent article on the subject, which is forthcoming in the Yale Law Journal. Below, I summarize some of the key points.
There are two important categories of refusal-to-deal cases, which I call “primary” and “secondary.” An example helps to illustrate the distinction. Suppose the defendant developed a network of satellites to deliver high-speed internet service to consumers in rural areas that lack access to conventional internet service. Assume there are no competing satellite internet providers. Then the defendant will hold a monopoly over high-speed internet in rural areas.
In a primary refusal case, the defendant refuses to share its technology with prospective competitors in its “primary” product market, which in this case is internet service. Imagine that a prospective rival is unable to develop its own satellite network. Instead, it asks to pay to use the defendant’s satellites so that it can deliver its own competing satellite internet service. (Assume this would not prevent the defendant from continuing to offer its own service.) By refusing the rival, the defendant will preserve its internet monopoly.
By contrast, in a secondary refusal case, the defendant is vertically integrated into some adjacent (“secondary”) product market that competitors are capable of entering on their own. For example, suppose the defendant also sells an internet-based phone service. Competitors have already developed their own internet-based phone services. But suppose the defendant blocks them all—that is, it refuses to let competing internet phone services run on its satellite network. This will give the defendant an additional monopoly over the secondary product, but not because rivals are unable to develop competing versions of it. On the contrary, they have already done so.
Importantly, a secondary refusal raises essentially the same theory of anticompetitive harm as a tying arrangement. In a textbook tying case, the defendant already has a monopoly over a primary product; it then tells consumers they can buy it only if they agree to buy some additional secondary good from him as well. In such cases, the plaintiff’s theory of harm is usually that the defendant is using its primary market monopoly to distort competition in the secondary market by foreclosing competitors—impairing their ability to compete effectively.
The analogy does not end there. In a tying case, the plaintiff typically does not dispute that the defendant earned its primary market monopoly on the merits (i.e., without engaging in anticompetitive conduct). And its desired remedy—an order to stop tying—would leave that monopoly intact. Antitrust intervention would merely prevent the defendant from exploiting its primary monopoly to injure competition in a separate market. This is important in terms of protecting investment incentives, since the defendant is not being deprived of any profits it earned on the merits.
Similarly, in a secondary refusal case, imposing a duty to deal with secondary market rivals would not disturb the defendant’s monopoly over the primary product. In the example, the defendant would retain its satellite internet monopoly. Antitrust intervention would merely prevent it from excluding rival producers of internet-based phone service. Of course, as with ordinary tying, antitrust intervention would be appropriate only in cases where the defendant’s refusal is shown to harm competition by foreclosing competitors. But, conditional on such a showing, intervention does not harm incentives for investment—it enhances them.
By contrast, a primary refusal to deal is not analogous to any traditional type of exclusionary practice. In the example, the defendant obtained its monopoly over satellite internet by building a satellite array—a costly and difficult undertaking that rivals could not feasibly replicate. Clearly it is not anticompetitive to build satellites. And the usual rule is that a firm does not invite antirust liability by earning a monopoly on the merits. This is an effort to protect incentives for investment. We don’t want to penalize firms for building a better mousetrap. As Judge Learned Hand once wrote, “[t]he successful competitor, having been urged to compete, should not be turned upon when he wins.” The same logic ought to apply in cases where the only allegedly unlawful act is the defendant’s refusal to give away a monopoly it earned on the merits. Therefore, the usual critiques of the refusal-to-deal doctrine—most notably its propensity to diminish investment—make a lot of sense in primary refusal cases.
Thus, there are crucial differences between primary and secondary refusals. The problem is that the refusal-to-deal doctrine lumps them into the same bucket, applying the same legal standard in both types of cases. Courts are rightly concerned that intervention in some cases (namely, those involving primary refusals) could chill investment, and so they have erected suffocating evidentiary rules that make it virtually impossible for plaintiffs to win. But these rules also prevent meaningful antitrust scrutiny of secondary refusals, even though intervention in these cases would be perfectly consistent with longstanding antitrust policy toward vertical restraints (such as tying).
The solution is simple: courts should apply different legal standards in primary and secondary cases. They should continue to disfavor intervention in primary refusal cases, where the usual concerns apply. But a plaintiff should be permitted to challenge a secondary refusal as a de-facto tie or similar restraint. Significantly, this aspect of the proposal would not contravene any Supreme Court decisions, as all conflicting precedents were created by lower courts. Thus, if it were so inclined, the Supreme Court could recognize a distinction between primary and secondary refusals without having to overturn any of its prior decisions.
Among other things, this would allow for meaningful and sensible antitrust scrutiny of digital platforms that refuse to deal with or discriminate against competitors, as the large majority of such cases are secondary. In a typical case, the defendant’s primary product is its platform, but it is vertically integrated into some secondary product market. It is in the latter market that rivals are allegedly being excluded. Unlike a primary refusal case, these rivals have independently developed their own competing products. But they nevertheless seek to deal with the defendant in order to facilitate distribution or marketing of their products, or else to ensure their interoperability with the primary product. In most cases, the defendant already engages in such dealings with third-party customers; it is only competitors who are rebuffed.
In fact, this proposed distinction between primary and secondary refusals is exactly analogous to how antitrust already treats other forms of unilateral conduct. When considering other types of unilateral conduct, courts hold that they are typically beyond antitrust scrutiny. But there is an important exception: if the conduct serves to effectuate a de-facto vertical restraint, then plaintiffs can challenge it as such. A good example is unilateral price-setting. When a monopolist sets a high price, this decision is not subject to antitrust scrutiny. But in some cases the monopolist also offers a bundle comprising its monopoly good and a secondary product. In these cases, the monopolist may engage in “bundled discounting,” a multiproduct pricing arrangement that resembles tying, although it may be purely unilateral. In such cases, courts permit plaintiffs to challenge the arrangement as de-facto tying. There is no reason why this should not be permitted in secondary refusal cases as well.
There are a lot of important details that this brief write-up leaves out. But the key point is that, as a matter of antitrust economics, secondary refusals have a lot more in common with conventional vertical restraints than with primary refusals. A corollary is that expanded intervention in unilateral refusal cases need not entail a significant departure from bedrock antitrust principles. On the contrary, antitrust experts have long recognized that, although a firm should not be penalized for earning a monopoly on the merits, it is not entitled to exploit that monopoly to impair competition in other markets. However, by applying the same legal standard to all types of unilateral refusals, modern antitrust law defies this principle.
Learn more about our disclosure policy here.