Herbert Hovenkamp has posted "Antitrust and Platform Monopoly" on SSRN. The abstract is below. I was particularly struck by: "The history of antitrust law is replete with firms that are organized as single entities under corporate law, but that function as competitors and [are] treated that way by antitrust law."
This article first considers an often-discussed question about large internet platforms that deal directly with consumers: Are they “winner take all,” or natural monopoly, firms? That question is complex and does not produce the same answer for every platform. The closer one looks at digital platforms they less they seem to be winner-take-all. As a result, we can assume that competition can be made to work in most of them.
Second, assuming that an antitrust violation is found, what should be the appropriate remedy? Breaking up large firms subject to extensive scale economies or positive network effects is generally thought to be unwise. The resulting entities will be unable to behave competitively. Inevitably, they will either merge or collude, or else one will drive the others out of business. Even if a platform is not a natural monopoly but does experience significant economies of scale in production or consumption, a breakup can be socially costly. In the past, structural relief of this type has led to higher prices or business firm failure. If breakup is not the answer, then what are the best antitrust remedies? One possibility is to break up ownership and management rather than assets. The history of antitrust law is replete with firms that are organized as single entities under corporate law, but that function as competitors and treated that way by antitrust law. This permits productive assets to remain intact, but forces decision makers to behave competitively.
Finally, this paper takes a look at the problem of platform acquisition of nascent firms, where the biggest threat is not from horizontal mergers but rather from acquisitions of complements or differentiated technologies. For these, the tools we currently use in merger law are poorly suited. Here I offer some suggestions.