This paper uses network theory to argue that the consequences of horizontal ownership by large investment institutions are more complicated than, and sometimes the complete opposite of, what conventional economic theory predicts. Horizontal ownership occurs when a large investment institution, such as Vanguard or BlackRock, simultaneously holds large stakes in many different companies in the same industry. Legal scholars and economists have argued that these large investors have little incentive to encourage competition in the industries in which they have horizontal ownership because the investors are just as likely to hold shares in companies that might lose from competition as they are to hold shares in companies that might gain.
Against this background, this paper advances two claims. First, it shows that the policy proposals that have been advanced to address the alleged anticompetitive effects of horizontal shareholding could backfire and further reduce the level of competition in the affected markets. Second, it highlights that the consequences of horizontal shareholding are nuanced because things that happen in one industry inevitably affect other industries. For instance, increased ticket prices among airlines might be good for airlines but bad for their suppliers. Therefore, determining whether reduced competition in a given industry would benefit an investor requires us to compare the gains it would generate in the relevant market with the losses it would impose on other firms in the investor’s portfolio.
I work through the mechanics of these calculations and identify a method already developed in network theory that could help us perform them. I also show that in some markets (that is, “central markets”), horizontal shareholders might have greater incentives than undiversified shareholders to promote aggressive competition. I then outline a new set of regulatory tools, which I call “Network Sensitive Regulations,” that could address the anticompetitive effects of horizontal shareholding in a manner that would be sensitive to the nuances of these network effects.