Wednesday, May 1, 2013
Posted by Abe Wickelgren
Hempill and Wu’s (2013) paper, “Parallel Exclusion,” illustrates that inefficient exclusionary behavior is not necessarily just a concern in monopoly markets. They provide examples where exclusion has occurred in oligopoly markets and present some informal analysis of why this occurs. As with any analysis of exclusionary behavior, the standard “Chicago School” question must be answered: can exclusion be both profitable and inefficient? The question is potentially more difficult to answer in the affirmative for oligopoly exclusion to the extent that the firm bearing the cost of the exclusive action is likely sharing the benefit with other firms in the market.
Hemphill and Wu recognize this. Section III of their paper is devoted to addressing the externality issue. While their examples present helpful illustrations, examples cannot illustrate the necessary and or sufficient conditions for parallel exclusion to be profitable and inefficient. This is important since identifying these conditions is helpful in alerting antitrust enforcers to situations where they
should be concerned with such exclusion and situations where they should not. Thus, it might be helpful to consider some of the leading models of monopoly exclusion and ask if these models can also provide a basis for profitable and inefficient parallel exclusion.
Simpson and Wickelgren (2007) show that a monopolist can use exclusive dealing contracts to inefficiently exclude entry of a lower cost rival when its buyers are competitors in a downstream market. In this model, a buyer who does not sign an exclusive contract induces the monopolist to offer its input at a lower price to the buyers that did sign. The monopolist does this so that its buyers
can more effectively compete with a non-signing buyer who can purchase from the lower cost entrant. As a result, the non-signing buyer does not benefit significantly from spurring competition in
the upstream market and obtaining the input at a lower price. This benefit is almost entirely passed-on to the downstream consumers. Because of this, it costs the incumbent monopolist very little to induce all buyers to agree to an exclusive dealing contract.
How would this model work in an oligopoly market? On the one hand, the fact that this model shows it costs very little to induce competing buyers to sign exclusive deals, suggests that the same strategy should be effective in an oligopoly setting. Even if the incumbent firms had to share the benefit from keeping out an entrant, the cost could still easily be low enough that each firm would find doing so profitable. There are two potential complications, however. First, the firms would need to coordinate on which buyers each firm was supposed to contract with. If firm A thought that firm B would offer each firm a small payment for agreeing not to purchase from an entrant, then
firm A would not do so. It is natural to imagine an equilibrium where A and B split up the buyers, but some coordination would be required to decide which buyers were A’s responsibility and which were
B’s. That coordination may or may not subject the firms to Section 1 liability.
Second, oligopoly not only requires sharing the benefit from exclusion, it often means the combined benefit is smaller as well. For example, in the Simpson and Wickelgren model, the upstream firms produce a homogenous input and compete via Bertrand competition. Thus, if there were even two firms, profits would be zero, providing no incentive to exclude a third. There are two natural
ways to tweak this model so that oligopoly is profitable. The first is to imagine that the firms are
colluding on price, so that they are sharing positive industry profits. Entry would then lower their profit either by forcing them to share this collusive profit three ways instead of two or by disrupting the collusive outcome altogether. In either case, the loss from entry is smaller than the loss to the monopolist in the base Simpson and Wickelgren model, but since the cost to exclusion is so small, exclusion is still individually profitable.
The second is to change the assumption about the upstream competition to either Cournot competition or differentiated Bertrand competition. Either of these would create a loss to each firm from entry. Furthermore, under Cournot competition it would still be the case that exclusion would be close to free. A non-signing firm that induced entry would increase quantity in the market and lower the price for all of the competing buyers, again leading the benefits from the price reduction being passed-on to final consumers.
Under differentiated Bertrand competition, however, exclusion would probably be costly to the excluding firms. The fact that upstream competition is differentiated suggests that there is some benefit to variety among the downstream firms. One possible story is that the production of the final consumer good is cheaper when the downstream firms use more different types of inputs. In most such models, it would not be optimal for the entrant to price its input to completely eliminate
the profit of the non-signing firm. This would mean that each firm would have to receive a non-trivial payment to sign the exclusive contract, making exclusion more difficult under oligopoly than
monopoly since each firm’s share of the benefit could now be less than this non-trivial cost. A similar phenomenon would likely occur if the upstream differentiation were due to final consumers having a preference for retailers having a variety of products.
This does not necessarily mean that inefficient exclusion is impossible in such a model. It only means that the profitability of inefficient exclusion does not obviously generalize from the monopoly model to the duopoly model. More formal analysis would need to be done to determine the exact conditions under which inefficient exclusion would be profitable in an oligopoly with upstream differentiated Bertrand competition.
One could, and should, perform a similar analysis for how other models of monopoly exclusion might be extended to cover the case of oligopoly exclusion. In so doing, one could more accurately determine the range of circumstances in which parallel exclusion could be profitable and
inefficient. Hemphill and Wu’s valuable contribution is to have alerted the antitrust community to an important issue, but more work needs to be done in order to transform this insight into clearer
guidance for antitrust enforcers and the courts.
That said, it is possible that the benefit from a third variety of the input is less than the benefit from the second, which might mean that it would actually cost the monopolist more to exclude the second firm than it would cost duopoly firms to exclude the third firm.