Stochastic Market Sharing, Partial Communication and Collusion
Heiko A. Gerlach
2007
Social Science Research Network
This paper analyzes the role of communication between firms in an infinitely repeated Bertrand game in which firms receive an imperfect private signal of a common value i.i.d. demand shock. It is shown that firms can use stochastic, inter-temporal market sharing as a perfect substitute for communication in low demand states. Therefore, partial communication in high demand states is sufficient to achieve the most collusive, full communication outcome. And partial communication in low demand
more »
... does not improve on the equilibrium without communication. Communication in high demand states allows firms to coordinate their pricing, choose the most efficient uninformed price and avoid price wars. I demonstrate that under some conditions consumers are better off with communication among colluding firms. JEL-classification: L41, L13, D82 1 to be able to observe the use of facilitating practices. While it seems less obvious to detect information sharing or communication between firms, the Lombard Club case and other recent high-profile cartel cases such as Citric Acids or Vitamins suggest that communication typically leaves hard evidence that can potentially be seized. In this paper I concentrate on the first condition and analyze the importance of communication between firms for the sustainability of collusion. While there is some consensus about the fact that communication facilitates collusion, the question here is rather how much communication is actually needed and in which circumstances firms have stronger incentives to communicate. To this end, I consider an infinitely repeated Bertrand game with independent, common value demand fluctuations (low or high). At the beginning of each period, each firm receives a private, independent (over time and firms) signal about the current demand level which is either perfectly informative or not informative at all. The resulting asymmetric information between firms implies that firms do no longer agree on the most collusive industry price and have an incentive to engage in communication. At the following communication stage, firms simultaneously send messages to all other firms in the industry. Then firms set prices, the demand level is disclosed and profits are realized. I analyze and compare symmetric perfect public equilibria (SPPE) in three different modes of communication: no communication, full communication (i.e. communication in all states of demand) and partial communication (i.e. communication in one demand state only). The unobservability of private signals introduces the possibility of opportunistic price cuts in the sense that firms can deviate to prices that are assigned to different signal types. These deviations are not detectable with probability one and impose additional constraints on the optimal organization of collusion. Two such on-schedule deviations have to be accounted for in this context: the deviation of a firm with an informative signal that demand is high (low) to the equilibrium price for firms with an uninformative signal. The analysis of optimal collusion without communication among firms shows that the most restrictive on-schedule constraint is the one that prevents firms with a high demand signal to post the price of an uninformed firm. However, firms have three instruments to relax this condition and the optimal organization of collusion trades off the benefits and costs of these instruments. First, firms can impose price rigidity, i.e. they simply equate the price for the two types in every period. Second, they can distort the uninformed price downwards to make deviations less attractive. And thirdly, firms can start price wars after price distributions they deem suspicious of on-schedule deviations. I show that if firms are patient, the leverage of the price war threat is sufficiently strong such that firms only rely on the latter two instruments. Firms optimally start price wars after observing uniformed prices from all firms and distort the uninformed price downwards but always above the low demand monopoly level. The on-schedule constraint for low demand signal firms is never binding and the optimal organization of collusion approaches the unconstrained maximum without communication if the discount factor goes to one.
doi:10.2139/ssrn.982323
fatcat:ohdwd5gw2jd25cqdkrcyxgu4my