NONCOOPERATIVE COLLUSION UNDER IMPERFECT PRICE INFORMATION

Recent work in game theory has shown that, in principle, it may be possible for firms in an industry to form a self-policing cartel to maximize their joint profits. This paper examines the nature of cartel self-enforcement in the presence of demand uncertainty. A model of a noncooperatively supported cartel is presented, and the aspects of industry structure which would make such a cartel viable are discussed. LONG-STANDING QUESTIONS about how widespread is the occurrence of collusion in industries having several firms, and about the extent to which the performance of industries experiencing such collusion departs from the competitive norm, continue to provoke spirited debate. In this paper we offer a theory of collusive industry equilibrium which will provide a means of clarifying these questions. In his classic paper "A Theory of Oligopoly" [15], George Stigler appealed to dynamic considerations to explain how apparently cooperative industry performance might result from noncooperative motives. According to this theory, the firms of an industry form a cartel, which is designed to enforce monopolistic conduct in a self-policing way. "Self-policing" means precisely that the agreedupon conduct is noncooperatively viable and that it remains so over time. Stigler's theory differs markedly from traditional oligopoly theories based on static equilibrium concepts (e.g., Cournot and Stackelberg). This difference is particularly striking in the case of an industry structure which is essentially immune from entry. The traditional theories would suggest that the performance of such an industry should be largely determined by its degree of concentration -the number of firms in the industry and their relative sizes-and by the extent to which substitute goods are available. In contrast, Stigler suggested that the greatest obstacle to collusion in the absence of entry would be what he characterized as "secret price cutting." By informally relating concentration and various other features of industry structure to the immunity of a cartel from entry and to its ability to deter inimical firm behavior, and by assuming that industry profitability reflects successful operation of a cartel, he justified the use of cross-industry regressions to test his theory. The obvious interpretation of Stigler is that he made explicit a theory of oligopoly which implicitly conceived of a cartel as a "policeman" which with some frequency is required to punish destabilizing "offenses" of individual cartel