CAREER: Private Information in Auctions, Pricing Games, and Ongoing Relationships

Massachusetts Institute Of Technology, Cambridge MA

Investigators

Abstract

In many relationships, agents interact repeatedly over time with the goal of achieving a self-enforcing cooperative outcome. Problems of this nature are widely studied within economics, and a primary application is collusion among firms. I propose to incorporate into such models the presence of privately observed shocks to the utility of each actor, shocks that affect the efficient actions for the group. I will develop a theory of such relationships, characterizing optimal collusive schemes, and analyzing how these schemes change with the economic environment (such as the nature of anti-trust regulation). I argue that the nature of a collusive scheme hinges critically on the ability of firms to track the identities and actions of individual cartel members over time, and implement schemes that favor or disfavor individual firms. When firms cannot use future market-share favors to reach agreement, optimal collusive schemes can be characterized by high prices and productive inefficiency, and the absence of equilibrium-path price wars. In contrast, when firms have access to future market-share favors, they use them to provide incentives for truthful revelation of costs, without resorting to low prices or price wars. Characterizations of the optimal collusive schemes in this context are challenging, and firms may incur productive inefficiency when trying to implement asymmetric outcomes. Nonetheless, I outline a methodological approach to this problem, with both theoretical and computational parts, and I provide some initial results. With these tools in place, it is possible to evaluate the role of alternative instruments that might be available to the cartel. I propose to analyze the role of explicit communication about cost conditions, highlighting a tradeoff between improved coordination and market-share allocation, and heightened incentives to deviate from a collusive agreement following the revelation of cost information. I further consider monetary side-payments that incorporate some transaction cost, for example, due to the probability that the side-payment will be detected by anti-trust enforcement. Such "bribes" can be a substitute for future market share favors, but if they have any transaction cost they will not fully replace market share favors: optimal collusive schemes will generically be non-stationary. Proposed extensions to this framework include the incorporation of business cycle fluctuations and alternative assumptions about the sources of private information. For example, I propose a model where demand is stochastic and firms privately forecast demand, but the final realization of demand is publicly observed. In such settings, firms will be willing to undergo price wars on the equilibrium path as a punishment for pricing behavior that does not "match" the realized demand level. More generally, I also plan to apply the framework to other contexts, including the provision of public goods in ongoing relationships, and problems of organizational design. The final component of my proposal concerns the empirical measurement of the importance of private information in auctions, with a focus on Forest Service timber auctions. I propose a variety of approaches, both "structural" and "non-structural," to identify the presence of, and extent of, private information about an unknown attribute of the object that affects the utilities of all bidders. Quantifying the extent of the "winner's curse" in such settings provides insight into strategic use of information in auctions and other settings characterized by adverse selection, such as insurance markets. In terms of education, I plan to integrate graduate and undergraduate students into the research process, train them in the new methods I propose, and advise them on related thesis topics. I further intend to continue to aggressively seek out undergraduates, mentor them, and help them apply to graduate school.