Information Games and Robust Trading Mechanisms

Abstract

Agents about to engage in economic transactions may take costly actions to influence their own or others' information: costly signaling, information acquisition, hard evidence disclosure, and so forth. We study the problem of optimally designing a mechanism to be robust to all such activities, here termed "information games." The designer cares about welfare, and explicitly takes the costs incurred in information games into account. We adopt a simple bilateral trade model as a case study. Any trading mechanism is evaluated by the expected welfare, net of information game costs, that it guarantees in the worst case across all possible games. Dominant-strategy mechanisms are natural candidates for the optimum, since there is never any incentive to manipulate information. We find that for some parameter values, a dominant-strategy mechanism is indeed optimal; for others, the optimum is a non-dominant-strategy mechanism, in which one party chooses which of two trading prices to offer.