Implementation under ambiguity

We introduce the idea of implementation under ambiguity. In particular, we study maximin efficient notions for an ambiguous asymmetric information economy (i.e., economies where agents' preferences are maximin à laWald, 1950). The interest on the maximin preferences lies in the fact that maximin efficient allocations are always incentive compatible (de Castro and Yannelis, 2009), a result which is false with Bayesian preferences. A noncooperative notion called maximin equilibrium is introduced which provides a noncooperative foundation for individually rational and maximin efficient notions. Specifically, we show that given any arbitrary individually rational and ex-ante maximin efficient allocation, there is a direct revelation mechanism that yields the efficient allocation as its unique maximin equilibrium outcome. Thus, an incentive compatible, individually rational and efficient outcome can be reached by means of noncooperative behavior under ambiguity.