We prove the folk theorem for the Prisoner's dilemma using strategies that are robust to private monitoring. From this follows a limit folk theorem: when players are patient and monitoring is sufficiently accurate, (but private and possibly independent) any feasible individually rational payoff can be obtained in sequential equilibrium. The strategies used can be implemented by finite (randomizing) automata. * Thanks to Görkem Celik for valuable research assistance.

No abstract

We construct a model where the reputational concern of the long-run player to look good in the current period results in the loss of all surplus. This is in contrast to the bulk of the literature on reputations where such considerations mitigate myopic incentive problems. We also show that in models where all parties have long-run objectives, such losses can be avoided. Downloaded from2. The formal expression of this statement is the following. Without any external incentives the agent could expect at least his worst, and at most his best, Nash equilibrium payoff. Fudenberg and Levine [1992] show that reputation effects restrict the set of payoffs a long-run agent can expect, and they provide explicit lower and upper bounds. The Fudenberg and Levine [1992] lower and upper bounds weakly exceed the worst and best Nash equilibrium payoff, respectively. In relation to these observations, an interpretation of our results is that a tighter upper bound is possible. Downloaded from4. In particular, we cannot bound in advance the absolute number of stages that are required before the mechanic's reputation is established or lost. In the analysis developed by Fudenberg and Levine [1989] for the case of commitment types, such a bound exists independently of the discount factor, and this bound translates into a bound on the payoffs of the long-run player. With strategic types, the number of engine replacements which leads to the loss of reputation depends on the equilibrium strategy of the bad mechanic, and this in turn depends on the discount factor.

We consider a general mechanism design setting where each agent can acquire (covert) information before participating in the mechanism. The central question is whether a mechanism exists that provides the efficient incentives for information acquisition ex-ante and implements the efficient allocation conditional on the private information ex-post.It is shown that in every private value environment the Vickrey-Clark-Groves mechanism guarantees both ex-ante as well as ex-post efficiency. In contrast, with common values, ex-ante and ex-post efficiency cannot be reconciled in general. Sufficient conditions in terms of sub-and supermodularity are provided when (all) ex-post efficient mechanisms lead to private under-or over-acquisition of information. for several helpful discussions. We are especially grateful for suggestions from two anonymous referees and a co-editor. Comments from seminar participants at Minnesota, U.C.L.A., and Yale are greatly appreciated.

Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number.