Contracting with Repeated Moral Hazard and Private

A repeated moral hazard setting in which the Principal privately observes the Agents output is studied. It is shown that there is no loss from restricting the analysis to contracts in which the Agent is supposed to exert e¤ort every period, receives a constant e ¢ ciency wage and no feedback until he is \u85red. The optimal contract for a nite horizon is characterized, and shown to require burning of resources. These are only burnt after the worst possible realization sequence and the amount is independent of both the length of the horizon and the discount factor (). For the in\u85nite horizon case a family of \u85xed interval review contracts is characterized and shown to achieve rst best as ! 1. The optimal contract when << 1 is partially characterized. In-centives are optimally provided with a combination of e ¢ ciency wages and the threat of termination, which will exhibit memory over the whole history of realizations. Finally, Tournaments are shown to provide an alternative solution to the problem. Mainly due to the availability of data, most of the contracting literature has focused on occupations such as chief executive o ¢ cers, golfers, mutual fund managers, tree cutters, windshield installers, and so on, for whom it is possible to construct objective measures of output. As pointed out by Canice Prendergast (1999), the analysis of these occupations has resulted in interesting insights on how incentives work in theory and practice. However, as Im particularly indebted to Andy Skrzypacz and Tom Sargent for their valuable advice and suggestions