Measurement distortion and missing contingencies in optimal contracts

Abstract

Theory suggests that optimal contracts should include many contingencies to achieve optimal risk sharing. However, in practice, few contracts are as complex as theory suggests. This paper develops a model which is consistent with this observation. The lack of risk sharing results from the interplay of two factors. First, contingencies must be based on information produced by measurement systems, which may be manipulable. Second, when two parties to a contract meet, they often have incomplete information. The type of contract offered may reveal information about the party who proposes it. Different types of agents have different preferences over contingent contracts, because they have different abilities to manipulate the measurement system. These differences in preferences allow the parties to signal their types through the contracts they offer. Noncontingent contracts may be chosen in equilibrium because they are the only contracts which do not give any type an incentive to distort the measurement system and, hence, do not reveal information about the party proposing the contract.