Relational Contracts with Subjective Peer Evaluations

We study optimal contracting in a setting where a firm repeatedly interacts with multiple workers, and can compensate them based on publicly available performance signals as well as privately reported peer evaluations. If the evaluation and the effort provision are done by different workers (as in a supervisor/agent hierarchy), we show that, using both the private and public signals, the first best can be achieved even in a static setting. However, if each worker is required to both exert effort and report on his co-worker's performance (as in a team setting), the worker's effort incentives cannot be decoupled from his truth-telling incentives. This makes the optimal static contract inefficient and relational contracts based on the public signals increase efficiency. In the optimal contract, it may be optimal to ignore signals that are informative of the worker's effort