Making and Breaking Promises When Their Costs Are Private Information

We discuss how trust is optimally maintained when promise-makers are privately informed about the costs of keeping their promises and efficient transfers are not feasible. To this end, we present a simplified version of the model in Li and Matouschek (2013) in which a principal and an agent are in a infinitely repeated relationship. The agent's effort and output are observable but not contractible and the principal is privately informed about the per-dollar cost of paying the agent, which is either one or infinite. We characterize the optimal relational contract, illustrate the methods used in solving games with one-sided asymmetric information and inefficient transfers, and discuss further applications.