A group of principals collectively and sequentially screen an agent. The principals hold heterogeneous values from the relationship that may evolve over time. At each date, the principals use a collective decision rule to propose a contract to the agent. We unearth circumstances in which every non-dictatorial decision rule yields inefficiency in the form of excessive learning: relative to a single-principal benchmark, collective principals screen the agent too aggressively. They do so in order to reduce uncertainty about the agent’s preferences, and alleviate dynamic conflicts of interest between the principals that exist regardless of their static alignment.

scoco@ceibs.edu