Renegotiation in Agency Contracts and Information Acquisition before Renegotiation

This article analyzes renegotiation in an agency model where a principal observes a noisy signal about the agent’s effort before the renegotiation. This signal is not verifiable by an outsider or is not contractible in the initial contract. Although any single effort level except the minimal one is not sustainable, the signal is used by the principal to improve efficiency. The agent randomizes his efforts no matter how accurate the signal is but the distribution of the effort levels and the principal’s payoff converges to the first best as the signal converges to a completely accurate one.