Theoretical Economics 16 (2021), 1471–1512

I study a repeated principal-agent game with long-term output contracts that can be renegotiated at will. Actions are observable but not contractible, so they can only be incentivised through implicit agreements formed in equilibrium. I show that contract renegotiation is a powerful tool for incentive provision, despite the stationarity of the environment. Continuation contracts are designed to punish deviations in noncontractible behaviour. If the equilibrium actions are observed, these contracts are renegotiated away. This form of anticipated renegotiation results in welfare improvements over outcomes attainable by one-period contracts, or by long-term contracts which are not renegotiated. When the principal is not protected by limited liability, first-best outcomes are attainable regardless of the impatience of the players. Equilibrium strategies are shown to satisfy various concepts of renegotiation-proofness.

Keywords: Long-term contracts, renegotiation, relational contracts

JEL classification: C73, C78, D86