I consider a repeated principal-agent model with moral hazard, in which the agent has βδ-preferences, which are widely used to capture time-inconsistency. I analyze the case where the agent is sophisticated in the sense that he is fully aware of his inconsistent discounting. I characterize the...