Information Design and Moral Hazard

We analyze an information design problem within a moral hazard framework with implicit incentives, where the agent’s effort cost depends on their initial talent. The probability of an agent becoming competent depends on both their type (initial talent) and exerted effort, with effort costs negatively correlated with talent. After exerting effort, the agent takes a test with a publicly observed outcome (success or failure), where competent agents are more likely to succeed. The agent’s payoff is the market’s expected assessment of their competency, conditional on observed information. A designer, aiming to maximize expected competency, selects a signal about the agent’s initial talent. We characterize the designer’s indirect utility over posterior distributions and show that it depends solely on the posterior mean, exhibiting an S-shaped form. Consequently, the optimal signal structure follows an upper censorship rule: fully separating types below a threshold while pooling all types above it. The designer’s incentive to separate types, which is novel in the literature, arises from complementarities between beliefs that enhance the informativeness of the task’s outcomes and the fact that more talented agents incur lower effort costs.