Abstract
We develop a principal–agent model with a moral hazard problem in which the principal has access to a hard signal (the level of output) and a soft behavioral signal (the supervision signal) about the agent's level of effort. In our model, the agent can initiate influence activities and manipulate the behavioral signal. These activities are costly for the principal as they detract the agent from the productive task. We show that the agent's ability to manipulate the behavioral signal leads to low-powered incentives and increases the cost of implementing the efficient equilibrium as a result. Interestingly, the fact that manipulation activities entail productivity losses may lead to the design of influence-free contracts that deter manipulation and lead to high-powered incentives. This result implies that the optimal contract (and whether manipulation is tolerated in equilibrium or not) depends on the magnitude of the productivity-based influence costs. We show that it may be optimal for the principal not to supervise the agent, even if the cost of supervision is arbitrarily low.