| This dissertation consists of three essays. The first essay studies a long-term principal-agent model in which the agent can exert effort in multiple tasks. The analysis shows that along the path of the optimal contract, the selection criterion for tasks deviates from the net present value (NPV) and takes the form of a task-specific markup which depends on the task's risk-return ratio. The optimal contract can be implemented with an equity stake in the firm and task-specific bonus payments. The second essay examines a long-term principal-agent model when the principal is subject to a form of ambiguity aversion. Specifically, the effort cost of the agent may change over time and the principal offers the contract which is optimal under the worst-case realization of the effort cost process. This model rationalizes excessively strong incentives, which are reflected in a large equity stake, as well as over- or undercompensation for agents who have been respectively successful or unsuccessful in the past. The two essays above are formulated in continuous time and rely on the Hamilton-Jacobi-Bellman (HJB) apparatus. The third essay, which is based on joint work with Bruno Strulovici, provides general conditions under which stochastic control problems can be solved via this method. In particular, we consider both the existence of an optimal control which admits a strong solution of the associated controlled diffusion process and the existence of a twice differentiable solution of the HJB equation. |
