Dynamic programming approach to the robust principal-agent problem

Principal-agent models are studied to incorporate the moral hazard where the agent has unobservable behavior. This paper considers a special formulation of the principal-agent problem with finite time lump-sum payment, which can be interpreted as the two-player stochastic differential game. Inspired by the latest works, we exploited the dynamic programming approach to solve the stochastic control problem. In addition, we investigated the robust formulation by introducing uncertainty to the drift function, resulting in an inaccurate observation about the output process