principal-agent problem: Nonlinear Function