Optimal contract under moral hazard with soft information

I study a model of moral hazard with soft information: the agent alone observes the stochastic outcome of her action; hence the principal faces a problem of ex post adverse selection. With limited instruments the principal cannot solve these two problems independently; the ex post incentive for misreporting interacts with the ex ante incentives for effort. This affects the shape and properties of the optimal contract, which fails to elicit truthful revelation in all states. In this setup audit and transfer become strategic complements; this is rooted in the nonseparability of the problem.