Working Paper du GATE 2003-05This paper investigates to which extent the optimal centralized mechanism as defined by the Revelation Principle may optimally prescribe to follow the agent's recommendation. We define two sub-cases, called delegation and authority, being parts of the consideredcentralized organization, a decision being delegated to the agent when his recommendation is followed by the principal. We assume that there are both adverse selection and moral hazard under delegation and moral hazard under authority and the agent is risk neutral but protected by limited liability on utility. We find that delegation can be efficient even if the principal'stalent is higher than the expected talent of the unknown agent.Dans quelle mesure l...