Collusion in a three-tier hierarchy: Credible beliefs and pure self-interest

Abstract

The assumption of a purely self-interested supervisor in the principal–supervisor–agent framework gives rise to the possibility of supervisor–agent collusion. In this paper, I show that equilibrium refinement assumptions on type belief updating during side contract negotiations can guarantee that collusion occurs. Even if the principal is allowed to manipulate the incentives in the contract he offers, the only way to prevent collusion is to ignore the supervisor. Thus the role of a purely self-interested supervisor, in this case, is questioned. Lastly, I discuss some plausible alternatives and explanations of the assumption of pure self-interest.