Unique Information Elicitation

This study investigates the unique information elicitation problem. A central planner attempts to elicit correct information from multiple informed agents through mutual monitoring. There is a severe restriction on incentive devices: we assume neither public monitoring technology nor allocation rule is available; thus, the central planner only uses monetary payment rules. It is well-known that if all agents are selfish, it is impossible to elicit information as a unique equilibrium. We consider an epistemological possibility that some agents could be motivated by an intrinsic preference for honesty, while we allow that honest agents are mostly motivated by monetary interest. We prove that, once we introduce an epistemic type space that allows agents to be (weakly) honest, then the impossibility theorem reduces to a knife-edge case: The central planner can elicit correct information from agents as a unique Bayes Nash equilibrium behavior if and only if it is never common knowledge that all agents are selfish.