Peer selection with external referees

The problem of peer selection is a problem facing a social planner who aims to choose a project of the highest quality from the set of projects submitted by a number of strategic agents. To resolve this problem the planner asks the agents to evaluate the projects and report on their quality. We study the implementation problem resulting from the peer selection problem and consider a scenario where projects are also reviewed by a less informed but impartial external referee. We consider both dominant and Bayesian incentive ompatibility (DIC and BIC) and show that the latter allows the planner to use the reports submitted by the non-expert to discipline the agents and to extract some truthful information about the projects. We characterize both the DIC and the BIC mechanism and use this characterization to obtain the compute to what extent the used of an external mechanism helps the social planner in the case of the BIC mechanism.