Adam Hamdan

A principal wishes to rank and reward teams in an organization, modelled by cliques in a graph. Agents in the organizational network perfectly observe the ranking of the teams they are part of, and report local comparisons of their teams. I study the design of deterministic team-ranking mechanisms in settings where agents care about all of their teams' ranks. I show that the only network structures that guarantee a complete ranking of teams are the triangle and the star graph. Unless the network is a star, the principal cannot design an ex-post incentive compatible and efficient mechanism that ranks every clique.