Common Agency and the Revelation Principle

In the common agency problem multiple mechanism designers simultaneously attempt to control the behavior of a single privately informed agent. The paper shows that the allocations associated with equilibria relative to any ad hoc set of feasible mechanisms can be reproduced as equilibria relative to (some subset of) the set of menus. Furthermore, equilibria relative to the set of menus are weakly robust in the sense that it is possible to find continuation equilibria so that the equilibrium allocations persist even when the set of feasible mechanisms is enlarged. In agency problems with a single principal, no loss of generality is involved when the principal is restricted to offer standard direct mechanisms. When there are multiple principals things are more complicated because of the fact that the agent knows about the mechanisms that have been offered by all the principles at the time that he communicates with any particular principal. A principal who recognizes this can construct a mechanism that makes the ∗Thanks are due to Larry Epstein, Meg Meyer, Carolyn Pitchik, Lars Stole and Sergei Severinov for helpful comments. The financial support of the Social Sciences and Humanities Research Council of Canada is gratefully acknowledged.