Convex Potentials with an Application to Mechanism Design

This paper establishes a general form of the "payoff equivalence" result in mechanism design theory: under certain conditions, the utility of any type in an incentive-compatible mechanism is determined up to an additive constant by the allocation rule alone. When types are single-dimensional the result is well known (see, for instance, Myerson (1981)). When types are multi-dimensional the result follows from the Fundamental Theorem of Calculus once sufficient smoothness is assumed. We obtain a more general result by using an extension of the Fundamental Theorem to nonsmooth convex functions and more generally, to the class of regular Lipschitzian functions.