Abstract. We describe a set of mechanisms we refer to as reciprocal mechanisms. These play the same role as direct mechanisms in single mechanism designer problems in that they provide a 'canonical', though abstract, way of representing equilibrium outcomes. We use them to show that the set of outcome functions supportable as Perfect Bayesian equilibrium in regular competing mechanism games is equivalent to the set of outcome functions supportable as Perfect Bayesian equilibria in a reciprocal contracting game. We provide a full characterization of this set of outcome functions. This characterization makes it possible to by pass game theoretic complexities in order to understand the impact of competition in competing mechanism games using a set of inequalities.