Private Disclosures in Competing Mechanisms – ACMP2023

The proposed project aims at studying a broad class of situations in which multiple designers compete by offering mechanisms to multiple agents, as in the case of platforms competing to attract users, auctioneers competing to attract bidders, or marketplaces competing to attract trading partners to their venues as in competitive search. Attar, Campioni, Mariotti and Pavan (2021) have shown that allowing the principals to engage in private disclosures (i.e. information about certain elements of their mechanisms disclosed privately to the agents before they act in the mechanism) has a profound effect on the outcomes of such games and on the corresponding predictions and policy recommendations. In particular, equilibria of games in which the principals are restricted to standard mechanisms (with no private disclosures) need not be robust. Furthermore, allocations that cannot be sustained in equilibrium when principals offer mechanisms without private disclosures can be sustained when private disclosures are allowed. Taken together, these results challenge the existing approach to these economic situations and call for a novel view in which principals' private disclosures play are central to the theory of competing mechanisms. The research that we plan to conduct involves different fronts. First, to construct a class of mechanisms that is fully canonical, satisfying two key properties: (1) any equilibrium outcome sustained by restricting the designers to mechanisms from such a class is also an equilibrium outcome in any game with richer message and disclosure sets (robustness); and (2) any equilibrium outcome in the enlarged game is also an equilibrium outcome in the canonical game (universality). Second, to identify a class of tractable mechanisms, to be used in applications. Third, to apply the new methodology to a few settings of interest to identify novel predictions and policy recommendations.