Dynamic mechanism design on social networks

Abstract

This paper studies a mechanism design problem with networked agents and stochastically evolving private information. In contrast to the canonical mechanism design theory focusing only on information asymmetry, we also pay attention to the topology of social network among agents. We find that the standard first-order approach for mechanism design is invalid in dynamic environment. As a remedy, a novel ironing technique is proposed, which produces a perfectly sorting allocation. Based on the optimal dynamic mechanism obtained, we define and compare some important nodes and edges in a network for different ranges of synergy parameter. We further discuss the network intervention problem, in which the principal can intervene to change the ex-ante distribution of individual types.