Imitation, local interactions, and efficiency

We consider a circular city model as in Ellison [Ellison, G., 1993. Learning, local interaction, and coordination, Econometrica 61, 1047-1071], where agents follow imitation rules rather than myopic best-response. If interactions are neither global nor limited to the immediate neighbors, payoff-efficient equilibria, and not risk-dominant ones, can be uniquely selected.