Social interactions and complex networks

This paper studies the impact of interaction topologies on individual and aggregate behavior in environments with social interactions. We study social interaction games of an infnitely large population with local and global externalities. Local externalities are limited within agents' ego-networks while the global externality is derived from aggregate distribution in a feedback manner. We consider two forms of heterogeneity, that due to individual intrinsic tastes and that due to ego-networks. The agents know the potential number of other agents they will interact with but do not posses complete information about their neighbors' types and strategies so they base their decisions on expectations and beliefs. We characterize the existence, uniqueness and multiplicity of equilibrium distribution of strategies. By considering arbitrary interaction topologies, we show that the interaction structure greatly determines the uniqueness and multiplicity of equilibrium outcomes, as well as the equilibrium aggregate distribution of strategies as measured by the mean strategy.