Clustering in communication networks with different-minded participants

Abstract

We consider a game with players who build links to exchange their opinions on the relevant state and reduce their uncertainty on it. Opinions transmit private signals with frictions (termed interpretation noise) because the players hold imperfectly correlated subjective priors on the state. Communication is limited to two rounds. For tractability, we focus on an arbitrarily small correlation of priors and allow for misspecified beliefs on a distant network. We find that player's uncertainty remaining after communication is given by and decreasing in: the number of his friends with only common connections, and the total number of his local and distant friends. The egalitarian efficient network is: a complete component if the interpretation noise is sufficiently high, and a flower otherwise. It constitutes a Nash equilibrium. We show numerically that these results may hold in general, if the correlation of players' priors is sufficiently weak. Furthermore, egalitarian efficiency criterion may be replaced with utilitarian one.

Keywords

JEL codes

Reference

Thibault Laurent, and Elena Panova, “Clustering in communication networks with different-minded participants”, TSE Working Paper, n. 20-1147, September 2020, revised August 2024.