Structural Estimation of Pairwise Stable Networks: An Application to Social Networks in Rural India

Abstract

This paper studies what we can learn from pairwise stable networks. Recent literature on empirical models of strategic network formation confronts problems such as the curse of dimensionality and multiple equilibria. To solve these problems, I consider the probability that the observed network is pairwise stable, instead of the probability that a certain equilibrium outcome is observed. Pairwise stability provides conditions under which no pairs of individuals have an incentive to deviate from the current network configuration. Pairwise stability and the assumption of myopic agents contained in it give strong identification power when we consider the probability that the observed network is pairwise stable. I propose a semiparametric maximum score estimator which is simple and computationally feasible. I applied the empirical model to different types of social networks in rural India. Estimation results show that individuals have strong homophily on castes in all types of social networks.