Learning Dynamics in Social Networks

Simon Board (University of California - Los Angeles)

March 11, 2019, 14:00–15:30

Room MS 001

Industrial Organization seminar


This paper proposes a tractable model of Bayesian learning on social networks in which agents choose whether to adopt an innovation. We study the impact of network structure on learning dynamics and diffusion. In tree networks, we provide conditions under which all direct and indirect links contribute to an agent’s learning. Beyond trees, not all links are beneficial: An agent’s learning deteriorates when her neighbors are linked to each other, and when her neighbors learn from herself. These results imply that an agent’s favorite network is the directed star with herself at the center, and that learning is better in “decentralized” networks than “centralized” networks.