Herd behavior and the quality of opinions

This paper analyzes a decentralized decision model by adding some inertia in the social leaning process. Before making a decision, an agent can observe the group opinion in a society. Social learning can result in a variety of equilibrium behavioral patterns. For insufficient ranges of quality (precision) of opinions, the chosen stationary state is unique and globally accessible, in which all agents adopt the superior action. Sufficient quality of opinions gives rise to multiple stationary states. One of them will be characterized by inefficient herding. The confidence in the majority opinion then has serious welfare consequences.