In the presence of dispersed information, agents may decide to take into account the actions of other agents because of the possible additional information conveyed by these actions. We call the act of using other agents’ actions in the individual decision process social learning. This paper argues that social learning aimed at increasing the precision of individual information may lead to aggregate fluctuations. We consider a setting where firms receive independent noisy signals about a common fundamental and can observe other firms’ actions through a network of informational links. We show that, when firms can observe each other’s decisions, they are able to increase the accuracy of their actions. While reducing volatility at the individual level, social learning may lead to an increase in volatility at the aggregate level depending on the network topology. Moreover, if the network is very asymmetric, aggregate volatility does not decay as predicted by the law of large numbers.
business cycle social networks dispersed information aggregate volatility