Keeping an Eye on the Competition Observational Learning in Analysts’ Forecasts

Abstract

Substantial research shows the importance of different network structures in the process of learning from others. Learning in most of these networks takes place when knowledge is transferred via social or business ties. I argue that network structure matters differently for observational learning, defined as an arm’s length process of observing and incorporating information from others who are unable or unwilling to transfer knowledge. I argue that, even in the absence of sociality and collaboration, closed networks improve decision makers’ performance by facilitating observational learning. Closed networks improve performance by focusing decision makers’ attention on relevant sources of information and providing context that is useful for interpreting what is observed. I find that security analysts’ earnings forecasts are more accurate when their stock coverage network is closed. This effect is robust only for analysts who reacted to the opinions their competitors published, which supports observational learning as the underlying mechanism. Learning from competitors’ forecasts is further supported using a natural experiment. I find that the sudden absence of analysts who perished in the September 11, 2001 attacks on the World Trade Center reduced survivors’ forecasting accuracy as a function of the damages to their coverage network.