Herding Behavior: Explanations and Implications

Abstract

This paper empirically tests theoretical explanations for herding behavior in an earnings forecast context. An experimental setting was utilized to capture the complex dynamics forecasters generally experience and isolate the herding behavior phenomenon. This approach is uniquely suited to study herding behavior as it allows the amount and type of information to be controlled, making it possible to distinguish between herding behavior and alternative explanations for a correlation among observed forecasts. Three variables, reputation concern, consensus forecast credibility, and forecast ability affected herding behavior. Subjects with greater concerns about their reputations and consensus forecast credibility exhibited higher levels of herding behavior. Forecast ability was inversely related to the level of herding behavior. In addition to the insights this study offers into earnings forecast behaviors, implications are derived for aggregate market behavior, such as, the forecast bias observed in archival studies.

JEL Classification: M41, G29