Dynamic Forecasting Behavior by Analysts: Theory and Evidence

We examine the dynamic forecasting behavior of security analysts in response to their prior performance relative to their peers within a continuous time/multi-period framework. Our model predicts a U-shaped relationship between the boldness of an analyst's forecast, that is, the deviation of her forecast from the consensus and her prior relative performance. In other words, analysts who significantly out perform or under perform their peers issue bolder forecasts than intermediate performers. We then test these predictions of our model on observed analyst forecast data. Consistent with our theoretical predictions, we document an approximately U-shaped relationship between analysts' prior relative performance and the deviation of their forecasts from the consensus. Our theory examines the impact of both explicit incentives in the form of compensation structures and implicit incentives in the form of career concerns, on the dynamic forecasting behavior of analysts. Consistent with existing empirical evidence, our results imply that analysts who face greater employment risk (that is, the risk of being fired for poor performance) have greater incentives to herd, that is, issue forecasts that deviate less from the consensus. Our multi-period model allows us to examine the dynamic forecasting behavior of analysts in contrast with the extant two-period models that are static in nature. Moreover, the model also differs significantly from existing theoretical models in that it does not rely on any specific assumptions regarding the existence of asymmetric information and/or differential analyst abilities.