Essay on Analyst Herding

This study investigates a dynamic model of analyst forecasting where the ordering of forecasts and analysts' information endowments are endogenously determined. Analysts are probabilistically informed, potentially biased, and can increase their informedness through information acquisition. I characterize the unique equilibrium which holds for general distributions. The results show that analysts with less bias, greater precision, or a greater likelihood of being informed forecast earlier. Moreover, the main results show (perhaps surprisingly) that analysts always choose to be imperfectly informed, even though information acquisition is costless. This arises from the incentive to induce more timely forecasting by the other analyst. Likewise, analysts choose a positive bias level in equilibrium in order to gain a strategic advantage in their forecast timing. I discuss a number of empirical implications and extend the model to allow analysts to learn over time