An Analysis of Stock Recommendations

Abstract

We analyze the information content of stock recommendations by a sell-side equity analyst when investors are uncertain about the analyst's incentives. In our model, an analyst can either be "unbiased", having incentives that are congruent with those of the investor, or "biased", having incongruent incentives. We find that investor uncertainty about an analyst's incentives dramatically reduces the information content of stock recommendations. Specifically, when the investor cannot infer the analyst's bias simply from his recommendation, all perfect Bayesian equilibria of the game are equivalent to partition equilibria - equilibria where the analyst's recommendations correspond to a finite partition of the (continuous) state space. Partition equilibria are consistent with the traditional three-category system (buy, hold, and sell) frequently employed by analysts. These equilibria also have the properties that: (a) the possibility that an analyst is biased eliminates the incentives of even unbiased analysts to reveal all relevant information in their recommendations; (b) both types of analysts issue overly-optimistic recommendations, but the degree of over-optimism is greater for biased than for unbiased analysts; and (c) despite the information loss arising from uncertainty about the incentives of analysts, stock recommendations do contain some information and hence price responses to recommendations are consistent with rational investor behavior.