Information content of analyst stock recommendations, the

We investigate the relation between analyst stock recommendations and eight concurrently available variables that have predictive power for stock returns. We find that analysts generally pay little attention to the large sample predictive attributes of these variables. In seven out of eight cases, analysts’ stock recommendations are directionally opposite to the variable’s normative usage in returns prediction. In general, analysts exhibit a strong bias in favor of glamour stocks with growth characteristics. Despite this general bias, analyst recommendations have incremental predictive power for future returns. In fact, after controlling for the other predictive variables, the predictive power of the level of the analyst recommendation increases. These findings suggest that analyst stock recommendations contain information that is largely orthogonal to the information in the other predictive variables. We discuss the implications of these results for analysts, and for investors who rely on their recommendations.