Using a model without conflicts of interest and with identical information available to equity analysts, we show that bias and herding in their stock recommendations occur due to incentives provided by relative performance evaluation and top awards. Furthermore, these incentives also lead to dispersion of recommendations. In particular, and contrary to commonly held views, high dispersion is more likely to arise for stocks with low volatility, for which bold recommendations increase chances of attaining top analyst status. Our empirical analysis supports this negative relationship between return volatility and recommendation dispersion, especially for large stocks, for which less information asymmetry among analysts is likely.