Beta Reversal and Expected Returns

In this paper we show that the failure of the CAPM beta to predict individual stocks' expected returns documented by Fama and French (1992) is largely driven by a small group of stocks with large betas and high idiosyncratic volatilities. These stocks' betas tend to reverse. Therefore, even when the CAPM holds period-by-period, the cross-sectional evidence on market beta is weak at best due to the confounding effect of beta reversal and instability. We further show that such a beta reversal is partly predictable by idiosyncratic volatility. As a result, the current beta estimates of individual stocks can significantly explain the cross-sectional differences in future returns whit a simple control for such a reversal effect. In fact, the market risk premium estimated from cross-sectional regression is close to that of the historical average. All results are robust with respect to different measures of beta and idiosyncratic volatility as well as different subsamples. In addition, we explore several possible causes for the beta reversal phenomenon.