Estimating security betas using prior information based on firm fundamentals

This paper proposes a novel approach for estimating time-varying betas of individual stocks that incorporates prior information based on fundamentals. We shrink the rolling window estimate of beta towards a firm-specific prior that is motivated by asset pricing theory. The prior captures structural changes in beta while the sample estimate picks up short-term fluctuations. Shrinkage is most important for small firms and companies with high book-to-market and leverage ratios. Our method sharply increases the accuracy of out-of-sample beta forecasts. We find that when beta is estimated more precisely, it is priced in the cross-section even after controlling for firm characteristics