A Bayesian panel data approach to explaining market beta dynamics

We characterize the process that drives the market betas of individual stocks by setting up a hierarchical Bayesian panel data model that allows a flexible specification for beta. We show that combining the parametric relationship between betas and conditioning variables specified by economic theory with the robustness of an autoregressive specification delivers superior estimates of firm-specific betas. Our model also improves the accuracy of betaforecasts, which we use to construct optimal portfolios subject to target beta constraints.We further provide empirical support for the prediction of conditional asset pricing theory that individual stocks exhibit different risk dynamics. Finally, we document strong cross-sectional heterogeneity in firm-specific betas within the 25 size-B/M portfolios that are commonly used to test asset pricing models.