FRB logo


Abstract: We examine the evidence on excess stock return predictability in a Bayesian setting in which the investor faces uncertainty about both the existence and strength of predictability. Departing from previous studies, we allow the regressor to be stochastic. When we apply our methods to the dividend-price ratio and payout yield, we find that even investors who are quite skeptical about the existence of predictability sharply modify their views in favor of predictability when confronted by the historical time series of returns and predictor variables. We find that taking into account the stochastic properties of the regressor has a substantial impact on the investor's inference about returns.

Keywords: Equity premium, return predictability, Bayesian methods

Full paper (451 KB PDF) | Full paper (Screen Reader Version)