Asset Pricing Under Prior-Induced Beta Uncertainty

We show theoretically that when Bayesian investors face time-series uncertainty about assets' risk exposures, differences in their priors affect the pricing of risk in the cross-section: different priors for the same asset can generate differences in perceived risk exposures, and thereby differences in required returns. The main testable implication is that the relation between required return and risk factor betas is steeper under a low-beta prior than under a high-beta prior. Using novel proxies for investors' priors about assets' exposures to risk factors, we find strong empirical support for our main prediction. Our results have important implications for understanding how prior-induced parameter uncertainty affects asset returns