An asset pricing model that corrects for variance expectation bias

Under conditions of predictably-changing variances, traditional estimations of the CAPM beta are biased. This dissertation introduces an estimation procedure that endogenizes both the expected changes in the variance of asset excess returns and the expected changes in the variance of market index excess returns. This dissertation derives, estimates, and tests the resulting predicted betas, then explores the implications for existing market anomalies. The predicted betas are statistically significant in modeling out-of-sample stock excess returns for January and non-January samples in cross-sectional regressions, even when controlling for the level of traditional betas.