Tests of the CAPM with Time-Varying Covariances: A Multivariate GARCH Approach



    Search for more papers by this author
    • Assistant Professor, Department of Finance, University of Texas at Austin. This paper is based on part of my dissertation at the Wharton School, University of Pennsylvania. I am very grateful to Simon Benninga, Yin-Wong Cheung, Campbell Harvey, Robert Litzenberger (Main Advisor), Craig MacKinlay, Krishna Ramaswamy, Robert Stambaugh, Rex Thompson, Seha Tinic, the editor Ren M. Stulz, and two anonymous referees for their helpful comments.


This paper examines an asset pricing model in which the Sharpe-Lintner CAPM and the zero-beta CAPM are special cases. The model allows the ratio of expected market risk premium to market variance, the conditional expected excess returns, and the risks to change over time. The results are found to be sensitive to the choice of the portfolio formation techniques. Significant time variability is shown in the conditional expected excess asset returns and risks and also in the reward-to-risk ratio.