Testing the conditional CAPM using multivariate GARCH-M

Publisher Website
Google Scholar
Abstract
The relation between expected return and time varying risk on the Swedish stock market for the period 1977 to 1990 is examined. Using a parsimonious multivariate GARCH-M model, the conditional Sharpe - Lintner - Mossin CAPM is tested against six alternative hypotheses, including the zero-beta version of CAPM, a conditional residual risk model, and models which nest the international CAPM and the consumption CAPM. The hypotheses are tested using beta-ranked, size-ranked, and industry-sorted portfolios. The estimates for the null hypothesis show that the price of risk is positive and significant for all portfolio groupings. Using robust LM-tests, the null hypothesis cannot be rejected in favour of any of the alternative hypotheses. In contrast to international evidence, where the traditional CAPM very often is rejected in favour of asset pricing models that rely on more general measures of risk, these results provide strong support for the Sharpe - Lintner - Mossin version of the conditional CAPM.