Tests of conditional mean-variance efficiency of the U.S. stock market

Abstract

We test the mean-variance efficiency (MVE) hypothesis using a method that allows conditional expected returns to vary in relatively unrestricted ways. The method takes advantage of the predictability of conditional variances. The data estimate reasonably the price of risk, and the MVE model is valuable in explaining expected equity returns. Nevertheless, we reject the restrictions imposed by MVE on the alternative hypothesis. Unlike with most tests of MVE, we can put an explicit interpretation on the alternative hypothesis — a general linear Tobin portfolio choice model.