Popis: |
The usual cross-sectional tests of asset pricing models suffer from lack of power because they do not impose the null hypothesis of zero pricing errors on a full set of test assets. This paper proposes a simple remedy using full-rank maximum correlation portfolios obtained by adding extra return bases to test assets, allowing the cross-sectional tests to be performed with full degrees of freedom. I show that the resulting test statistic is equivalent to the Hansen-Jagannathan distance with an expanded set of returns. The statistical test has more power and rejects most of the recent asset pricing models in the literature. |