Nagoya City University

The Effects of Nonnormality on the Market Model in the Class of Elliptical Distributions

Email address: hodoshi@econ.nagoya-cu.ac.jp

Keywords: Mean-variance efficiency, market model, nonnormality, scale mixture of normal distributions, elliptical distributions

JEL Classifications: C51 C52 G12

Abstract:
This paper considers the effects of nonnormality on the statistical inference of the market model when the joint distribution of asset returns and a given portfolio return is elliptically distributed. The standard covariance matrices derived under notmality for the least squares estimators of alpha and beta are shown to overstate the accuracy of the least squares estimators whenever the joint distribution is in fact nonnormal in the class of elliptical distributions satisfying the representation of scale mixtures of multivariate normal distributions. Therefore, non-robustified tests of mean-variance efficiency, constructed with the standard covariance matrix derived under normality, result in rejecting the mean-variance efficiency hypothesis too often whenever the joint distribution is in fact nonnormal in the same class.

PDF file of paper: hodoshima.pdf

Session: Asset Prices I

Time: Saturday, 7 July, 2:15pm - 3:45pm

Room: E