Statistical Tests for Return-Based Style Analysis

Return-based style analysis investigates the exposure of mutual funds to a number of style indices. Because the style weights need to meet particular constraints, traditionally only point estimates of the style exposures have been reported. In this paper we include the entire asymptotic distribution of the style weights. These results are obtained by applying a combination of the Kuhn-Tucker optimization algorithm and standard bootstrapping. This allows us to infer confidence intervals for the style coefficients, and to carry out statistical tests on the parameters. Empirical tests on a sample of UK equity funds, demonstrate the usefulness of this extra information, in the light of the mutual fund misclassification phenomenon