Empirical studies on the mean/variance efficiency of levy and roll (2010) in the korean stock market

Abstract

While the numerous studies have shown that the mean/variance efficiency of various market proxies does not hold empirically, thus casting doubt about the capital asset pricing model (CAPM), the study of Levy and Roll (2010) adopts a reverse approach: given a market proxy, they apply minimal variations to sample parameters such that they are within statistical error bounds and at the same time make the proxy mean/variance efficient. This thesis follows the same approach and finds that the same phenomenon holds regardless of specific market condition - namely, the Korean stock market. All the tests were performed - univariate and multivariate tests and the comparison with Gibbons, Ross, and Shanken (1989) test, which all show a largely consistent picture with the original paper thus yielding the same implications that many conventional market proxies are useful for estimating expected returns.