On selectivity and market timing ability of U.S.-based international mutual funds: Using refined Jensen's measure

Abstract

As shown in previous studies of international finance, it is well recognized that when there are no investment restrictions, diversifying into foreign capital markets reduces systematic risk significantly and provides investors with substantial gains. However, various kinds of restrictions on investments still exist and the cost of information in the international capital markets is high. One way for U.S. investors to achieve the potential gains from international diversification while minimizing the costs of information acquisition is by investing in the international mutual funds available in the U.S. market. There are approximately 50 internationally diversified mutual funds available to U.S. investors. Those international mutual funds have gained popularity not only due to the expanded foreign investment opportunities that they provide, but also due to the belief by investors that these funds outperform most of the domestic funds. Howeve& only a limited amount of research has been performed regarding international mutual fund performance (Cumby & Glen, 1990). Most of the studies regarding performance evaluation of mutual fund managers employ a method developed by Jensen (1968,1969). The Jensen measure of performance evaluation, based on the Security Market Line (SML), concentrates on a manager’s stock selection ability However, it is widely recognized that Jensen’s measure is subject to errors, especially when the manager possesses information about the market and thus engages in market timing activity Some studies have attempted to correct the bias in Jensen’s measure when timing activity exists. For example, Bhattacharaya and Pfleiderer (1983) show that an unbiased estimate is obtainable by a regression technique. Grinblatt and Titman (1989) attempted to separate selectivity and timing abilities of fund managers by proposing an alternative measure: the positive period weighing test. However, that measure is