PMPT 기반 성과지표를 이용한 펀드 투자전략의 성과평가

We estimated the performance of mutual funds using performance indicators based on the post modern portfolio theory (PMPT indicators). PMPT indicators can segregate downside risk from total risk unlike performance measures based on the modern portfolio theory (MPT indicators). We diagnosed the validity of these indicators by evaluating portfolios constructed by MPT and PMPT indicators. The main results are summarized as follows. First, more than 90% of domestic mutual funds rejected normal distribution hypothesis in the Shapiro-Wilk test. This suggests that mutual funds composed of various asset classes and portfolios have high skewness and kurtosis risk that are not captured by normal distribution. Therefore, we applied NIG(negative inverse Gaussian) distribution to estimating PMPT indicators. NIG distribution is known to capture high-order moment risk. Second, we found high within-group correlation and low between-group correlations for MPT and PMPT indicators. This implies that fund performance estimation can be very different according to risk attitudes of fund investors and investment styles of fund managers. Third, we constructed zero cost portfolio by short selling funds with high performance indicator and buying funds with low performance indicator. These portfolios showed the high-quality of distribution property and outperformed KOSPI index in terms of MPT and PMPT indicators. Fourth, we conjecture that the reversal of MPT and PMPT performance indicator is related with portfolio pumping behavior of fund managers. In other words, fund managers have reasonable incentive to temporarily inflate the net asset value of mutual fund through window dressing because portfolio management ability is evaluated by net asset value (NAV) at the time of evaluation. We conclude that this incentive induces fund managers to have short-term myopia and causes ex-post poor performance after the evaluation period.