| Summary: | 碩士 === 國立中興大學 === 財務金融系所 === 96 === The performance indexes of mutual funds have been developed for more than 40 years since Treynor’s index in 1965. Traditional indexes usually focus on the relationship between single risk variable and excess return only. The DEA indexes consider multiple inputs and multiple outputs compare to traditional indexes. However, DEA indexes can only measure relative performance rather than the absolute performance measurement as like traditional indexes can do.
In this paper, we compared the relative and the absolute performance measurement results of mutual funds in Taiwan. We compared performance results using four traditional indexes (Sharpe, Treynor, reward to half variance, Jensen) and four DEA indexes (DPEI, IDEA_1, IDEA_g, IDEA_g with VaR). The main differences among four DEA indexes are subject to different risk variable and take into account of traditional indexes in output variables or not. DPEI index using the annual standard deviation as risk variable, IDEA_1 index using both β coefficient and root of the lower half variance and IDEA_g considered traditional indexes in their risk measurement. Finally, ”IDEA_g with VaR” considered Value at Risk (VaR) as risk variable .
The main empirical results are concluded as follows:
1. By ranking the performance results of traditional versus DEA indexes separately, we found their results are not consistent between absolute and relative performance measurement.
2. We found the results keep consistently among traditional indexes and among DEA indexes.
3. We also found the more variables in the DEA model, the more efficient it will be. However, when variables number of DEA model exceed half of DMU numbers, the DEA indexes will became inaccurate.
4. Finally, we found there exists high correlation coefficient among four DEA indexes. It means that more variables in the DEA model is not necessary in performance measurement of mutual funds.
|
|---|
