國內股票型基金經理人選股能力研究 －Bootstrap分析法應用

Abstract

我們應用Bootstrap統計分析法，來檢定國內績效特出的（特別好或特別壞）共同基金經理人，其績效是否來自經理人自身操作的選股能力，還是單純來自於運氣。資料採用投資國內的57支股票型基金為研究對象，研究期間自1997年4月到2007年3月，共10年。首先利用Carhart四因子模型迴歸出個別基金的 值與殘差項；接著透過對 值排序找出表現特別優異與特別落後的基金進行Bootstrap分析法，針對其殘差項做重新取樣來檢定這些表現特出的 值究竟是來自經理人具有的選股能力，還是單純只是運氣比較好以及運氣比較差而已。我們的研究結果發現：（1）表現優異的 值，透過Bootstrap法重新檢定皆呈現顯著正值，表示這些基金經理人的確具備選股能力而並非只是運氣好。（2）表現優異的 值其顯著性皆較表現落後的 值高。（3）將研究期間分成前5年與後5年觀察，發現 值表現優異的基金在兩個時間區段重複出現的頻率較 值表現落後的基金高。

For studying the question, “can star fund managers really pick stocks?”, we apply a new bootstrap statistical technique to examine those funds with very outstanding and very poor alpha. We choose the 57 domestic equity mutual funds as research target over the 1997 to 2007 period. To begin with the Carhart four-factors model, we find out the alpha and the residuals of each fund. For those funds with very outstanding （or very poor） alpha, through bootstrap analysis we examine if those alphas originate from the fund managers’ stock-picking ability or merely luck（bad luck）. Our bootstrap approach uncovers findings which show that（1）all those outstanding alphas are significantly positive which imply the outstanding performances do not come from merely luck but from fund managers’ stock-picking ability；（2）the outstanding alphas are more statistically significant than the poor alphas；（3）by dividing the research period into two parts, each 5 years, the funds with outstanding alpha have higher frequency of appearing in both parts than the poor alpha funds have.