本研究探討各個基金特徵與基金績效持續性的關係，為避免樣本數目不足或樣本期間、期長差異影響實證結果的穩健性與一般性，納入2001年初所有可用的台灣共同基金母體資料。因為績效指標種類的不完全，或是傳統指標可能涉及的市場投資組合定義問題，本文也同時計算立基於CAPM之傳統指標群，和不需用到市場投資組合的 PET 指標群，希望釐清該兩項問題對台灣基金績效持續性結論可能的影響。最後，本文依「投資特徵」與「組織特徵」兩類變數做進一步基金分類，探討不同特徵與績效持續性的關係，並驗證相關文獻提出的想法或假說，探究台灣共同基金績效持續性背後的因素。由各項績效指標在各長短窗期期長與各潛在變數下的實證數據，本文發現：(1)「市場投資組合」定義的差異應不至於使基金績效的持續性顯著虛增，惟不同構面下績效持續性程度有異；(2) 台灣共同基金「風險」與「風險調整」構面的績效持續特性穩定且顯著，這和基金的交易標的、投資地區、風險、歷史存續期間長短等因子有顯著關係；(3) 台灣共同基金績效持續特性，和基金報酬率等級、經理人在位日數長短、以及成立規模等變數沒有直接關係。

This study explores the persistence in performance for Taiwan’s mutual funds. Contrast with contemporary papers in the related fields, our broad based study adopts the complete sample set for Taiwan’s mutual funds. Furthermore, we include a much richer set of performance/style indicators. Moreover, we present tests with various holding period horizons to enhance the robustness. Such settings enable us to gauge the possible disturbance from inefficient benchmarks and to look into the performance persistence in a more comprehensive way. With the characteristics of mutual funds being controlled, we examine the relationship between performance persistence and the intuitional variables, testing the hypotheses and notions raised by prior researchers. Our empirical results support the following notions. First, despite that the persistence in fund performance differs for different dimensions, the selection of the market portfolio proxy does not appear to bias the findings towards the persistence hypothesis. Second, the “risk” as well as the “risk-adjusted measure” appears to be significantly persistent during the sample period. Specifically, persistence in performance appears to be associated with target investment category, target market segment, risk level, and age of the funds. Third, sample period fund returns, durations of fund managers’ tenures, and original fund size appear to have meager explanatory power to the persistence measure.