運氣好或操作策略好？－拔靴法下共同基金之績效衡量

Abstract

本文的重點在於導入拔靴法(bootstrap)解決國內基金普遍成立年限不足、觀察值過少、以及傳統方法在分配假設上的困擾。傳統的共同基金衡量方法在樣本期間不足及樣本數目過少的情況下，所衡量出之共同基金績效很有可能是因為統計誤差或運氣成分使然，本文則利用拔靴法來衡量共同基金績效時，主要是強調共同基金的績效必須要能高於運氣成分下的績效，才能算是真正的操作績效。本文指出，以傳統Jensen's α衡量下的共同基金沒有明顯的超額報酬，而在Carhart四因子模型下，台灣的共同基金存在有異常報酬。但是，拔靴法指出：Carhart模型下的共同基金異常報酬其實是運氣成分使然，而不能真正歸因於基金經理人的操作能力。

In this paper, we focus on a bootstrap approach to measure the mutual funds performance. Compared to the traditional mutual funds performance measures, a bootstrap approach can avoid several severe econometric problems such as short time horizon, small sample size, and the assumption of distribution. Due to the econometric problems, mutual funds performance measured by traditional measures could simply be attributed to luck or sampling bias. Our results show that Taiwan mutual funds outperform under Carhart four-factor model by traditional measurement. However, bootstrapping indicates that performance under Carhart model can be earned by chance and implies that fund managers do not perform by their superior investment strategy instead they perform by luck.