無資料窺探偏誤的檢定評估共同基金績效
Other Title
Testing the Performance of Taiwan Mutual Funds Based on the Tests without Data Snooping Bias
Journal
證券市場發展季刊
Journal Volume
22
Journal Issue
3
Start Page
181
End Page
206
ISSN
1023-280X
Date Issued
2010-10
Author(s)
莊惠菁
Abstract
本文採用新的計量檢定方法來評估臺灣共同基金的績效是否具有統計上的顯著性。為避免在實證方法中常見的資料窺探（data snooping）所造成的推論錯誤，我們採用Hsu，Hsu，and Kuan（2010）的逐步檢定法來認定具有顯著績效的基金。實證結果顯示，臺灣股票型共同基金於2002到2007年間，若以基金月報酬均數為績效評估標準時，沒有任何基金顯著的高於臺股加權指數等的大盤報酬，而只有一至二支基金的夏普值顯著的超越大盤之夏普值。當比較標準為三因子模型之報酬時，只有不到百分之三的基金存在顯著的異常報酬。
Data snooping is common in empirical finance studies. In this paper, we adopt the SPA test of Hansen (2005) and the stepwise SPA test of Hsu, Hsu, and Kuan (2010) to examine the significance of fund performance. These tests are free of data snooping bias. Our empirical study shows there is no fund significantly outperforms the monthly returns of Taiwan stock index, MSCI and Taiwan 50 indices. Only 1 to 2 mutual funds have significantly higher Sharpe ratios than these indices. Moreover, there are less than 3% funds that have abnormal returns in Fama and French's three factor model.
Subjects
共同基金
真實性檢定
SPA檢定
資料窺探
逐步檢定法
mutual fund
reality check
SPA test
data snooping
Type
journal article