此篇研究使用均等加權平均法、指數加權移動平均法、歷史模擬法、拔靴法及穩健指數加權移動平均法來測量21檔個別國家指數和12檔個別產業型指數股票型基金的風險值,並使用回溯測試及對數概似法來衡量模型的準確性。實證結果為在歐洲和美洲型基金分別以均等加權平均法和指數加權移動平均法表現較好,使用的歷史期間為250天,顯著水準為95%。亞洲型基金以指數加權移動平均法較好,在大洋洲和非洲則以歷史模擬法較好,使用的歷史期間和顯著水準分別是500天和95%。對產業型基金而言,建議使用250天的歷史期間和95%顯著水準的指數加權移動平均法來計算風險值。大致上來說,在95%水準下,參數法和非參數法的效果差不多。在99%水準下,則以非參數法的效果較好。
This paper use EQMA、EWMA、HS、BS and REWMA to find adequately methods to calculate value-at-risk measure of country-specific ETFs and sector-specific ETFs. And then, we adopt backtest proposed by the Basle Rule and LR statistic test to examine methods we used here. Our empirical results show that in European and American ETFs, we should use EQMA and EWMA respectively. Historical observations are 250 days and confidence level is 95%. In Asia, Pacific and African ETFs, performance of EWMA and HS is better. Using 500 days historical observations and 95% confidence level is best. As to sector-specific ETFs, performance of REWMA with 250 days historical observations and 95% confidence level is best. General speaking, performances of parametric and nonparametric are similar under 95% confidence level, but nonparametric method is better than parametric method when using 99% confidence level.