台灣的金融市場隨著國際化的腳步開放與自由化,各金融商品間的關聯性日益密切;自90年代後期金融事件接踵而至,如何在金融市場中建構一個完善的風險管理評價機制中,成為相當重要課題。過去學者針對台灣指數股票型基金的研究,鮮少以風險值的方法分析之,而本研以台灣已發行之三檔指數股票型基金為例,運用歷史模擬法、RiskMetrics 模型、TGARCH 模型等三種衡量風險值之方法,試圖尋找其最適合估計台灣指數股票型基金之風險值模型。 本研究之實證結果發現: 整體而言:若就成功率來比較各模型,以歷史模擬法所計算出的成功率相對於其他模型較為適用;若以概似比檢定比較,則以TGARCH 模型所估計出來的失敗率較為接近顯著水準的假設,若以誤差均方根法而言,以歷史模擬法所計算出的 RMSE 值,相較於其他模型會有較小的結果。因此就歷史模擬法、 RiskMetrics 模型與 TGARCH 模型而言,對於台灣指數股票型基金報酬率之風險值的估計,似乎以歷史模擬法較為適合。 個別而言: 1.在估計寶來台灣卓越 50 證券投資信託基金之風險值時,以成功率來比較各模型時,在持有資產為 1 日、5 日、10日時,其值以歷史模擬法較佳。 2.在估計寶來台灣中型 100 證券投資信託基金之風險值時,以成功率來比較各模型時,若持有資產為 1 日時,其值以歷史模擬法較佳;而持有資產 5 日,其值以 TGARCH 模型較佳;持有資產 10 日,其值以 RiskMetrics 模型較佳。 3.在估計富邦台灣科技指數證券投資信託基金之風險值時,以成功率來比較各模型,若持有資產1日、5日、10日時,其值以歷史模擬法較佳。
With the opening and liberalization by internationalization in the financial market of Taiwan, it is more related and closer day by day among all financial goods. Since one after another financial incidents coming on later 1990s, it’s becoming a very important issue for how to build a perfect risk management and appraising system in the financial market. In the past, the scholars seldom researched the Taiwan ETF with VaR. There are three types index stock fund of Taiwan ETF as instances in this research. Using Historical Simulation , RiskMetrics model, and TGARCH model to find out the most suitable ways estimating the risk value model of Taiwan ETF. As the result from the research: For entirety:Comparing with the success rate of above models, using Historical Simulation is more suitable than other models. Comparing using RiskMetrics, using TGARCH model to estimate the failure rate is closer a standard of assuming. When using RMSE, the Historical Simulation will estimating smaller value of RMSE as other models. Therefore with previous section of these three models, the risk value model Taiwan ETF is more suitable on using Historical Simulation. For individual: 1.On estimating risk value of Taiwan Top 50 Tracker Fund (TTT): comparing with the success rate of above models, it’s better of using Historical Simulation when holding the assets for 1 day, 5 days, or 10 days. 2.On estimating risk value of Polaris Taiwan Mid-Cap 100 Tracker Fund: comparing with the success rate of above models, it’s better of using Historical Simulation when holding the assets for 1 day, greater for the TGARCH model of 5 days; and RiskMetrics model for 10 days. 3.On estimating risk value of Fubon Taiwan Technology Tracker Fund : comparing with the success rate of above models, it’s better of using Historical Simulation when holding the assets for 1 day, 5 days, or 10 days.