隨著金融市場的開放與自由化，金融商品間的關聯性日益密切；使得在金融市場中如何建立一個完善的風險管理評價機制中，成為目前相當為重要的課題。目前國內外許多研究發現，資產之報酬通常並不符合常態分配，且報酬是具有高峰及厚尾的現象，因此，如以常態分配來推論的話，可能會造成偏差的結果。本研究主要是以GARCH、EGARCH與GJR-GARCH模型，分別假設誤差項為漸進常態分配、t分配與GED分配，來計算臺指選擇權之風險值；另外風險值的評估方面，分別採用失敗次數、失敗率與Kupiec(1995)之概似比檢定來檢定風險值之模型之適合度。最後本研究之實證結果如下:(1)三種模型中以EGARCH模型表現較佳，臺指選擇權報酬率之變異具有不對稱性的情形。(2)在比較三種分配狀態後發現t分配較能正確的捕捉資料之波動性。
With the liberalization and opening of financial markets, the relationships among financial commodities become much closer day-by-day. Therefore, how to establish a perfect measure to evaluate risk in the financial markets has become an important topic. From lots of domestic and foreign studies, we find that the returns of financial properties usually do not follow normal distribution, and the phenomenon of leptokurtic and thick tails do exist as usual. So, if researchers assume normal distribution to analyze equity returns, they would probably get some wrong results. The purpose of this research is to compare with the value of risk of TSE index options, while we apply the GARCH, EGARCH and GJR-GARCH model to analyze, with assuming that the error terms are asymptotic normal distribution, or student t & GED distribution individually. Besides, we adopt failure tests, proportion of failures test and LR test to examine the models of fitting. Finally, we get two conclusions as followings (1) Comparing with these three models, we find that EGARCH model is a better method. The returns of TSE index options appear asymmetry situations. (2) After comparing these three distributions, we find that the student t can more accurately capture the volatility of the data.
