Research of VaR on Taiwan Financial Futures and Commodity Futures

• Main Authors: Huan-Jen Lai, 賴煥仁
• Other Authors: Chui-Chun Tsai
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/80054786436637543433

碩士 === 靜宜大學 === 會計學系研究所 === 96 === Be derivative financial instrument in the flourish development of the world, The whole futures trade market of Taiwan trading volume 1895 volumes on average day from1998 TAIEX Futures after listing, go through nearly ten tears development, already there are 136709 volumes till October of this year, Observe the internal market trade volume more have scale placings of Finance Sector Index Futures, TAIEX Futures、Electronic Sector Index Futures、Finance Sector Index Futures、MiNi-TAIEX Futures valuating with Taiwan dollar, and adopt the TAIFEX MSCI Taiwan Index Futures that U.S. dollar valuates relatively representative；Commodity Futures only have one TAIFEX Gold Futures that U.S. dollar valuates. So this study division internal futures to Finance Sector Index Futures and Commodity Futures, and the trading volume is greater than five kinds of indexes futures and TAIFEX Gold Futures of 100 volumes at least on average every day in internal as the research object ,among them, TAIEX Futures、Electronic Sector Index Futures、Finance Sector Index Futures、MiNi-TAIEX Futures are extracting from January 2,2004；TAIFEX MSCI Taiwan Index Futures and TAIFEX Gold Futures listed relatively late, so since listing for March 27, 2006, daily closing price so far by November 30, 2007, after data obtaining, respectively apply the Jarque-Bera Value、the ADF t-value of Liner Regression and Non- liner Regression、LM Q-Value, with the closing price of normality、constancy and AGCH effects to test. The result of experiment detect, through the ADF test of Liner Regression and Non- liner Regression、Normal test and Homogeneity, rate of return by first difference is stationary, However, the rate of return are all non-normal and its variance is Heterogeneity, accordingly, this study estimates with QMLE method, and at the same time the variance-covariance matrix variance disassemble the skill to carries on the model estimate which introduce by Bollerslev & Wooldridge(1992),The conclusion detect：(1)The dual comparison result on AIC value and SBC value with TGARCH and EGARCH comparatively surpasses in Linear Regression ARCH and GARCH; (2)Further learn from ARCH, GARCH, EGARCH and TGARCH model coefficient, the current period risk mainly receives the previous period risk、previous period error、previous period do not expect residual of large margin change 、bad news, and the effect by previous period error change intensity; (3)Bring the rate of returns into ARCH in all kinds of models, most results of calculation of value at risk obtain the value by the TGARCH model and the EGARCH model to be lower; (4)And because of the internal futures earnest money calculate the foundation is based on 99% of the risk value confidence level, in Back Testing, With the proportion of failure test put forth by Kupiec(1995), when proportion of failure test measures by Penetration ratio, The proportion of failure of TGARCH model is also lowest, and relatively conform with the theory and set up the data 1%, The TGARCH model is applied to the valuation of internal futures market VaR seem to be better than other modes.