An Empirical Investigation of Value at Risk of Taiwan Stock Indexes of Spot, Futures, and Call Options: Using Tri-EGARCH Models

• Main Authors: HSU-MEI YUM, 徐美雲
• Other Authors: GOO,YEONG-JIA
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/46977012921997440234

碩士 === 國立臺北大學 === 企業管理學系 === 95 === Previous researchers usually use GARCH models in estimating volatility for evaluation of value at risk (VaR) performance. There are very few academic articles talked about that interaction with assets would affect the VaR on these assets. A 744 daily data of Taiwan stock indexes of spot, futures, and call options ranging from Jan. 1, 2004 to Dec. 31, 2006 were collected. In this study, Tri-EGARCH models were adopted in estimating volatility. The estimated volatility is then estimated VaR using Monte Carlo Simulation method. We established a control model to estimate VaR which simply using Monte Carlo Simulation method. The VaR performances of these two models are compared using back-testing and Lopez loss function. There are four purpose of this study. First, using Tri-EGARCH analyzes three markets to discover how interaction affect rate of return in these three markets. To apply conditional variance equation describes volatility clustering and volatility spilling effect in these markets which is second purpose. Third purpose is to confirm that asymmetric effect existence whether or not. Forth purpose, the main purpose, is to establish estimating method using Tri-EGARCH and Monte Carlo Simulation method, and to confirm its’ performance. The result which is research in how interaction affect rate of return in these three markets is showed that spot market will be positive affected by futures about one day ago. Futures will be positive affected by spot and itself about two days ago. Call option will be negative affected by itself about one day and two days ago, others are not significant. And the result which is research with second purpose is showed that three markets all have volatility clustering and volatility spilling effect. The research into the effect of asymmetric effect, we find the effect is opposite of Tri-EGARCH model and single EGARCH model using in these three markets. The result on model of estimating VaR is showed that Tri-EGARCH model estimating performance is the best whether in back-testing or Lopez loss function.