近年來興新市場出色的表現,使得投資者高度重視並積極參與投資。因為風險與報酬是同等重要的,如何衡量風險成為重要的課題。現今風險值已經成為衡量風險的重要指標,然而波動的形態在新興市場與已開發市場迥異,單純應用已開市場的風險值經驗可能導致偏誤。本篇研究除了比較現存風險值方法,並引用較新的核函數估計法(Kernel Estimator Approach),應用在新興市場與已開發市場的股價指數上的表現。 本篇論文主要的結論,整體而言衡量風險值方法在新興市場的表現,相對而言較已開發市場差。若依照回朔測試以及無條件覆蓋比率檢定而言,核函數的方法表現最為優異,但以常態性為基礎的參數方法會因為報酬有高峰態與偏態的特性,將會低估風險值使得整體而言表現較差,條件自回歸風險值法(CAVaiR)中以適應模型表現相對較其他模型佳。若依據獨立覆蓋比率檢定結果,則會得到非常不同的結果,在無條件覆蓋比率檢定下結果有有較佳結果的方法,在獨立覆蓋比率檢定下則表現不佳;結果較差的在這檢定下有較佳的結果。此外在此獨立性檢定下參數法估計期間較短有較佳的結果與無條件覆蓋比率檢定相反。因此,風險值估計的結果,主要還是在於參數的假設、檢定、以及波動性的形態。
The performance of emerging markets makes investor emphasize them in recent years. Because risk is as important as return, how to measure becomes an important subject. Value at Risk has become an important indicator in risk measurement. However the patterns of volatility of emerging markets are different form developed market. Risk Measurement of emerging market might cause some biases for misusing the experiences of risk measurement in developed market. This study compares the performance of current VaR approaches in stock indices between developed and emerging markets. Besides, it utilizes recently approach – Kernel estimator approach. The main conclusion of this study is that VaR approaches in emerging market perform worse than those in developed markets. According to back testing results and unconditional coverage tests, Kernel estimator approaches perform outstanding. Parametric approaches basing on Normality will underestimate VaR because of skewness and kurtosis and adaptive model perform better than other CAVaiR approaches. However in independent coverage test, we get extremely different results. Some approaches that meet unconditional coverage test well do not meet independent coverage test well. Besides, parametric approaches basing on shorter estimation period meet the test well in independent coverage test but it does worse in unconditional coverage test. Therefore, VaR estimation results are extremely depended on parameter setting, focusing test and patter of volatility.