基于EVT-POT-SV-MT模型的极值风险度量

针对金融资产收益的异常变化,采用SV-MT模型对风险资产的预期收益做风险补偿并捕捉收益序列的厚尾性、波动的异方差性等特征,将收益序列转化为标准残差序列,通过SV-MT模型与极值理论相结合拟合标准残差的尾部分布,建立了一种新的金融风险度量模型——基于EVT-POT-SV-MT的动态VaR模型.通过该模型对上证综指做实证分析,结果表明该模型能够合理有效地度量上证综指收益的风险.

Extreme Risk Measurement Based on EVT-POT-SV-MT Model

VaR(Value at Risk) is an important method of risk management,which is used to measure the maximum possible loss of a portfolio in a selected confidence level at a given period of time in the future.The primary purpose of this method is to describe the volatility of financial time series as accurate as possible.Analysis methods are primarily used to predict management risk in VaR.The accuracy of the analysis result is mainly reflected in the financial asset of the error term distribution settings and volatility forecast.Because of the variability of financial markets,the traditional non-conditional normal distribution assumption is no longer applicable,and the conditional normal distribution or conditional distribution which is more characterized by fat tail is more in line with actual market volatility and return distribution. The relationship between the risk and expected rate of return is an important part of the modern financial theory.Many modern asset pricing models are based on the assumption that investors are risk-averse.Risk-averse investors holding risky assets require appropriate risk premium.Because the risk of a risky asset can be measured by the variance of returns,the risk premium is the increasing function of the conditional income variance.The traditional financial theory and statistical methods designed to deal with the extreme financial market volatility have many kinds of flaws.More scholars begin to try to introduce extreme value theory to financial risk management. Although extreme value theory does not have to make assumptions about the distribution of asset returns,the theory fits the tail of data distribution directly.As such,the theory can deal with heavy-tailed phenomena more effectively.VaR measurement based on the extreme value theory for the risk of financial assets are concentrated mostly in the GARCH model under the framework of the distribution function of standardized residuals.The application of SV model to depicting the financial timing is mostly limited to the normal VaR estimation method and does not consider the end characteristics of the assets return. Facing with the characteristic of thick tail and fluctution heteroscedasticity about financial assets,the article uses SV-MT models giving risk compensation for expected return on risky assets and combines with the Extreme Value Theory to establish a new financial risk measure model——the dynamic VaR model based on EVT-POT-SV-MT.SSE Composite Index. Our empirical analysis results show that the dynamic VaR model based on the integration of the SV-MT model and Extreme Value Theory is rational and effective in measuring earning risks of SSE Composit Index.