基于EVT-POT-SV模型的极值风险度量

金融市场常受各种因素的影响造成剧烈波动,资产收益也会因此产生异常变化。针对金融资产收益的厚尾性、波动的异方差性等特征,采用基于Markov链的Monte Carlo模拟积分方法,对随机波动模型进行参数估计并取得标准残差序列,应用极值理论与随机波动模型相结合,建立了基于EVT-POT-SV的动态VaR模型。通过对上证综指收益做实证分析,结果表明：该模型能很好地刻画收益序列的波动性及尾部分布特征,在度量上证综指收益的风险方面更加合理而有效。

Extreme Risk Measures Based on EVT-POT-SV Model

Facing with the fat--tail proceeds and the heteroskedasticity characteristics ot volatility from financial assets return, the article uses Markov chain Monte Carlo method to estimate the parameters of the Standard SV model. At the same time, it transfers return series into standard residuals and uses EVT-- POT methed to capture the fat tails of standard residuals. Then, the paper construsts a new dynamic VaR risk measure baded on EVT--POT--SV and applies it to daily returns of composite index of Shanghai stock market. The empirical anylysis indicats that the risk measure can describe index return＇ s dynamic VaR risk more exactly and reasonably.