平稳序列BMM模型与沪深股市极端风险研究

The Research of BMM Model for Stationary Sequence and Extreme Risk of Shenzhen and Shanghai Stock Market

基于VaR理论正态分布假设导致的尾部风险低估问题,研究了GEV分布下的BMM模型及区间关联下的极值VaR的建模,并实证分析了沪深股市极端风险.研究结果表明:BMM模型对金融风险的厚尾具有更合理的理论基础.然而,涨跌停板极大地抑制了沪深股市极值数据的异质性,形成"极值不极"现象,导致在较高置信度下BMM模型更为有效,而在较低置信度下反而存在低估问题,有效性尚不及VaR模型.

To revise the shortage of model VaR under the assumption the sequence is nor- mal distribution, this paper employs BMM model to estimate extreme risk, and develops the formula for calculating Extreme-VaR, based on which the extreme risk in Shanghai and Shen-zhen stock market is explored. The results show that BMM model is more effective than VaR model to study sequences with fat tail. However, the raising limit restrains the heterogeneity of extreme values, which result in underestimation for BMM model under the condition of lower confidence level.