摘要
极值理论近来常用于股市极端风险测度,然而涨跌停板严格抑制了极端风险数据的变异性,不但影响了极值POT模型的有效性,并常常造成其阈值难以确定,而当前约束条件下的相关研究却还较为匮乏。本文利用极值理论POT模型对沪深A股及港H股的极端风险进行了测度比较分析,引入峰度法较好地解决了POT模型阈值难以定量选取的问题,实证结果表明,涨跌停板以上下尾部不等效的方式显著地抑制了极端风险水平,使得即使在较低置信水平95%下,POT模型依然比VaR模型有效,而VaR模型非但没有低估反而普遍存在高估现象。最后,本文还定量分析了涨跌停板制度对股市极端风险数据分布结构的影响。
Extreme Value Theory(EVT)has recently been applied to measure extreme risk in stock markets.However,Raising Limit restrains the heterogeneity of extreme values strictly,which may influence the validity of EVT model and make it more difficult to select the threshold accurately,but the research of EVT in the constraint conditions has not to be accomplished in the literature.This paper applies the method of Kurtosisr to take threshold,and measure the extreme risk of China A and H shares.The results indicate that Raising Limit reduce the extreme risk greatly,and has greater effects on POT model than VaR model.Even in the 95%confidence interval,POT model is still more effective than VaR model which overestimates extreme risk instead of underestimation.Finally,this paper analyzes the effect caused by Raising Limit based on the structure of extreme value distribution.
出处
《系统工程》
CSSCI
CSCD
北大核心
2015年第7期65-72,共8页
Systems Engineering
基金
国家自然科学基金资助项目(71171209)
国家社会科学基金资助项目(14BJY188)
关键词
极值理论
阈值模型
广义帕累托分布
涨跌停板
Extreme Value Theory
POT Model
Generalized Pareto Distribution
Prices