基于已实现极差的上证综指波动长记忆性识别与风险度量研究  

RRV-Based Research of Identifying Long Memory Property and Risk Measurement for Shanghai Composite Index Volatility

作　　者：周文浩 王沁 张红梅 汪玲 ZHOU Wenhao;WANG Qin;ZHANG Hongmei;WANG Ling(College of Mathematics and Statistics,Southwest Jiaotong University,Chengdu 611756,China)

机构地区：[1]西南交通大学数学学院统计系,成都611756

出　　处：《西南大学学报（自然科学版）》2021年第11期142-150,共9页Journal of Southwest University(Natural Science Edition)

基　　金：国家自然科学基金项目(71671145);教育部人文社会科学研究基金项目(17YJC790119).

摘　　要：基于上证综指的高频信息,计算了已实现极差.针对已实现极差,进行R/S检验和单位根检验,发现已实现极差具有长记忆性.考虑到杠杆效应、异方差性和长记忆性,构建了基于已实现极差的ARFIMA-M-FIGARCH模型、ARFIMA-M-FIEGARCH模型和ARFIMA-M-HYGARCH模型,进行了拟合优度检验、VaR的失败率检验以及VaR的动态分位数检验,结果表明,基于已实现极差的ARFIMA-M-HYGARCH模型拟合效果最优,刻画了上证综指的双长记忆特征、杠杆效应和尖峰厚尾的特征,能更好地度量极端情况下的在险价值VaR.Realized range variance(RRV)is computed with the high-frequency data sampling from Shanghai Composite Index.R/S test and unit root test are carried out to prove its long memory property.Considering leverage effect,heteroscedasticity and long memory property,RRV-basedARFIMA-M-FIGARCH,ARFIMA-M-FIEGARCH and ARFIMA-M-HYGARCH models are established.Goodness-of-fit test,failure rate test and dynamic quantile test are used to evaluate the models.Results show that ARFIMA-M-HYGARCH,which reflects dual long memory property,leverage effect and“peak fat tail”property of Shanghai Composite Index,is better fitted and measures VaR(value at risk)more precisely in extreme cases.

关 键 词：已实现极差 双长记忆 R/S检验 ARFIMA-M-HYGARCH模型 在险价值 

分 类 号：F830.91[经济管理—金融学]