基于高频数据的沪指波动长记忆性驱动因素分析  

Long Memory-Driven Factors of Volatility in Shanghai Complex Index Based on High Frequency Data

作　　者：张波[1] 钟玉洁[1] 田金方[1,2] 

机构地区：[1]中国人民大学应用统计科学研究中心,北京100872 [2]山东经济学院统计与数学学院,山东济南250014

出　　处：《统计与信息论坛》2009年第6期21-26,共6页Journal of Statistics and Information

基　　金：教育部人文社会科学重点研究基地重大项目《基于高频数据的中国金融市场微观结构研究》(07JJD910244);国家自然科学基金项目《倒向随机微分方程,非线性数学期望及其应用研究》(10771214)

摘　　要：借助于高频数据的最优取样法,利用已实现波动率给出的上证指数波动率的有效估计,在研究已实现波动率特性的基础上,用计量模型探讨沪指波动的长记忆特征。发现HAR-RV模型比FARIMA模型更能有效地刻画沪指波动的长记忆性,且HAR-RV模型样本外预测效果远远优于FARIMA模型,这说明沪指波动具有伪长记忆性,表面特征显示的长记忆性是由短期投资、中期投资和长期投资形成的短记忆性叠加而成。同时由于HAR-RV模型综合考虑了不同时间水平上的已实现波动率,从而在深层次上验证了中国股票市场的异质性和波动率的杠杆效应。Based on realized volatility, this paper uses the optimal sampling method to give an effective estimate of the Shanghai Complex Index Volatility. With the analysis of its features, we investigate the volatility＇s long memory feature by econometric model. It is interesting to find that the HAR - RV model is more effective at depicting the long-term memory characteristic of volatility than FARIMA model is. What＇ s more, the former has better out - of- sample forecast performance than the latter does. This demonstrates that Shanghai Complex Index Volatility has pseudo - long memory, the apparently so- called long term memory is the result of cascades of short memory of short - term, medium - term and long- term investmer^ts. Moreover, the HAR- RV model comprehensively considers the different time horizontal realized volatility; it deeply validates the heterogeneity of the stock market of china and volatility＇s leverage effect.

关 键 词：已实现波动率 长记忆性 异质市场 HAR-RV模型 

分 类 号：F224.0[经济管理—国民经济]