沪深300指数收益率及已实现波动联合建模研究  

A Joint Model for CSI300 Index Return and Realized Volatility

作　　者：瞿慧[1] 刘烨[1] 

机构地区：[1]南京大学工程管理学院,南京210093

出　　处：《管理科学》2012年第6期101-110,共10页Journal of Management Science

基　　金：国家自然科学基金(70932003;71201075;71203091);江苏省自然科学基金(BK2011561);高等学校博士学科点专项科研基金(20120091120003);中央高校基本科研业务费专项资金(1107011810;1118011804);教育部留学回国人员科研启动基金~~

摘　　要：金融资产的收益率和波动率是金融资产投资和风险管理等应用中的重要决定因素。针对收益率的新息过程与波动率的新息过程之间可能存在相关性的实际情况,将已实现波动区分为连续波动和跳跃波动,对收益率、连续波动和跳跃波动联合建模并刻画各时间序列模型新息之间的相关性,给出联合模型的最大似然估计法,使用2005年4月8日至2011年5月23日沪深300指数5分钟高频数据进行实证。研究结果表明,收益率、连续波动和跳跃波动的新息之间存在统计显著的相关性,对各时间序列单独建模估计的传统方法存在本质缺陷,沪深300指数已实现波动的杠杆效应及周日效应主要来自连续波动分量。联合模型通过对新息之间相关关系的合理刻画,提高了参数估计的有效性。The return and volatility of financial assets are critical for making investment and risk management decisions. Since there might be dependencies among innovations of return and volatility, we divide realized volatility into continuous-time volatility and jump volatility and then propose a joint model for return, continuous-time volatility and jump volatility, which also character- izes the dependencies among innovations. Finally the maximum likelihood estimation method for the joint model is provided. Em- pirical tests using CSI300 index＇s 5-minute high frequency data between 2005.4.8 and 2011.5.23 show that, there do exist sta- tistically significant correlations among innovations of its return, continuous-time volatility and jump volatility, which makes the traditional method of separately modeling and estimating each time series questionable. On the other hand, in addition, parame- ter estimation results of the joint model show that, the leverage effect and weekday effect of CSI300 index＇s realized volatility mostly come from the continuous-time volatility component. The joint model improves the efficiency of parameter estimation through appropriate modeling dependencies among innovations.

关 键 词：已实现波动 连续波动 跳跃波动 非独立新息 联合建模 高频数据 

分 类 号：F830.91[经济管理—金融学]