金融资产收益率波动预测——基于我国股票市场跳跃行为研究  

Forecasting Returns Volatility for Financial Assets Based on High-Frequency Data

作　　者：柳会珍[1] 张成虎[1] 李育林[2] 

机构地区：[1]西安交通大学经济与金融学院,陕西西安710061 [2]西北工业大学管理学院,陕西西安710072

出　　处：《当代经济科学》2011年第5期109-118,128,共10页Modern Economic Science

基　　金：中国博士后科学基金(20100471621);教育部人文社会科学研究青年基金项目(09XJC790011)

摘　　要：金融资产收益率波动是资产定价和金融风险管理的核心部分,而跳跃是收益率波动中的重要组成部分。基于修正Z-检验,本文检测识别我国股市波动中跳跃行为,并且研究了跳跃的时序特征,统计结果表明,在市场大波动时期,和连续成份相比,跳跃对于波动率具有极其重要的贡献。建立包含跳跃的已实现波动率非齐次自回归模型,在波动模型中纳入滞后绝对日收益率和杠杆效应预测股指收益率波动。实证分析结果显示,对于短期的波动预测,包含跳跃和两种影响因素的波动模型表现最好,然而对于提前1月的长期预测,跳跃和连续波动成份分离模型预测明显优于其它模型,这些事实说明跳跃对股指波动率预测具有重要的影响,好坏消息对波动率非对称性具有短期显著影响,而对长期水平的波动率预测影响不显著。Financial asset return volatility is the core for asset pricing and financial risk management while jump is the important component of returns volatility. Based on the corrected Z-statistical test of jump, this paper identifies the jumps behavior in domestic stock market volatility and studies the time series features of jump. Statistical results show that compared with continuous components, jumps contribute significantly to volatility during market turmoil. Heterogeneous autoregressive models including jumps are built, and by incor- porating lagged absolute daily returns and leverage effects, we fit and forecast stock index returns volatility. Empirical analyses indicate that the model incorporating the jumps and the two factors performs best in forecas- ting at short horizon, however, at longer horizon such as one month ahead, the model separating jumps from the continuous components of volatility outperforms the other volatility models. These facts show that jumps have important impact on volatility forecasts of stock index, while good or bad news have short-term significant influence on volatility asymmetry but have no significant effects on long-term horizon forecasts.

关 键 词：已实现波动率 跳跃 波动率预测 金融风险 

分 类 号：F224[经济管理—国民经济] F832.51