不同发展时期的中国股市波动跳跃：基于高频数据的实证分析

摘要

本文以跳跃扩散过程为理论基础．利用2003年1月至2010年12月的上证综指5分钟高频数据，通过波动跳跃显著性检验方法与HAR—RV-j模型，探讨了在不同发展阶段上中国股市波动的跳跃特征与波动率模型的预测能力。实证研究的结果表明，越是波动激烈的时期，发生显著跳跃的频率越多、波动跳跃的幅度与强度越大；中国股市的波动跳跃包含着许多有利于改善波动预测的有效信息；股市的异常波动使得对波动预测的准确性大打折扣，市场变得更加难以预测。%This paper studies the characteristics of volatility jumps and fore-casting abilities in China Stock Market at different stages of development from Jan 2003 to Dec 2010. Based on the jump diffusion theory and using high-fre- quency data of Shanghai Composite Index, we adopt the significant jump test of volatility jumps and HAR-RV-J model. The empirical results show that the jump is more likely to occur with much bigger strength when the market is more vola-tile; some information existing in jump is effective in improving volatility fore- casting; unusual jump destroys the accuracy of volatility forecasting, which in turn makes it more difficult to forecast the market.