摘要
以沪深300指数的高频数据为例,运用滚动时间窗的样本外预测方法以及比SPA检验更具优势的模型信度设定检验(MCS),实证分析了跳跃、符号跳跃变差及符号正负向跳跃变差对HAR-RV、HAR-RV-J、HAR-RV-CJ和HAR-RV-TCJ等4种基础波动率模型预测能力的影响。研究发现:符号跳跃变差不仅可以提高各波动率模型的拟合精度,而且还可以提高模型的预测精度;符号正向和负向跳跃变差相比符号跳跃变差对未来波动率具有更好的解释能力,且它们对未来波动率的影响是不对称的;加入符号正、负向跳跃变差的HARRV-TCJ模型的预测效果是众多模型中表现最好的,尤其是它的对数形式。
Taking 5-minute high frequency data of the CSI 300 index as an example, and applying the out- of-sample rolling time window forecasting combined with Model Confidence Set which is proved superior to SPA test, we explore the impact of the jump, signed and signed negative （and positive） jump variation on future volatility. The empirical results show that, signed jump variation not only increases the goodness of fit of the model, but also improves the forecasting accuracy of the model. By comparison, the signed jump variation, signed negative and positive jump variation have better explanations to the future volatility, which is asymmetry. The HAR-RV-SJV-D model is the best one among models discussed in this paper, especially its logarithmic form.
出处
《系统管理学报》
CSSCI
北大核心
2015年第5期700-710,共11页
Journal of Systems & Management
基金
国家自然科学基金资助项目(71372109)
教育部博士点基金资助项目(20120184110020)
四川省科技青年基金资助项目(15QNJJ0032)
