基于外部信息冲击的符号跳跃变差高频波动率模型

现阶段研究高频波动率的主流HAR-RV-跳跃模型仅考虑了与高频波动率有关的内生变量,忽视了外部信息冲击的影响,对高频波动率的估计和预测可能存在偏误.本文尝试将外部信息冲击引入到HAR-RV-跳跃模型中,构建基于外部信息冲击的符号跳跃变差高频波动率模型(HAR-VRV-跳跃模型).这类模型不仅兼顾内生因素和外部信息冲击对高频波动率的共同影响,还考虑了多元信息冲击的非对称效应.通过选取沪深300和中证500指数的高频交易数据作为研究样本,并利用滚动时间窗口预测和SPA检验对HAR-V-RV-跳跃模型的预测能力进行了评价,结果表明:HAR-V-RV-跳跃模型可以依据外部信息冲击的类型对高频波动率做出更准确的预测,其预测能力明显优于现有的HAR-RV-跳跃模型.但是,HAR-V-RV-跳跃模型对平稳期高频波动率的预测表现优于非平稳期.

A high-frequency volatility model with signed jumps variation based on external information impact

Most HAR-RV-JUMP models of high frequency volatility only include the endogenous variables related to high frequency volatility, but ignore the impact of external information, which may lead to errors and bias in estimation and prediction. This paper introduces the impact of the external information into the existing HAR-RV-JUMP models by constructing a high-frequency volatility model with signed jump variation. These new models take into account not only the common influence of endogenous factors and external information, but also the asymmetric effect of multi-information on high frequency volatility. Then, high-frequency trading data of CSI 300 Index and CSI 500 Index is collected to make estimates and prediction, and rolling time window prediction method and SPA test are used to evaluate the forecasting capacity of HAR-V-RV-JUMP model. The result shows HAR-V-RV-JUMP models boast higher prediction accuracy in high frequency volatility than HAR-RV-JUMP models. However, the result also shows the performance of HAR-V-RV-JUMP models is better in predicting high frequency volatility during the stable period.