采用修正的已实现门阈多次幂变差实证分析了中国股市高频波动率跳跃的特征,并运用自回归条件持续期模型、自回归条件风险模型以及扩展的自回归条件风险模型刻画了跳跃持续期的特征.实证研究表明,中国股市高频波动率发生显著跳跃的比例较高,并且跳跃具有聚集的特征,跳跃的幅度、强度以及跳跃幅度的分布都具有时变性,而跳跃对高频波动率的贡献却具有相对稳定性;在样本期,中国股市高频波动率跳跃表现出较强的正自相关性,且跳跃的持续期存在较强的长记忆性和周日历效应.
Using corrected realized threshold multipower variation (C_TMPV), this paper investigates the jump dynamics of volatility of the high-frequency data from SSEC, and analyzes the jump duration based on the ACD model and the ACH model. The results show that the size, the intensity and the distribution of the jumps in Chinese stock market are time-varying, but the jump's contribution to high-frequency volatility is relatively stable. In addition, the jumps of high-frequency volatility exhibit strong positive auto-correlations, and the jump durations have the long term memory and weekday calendar effects.