论文部分内容阅读
【目的】对股票市场的VaR动态风险价值进行研究。【方法】采用小波多分辨技术将高频已实现波动率分解为近似信号和细节信号,建立MRA-RV-ARFIMA GARCH-VaR类模型,分别在1~2d、2~4d、4~8d和8~16d的尺度下进行动态风险价值度量。【结果】实证表明该模型能很好地捕捉到市场的信息,对风险预测效果较好。【结论】经过多分辨分解后的信号能有效地捕捉到不同时间尺度上的波动信息,近似信号能很好的反应波动的变化趋势,资产波动对短期交易反应敏感,不同时间尺度拟合的VaR比低频GARCH类模型效果更好。
【Objective】 The VaR dynamic VaR of stock market is studied. 【Method】 The MRA-RV-ARFIMA GARCH-VaR model was established by decomposing the high frequency realized volatility into approximate signal and detail signal using wavelet multiresolution technique. The model was established at 1 ~ 2d, 2 ~ 4d, 4 ~ 8d and 8 ~ 16d scale dynamic risk value measurement. 【Result】 Empirical results show that the model can capture the market information well and predict the risk better. 【Conclusion】 The signals after multiresolution decomposition can effectively capture the fluctuation information on different time scales. The approximate signal can reflect the changing trend of volatility very well. Asset volatility is sensitive to the short-term transaction response. The VaR fitted at different time scales Better than the low-frequency GARCH class model.