基于GARCH-MIDAS模型对股市波动率预测

Forecasting the Stock Market Volatility Based on GARCH-MIDAS Model

极端冲击(例如战争和金融危机)会导致股市波动剧烈.提出5种增强的混频模型(GARCH-MIDAS),可以捕捉非对称和极端冲击对股市波动率的影响.样本内的结果表明,我国股市存在明显的波动率聚集效应和杠杆效应,并且负面的极端震荡会导致较高的波动性.而样本外的MCS和DM检验结果则清楚地显示出EGARCH-MIDAS-ES模型最适合预测股市波动率,该模型在短期波动中纳入了非对称效应,长期趋势中加入了极端冲击的影响.此外,稳健性检验证实,与标准GARCH-MIDAS模型相比,增强的波动率模型在统计和经济方面均可产生更好的预测结果.通过考虑极端冲击,为股市波动率预测提供了新的见解.

Extreme shocks(such as wars and financial crises)can cause volatility in the stock market. Five enhanced mixing models(GARCH-MIDAS)are proposed to capture the impact of asymmetric and extreme shocks on stock market volatility. The results in the samples show that there are obvious volatility aggregation effects and leverage effects in China stock market,and negative extreme shocks will result in higher volatility. The out-of-sample MCS and DM test results clearly show that the EGARCH-MIDAS-ES model is best for predicting stock market volatility rate,the model incorporates asymmetric effects in short-term volatility and extreme shocks in long-term trends. In addition,robustness tests confirm that the enhanced volatility model can produce better prediction results statistically and economically than the standard GARCH-MIDAS model. By considering extreme shocks,our research provides new insights into stock market volatility predictions.