This paper extends the time-varying higher-order moment model to high-frequency data using the multiplicative component GARCH (MC-GARCH) framework of Engle and Sokalska (2012). We propose two multiplicative component time-varying higher-order moment models for intraday returns: the MC-GJRSK and MC-ARCD models. Empirical analysis based on the Shanghai composite and Shenzhen component indices reveals that intraday returns exhibit significant and persistent higher-order moment dynamics. Compared to the benchmark MC-GARCH model, the two proposed models, which incorporate higher-order moment information not only achieve superior in-sample fit, but also produce more accurate out-of-sample forecasts of Value-at-Risk (VaR) and Expected Shortfall (ES). Further analysis demonstrates that the superior forecasting performance of the proposed models remains robust across both high and low volatility periods. Moreover, the proposed models offer more accurate forecasts of tail conditional densities, thereby enhancing their effectiveness in intraday risk forecasting.
扫码关注我们
求助内容:
应助结果提醒方式:
