Volatility Forecasting of Crude Oil Market: Which Structural Change Based GARCH Models have Better Performance?
研究了平滑和突变两种结构变化对原油价格波动率预测的影响，发现考虑平滑变化的FFF-GARCH模型在预测和投资组合表现上优于传统GARCH和马尔可夫机制转换GARCH模型。
GARCH-type models have been widely used for forecasting crude oil price volatility, but often ignore the structural changes of time series, which may lead to spurious volatility persistence. Therefore, this paper focuses on the smooth and sharp structural changes in crude oil price volatility, i.e., smooth shift and regime switching, respectively, and investigates which structural change based GARCH models have better performance for forecasting crude oil price volatility. The empirical results indicate that, first, the flexible Fourier form (FFF) GARCH-type models considering smooth shift can accurately model structural changes and yield superior fitting and forecasting performance to traditional GARCH-type models. Second, the Markov regime switching (MRS) GARCH model incorporating regime switching exhibits superior fitting performance compared to the single-regime GARCH-type models, but it does not necessarily beat the counterparts for forecasting. Finally, the FFF-GARCH-type models outperform MRS-GARCH for forecasting crude oil price volatility and portfolio performance.