Generalized Modeling of Oil Futures Volatility Through Uncertainty Indicator Selection: A GARCH–MIDAS–AES Framework

Abstract

Building on prior literature that has demonstrated the effectiveness of various uncertainty-related indicators in enhancing the accuracy of crude oil volatility forecasting, this paper first investigates the type and persistence of the impact of changes in these indicators on volatility and then compares these indicators across different scenarios to determine the optimal strategy for their implementation. We employ a more generalized approach by utilizing the GARCH–MIDAS–AES model, which accommodates features that vary with different indicators. The empirical results, based on data from 1997 to 2022, underscore the importance of considering threshold and leverage effects. We also identify two types of impact: directional and nondirectional. Furthermore, among the uncertainty indicators examined, our findings affirm the predictive prowess of the Financial Uncertainty indicator in the majority of cases. However, during periods of global crisis, the Index of Global Real Economic Activity emerges as a more practical choice.