Jumps in the Chinese crude oil futures volatility forecasting: New evidence

Abstract

This study examines whether incorporating jumps and jump size can enhance the predictive ability of Chinese crude oil futures market realized volatility within the MIDAS model framework, incorporating Markov-switching. Our out-of-sample analysis indicates that the Markov-switching mixed data sampling model incorporating large and small jumps (MS-MIDAS-LMLS) performs better than other models in forecasting the volatility of Chinese crude oil futures. To ensure the robustness of our findings, we conducted various sensitivity test. The results consistently exhibit strong predictive performance of the MS-MIDAS-LMLS model. Additionally, we explored the impact of jump size on forecasting performance using an alternative jump test method (the ABD test) and different volatility regimes, with similar results indicating the positive impact of incorporating jump size on forecasting accuracy. Overall, our results underscore the potential benefit of integrating jumps and jump size within the MIDAS model framework to enhance the accuracy of volatility forecasting of Chinese crude oil futures market volatility.