Environmental attention and the predictability of crude oil volatility: Evidence from a new MIDAS multifractal model

Recent literature reveals that the volatility of Chinese crude oil futures exhibits significant multifractality and is influenced by investor environmental attention. In response, we incorporate the mixed frequency data sampling (MIDAS) framework into the long-term volatility component of the Markov switching multifractal (MSM) model, resulting in the MSM-MIDAS model. Additionally, we construct an environmental attention index for Chinese investors using the Baidu Index and principal components analysis methodology, which we integrate into the MSM-MIDAS model, creating the MSM-MIDAS-EA model to enhance crude oil price volatility predictions. Our empirical results indicate that the MSM-MIDAS models, which account for the multifractality of the Chinese crude oil market, outperform GARCH-MIDAS models in predicting the short-, medium-, and long-term volatility of Chinese crude oil futures. Furthermore, the MSM-MIDAS-EA model, which incorporates multifractality in short-term volatility and the environmental attention index for long-term volatility, achieves higher prediction accuracy across various forecast horizons.