Stock market volatility predictability: new evidence from energy consumption

Abstract

Abstract This research develops a group of novel indicators from the energy consumption perspective and assesses their ability to forecast stock market volatility using various techniques. Empirical evidence reveals that novel indicators, notably industrial non-renewable energy consumption, significantly enhance the forecasting of stock market volatility. The MIDAS-LASSO model, which integrates a mixed-data sampling method, effectively captures key information and outperforms other models in predictive accuracy. Further analysis reveals that the novel indicators contain useful forecasting information over the business cycle and crisis periods. Additionally, we indicate the forecasting ability of the novel indicators from the standpoint of investor sentiment variation. Our findings yield useful insights for the forecasting of stock market volatility, emphasizing the significant role of energy consumption.