Macroeconomic variables and time-varying volatility persistence of carbon market

Several studies find structural breaks in carbon market volatility, which could be well accounted for by introducing time-varying GARCH coefficients (volatility persistence). Most of the related literature suggests that the main driver of time-varying volatility persistence (TVP) is returns or the realized volatility (RV), and few studies investigate the time-varying persistence of volatility from the perspective of external macroeconomic factors. Building on the EGARCH model, we therefore develop the standard TVP-EGARCH-MIDAS model and its specification with macroeconomic variables (TVP-EGARCH-MIDAS-M models), which explicitly consider structural breaks by linking time-varying GARCH coefficients to explanatory variables (RV and macroeconomic variables) using MIDAS techniques. An empirical application to European Union Allowance (EUA) futures data shows that macroeconomic variables have a significant impact on carbon market volatility persistence. More importantly, the standard TVP-EGARCH-MIDAS model and TVP-EGARCH-MIDAS-M models outperform the EGARCH model in terms of in-sample fit and out-of-sample forecasts. Finally, the out-of-sample results show that the TVP-EGARCH-MIDAS-M models have superior predictive ability compared to the TVP-EGARCH-MIDAS model. In particular, the TVP-EGARCH-MIDAS model with European economic policy uncertainty index (EEPU) (TVP-EGARCH-MIDAS-EEPU model) exhibits the best forecasting performance, and the results are robust to different forecasting windows.