Volatility forecasting of carbon prices using factor models

This article develops a forecasting exercise of the volatility of EUA spot, EUA futures, and CER futures carbon prices (modeled after an AR(1)-GARCH(1,1)) using two dynamic factors as exogenous regressors that were extracted from a Factor Augmented VAR model (Bernanke et al. (2005)). The dataset includes 115 macroeconomic, financial and commodities indicators with daily frequency from April 4, 2008 through January 25, 2010 totalling 463 observations that capture the strong uncertainties emerging on the carbon market. The main result shows that the best forecasting performance for the volatility of carbon prices is achieved for the model including the dynamic factors as exogenous regressors, which can be useful to inform hedging or speculative trading strategies by energy utilities, financial market players and risk managers.