Time-varying risk analysis for commodity futures

Our work presents risk analysis for twelve major global commodity futures during the financial crises and post -crisis period. We perform in-sample and out-of-sample risk analysis which includes equal predictive accuracy model and univariate GAS models for during and post-crisis periods. We also perform a backtesting procedure for providing better information about the predictive strength. We report that the models of all commodities show equal predict accuracy except for Gold whose models exhibit differing predictive accuracies. Among all models, ALD appears as best fitted for Natural Gas, Crude Oil-WTI, Gold, Silver, Aluminum, and Zinc under crises (Eurozone and global financial crises) and post-crisis period. However, SNORM performs best for Diesel and Natural Gas under crises and post-crisis period, respectively. Our paper entails implications for policymakers and investors.