Factors of predictive power for metal commodities

Abstract

There are numerous forecasting studies on commodity prices using various micro- and macroeconomic indicator sets. However, commodity markets have undergone a substantial transformation in the last 20 years, with periods of the financialization, and possibly also a de-financialization, which should also be reflected in the commodity price forecasts. To identify the changes in price predictors and determinants, we individually forecast 24 metal prices one-month ahead in the pre- and post financial crisis period, where we identify the autoregressive price components having a large impact across all commodities and periods. However, interest rates are of larger impact in the first sub-sample, whereas commodity- and financial market indices are dominating in the second sub-sample. Further, we perform an out-of-sample forecast over the entire timespan, where we are able to significantly outperform the predefined benchmark forecast models, a random-walk and a random-walk with drift, in 12 of the 24 cases.