Deliverable 8.3: Forecasting Commodity Prices Under Specification Uncertainty: A Comprehensive Approach

Paper authors
Jesus Crespo Cuaresma (IIASA)
Jaroslava Hlouskova (IIASA)
Michael Obersteiner (IIASA)
Apr 5 2017
Description We present a comprehensive modelling framework aimed at obtaining short-term forecasts (one to twelve months ahead) of commodity prices and apply it to short and medium run predictions of Arabica coffee, wheat, soybeans and corn. We entertain a large number of univariate and multivariate time series models, including specifications that exploit information about market fundamentals, macroeconomic and financial developments and climatic variables. A comprehensive set of forecast averaging tools is implemented to explicitly address model uncertainty. Our results indicate that variables measuring market fundamentals and macroeconomic developments (and to a lesser extent, financial developments) contain systematic predictive information for out-of-sample forecasting of commodity prices.
Authors at SUSFANS
Work package
Commodity prices, forecasting, vector autoregressive models, model uncertainty, forecast averaging