forecastingprinciples.com Reviews of Important Papers on Forecasting
Before 1985 Reviews
Review of:

Richard E. Just and Gordon C. Rausser (1981), "Commodity price forecasting with large scale econometric models and the futures market," American Journal of Agricultural Economics, 63, 197-208.


A readily available method of forecasting commodity spot prices is to use currently available prices on corresponding futures markets. Such future price quotes can serve as a direct predictor or alternatively various filtering methods or transformations can be applied to observable futures prices. The authors use various filters of future prices as a base comparison from which to evaluate five forecasting series: Chase, Doanes, Data Resources, Wharton, and the U.S. Department of Agriculture. The comparison is made monthly for a time lead of one to four quarters. In some instances, the futures markets perform admirably in this detailed comparison based in part on non-parametric tests. In other instances, an opposite result obtains. The effect of forecast horizon on absolute forecast accuracy depends on a large extent on the characteristics of individual markets. In general, future markets were more accurate in terms of bias whereas the large-scale econometric models were more accurate in terms of variance. The results may of course be dependent on the choice of the time period of comparison, 1977-78. Finally, although the authors do not attempt to determine which method of forecasting is best ex ante, their results strongly suggest that composite forecasts of future markets and econometric models using time varying parameters schemes should generate the 'optimal' ex ante forecast.