Basis Forecasting Performance of Composite Models: An Application to Corn and Soybean Markets

Numerous studies have examined the performance of different models on basis forecasting while none of them has compared relative performance of composite models. In order to further improve basis forecasting accuracy, the crux of hedge management strategies, we investigate basis forecasting performance of selected composite models, as well as various individual models. Empirical results based on weekly futures and cash prices for major North Carolina corn and soybean markets indicate that composite models have more stable and better performance in forecasting basis compared to individual models’ forecasts. The informationtheoretic forecast combination method is found to be superior among the composite models considered.