Corn and soybean basis behavior and forecasting: fundamental and alternative approaches

A model is developed to analyze the nearby basis behavior of corn and soybeans in several markets across the U.S. Results suggest that basis behavior has a seasonal pattern, and that different variables affect nearby corn and soybean basis at various periods. The most important factors affecting basis are storage cost (opportunity cost), transportation cost (barge rates), and supply (regional production relative to regional storage capacity). Nine conventional (naive three-year-average forecasts, econometric, ARIMA and composite forecast models) and less conventional forecasting techniques (State Space and Neural Networks models) are utilized to forecast out-of-sample basis for one-month up to 12-month ahead. The performance of all methods in forecasting 1991-1995 basis is analyzed. The forecasting performance comparison shows that adding current market information to the three-year-average model and the seasonal ARIMA model can slightly improve basis forecasts compared to the benchmark simple three-year-average forecast model.