Forecasting Corn and Soybean Basis Using Regime-Switching Models

Corn and soybean producers in the core production areas of the U.S. have experienced a notable jump in basis volatility in recent years. In turn, these increasingly erratic swings in basis have increased producers’ price risk exposure and added a volatile component to their marketing plans. This paper seeks to apply regime-switching econometrics models to basis forecasting to provide a model that adjusts to changing volatility structures with the intent of improving forecasts in periods of volatile basis. Using basis data from 1981 through 2009 from ten reporting locations in Ohio, we find that although models using time series econometrics can provide better short run basis forecasts, simple five year moving average models are difficult to improve upon for more distant forecasting. Moreover, although there is statistical evidence in favor of the regime-changing models, they provide no real forecasting improvement over simpler autoregressive models.