Using Futures Prices to Forecast US Corn Prices: Model Performance with Increased Price Volatility

A futures price forecasting model is presented which uses monthly futures prices, cash prices received, basis values (cash prices less futures) and marketing weights to forecast the season-average farm price for US corn. Performance of the model forecasts is examined using standard measures, such as mean absolute error, mean absolute percentage error and mean squared error. Tests for statistical differences between the futures model forecast and price projections from the US Department of Agriculture (USDA) are conducted using the Modified Diebold Mariano test statistic. A measurement of price volatility identified the past 3 crop years, 2006/2007-2008/2009 with increased volatility compared to the prior 6 years, 2000/2001-2005/2006. Forecast errors from the futures forecast model increased during these volatile price years compared to the prior 6 year period which exhibited more stability. Suggestions are made to improve model price forecasts during periods of price volatility.