An econometric forecasting model for Tennessee feeder cattle prices

Feeder cattle prices have been highly variable. This variability creates difficulty on the part of producers and users in correctly anticipating future prices. This study was intended to develop a price forecasting mechanism which would assist in predicting the price of 500 to 600 pound feeder steers six months in the future in Tennessee.

Two separate forecasting models were developed: (1) an econometric model based on causal variables, and (2) a model based on futures market prices. Data for 1972 through 1981 were used to estimate the parameters of the two models. Data for 1982 were used to test the ex ante forecasting accuracy of the models.

The econometric model used a single equation which specified feeder cattle price to be dependent upon the following components of feeder cattle demand and supply: feedlot demand, stocker demand, nonfed slaughter demand, calf production, and quantity of available feeder cattle. Measures of the variables were developed from secondary data. The futures market model used current prices for the feeder cattle futures contract deliverable six months in the future and the first differences of these prices as independent variables to predict prices six months in the future.

The econometric model indicated that feedlot demand eight months in the future, calf production, and current nonfed slaughter demand were statistically significant indicators of Tennessee feeder cattle prices six months in the future. The futures market model indicated that the current futures price for the contract deliverable six months in the future was a significant indicator of prices six months in the future.

Using Theil's U₂ and root-mean-squared-error measures of forecast quality, it was determined that the econometric model outperformed the futures market model for the within-sample period (1972-1981). When the two models were used to forecast prices for 1982 (out-of-sample period) the futures market model outperformed the econometric model. These results made it difficult to determine which model was the best forecaster. Both models outperformed current prices in forecasting futures prices as indicated by Theil's coefficients less than 1. Thus, both models should be of value in forecasting Tennessee feeder cattle prices.