Forecasting U.S. Pork Production Using a Random Coefficient Model

Abstract

AbstractA random coefficient regression model is found to be superior to a fixed coefficient model for short‐ and intermediate‐term forecasting of quarterly U.S. pork production. The random coefficient model portrays some regression parameters as the sum of a systematically changing component and random error. Use of such models is discussed. Pork supply is hypothesized as a function of seasonal shifters with geometric lags on hog and feed prices. Results show seasonal effects declining, feed price not being a significant explanatory variable, and pork production adjusting faster to lagged price conditions than indicated by the constant coefficient model.