Economic Losses from Forecasting Error in Agriculture

Abstract

This paper develops a new method for estimating economic losses from forecasting error. The method is an alternative to the traditional distributed lag approach to estimating expectations. An application to Kansas agriculture over the period 1874-1933 reveals significant variation in farmers' price forecasting performance across counties and over time. These differences can be explained in terms of market improvements, technological and geographical factors, and county characteristics suggestive of human capital differences. The magnitudes, timing, and cross-sectional pattern of the losses from forecasting error confirm the importance of forecasting skill and allocative ability in American agriculture during the period examined.