Multivariate analyses of acreage abandonment patterns in the U.S. Great Plains winter wheat region indicate that the major mode of variation is an in-phase oscillation confined to the western half of the overall area, which is also the area with lowest average yields. This is one of the more agroclimatically marginal environments in the United States, with wide interannual fluctuations in both climate and profitability.
We developed a multiple regression model to determine the relative roles of weather and expected price in the decision not to harvest. The overall model explained 77% of the spatial and temporal variation in abandonment. The 36.5% of the non-spatial variation was explained by two simple transformations of climatic data from three monthly aggregates—September–October, November–February and March–April. Price factors, expressed as indexed future delivery quotations,were barely significant, with only between 3 and 5% of the non-spatial variation explained, depending upon the model.
The model was based upon weather, climate and price data from 1932 through 1975. It was tested by sequentially withholding three-year blocks of data, and using the respecified regression coefficients, along with observed weather and price, to estimate abandonment in the withheld years. Error analyses indicate no loss of model fidelity in the test mode. Also, prediction errors in the 1970–75 period, characterized by widely fluctuating prices, were not different from those in the rest of the model.
The overall results suggest that the perceived quality of the crop, as influenced by weather, is a much more important determinant of the abandonment decision than are expected returns based upon price considerations.