SOUTHERN JOURNAL OF AGRICULTURAL ECONOMICS DECEMBER 1992 A COMPARISON OF SUBJECTIVE AND HISTORICAL CROP YIELD PROBABILITY DISTRIBUTIONS

Forecast distributions based on historical yields problems associated with such aggregate data and subjective expectations for 1987 expected crop for risk applications. Nevertheless, the avail-yields were compared for 90 Western Kentucky ability of county yield series continues to en-grain farms. Different subjective probability elicita- couragetheiruse in manyresearch anddecision tion techniques were also compared. In many indi- support efforts. vidual cases, results indicate large differences 2. Yield Series from Agronomic Experiments. between subjective and empirical moments. Over- Some experiment stations have maintained all, farmer expectations for 1987 corn yields were regular historical records of crop yields. Day below those predicted from their past yields, while provides the first extensive analysis of yield soybean expectations were above the historical fore- distributions for corn, cotton, and oats, using cast. Geographical location plays a larger role than Mississippi Experiment Station data. Distribu-crop in comparisons of relative variability of yield. tions constructed from such yield series have Neither elicitation technique nor manager charac- been used to calibrate plant growth models and teristics have significant effects on the comparisons thus to represent farm level yield variability. of the forecasts. These distributions, however, tend to reflect un