Spatial disaggregation of agricultural production data by maximum of entropy.

Summary The authors develop an econometric method to estimate crop choices at a disaggregated level (district level), using more aggregated data (regional level). This disaggregation method requires two steps. In the first step the authors estimate a dynamic land use model at the regional level. In the second step, they disaggregate the results of the regional model using a maximum entropy (ME) method. This ME disaggregation method is applied to a sample of California data. The sample includes six districts located in the Central Valley and eight possible crops, namely: alfalfa, cotton, forage, grain, melons, tomatoes, vegetables, and subtropical crop. The disaggregation process reconstructs land use at the district level with an out-of-sample prediction error of 16%. This result shows that the microeconomic behavior inferred from the aggregated data with this disaggregation method appears to be consistent with observed behavior.