Spatial disaggregation of agricultural production data using maximum entropy

We develop a dynamic data-consistent method to estimate agricultural land use choices at a disaggregate (district) level, using more aggregate (regional-level) data. The disaggregation procedure consists of two steps. First, we estimate a dynamic model of land use at the regional level, then we disaggregate outcomes of the aggregate model using maximum entropy (ME). The ME disaggregation procedure is applied to a sample of California data including six districts and eight crops. The disaggregation procedure results in the recovery of district-level cropping areas with an average prediction error of 16.2 per cent. Copyright 2003, Oxford University Press.