Comparative Performance of Selected Mathematical Programming Models

This study compares the predictive performance of several mathematical programming models. Using the cropping patterns, yields and crop gross margins of eighteen farms over a period of five years we compare the models' optimum solutions with observed crop distributions after the Reform of the EU Common Agricultural Policy of 1992. The results show that the best prediction corresponds to a model that includes expected profit and a qualitative measure of crop riskiness. The results suggest that, in order to obtain reliable predictions, the modelling of farmers' responses to policy changes must consider the risk associated with any given cropping pattern. Finally, we test the ability of the proposed model to reproduce the farmers' observed behaviour with equally good performance under conditions of limited data availability.