Comparing Environmental Impact of Alternative CAP Scenarios Estimated Through an Artificial Neural Network

The paper aims to assess environmental impact produced by alternative Common Agricultural Policy (CAP) scenarios in the Italian Marche region for the period 2000-2002. Scenarios concern alternative hypotheses about direct payments for arable crops related to Agenda 2000. For this aim, a Multilayer Feedforward Neural Network model (MFNN) was applied. Different from traditional models, MFNN is able to analyze complex patterns quickly and with a high degree of accuracy. Moreover, MFNN makes assumptions about neither the underlying population nor the existence of optimising behaviour and uses the data to develop an internal representation of the complexity characterising the system analysed. The results indicate that direct payments produced positive environmental effects compared to the hypothesis of absence of direct payments. Moreover, they show that it would have been even better, from an environmental point of view, if Agenda 2000 had been more radical in comparison to the 1992 Mac Sharry reform, by introducing decoupled direct payments.