More powerful computers, the better availability of micro-data, and the development of new modeling techniques, such as multi-agent systems, allows to analyze agricultural policies from the bottom up. We present such an approach that considers the spatial interaction of thousands of individually behaving heterogeneous farms and apply it to analyze agri-environmental policies for a selected intensive production region in the southwest of Germany.