Improved Program Planning Generates Large Benefits in High Risk Crop Farming – A Profitable Application of Time Series Models and Stochastic Optimization

Agricultural production relies to a great extent on biological processes in natural environments. In addition to volatile prices, it is thus heavily exposed to risks caused by the variability of natural conditions such as rainfall, temperature and pests. With a view to the apparently lacking support of risky farm production program decisions through formal planning models, the objective of this paper is to examine whether, and eventually by how much, farmers’ “intuitive” program decisions can be improved through formal statistical analyses and stochastic optimization models. In this performance comparison, we use the results of the formal planning approach that are generated in a quasi ex-ante analysis as a normative benchmark for the empirically observed ones. To avoid benchmark solutions that would possibly exceed the respective farmer’s risk tolerance, we limit the formal search to a subset of solutions that are second-degree stochastically dominant compared to the farmer’s own decision. We furthermore compare the suitability of different statistical (time series) models to forecast the uncertainty of single gross margins.