Better simulation modelling to support farming systems innovation: Review and synthesis

Simulation modelling is a methodology that appears highly suitable for use in farming systems innovation. However, if the aim is to improve farming systems by supporting farmer behaviour change, model-based approaches seem to have delivered surprisingly little observable benefit to date. This paper identifies the problems underlying this apparent lack of impact, and proposes better approaches to improve on-farm benefits from farming systems modelling. A key principle that has been neglected in farm simulation modelling is intimate involvement of clients (defined as "the individuals or groups whose approval is needed for change to be implemented") throughout the innovation process. We argue that client participation is essential in the problem definition, model design and testing and policy design and evaluation phases of model-based research projects. The role of a simulation model within the innovation process, then, is to be a jointly-created "virtual world" wherein experiments may be conducted to facilitate learning about the relevant system. We argue that whole farm simulation models that use decision rules to specify alternative farm management strategies are the best available form of virtual world models of farming systems. Besides appropriate client input, high quality models require excellent software development practices and strenuous attention to building user confidence. The latter should include analysing the model to assess its stability and sensitivity properties, before using it to simulate experiments that compare several management alternatives under a range of environmental and local conditions. This approach allows estimation of the variations in farm system performance that are likely to result from interactions between initial farm state, farm management policy and future weather and markets. On the other hand, using simulation models to discover "optimal" farm systems would usually be inappropriate due to the complexity arising from multiple-stakeholder views, multiple-criteria, and the dynamic nature of farming systems problems. An improved systems modelling methodology is proposed that should be better able to provide benefits into farming practice.