Crop Modeling in the Insurance Sector: Beyond the Limits of Forecasting

Stakeholders in the agricultural sector often consider observational data and modeling in the context of ‘what can it tell me about current conditions?’ or ‘what can it forecast about the end of the growing season?’ In locations where crop insurance plays a significant role, however, answers to these questions may be insufficient to quantify the major risks faced by stakeholders.

Multiple peril crop insurance (MPCI) programs dominate the market in some of the largest agricultural regions of the world, e.g., the United States, China, and India -- insuring crops against the disruption to normal outcomes (yield, revenue, etc.) for a large number of weather perils. The operation of an MPCI program is dependent upon a detailed, large-scale reckoning of the probability distribution of weather events impacting crops and livestock and an understanding of the expected (mean) outcome on a granular level. In addition, from the perspective of individual farmers, the risk analysis they need is one that accounts for the expectations for the current growing season and for the expectations of claims to their crop insurance policy.

Into this void, the field of probabilistic crop insurance modeling has emerged – combining the probabilistic generation of weather scenarios with crop and financial modeling. The challenges in this field are numerous. Observational data is continually improving in quality, but this means that some of the best observations lack the long time series required to describe a range of weather scenarios. Crop models also improve over time, but for the purposes of assessing crop insurance claims, the sophistication and quality of an underlying crop model must be uniform across major crops and over the geographic domain of the program. In addition, depending on the terms of a crop insurance program, multiple assessments of crops may be required: prevented planting, final yield, quality of product, and post-harvest losses. Finally, access to risk analyses through crop insurance modeling is not uniform and the asymmetry of information can have significant consequences for farmers and insurers. In this presentation, we will discuss these challenges in the context of bringing new crop insurance modeling solutions to domestic and global markets.