Weather index-based crop insurance: Exploring the benefits of Bayesian and Deep Learning models in crop yield prediction

The agricultural sector is highly vulnerable to a wide range of weather- and climate-induced risks. Challenges with modeling the complex weather and climate dynamics bring to the forefront statistical issues linked with analyzing massive multi-resolution, multi-source data with a non-stationary space-time structure, a nonlinear relationship of weather events and crop yields, and the respective actuarial implications due to imprecise estimation of risk. Conventional parametric statistical and actuarial approaches are constrained in being able to address these problems. However, state-of-the-art machine learning (ML) and artificial intelligence (AI) methods provide fast, automatic learning of hidden nonlinear dependencies and nonstationary structures within large complex datasets and are proving to outperform other methods across a wide variety of applications, from credit card fraud detection to the next best product offer and customer segmentation. Yet, their potential in actuarial sciences and, particularly, agricultural insurance, remains largely untapped. In this project, we investigate the utility of a novel methodology for evaluation of basis risk in agricultural index-based insurance, using a flexible framework of Deep Belief Nets (DBNs) and Copula Bayesian Networks (CBNs) ML methods. This study aims to provide a better understanding of nonlinear relationship of crop yields and weather events at disparate space-time scales, identify optimal indicators that reliably track future risks of climate and weather to crop production and better identification, quantification and propagation of uncertainty to improve crop production basis risk estimation for more reliable index-based insurance rate-making.