Abstract:

This paper proposes a new optimization framework to minimize geographical basis risk in weather index insurance by integrating extreme weather penalties and spatial regularization into strike temperature design. The model balances three objectives: (i) reducing payoff mismatches between local and reference stations, (ii) penalizing strike temperature that fail to account for heatwaves and cold spells, and (iii) ensuring smooth strike variations across geographically correlated stations. To solve the resulting high-dimensional, non-convex problem, we employ a hybrid Genetic-Nelder–Mead algorithm, where a genetic algorithm provides global exploration and Nelder–Mead delivers local refinement. Using daily temperature data from 28 weather stations in Ontario (2000–2023), the model produces station-specific strike temperatures that adapt to both spatial heterogeneity and annual climatic extremes, such as the 2018 heatwave and 2019 cold spell. Results demonstrate significant reductions in basis risk and more equitable insurance payouts. The contribution lies in the optimization model’s design, which enhances the fairness, resilience, and practical reliability of weather index insurance under climate variability.

Speaker: Dr. Samuel Asante Gyamerah | Department of Mathematics  | Toronto Metropolitan University

Chairperson:  | Department of Mathematics | University of Ghana 

Meeting ID: 375 747 304 216 9 | Passcode: 46AN7Be2 | Teams: Link 

All are cordially invited.

Bio