MODELING EXTREME CLIMATE RISK IN AGRICULTURE: AN EVT-BASED APPROACH TO WHEAT YIELD LOSSES AND INSURANCE DESIGN IN UZBEKISTAN

Abstract

This study develops an econometric framework to assess extreme climate risks affecting agricultural production in Uzbekistan, with a particular focus on wheat yield losses in the Surkhandarya region. Using annual production data (2010–2024) and long-term hydroclimatic observations (1981–2025), the research applies Extreme Value Theory (EVT) and the Peak Over Threshold (POT) method to model tail risks associated with drought and temperature extremes. To isolate climate-induced shocks, wheat yields are detrended using linear regression techniques, and climate anomalies are constructed from long-term monthly averages. The empirical results indicate that rainfall deficits represent the dominant source of catastrophic yield losses, while temperature extremes exhibit relatively lower tail risk. The estimated shape parameters (ξ < 0) suggest bounded but persistent extreme risk. The 99.5% Possible Maximum Loss (PML) is estimated at approximately 29.6 units for rainfall-related losses and 4.6 units for temperature-related losses. The findings demonstrate that traditional indemnity-based insurance mechanisms systematically underestimate climate-induced tail risks and fail to provide adequate financial protection. Based on international best practices, the study proposes an EVT-calibrated index-based insurance framework integrated with reinsurance and early warning systems. This approach enhances climate resilience, improves insurer solvency, and supports sustainable agricultural development in Uzbekistan.