Forecasting Catastrophe: Integrating GAN-ALSTM and Monte Carlo Simulations for Insurance Underwriting in Extreme Weather

To help insurers assess underwriting risks in areas prone to extreme weather, we introduce the Extreme Weather Underwriting Prediction Framework (EWUP). This framework employs Monte Carlo simulations for catastrophic loss bounds and integrates Generative Adversarial Networks-Attentive Long Short-Term Memory (GAN-ALSTM) for precise loss forecasting, coupled with a Comprehensive Profit Model (CPM) for profitability analysis. Our research suggests that the U.S. may face financial deficits in 2–3 years without policy adjustments, whereas China could remain profitable until 2040, thanks to its large population and lower disaster-related losses.