Catastrophe Models and Capital Markets in the Insurance Industry

Catastrophe bond investors are earning less for bearing peril risk. According to a Woodwell Climate Research Center analysis, the average spread on new cat bonds fell from 3.71 times modeled expected loss in 2024 to 2.4 times in 2026 to date, and expected loss rose in the first half of the year even as spreads narrowed. The authors argue that a thinner cushion is only safe if the underlying catastrophe models are accurate, and that the models used by insurers and by capital markets rest on different climate assumptions.

Risk moves from homeowners through insurers, reinsurers and, ultimately, the capital markets via catastrophe bonds, where investors receive interest but forfeit principal if a predefined disaster occurs. Catastrophe models, which simulate extreme events across thousands of years of hypothetical activity, are central to pricing at each step. Many U.S. models, however, are built on backward-looking data and assume stationarity, meaning the climate of today is treated as the same as the past century. Warm sea surface temperature models, adopted after the 2004 and 2005 hurricane seasons, reflect present-day conditions and are used by reinsurers and cat bond investors. The authors cite Ian Gray's 2021 paper in Economy and Society, which documents that regulatory and political pushback after 2005, when some warm-SST models showed loss increases of up to 40%, led most modelers to revert to long-term views. Moody's RMS withdrew its model from Florida Commission on Hazard Loss Projection Methodology consideration in 2007.

The Florida regulatory framework has broader implications. Models used for Florida pricing must be approved by the FCHLPM, and states across the Gulf and Atlantic coasts look to that approval. According to the authors, only Karen Clark & Company's model includes explicit climate change accounting in Florida filings, and even that model produces fewer major hurricanes than the historical record. Verisk has stated it is unclear whether current warm sea surface temperatures reflect climate change. South Carolina has outlawed models representing present-day sea surface temperatures, and all but one Louisiana insurer uses the long-term model, according to a Woodwell review of public Louisiana rate filings.

The authors say reinsurance pricing has become a major driver of premiums. They attribute 45% of the 2024 increase in insurance premiums to reinsurance pricing, citing Keys and Mulder, and report that the premium gap between highest- and lowest-risk areas widened from 30% in 2014 to 53% in 2024. In 2022, foreign companies not subject to U.S. state insurance regulation accounted for almost two-thirds of the U.S. reinsurance market. The authors note that studies show hurricanes are delivering more damage through higher wind speeds, heavier rainfall and slower movement, and that inflation-adjusted disaster data shows an upward trend over 45 years.

For allocators, the authors highlight two implications: due diligence on the climatology underlying insurance-linked securities models, and a shift of capital toward climate risk-aware investments. They note investors already assume a warmer hurricane view than most U.S. primary insurers, so the gap alone does not indicate mispricing. The concern is that even present-day models may understate risk, leaving less cushion at current multiples. The authors suggest managers be asked what time period their estimates reflect and how holdings change under alternative assumptions.