California wildfires and the future of wildfire risk modeling

By the end of January 2025, wildfires in Los Angeles had destroyed or damaged over 18,000 structures, claimed at least 29 lives and forced approximately 200,000 residents to evacuate. This event is destined to become the costliest wildfire in U.S. history, exceeding the ~$12 billion (2024 USD) loss for California’s 2018 Camp Fire. The two most damaging fire complexes, the Palisades and Eaton fires, burned around 24,000 and 14,000 acres respectively.

Current insured loss estimates stand at up to $45 billion – representing roughly one-third of the global insured losses for 2024 – and occurred within just the first two weeks of the year. This means that many U.S. primary insurers with direct exposure to these fires will have already used up a significant proportion of their catastrophe loss budgets for the year. Some insurers will likely have also eaten into part of their reinsurance coverage that they purchased only a few weeks ago.

At the market level, the impact on reinsurers is expected to be manageable, in part due to recent shifts in risk appetite. The hard market over the last few years has led many reinsurers to withdraw from covering lower-layer catastrophe risks, which have been increasingly affected by frequent losses from “secondary perils” such as wildfires, floods, and convective storms. However, claims related to the Los Angeles fires are still developing, and at least one U.S. primary insurer has already reported a total loss on their reinsurance coverage, suggesting that some reinsurers may see notable losses.

As the U.S. drives global reinsurance markets, an unexpected loss of this size so early in the year could lead to further hardening of reinsurance rates at the June 1 renewals, which are primarily focused on the Florida/Southeastern U.S. markets ahead of the Atlantic hurricane season. Increases in global reinsurance costs will ultimately be passed onto policy holders.

For many years, California insurers have relied on historical loss data for ratemaking, as the use of catastrophe models (see below) was not permitted by regulators. But in late 2024, new regulations^[1] allowed the use of these models, while in return requiring insurers to expand coverage in high-risk areas and reward mitigation efforts. These regulatory updates are aimed at improving risk assessment and market stability. All eyes will therefore be on California after the Los Angeles wildfires to see if these changes succeed.

Why wildfire risk is rising

The combined effects of climate change and increased development in fire-prone areas drive greater exposure to potential losses. Rising temperatures, prolonged droughts, and changing vegetation patterns are intensifying fire activity, making wildfire seasons longer and more severe. In some regions, like California and Australia, fire seasons now extend nearly year-round. This has contributed to a steady increase in California’s annual total burned areas, as highlighted in WTW’s 2024 H2 Natural Catastrophe Review. Similarly, the number of large wildfires in the U.S. has nearly doubled since the 1980s. 

The wildland-urban interface (WUI) – areas where homes and businesses meet wildland vegetation – has also expanded. This is particularly true in California, where 32% of housing units were in the WUI by 2020. As more people and assets are exposed to these high-risk areas the potential for catastrophic losses also increases.

These changes are not just a challenge for homeowners and insurers. Businesses face operational risks, such as damaged assets, supply chain disruptions, business interruption and increased liability. At the same time, wildfire smoke is a growing public health issue, with long-term costs for healthcare systems. Degraded air quality from wildfires has long-term health impacts – ranging from respiratory issues to psychological distress – with research attributing approximately 340,000 premature deaths globally to wildfire-induced air pollution each year.

Wildfire modeling solutions for risk management

Wildfires are one of the hardest natural hazards to model because humans influence both the hazard (how and where fires ignite and spread) as well as other factors that drive risk, such as exposure (the location and density of communities in fire-prone areas) and vulnerability (the materials and design of structures, or the effectiveness of mitigation measures). There are a variety of modeling approaches available for assisting in understanding and managing wildlife risk.

Catastrophe models are the primary tools used by insurers and corporations. These physically-based models simulate fire spread using principles like heat transfer and wind dynamics, calibrated with historical data. By modeling thousands of potential scenarios, catastrophe models estimate wildfire likelihood and financial impact, enabling risk assessment, premium setting, and mitigation planning.

The U.S. Forest Service, alongside other agencies, has also developed models to help understand and manage wildfire risk which simulate how fires spread based on factors such as weather, terrain, and vegetation. In addition to supporting risk assessment and mitigation strategies, these models are also often used for operational decision-making, aiding in real-time firefighting and ecological planning.  

Wildfire risk score models, such as those incorporated within WTW’s Global Peril Diagnostic tool, use datasets like historical fire footprints, population density, and WUI proximity to generate simple-to-interpret risk scores. WTW also has bespoke in-house vulnerability curves that can convert risk scores to loss estimates, which are often used for deterministic, site-specific risk assessments.

More recently, several AI-based wildfire models have come to market, leveraging machine learning to predict fire risk. Like traditional risk score models, they often produce a single risk score for a location and are trained large datasets—sometimes incorporating over 100 data layers, including real-time satellite imagery and weather observations.

Predicting and preparing for wildfires is greatly helped by having a good understanding of these modeling tools, but their effectiveness ultimately depends on their integration into broader risk management strategies. As wildfires become more frequent and intense, the demand for robust models and effective strategies will only increase, which will help insurers, policymakers, and communities better manage and reduce risk.

Reference

1. California's sustainable Insurance strategy Return to article