Analysing the drivers of rising wildfire risk in California

Large wildfires have become increasingly common in California over the past two decades, with one in every six acres of the state burning since 2000 and thirteen of the twenty largest wildfires on record occurring since 2015. However, much of this burned area has been in wildland regions with relatively low population and asset density. Less than 4% of the state’s wildland–urban interface (WUI) has burned over the past decade, meaning the area that carries the greatest loss potential has not, for the most part, been exposed to large-scale fire under today’s conditions.

The January 2025 Los Angeles fires illustrate how today’s underlying conditions can translate that latent exposure into large losses. They occurred in mid-winter, when conditions historically limited spread, and burned through some of the most valuable WUI in the world. The result was more than $40 billion in insured losses – the largest wildfire bill in history. A decade ago, losses of this scale would have been considered an extreme tail outcome.

What makes this event significant for (re)insurers is not simply that such a loss occurred, but what it implies about the unburned WUI more broadly: the potential for future loss is much larger than recent fire footprints suggest. If events of this scale now sit within the range of credible outcomes, wildfire needs to be considered a core contributor to portfolio volatility, rather than an infrequent or marginal loss source.

To understand today’s wildfire loss potential, we therefore need to look at three interacting forces: climate conditions, where exposure is located, and how expensive it is to rebuild when losses occur.

Average summer temperatures in California has warmed by +0.5°C per decade over the 1979-2025 period, increasing evaporative demand and drying vegetation. The state has also experienced prolonged drought, including thirteen of the driest 30 months on record between 2012-2016, with dry conditions returning in 2021 and 2022 after wetter years. At the same time, the rainy season has become shorter and more variable, providing less reliable fuel recovery. Together, these shifts mean vegetation stays flammable for longer and Santa Ana wind events are more likely to drive rapid spread.

These conditions have expanded the scale and timing of wildfire activity. Between 2006 and 2025, total burned area was 146% higher and the number of fires 54% higher than during 1986–2005 (Figure 1). Additionally, fires are no longer confined to the late-summer and autumn window: winter months have seen more than twice as many wildfires in the recent period compared with the earlier one.

The January 2025 Palisades and Eaton fires are a clear expression of this expanded fire window. These fires occurred in mid-winter, during a period that historically would have limited ignition and spread. Attribution analysis indicates that similar fire-weather conditions are now more than twice as likely in the present-day climate, and that the burned area during these events was approximately 2.3 times greater due to the influence of warming [1].

This expanded timing and geographic window matters because it increases the likelihood that large fires coincide with populated and high-value areas.

Just as the geographic and seasonal window for large fires has expanded, population and development have moved further into the same areas. Figure 2 shows that much of California’s population growth and new built-up area since the mid-1980s has occurred along the metropolitan edges of Southern California, the Bay Area, the Central Valley and the Inland Empire. This growth has expanded the footprint of the WUI: today, one in three Californians lives in the WUI, and an estimated 244,000 additional homes were built in these zones between 2010 and 2020.

This means that a larger share of the housing stock now sits in areas where fire is both more likely and more difficult to manage. At the same time, the concentration of economic value in these areas has increased. The median price of a single-family home in California over the past five years has been around $800,000, more than twice the level of the early 1990s. Market value is not the same as insured replacement cost, but it indicates how much economic value now sits in the WUI. Each individual structure now represents a larger potential loss when damaged.

Crucially, we can also now detect and quantify how this expansion affects fire outcomes. Analysis of the January 2025 Los Angeles fires indicates that fuel conditions in WUI areas explained the largest share of the burned-area extent, with weather contributing and ignitions playing a smaller role [1]. This means the scale of the event reflected where combustible fuels and development were located, not simply the presence of severe fire-weather conditions.

Reconstruction costs have also increased, meaning that the financial impact of wildfire is driven not only by the number of structures affected, but also by how expensive each one is to replace. Construction material costs rose sharply during the post-COVID supply chain disruption and have remained elevated, rather than reverting to prior baselines. On average, construction materials today are around 1.6 times higher than in the five years before 2020 (Figure 3). Over longer horizons, where differences largely reflect general inflation, lumber prices are roughly triple their mid-1980s level, and concrete costs are more than 70% above their early-2000s baseline.

Labour and capacity constraints add further pressure, as rebuilding often occurs simultaneously across multiple affected areas. Stricter building codes also increase costs: in California, requirements under Chapter 7A can mean partial repairs must be upgraded to fire-resilient standards, pushing costs toward full replacement even where structures are not fully destroyed.

In high-value housing markets, these factors mean that even relatively small changes in hazard or exposure can produce disproportionately large insured losses. The cost of returning damaged communities to their pre-fire baseline is materially higher than it was in previous decades.

Future wildfire losses in California are likely to exceed what recent fire footprints suggest. Fires can now occur across more of the year, more people and assets are located in flammable areas, and rebuilding is materially more expensive. With much of the high-value WUI yet to experience fire under these conditions, historical loss experience alone is no longer a reliable guide for pricing or portfolio steering.

To respond:

1. Evaluate how well current models represent today’s risk climate. Historical averages and older event sets can understate both the likelihood and scale of large-loss scenarios. Model evaluation should include sensitivity testing and checks against recent events to understand where adjustments or judgement are required.
2. Supplement exposure data where it matters most. Asset-level characteristics, such as defensible space, roof materials, and fuel continuity, meaningfully influence loss outcomes. Integrating higher-resolution exposure data, and validating it against claims, enables more accurate pricing and risk selection.
3. Calibrate loss assumptions to current cost realities.Replacement costs have risen faster than modelling frameworks assume, and building codes can mean partial repairs trigger near full replacement. Reviewing where model assumptions diverge from claims development is increasingly important when setting technical prices and portfolio strategy.

In combination, these steps help ensure that wildfire risk is understood and priced in a way that reflects the current drivers of loss, not the conditions of past decades.

[1] Kelley, D. I., et al. State of Wildfires 2024-2025. (2025).

Author: Daniel Bannister, Willis Research Network