Subscribe to our newsletter
The place for unfiltered climate news and Deep Sky developments.
Insurers Retreat as 2025 Wildfire Risk Reaches Dangerous Levels
Executive Summary
Insurance companies are abandoning homeowners in the highest risk wildfire areas. They cannot charge prices high enough to take on the risk they see coming. Exiting the market has become their last resort. In California, 1 in 5 homes in the most extreme fire risk areas has lost coverage since 2019. There are now over 150,000 uninsured households in these areas in California alone.
When the next extreme wildfires hit – perhaps in the next few months – thousands of uninsured houses may burn down. Regional property markets will collapse: property values will drop, FAIR plans1 will be way overdrawn, the government will be forced to buy abandoned properties to avoid rebuilding in dangerous areas, and taxpayers will foot the bill. The home real estate market in states like California is in a desperate situation.
And peak wildfire season hasn’t even started. Severe drought in the US Southwest is driving record fire risk across large swathes of Arizona, New Mexico, Texas, Southern California as well as Midwestern states: Nebraska, South Dakota, and others. Wildfires have already sprung up across Canada resulting in thousands of evacuations in Manitoba and two deaths. As of the end of May there were over 200 active fires burning across Canada, half of which were classified as 0% contained or “out of control”. This wildfire season is ramping up quickly and the data suggest the worst is yet to come.
Insurance markets are a leading indicator of how financial markets will deal with the climate crisis. Just as insurers won’t do business in key areas now, so too will investors avoid those kinds of risks. Insurance markets may be the first to show the effects of the climate crisis but the disruption won’t stop there.
Key Findings
- Home insurance premiums have shot up 42% in the most fire prone areas of California
- One in five homes in extreme fire risk areas of California has lost coverage since 2019
- Spring fire risk in the US Southwest and Northern Mexico has reached a ten-year record
Introduction
Some impacts of climate change are indirect, making attribution challenging due to the range of interrelated factors. Climate change’s impact on wildfires, however, is clear. Wildfires happen more often and are more dangerous in warmer, drier conditions. Climate change is raising temperatures and making droughts worse.
It’s not surprising then that wildfires are getting worse. That is true on just about every metric – many of which this report will dive into. Wildfires are happening more frequently, they are getting larger, and they are doing more damage to property and human lives than ever before.
So as temperatures rise across North America this spring and summer, what can we expect from wildfires?
A key indicator of the severity of the problem is how the home insurance is responding, and it provides a lesson for how financial markets will react to increasing climate risk.
Insurers are Worried about Wildfires
Insurance companies lose a lot of money if they underestimate the risk of wildfires. The devastating fires in Los Angeles in January made this painfully clear.
The total figure paid out by insurance companies to cover losses from the LA fires may be as high as $44.5B USD.2
Few private companies can withstand sudden losses of this magnitude, and it may take some of these companies many years to financially recover, if they ever do. This is why insurance companies throw huge resources at catastrophe or “CAT” modeling. This approach assesses the risk of such an event using the most up-to-date data and most advanced modeling techniques available. Failure to better model fire risk in the LA area cost these companies almost 50 billion dollars.
An insurance company’s survival depends on its ability to accurately predict risk. For this reason it can be instructive to analyse insurance companies’ actions when it comes to things like wildfire risk – they have to get it right.
Deep Sky Research received data from state insurance departments of California, Oregon, Texas, and Washington via Public Records Requests. The data allowed us to analyze how insurers’ actions differ in zip codes with different levels of wildfire risk. We compared insurer behaviour to wildfire hazard potential (WHP) data from the US Forest Service.3 WHP data is used by policymakers and insurers to assess fire risk throughout the United States. In particular, this analysis focused on the percent of a ZIP code’s area that is categorized as having “high” or “very high” WHP.
One lever an insurance company controls is how much it charges for its coverage. In zip codes where 90% of the area is classified as having high or very high WHP (extreme fire risk in Figure 1), the average cost of home insurance has skyrocketed 42% since 2009.
Figure 1: Home Insurance Premiums have Shot up 42% in Extreme Fire Risk Areas of California
One piece of evidence that these insurance companies’ CAT models are working is that this price hike happened before the LA fires. Figure 5 below shows that the LA fires occurred in areas with clear fire risk – raising premiums was how insurance companies sought to counterbalance that risk.
A second lever insurers control is their right to exit the market. Between 2015 and 2023, there was a 19% reduction in the number of home insurance policies in the highest risk wildfire areas. Since 2019 the number of policies in the highest risk areas has dropped dramatically. Some of this reduction is due to insurers issuing non-renewals - simply declining to renew an insurance policy. (See Figure 3 for non-renewal data specifically.) There were many well documented cases of this happening to homeowners in the LA fires eerily close to when the fires started. In all likelihood that is simply because the CAT models continued to predict greater and greater risk as fire weather conditions worsened.
Figure 2: 1 in 5 California Homes at Extreme Risk of Wildfires No Longer Has Insurance
Some of this reduction in coverage is also due to homeowners choosing to go without coverage. In extremely high risk areas however, that is not a true choice – it happens when homeowners cannot afford the rapidly increasing costs shown in Figure 1.
Figure 3: Since 2018 over 30,000 home insurance policies have been non-renewed in extreme fire risk parts of California.
In California, this is not because people are moving out of the most dangerous areas, as you might expect. In fact, the population in Extreme Fire Risk areas has actually grown since 2015, and grown faster than lower fire risk areas (see Methodology Figure 1). So beyond just the raw number of home insurance policies decreasing, the actual percentage of households in these areas with home insurance has dropped 24%.
This sharp drop-off in households with home insurance is concerning for a number of reasons. Most immediately it means the next severe wildfire in these areas will be financially devastating. In 2023 there were over 150,000 households in the highest risk zip codes with no home insurance. What will happen when thousands of uninsured homes burn down?
California is at the forefront of this trend but warning signs are beginning to flash in other states as well.
Figure 4: Non renewal rates are creeping up in high risk areas of Washington too.
Figure 5: California Fire Risk by ZIP Code
The home insurance market is in a state of crisis. The highest risk areas of California have effectively become uninsurable and will soon become unaffordable. Banks will not approve mortgages without home insurance, and few will buy a house without a mortgage (in a high risk wildfire area no less). Without significant policy intervention, these properties will eventually become worthless.
The California FAIR plan, and its equivalent in other states, provides exactly that policy intervention. As the insurer of last resort, it offers subsidized insurance for those that can’t find it on the private market. These subsidies are paid for by the very same insurance companies leaving the market. In order to contribute to the FAIR plan, they are forced to raise rates elsewhere. In other words, FAIR plans pool risk so that homeowners in low risk areas subsidize insurance for homeowners in high risk areas.
As the costs of living in these high risk areas continue to skyrocket, the system will fail. More homeowners will be forced to join the FAIR plan, which is already in financial trouble4, further raising costs for those not on the FAIR plan, who then may also be priced out of the private market. There is no easy exit from this vicious cycle. Meanwhile climate change continues to exacerbate wildfire risk.
Figure 6: Homeowners are being forced out of private insurance markets
California's FAIR plan grew from insuring 210,000 homes in 2020 to over 463,000 in 2024, with total exposure exceeding $450 billion. The LA fires alone may consume over 40% of the FAIR plan's annual resources, triggering emergency assessments on private insurers.
Figure 4 shows that California is further along this trend than some other states but Texas and Oregon are not far behind. The growth of FAIR plans may appear to be a policy success – as home insurance costs rise, publicly subsidized insurance is there to catch stranded homeowners. But FAIR plans have several fundamental limitations.
Unlike standard homeowners insurance, FAIR plans typically provide only basic dwelling coverage, excluding personal property, liability protection, and additional living expenses. In California, the maximum coverage is capped at $3 million, though most policies are much smaller. Homeowners often need to purchase separate "wrap-around" policies for comprehensive protection, significantly increasing their total insurance costs. For homeowners in high risk areas, the FAIR plan is not a get out of jail free card.
The explosive growth in FAIR plan enrollment creates systemic risk. FAIR plans operate on thin margins and lack the capital reserves of traditional insurers. When catastrophic losses exceed FAIR plan resources—as nearly occurred with the 2025 LA fires where the California FAIR plan faces $4.8 billion in losses—private insurers must cover the shortfall. This creates a vicious cycle: private insurers face losses both from their own policies and from FAIR plan assessments, incentivizing further market exits.
The growth of FAIR plans indicates that private markets are failing. FAIR plans avoid stranding homeowners in the short term, but they do not provide a long term solution.
Severe Fire Threat Looms for US Southwest
Fire conditions across North America are abnormally high for this time of year. A prolonged and deepening drought across the US Southwest and Mexico are causing record-breaking fire risk conditions.
In this analysis, Deep Sky Research calculated and analyzed a metric called Fire Weather Index (FWI) Anomaly. This metric compares 2025 FWI values to typical values from the same spring time period (March 1 - June 1). An FWI anomaly value of 0 indicates that fire risk is typical of this time of year. Positive FWI anomaly values indicate uncharacteristically high fire risk for spring, and negative, low risk for spring.
High FWI values have been predictive of extreme wildfires in the past5, and they are useful at this stage of the wildfire season in order to identify underlying risk conditions as we enter the hottest months of the year. Extreme wildfires like those in LA in January rely on specific weather conditions on the scale of weeks and days: very high temperatures, no rainfall, high winds, etc. But we don’t have to wait for those specific conditions in order to assess the level of underlying risk.
This report analyzes March 1 - June 1 data because spring conditions set the stage for peak fire season. Early season drying, snowpack depletion, and vegetation green-up create the fuel conditions that determine summer and fall fire severity. This "preconditioning period" is when climate change signals are often strongest—earlier snowmelt and hotter springs compound into extreme summer conditions.
Figure 7: Spring Fire Risk Has Reached Dangerous Levels in Parts of the US
Figure 7 shows how abnormally high fire risk is this year compared to previous years. In an average year the map would have as much blue as red. But in spring of 2025, risk is far above normal. Particularly in southern Arizona, New Mexico and Texas drought is making conditions ripe for dangerous wildfires. The Midwestern United States is seeing elevated risk as well particularly in South Dakota, Nebraska, Northeastern Colorado, and Kansas.
Figure 8 helps put the current conditions in the US Southwest and Northern Mexico into context compared to previous years. Fire risk has not been this high at this time of year for over 10 years.
Figure 8: Spring Fire Risk in US Southwest and Northern Mexico Has Reached a 10-year Record.
Drought conditions extend south into Mexico and are driving dangerous conditions there too. Northern Mexico typically sees high temperatures and dry conditions in the spring and summer, but Figure 9 shows that there too 2025 is bringing abnormally high fire risk.
Figure 9: Drought in Mexico Is Causing Record-breaking Fire Risk
Hydroclimate Whiplash
Dryer conditions generally increase wildfire risk, but climate change is demonstrating that the relationship between precipitation and wildfires is not quite so straightforward. One effect of climate change is extreme swings in levels of precipitation. One year brings above average rainfall, the next far below.
Figure 10: “Hydroclimate whiplash” in US Southwest and Northern Mexico
Daniel Swain has called this trend “hydroclimate whiplash”.6 Surprisingly, one effect of these swings in precipitation is worsening wildfires. High rainfall in one year drives vegetation growth. Then dry conditions turn this vegetation into easily accessible fuel for wildfires.
Figure 10 shows how annual precipitation has deviated from the 15-year climatological average for the region. This finding aligns with Swain et al. (2025) who found subseasonal hydroclimate whiplash has already increased 31-66% globally since 1950, with projected increases of 113% at 3°C warming.
These anomalous precipitation trends, combined with higher than normal temperatures, are driving record-breaking fire conditions. It’s not that there hasn’t been rain in recent years. In fact, 2021 and 2022 saw well above average rainfall across the region. Climate change is increasing incidents of atmospheric rivers which have been unloading huge quantities of precipitation on the west coast of the continent.7
The record FWI values in 2025 show the effects of rapid fuel accumulation followed by fuel desiccation - where wet periods promote vegetation growth that becomes critically flammable during subsequent hot and dry conditions. Swain et al. (2025) find that each 1°C of warming increases atmospheric water-holding capacity by ~7% (Clausius-Clapeyron), driving both more intense precipitation and higher evaporative demand - the dual drivers of fire weather extremes.
The trend over the last 15 years is that the extremes are getting more pronounced in both directions. As Figures 7 and 8 make clear, this is contributing to extremely dangerous wildfire conditions.
Outlook For Canada
Higher than normal fire risk is present in large parts of Alberta, Saskatchewan, and Manitoba as we head into the warmest months of the year. Early season wildfires like those in Manitoba in May offer warnings of what is ahead.8 In mid-May 1,000 people were evacuated from the small community of Lac Du Bonnet, and two people lost their lives. Then, at the end of May conditions deteriorated and thousands more people were forced to evacuate as the Premier declared a state of emergency and called for the deployment of the Canadian Armed Forces.
Figure 11: Elevated Fire Risk Is Already Sparking Deadly Fires Across Canada
Though the worst of the fires has so far struck Manitoba, the most abnormally high fire risk is actually in the provinces to the West as you can see in Figure 10. Figure 11 shows just how elevated risk is in that region of Northwestern Alberta and across the British Columbia border.
At the end of May there were already over 200 active fires burning, half of which were classified as 0% contained or “out of control”.9
Figure 12: Northwestern Alberta Is Facing Record Fire Threat
A vicious cycle
Even those in North America who have not experienced wildfire up close have probably breathed in the smoke. Canadian wildfires in 2023 spread smoke across vast areas of Canada and the US and the 2025 season appears to be following suit. There are unhealthy air conditions in several Canadian provinces, Minnesota, and smoke is expected to continue spreading as the fires continue raging.
Figure 13: May Wildfires in Canada Already Spreading Smoke Across the Continent
These huge quantities of smoke hint at another wildfire-related problem: carbon emissions. In 2023, emissions from Canadian wildfires dwarfed emissions from all other sectors of the economy.10 This is an example of what climate scientists call a “vicious cycle”: climate change causes worse fire conditions, which cause larger wildfires, which create massive carbon emissions, contributing to further climate change.
Conclusion
Home insurance companies are raising prices and retreating from high risk wildfire areas. These areas have been left with a weak safety when the next extreme wildfire hits. Unfortunately, the data show 2025 brings extraordinary levels of fire risk. Severe drought in the US Southwest and Northern Mexico have created dangerous fire conditions in that region. Fires are already wreaking havoc in Western Canada and spreading smoke across the continent. And the worst is yet to come.
Methodology
Insurance
Deep Sky Research submitted several Public Records data requests to state departments of insurance and received data from California, Oregon, Texas, and Washington. Some of the California data is also available for download on the California Department of Insurance website. These datasets informed the analysis in the insurance section of the report.
Figures 1 through 5 analyse insurer behaviour and fire risk at the zip code level. Oregon data did not come at the zip code level so Figure 6 displays aggregate state-wide data.
Figure 1 compares the cost of home insurance over time between areas with different levels of fire risk. Average homeowners’ insurance premium is calculated per policy, as opposed to per unit of exposure, and excludes FAIR plan policies in order to capture the behaviour of the private market alone.
Some analysis of home insurance costs does not compare premium prices per policy, instead using a comparison of premium price to exposure. This report uses the former approach for a few reasons: 1. It reflects the actual financial burden faced by homeowners in each market, which is central to understanding insurance affordability and availability; 2. Coverage amount reporting varies significantly across insurers and states, making rate-based comparisons less reliable; 3. The focus of this analysis is on homeowners ability to access any coverage, not the technical pricing of specific coverage amounts. Premium per policy is best suited to analyzing market exits and non-renewals as we do in this report.
In Figure 3, California non-renewal rate is calculated as the number of non-renewed policies as a percentage of total policies in force: new and renewed policies. This includes customer-initiated non-renewals as well as insurer-initiated non-renewals. In Figure 4, the Washington non-renewal rate was pre-calculated in the data received.
Home insurance data was combined with US Census household data at the ZIP code level in order to derive statistics like the 150,000 households without a home insurance policy in the most wildfire vulnerable ZIP codes of California.11
Wildfire Hazard Potential
Wildfire risk classification is consistent throughout the report. Zip codes above the state 90th percentile for percentage of area classified as “high” or “very high” Wildfire Hazard Potential (WHP) by the USDA Forest Service, Fire Modeling Institute are “Extreme Fire Risk”. Zip codes above the 70th percentile but below the 90th are “High Fire Risk” and zip codes below the 50th percentile are “Low Fire Risk”.
The report relied on this WHP data because a) it is geospatially granular data focused on the risk of destructive high-intensity wildfire, and b) it is used pervasively as an input to risk classification in insurance and policy analysis. Therefore it provides valuable insight into how insurers’ are understanding and responding to visible risk.
Fire Weather Index
Fire weather index (FWI) is a commonly used metric to understand fire risk in a given area – it combines data on temperature, moisture, fuel availability, and wind speeds to provide a comprehensive estimate of fire risk. This report calculates each location’s FWI anomaly by comparing 2025 FWI values across North America with average FWI values for the same 90-day period and same gridcell, across a baseline 30-year period from 1991 - 2020.
Methodology Figure 1
Methodology Figure 2
1 Fair Access to Insurance Requirements (FAIR) plans are state-managed insurance policies for homeowners who have been forced out of private insurance markets due to high risk factors.
3 Dillon, Gregory K. 2023. Wildfire Hazard Potential for the United States, version 2023 (270m). 4th Edition. Fort Collins, CO: Forest Service Research Data Archive. https://doi.org/10.2737/RDS-2015-0047-4.
5 https://www.deepskyclimate.com/blog/where-will-the-next-extreme-wildfire-be
6 Swain, D.L., Prein, A.F., Abatzoglou, J.T. et al. 2025. Hydroclimate volatility on a warming Earth. Nat Rev Earth Environ 6, 35–50. https://doi.org/10.1038/s43017-024-00624-z.
7 Zhou, Y., Wehner, M. & Collins, W. 2024. Back-to-back high category atmospheric river landfalls occur more often on the west coast of the United States. Commun Earth Environ 5, 187. https://doi.org/10.1038/s43247-024-01368-w.
8 https://www.theguardian.com/world/2025/may/14/canada-wildfire-manitoba-evacuations.
9 https://cwfis.cfs.nrcan.gc.ca/home
10 https://www.deepskyclimate.com/blog/where-will-the-next-extreme-wildfire-be
11 https://www.census.gov/programs-surveys/geography/guidance/geo-areas/zctas.html
