Why Home Insurance Keeps Getting More Expensive — the Climate Repricing Explained

Property insurance works on a simple bargain: a large number of people each pay a small amount so that the unlucky few can be made whole. The bargain holds as long as losses are frequent enough to be estimated and rare enough to be pooled. For roughly a century, insurers estimated those losses by looking backwards — decades of hail, wind, fire and flood claims, averaged into a rate.

That method quietly stopped working. The problem is not that any single disaster was unforeseeable; it is that the historical record has become a weaker predictor of the near future. The US National Oceanic and Atmospheric Administration's long-running tally of billion-dollar weather and climate disasters counted 28 separate billion-dollar events in 2023, a record at the time the series was maintained (the archive is still published by NOAA's National Centers for Environmental Information). Severe convective storms — hail and straight-line wind, not hurricanes — did much of that damage, and they are precisely the peril that older models underestimate.

The financial consequence showed up on the balance sheet. According to rating agency AM Best, the US homeowners insurance line paid out roughly $15.2 billion more in claims and expenses than it collected in premiums in 2023 — the worst underwriting result of this century, and roughly double the prior year's loss. When a line of business loses money at that scale, three things follow in order: reinsurance gets more expensive, primary insurers reprice, and the least profitable policies get shed.

From loss history to catastrophe models

The technical change is that carriers now price from catastrophe models rather than from claims averages. A cat model simulates tens of thousands of synthetic wildfire, hurricane, hail and flood seasons over a specific property — its roof material, elevation, slope, nearby vegetation and distance to a fire station — and returns an expected annual loss for that address.

The practical effect is granularity. Under the old system, a whole town shared a rate. Under the new one, two houses a few streets apart can be priced very differently because one sits in a wildland–urban interface zone and the other does not. That is defensible actuarial practice, and it is also the reason premium increases feel arbitrary to the people receiving them: the input that changed was a model, not their behaviour.

Regulators have started to formalise this. In December 2024 California adopted a first-of-its-kind regulation allowing forward-looking catastrophe models in rate filings — replacing rules that had stood for some 37 years — as part of its Sustainable Insurance Strategy. The trade was explicit: carriers get to use the models, and in exchange they commit to writing more policies in distressed, high-hazard areas.

Non-renewals are the leading indicator, not the price

The most useful signal in this market is not the premium; it is the non-renewal rate. A non-renewal is an insurer declining to continue a policy at the end of its term. It is not a cancellation, and it usually has nothing to do with the policyholder.

A study by the US Treasury's Federal Insurance Office, which examined roughly 246 million policies, found non-renewal notices climbed nearly 30% between 2018 and 2022 to more than 620,000 a year — and that non-renewal rates ran about 80% higher in the highest-risk ZIP codes than in the lowest-risk ones. A parallel Senate Budget Committee investigation, drawing county-level data from insurers representing roughly two-thirds of the national market, found the same pattern and made an important correction to the popular story: California is not the epicentre. Florida, Louisiana and North Carolina all posted higher statewide non-renewal rates. Southern New England, the Carolinas, New Mexico, Oklahoma, parts of the northern Rockies and Hawaii all show elevated rates, which tells you the driver is the full range of perils rather than any one headline hazard.

Where non-renewals cluster, prices follow. Research published by the National Bureau of Economic Research using mortgage escrow data (NBER working paper 32579) documents the sharp premium growth of the early 2020s and, more importantly, shows it concentrating in the most disaster-exposed counties rather than spreading evenly.

When the backstop becomes the market

Most US states run a FAIR Plan — Fair Access to Insurance Requirements — as an insurer of last resort. These are funded by the private carriers licensed in the state, not by taxpayers, and were designed as a temporary holding pen for a small number of hard-to-place properties.

California shows what happens when the holding pen becomes the destination. Its plan carried roughly 126,000 policies in 2018; by early 2025 it had grown past 555,000 residential policies, with aggregate exposure in the hundreds of billions of dollars. After the January 2025 Palisades and Eaton fires destroyed more than 18,000 structures — insured losses were estimated in the $25–40 billion range depending on the estimator — the plan booked several billion dollars in claims and levied a $1 billion special assessment on its member insurers, the first such call since the 1994 Northridge earthquake.

Two structural details matter more than the headline numbers. First, an assessment on member insurers is ultimately recovered from policyholders statewide, so a concentrated fire loss becomes a diffuse premium increase hundreds of miles away. Second, FAIR Plan coverage is deliberately narrower than a standard policy, and California's residential dwelling limit is $3 million — so homes above that value are structurally underinsured without a separate difference-in-conditions wrap.

What it actually does to households

Insurance is not a standalone bill. It is a gate. Almost every mortgage requires it, so an unaffordable premium becomes a housing problem rather than an insurance problem.

• Going bare. A Consumer Federation of America analysis put the number of uninsured US homeowners at about 6.1 million — roughly one in 13 — with lower-income households roughly twice as likely to be uninsured as the general population.
• Foreclosure pressure. Research from First Street finds that each 1% rise in premiums is associated with roughly a 1% rise in foreclosure rates, and projects that the share of foreclosures attributable to climate-related factors could climb from around 7% today toward 30% by 2035.
• Price discovery in housing. Once premiums are modelled at address level, they become a public estimate of physical risk. That estimate eventually shows up in what buyers will pay — which is how an insurance repricing turns into a property repricing.

What you can actually control

Location dominates, and you cannot move a house. But three things measurably shift the outcome. Harden the structure against your region's dominant peril — Class-A fire-resistant roofing, ember-resistant vents and defensible space in wildfire country; impact-rated roofing, reinforced garage doors and roof-to-wall connectors in wind and hail country. These change the model input, not just the discount code. Shop annually, and use an independent broker; in a market where carriers are entering and exiting continuously, loyalty is no longer priced. And check your dwelling limit against current rebuild costs, because construction inflation has left a large number of technically insured homes unable to fund a full rebuild.

The uncomfortable conclusion is that this is not a cycle waiting to turn. Insurers are pricing a hazard curve that is still moving, and the affordability question is migrating from the insurance industry to housing policy — which is where it will be settled.