Why home-insurance averages miss local pressure

Broad participation by insurers can coexist with worsening conditions for homeowners. A study released by the National Association of Insurance Commissioners on August 5th 2026 documents that tension in the American market. The release described broad insurer participation and improving underwriting results in 2024, alongside rising premiums, non-renewals and claims costs. Those findings concern data through 2024, not the price of a new policy in October 2026.

Across the association’s four geographic regions, average premiums per policy rose by between 18.3% and 43.3% after inflation from 2018 to 2024. Company-initiated non-renewal rates increased by between 96% and 216%, depending on the region. The latter figures measure relative increases in non-renewal rates. They do not mean that those percentages of homes became uninsured, or reveal the level of the rates on their own.

The findings describe different sides of the same business. A premium is revenue to the insurer and an expense to the household. Changes in prices, claims or the mix of risks accepted can improve a carrier’s financial performance without making coverage more affordable for a particular property. Aggregate profitability cannot settle the availability question.

What 715 companies can tell us

The report’s 2024 regulatory dataset records 715 homeowners companies with at least $50,000 in annual premium. It warns that many companies operate inside branded groups, so a company count differs from a count of independent groups. It also explains an important geographic omission: New York did not supply data for that part of the study and is absent from the Northeast results.

The geography of an insurer’s business matters as much as its presence in a company count. An insurer writing somewhere in a region is not necessarily offering a new policy for every building in it. Nor does a stable number of companies show that each has maintained its appetite for all locations. The report’s public overview notes that more than half the insurers in each zone reduced policy counts after 2018 even as overall participation held steady.

A regional total can therefore conceal substantial changes in where individual insurers are willing to write policies. The economic problem can be concentrated in a smaller set of properties, while a large book of other business continues to operate. Measuring that concentration requires information about where cover is offered, renewed or withdrawn and how the terms change.

An earlier Treasury study illustrates the value of looking more closely. Published in January 2025 and covering 2018–22, it found that average premiums in the highest-risk fifth of ZIP codes were 82% higher than in the lowest-risk fifth, using its measure of expected losses from climate-related building hazards. Its analysis excluded flooding from that hazard measure, since ordinary homeowners policies generally do not cover it. These are historical associations within a defined dataset, not quotations for comparable houses today.

Finding the gaps in coverage

State regulators’ 2026 homeowners data call requested more detailed records for 2018–25 across a wider range of policy forms. A May 21st memorandum extended the submission deadline to July 15th, citing the expanded scope and the need for validation. That submission date has passed. The published requirements alone do not establish whether every insurer supplied complete data or how much of the resulting analysis is public.

Better information can sharpen the policy choice. If difficulty is mainly rising rebuilding costs, the response differs from one aimed at a shortage of willing underwriters in a particular location. If policyholders are moving between forms of coverage, a count of cancellations alone may miss what replaces the old protection. Answering those questions requires the geographic and policy-level detail that broad market totals leave out.

The reports answer a different question from the one facing a household seeking a renewal. They describe prices and participation across large historical datasets; the household needs to know which insurers will cover its property, on what terms and at what cost. Connecting those views would make the policy debate more useful. Broad participation is relevant to market health, but the availability of a workable policy remains a local question.