The NAIC's June 2026 homeowners data call extended ZIP code-level collection through 2025, adding the three years the Federal Insurance Office's January 2025 report could not reach. All 50 jurisdictions are participating, covering every carrier writing at least $50,000 in homeowners premium in any year from 2018 to 2025.

The FIO dataset stopped at 2022. What follows 2022 is the non-renewal surge itself, and it now sits with 50 state regulators ahead of a public report planned for early 2027.

50
Jurisdictions participating in the NAIC's 2026 homeowners data call, covering every carrier writing at least $50,000 in homeowners premium in any year from 2018 to 2025
8
Policy years captured at ZIP code granularity, spanning the pre-crisis baseline (2018-2020) through the peak non-renewal cycle (2021-2025)
Early 2027
Planned public release of the NAIC's homeowners market report, with a public comment period before finalization, creating the regulatory benchmark for state DOI rate review proceedings

Key Takeaways

  • The FIO's January 2025 report covered 330-plus insurers and more than 246 million policies for 2018 through 2022, finding average premiums growing 8.7% faster than general inflation.
  • The new call adds 2023 through 2025 at ZIP resolution: the years State Farm stopped writing new California homeowners business and Citizens reached 1.4 million policies.
  • Eight data categories are collected at ZIP granularity, including peril-specific claims and losses and mitigation discounts applied at the policy level.
  • California surplus lines homeowners policies grew from roughly 50,000 in 2023 to more than 320,000 by end of 2025, a 540% increase driven by admitted market withdrawal.
  • Florida's non-renewal rate of 3.35% of policies in force was the highest of any state in the FIO data, with Louisiana posting the sharpest five-year jump.

What Changes Between the FIO Data and This One

The structural difference is not scale. It is which regime the record covers.

The FIO analysis, published January 2025 under a 2024 agreement with the NAIC, covered 330-plus insurers and more than 246 million policies aggregated to ZIP code level for 2018 through 2022, an annual average of 49.3 million. It found premiums growing 8.7% faster than general inflation, non-renewal rates materially higher in ZIP codes with elevated modeled climate losses, Florida's 3.35% non-renewal rate the national high, and Louisiana's five-year increase the steepest.

The 2026 call collects eight categories at ZIP granularity: policy type across home, renter, condo, and mobile home forms; premiums written and earned; peril-specific claims counts and losses; deductible structures; cancellations and non-renewals with cause coding; coverage limits by tier; replacement cost versus actual cash value elections; and mitigation discounts applied at the policy level.

Submissions were due June 15, 2026, extended once to July 15. The three added years are where the market actually moved.

State Farm stopped writing new California homeowners in May 2023, Allstate earlier that year. California surplus lines homeowners policies went from roughly 50,000 in 2023 to more than 320,000 by the end of 2025, a 540% increase. Citizens reached 1.4 million policyholders to become Florida's largest homeowners insurer, before a depopulation program moved 184,000 policies out in May 2026 alone.

So the dataset now spans two regimes: the competitive period through 2020, and the crisis and early-recovery period after it. A regulator can put a carrier's 2024 non-renewal rate in a coastal Mississippi ZIP code against that same carrier's 2019 rate in the same ZIP code, against the all-industry 2024 rate there, and against the all-industry 2019 baseline.

The Benchmark Changes Who Brings the Comparison

That four-cell comparison is the whole actuarial consequence, and it inverts the evidentiary posture of a rate proceeding.

Until now a carrier arrived at a state hearing with its own data: its loss experience, its non-renewal history, its peril-level loss costs. Regulators held market share filings and actuarial publications, but not an all-industry ZIP-level record that would let them place one carrier's pattern against the market aggregate in a specific ZIP code in a specific year. After the 2027 report they hold it, and it is the regulator who brings it.

The peril split is the part that reaches the indication rather than the narrative. Eight years of loss broken out by wind, hail, fire, and flood at ZIP resolution gives regulators something no individual carrier has: the all-industry aggregate by peril in each ZIP code. A carrier knows its own wind loss experience in a coastal ZIP. It has never had to explain why that figure sits above the market's in the same ZIP across the same eight years.

That turns a routine credibility question into a filing requirement. A carrier whose indicated peril-level loss cost in a ZIP code deviates materially from the published all-industry figure needs documented actuarial support for the divergence: rate inadequacy confirmed by subsequent development, aging housing stock inside FEMA-designated flood zones, or a class mix skewed to the highest-hazard coverage tier. That analysis is standard. What is new is that it has to live in the rate filing support file rather than in an internal memorandum a regulator would have to request.

Non-renewal cause coding carries the same shift. A carrier whose coding in high-hazard ZIP codes defaults to broad underwriting judgment language, rather than specific peril-related rationale, is building a record against a public map that will show exactly where its exits concentrated.

Whether the Benchmark Normalizes for Hazard Decides What It Measures

One methodological choice determines whether the report produces an actuarial reference point or a political one.

A 3.35% non-renewal rate in coastal Broward County is not the same actuarial event as a 3.35% rate in suburban Kansas City. If the report presents non-renewal rates without adjustment for underlying hazard and rate adequacy, the map will generate pressure on carriers in high-risk markets whether or not their withdrawal decisions were consistent with the hazard they faced.

The FIO report did not resolve this, and the industry critique of it had technical merit. Correlating non-renewal rates against climate-risk scores does not separate hazard exposure from the social inflation, litigation, and replacement cost pressures that concentrated in the same coastal and wildfire-exposed ZIP codes over the same years. High South Florida non-renewal in 2022 reflects inadequate rate levels under the prior ratemaking environment as much as it reflects wind exposure, and a cross-sectional correlation cannot tell those apart.

The 2026 dataset is thick enough to do better. Eight years of peril-specific loss at ZIP resolution supports multivariate work that separates wind-driven experience from hail, flood, and fire. Whether that capability gets used for peril attribution or for straightforward non-renewal mapping is the first thing worth reading in the report.

The mitigation discount question runs the same risk in reverse. Requiring carriers to report discounts applied at the policy level gives regulators, for the first time in aggregate, a way to test whether the credits are associated with lower observed loss frequency.

Florida's wind mitigation credit program, active since 2002 and generating roughly $1.5 billion in annual premium credits, is the obvious test case. If high-penetration ZIP codes show meaningfully lower wind loss frequency per policy than comparable ones, the actuarial support for the credits strengthens. If they do not, the credits become the thing in question at the next Florida rate hearing, on a dataset no carrier assembled and none can rebut with its own book alone.

Further Reading

Sources