LexisNexis: Record 25.9% Home Severity Is Hiding Behind a Frequency Drop

LexisNexis clocked all-peril homeowners severity at a record 25.9% year-over-year increase in 2025, even as claim frequency fell 23.8%, netting a loss cost decline of just 4.4% that was still the third-highest in seven years and 50.0% above 2019 (LexisNexis Risk Solutions, July 2026). The frequency drop is doing the work in that math, not an improving hazard.

A Loss Cost Number That Is a Composition Effect, Not Relief

An indication anchored to the all-peril loss cost figure alone is making an implicit bet: that the 23.8% frequency decline holds through the next policy period. LexisNexis's own framing makes the fragility of that bet explicit. "Although loss cost across all perils decreased in 2025, it was still the third most expensive year for loss cost for the U.S. home insurance industry within the last seven years" (Insurance Journal, July 24, 2026). George Hosfield, LexisNexis Risk Solutions' vice president and general manager for home insurance, put the underlying pressure more bluntly: "U.S. home insurers continue to face increasing pressure and uncertainty as they contend with a 'perfect storm' of rising severity, rising inflation-driven replacement costs and reshaping loss patterns" (LexisNexis Risk Solutions, July 2026).

The frequency decline itself is plausibly structural rather than a one-year blip: higher percentage deductibles, tighter roof-age and cosmetic-damage underwriting, and the broader industry shift toward actual-cash-value-only settlement provisions on aging roofs and siding all raise the effective threshold for filing a claim, which is the same suppression mechanism actuary.info flagged in Verisk's Q1 2026 property claim data. If that is the driver, the claims that keep getting filed skew toward the larger, harder-to-avoid losses, which is consistent with severity accelerating precisely as frequency falls. On-leveled against a trend that treats the frequency side as durable and the severity side as noise, a homeowners indication built off the blended -4.4% number is pricing an assumption, not a fact the data supports.

Fire and Lightning Rewrite the Peril Mix

Fire and lightning loss cost rose 76.8% year over year, on a 67.3% severity increase and a 6.0% frequency increase, the sharpest peril-level move in the report (LexisNexis Risk Solutions, July 2026). The January 2025 Los Angeles wildfires, the Palisades and Eaton fires, drove the bulk of it: an estimated $61.2 billion in damages, the costliest wildfire event in U.S. history and the single most expensive climate disaster of 2025. That fire and lightning is both the fastest-growing loss cost and one of only two perils (with liability) where frequency rose rather than fell separates it from the frequency-suppression story running through the rest of the book: LA's wildfire severity was large enough to pull the frequency count up with it.

That is precisely the case against trending on the all-peril blend. A single homeowners severity trend applied uniformly across all perils nets a 76.8% loss cost swing against a -50.4% wind swing and lands somewhere in the middle that describes neither peril accurately. Wildfire exposure concentrated in California, the Mountain West, and pockets of the Southeast needs its own severity trend built off the 67.3% print and the widening tail risk that a $61.2 billion single-event loss represents for excess and surplus lines capacity, not a blended number diluted by a calm wind year elsewhere in the country.

Wind and Hail Look Calm Because 2025 Was Calm

Wind loss cost fell 50.4%, on a 12.0% severity decline and a steeper 43.9% frequency decline; hail loss cost fell 38.4% from its 2023 peak on a 35.4% frequency decline, with severity essentially flat (LexisNexis Risk Solutions, July 2026). Both declines read as favorable loss experience. They are more accurately read as a light convective-storm season. A mid-March central U.S. tornado outbreak still produced an estimated $11 billion in damages, the second-costliest single event of 2025 behind the LA fires, inside a year that still logged 23 separate billion-dollar disasters totaling $115 billion, the third-highest annual total on record (Insurance Journal, July 24, 2026). A year that produces an $11 billion tornado outbreak is not a year without severe convective storm risk; it is a year that got fewer of them than 2023 did. Banking the 43.9% wind frequency decline directly into next year's trend selection assumes the storm count itself is now structurally lower, when the more defensible read is that 2025 sat on the favorable side of a still-volatile distribution.

The reinsurance market is compounding the risk of misreading that distinction. Global property catastrophe rate-on-line fell 16% through the mid-year 2026 renewals, the steepest year-over-year decline since the late 1990s, as abundant capacity and a run of benign loss years pushed pricing down across every major region (Guy Carpenter, via Reinsurance News, July 2026). Primary carriers are getting ceded-cost relief at the same moment ground-up severity is setting a record. That combination flatters combined ratios this cycle. It does the opposite the moment either the wind and hail frequency reverts toward its longer-run average or a single large event forces reinsurers to reprice, and a book that priced its net retention off a -16% ROL environment plus a -4.4% loss cost print has stacked two favorable but non-independent assumptions on top of each other.

Liability and Water Carry Their Own Severity Signal

Two smaller lines inside the homeowners book deserve separate attention because their severity trends run in the opposite direction from the headline number even as their frequency falls. Liability loss cost declined 4.0% on a 14.6% frequency drop, but severity rose 12.8%, a spread wide enough to read as a social-inflation signal riding inside the home liability line rather than a genuine improvement in exposure (LexisNexis Risk Solutions, July 2026). Non-weather water followed a similar pattern: loss cost fell 6.4% and frequency fell 7.8%, but severity crept up 2.5% for the year and 63.16% since 2019, a persistent climb LexisNexis attributes largely to rising material and labor costs in water remediation. Both perils are small enough that they do not move the all-peril blend much on their own, but both are compounding in a direction the aggregate number obscures, and both belong in a peril-level reserve review rather than folded into the same trend factor as wind and hail.

What Moves at the Next Filing

Florida's Citizens Property Insurance Corporation offers a live example of how loss-cost relief actually reaches a filing. Its December 2025 recommended rate filing put the actuarially sound premium for 2026 at $3,617, down 43% from $6,347 in 2024, with the reduction attributed largely to SB 2-A litigation reform and to actual losses trending below prior projections after Florida avoided a direct major hurricane landfall in 2025 (Citizens Property Insurance Corporation, Recommended Rate Filing, December 2025). That relief is defensible on Florida-specific grounds, tort reform and a quiet hurricane year are real and documented drivers. It is also exactly the kind of filing a national LexisNexis print with a record severity underneath it should invite a second look at: is the loss-cost decline in a given book driven by the same durable, line-specific mechanism Citizens can point to, or is it riding the same frequency-driven composition effect running through the national data.

Three things belong on the desk before the next homeowners indication goes out the door. First, replace the blended all-peril severity trend with peril-level selections, at minimum splitting fire and lightning and non-weather water, both running well above the aggregate, from wind and hail, both artificially calm on a light 2025 cat year. Second, ask the reinsurance broker directly, ahead of the January 1, 2027 renewal, how much of the quoted rate-on-line relief assumes 2025's benign wind and hail activity repeats, and stress the net cost of capital against a return to 2023-level convective storm frequency. Third, watch whether LexisNexis's fire and lightning severity print holds or accelerates further when the 2027 edition of this report publishes next summer, since a second consecutive year of high-70s loss cost growth in that peril would confirm this year's spike as a trend rather than a single-event artifact tied to one California wildfire season.