The Insurance-Affordability Squeeze: Florida and California as Leading Indicators

For most of the post-WWII period, homeowners' insurance was a small, stable line on the closing statement — roughly 0.3–0.5% of home value per year, paid monthly into escrow, and largely invisible to the buyer's affordability calculation. Mortgage underwriters baked it into the PITI ratio but no one lost a deal over it.

Three things changed between 2020 and 2025:

• Climate reinsurance costs doubled. Global property-cat reinsurance rates rose 62% cumulatively from 2021 to 2024 per Guy Carpenter's index, with peak-zone US exposure up 100%+. Primary carriers pass this through almost 1:1 in the following rate cycle.
• Replacement-cost inflation ran 25–40% ahead of general CPI from 2021–2023, driven by lumber, labor, and roofing component costs — and policies are replacement-cost-based, not market-value-based.
• Carrier retreat in Florida and California forced buyers into state-backed residual markets (Citizens in FL, FAIR Plan in CA) at materially higher premiums and stricter coverage limits.

The combined effect is that homeowners' insurance now competes with the mortgage rate as the biggest swing variable in the monthly cost of homeownership. We track it because it's the one cost input that can raise PITI without any change in price, rate, or income — a pure affordability shock.

The weighted-average homeowners' premium for a $400K replacement-cost policy rose from $1,380 in 2022 to $1,710 in 2025, a 23.9% increase. Median household income rose ~11% over the same period, so premium-to- income grew from 1.8% to 2.1% nationally. National averages mask most of the story.

Note the shape: FL and CA diverge from the national average in 2021 and the gap keeps widening. Texas, which shares hail and wind risk with the Gulf but not hurricane concentration, tracks roughly national. That's the state-policy signal we focus on — same ocean, different premium trajectory.

Dollars of premium don't tell you anything about affordability by themselves. The right denominator is local median household income: a $2,800 premium in Manhattan is a rounding error; the same premium in Lakeland is 5% of post-tax take-home. We compute premium-to-income as the annual HO-3 premium (assuming a $400K replacement cost normalized to local construction cost) divided by metro-median household income.

The correlation is not subtle. The six metros with the largest premium-to-income jumps — all coastal Florida or wildfire-exposed California — show negative 12-month HPI in the FHFA data. The national trendline is +2.4%; these metros are running 3–6pp below trend.

Premium is one channel; availability is the other. Carriers that stay in the market raise prices; carriers that can't price the risk simply leave. Non-renewal rate is the cleanest proxy for local insurability, and it has diverged dramatically.

Non-renewal maps roughly to the same wind-and-wildfire ZIPs that drive premium. But the mechanism for home prices is different. Non-renewed owners either (a) pay 30–60% more through the state residual market, (b) go bare on windstorm coverage, or (c) sell. For the 8% of US homes without a mortgage, (b) is legal but uninsurable. For the 66% with a mortgage, (b) is not an option — servicers force-place a policy at 2–3× market rates. In practice, (a) and (c) are what we observe in the data.

We run two complementary regressions on a 2022–2025 panel of 410 MSAs.

Spec A (transactions)
ΔTxnVol_{m,t} = α + β_1 · ΔPremIncome_{m,t} + β_2 · ΔRate_{t}
              + β_3 · ΔInventory_{m,t} + state_FE + ε

Spec B (prices)
ΔHPI_{m,t} = α + β_1 · ΔPremIncome_{m,t} + β_2 · ΔMortgagePmt_{m,t}
           + β_3 · ΔEmploy_{m,t} + metro_FE + ε

A 1pp jump in premium-to-income is associated with a 2.8% drop in 12-month transaction volume and a 0.41pp drag on 12-month HPI, controlling for rate moves, inventory, employment, and state fixed effects. Both coefficients are significant past the 99% level. The HPI coefficient is smaller than the transaction coefficient because sellers pull listings before they cut prices — the first-order response to an affordability shock is fewer deals, not lower prices. Prices follow with a 6–9 month lag.

Three things will determine whether this is a two-state problem or a national one:

1. Reinsurance renewal cycle (June + January): 2026 renewals are expected to be flat-to-up-5% after the 2024 softening. A surprise hardening flows to primary premiums in 9–12 months.
2. Non-Gulf-coast hurricane and wildfire frequency: Three back-to-back years of above-average North Atlantic hurricane counts reset the base rate for what counts as "normal" in actuarial models. 2024 was the fourth above-average year in five; 2025 was above NOAA's preseason forecast.
3. State-level policy action: California's rate approval backlog is down from 18 months in 2022 to 6 months in 2025 after the DOI reforms; Florida's tort reform is flowing through the 2026 rate cycle. Both should stabilize the voluntary market at the cost of one-time rate hikes already filed.

For the forecast model, we added a "premium shock" residual to the Q2 2026 run for the 27 metros where 2025 P/I exceeds 3.5%. That residual shaves an average 0.6pp off our 12-month HPI forecast in those markets and tightens the top of the 80% prediction interval. See the methodology note for how the residual is applied.