Flood-prone coastal counties host ~25% of new US home construction but ~70-78% of national flood-insurance claims and losses

Summary

Using FEMA's own coastal high-hazard flood-zone codes (V/VE, the open-coast, wave-action zones) to define "flood-prone coastal areas," 193 U.S. counties qualify. From 2009-2024 these counties absorbed 4.69 million of the nation's 18.95 million newly permitted housing units (24.8%) and $1.01 trillion of $3.91 trillion in permit valuation (25.8%) — construction has kept pace with, not fallen behind, the national building cycle. Those same 193 counties also accounted for 676,905 of 966,722 NFIP paid claims nationally (70.0%) and $43.6 billion of $55.8 billion in paid flood losses (78.2%) since 2010. On repeated claims specifically: the connector's claims data has no property-level identifier, so a true FEMA-standard repetitive-loss count (same address, 2+ paid claims within 10 years) could not be computed directly. Falling back to FEMA's own published Multiple Loss Properties registry, 240,651 structures nationwide currently meet repetitive- or severe-repetitive-loss criteria, and a 2004 GAO analysis found such properties are about 1% of all NFIP policies but generated 38% of all claims from 1978-2004 — a concentration other researchers have since put as high as 30% of all claims paid since 1978 from repetitive-loss properties, and roughly 40% of paid losses from the smaller "severe repetitive loss" subset (about 1% of insured properties). Given that coastal counties already account for 70-78% of the nation's flood claims and dollars paid, and that repetitive-loss properties are known to be concentrated overwhelmingly in coastal states (Florida, Louisiana, Texas together account for roughly half of all NFIP claims ever filed), it is a reasonable inference — not a directly computed figure — that a large share of the repetitive-loss inventory sits inside these same 193 counties. That inference could not be confirmed at the property level with the data available.

How "flood-prone coastal areas" was defined

There is no single official U.S. list of "flood-prone coastal counties." NOAA's own coastal-county definitions (452 shoreline counties, or a broader 15%-of-land-in-a-coastal-watershed definition) are geographic, not flood-risk-based, and were not available as a ready-made table in this connector. Rather than guess at a boundary, the analysis anchored the definition in FEMA's own risk classification: a county was counted as "flood-prone coastal" if it has ever recorded a paid NFIP claim in a V or VE flood zone — the FEMA zone codes reserved for open-coast areas subject to storm-wave action, the most severe coastal flood hazard category. That test returned 193 counties, concentrated in Florida, the Gulf Coast, and the mid-Atlantic/Northeast shoreline (the ten highest-claim counties were all in FL, NC, NJ, NY, LA, AL, CT, TX).

This is a conservative, hazard-grounded definition, but it understates true coastal flood exposure two ways: (1) many oceanfront and near-coast properties sit in AE or A zones rather than V/VE, so this undercounts the full coastal floodplain; (2) it only captures counties that have actually filed a V/VE claim during 2010-2026 — a coastal county that has so far avoided a wave-damage claim would not appear even if it carries V/VE-zoned properties.

Building activity in these counties

Source: Census Bureau Building Permits Survey (housing.building_permits), county-annual totals, 2009-2024 (the table's full, gap-free observed window).

The annual coastal share ranged narrowly (roughly 22-28%) across the full cycle — it was not a one-time spike. Annual coastal permitting nearly tripled between 2009 (128,288 units) and its 2022 peak (419,465 units), tracking the same recovery-boom-plateau shape as the national series (see chart). In short, building in these flood-hazard coastal counties has moved in lockstep with the national housing cycle rather than slowing down after loss events.

Flood claims and losses concentrated in the same counties

Source: FEMA OpenFEMA NFIP paid claims (disasters.nfip_claims), 2010-2026 (this table has a confirmed interior gap: no rows loaded for 2018, so all totals below are undercounts by one year's claims).

These 193 counties are about 6% of all U.S. counties/county-equivalents by count, yet they generate seven out of every ten NFIP claims and nearly four out of every five dollars paid out nationally — a concentration far beyond their 25% share of new construction.

Repeated claims / repetitive-loss properties — what could and could not be computed

FEMA defines a Repetitive Loss (RL) property as one with two or more paid claims of $1,000+ within any 10-year period, and a Severe Repetitive Loss (SRL) property as one with four or more claims each exceeding $5,000 (cumulative $20,000+). Both definitions require tracking claims back to a specific insured structure over time.

The askamerica NFIP claims table (disasters.nfip_claims) carries no policy or property identifier — only a census tract, ZIP code, and latitude/longitude rounded to one decimal degree (roughly county-scale precision, far too coarse to distinguish individual addresses). A tract-level proxy was tried (grouping claims by census tract and flagging tracts with 2+ claims over the full 2010-2026 span), but this measures something different and much looser than FEMA's address-level definition — over a 16-year window spanning many separate storms, 11,443 of 14,039 coastal tracts (99.0% of coastal claims) had at least two claims, which mostly reflects that flood-prone tracts get hit by more than one storm over 16 years, not that individual properties are repeat claimants. That proxy was discarded as too crude to report as a finding.

Falling back to FEMA's own published property-level dataset instead: OpenFEMA's NFIP Multiple Loss Properties registry (last refreshed August 2026) lists 240,651 structures nationwide currently meeting RL or SRL criteria. A 2004 GAO study (cited via Resources for the Future's retrospective analysis) found that such properties made up about 1% of all NFIP policies but generated 38% of all paid claims from 1978-2004; more recent tallies put repetitive-loss claims at roughly 30% of all claims paid since 1978, and severe-repetitive-loss properties alone (about 1% of insured properties) at roughly 40% of paid losses. The same literature shows claims are heavily concentrated in coastal states — Florida, Louisiana and Texas together account for close to half of all NFIP claims ever filed, and Florida alone carries over 33,000 multiple-loss properties including 5,400+ severe repetitive loss properties. Because this connector could not join the national repetitive-loss registry to the 193-county coastal set at the property level, the true "how much coastal new construction sits in repeat-claim areas" question is answered qualitatively here (coastal counties dominate both the national repetitive-loss inventory in the cited literature and the raw claims/dollars in askamerica's own data) rather than with a single computed percentage.

What This Report Does Not Answer

• Quantify how much of that building is in places with repeated insurance claims: Could not be computed directly: the NFIP claims table has no property-level identifier, so FEMA's actual repetitive-loss definition (same address, 2+ claims in 10 years) cannot be applied, and therefore cannot be joined back to specific building-permit locations. Reported instead: (a) the 193 coastal counties' claims/dollar share of the national total (70-78%) as a place-level proxy for loss concentration, and (b) FEMA's own national repetitive-loss property counts and their historical claims share, cited from FEMA/GAO sources, with the gap to a true coastal-building-permits-in-repeat-loss-areas figure stated explicitly.