227,000 homes keep flooding; FEMA fixed 13,000 since 1989
Summary
A March 2026 GAO testimony to Congress finds that NFIP's 'repetitive loss' properties -- those that have flooded and collected an insurance payout two or more times -- make up about 2.5 percent of the National Flood Insurance Program's policies but accounted for 48 percent of its claims by dollar value as of December 2021. The unmitigated backlog has grown from about 131,000 properties in 2009 to 227,000 in 2026, a 73 percent increase, while FEMA has permanently mitigated just 12,972 repetitive-loss properties across all its grant programs since 1989 -- about 5.7 percent of the current backlog. In North Carolina, at least 575 of the 657-plus Hurricane Helene acquisition applications logged as of February 2026 remained unapproved -- 17 months after the storm. GAO, which put NFIP on its High-Risk List in 2006, projects the program's own premium shortfall at $26.7 billion through 2037 even after a 2023 pricing overhaul, and NFIP's current authorization expires September 30, 2026.
A backlog FEMA clears slower than it grows
can permanently reduce a property's flood risk four ways: buying and demolishing it (acquisition), raising it above the flood line (elevation), moving it (relocation), or hardening it against water (floodproofing). Since 1989, has completed 95,762 of these projects nationwide -- but only 12,972 of them, about 14 percent, were repetitive-loss properties, the ones says drive the disproportionate claims. The rest of the repetitive-loss inventory has simply kept accumulating: up 73 percent since 2009, according to 's own count.
View data as table
| 2009 | 131,000 |
|---|---|
| 2018 | 176,000 |
| 2026 | 227,000 |
Why the fix moves so slowly
The mitigation does complete works mostly through one channel: 77 percent of all property mitigations since 1989 were funded by the Hazard Mitigation Grant Program (HMGP), which only opens up after a president declares a major disaster -- meaning a home has to flood badly enough to trigger a declaration before the money to fix it becomes available. Acquisition, the method that eliminates flood risk outright rather than just reducing it, accounts for 72.5 percent of everything has mitigated; elevation covers most of the rest. But says the acquisition process 'typically takes at least 2 to 3 years, and often longer,' and states took an average of 16 months after a disaster declaration just to submit an application, because local governments often lack the grant-management staff to navigate environmental reviews and a nonfederal cost share -- normally 25 percent of the project -- that can leave homeowners owing money even after their house is bought out.
Hurricane Helene, which struck North Carolina in September 2024, shows what that delay looks like in the field: as of February 2026 -- 17 months after the storm -- had approved just 82 acquisitions of Helene-affected properties, according to the state's Department of Public Safety, with more than 575 additional applications submitted but not yet approved -- at least 657 in all, and possibly more. Every month those applications sit unapproved is a month the underlying properties stay exactly as they were -- unmitigated, still insured, still eligible to flood and claim again.
View data as table
| Acquisition | 72.5% | 69,415 properties |
|---|---|---|
| Elevation | 23% | 22,039 properties |
| Relocation + floodproofing | ~4.5% | remainder of 95,762 total |
The bill sitting behind the backlog
Mitigation is the fix on the cost side; the other lever is the price of the policy itself, and NFIP is statutorily barred from charging what full flood risk actually costs. As of December 2022, 66 percent of NFIP policies still carried discounted premiums, and estimated it would take until 2037 for 95 percent of policies to reach full-risk pricing -- even after 's April 2023 overhaul that began pricing individual properties by risk rather than by flood-zone averages. 's own arithmetic puts the resulting shortfall at $2.7 billion in 2023 alone and $26.7 billion cumulatively through 2037. Since 2005, that gap has forced to borrow roughly $38.5 billion from the U.S. Treasury just to keep paying claims -- debt that, unlike a normal insurer's, taxpayers ultimately stand behind.
has recommended that Congress authorize and fund means-based premium assistance for policyholders who can't afford full-risk rates, so pricing reform doesn't just default to slower discounts, and that target mitigation dollars at the repetitive-loss properties driving the shortfall rather than spreading them evenly. As of February 2026, Congress had not enacted the assistance recommendation. Across 's 2017-2023 body of work on NFIP, it made nine recommendations to and eight to Congress; has fully implemented four of its nine -- meaning five recommendations agreed to, in some cases years ago, remain undone, with no statutory deadline forcing action. NFIP itself needs reauthorizing again by September 30, 2026, the same deadline every version of this program has run up against since 2017.
The takeaway
- A small slice of policies drives the program's losses, and it keeps growing. Repetitive-loss properties are about 2.5 percent of NFIP policies but 48 percent of its claims by dollar value; the unmitigated count has grown 73 percent since 2009, reaching 227,000 in 2026.
- The one fix that works is too slow to keep up. has mitigated just 12,972 repetitive-loss properties in 37 years -- about 5.7 percent of the current backlog -- because acquisitions typically take 2 to 3 years and depend on a disaster declaration to unlock funding in the first place; at least 575 of 657-plus Hurricane Helene applications in North Carolina were still pending 17 months after the storm.
- 's own fixes are half-adopted and Congress hasn't acted on the rest. Four of 's nine -directed recommendations are implemented; five are not, some dating to 2020 or earlier. Congress has not funded the means-based premium assistance recommended in July 2023, and NFIP's authorization -- lapsed twice in the past year -- must be renewed again by September 30, 2026.
All figures are from -26-109045, 'Flood Risk Mitigation: Reducing Fiscal Exposure and Improving Affordability,' the written testimony of Director Alicia Puente Cackley before the House Financial Services Subcommittee on Housing and Insurance, delivered March 26, 2026 and read in full. The testimony synthesizes and updates four earlier audits (-17-425, -20-508, -22-106037, -23-105977) with data current through January-March 2026; where this piece cites a year for a figure, it is the 'as of' date itself gives, not this outlet's estimate. The mitigation-rate, backlog-growth, Helene-backlog-share, and repeat-loss-share-of-total-mitigated percentages are this outlet's own arithmetic on 's reported counts (methods and caveats in analysis.json); 's own reported growth rate ('about 73 percent,' 2009-2026) independently matches our recomputation. 's Helene figures ('more than 575' pending) are themselves a lower bound, so the 657-plus total and 87.5% pending share are reported here as 'at least,' not exact. gao.gov blocks automated PDF access (HTTP 403), so the capture rides a Wayback Machine snapshot of the same document.
Sources(1) ▾
- U.S. Government Accountability Office, Flood Risk Mitigation: Reducing Fiscal Exposure and Improving Affordability (GAO-26-109045) (2026-03-26) — Written statement of Alicia Puente Cackley, Director of Financial Markets and Community Investment, delivered March 26, 2026 before the House Financial Services Subcommittee on Housing and Insurance. Draws on and updates figures from four underlying audits (-17-425, -20-508, -22-106037, -23-105977) with data current as of January-March 2026. Source for the repetitive-loss property counts and growth trend, the 2.5%-of-policies/48%-of-claims disproportion, the mitigation totals and methods, the Hurricane Helene acquisition-backlog figures, the premium-shortfall projection through 2037, and the nine /eight Congress recommendation count. gao.gov · original document
Comments
Always open. Logged-in readers can annotate paragraphs in place.
A March 2026 GAO testimony to Congress⧉ says a class of homes insured by the National Flood Insurance Program (NFIP) -- 's program that sells flood coverage most private insurers won't, because Congress required it after the private market retreated in the 1960s -- keeps generating an outsized share of the program's losses. These 'repetitive loss' properties, ones that have flooded and collected an insurance payout two or more times, make up about 2.5 percent of NFIP's policies but accounted for 48 percent of its claims by dollar value as of December 2021. -- Congress's own auditor, whose findings federal agencies are expected to act on -- put NFIP on its High-Risk List in 2006 over exactly this exposure. Two decades later, the properties driving it have kept multiplying anyway: the unmitigated backlog grew from about 131,000 in 2009 to 227,000 in 2026, even as has permanently fixed only 12,972 of them in 37 years.