The National Flood Insurance Program: 22.5 Billion in t...

There is a federal insurance program that most Americans have never bought a policy from, that has never once paid its own way, and that currently owes the United States Treasury about 22.5 billion dollars with no realistic plan to repay the principal. In October 2017 Congress erased 16 billion dollars of that debt in a single stroke, the first time it had ever done so. Within eight years the program was back underwater by more than the amount it had been forgiven. The program is the National Flood Insurance Program, and for the overwhelming majority of households in America it is the only place to buy flood coverage at all. Whether you read it as a fiscal sinkhole or as a necessary backstop depends almost entirely on which of its numbers you decide to look at.

What it is, and who actually runs it

The National Flood Insurance Program was created by the National Flood Insurance Act of 1968 and is administered by the Federal Emergency Management Agency, according to the Congressional Research Service. It exists because flood is close to an uninsurable peril for a private market. Flood losses are spatially correlated and catastrophic: when one house on a street floods, so do the next fifty, all at once, which is the opposite of the independent, diversifiable risk that ordinary insurance is built to spread. For most of the twentieth century private insurers simply would not write residential flood coverage, and Congress stepped in to fill the gap.

The program is large. FEMA reports more than 4.7 million policies in force providing over 1.3 trillion dollars in coverage across more than 22,000 participating communities. Those are rounded headline figures. A precise snapshot from the Federal Register puts it at 4,664,515 policies in force as of April 30, 2024, of which about 3.25 million were residential dwelling policies, roughly 1.1 million condominium-association policies, about 291,000 commercial, and a small remainder of group policies. Annual premium revenue runs around 4.3 billion dollars.

Yet coverage is thin outside mapped high-risk zones. Only about 3.3 percent of United States households carry an NFIP policy, a secondary analysis of FEMA data that also shows enormous state-by-state variation: over 20 percent take-up in Louisiana and nearly 18 percent in Florida, but under 1 percent across more than two dozen states. The program underpins mortgage lending in flood-prone regions because federal law requires flood insurance for federally backed mortgages on properties inside Special Flood Hazard Areas. For millions of households, in other words, the policy is not optional, and the NFIP is the only vendor.

How it went underwater

For its first three decades the program roughly broke even in ordinary years and quietly borrowed from the Treasury in bad ones. Then came the catastrophes. Hurricane Katrina in 2005 and Hurricane Sandy in 2012 drove the program's debt from near zero into the tens of billions, the Congressional Research Service reports. Katrina alone pushed the cumulative deficit past 17 billion dollars. In the wake of Sandy, Congress raised the program's statutory borrowing authority, the legal ceiling on how much FEMA may owe the Treasury, to 30.425 billion dollars in 2013.

The 2017 hurricane season broke the arithmetic again. Facing claims from Harvey, Irma, and Maria with the program already owing about 24.6 billion dollars and pressed against its ceiling, Congress cancelled 16 billion dollars of NFIP debt in October 2017, the first cancellation in the program's history, enacted through Public Law 115-72. The cancellation was sized to restore borrowing headroom rather than to match that single year's losses, and it comfortably exceeded them: the Congressional Research Service records that only about 10.2 billion dollars was ultimately paid for Harvey, Irma, and Maria, a gap that one advocacy critique from the R Street Institute has pointed to as evidence the forgiveness outran the actual claims. Treat that framing as advocacy-flavored, but the underlying figures hold up.

Forgiveness did not fix the structure. Just weeks after the cancellation, the program borrowed 6.1 billion dollars more on November 9, 2017, bringing its debt to 20.525 billion. Then in February 2025, following heavy Hurricane Helene and Milton claims, FEMA borrowed another 2 billion dollars, raising the debt to 22.525 billion dollars and leaving only about 7.9 billion dollars of unused authority under the 30.425 billion cap. That is the current figure. It is worth keeping the two debt numbers straight: the roughly 22.5 billion dollars still outstanding is smaller than the roughly 36.5 billion the program has cumulatively borrowed since 2005, because cancellations and partial repayments sit between the two.

One more caveat matters for anyone reading this in 2026. The borrowing cap and the program's own authority to write new policies are not permanent. The NFIP operates on short-term reauthorizations, has lapsed briefly more than once, and has on occasion been reauthorized retroactively after a gap. Absent reauthorization the borrowing authority reverts to a fraction of the current ceiling. The specific expiration date and reauthorization status shift with each act of Congress, so verify the live status before relying on the 30.425 billion figure as settled.

Why it looks like a loss

Line the program up against ordinary actuarial soundness and it fails, by design. FEMA is statutorily required to charge some rates below full risk, and increases are capped, so the program has never been able to price its way to solvency. The sharpest illustration is repetitive loss. The Government Accountability Office reports that unmitigated repetitive-loss properties are about 2.5 percent of NFIP policies but account for roughly 48 percent of claims by dollar value, a figure FEMA attributes to properties with two or more losses. A small sliver of structures drives nearly half the payouts.

That is a textbook moral-hazard dynamic: subsidized insurance rebuilds the same flooded houses again and again instead of funding a buyout or an elevation that would end the cycle. The exact ratio depends on how you define the term and which year you measure. Older analyses using narrower definitions put repetitive-loss properties nearer 1 percent of policies and something like a quarter to a third of claim dollars, and the Pew Charitable Trusts documented the same disproportion in 2016 using its own accounting. Do not treat any single pair of numbers as canonical. What is durable across every vintage is the shape: a tiny share of properties producing an outsized share of losses.

The stock of these properties is large and growing. By one dated count there are more than 160,000 repetitive-loss properties and roughly 21,000 severe-repetitive-loss properties nationwide, though FEMA does not publish a single continuously updated public total, so treat those counts as order-of-magnitude figures from a particular snapshot rather than a live tally. The statutory definitions are precise: a repetitive-loss property has two or more claims of at least 1,000 dollars in any ten-year period since 1978, while severe repetitive loss, defined at 42 U.S.C. 4104c, requires four or more claims over 5,000 dollars each with cumulative payments above 20,000 dollars, or two or more claims whose total exceeds the building's value.

Because the affordability caps are written into law rather than chosen by FEMA, the program lands on the Government Accountability Office's High-Risk List and has stayed there continuously since March 2006, added after the 2005 hurricanes left a deficit then near 17 billion dollars. GAO's standing recommendation is that Congress pursue comprehensive reform across six fronts: the existing debt, the legislative barriers to full-risk pricing, affordability, consumer participation, barriers to private-sector involvement, and flood resilience and mitigation.

Ratio-adjusted for who it serves

Now change the measure, and the ledger reads differently. The reform FEMA is actually attempting is Risk Rating 2.0, an actuarial overhaul it branded "Equity in Action" that replaced a rating approach largely unchanged for about fifty years. It rolled out in two phases, Phase 1 on October 1, 2021 for new policies and voluntary renewals, and Phase 2 on April 1, 2022 for all remaining renewals. Instead of pricing a property to its flood-zone map, it prices each property to its own full risk. It is frequently misreported as a blanket rate hike. It was not: FEMA data show a large share of policyholders, particularly owners of lower-value homes who had been overpaying relative to their risk, saw immediate decreases or only small changes.

Where premiums do rise toward full risk, statutory guardrails slow the climb. Under the 2014 Homeowner Flood Insurance Affordability Act, most primary-residence increases are capped at 18 percent per year until the full-risk rate is reached, at which point they stop. Non-primary residences, businesses, severe-repetitive-loss properties, and substantially damaged or improved buildings face a higher cap of up to 25 percent per year. These caps are the statutory glide path, not an invention of Risk Rating 2.0, and they are the direct reason the program stays in the red: they are Congress choosing affordability over solvency, year after year, and the debt is the running receipt for that choice.

Set against that reading, the debt looks less like mismanagement and more like the accumulated cost of a deliberate policy. The program pays out far more in flood claims than any private market historically would have covered, and it produces public goods that never show up as claims at all: FEMA's flood mapping, the floodplain-management standards adopted by more than 22,000 participating communities, and the buyout and elevation programs that permanently retire the worst repetitive-loss structures. The private flood market that critics point to as an alternative exists today, but it remains thin, and much of it depends on the very maps and standards the federal program produces.

The ledger reading

The National Flood Insurance Program is, at the same moment, a subsidy that rebuilds the same flooded houses over and over and the only reason millions of Americans can hold a mortgage in a flood zone at all. Both are true, and they are answers to two different questions. Measured as an insurer, it is insolvent by construction, kept afloat by a Treasury credit line and periodic debt forgiveness, with a moral-hazard core that GAO has flagged for two decades. Measured as public infrastructure for a peril the market abandoned, it is close to irreplaceable, and its 22.5 billion dollar debt is not evidence of a scandal so much as the price tag of affordability caps that Congress writes and rewrites on purpose.

What makes the arithmetic worth watching is that the forgiveness does not fix the structure. Cancel 16 billion dollars and the program borrows its way back to a larger number within a decade, because the thing driving the debt is not a one-time accident but a standing decision to charge less than the risk costs. Risk Rating 2.0 is the first serious attempt in fifty years to close that gap, and the 18 and 25 percent caps are Congress deciding how slowly it is allowed to close. Which number you lead with, the debt or the coverage, is usually a decision you have already made before you open the spreadsheet.

Fact-check notes and sources

Created by the National Flood Insurance Act of 1968, administered by FEMA, with Katrina (2005) and Sandy (2012) driving its debt into the tens of billions: the Congressional Research Service program overview.
More than 4.7 million policies, over 1.3 trillion dollars in coverage, and 22,000-plus communities as headline figures; about 4.3 billion dollars in annual premium: FEMA. The precise 4,664,515 policies in force as of April 30, 2024, with the residential, condo, commercial, and group breakdown: the Federal Register installment-payment-plan rule, which quotes FEMA's own data.
Only about 3.3 percent of United States households carry NFIP coverage, with wide state variation: ValuePenguin's analysis of FEMA data, a secondary source, reliable but not the primary record.
The 30.425 billion dollar borrowing authority set in 2013; the 16 billion dollar debt cancellation of October 2017 (Public Law 115-72), with the program owing about 24.6 billion beforehand; the 6.1 billion dollar November 2017 borrow to 20.525 billion; and about 10.2 billion dollars actually paid for Harvey, Irma, and Maria: the Congressional Research Service on NFIP borrowing. The February 2025 borrow of 2 billion dollars to a current 22.525 billion dollar debt with roughly 7.9 billion in remaining authority: the FEMA press release. The reauthorization and lapse history noted in the text is time-sensitive and should be verified against the current CRS reauthorization product at publish time.
The R Street framing that the 16 billion dollar cancellation exceeded actual 2017 claims: Insurance Journal's R Street commentary, an advocacy source; the underlying claim figures are corroborated by CRS.
Repetitive-loss properties at about 2.5 percent of policies but roughly 48 percent of claims by dollar value: the Government Accountability Office, March 2026, attributing the figure to FEMA. The older, definition-dependent estimates and the durable disproportion: the Pew Charitable Trusts, 2016. The stock counts of roughly 160,000 repetitive-loss and 21,000 severe-repetitive-loss properties are order-of-magnitude figures from a dated snapshot; FEMA publishes no single continuously updated public total. The statutory definitions: 42 U.S.C. 4104c.
Risk Rating 2.0 phase dates (October 1, 2021 and April 1, 2022) and the 18 percent primary-residence HFIAA cap: FEMA on risk rating. The up-to-25 percent cap for non-primary, business, and certain high-risk categories: FEMA and FloodSmart FAQs.
NFIP on the GAO High-Risk List since March 2006 and the six-part reform framework: the Government Accountability Office High-Risk page and its explanation of why the program was listed.

This post is informational and journalistic, describing a public program and public records. It is not legal, financial, or policy advice. Figures are drawn from government reports and public law, with derived and dated figures noted as such; where advocacy or secondary sources are cited, they are labeled. The program's reauthorization and borrowing-authority status can change with any act of Congress, so verify current figures before relying on them.

The National Flood Insurance Program: 22.5 Billion in the Red, and Why That Was the Plan