Moody's $375B Flood Gap: Property Catastrophe Tail Risk Beyond the Insured Loss

A 1-in-100-year U.S. flood could leave $375 billion of aggregate uninsured residential loss, with 65% of projected economic flood damage falling outside any policy, on Moody's RMS modelling published in May 2026. At the 1-in-500-year level the uninsured figure passes $1 trillion. The number is not an insured loss estimate, which is exactly what makes it relevant to carriers writing no flood at all.

Key Takeaways

• $375 billion uninsured at the 1-in-100-year level, rising above $1 trillion at 1-in-500. These are economic losses outside the insured system, not modelled insured losses from a defined event set.
• Fewer than 2% of counties across 11 states carry 65% of national uninsured exposure, with individual counties in Florida, Louisiana, South Carolina and Texas each above $5 billion.
• Buncombe County, North Carolina ran an 88% protection gap through Helene's 1-in-1,000-year rainfall: roughly 12% of residential flood damage had any insurance recovery.
• 32% of affected mortgage borrowers in a studied North Carolina flood event could not finance repairs through income or home equity, with 66% of property damage uninsured.
• The NFIP owed $22.525 billion to Treasury as of February 2025 with $7.9 billion of borrowing authority left under a $30.425 billion cap, against a structural annual deficit near $1.4 billion.

What Kind of Number $375 Billion Is

The RMS US Inland Flood HD model covers fluvial flooding, pluvial flooding and coastal storm surge. FEMA's Special Flood Hazard Area maps are calibrated mainly to riverine flooding at the 1-in-100-year return period and do not integrate extreme precipitation or surge, which are the flood types most likely to hit properties with no NFIP policy.

Source: Moody's RMS US Inland Flood HD model, May 2026

Scenario	Aggregate Uninsured Loss	Protection Gap	States Exceeding 10% Loss-to-Replacement
1-in-100-year (current)	$375 billion	65%	6
1-in-500-year (current)	>$1 trillion	>70%	16
1-in-100-year (2050, RCP 4.5)	~$472 billion	65%	~8

Six states already cross a 10% loss-to-residential-replacement ratio at the 1-in-100-year severity: Florida, Kentucky, Louisiana, South Dakota, South Carolina and Texas. At 1-in-500 the count reaches 16 and the footprint widens inland to Pennsylvania, Illinois, New Jersey and New York. The 2050 RCP 4.5 pathway puts a 1-in-100-year event at roughly $472 billion uninsured, 25% above the current baseline, on a static coverage assumption: the gap widens because economic loss grows, not because take-up falls further.

The $375 billion is an aggregate of residential uninsured exposure across counties at a severity level, not a modelled single-event insured loss. A vendor cat model measures loss inside a defined insured event set. This measures the universe sitting outside it.

Buncombe County is the empirical check. Helene delivered 1-in-1,000-year rainfall intensity to Asheville in September 2024, and the county's protection gap across the 1-in-100 to 1-in-500 range runs near 88%. It had no significant riverine flood history in FEMA maps, so most homeowners were neither required to carry NFIP coverage nor chose to.

The Transmission Runs Through Collateral, Not Claims

The channel that reaches an admitted P&L does not go through flood claims. It goes through property values and mortgage credit, and it reaches carriers that write no flood.

A 2026 paper in Natural Hazards and Earth System Sciences examined residential mortgage borrowers in a North Carolina county flood event and found 32% lacked the income or collateral to finance repairs through home equity borrowing, with 66% of property damage outside any policy. Households in that position either fall behind on the mortgage or sell into a distressed market. Both compress values in the affected geography, and compressed values in a flood-exposed county are what pushes admitted homeowners writers to restrict new business.

FAIR plans absorb what the voluntary market leaves. California's FAIR Plan enrollment rose 43% between September 2024 and December 2025 after wildfire-driven private exits. The sequence transfers cleanly to flood: availability narrows, residual market participation grows, and losses arrive at a residual market that needs assessment to pay them.

That assessment is allocated to admitted writers in proportion to voluntary market share in the state. A carrier with no direct flood exposure in its admitted book can receive an assessment from a state where flood losses drove residual market growth. A cat model run against the admitted portfolio does not see it, and neither does a flood stress test scoped to the insured layer.

The same event compresses the assessed value base securing municipal general obligation debt. Moody's moved Buncombe County's outlook from negative to stable in April 2026, roughly 18 months after Helene, on recovering fund balances, with federal disaster assistance and state resources doing much of the stabilizing. That pipeline does not scale at the same rate when several counties in one state are hit at once.

The Insured 35% Contains Its Own Underinsurance

The NFIP carries roughly 4.6 million policies and about $1.3 trillion of aggregate coverage, and it is the primary backstop for the insured 35% of economic flood loss at the 1-in-100-year level. That share is thinner than the count suggests.

Residential NFIP structural coverage is capped at $250,000. In coastal Florida, South Carolina and Texas, that cap describes a large share of housing stock as structurally underinsured before any protection gap calculation begins. Private flood remains roughly 10% of insured flood risk nationally, so the excess layer above the cap is largely unfilled.

The program's own balance sheet constrains the rest. Annual premium of $4.3 billion runs against roughly $5.8 billion of programmatic cost, a structural deficit near $1.4 billion a year (Congressional Research Service). Katrina alone triggered $16.3 billion of NFIP claims and started borrowing the program has never repaid. A scaled 1-in-100-year event could exhaust the remaining $7.9 billion of authority in one claims cycle, before the uninsured share of that same event is counted.

Take-up is moving the wrong way in the places that matter. The 65% gap is a national average; county-level gaps run from 45% to above 90% depending on participation, the share of property value above the $250,000 cap, and the rate path under Risk Rating 2.0, phased in from 2022. Counties where participation fell after those premium increases are the counties where the uninsured fraction is growing fastest, which means the national average understates the gap on exactly the exposure a coastal book is holding.