The Insurer of Last Resort

The problem

Home insurance is quietly becoming the mechanism by which climate risk gets priced into American life, and it is doing it faster and more brutally than any policy debate.

In California, major national carriers stopped writing new homeowners policies. The FAIR Plan, the state's insurer of last resort, more than doubled its policy count in four years and its exposure passed $450 billion. When the January 2025 Los Angeles fires hit, the plan levied an assessment on member insurers of roughly $1 billion, which is a cost that flows back to every other policyholder in the state.

In Florida, Citizens Property Insurance, also designed as a last resort, at its peak carried around 1.4 million policies and was among the largest property insurers in the country. A last resort holding a plurality of the market is not a last resort. It is the market, with a state balance sheet behind it.

Average American homeowners premiums have risen roughly 30 to 40 percent in four years, far faster than inflation. The United States recorded 28 separate billion-dollar disasters in 2023 and a comparable number the following year, against a long-run average in the single digits.

The National Flood Insurance Program owes the Treasury on the order of $22.5 billion and has never charged actuarially sufficient rates. Roughly 1 percent of its insured properties, the repetitive-loss properties, generate something like a quarter to a third of all claims. These are houses that flood, get rebuilt with federal money, and flood again, in a cycle that has run for decades and serves nobody living in them.

Here is why this is a covenant problem rather than an insurance problem.

A house that cannot be insured cannot be mortgaged, because lenders require coverage. A house that cannot be mortgaged can only be sold for cash. Uninsurability collapses property value, and for most American families the house is the entire balance sheet. The mechanism by which a retired couple in Louisiana or Paradise, California loses everything is not the fire or the flood. It is the renewal notice.

And roughly one in eight American homeowners now carries no insurance at all, a number that rises sharply as premiums do, which means the exposure is migrating from insurers' books to households' lives.

The proposal

A federal reinsurance backstop for catastrophic tail losses, priced and capped, in exchange for honest risk-based rates, mandatory mitigation funding, and a real buyout option for the houses that cannot be saved.

How it would work

1. Backstop the tail, not the risk. A federal reinsurance facility that attaches only above a very high industry-wide loss threshold in a single event or year, charging an actuarially estimated premium and capped in total exposure. The precedent exists and works: the Terrorism Risk Insurance Act has provided exactly this structure since 2002, has been reauthorised repeatedly, and has cost the taxpayer nothing while keeping a market alive that would otherwise not exist.

1. The trade: rates tell the truth. States accessing the backstop must permit risk-based pricing in the primary market. Suppressed rates are why carriers left California, and a backstop layered on top of price controls simply moves the loss to the federal balance sheet while keeping the underlying signal switched off.

1. Mitigation, funded and mandatory. A federal grant covering the large majority of the cost of hardening a home to a published standard, roof, openings, defensible space, elevation, targeted first at the highest-risk and lowest-income households. The engineering here is unusually well established: the FORTIFIED standard measurably reduces losses, Alabama has run a working state programme for years, and the National Institute of Building Sciences puts the return on mitigation spending at roughly $6 saved per $1 spent, higher for some hazards.

1. Mitigation must be priced. Insurers accessing the backstop must offer defined premium credits for certified mitigation. Today a homeowner can spend $15,000 hardening a roof and see no change in their bill, which is a complete failure of the price signal to do the one job this proposal depends on.

1. A standing buyout offer for repetitive loss. Any property meeting a published repetitive-loss threshold receives a standing, voluntary, pre-funded offer at pre-disaster fair market value, available at any time rather than only in the chaotic weeks after a flood. Accept and the land converts permanently to open space with a deed restriction. Decline and keep the house, with future federal rebuild assistance for that property reduced on a published schedule. Nobody is moved. Everybody knows the terms in advance.

1. Stop rebuilding in the footprint. Federal disaster and mortgage support conditioned on rebuilding to the mitigation standard, and withheld for reconstruction inside mapped extreme-hazard zones. This is the clause that will draw the most opposition and it is the one that determines whether the rest is a fix or a subsidy.

1. Do not let it become the market. The facility carries a statutory sunset and a published market-share trigger: if primary carriers do not return, the programme has failed at its purpose and requires re-legislation rather than quiet expansion.

The numbers

The backstop. Attaching above, say, $50 billion of insured loss in a single event, the facility would have been triggered only a handful of times in modern American history. Priced at an actuarial premium with a cap, its expected annual cost is plausibly near zero in most years, exactly as TRIA's has been, with a large contingent exposure that is the entire point of a government being the one to hold it. Government is the only balance sheet that can absorb a correlated national tail loss, and that is the sole legitimate reason for it to be here.

Mitigation. Hardening a house to the FORTIFIED standard runs roughly $5,000 to $15,000 on a retrofit. Doing 200,000 homes a year at an average federal share of $10,000 is $2 billion a year. At a 6:1 return, that is $12 billion of avoided loss annually, and the avoided loss falls on households, insurers and the federal disaster budget alike.

Buyouts. Roughly 30,000 to 40,000 repetitive-loss properties are on the NFIP's books. At an average of $250,000, buying all of them over a decade is about $1 billion a year, against a claims history on those same properties that has already exceeded their value, in some cases several times over.

Total: roughly $3 billion a year of direct spending, plus a contingent tail exposure priced and capped.

Against: a $22.5 billion flood programme deficit that grows, an insurance market withdrawing from multiple states, and the largest asset most American families own becoming unfinanceable.

The honest objections

"This is a federal subsidy for people who chose to build in fire and flood zones, paid by people who did not." The strongest objection, and items 2, 5 and 6 exist to answer it. The backstop is priced and attaches only at a catastrophic tail; the rates underneath must be actuarially honest, which is the opposite of a subsidy; and federal support for rebuilding in the worst locations is reduced rather than extended. What remains is a genuine cross-subsidy in the mitigation grants, which we defend on the grounds that hardening a house benefits the disaster budget and the insurance pool as well as its owner, and that the alternative is paying more later through FEMA.

"Risk-based pricing will force elderly and low-income homeowners out of houses they have owned for forty years." True, and it is the real human cost of item 2. Someone who bought in 1978 did not misprice anything; the risk changed underneath them. This is why item 3 targets mitigation grants at exactly that population first, and why item 5's buyout is at pre-disaster value rather than post-disaster value, which is the difference between a dignified exit and a ruinous one. A means-tested premium voucher, transparently on-budget and paid to the household rather than through suppressed rates, is the right instrument for the remaining hardship. Hiding the subsidy inside the rate is what produced the current situation.

"Managed retreat is politically radioactive and you have written it into a federal statute." Which is why item 5 is voluntary, standing, and paid at pre-disaster value, with the only consequence of declining being a published reduction in future rebuild assistance rather than any compulsion. Existing buyout programmes fail largely because the offer arrives months after the flood, when people have already rebuilt or left, and the money runs out. A standing offer that a family can consider calmly in an ordinary year is a different instrument with the same name.

"Insurance is regulated by the states under McCarran-Ferguson. This is a federal takeover." The conditionality in item 2 is attached to voluntary participation in a federal reinsurance facility, not imposed on states directly, which is a well-worn constitutional path. It is still real pressure on state regulators, and states that prefer to keep rate suppression may decline the backstop and keep both. Whether that choice remains available in practice once neighbouring states join is a fair question, and the answer is probably not entirely.

"You are insuring a risk that is growing without bound. Eventually the tail eats the backstop." The most honest long-run objection. A backstop makes the transition survivable; it does not stop the hazard from worsening, and if losses keep compounding, the facility becomes permanent and enormous. That is why item 7 has a sunset, and it is also why this proposal is a companion to № 038 rather than a substitute for it. Insurance prices risk. It does not reduce it.

Sources

• California FAIR Plan policy growth and exposure; 2025 member assessment following the January 2025 Los Angeles fires (cfpnet.com)
• Citizens Property Insurance Corporation policy counts and depopulation programme (citizensfla.com)
• Billion-dollar weather and climate disaster counts; NOAA National Centers for Environmental Information (ncei.noaa.gov)
• National Flood Insurance Program debt to the Treasury and repetitive-loss property claim concentration; GAO and CRS reporting (gao.gov)
• Terrorism Risk Insurance Act of 2002 and subsequent reauthorisations, as the structural precedent for a priced federal reinsurance backstop
• Insurance Institute for Business and Home Safety, FORTIFIED standard; Strengthen Alabama Homes programme (ibhs.org)
• National Institute of Building Sciences, Natural Hazard Mitigation Saves (2019), estimating approximately $6 saved per $1 of mitigation investment
• Share of US homeowners without insurance; Consumer Federation of America
• Proposals № 026 (The Landing); № 032 (The Housing Dividend); № 034 (The Water Compact); № 038 (The Carbon Dividend)