Insurance & Actuarial Risk
The system that prices risk, pools losses, and - increasingly - withdraws from the places that need it most.
(Swiss Re sigma; about 7% of world GDP)
(Swiss Re; below the ~$140 billion trend, with wildfires, storms and floods a record 92%; first half of 2026: $42 billion)
(the "protection gap"; in 2025 insurers covered 49% of the $220 billion economic loss, the highest share Swiss Re has recorded)
A note on framing. Insurance is not a topic that generates popular attention until it fails: a hurricane destroys homes and insurers refuse to pay, or premiums double in a single year and families cannot afford coverage. But insurance is a structural force that shapes where people can live, what they can build, and how risk is distributed across society. This page describes the mechanics and the current stresses, particularly the retreat of insurers from high-risk areas as catastrophe losses rise. Sources include Swiss Re sigma, Munich Re NatCatSERVICE, Aon, the Geneva Association, Lloyd's, the NAIC (National Association of Insurance Commissioners), and the Federal Emergency Management Agency (FEMA).
How insurance works
Insurance is risk pooling. A large group of people each pay a small premium into a common fund. When one of them suffers a loss, the fund pays out. The system works when losses are independent (not everyone floods at the same time), reasonably predictable in aggregate (actuaries can estimate the frequency and severity of claims), and large enough to be worth pooling (nobody insures against dropping a coffee cup). Actuarial science - the mathematics of risk pricing - has been refined since the 17th century, when Edmund Halley (of comet fame) published the first life table.
Two classic problems plague insurance markets. Adverse selection: people who know they are high-risk are more likely to buy insurance, which drives up premiums, which pushes low-risk people out, which drives up premiums further - a spiral that can collapse markets entirely (George Akerlof's "lemons" problem, work that earned him a share of the 2001 Nobel in Economics). Moral hazard: once insured, people may take more risk than they would without coverage (driving faster with good auto insurance, building in flood zones because flood insurance exists). Every insurance system design involves managing these two tensions.
Climate and the insurability crisis
The most consequential stress on the insurance system is the rise in catastrophe losses. Swiss Re estimates that insured losses have grown 5-7% a year in real terms since 1996, and Aon counts six straight years above $100 billion to 2025. How much is attributable to climate change versus increased building in risk-prone areas versus higher property values is debated, but the aggregate trend is clear: losses are rising faster than premiums can sustainably cover in certain regions.
In the US, nonrenewal rates in 2018-22 ran about 80% higher in the riskiest fifth of ZIP codes than in the safest (Treasury). The strain is most visible in California and Florida. State Farm, the largest US property insurer, stopped writing new homeowner policies in California in 2023. The Los Angeles fires of January 2025 caused about $40 billion of insured losses, a wildfire record. California's insurer of last resort, the FAIR Plan (Fair Access to Insurance Requirements), has more than doubled since 2022 to about 697,000 policies and needed $1 billion from private insurers for fire claims. Florida's state-run Citizens Property Insurance shrank from 1.41 million policies in 2023 to about 266,000 in August 2026 as private insurers returned, a recovery it credits to laws curbing litigation. The underlying dynamic is that actuarially priced insurance in high-risk coastal and wildfire zones is unaffordable for many homeowners. Yet premiums held down by regulation or public subsidy create moral hazard, encouraging building where the risk is rising.
The pattern is not uniquely American. In Australia's cyclone-prone north, the government has run a reinsurance pool since July 2022 to bring premiums down. Parts of coastal Bangladesh and low-lying Pacific island nations are effectively uninsurable for flood and storm risk. Turkey's mandatory earthquake insurance (DASK), introduced after the 1999 Marmara earthquake, covered about 56% of homes in 2024 - a high penetration rate by global standards, but the 2023 earthquakes in southeastern Turkey exposed gaps in coverage as well as weak buildings.
Health insurance: where pooling is hardest
Health insurance operates under different economics than property insurance because healthcare demand is not random - it is correlated with age, chronic conditions, and socioeconomic status, making the adverse selection problem particularly severe. The US employer-based system, a historical accident of World War II wage controls, covered about 53% of Americans in 2025 (Census Bureau) but leaves significant gaps. The Affordable Care Act (2010) addressed some of these through individual mandates and subsidized exchanges, but the US remains the only wealthy country without universal coverage, and healthcare costs continue to rise faster than inflation. The enhanced premium tax credits for exchange plans expired at the end of 2025. The health-policy research group KFF estimates enrollees' premium payments rose 58% on average in 2026, and insurers propose a median 15% rise for 2027, citing medical costs and a sicker pool.
Other models exist. The UK's National Health Service is a single-payer, tax-funded system. Germany uses a mixed system of statutory (public) and private insurance with mandated coverage. Japan's universal system combines employer-based and community-based insurance with government subsidies, achieving universal coverage at roughly 11% of GDP (versus about 17% in the US; World Bank, 2023). Singapore's system - mandatory health savings accounts (Medisave), catastrophic insurance (MediShield Life), and government subsidies for low-income patients (Medifund) - achieves universal coverage at about 4.5% of GDP, though with significant out-of-pocket costs. No system perfectly resolves the tension between access, quality, and cost, but the range of outcomes demonstrates that institutional design choices matter as much as the underlying economics.
Flood insurance and managed retreat
Flood risk illustrates the insurance problem most starkly. In the US, the National Flood Insurance Program (NFIP) was created in 1968 because private insurers would not cover flood risk at affordable rates. The program owes the US Treasury about $22.5 billion against a statutory borrowing limit of $30.4 billion, largely because premiums were set below actuarial cost for decades - an explicit political choice to keep insurance affordable in flood-prone areas. Even that figure understates the losses: Congress canceled $16 billion of the program's debt outright in 2017, and it has borrowed its way back to this level since.
FEMA's Risk Rating 2.0 (phased in over 2021-23) attempts to move NFIP premiums toward actuarial pricing. The transition creates affordability crises for homeowners whose full-risk premiums are far above what they now pay: the law caps most annual increases at 18%, so their bills rise year after year. The question of managed retreat - systematically relocating communities from areas where the risk-to-cost ratio makes habitation economically irrational - is one of the most politically difficult issues in climate adaptation. The UK's Flood Re scheme (2016) provides a different model: a reinsurance arrangement funded by a levy on all home insurance policies, designed to keep flood insurance affordable until its planned end in 2039 while at-risk areas are adapted. France's Cat Nat system (catastrophes naturelles) spreads risk across all policyholders through a mandatory surcharge on every property insurance policy.
Reinsurance: the insurance of insurance
Behind every insurance company stands a reinsurer - a company that insures insurers against catastrophic losses. The global reinsurance market is dominated by a handful of firms: Swiss Re (Switzerland), Munich Re (Germany), Hannover Re (Germany), Berkshire Hathaway (US), SCOR (France), and Lloyd's of London (technically a market, not a company). When a hurricane causes $50 billion in insured losses, primary insurers pay the first layer; reinsurers absorb the excess.
Reinsurance pricing acts as an early warning system for systemic risk. When reinsurers raise prices or withdraw from certain categories of risk (as several did for US windstorm and wildfire coverage in 2023-24), it signals that the risks being priced are changing faster than the models can comfortably absorb. Prices also follow capital, now at record levels: property-catastrophe rates fell about 15% at the January 2026 renewals and 15-25% for Florida insurers in June, according to the brokers Howden Re and Guy Carpenter. Either way, the industry's moves cascade through the insurance chain to the homeowner or business paying premiums.
Where analysts disagree
How much of the loss trend is climate versus exposure? Munich Re's chief climatologist says a warming world makes extreme weather disasters more likely; World Weather Attribution found warming made the hot, dry, windy conditions behind the Los Angeles fires about 35% more likely. Swiss Re finds that exposure growth explains more than 80% of the global rise in weather-related insured losses, though North American wildfire and European thunderstorm losses have grown twice as fast as exposure. Roger Pielke Jr., a political scientist reviewing 54 studies that adjust losses for growth in wealth, found little evidence that human-caused climate change explains any part of the global rise in economic losses. Most analysts accept that both contribute; disentangling them precisely is methodologically difficult.
Should government be the insurer of last resort? When private markets withdraw, governments face pressure to step in - as the US did with flood insurance and as various states have done through residual-market programs (Florida's Citizens Property Insurance, California's FAIR Plan). Critics argue that government insurance creates moral hazard by underpricing risk. Advocates argue that some risks are too large and too correlated for private markets and that government has a role in ensuring that essential risk-sharing remains available.
Is parametric insurance the future? Traditional insurance reimburses actual losses, which requires claims adjustment and dispute resolution. Parametric insurance pays a fixed amount when a measurable trigger occurs (e.g., wind speed exceeds 120 mph, rainfall exceeds a threshold). This model is gaining traction in developing countries where traditional claims infrastructure does not exist - the African Risk Capacity pool and the Caribbean Catastrophe Risk Insurance Facility both use parametric designs. Whether parametric models can scale to replace traditional insurance in developed markets is debated.
Insurance is the system through which societies distribute risk. When it works, losses that would be catastrophic for individuals are absorbed by the collective. When it fails - when premiums become unaffordable, when insurers withdraw, when the protection gap widens - risk falls back on individuals, on governments, and ultimately on the communities that bear the physical consequences. Understanding how insurance works is a prerequisite for understanding climate adaptation, healthcare policy, housing markets, and the distributional choices every society faces.
The paths from here
Dated September 2026, after 18 months of below-trend catastrophe losses and with reinsurance prices falling. Insurance is unusual among the topics on this site in that its future is being set by an arithmetic that is already visible. Losses are rising faster than premiums can sustainably cover in specific regions, and every path below is a different answer to who absorbs the difference.
Orderly repricing and retreat
Insurers keep withdrawing from the highest-risk areas, premiums in those areas rise toward the actuarial cost, and building slows or stops where cover becomes unavailable. Risk gets priced at its full cost and the market does the adaptation that policy has not.
Will it happen? It is already happening in parts of Florida and California. California now lets insurers price with catastrophe models and pass on reinsurance costs if they write more policies in wildfire-prone areas; six large home insurers had signed up by May 2026. The obstacle is political rather than technical: full-cost pricing means telling existing owners their asset is worth less, which no elected government has yet been willing to preside over at scale.
The state becomes the insurer of last resort
Public schemes expand to cover what private markets abandon, as they already have for flood in the United States and terrorism in several countries. The immediate affordability problem is solved and the fiscal exposure moves onto the public balance sheet, usually without being scored as a liability until a bad year arrives.
Will it happen? Partly, and it is the default outcome when the first path becomes politically intolerable. It can also reverse, as Florida's Citizens has since 2023. The pattern to watch is whether the public schemes charge risk-based premiums or subsidized ones. Subsidized premiums keep people in harm's way, which critics see as one cause of the current exposure.
The protection gap widens
Cover simply thins. Households in exposed regions go uninsured or underinsured, and the loss lands on them directly after each event. Recovery slows, property values diverge sharply between insurable and uninsurable areas, and the wealth effects compound across a generation.
Will it happen? This is the base case for most of the developing world, where Swiss Re puts the uninsured share of catastrophe losses at 80 to 90%, and an increasingly plausible one for parts of rich countries. It is the least visible path because nothing announces it.
Adaptation makes the risk insurable again
Building codes, managed retreat from the worst-exposed land, flood defenses, wildfire-resistant construction and better land-use rules reduce expected losses enough that private cover returns at affordable prices. Insurance pricing becomes the signal that drives the adaptation rather than the casualty of its absence.
Will it happen? In places, and there are working examples - post-cyclone building standards in Queensland, Dutch flood engineering, wildfire-hardening requirements in parts of California. The constraint is that adaptation is paid for now by people who benefit later, which is the same political problem that runs through most of this site.
The base case is a mixture: orderly repricing where politics allows it, public backstops where it does not, and a widening gap in the places with the least fiscal room. The variable worth watching is not the premium but the protection gap, because that is the number that records who has been left holding the risk.
What this means for you
If you own or are buying property
Get an insurance quote before you commit, not after. Treat a refusal or a very high premium as information about the asset rather than an administrative obstacle. Insurers price flood, fire and wind exposure with far more data than a buyer has, and their number is the closest thing to an independent assessment of physical risk that most people can obtain for free.
If your premium jumped
Large increases usually reflect costs outside your control rather than anything about you: rebuilding costs, catastrophe losses in your region, and reinsurance, which reprices globally after bad catastrophe years and passes through. Reinsurance prices fell sharply in 2026, so where local risk has not changed, comparing quotes at renewal can pay off.
If you are reading about a disaster
Two numbers get quoted and they are not the same: insured losses and total economic losses. The gap between them is the protection gap, and it is where the human consequences sit. A disaster with low insured losses is often the worse one, because it means the cost fell on people who had no cover.
If you are weighing the fairness argument
Risk-based pricing and affordability pull against each other, and there is no arrangement that delivers both. Charging everyone the true cost of their exposure prices some people out of their homes. Averaging the cost across everyone means low-risk households subsidize building in dangerous places. Every real system picks a point between these, and the honest debate is about where, not about whether.
The mechanics behind this
Insurance is an applied case of several mechanisms this site builds up from first principles. If the reasoning above rests on ideas you want properly grounded, start with these.
