The Insurability of Physical Climate Risk

1. What makes a risk insurable?

The classical criteria of insurability

Before asking whether climate risk is insurable, it is worth recalling what makes a risk insurable in the first place. The canonical reference remains the list of criteria established by Baruch Berliner in his 1982 work on the limits of insurability¹. A risk is more insurable to the extent that losses occur randomly and independently of one another, that their average frequency and severity can be estimated from sufficient historical data, that the maximum possible loss remains contained and financeable by the insurer’s capital, that the required premium remains both economically viable for the insurer and acceptable to the insured, and that adverse selection and moral hazard remain manageable. These technical conditions are supplemented by market and social acceptability conditions, including the existence of coverage limits and the absence of any public policy or legal obstacle to underwriting.

This framework is not a mere academic checklist. It describes a risk transfer mechanism in which every link is necessary. If any single condition fails, insurability does not collapse abruptly; it deteriorates continuously, through premium increases, reduced coverage, the introduction of deductibles or sub-limits, and finally, at the extreme, the outright withdrawal of supply. Physical climate risk has the rare property of simultaneously undermining several of these conditions, which explains why its deterioration is faster and more visible than in most other lines of business.

Physical risk and transition risk

A preliminary distinction is needed. Climate risk decomposes into transition risk, linked to the economic consequences of decarbonization (the depreciation of fossil assets, regulatory change, and emitter liability) and physical risk, linked to material damage caused by natural hazards themselves. This article focuses exclusively on the latter. Physical risk subdivides into acute hazards, meaning sudden and violent events such as cyclones, floods, wildfires or hailstorms, and chronic hazards, meaning slow-onset degradations such as sea-level rise, long-term drought or clay shrink-swell. The former stress the insurer’s capacity to absorb a shock; the latter erode the very stationarity of loss experience, and thereby constitute the deepest theoretical challenge.

2. The end of stationarity

The implicit bet of actuarial science

All actuarial pricing rests on a hypothesis rarely made explicit but ever-present: stationarity. The insurer observes the historical loss distribution, estimates its moments (primarily the mean and variance) and derives a pure premium equal to the expected average loss, augmented by a safety loading proportional to volatility. This approach only makes sense if yesterday’s observed distribution remains representative of the one that will generate tomorrow’s losses. That is precisely the hypothesis climate change undermines.

The most celebrated formulation of this problem comes not from the insurance world but from hydrology. In an article published in Science in 2008, Milly and co-authors argued that stationarity was dead, in the sense that the probability distributions describing streamflows, precipitation and floods could no longer be considered time-invariant². Their argument targeted hydraulic infrastructure planning, but it applies word for word to property and casualty insurance pricing. If the underlying distribution shifts, then historically-based estimation systematically underestimates future loss experience, and yesterday’s premium becomes structurally insufficient tomorrow.

Three distinct aggravating factors

It is important not to attribute the entirety of the rise in insured losses to climate alone. The Swiss Re Institute has long distinguished three drivers of the increase in insured losses, and analytical honesty requires separating them. The first, and by far the historically most important, is economic development itself: the accumulation of insurable assets, urbanization and the concentration of value in exposed areas, particularly coastlines and wildland-urban interfaces. The second is inflation, including the rising costs of reconstruction and repair, which mechanically inflates the nominal amount of each loss. Only the third is the intensification of hazards themselves under the effect of warming.

This decomposition is not an argument for minimizing the risk; on the contrary. Over the three decades from 1994 to 2023, insured losses from natural catastrophes grew at approximately 5.9% per year in real terms, compared to 2.7% per year for global GDP over the same period³. In other words, the insurance burden of catastrophes has more than doubled as a proportion of the economy, and the reinsurer estimates it could double again over the next decade if the underlying trend of 5 to 7% per year is maintained. Even attributing the majority of this dynamic to exposure and inflation, the climate component layers on top of an already strongly rising trajectory, and it is this superposition that makes the situation difficult to sustain.

The rise of secondary perils

One phenomenon has particularly destabilized models in recent years: the shift of losses toward so-called secondary perils. For a long time, the industry concentrated its attention on primary perils, specifically tropical cyclones and earthquakes, rare events with very high individual potential. Yet the bulk of recent insured losses now comes from an accumulation of medium-intensity events, particularly severe convective storms, hail and localized flooding, whose high frequency produces considerable aggregate loss experience. In 2023, the market recorded 142 loss-generating catastrophes, a record, and the category of events between one and five billion dollars proved the most dynamic³. These secondary perils are precisely those that catastrophe models calibrated least precisely, having historically not judged them a priority.

3. The anatomy of a withdrawal: California

The disengagement spiral

California offers the most complete illustration of what happens when the insurability of a peril deteriorates faster than the regulatory environment allows. The homeowners insurance market there experienced a progressive withdrawal by major national insurers from 2023. State Farm, the state’s leading home insurer, stopped underwriting new policies that year, then announced in March 2024 the non-renewal of approximately 72,000 policies in areas deemed most exposed to wildfire⁴. Allstate had already suspended new policy underwriting in 2023. The reason cited by insurers is consistent: the inability to correctly price a wildfire risk that had become too frequent and too severe within a regulatory framework that prevented them from fully incorporating prospective models and the cost of reinsurance into their rate schedules.

The January 2025 wildfires in southern California, particularly the Palisades and Eaton fires, transformed this latent deterioration into an open crisis. These events destroyed more than 18,000 structures and caused at least thirty deaths⁵. State Farm estimated its losses from these fires in Los Angeles County alone at approximately $7.6 billion, and obtained from the regulator an emergency interim rate increase of 17% in June 2025, alongside a request for additional increases⁵. The exceptional geographic concentration of affected policyholders compounded the shock, as the hardest-hit areas counted among those most densely covered by the insurer of last resort.

The insurer of last resort on the front line

The California mechanism known as the FAIR Plan, originally designed as a residual safety net for risks refused by the private market, has become a primary insurer for a growing share of the population. It covered more than 555,000 policies in spring 2025, nearly four times its 2015 volume, and its cumulative exposure had reached approximately $599 billion in March 2025, up 31% in just the previous six months⁴. The January 2025 fires imposed approximately four billion dollars in losses, obliging it to raise a special assessment of approximately one billion dollars from its member insurers to preserve solvency, a portion of which may be passed on to policyholders⁵.

The regulatory response and its ambiguities

The response of California’s authorities illustrates the central dilemma of insurance regulation facing climate change. To bring back insurers, the regulator authorized in 2024 and 2025 the use of prospective catastrophe models and the inclusion of reinsurance costs in rates, in exchange for a commitment to underwrite in distressed zones at a minimum share of their portfolio. But these pricing concessions run head-on into the affordability requirement, and the same public authorities brought actions against insurers for their handling of claims following the fires⁶. This tension (authorizing rate increases to preserve supply while penalizing practices deemed abusive) encapsulates the impossibility of sustainably reconciling the insurer’s technical viability and the insured’s solvency without external pooling.

4. The French model: mutualizing to insure the uninsurable

The architecture of the CatNat scheme

Where the California market long attempted to address climate risk through private mechanisms alone, France has since 1982 made the opposite choice: a public-private hybrid scheme based on national solidarity. The natural disaster compensation scheme, known as the CatNat scheme, rests on a compulsory coverage extension attached to every property and casualty insurance contract, financed by a uniform additional levy, the CatNat surcharge⁷. When a natural disaster is recognized by ministerial decree, the insurer is required to compensate, and may reinsure with the Caisse centrale de réassurance, a company wholly owned by the state and backed by an unlimited public guarantee. The scheme is explicitly designed to cover risks that the private sector alone would deem uninsurable, particularly flooding and geotechnical drought.

The distinctive feature of this arrangement is its uniform rate. The surcharge applies at the same percentage of the property premium for all policyholders, regardless of their actual exposure. A household in a flood-prone area pays the same additional levy as one with no exposure whatsoever. This actuarially non-priced choice is a deliberate solidarity decision, whose primary original objective was to ensure coverage accessibility for lower-income households, often located in the most exposed areas because of lower land prices.

A scheme under climate pressure

This long-stable equilibrium has deteriorated under the combined effect of climate change and pricing inertia. The surcharge rate, originally set at 5.5% then raised to 9% and then to 12% in the late 1990s, remained unchanged for a quarter of a century. Meanwhile, the scheme entered technical deficit from 2016, and the reserves of the Caisse centrale de réassurance fell by approximately 46% between 2015 and 2022⁸. The conclusion drawn by public authorities is unambiguous: maintaining the scheme’s balance requires a substantial increase in resources. By decree of December 2023, the surcharge rate was therefore raised from 12% to 20% on 1 January 2025 for residential and commercial property contracts, the first increase since 1999, from which the Treasury expects approximately 1.5 billion euros in additional annual revenues⁹.

The first effects of this reform were felt in the 2025 financial year. Thanks to the surcharge increase and a year of moderate loss intensity, the Caisse centrale de réassurance returned to a positive technical result and began rebuilding its reserves, after eight years of deficit¹⁰. This recovery remains fragile and primarily cyclical. Drought, through clay shrink-swell which cracks buildings, remained the scheme’s leading loss category in 2025, a year ranked among the hottest ever recorded in France, and the overall cost of climate-related claims for the industry reached 5.2 billion euros, versus 3.9 billion in 2024¹¹. Projections by the Caisse centrale de réassurance also place the scheme’s structural financing shortfall at least 420 million euros per year by 2050 under the effect of climate change, meaning the increase to 20% was necessary but insufficient⁸.

The dilemma of solidarity and fairness

Climate pressure reveals a latent contradiction in the French model. Uniform pricing, originally conceived as a protective instrument for the least well-off, produces a redistributive effect that reverses as risk concentrates. As shown by the work of Charpentier, Barry and James, rising sea levels and growing coastal exposure mean that the primary beneficiaries of a flat-rate scheme now tend to be wealthy owners of high-value coastal properties, whose potential losses are compensated by a levy identical to that paid by an unexposed household¹². Pricing solidarity thus increasingly subsidizes the occupation of hazardous zones.

This creates a tension that cannot be resolved by pricing alone. Introducing an element of risk-based pricing, as some economists suggest, would restore actuarial fairness and send a prevention signal to those settling in exposed zones; but it would make coverage unaffordable precisely where need is greatest, and transfer residual risk to the community through another channel: post-disaster public compensation. This is the core of the problem: physical climate risk can only be made insurable by the market at the cost of excluding the most exposed, and can only be made accessible to the most exposed at the cost of socializing the burden.

5. The protection gap as systemic risk

Measuring a gaping shortfall

The indicator that synthesizes all these dynamics is the protection gap: the difference between total economic losses and losses actually covered by insurance. Globally, natural catastrophes caused economic losses of $318 billion in 2024, of which only 43%, or $137 billion, was covered by insurance, leaving a protection gap of $181 billion for the year alone³. The gap is even wider in Europe, where, according to EIOPA’s dashboard, only approximately one quarter of losses from extreme events was insured over the period 1980-2024¹³. This gap is not merely a market statistic; it measures the economic fragility of societies when faced with a shock, since every uninsured loss is one that households, businesses or the state must absorb directly.

Contagion into credit and asset values

The withdrawal of insurance does not remain confined to the insurance sector; it propagates through the financial system as a whole. EIOPA notes that in high-risk areas, the unavailability of coverage mechanically entails the unavailability of mortgage credit, since insurance is a standard condition for granting a mortgage loan¹⁴. A property that has become uninsurable becomes difficult to finance, therefore difficult to sell, which depresses its value and, in aggregate, that of entire neighbourhoods. The German insurance federation has warned that property and casualty premiums could double within a decade under the effect of climate loss experience, while the demand shortfall worsens the problem: barely 17% of respondents to the 2025 European survey declared holding coverage for natural disaster damage¹⁴. Physical climate risk thus becomes a transmission channel into financial stability, through the double effect of uninsurability on credit and on collateral value.

Reinsurance capacity and its limits

Facing correlated losses that classical pooling cannot diversify, the industry offloads tail risk onto reinsurance and capital markets. Global traditional reinsurance capital is estimated at around $500 billion, supplemented by approximately $50 billion of alternative capital from the catastrophe bond market, which in principle allows the absorption of a scenario of annual losses in the order of $300 billion (equivalent to a modeled century event at a ten-year frequency)³. This capacity is real but not unlimited, and above all it has a cost. The brutal hardening of reinsurance conditions at the 2023 renewals directly fueled the pricing crisis in exposed primary markets, and the sustainability condition set by reinsurers themselves is clear: reinsurance capital must grow at the pace of exposure and remunerate its cost over time, failing which the risk transfer function seizes up.

Toward prevention, adaptation and managed retreat

The conclusion that emerges from this analysis is that the frontier of insurability cannot be stabilized by financial instruments alone. If the loss distribution shifts faster than the premium can follow, the only durable adjustment variable is the hazard itself, meaning the exposure and vulnerability of the built environment. This is why actors are now converging on a logic of prevention-conditioned insurance, in which coverage and its price depend on the vulnerability-reduction measures taken by the insured and the community. The Swiss Re Institute estimates that flood protection infrastructure (levees and engineering works) can be up to ten times more cost-effective than post-loss reconstruction³.

This shift has a consequence that the insurance industry has long been reluctant to state explicitly: the existence of zones where no combination of premium, prevention and risk pooling will make continued habitation sustainable, and where the only rational response is managed retreat (the planned withdrawal of construction from the most dangerous zones). Insurance is not neutral in this process: its withdrawal constitutes de facto the first market signal of future uninhabitability, a signal all the more brutal for having been delayed by non-actuarial pricing. Over the coming decades, the question is therefore not so much whether physical climate risk will remain insurable (it will, partially and at growing cost), but who will bear the uninsurable share: the insured through the premium, the taxpayer through the public guarantee, or the property owner through asset depreciation. It is this allocation, and not actuarial technique alone, that will constitute the real political battleground of insurance over the next two decades.

1. Berliner B., Limits of Insurability of Risks, Prentice Hall, 1982; see also Berliner B., Large Risks and Limits of Insurability, The Geneva Papers on Risk and Insurance, vol. 10, no. 37, 1985, pp. 313-329.

2. Milly P. C. D., Betancourt J., Falkenmark M., Hirsch R. M., Kundzewicz Z. W., Lettenmaier D. P., Stouffer R. J., Stationarity Is Dead: Whither Water Management?, Science, vol. 319, no. 5863, 1 February 2008, pp. 573-574, DOI 10.1126/science.1151915.

3. Swiss Re Institute, sigma 1/2025: Natural catastrophes: insured losses on trend to USD 145 billion in 2025, April 2025; sigma 1/2024 for the decomposition of loss drivers and the record number of 2023 events.

4. California FAIR Plan, exposure and portfolio data; California Assembly Insurance Committee, oversight hearing of 28 May 2025; AM Best and Northeastern University Global Resilience Institute, post-fire analyses, 2025.

5. Insurance Journal, California’s Property Insurance Crisis and the Path to Reform, May 2025; reports on State Farm losses and FAIR Plan special assessments following the January 2025 fires.

6. California Department of Insurance, communications from Commissioner Ricardo Lara, 2024-2026; regulatory actions regarding claims handling after the fires.

7. Code des assurances, provisions on the natural disaster guarantee (Act of 13 July 1982); Caisse centrale de réassurance, presentation of the CatNat scheme.

8. Senate Finance Committee, Report on the natural disaster compensation scheme, May 2024; budget.gouv.fr, Guaranteeing the sustainability of the natural disaster scheme.

9. Decree of 28 December 2023 raising the CatNat surcharge from 12% to 20% on 1 January 2025; revenue estimates of the Treasury Directorate-General.

10. Caisse centrale de réassurance, 2025 results announced 12 March 2026; L’Argus de l’assurance, CCR rebuilds reserves and begins rebalancing the CatNat scheme, March 2026.

11. France Assureurs, press conference of 25 March 2026 on the cost of climate claims in 2025; CCR, estimate of the cost of clay shrink-swell damage.

12. Charpentier A., Barry L., James M. R., Insurance against natural catastrophes: balancing actuarial fairness and social solidarity, The Geneva Papers on Risk and Insurance; see also Charpentier A., Insurability of Climate Risks, The Geneva Papers, 2008.

13. EIOPA, Dashboard on insurance protection gap for natural catastrophes, updated December 2025; historical loss data 1980-2024.

14. EIOPA, Climate insurance protection gaps: a demand-side challenge, March 2026; Eurobarometer 2025; warning from the Gesamtverband der Deutschen Versicherungswirtschaft on premium doubling; joint EIOPA-ECB proposal for shared reinsurance layers.