What the Insurance Industry Knows About Climate Risk That Governments Ignore

The Actuarial Architecture of Climate Risk

The global insurance and reinsurance system is a distributed risk assessment apparatus of extraordinary sophistication. At the base are primary insurers — home, auto, commercial property, crop — who write policies directly with individuals and businesses. Above them are reinsurers — Swiss Re, Munich Re, Hannover Re, Everest Re, and about 30 other major players — who insure the insurers, accepting risk in exchange for a portion of premiums. At the apex is the retrocession market, where reinsurers lay off portions of their own risk, and the catastrophe bond market, where large institutional investors take on tail risk in exchange for yield.

Each layer prices risk using proprietary catastrophe models — computational systems that simulate millions of weather scenarios and their impact on insured portfolios. These models were sophisticated by the 1990s, driven partly by the $25 billion loss from Hurricane Andrew in 1992, which bankrupted several insurers and forced the industry to invest heavily in risk modeling. They have been continuously refined since. The models now incorporate climate projections explicitly, updating loss expectations based on observed changes in storm frequency and intensity, flood return periods, and wildfire risk.

The key point for policy analysis is that catastrophe models are actuarially accountable. If a model systematically underestimates losses, the company using it pays claims it did not price for and may become insolvent. The market disciplines inaccurate models over time. This creates a very different institutional incentive than political systems, where underestimating risk is often electorally convenient. The catastrophe modeling community is not perfectly calibrated — significant uncertainty remains — but it is structurally incentivized to be accurate in ways that electoral institutions are not.

The Signal in Premium Data

Insurance premium data is public or semi-public in most jurisdictions — insurers must file rate changes with state regulators in the US, for example — and it constitutes a massive distributed signal about risk perception. When premiums rise rapidly in specific geographic areas, it reflects actuarial judgment that loss probabilities have increased. When coverage is withdrawn entirely, it reflects judgment that no premium level makes the risk insurable within political constraints.

The geographic patterns in premium data track climate risk geography closely. Coastal Florida premiums have increased by 30 to 50 percent annually in some markets since 2020, reflecting hurricane track intensification, sea level rise increasing storm surge, and expanding flood zone designations. California wildland-urban interface premiums have increased by 100 to 500 percent since 2017's catastrophic fire years. Australian flood-plain communities have experienced similar premium trajectories. In all these cases, the market signal preceded political recognition of the underlying risk shift.

The lag between market signal and political response is where damage accumulates. If coastal property insurance had been priced at actuarially sound levels for the past 30 years, the construction of $3 trillion worth of coastal real estate in high-flood-risk areas would have been dramatically curtailed. The artificially low premiums supported by political pricing decisions and public backstop programs subsidized coastal development that the unadulterated market signal would have deterred. The accumulated risk is now on public balance sheets.

The National Flood Insurance Program as Case Study

The US National Flood Insurance Program was created in 1968 when private insurers largely withdrew from flood coverage after a series of catastrophic losses. The program was supposed to be actuarially sound — premiums sufficient to cover expected losses. It has never been. Political decisions have systematically held premiums below actuarially sound levels, particularly for existing homes in high-risk areas, to avoid imposing losses on existing property owners whose values depended partly on artificially cheap insurance.

The program's debt to the US Treasury stood at $36 billion by 2017. After Hurricane Harvey, Maria, and Irma created $16 billion in new losses that year, Congress cancelled $16 billion of the debt — effectively treating the losses as a government grant rather than an insurance obligation. The program then began Risk Rating 2.0, an actuarial reform effort that for the first time priced individual properties based on their specific flood risk rather than using coarser zone-based pricing. The result was premium increases of hundreds of percent for some high-risk properties — increases that correctly reflect their actual risk but that create political conflict because they would force revaluation of coastal property prices that have been partly sustained by subsidized insurance.

The fiscal exposure the NFIP represents is not bounded. As sea levels rise and storm intensities increase, the losses from coastal flooding will increase. The current actuarial reform is moving in the right direction, but the political pressure to maintain below-market premiums — from coastal property owners, real estate interests, and their representatives — is persistent. The gap between actuarially correct pricing and politically acceptable pricing represents a growing contingent liability on the federal government's balance sheet that does not appear in official debt statistics.

What Insurer Withdrawal Means Geographically

Insurance market withdrawal from specific geographies is an unprecedented phenomenon in the modern era. Through most of the 20th century, insurance availability was a function of market competition — insurers competed for customers across most geographies, with differences in price but general availability. The post-2017 period has seen qualitative change: entire geographies becoming effectively uninsurable at market-rate premiums, not because of market failure in the conventional sense but because the underlying risk is genuinely too high.

The geography of withdrawal currently covers: most of coastal Louisiana outside major urban centers; large portions of Florida's coastline and inland flood zones; California's wildland-urban interface communities, particularly in Paradise's former location and similar foothill towns; parts of the Texas Gulf Coast; portions of coastal North Carolina and South Carolina; and, internationally, many communities in coastal Australia, flood-prone areas of central Europe, and wildfire-risk areas of southern Europe.

The social consequences of withdrawal are not evenly distributed. Wealthy property owners can self-insure — absorbing the financial loss if disaster strikes. Middle-class homeowners whose primary asset is their home, and who could not absorb a total loss, face a genuine threat to their financial security when insurance becomes unavailable or unaffordable. Lower-income renters and property owners in high-risk areas are in the worst position: they cannot self-insure, government backstop programs are often insufficient, and insurance market withdrawal removes the institutional mechanism that would otherwise help them recover after disasters.

What Governments Are Actually Doing

The gap between what actuarial data shows and what governments plan for is measurable. The Congressional Budget Office periodically scores the federal government's climate-related fiscal exposure — the contingent liabilities from flood insurance, crop insurance, disaster relief, infrastructure replacement, and military facility adaptation. The numbers are large and growing, and they are largely not incorporated into federal budget projections, which extend only ten years.

State and local governments are somewhat more responsive because they face nearer-term budget constraints and because their voters are more directly affected by local disasters. Miami-Dade County's sea level rise adaptation planning, New York City's post-Sandy resilience investments, and Houston's Harris County Flood Control District's post-Harvey buyout and detention programs are examples of local government translating climate risk data into planning action. The scale of investment, however, remains far below what actuarial models suggest is warranted.

The mortgage industry is beginning to incorporate climate risk in ways that have distributional consequences. Fannie Mae and Freddie Mac, which guarantee the majority of US mortgages, have begun developing climate risk screening that could affect which mortgages they will back. If high-risk properties become ineligible for government-backed financing, property values in those areas could decline sharply — a market correction that would create winners (those without property in affected areas) and losers (those whose primary financial asset is an at-risk property). The political management of this transition will be complex regardless of its ecological correctness.

The Liability Wave and Stranded Asset Risk

Beyond the insurance and government budget channels, climate risk is beginning to appear in corporate liability frameworks. If a fossil fuel company can be shown to have known the climate risks of its products and deceived the public about them, liability claims — similar to tobacco litigation — become possible. Climate attribution science, which can now quantify the fraction of specific weather event damage attributable to human-caused climate change, creates a causal chain that liability lawyers can use.

The stranded asset risk is related but distinct. If climate policy accelerates — carbon pricing, fossil fuel regulations, transition mandates — assets valued on the assumption of continued fossil fuel use become worth significantly less. The Financial Stability Board's Task Force on Climate-related Financial Disclosures (TCFD) was established precisely to bring this risk into financial markets in a systematic way, enabling investors to price transition risk into asset valuations. The transition risk and the physical risk (from climate events) are both now in the range of values that can affect systemic financial stability — which is why central banks have begun developing climate stress tests for financial institutions.

What Honest Planning Looks Like

The insurance industry's implicit planning methodology is: assess the probability distribution of losses, including tail risks; price that distribution accurately; hold adequate reserves; and withdraw from markets where accurate pricing exceeds what markets or regulations will bear. Governments need something analogous: systematic assessment of climate risk across public assets and public obligations, incorporation of that assessment into long-run budget projections, use of the risk signal to drive land use and infrastructure decisions, and political willingness to price public risk at levels that reflect actual probabilities.

This requires making uncomfortable decisions. Telling coastal communities that their homes are uninsurable and that public infrastructure investment will be withdrawn from their areas within a planning horizon. Telling agricultural communities that they need to transition to drought-tolerant crops or face crop insurance withdrawal. Telling cities on flood plains that future development will not be publicly backed. These decisions create losers in the near term and distribute those losses onto constituencies with political power.

The insurance industry makes these decisions through price signals rather than political authority, which is why the signals work but are resisted politically and why government backstops are created to blunt them. A planning framework that uses the insurance signal — that takes actuarial data as the most accurate available assessment of where climate risk has already materialized — and builds policy around it, is working from the best available information. Governments that ignore this data are not making uninformed decisions. They are making politically motivated decisions to defer the recognition of risk that the actuarial system has already priced. The deferral does not reduce the risk. It transfers it to future budgets, future governments, and future populations who will pay for it without having made the decision that created it.