SCIENCE
Climate Risk and Insurance
Abstract. Climate risk and insurance investigates the changing distribution of weather-related and climate-linked harm, its translation into financial losses and insurance obligations, and the role of private and public risk transfer in economic resilience. This fundamental monograph integrates the physical science of climate extremes, disaster attribution, exposure and vulnerability assessment, actuarial pricing, catastrophe modeling, contractual indemnification, claims and reserving, solvency, reinsurance, insurance-linked capital, adaptation, consumer conduct, protection gaps, sovereign financing and international supervision. It distinguishes climate-attributable hazard change from observed loss trends driven jointly by physical hazards, expanding exposure, structural vulnerability, inflation, and coverage. It evaluates evidence from IPCC assessments, WMO observations, the IAIS 2025 supervisory paper, EIOPA research, the 2026 UNEP adaptation report, insurance-market studies and peer-reviewed economics. Analytical conclusions emphasize the necessity of consistent hazard-to-claim modeling, measurable loss prevention, financially sustainable contracts, transparent public assistance and equitable protection. A documented caution concerning retracted physical-damage research in NGFS Phase V scenarios illustrates the importance of model provenance. The work offers evidence-based explanatory synthesis and illustrative mathematics without asserting original calibrated predictions, newly invented theory or external peer review.
Keywords: Climate risk; insurance; climate change; catastrophe modeling; physical risk; transition risk; liability risk; insurance economics; actuarial science; nonstationarity; extreme events; flood insurance; wildfire; extreme weather attribution; reinsurance; solvency; ORSA; parametric insurance; basis risk; adaptation; loss prevention; climate resilience; insurance protection gap; public–private risk sharing; sovereign disaster risk financing; risk pooling; scenario analysis; climate justice; insurance regulation.
Research Method and Source-Control Note
The analytical cutoff for this archived manuscript is November 6, 2026. Institutional and publisher records linked below were examined for their specific claims; peer-reviewed papers were checked against publisher or bibliographic records and accessible abstracts unless their content was available in other verified summaries. Numerical demonstrations are expressly illustrative. Reported third-party loss figures retain provider provenance and applicable differences in scope. Internal manuscript review is distinct from external peer review and from public release.
Introduction: Scope, Questions, and Research Design
Climate risk and insurance is the scientific study of how changes in the climate system alter insurable hazards, losses, contractual obligations, risk pooling, capital requirements, financial protection, and incentives for adaptation. Its object extends from the physical generation of hazardous events to the distribution of their economic consequences among households, firms, insurers, reinsurers, governments, investors, and communities. The discipline meets climate science at the boundary between a meteorological event and its distribution; it meets actuarial science at the boundary between a loss distribution and a premium; and it meets public economics at the boundary between private coverage and collective resilience.
This monograph examines climate risk as a changing, spatially correlated, institutionally mediated distribution of possible consequences. It integrates insurance economics, actuarial mathematics, atmospheric and Earth-system science, statistics, regulation, financial accounting, behavioral research, disaster studies, and adaptation policy. The governing questions are how climate-related losses can be estimated with defensible uncertainty, which elements are transferable through insurance contracts, how risk reduction and insurance affect one another, how insurance systems remain financially viable as conditions change, and which institutions can protect people whose risks exceed commercial affordability.
The research method is critical synthesis. Physical scientific assessments are taken from the IPCC and WMO; supervisory definitions and expectations from the IAIS and EIOPA; financial reporting provisions from the IFRS Foundation; loss observations from explicitly attributed industry sources; and empirical insurance propositions from peer-reviewed studies. The work provides formal explanations and illustrative calculations, each distinguished from an empirical calibration. It reports differences among datasets rather than averaging incompatible estimates. The reference base was checked against the cited institutional pages, publisher records, and accessible descriptions as of November 6, 2026. Full independent replication of proprietary catastrophe models or inaccessible underlying datasets is outside the present evidence.
The central analytical finding is that insurance capacity depends jointly on physical peril, exposed value, vulnerability, contract design, capital, and social institutions. A higher insured-loss total by itself identifies none of these factors. This finding is an inference drawn from established hazard–exposure–vulnerability analysis, actuarial loss decomposition, and market economics, rather than a claim to a new proprietary theory. Every subsequent chapter separates observed evidence, mathematical implications, and policy evaluation. (see IPCC climate assessment; IAIS supervisory guidance)
Historical Development of Climate Risk and Insurance
The relevant history begins long before the contemporary climate debate. Marine, fire, agricultural, and mutual insurance created contractual techniques for pooling uncertain losses; modern probability theory made expected loss, variance, and ruin calculable; reinsurance and catastrophe models subsequently expanded geographic and financial capacity for low-frequency severe events. Flood-control projects, building standards, public relief, and agricultural support developed alongside private risk transfer. This parallel history matters because climate insurance inherits institutional constraints established by land use, property law, state disaster policy, and insurance regulation.
Late twentieth-century climate assessment introduced systematic evidence that the statistical properties of some weather-related perils would change as greenhouse gases accumulated. By 2005 Evan Mills had explicitly assessed insurance in a changing climate, emphasizing the multiple exposures of insurance businesses and opportunities for loss prevention. Later insurance scholarship studied disaster recovery, incentives for mitigation, and the changing composition of insured risks. Climate risk consequently moved from a possible adjustment in catastrophe assumptions to a subject within underwriting, investments, reserving, solvency, corporate governance, and consumer protection. (see Mills (2005); Kousky (2019))
The Paris Agreement, adopted in 2015 and entering into force in 2016, established an international mitigation and adaptation context. The IPCC Sixth Assessment cycle supplied extensive assessments of changing extremes, impacts, adaptation options, and residual losses. Its 2023 Synthesis Report consolidates the physical, impact, adaptation, and mitigation assessments across the cycle. (see IPCC climate assessment) In parallel, insurance supervisors developed climate-related expectations under existing prudential frameworks. EIOPA issued scenario-analysis guidance for insurers' own risk and solvency assessments in 2022; the IAIS replaced its 2021 application paper with a comprehensive supervisory paper in April 2025. These measures did not create a single worldwide insurance law. They document the incorporation of climate risk into existing governance and supervisory systems. (see United Nations Framework (2015); IPCC climate assessment)
A further historical development is the use of insurance-market data to illuminate wider social vulnerability. EIOPA's catastrophe protection-gap dashboard combines peril estimates, losses, and coverage, while the U.S. Federal Insurance Office analyzed a geographically detailed homeowners dataset covering 2018–2022. By 2026, financial protection and adaptation had become a shared question for national authorities, insurers, development institutions, and climate policy. The history thus links commercial risk selection to the public determination of where rebuilding, adaptation, and relocation remain feasible. (see EIOPA guidance; U (2025))
Foundational Concepts and the Structure of Climate Loss
A climate hazard is a potentially damaging climatic event, trend, or physical condition. Exposure denotes people, assets, productive activities, or ecological systems subject to that hazard. Vulnerability denotes the degree to which exposed entities suffer harm. Financial loss expresses selected consequences in monetary terms. An insured loss is the portion of such consequences paid or payable under applicable insurance contracts, after policy limits, deductibles, exclusions, conditions, and claims adjustment. Climate risk is defined here as the prospective distribution of climate-related adverse consequences, conditioned on these linked factors and the state of knowledge.
These definitions support a basic separation. Let H denote hazard intensity, E exposure, V vulnerability, and C contractual coverage. The modeled insured loss can be expressed as L = g(H,E,V,C). The notation is conceptual: its functional form must be estimated for each peril, geography, building class, and coverage type. Climate affects H, but H does not independently fix E, V, C, or claim-settlement inflation. A major flood in an uninhabited region can produce enormous physical intensity and negligible insured loss; a less extreme storm can cause large claims where property values, rebuilding costs, and coverage penetration are high.
Physical risk encompasses acute phenomena such as floods, storms, wildfires, and heatwaves, as well as chronic processes such as sea-level rise and persistent heat. Transition risk arises from shifts in policy, technology, law, demand, and asset valuations during movement toward a lower-emissions economy. Liability risk concerns possible legal responsibility associated with climate harms or disclosures; its insurability and contractual treatment depend on jurisdiction, legal cause of action, causation, policy wording, exclusions, and claims-made or occurrence structures. Climate-related financial risks are interdependent with wider economic and political conditions. (see IPCC climate assessment; IAIS supervisory guidance)
Risk, uncertainty, and ambiguity also require separate treatment. Sampling uncertainty concerns estimated parameters from finite observations. Model uncertainty concerns structural choices such as damage functions and event generators. Scenario uncertainty concerns alternative emissions, adaptation, and socioeconomic pathways. Deep uncertainty denotes cases without a defensible single distribution. The choice between a probabilistic forecast and a scenario narrative affects underwriting and capital conclusions; presenting one as the other creates false precision.
Insurance protection must be evaluated at three levels: nominal coverage, effective indemnification after contractual conditions, and the ability to recover economically after payment. A policy may provide formal coverage while leaving a large deductible or uncovered peril. This layered understanding will be used throughout the monograph. (see EIOPA guidance)
Climate Physics, Observations, and Changing Hazards
The physical basis is established most clearly where attribution evidence is strong. The IPCC's 2021 Working Group I assessment reports that human-induced climate change already affects weather and climate extremes across regions. It finds that hot extremes have become more frequent and intense across most land regions since the 1950s, with strong attribution to human influence. Heavy precipitation frequency and intensity have increased since the 1950s over most land regions with adequate data, and human influence is likely the main driver. Its assessment is more qualified for worldwide trends in the frequency of all-category tropical cyclones; it nevertheless identifies a human contribution to heavy precipitation associated with tropical cyclones. These are peril-specific scientific findings, not a single statement about every hazard everywhere. (see IPCC climate assessment)
The WMO's State of the Global Climate 2025, published March 23, 2026, identifies 2015–2025 as the eleven warmest years in the observational record. It estimates 2025 global temperature at approximately 1.43°C above the 1850–1900 baseline, with observational uncertainty. This global mean figure describes climate state, not a deterministic multiplier applicable to individual property losses. Local elevation, hydrology, vegetation, drainage, coastal geography, urban design, and structural characteristics determine how a change in physical forcing affects claims. (see World Meteorological Organization (2026))
Flood risk includes river, surface-water, coastal, and compound flooding, each requiring a distinct hydraulic or statistical description. Wildfire damage reflects ignition, fire weather, fuel management, building material, wildland–urban interface patterns, suppression capacity, and evacuation. Drought risk affects crop yields, soil, water, energy, and supply chains, but insurance attachment may be indirect. Heat affects morbidity, labor productivity, power demand, and potentially liability and health insurance. Tropical-cyclone losses depend on wind, surge, rainfall, track, coastal exposure, and building-code performance. Rainfall-driven loss often depends on drainage and topography even when precipitation increases.
Perils also interact. Heat and drought may worsen fire weather; rainfall on burned watersheds may increase sediment and flood consequences; coastal surge and river discharge may compound; simultaneous regional events may reduce diversification. A useful scientific conclusion follows: the insurer's relevant climate signal is a peril-specific, spatially resolved change to an event and vulnerability distribution, and any conversion of that signal into financial claims requires additional evidence. (see IPCC climate assessment)
Event Attribution, Loss Attribution, and Causal Inference
Event attribution asks how anthropogenic climate change affects the probability or intensity of a defined physical event under a specified counterfactual climate. Insurance loss attribution asks how much of an observed or modeled financial loss arises from changed climate hazards rather than from exposure, vulnerability, coverage, inflation, and other influences. These are different estimands. An event can become much more probable in a warmer climate while the observed change in claims is dominated by property accumulation. Conversely, flat inflation-adjusted losses may conceal a worsening physical hazard offset by better building standards.
Suppose an event has factual probability p1 and counterfactual probability p0 under a model. A risk ratio RR = p1 / p0 describes modeled change in event probability for the precise event definition and counterfactual. A commonly used fraction of attributable risk, 1 − p0/p1, is then an event-probability construct. It does not automatically determine the fraction of an insurer's paid claims caused by climate change. Its use requires attention to ensemble methods, event threshold, observational constraints, and uncertainty intervals. For nonstandard event definitions or p0 near zero, estimates may be especially unstable.
A financial decomposition can compare normalized losses after controlling for inflation, population or asset growth, insured penetration, claims practices, and vulnerability. Yet any normalization is partly model-based, and the omitted interactions may be substantial. Market-wide insured totals may rise because previously uninsured assets enter the formal market. An insurer's local premium growth may reflect improved risk differentiation or the repricing of previously underpriced exposures rather than a contemporaneous increase in physical hazard. (see Swiss Re Institute (2026); Munich Re (2026))
The proper causal question identifies its unit, population, intervention or exposure, counterfactual, and time horizon. A catastrophe database is usually observational evidence rather than a controlled attribution experiment. Hazard attribution from climate models, engineering studies of damage vulnerability, and econometric studies of insurance decisions should be triangulated, with uncertainty preserved. A robust report states the event definition, hazard counterfactual, exposure adjustment, contract scope, and sensitivity range before making a causal claim. The distinction is essential when insurance evidence informs adaptation policy or public debate. (see IPCC climate assessment)
Nonstationarity, Probability, and Deep Uncertainty
Classical actuarial estimation often assumes that the distribution of sampled claims is sufficiently stable to inform future expectations after known changes are accounted for. Climate change challenges this assumption for some hazards and locations. Nonstationarity means that parameters or distributional structure depend on time and covariates, including temperature, sea level, land cover, or exposure. It does not make probability theory unusable; it makes the conditioning structure and estimation horizon more consequential.
For one exposure, a stationary annual loss model might use E[L] = integral of loss l with respect to F(l). Under changing conditions, the appropriate object is E[L_t | X_t, S], where X_t includes exposure, protective features and climatic covariates, and S describes an assumed development and emissions pathway. A premium for the coming insurance year is a decision over the near-term conditional distribution; a twenty-year investment or flood-reinsurance transition requires attention to longer horizons and structural change. Both can be consistent if their different horizons are stated.
Sampling variability can be severe for rare events. A one-in-one-hundred annual event is a shorthand for a probability under a model and reference period, not a schedule of recurrence. Dependence between events, event clustering, changing landscapes, and limited historical measurements complicate estimation. Extreme-value techniques such as generalized extreme-value and generalized Pareto families can be useful for tail estimation when statistical assumptions are defensible. Their numerical outputs require sensitivity analysis for threshold choice, temporal dependence, spatial pooling, and tail shape. An apparently precise return period may conceal substantial parameter uncertainty.
Deep uncertainty is particularly important where historical analogues do not represent future combinations of hazard and exposure. Decision makers can use stress scenarios, robust design, sensitivity bounds, and staged investment rather than force a single probability onto an unknowable pathway. Scenario analysis is valuable for testing resilience under specified conditions; it does not produce a forecast merely because a computational model prints an exact number. The 2025 NGFS clarification regarding retracted damage-function research is a concrete reminder that scenario assumptions require version control and scientific reassessment. (see Network for Greening (2025); National Oceanic and (2025))
The policy implication is temporal consistency: short-duration insurance contracts can be repriced, while buildings, infrastructure, mortgage portfolios, and settlement patterns last decades. Climate insurance analysis must jointly acknowledge those different clocks.
Actuarial Foundations: Frequency, Severity, Dependence, and Tail Risk
Insured catastrophe loss is often represented by an annual aggregate random variable S = sum_{i=1}^{N} X_i, where N counts covered events or claims and X_i is an individual insured severity. In a simple compound model with independent identically distributed severities, E[S] = E[N]E[X]. This equality is useful for exposition, but portfolio applications must model dependencies between counts, claim severities, risk characteristics, and coverage limits. Climate-driven changes may operate through event frequency, intensity, spatial footprint, claims severity, and relationships between perils.
Expected annual loss (AAL) is the mean of a modeled annual loss distribution. Probable maximum loss (PML) is a model-specific term commonly linked to selected exceedance probabilities or return periods; the definition must be disclosed. Value at Risk at confidence q is the q-quantile of loss. Tail Value at Risk or expected shortfall describes the mean loss in the specified extreme region under the relevant mathematical conditions. AAL informs prospective underwriting cost, while tail measures inform capital, reinsurance, and stress management. The measures cannot be substituted for one another.
For illustration only, suppose annual catastrophe losses in arbitrary currency units take values 0 with probability 0.90, 10 with probability 0.09, and 100 with probability 0.01. AAL equals 1.9. If a portfolio were priced at 1.9 without expenses, uncertainty load, capital cost, or contractual adjustments, its expected indemnity might be covered, but its tail capital needs would remain substantial. This example is an identity-checking teaching calculation, not an estimate of actual climate or insurance-market probabilities.
Diversification is limited by correlated physical hazards and economic connections. Distant risks may share climate modes, dependence in reinsurance capacity, or simultaneous post-disaster construction-cost inflation. Insurers also face model uncertainty from incomplete exposure data and the possibility that hazards outside historical training data occupy the tail. Correlation structures inferred from short periods may be fragile. Actuarial decisions therefore require plausible dependence stress tests, granularity by peril and geography, and explicit limits to model reliability.
An appropriate climate-sensitive actuarial workflow maps historical claims to policy terms and exposure; adjusts for changing values, coverage and vulnerability; estimates hazard relationships; tests models out of time and geography; and reports ranges rather than a single apparently objective probability. (see IAIS supervisory guidance; IPCC climate assessment)
Catastrophe Models and Spatial Accumulation
A catastrophe model generally joins four functional components: an event or hazard module, a geocoded exposure inventory, vulnerability or damage functions, and a financial module that applies contracts and insurance structures. Hazard events may be simulated from historical records, stochastic generators, physical simulations, or hybrid approaches. The exposure inventory requires reliable locations, occupancy, use, insured values, building characteristics and defenses. Damage functions relate hazard intensity to physical damage, with uncertainty. Financial terms convert physical damage into losses after deductibles, coverage conditions, policy limits, coinsurance, demand surge, and reinsurance participation. (see Network for Greening (2025))
Spatial accumulation is central. A single event can damage thousands of individually small policies; a long-duration event can interrupt infrastructure, suppliers, transport, and electricity far beyond the hazard footprint. A model that prices one building correctly can underestimate portfolio loss when it omits regional correlations, contingent business interruption, or reconstruction inflation. The insurer must therefore validate both property-level vulnerability and aggregate loss relationships.
Climate sensitivity can enter the hazard component through changing event intensity, frequency, or spatial distribution. Exposure and vulnerability must be modeled separately or the effect of adaptation will be obscured. Simple methods that uniformly scale historical paid claims by a temperature coefficient can fail where hazard-response functions are nonlinear and spatially heterogeneous. Engineering evidence, satellite observations, land-surface models, and damage claims can help, but their scales and definitions may differ.
Model validation combines hindcasts, out-of-sample evaluation, event-by-event forensic comparison, sensitivity testing, and independent challenge. Historical fit alone is insufficient when the target distribution is changing. Open disclosure of assumptions, input provenance, version, model purpose, and known blind spots improves scientific accountability even where proprietary code cannot be disclosed. A model should be evaluated against its intended use: annual underwriting, capital adequacy, adaptation investment, or public policy may demand different validation standards.
The IAIS has described how supervisors should engage with insurer climate models and scenario practices; its 2026 member-only material on catastrophe models is not treated here as an independently examined public primary source. The published 2025 supervisory paper supplies the verified public framework. (see IAIS supervisory guidance)
Insurance Pricing, Affordability, and Market Selection
A technically adequate prospective premium must cover expected insured claims, claims handling, expenses, applicable taxes, cost of risk capital, reinsurance and a sustainable margin, subject to market and legal constraints. Risk-based prices can inform choices about location and mitigation, but they also expose the limits of affordability for households with low income, high risk, or few relocation options. Purely redistributive premium suppression without accompanying financing can hide risk, accumulate fiscal liabilities, or weaken insurer solvency. An unqualified rule demanding immediate fully risk-reflective premiums can instead generate severe coverage withdrawal. Both mechanisms require institutional analysis.
Demand responds to price, perceived probability, wealth, access to credit, legal requirements, trust, and contract clarity. Supply responds to expected claims, capital, reinsurance pricing, competition, model confidence, investment opportunities, and permitted premium adjustments. Thus an increase in rates can reflect risk adjustment or regulatory catch-up as well as changing hazard. Cancellation, nonrenewal, reduced limits, and rising deductibles are distinct indicators of availability and quality. (see U (2025); EIOPA guidance)
The U.S. Federal Insurance Office's January 2025 study analyzed over 246 million homeowners-policy observations from 2018–2022, aggregated at ZIP Code scale. It found that premiums per policy grew 8.7% faster than inflation on average over the sample period, and that the highest-quintile climate-peril risk areas paid substantially more than the lowest quintile. The study is direct evidence of an association between its specified peril-risk index and observed insurance outcomes. It is not a controlled experiment isolating the causal effect of anthropogenic warming; general homeowners policies also typically exclude flood coverage. (see U (2025))
Under competitive pressure, insufficient pricing may temporarily sustain apparent affordability while increasing the risk of insurer failure or later abrupt repricing. Delayed mitigation can also deepen long-run loss costs. A sustainable affordability policy should therefore distinguish an insurer's marginal cost of bearing risk from the distributional decision about who pays that cost. Public grants, targeted subsidies, risk-pooling mandates, mitigation support, or income-based assistance each allocate cost differently. Their evaluation must report financing, incidence, incentives, access and solvency consequences, not only the headline premium.
Contract Architecture, Coverage, Exclusions, and Legal Risk
Insurance protects a defined interest against a defined event or liability according to the contract and applicable law. Climate hazards become insured losses only through these legal and operational arrangements. Flood, wind, storm surge, wildfire, business interruption, crop yield, and contingent business interruption may be covered differently across products and jurisdictions. The insured's reasonable understanding of a policy may diverge sharply from actual wording, sublimits, waiting periods, exclusions, aggregation rules, deductibles, and proof-of-loss duties. The protection gap is partly a contract interpretation problem.
Property and casualty policies often have annual terms and can adapt relatively quickly through underwriting renewal, while liability exposures can develop over longer periods. Claims relating to climate-linked legal duties, professional conduct, emissions disclosures, or environmental harm involve specific triggers, causation standards, defense costs, exclusions and jurisdiction. A physical event's connection to global warming is insufficient on its own to establish covered legal liability. Professional coverage analysis must avoid presuming either universal insurability or universal exclusion. (see IAIS supervisory guidance)
Climate adaptation projects can create new contract questions. Insurers may consider resilient rebuilding upgrades, code compliance, green repair, and the cost of replacing damaged assets with higher-standard structures. Each affects insured value, claims settlement and coverage scope. Insurance that funds only restoration of pre-disaster vulnerability may reproduce the same damage pathway. Yet replacement at higher standards introduces an additional benefit whose payer and eligibility conditions must be explicit.
Contractual clarity has become an important supervisory issue. EIOPA's 2026 discussion of natural-catastrophe protection gaps reports that consumers may mistakenly believe they have adequate coverage when important perils or conditions remain outside the policy. It identifies comprehension, prior claims experiences, perceived affordability, income, and expectation of state assistance among demand-side factors. Legal design and consumer understanding therefore affect effective insurance protection as directly as nominal take-up. (see EIOPA guidance)
The policy standard is to map the whole chain from hazard to claim: insured subject, peril definition, attachment, trigger, payout, limits, dispute mechanism, and post-loss financing. A policy with an inexpensive nominal premium may be poor financial protection if its central expected catastrophe peril is excluded.
Claims, Reserving, Reconstruction Inflation, and Insurance Accounting
Climate-linked catastrophe claims have several temporal components. The insurer initially estimates a reported or incurred-but-not-reported liability; later development reflects new claims, repair estimates, litigation, business interruption duration, salvage, subrogation and settlement inflation. Severe events can produce demand surge: a simultaneous shortage of contractors, materials, accommodation, and infrastructure increases rebuilding costs above ordinary price inflation. Inflation adjustments based only on general consumer prices may understate rebuilding costs in affected markets.
Reserving questions extend beyond the immediate storm or fire. Longer-duration insurance liabilities can be affected by evolving court decisions, ecological impairment, health conditions, and economic transitions. Any claim of future climate liability must respect actual contract wording and the state of legal evidence. Different lines require different development methods, tail assumptions, and scenario horizons. An asset's physical climate damage can also affect collateral and the insurer's investment valuations before a related insurance claim exists.
IFRS 17 establishes principles for recognizing, measuring, presenting, and disclosing insurance contracts and reinsurance held, including the use of current estimates of future cash flows and the presentation of insurance service results. It became effective for annual reporting periods beginning on or after January 1, 2023, subject to its scope and applicable adoption. Climate analysis supplies relevant inputs to the measurement of contractual cash flows where material; IFRS 17 does not itself prescribe a universal scientific climate-damage model. (see IFRS 17)
An illustrative reserve exercise separates a catastrophe's reported losses, late-reported exposure, reconstruction cost trend, coverage disputes, and expected recoveries. A stress that increases claims by 20% should identify whether it changes hazard intensity, damage ratio, rebuilding costs, or legal settlement, because those changes have different implications for subsequent periods. Reinsurance recovery must also be tested for counterparty performance and attachment definitions.
The scientific requirement is to preserve an audit trail between raw claim evidence, actuarial assumptions, climate hypotheses, financial recognition and management judgment. This makes it possible to correct a hazard assumption without silently rewriting contract or accounting history.
Risk Pooling, Reinsurance, and Capital Markets
Insurance operates by pooling many uncertain outcomes and financing claims from premiums, capital, and other resources. Pooling is most effective when losses are not perfectly correlated; catastrophe risk limits the benefits because one storm, flood, or fire can damage a large share of a portfolio. Reinsurance allocates defined layers of insurance loss to another balance sheet, helping primary insurers manage accumulation, earnings volatility and capital. Reinsurance does not eliminate physical damage or societal loss. It reallocates contractual claims and introduces counterparty, pricing, basis and renewal risks.
Common structures include proportional quota-share arrangements, per-risk excess of loss, per-event catastrophe excess of loss, and aggregate stop-loss protection. The economic effects differ by attachment, exhaustion, reinstatements, hours clauses, and aggregation. A primary insurer with an adequate annual aggregate limit may still face unacceptable net retention after multiple correlated events or a major reinsurance price increase.
Catastrophe bonds and other insurance-linked securities transfer specified modeled or contractual catastrophe risks to capital-market investors. Triggers may be indemnity-based, industry-loss, modeled-loss or parametric. Faster and more objective triggers may reduce claims-adjustment disputes and settlement delays but can increase basis risk relative to actual insured loss. Investors require compensation for expected catastrophe losses, risk uncertainty, liquidity, and correlation with their other assets. After a major event, trapped collateral and changing risk perceptions can affect renewal capacity.
Climate change may alter both the expected cost of layers and their tail distribution, making historic return-period conventions insufficient without time-horizon and pathway statements. Reinsurance renewal can give comparatively short-duration contracts new information faster than some long-lived assets adjust. That speed is economically valuable, but also capable of transmitting sudden repricing or capacity shortages to consumers.
Portfolio decisions should compare gross and net exceedance curves, stress the dependence of different perils and geographies, identify uninsured social loss, and test the financial resilience of contractual counterparties. Optimizing net insurance loss alone cannot show that the community is resilient. This conclusion follows from the division between economic, insured, and retained losses described in the IPCC and catastrophe insurance literature. (see IPCC climate assessment; Swiss Re Institute (2026))
Solvency, Own Risk Assessment, and Supervisory Governance
Climate-related risks can affect insurers' solvency through underwriting, reserving, market values, credit exposures, operational continuity, and reputational or legal conditions. Supervisors therefore assess climate risk across the entire insurer rather than as an isolated catastrophe line. The IAIS April 2025 Application Paper discusses supervisors' integration of climate risk into governance, internal control, valuation, investment practices, disclosure, group supervision, macroprudential monitoring, and scenario analysis, using existing Insurance Core Principles. An IAIS Application Paper provides supporting guidance rather than establishing additional binding requirements independently of applicable national law. (see IAIS supervisory guidance)
European practice provides a documented example of forward-looking governance. EIOPA's 2022 application guidance explains climate-change materiality assessments and climate scenarios in the Own Risk and Solvency Assessment (ORSA). Its earlier 2021 opinion described proportionate assessment of material climate exposure and exploration of relevant long-term pathways, including scenarios below and above 2°C where appropriate. These tools inform insurer strategy and supervisory review; an ORSA scenario is not a contractual guarantee or a prediction. (see EIOPA guidance)
A scientifically credible ORSA identifies material risk pathways; maps each to exposures, lines, assets and cash flows; selects scenarios and horizons; evaluates sensitivity to uncertainty; documents management actions; and explains limitations. Relevant response actions include changing underwriting concentrations, increasing reinsurance or capital, supporting exposure reduction, diversifying investments, or revising products. The modeled effectiveness and legal feasibility of management actions deserve scrutiny: simultaneous action by many insurers may raise reinsurance prices or reduce market capacity.
Capital adequacy assessment also needs reverse stress testing. Instead of asking only how much capital is required under a conventional event, a reverse test asks which combinations of catastrophe losses, asset impairment, reinsurance default, liquidity needs, and market reactions would compromise the insurer's viability. The exercise must distinguish gross catastrophe damage, net insured loss, liquidity outflow, and solvency capital.
Supervisory comparability remains difficult because hazard exposure, policy wording, accounting, model access, national catastrophe schemes and emissions pathways differ. Disclosing assumptions, data lineage, and limitations is indispensable for meaningful comparisons. Governance quality is evidenced by decisions demonstrably linked to assessments, rather than by the mere existence of a climate-risk report.
Insurance Demand, Risk Perception, and Consumer Conduct
Climate risk is perceived through experience, information, social networks, trust, financing constraints, and expectations of public relief. People may overestimate memorable extreme events yet underestimate cumulative exposure over decades; both are possible, and actual responses differ by population. Insurance purchase decisions therefore cannot be explained solely by actuarial expected loss. Premium affordability, understanding of exclusions, administrative friction, mortgage obligations, and confidence in claim settlement all affect uptake.
A standard expected-utility account predicts benefits from reducing consumption volatility for a risk-averse household, but real choices involve limited liquidity and imperfect information. A premium that would improve expected utility over a long horizon may be unaffordable at a particular moment. Conversely, an affordable policy may go unpurchased if the consumer expects the state to compensate disaster losses or cannot interpret the product. These mechanisms can coexist. The welfare evaluation must consider household debt, essential consumption, exposure to displacement, and financial recovery.
The 2026 EIOPA examination of natural-catastrophe protection gaps reports barriers linked to affordability perceptions, clarity of terms, earlier negative claims experiences, and expectations of government aid. It also describes mistaken beliefs about policy coverage, sometimes called an insurance illusion. This evidence underscores that a protection gap is partly a communication and conduct problem, not simply a missing premium subsidy. (see EIOPA guidance)
Consumer policy can improve standardized peril disclosures, clearer deductibles and exclusions, access to independent comparison, prompt claim payments, and credible dispute mechanisms. Automatic inclusion or bundled catastrophe protection can raise take-up, but the distribution of cost and incentives must be made explicit. Transparent subsidy design can support low-income households while retaining a meaningful signal for loss prevention. Product complexity should be justified by genuine risk differentiation rather than by the creation of avoidable consumer confusion.
The empirical literature on disaster insurance and risk reduction is mixed. Kousky's 2019 review finds evidence supporting recovery benefits while observing that rigorous studies of pre-event risk reduction are relatively limited and effects may be modest. Policy should not assume that the purchase of insurance automatically produces physical adaptation. (see Kousky (2019))
Protection Gaps and Their Measurement
An insurance protection gap measures the difference between financial losses that are economically relevant and losses compensated by insurance under a specified definition. A common event-level indicator is G = (economic loss − insured loss) / economic loss for positive economic loss. Its interpretation depends critically on consistent coverage of geographies, perils, dates, price bases, and definitions. The fraction is not itself a causal estimate of climate change and does not measure every form of social vulnerability.
Protection gaps may arise because risks are uninsurable at feasible premiums, households lack income, products exclude particular perils, distribution channels are weak, legal uncertainty is high, coverage information is unclear, or governments have historically provided ex post assistance. Public infrastructure, ecosystems, informal assets, uninsured business interruptions, and human well-being may be underrepresented in monetary loss statistics. An insurance gap is therefore one dimension of a broader resilience gap.
EIOPA's protection-gap dashboard, last updated December 5, 2025 at the time of this research, compiles data on economic and insured losses, risk estimates, and coverage from 30 European countries. Its historical perspective covers 1980–2024 and its modeled current view distinguishes peril and country. Differences between its historical, modeled and coverage views must not be collapsed into a single universal European number. (see EIOPA guidance)
Industry global loss figures illustrate the importance of measurement. Swiss Re Institute's March 2026 sigma study estimated 2025 natural catastrophe insured losses at USD 107 billion and associated economic natural-catastrophe losses at USD 220 billion. Munich Re's January 2026 estimate gave approximately USD 108 billion of insured and USD 224 billion of total natural-disaster losses. These figures come from distinct proprietary databases and definitions. Their proximity is informative; it does not establish interchangeable event sets, identical inflation adjustments, or causal attribution of loss to climate change. (see Swiss Re Institute (2026); Munich Re (2026))
A scientifically responsible dashboard publishes the numerator, denominator, peril inclusion, source, revision date, geographic scope, and distinction between modeled and observed values. The normative objective is to reduce damaging uninsured vulnerability while maintaining financially sound compensation systems, not merely to maximize premium volume.
Climate Adaptation, Engineering, and Loss Prevention
Adaptation changes the consequences of climate hazards by altering exposure, vulnerability, and the ability to cope. It includes land-use planning, relocation from persistently hazardous sites, flood defenses, urban drainage, wildfire-resistant design, vegetation and fuel management, cooling and public-health systems, early warning, resilient infrastructure, and building-code improvements. Insurance can finance residual losses after adaptation and may help evaluate the expected cost of different interventions. It cannot substitute for physical measures when hazards make normal habitation or infrastructure functions unsustainable. (see IPCC climate assessment)
Risk reduction can be valued through avoided expected losses, reduced tail risk, preservation of services, and lower social disruption. Suppose an illustrative investment costs 100 and reduces annual expected damage by 8 for twenty years. Ignoring discounting, maintenance and uncertainty gives gross avoided damage of 160. A correct cost–benefit assessment would discount cash flows, model lifetime changes in hazard and exposure, and include environmental, distributional, operational, and residual risks. The simple arithmetic only demonstrates why project horizons and discount rates matter.
Insurance pricing may reward verified mitigation through lower expected-claim costs and improved underwriting eligibility. To avoid a subsidy appearing as a technical risk reduction, insurers should distinguish engineering evidence from policy incentives. Empirical results on complementarity require qualification. Botzen, Kunreuther, and Michel-Kerjan's post-Hurricane Sandy homeowner study found that flood insurance could complement earlier investments in loss reduction while substituting for some immediate emergency-preparedness measures. That context-specific finding does not imply the same behavior for all perils or populations. (see Botzen (2019))
Adaptation policy should track both physical and financial outcomes. A household that obtains affordable insurance but remains repeatedly flooded has improved compensation yet persistent physical harm. A household that improves resilience but cannot finance reconstruction after a residual event remains financially vulnerable. Sustainable protection joins prevention, accessible risk financing, safe rebuilding, and credible public planning.
The IPCC assesses insurance as a potentially useful risk-management option while warning that design and institutional conditions determine outcomes. Maladaptation is possible when coverage encourages hazardous development or shifts costs onto highly vulnerable groups. The appropriate question is whether each intervention reduces total expected harm over its full lifetime and preserves acceptable risk for the people affected. (see IPCC climate assessment)
Parametric Insurance, Index Products, and Basis Risk
Parametric insurance pays when a predefined measurable index or trigger crosses a contractually specified threshold. The trigger may be modeled wind speed, excess rainfall, drought conditions, earthquake intensity, or a composite event index. Its main operational attraction is speed and transparency when the trigger is measured independently. It avoids some of the time required for individual loss adjustment, but a quick payout is only useful to the extent that it aligns with the insured's real financing need.
Basis risk is the mismatch between an index payout and the insured's actual loss. A household can suffer major crop losses while the regional rainfall index remains above the payout threshold, or receive a payout despite smaller realized damage. Basis risk has both statistical and distributive consequences: it can undermine trust, reduce welfare, and leave precisely the most vulnerable households exposed. Product design should evaluate the joint distribution of index and individual loss, including tail mismatch, geographic heterogeneity and change over time.
Chantarat, Mude, Barrett, and Carter's 2013 study described index-based livestock insurance for pastoralists in northern Kenya. The contract used remotely sensed vegetation and modeled area-average livestock mortality; the study's simulations suggested material reduction in covered livestock mortality risk under its particular calibration. This is evidence for one design, population and risk model. It does not validate all index insurance products. Carter, de Janvry, Sadoulet, and Sarris's 2017 review examined the promise and constraints of agricultural index insurance more broadly. (see Chantarat (2013); Carter (2017))
CCRIF SPC demonstrates a sovereign and public-service use of parametric instruments. It reports rapid liquidity provision to participating Caribbean and Central American members following specified events. In its 2024–2025 annual reporting, CCRIF describes USD 84.5 million of Hurricane Beryl-related payouts to seven members and associated entities. This is a documented institutional example, not proof that every affected loss was compensated or that all trigger mismatches were absent. (see Caribbean Catastrophe Risk (2025))
A rigorous product assessment estimates basis risk under current and alternative climatic conditions; checks observation and satellite reliability; verifies governance of trigger data; tests rapid payment operations; evaluates consumer comprehension; and compares product welfare against reserves, emergency credit, indemnity insurance and social protection. Climate change increases the importance of ongoing recalibration because an index estimated from historical conditions may no longer capture future local losses.
Agricultural, Health, Life, and Business Interruption Exposures
Climate insurance extends beyond residential property. Agricultural production is affected by drought, precipitation extremes, heat, shifting growing seasons, pests and input-cost volatility. Traditional yield insurance, revenue insurance, livestock products, and index-based contracts allocate different combinations of production and price risk. Spatially correlated drought makes agricultural insurance particularly dependent on reinsurance, public support, and the credibility of long-term loss estimates. Index designs must attend to the basis-risk problem discussed above. (see Chantarat (2013); Carter (2017))
Health and life-related effects arise through extreme heat, altered disease environments, air pollution associated with wildfire, disrupted health services, and disaster-related mortality or displacement. The financial implications differ across medical reimbursement, life insurance, disability, workers' compensation and public health systems. Attribution is complicated by medical causation, adaptation, changing populations, and the design of health coverage. Insurers should not infer a universal claim multiplier from temperature alone; lines of business have distinct exposures and benefit triggers.
Business interruption exposes firms to lost operating income following covered property damage, contingent supply-chain disruptions, power outages, closed transportation routes, and workforce displacement. Whether a loss is insured often depends on physical-damage clauses, named-peril definitions, waiting periods, sublimits and exclusion language. A regional flood may cause severe business consequences outside the policy's indemnity provisions. Climate-sensitive business-risk assessments need network analysis as well as local hazard mapping.
Agricultural and business exposures connect to systemic risk through production, trade, food prices, credit, and public budgets. Physical events can transmit through supply chains long after the immediate hazard subsides. Hsiang and colleagues' 2017 study of U.S. climate damages integrated climate, econometric and process-model evidence across several economic sectors; Burke, Hsiang, and Miguel's 2015 work found nonlinear historical relations between temperature and economic output. These studies help establish the breadth of socioeconomic sensitivity, while their macroeconomic estimates cannot be directly entered as insurance claims without translating them through particular covered contracts. (see Hsiang (2017); Burke (2015))
Effective insurance design in these sectors demands multi-horizon exposure mapping, benefit-specific data, risk-financing combinations and attention to health and food security. Financial protection becomes scientifically meaningful only when the relevant loss-bearing entity and recovery objective are defined.
Public–Private Insurance Systems and Risk Allocation
Public–private systems emerge where commercial insurance alone cannot achieve socially desired access or continuity. They can include mandatory pooling, public reinsurance, disaster funds, premium support, industry levies, building-code conditions and explicitly financed catastrophe layers. Their design must distinguish who owns the risk, who prices it, who provides liquidity after an event, who bears the ultimate fiscal loss, and how mitigation is rewarded.
The United Kingdom's Flood Re is a joint insurance-industry and government arrangement supporting affordable flood insurance for eligible high-risk households. Its legislation-based transition is scheduled to end the scheme in 2039, after which the ambition is for a functioning market with more risk-reflective pricing. The transition date is part of the institutional design; it does not itself ensure that future physical risk will be affordable. Flood Re's public description links future availability to improved risk management, understanding of exposure, and resilience. (see Flood Re (2024))
France and other countries also employ mixed arrangements for natural catastrophes, but institutional details, eligible perils, financing rules and government guarantees differ. They should be studied at the level of operative law and current contracts rather than treated as identical national versions of a generic public backstop. The EU level remains a proposal area rather than an enacted uniform catastrophe insurance scheme.
In December 2024, EIOPA and the ECB proposed two complementary European pillars for discussion: a public–private reinsurance arrangement to pool private catastrophe risk and a public disaster-financing fund tied to risk-reduction conditions. The proposal identifies possible economies of geographic diversification and the importance of mitigation incentives. It is a policy proposal and must not be described as an implemented EU-wide guarantee. (see EIOPA guidance)
The strongest public–private arrangements make contingent liabilities transparent. A government can create affordable premiums by assuming a disaster layer, but the underlying expected cost remains real. Targeted support may be preferable to blanket implicit guarantees where subsidies disproportionately accrue to high-value properties. Conversely, reliance solely on private prices can exclude households with limited income or mobility. Scientific evaluation compares fiscal sustainability, coverage, mitigation, equity and speed of recovery, including effects on future land use.
Sovereign Risk Financing, Development, and Social Protection
Governments face disaster-related expenditure, lost revenue, emergency infrastructure needs, and pressure to assist uninsured households. Sovereign disaster risk financing combines reserves, budget contingencies, contingent credit, market-based risk transfer, and international support according to the frequency and severity of expected events. The World Bank describes a layered approach: relatively frequent lower-severity losses can be financed by budget instruments; medium events may use contingent credit; catastrophic layers may use insurance or other risk transfer. The correct mix depends on national fiscal capacity, borrowing costs, market access and the social objectives of recovery. (see World Bank Group (2026))
Insurance can provide predetermined claims payment and liquidity, but public borrowing and fiscal buffers may be more efficient for other layers. Parametric sovereign cover offers quick disbursement but leaves basis risk and often covers only a fraction of total national losses. Insurance premium subsidies, donor capitalization, and market access alter the true cost to the country. Cost comparisons should include expected payments, transaction costs, speed, financing conditions, basis risk, and opportunity cost.
Social protection is central because damage to informal housing, subsistence livelihoods, health, and local infrastructure may not map cleanly to commercially insurable property. Scalable cash-transfer systems, emergency healthcare, social insurance, and locally coordinated recovery can complement catastrophe policies. Poor households may need protection against recurring income interruptions as much as against a singular large asset loss. Fiscal policy has to account for distribution and continuity of public services as well as insured payouts.
The 2026 UNEP Adaptation Gap Report, publicly issued on November 4, estimates developing-country adaptation financing needs at approximately USD 305 billion per year by 2035, compared with international public adaptation finance flows of USD 28.2–31.7 billion in 2024. These are adaptation-finance figures, not insurance premiums or uninsured catastrophe losses. UNEP also discusses loss-and-damage funding pressures. The quantities indicate the scale of the wider financing challenge and must not be represented as a direct calculation of a global insurance protection gap. (see United Nations Environment (2026))
The economic value of risk financing lies in preserving critical functions before, during and after disaster. Sovereign policy therefore should test the timing and sufficiency of post-event funds, debt sustainability, mitigation incentives and access by the most vulnerable groups, rather than assess a program solely by how large a headline insured sum it announces.
International Comparative Evidence and Institutional Diversity
Climate-risk insurance is internationally heterogeneous. In the United States, homeowners insurance markets, state-based regulation, geographically detailed catastrophe risk, and a separate structure for much flood coverage create multiple overlapping protection questions. The 2018–2022 Federal Insurance Office dataset makes visible differences in premium, nonrenewal, and loss experience at ZIP Code scale. Its period and peril definitions are fundamental: it cannot establish the state of the 2026 market without updated observations. (see U (2025))
In Europe, EIOPA combines country and peril data and examines national insurance arrangements. The region's catastrophe exposures vary from flood and storm to drought and wildfire, while contract terms and public intervention differ between countries. Aggregating the region can hide high local protection gaps. The EIOPA–ECB proposal illustrates continuing debate about whether broader pooled backstops can preserve access while safeguarding prevention incentives. (see EIOPA guidance)
In the Caribbean and Central America, CCRIF's parametric sovereign and utility products address rapid liquidity for specified weather and earthquake triggers. They do not purport to replace all reconstruction finance. In the United Kingdom, Flood Re has an explicitly time-limited transition design oriented toward affordable household flood protection. These cases show different public objectives, beneficiary groups and operational instruments. (see Caribbean Catastrophe Risk (2025); Flood Re (2024))
Low- and middle-income economies face further challenges from limited insurance penetration, weaker exposure data, high agricultural dependence, inadequate public infrastructure, and fiscal constraints. Insurance access can support recovery, but donor support and public policy may be necessary to make it workable. The IPCC emphasizes the unequal capacity to manage residual climate risk and the danger that poorly designed insurance shifts burdens toward vulnerable populations. (see IPCC climate assessment)
A scientifically valid country comparison should use a structured matrix: eligible perils, insured populations, mandatory status, premium funding, public guarantee, risk-based price adjustments, reinsurance, trigger type, claims timeliness, mitigation obligations, data transparency, and distributional outcomes. It should also assess governance capacity, local hazard history and climate projections. The logical consequence is that no one scheme is universally transplantable; instruments work through national contracts, institutions and hazards. Comparative learning remains valuable when these underlying conditions are made explicit.
Climate Scenarios, Stress Testing, and Model Governance
Climate scenarios describe internally coherent possible pathways conditioned on assumptions about policy, emissions, technology, socioeconomic development, and physical hazards. They are analytical tools for decision-making under uncertainty. Supervisors and insurers use them to examine longer-term solvency, investment, underwriting and adaptation strategies, particularly when historical experience does not cover future conditions. A scenario is neither a forecast nor an empirically measured distribution of annual insurance claims. (see Network for Greening (2025); EIOPA guidance)
A well-governed assessment begins with a decision: for example, evaluating future reinsurance capacity for coastal property, the solvency impact of a clustered catastrophe year, or the affordability of flood coverage over two decades. The analyst then selects the relevant horizon, baseline, hazard and socioeconomic pathway, exposure and vulnerability assumptions, contractual conditions, financing arrangements, and outcome metrics. Short-horizon catastrophe stress and century-scale chronic damage projections answer different questions.
Scenario uncertainty can be decomposed into emissions pathways, climate response, natural variability, downscaling, exposure, vulnerability, macroeconomic translation, and financial-market responses. Sensitivity analysis should reveal which uncertainties drive decisions. Backtesting may be meaningful for near-term hazards and claims, while century-scale outcomes cannot be directly validated against observed future losses. Numerical precision must not be confused with evidentiary strength.
A documented scientific caution is required for the NGFS Phase V long-term scenarios. In a statement dated December 3, 2025, NGFS disclosed that the Kotz et al. (2024) article used in some physical damage estimates had been retracted by Nature, and identified the affected outputs. It clarified that its scenarios are not forecasts and that other unaffected scenario components and short-term NGFS scenarios were not implicated in the same way. Accordingly, this monograph does not treat the retracted paper's quantitative damage estimates as validated evidence. Any future analysis using relevant NGFS data must identify dataset version and whether a variable depends on that paper. (see Network for Greening (2025))
Governance requires source provenance, model inventories, scenario version locks, independent challenge, documented limitations and decision-specific signoff. The existence of a stress output cannot substitute for scientific review. The purpose is disciplined choice under uncertainty, not a claim to certainty.
Metrics, Disclosure, Data Quality, and Reproducible Research
Climate-risk insurance uses multiple measurement systems: physical measurements, geospatial exposures, damage inventories, paid and incurred claims, premium and policy counts, solvency capital, reinsurance attachment, and recovery outcomes. Their denominators differ. Insured catastrophe loss is a financial measure; climate-attributable hazard change is a scientific counterfactual measure; insurance take-up is a consumer-market measure; and adaptation effectiveness is a real-world reduction in adverse consequences. These metrics must not be merged without a stated conversion method.
A credible dataset records observation dates, geography, peril definitions, value normalization, currency, coverage terms, reporting lag, sampling uncertainty and missingness. Catastrophe loss estimates may be revised after events as claims settle. Industry estimates sometimes include non-weather geological perils, whereas the climate-science object concerns climatic hazards. For this reason, a natural catastrophe total should never be labeled entirely climate-induced.
The May 2025 U.S. National Oceanic and Atmospheric Administration notice that its Billion-Dollar Weather and Climate Disasters product would cease updating after calendar year 2024 illustrates a data continuity issue. Its historical archive remains available, but a researcher should not silently cite the archived series as current official 2025 or 2026 observations. Other datasets must be identified by provider and methodology. (see National Oceanic and (2025))
Disclosure frameworks also differ. IFRS S2 addresses climate-related sustainability disclosures for entities applying it; IFRS 17 governs the accounting treatment of insurance contracts. IAIS supervisory guidance addresses materiality, governance, reporting and the relevance of climate scenarios in insurance supervision. These frameworks have distinct legal and analytical functions. Compliance with one cannot demonstrate satisfaction of another without a separate mapping. (see IFRS Foundation, International (2023); IFRS 17)
Reproducibility can be improved by preserving data dictionaries, event-set versions, code descriptions, source files where legally shareable, uncertainty intervals, and provenance for observed and modeled results. Independent researchers should be able to reconstruct source–claim relationships even where commercial model code is proprietary. This monograph documents its checked public sources and distinguishes source-stage verification from external replication.
Climate Losses, Macroeconomic Feedback, and Systemic Risk
A large climate event may damage an insurer's portfolio directly, impair the economic assets supporting those policies, cause widespread infrastructure failure, and tighten financial conditions simultaneously. Loss transmission includes business interruption, impaired collateral, government expenditure, sovereign credit stress, migration, and reduced household consumption. This creates potential systemic channels even where individual insurers remain solvent.
The empirical macroeconomic literature supports important but model-dependent pathways. Burke, Hsiang, and Miguel analyzed nonlinear relations between temperature and economic output across countries; Hsiang and colleagues combined spatial climate and economic models to estimate potential U.S. damages; Kahn and coauthors estimated longer-term macroeconomic impacts of persistent climate changes. Their assumptions, periods, spatial aggregation and outcome variables differ. One should not add together their percentages or interpret a modeled GDP deviation as an insurance claim. (see Burke (2015); Hsiang (2017))
Systemic risk arises when losses and financial responses interact. For example, a major storm can raise claims, reduce asset prices, increase demand for liquidity, strain public budgets, raise reinsurance prices and lead multiple insurers to curtail high-risk coverage. The resulting withdrawal can affect mortgage availability, property values and local tax bases. This causal chain is plausible and institutionally meaningful, but its magnitude requires linked financial and economic data; a single catastrophe statistic does not establish it.
Diversification across countries can be weakened by the co-movement of investment assets, global reinsurance market cycles, shared supply-chain dependencies and multi-region climatic conditions. At the same time, broader pooling can reduce concentration where correlations are genuinely lower. The appropriate conclusion depends on peril-level dependence and financial stress behavior. Forward-looking supervision therefore needs a macroprudential as well as a firm-level view, as reflected in the IAIS 2025 guidance. (see IAIS supervisory guidance)
The societal objective is resilience of essential functions, not merely survival of insurers' balance sheets. A country can maintain strong insurance solvency while large parts of its population remain uninsured. Conversely, a financially generous but undercapitalized scheme can fail precisely when protection is most necessary. This tension must be measured and governed.
Equity, Climate Justice, and the Distribution of Financial Protection
Climate harm, insurance access and capacity to adapt are unequally distributed. Households with fewer resources may live in high-risk locations because housing costs, employment, historical settlement, or legal constraints limit choice. They may also lack affordable coverage, access to credit, and reserves for deductibles or temporary displacement. Richer owners may hold larger insured asset values and receive a larger share of public support under untargeted subsidies. A simple coverage rate cannot show whether the financial protection system distributes benefits fairly.
Risk-based pricing communicates expected loss where estimates are credible, while equity policy determines whether and how public resources support households facing unaffordable premiums. These functions can be separated in financial accounting even when they are combined administratively. A transparent system can calculate actuarial cost, specify a social contribution or income-sensitive support, and disclose who finances the difference. This allows debate over distributional objectives without concealing underlying risk.
The IPCC identifies financial and institutional constraints on adaptation, including especially severe limitations in vulnerable economies. UNEP's 2026 adaptation-finance assessment documents the shortfall between international financing flows and modeled needs. These findings support a broad view of climate financial protection, but do not justify assuming that the insurance market can finance all adaptation. (see IPCC climate assessment; United Nations Environment (2026))
Ethical assessment should include involuntary exposure, access to information, treatment in claims, intergenerational consequences, and the distribution of public guarantees. It should also consider when preserving settlement in extremely hazardous places may lock in recurrent harm. Assistance for relocation or changed land use can sometimes be more protective than indefinite premium support; decisions require local evidence, rights-based processes and consideration of affected communities.
The research standard is explicit incidence analysis: identify who benefits, who pays, who bears residual loss, and what incentives are created. An insurer's actuarial adequacy and a society's equitable protection are related but distinct outcomes, each requiring its own evidence.
Research Problems, Competing Approaches, and Limits of Knowledge
The field contains several overlapping analytical traditions. Physical risk science emphasizes climate dynamics, event attribution, and compound hazards. Actuarial science emphasizes probability distributions, solvency, and risk pooling. Insurance economics emphasizes demand, information, incentives, selection, and market structure. Engineering emphasizes the vulnerability of exposed assets and practical risk reduction. Institutional and legal scholarship emphasizes contracts, regulatory duties, social protection and public financing. None of these perspectives alone supplies a complete answer to how financial protection can be sustained.
Persistent research problems include poor geospatial exposure data; sparse tail observations; uncertainty in downscaled hazard estimates; the changing vulnerability of building stocks; economic feedback and insurer behavior; incomplete coverage data; poorly measured informal losses; price elasticity under high-risk conditions; fairness of subsidies; parametric basis risk; long-tail climate litigation; and the possibility of correlated failures in risk-transfer and investment markets.
An additional methodological challenge is the difference between a successful modeled result and a demonstrated policy effect. A catastrophe model may accurately estimate damage for a defined portfolio yet provide no causal evidence that a premium incentive will improve household adaptation. A randomized trial of insurance take-up may identify behavior for a specific study population while saying little about nationwide catastrophe accumulation. A climate projection may establish a changed hazard signal without resolving policy-wording or reconstruction inflation. The evidentiary bridge must be constructed explicitly in each application.
The NGFS retraction notice shows why model governance is a scientific matter, not only a compliance matter. Scientific estimates can change after publication, and downstream users must be capable of identifying affected variables and revising decisions. The 2025 NOAA archive discontinuation similarly shows that data availability and temporal coverage require continuous monitoring. (see Network for Greening (2025); National Oceanic and (2025))
Future empirical progress should focus on linked event–property–contract–claim data, independently evaluated engineering interventions, transparent longitudinal insurance take-up records, reproducible peril-specific attribution, risk-layer cost comparison, and evaluations that combine financial recovery with welfare and physical resilience. These are research priorities derived from the synthesized literature rather than claims that the present monograph has experimentally established a new causal model.
Original Analytical Synthesis: The Conditions of Sustainable Climate Insurance
The evidence permits a consolidated scientific interpretation. Insurance has three interlocking functions in a changing climate: it estimates and prices selected uncertainty, redistributes the financial consequences of covered events, and can support adaptation through information and contractual incentives. Their effectiveness depends on the surrounding system of physical risk reduction, household finances, legal rights, public institutions and capital provision. The functions must be evaluated simultaneously because optimizing one may weaken another.
The first condition is epistemic adequacy: a risk estimate must describe the relevant peril, exposure, vulnerability, coverage and horizon, with material uncertainty disclosed. The second is contractual adequacy: the payout must correspond to economically important loss with understandable terms and effective claims handling. The third is financial adequacy: premiums, reserves, reinsurance, capital and liquidity must support obligations under both expected and severe conditions. The fourth is social adequacy: coverage and assistance must reach those whose welfare and recovery are at stake, without perpetuating avoidable harm. These four conditions are an explanatory organization of established scientific and policy considerations, not a separately proposed proprietary doctrine.
The main analytical tension is that rapidly adjusting annual insurance prices can communicate deteriorating risk sooner than households, buildings and infrastructure can adapt. This mismatch can increase short-run withdrawal and financial exclusion even when risk pricing is technically sound. It creates a specific rationale for coordinated policy: adaptation infrastructure, resilience grants, public financing for transitional affordability, transparent future-risk information, and credible risk-based capital management can work on different timelines. The public role should be explicit, costed and subject to mitigation incentives.
A second tension concerns aggregate claims statistics. The 2025 natural catastrophe totals from Swiss Re and Munich Re confirm substantial insured loss, while their distinct methods and event mixes caution against simple historical attribution. WMO and IPCC evidence establishes change in relevant physical extremes, but economic losses also reflect exposure and vulnerability. Scientific integrity requires that the two bodies of evidence be joined through peril-specific modeling rather than rhetorical inference. (see IPCC climate assessment; World Meteorological Organization (2026))
A third tension concerns what the protection gap signifies. Insurance expands financial compensation, but a narrow increase in insured-loss share may be accompanied by unresolved physical vulnerability. Conversely, effective mitigation can reduce expected losses without increasing premiums or insured payouts. Performance should therefore be judged by a plural set of observed outcomes: lower physical damage, sustainable coverage, solvency, claims timeliness, fiscal resilience, and fair household recovery. (see IPCC climate assessment; EIOPA guidance)
The central conclusion is that climate risk and insurance is a joint science of hazard, economic exposure, institutional risk transfer, and adaptation. Sustainable protection arises when those elements are governed as an integrated system while preserving their distinct methods and evidence.
Methodological Appendix: Worked Interpretations and Verification Rules
The following worked interpretations are pedagogical and contain no new empirically estimated parameters.
Illustration A — protection gap. Suppose independently verified economic catastrophe losses are 500 million currency units and the corresponding contract-defined insured losses are 150 million. The event-specific protection gap is (500 − 150)/500 = 70%. That figure is meaningful only if loss geography, peril inclusion, date, currency and definition agree. It says nothing alone about what proportion of the event was caused by anthropogenic climate change.
Illustration B — conditional expected loss. Suppose an insured asset has damage severity of 20,000 with conditional annual probability 2% in one well-defined scenario and 3% under another scenario, with exposure, vulnerability and coverage held constant. Conditional expected damage rises from 400 to 600 currency units, a 50% difference. The example demonstrates sensitivity to assumptions. It is not an empirical forecast, and actual premium differences may include risk capital, fixed expenses, catastrophe dependence and regulatory constraints.
Illustration C — parametric basis risk. A contract pays 10,000 when an index crosses a specified threshold. If a household incurs 20,000 of qualifying economic harm while the index does not trigger, its payout is zero under this simplified trigger, producing material negative basis risk. If the index triggers when actual damage is small, the financial relationship differs in the other direction. The evaluation must assess the distribution of mismatch, not only the trigger's average accuracy.
Illustration D — return period. Under a fixed annual event probability p = 0.01 and independent years, the probability of at least one such event in a ten-year horizon is 1 − (1 − 0.01)^10, approximately 9.56%. The result changes if annual probabilities vary or events are dependent. A nominal one-in-one-hundred-year event can recur in consecutive years.
Verification rules used in preparing this monograph are: retain exact source and publication dates; identify jurisdiction and scope for regulation; separate observations, estimates, scenarios and proposals; treat an abstract-only scientific paper as bibliographically checked but not independently replicated; exclude retracted damage estimates as primary validation; never translate uninsured loss automatically into covered claims; and assign no public publication date before production release. In the evidence record, each reference is linked to a particular checked claim. The source-stage manuscript is prepared for further technical release operations without representing that publication has occurred.
Conclusion
Climate risk reshapes insurance through changes in physical hazards, exposed property and populations, structural vulnerability, correlated losses, market behavior, contractual obligations, and financial capacity. Its scientific treatment requires a transparent sequence from climate evidence through peril-specific risk assessment to economic damage, insured claims, capital requirements, and social recovery.
The reviewed evidence establishes changes in several climatic extremes, material catastrophe claims, heterogeneous insurance accessibility, and extensive institutional efforts to adapt supervision and public financing. It also establishes important limits: loss trends have multiple causes; scenario outputs have different evidentiary status from forecasts; market affordability cannot be inferred from actuarial adequacy alone; and financial compensation cannot replace adaptation. The NGFS Phase V retraction episode exemplifies the need to trace model dependencies before relying on numerical damage estimates.
The integrated conclusion is that durable climate insurance requires risk-sensitive knowledge, clear contracts, financially sound underwriting and reinsurance, physical loss prevention, transparent public financing, and measurable social protection. Scientific policy must evaluate their combined consequences across perils, territories and time horizons. Public institutions, insurers, engineers and communities act on different parts of the same risk process; effective governance recognizes their distinct responsibilities and mutual dependence.
The monograph supplies an evidence-attributed synthesis and a structured comparison of methods, institutions and outcomes. It does not claim a newly invented actuarial theory, numerical catastrophe calibration, independent replication of restricted models, or externally reviewed empirical novelty. The remaining research agenda is to deepen linked microdata, test causal interventions, and update climate and insurance assumptions as evidence changes.
Editorial and Publication Provenance
Author: Angela Bogdanova (Artificial Sapiens). Publisher: Aisentica Development. Published by Aisentica Development. This source-stage scientific manuscript has a reserved canonical URL and is not a publicly released Science page. A first-publication date, global publication identifier and production verification are assigned only by a separate, independently verified publisher release. Scientific source correspondence and the actual internal editorial disposition are recorded in the associated EVIDENCE.json and PASSPORT.json files.
Research Dossiers
143 approved research subjects are organized into 6 thematic families. Individual titles become links when their complete scholarly monographs are published and verified.
Foundations, History, and Concepts2
Law, Regulation, and Governance12
- Climate Adaptation Metrics for Insurance Regulatory Reporting
- Climate Insurance Sovereign Risk Pool Design and Governance
- Climate Litigation Pathways and Liability Insurance Coverage Triggers
- Climate Model Update Governance and Underwriting Rate Sensitivity
- Climate Risk Governance and Insurer Strategic Decisions
- Climate Risk Governance in Insurance Regulatory Frameworks
- Climate Risk Insurance Capital Modeling Supervisory Validation
- Directors and Officers Climate Disclosure Litigation Coverage
- Insurance Regulatory Stress Testing and Supervisory Scenarios
- Insurance Supervisory Response to Protection Gaps and Affordability
- Insurer Early Intervention Plans and Regulatory Escalation
- International Supervision of Climate Risk in Insurance Groups
Risk Science, Data, and Modeling96
- Actuarial Catastrophe Loss Modeling
- Actuarial Climate Scenario Analysis
- Agricultural Insurance Protection Gaps and Climate Vulnerability
- Behavioral Barriers to Climate and Disaster Protection
- Business Interruption and Contingent Business Interruption Catastrophe Loss
- Catastrophe Bond Risk Modeling and Expected Loss Calibration
- Catastrophe Event Attribution and Counterfactual Loss Normalization
- Catastrophe Loadings in Insurance Ratemaking
- Catastrophe Reinsurance Demand and Rate on Line Cycles
- Catastrophe Risk Capital Calibration
- Climate Adaptation Incentives in Property Insurance Contracts
- Climate Adaptation Malinvestment and Insurance Moral Hazard
- Climate Adaptation Premium Credits and Verification
- Climate Adaptation Return on Investment for Insurance Underwriting
- Climate and Cyber Stress in Insurance Solvency Assessments
- Climate Attribution of Insured Catastrophe Loss Trends
- Climate Change and Insurance Reserve Adequacy
- Climate Change and Public Insurance Residual Markets
- Climate Change and Reinsurance Contract Duration Mismatch
- Climate Change and Severe Convective Storm Insurance Hazard Assessment
- Climate Change Signal Detection in Insurance Loss Data
- Climate Change Thresholds in Insured Loss Severity Functions
- Climate Disaster Microinsurance for Informal Settlements
- Climate Disaster Recovery and Social Protection Integration
- Climate Disaster Relief and Insurance Crowd-Out
- Climate Displacement and Insurance Portability
- Climate Event Causation and Insurance Claims Evidence
- Climate Exclusions and Insurance Contract Interpretation
- Climate Exposure and Property Insurance Requirements in Mortgage Lending
- Climate Hazard Downscaling for Insurance Catastrophe Models
- Climate Hazard-to-Claim Causal Chain Validation
- Climate Insurance Affordability and Household Income Risk
- Climate Insurance Disclosure under IFRS S2 and Jurisdictional Standards
- Climate Model and Insurance Loss Out-of-Sample Validation
- Climate Model Ensemble Calibration for Insurance Catastrophe Risk
- Climate Nonstationarity and Insurance-Linked Securities Tail Risk
- Climate Reinsurance Basis Risk and Retrocession
- Climate Risk and Coastal Community Insurance Equity
- Climate Risk and Nature Loss Insurance Interactions
- Climate Risk Concentration in Multinational Insurance Groups
- Climate Risk Data Sovereignty and Cross-Border Access
- Climate Risk Data Vendor Transparency and Audit Rights
- Climate Risk in Insurance ORSA and Capital Planning
- Climate Risk Insurance Price Volatility and Policyholder Retention
- Climate Risk Mapping and Underwriting Geographic Fairness
- Climate Risk Uncertainty and Nonstationarity
- Climate Scenario Dependence and Multi-Peril Insurance Accumulation
- Climate Scenario Impacts on Insurance Valuation and Technical Provisions
- Climate Scenario Probability Interpretation for Insurance Decisions
- Climate Scenario Transmission to Insurer Asset Liability Management
- Climate Sovereign Risk Financing and Contingent Credit
- Climate Stress Correlation between Insurer Assets and Liabilities
- Climate Transition Credit Risk and Insurance Group Capital
- Climate Transition Project Completion and Performance Insurance
- Climate Transition Risk in Insurer Investment Portfolios
- Climate Transition Supply Chain Liability Insurance
- Climate-Adjusted Deductible and Coverage Limit Optimization
- Climate-Driven Flood Probability and Insured Loss Attribution
- Climate-Driven Insurance Protection Gaps in Emerging Economies
- Climate-Related Greenhouse Gas Emissions Liability Coverage
- Climate-Resilient Building Codes and Insured Damage Outcomes
- Climate-Sensitive Disease Burden and Health Insurance Exposure
- Climate-Sensitive Health Insurance Protection Gaps
- Commercial Insurance Nonstationary Physical Hazard and Site Exposure
- Compound and Cascading Hazard Risks
- Corporate Climate Transition Risk and Commercial Insurance Coverage
- Demand Surge Repair Cost Inflation and Catastrophe Claims Severity
- Dynamic Risk and Time-Dependent Exposure
- Forecast-Based Financing and Anticipatory Insurance Payouts
- Glacial Lake Outburst and Mountain Catastrophe Risk
- Global Catastrophe Reinsurance Capacity and Tail Supply Shocks
- Heatwave and Extreme Temperature Catastrophe Loss Modeling
- Insurance and Catastrophe Experience
- Insurance Climate Transition Accountability and Greenwashing
- Insurance Climate-Related Consumer Communication and Coverage Understanding
- Insurance Coverage of Slow-Onset Climate Damage
- Insurance Incentives for Climate Adaptation
- Insurance Loss Data Normalization for Climate Attribution
- Insurance Market Climate Risk Disclosure Reporting Comparability
- Model Drift Detection and Forecast Backtesting in Insurance
- Multi-Year Insurance Contract Design under Climate Nonstationarity
- Nonstationary Dependence in Climate Insurance Portfolio Tail Risk
- Ocean Warming and Marine Insurance Hazard Change
- Physical Climate Risk and Insurance Product Withdrawal
- Physical Climate Risk in Insurer Real Estate Investments
- Post Catastrophe Reconstruction Finance and Building Back Better
- Property Insurance Secondary Perils and Compound Disaster Exposure
- Property Risk Portfolio Diversification Under Climate Change
- Risk as Hazard, Exposure, and Vulnerability
- Sea Level Rise and Chronic Coastal Insurance Exposure
- Socioeconomic Pathway Integration in Insured Exposure Forecasts
- Sovereign Parametric Risk Transfer and Basis Risk
- Spatial Statistics and Geographic Risk Prediction for Insurance
- Wildfire Climate Signals and Insurance Loss Attribution
- Wildfire Defensible Space and Property Risk Assessment
- Wildfire Structure Hardening and Ember Exposure Mitigation
Finance, Capital, and Markets11
- Agricultural and Weather Reinsurance
- Carbon Credit Reversal and Environmental Insurance
- Compound Heat and Drought Insurance Portfolio Loss
- Disaster Insurance and Fiscal Externalities
- Insurance of Carbon Market Settlement and Delivery Risks
- Insurance Protection Gaps and Financial Vulnerability
- Insurance Valuation of Public Flood Protection Infrastructure
- Nonstationary Risk and Forward-Looking Insurance Pricing
- Property Insurance Market Nonrenewal and Geographic Retreat
- Public-Private Insurance and Disaster Risk Finance
- Resilience Bonds and Insured Loss Reduction Finance
Insurance Products, Coverage, and Claims11
- Carbon Capture and Storage Long-Tail Liability Insurance
- Community Adaptation Spillovers in Insurance Risk Pools
- Disaster Memory Decay and Renewal of Coverage
- Disaster Salience and Insurance Uptake
- Environmental Impairment Liability Insurance and Pollution Claims
- Insurance and Loss Prevention
- Morbidity Trend and Care Cost Inflation in Long-Term Care Insurance
- Multiyear Drought Propagation across Insurance Lines
- Nature Based Solutions for Flood and Storm Surge Insurance Risk Reduction
- Renewable Energy Construction and Operational Insurance
- Urban Drainage Expansion and Flood Insurance Risk Reduction
Specialized and Interdisciplinary Research11
- Automobile Flood Hail and Wildfire Damage Accumulation
- Availability Heuristics and Rare-Loss Judgments
- Commercial Environmental Liability and Restoration Program Design
- Extreme Heat Stress and Critical Facility Adaptation
- Insurability and Its Boundaries
- Insurability of Catastrophic and Systemic Risks
- Insurer Emerging Risk Identification and Horizon Scanning
- Nonstationarity and Time-Varying Risk
- Permafrost Thaw and Insured Infrastructure Damage
- Real-Income Durability Under Annuity Inflation Stress
- Tropical Cyclone Intensification and Insured Wind Loss Dynamics
References
- IPCC. 2021. Climate Change 2021: The Physical Science Basis, Working Group I, Summary for Policymakers. https://www.ipcc.ch/report/ar6/wg1/chapter/summary-for-policymakers/
- IPCC. 2022. Climate Change 2022: Impacts, Adaptation and Vulnerability, Working Group II, Chapter 17: Decision-Making Options for Managing Risk. https://www.ipcc.ch/report/ar6/wg2/chapter/chapter-17/
- IPCC. 2022. Climate Change 2022: Impacts, Adaptation and Vulnerability, Working Group II, Chapter 16: Key Risks across Sectors and Regions. https://www.ipcc.ch/report/ar6/wg2/chapter/chapter-16/
- IPCC. 2023. Climate Change 2023: Synthesis Report of the Sixth Assessment Report. https://www.ipcc.ch/report/ar6/syr/
- World Meteorological Organization. 2026. State of the Global Climate 2025 (published March 23, 2026). https://public.wmo.int/publication-series/state-of-global-climate/state-of-global-climate-2025
- International Association of Insurance Supervisors. 2025. Application Paper on the Supervision of Climate-related Risks in the Insurance Sector (April 16). https://www.iais.org/activities-topics/climate-risk/
- European Insurance and Occupational Pensions Authority. 2025. Dashboard on Insurance Protection Gap for Natural Catastrophes (updated December 5). https://www.eiopa.europa.eu/tools-and-data/dashboard-insurance-protection-gap-natural-catastrophes_en
- United Nations Environment Programme. 2026. Adaptation Gap Report 2026: Ready? Not (released November 4). https://www.unep.org/en/publications/report/adaptation-gap-report-2026
- Swiss Re Institute. 2026. sigma No. 1/2026: Natural Catastrophes in 2025: The Persistent Rise of Wildfire and Storm Risk (March). https://www.swissre.com/institute/research/sigma-research/sigma-2026-01-natcat-2025-wildfire-storm-risk.html
- Munich Re. 2026. Natural Disaster Figures 2025: Climate Change Presses On (January 13). https://www.munichre.com/en/company/media-relations/media-information-and-corporate-news/media-information/2026/natural-disaster-figures-2025.html
- U.S. Department of the Treasury, Federal Insurance Office. 2025. Analyses of U.S. Homeowners Insurance Markets, 2018–2022: Climate-Related Risks and Other Factors (January). https://home.treasury.gov/system/files/311/Analyses_of_US_Homeowners_Insurance_Markets_2018-2022_Climate-Related_Risks_and_Other_Factors_0.pdf
- Network for Greening the Financial System. 2025. Statement Regarding Physical Risk Estimates in Phase V of NGFS Long-term Scenarios (December 3). https://www.ngfs.net/en/press-release/statement-regarding-physical-risk-estimates-phase-v-ngfs-long-term-scenarios
- European Insurance and Occupational Pensions Authority. 2022. Application Guidance on Climate Change Materiality Assessments and Climate Change Scenarios in ORSA (August 2). https://www.eiopa.europa.eu/publications/application-guidance-climate-change-materiality-assessments-and-climate-change-scenarios-orsa_en
- IFRS Foundation. 2017, as amended. IFRS 17 Insurance Contracts, effective from annual periods beginning January 1, 2023. https://www.ifrs.org/issued-standards/list-of-standards/ifrs-17-insurance-contracts/
- IFRS Foundation, International Sustainability Standards Board. 2023. IFRS S2 Climate-related Disclosures (issued June 26). https://www.ifrs.org/supporting-implementation/supporting-materials-for-ifrs-sustainability-disclosure-standards/ifrs-s2/
- European Insurance and Occupational Pensions Authority and European Central Bank. 2024. Towards a European System for Natural Catastrophe Risk Management (December 18). https://www.eiopa.europa.eu/publications/eiopa-and-ecb-joint-paper-towards-european-system-natural-catastrophe-risk-management_en
- World Bank Group. Disaster Risk Finance and Insurance: Program and Risk-Layering Approach (institutional guidance, accessed 2026). https://www.worldbank.org/ext/en/topic/financial-sector/disaster-risk-finance-and-insurance
- Caribbean Catastrophe Risk Insurance Facility SPC. 2025. Annual Report 2024–2025; program and payout reporting. https://www.ccrif.org/en/publications
- Flood Re. 2024. Transition and the 2039 End of the Scheme (institutional transition statement). https://www.floodre.co.uk/ara-2024/transition/
- Kousky, Carolyn. 2019. The Role of Natural Disaster Insurance in Recovery and Risk Reduction. Annual Review of Resource Economics 11:399–418. DOI: 10.1146/annurev-resource-100518-094028. https://www.annualreviews.org/content/journals/10.1146/annurev-resource-100518-094028
- Mills, Evan. 2005. Insurance in a Climate of Change. Science 309(5737):1040–1044. DOI: 10.1126/science.1112121. https://pubmed.ncbi.nlm.nih.gov/16099975/
- Botzen, W. J. Wouter, Howard Kunreuther, and Erwann Michel-Kerjan. 2019. Protecting against Disaster Risks: Why Insurance and Prevention May Be Complements. Journal of Risk and Uncertainty 59(2). DOI: 10.1007/s11166-019-09312-6. https://ideas.repec.org/a/kap/jrisku/v59y2019i2d10.1007_s11166-019-09312-6.html
- Hsiang, Solomon, et al. 2017. Estimating Economic Damage from Climate Change in the United States. Science 356(6345):1362–1369. DOI: 10.1126/science.aal4369. https://pubmed.ncbi.nlm.nih.gov/28663496/
- Burke, Marshall, Solomon M. Hsiang, and Edward Miguel. 2015. Global Non-linear Effect of Temperature on Economic Production. Nature 527:235–239. DOI: 10.1038/nature15725. https://www.nature.com/articles/nature15725
- Kahn, Matthew E., et al. 2021. Long-term Macroeconomic Effects of Climate Change: A Cross-country Analysis. Energy Economics 104:105624. DOI: 10.1016/j.eneco.2021.105624. https://www.sciencedirect.com/science/article/pii/S0140988321004898
- European Insurance and Occupational Pensions Authority. 2026. Insurance Protection Gaps in a Changing Climate (April 16). https://www.eiopa.europa.eu/insurance-protection-gaps-changing-climate-2026-04-16_en
- National Oceanic and Atmospheric Administration. 2025. Notice of Retirement of Billion-Dollar Weather and Climate Disasters Product (May 8). https://www.nesdis.noaa.gov/news/billion-dollar-weather-and-climate-disasters
- Chantarat, Sommarat, Andrew G. Mude, Christopher B. Barrett, and Michael R. Carter. 2013. Designing Index-Based Livestock Insurance for Managing Asset Risk in Northern Kenya. Journal of Risk and Insurance 80(1):205–237. DOI: 10.1111/j.1539-6975.2012.01463.x. https://onlinelibrary.wiley.com/doi/10.1111/j.1539-6975.2012.01463.x
- United Nations Framework Convention on Climate Change. 2015. The Paris Agreement (adopted December 12, 2015; entered into force November 4, 2016). https://unfccc.int/process-and-meetings/the-paris-agreement
- IPCC. 2021. Climate Change 2021: The Physical Science Basis, Working Group I, Chapter 11: Weather and Climate Extreme Events in a Changing Climate. https://www.ipcc.ch/report/ar6/wg1/chapter/chapter-11/
- Carter, Michael, Alain de Janvry, Elisabeth Sadoulet, and Alexandros Sarris. 2017. Index Insurance for Developing Country Agriculture: A Reassessment. Annual Review of Resource Economics 9:421–438. DOI: 10.1146/annurev-resource-100516-053352. https://www.annualreviews.org/content/journals/10.1146/annurev-resource-100516-053352