RESEARCH DOSSIER
Global Catastrophe Insurance Markets and Government Risk Pool Penetration
Abstract. Catastrophe insurance markets organize the transfer, financing and settlement of unusually concentrated natural hazard losses across households, businesses, insurers, reinsurers, investors and public institutions. Government-supported risk pools participate in this system through statutory extensions of private contracts, public primary insurance, publicly supported reinsurance and regional sovereign risk transfer. Their penetration is measurable only after the relevant policy population, risk layer, institutional obligation and reporting denominator have been identified. This monograph develops a comparative institutional analysis of catastrophe protection supply, public-pool participation and insured losses, combining official market records, national legislation, program reporting, European supervisory documentation, international catastrophe loss estimates and published research. Selected evidence from France, Spain, the United Kingdom, Japan, Türkiye, New Zealand, Australia, the United States, the Caribbean and African and Pacific sovereign risk pooling illustrates different allocations of distribution, underwriting, retention and fiscal responsibility. A reproducible examination of aggregate loss ratios, Turkish compulsory earthquake insurance coverage and Japanese attachment and household policy ratios establishes the importance of measurement boundaries. The analysis explains how increases in public participation can reflect wider coverage, migration from incumbent insurers, changed reinsurance allocation, or deteriorating private capacity, and how observed insured losses combine exposure, severity, contractual coverage and measurement. Public risk bearing is therefore evaluated alongside the protection households and firms actually obtain and the residual exposure borne by fiscal authorities. The findings are an original analytical synthesis of documented institutions and established risk principles, with no claim of new empirical discovery or proprietary theory.
Keywords: Catastrophe insurance; natural hazards; public-private insurance; risk pools; government reinsurance; insurance penetration; protection gap; earthquake insurance; flood insurance; residual markets; sovereign catastrophe insurance; catastrophe risk financing; basis risk; risk layering; contingent fiscal liabilities; insurance markets.
Introduction
A catastrophe insurance market distributes the financial consequences of an event whose losses are concentrated in space, time and contractual portfolios. Its central institutional problem is the provision of credible claims-paying promises at a price and on terms that exposed people and enterprises can obtain. An earthquake can simultaneously damage thousands of insured buildings; a flood can affect property, business activity, municipal infrastructure and credit portfolios. A sequence of severe storms can exhaust annual aggregates without a single event approaching the maximum modeled disaster. The insurer's promise is consequently part of a wider system of financing: premium income, technical provisions, equity, reinsurance, financial-market instruments, statutory levies, and sometimes an explicit state guarantee.
Public participation is extensive in this field because the location of catastrophe losses influences both the willingness to purchase protection and the cost of maintaining capacity. Governments already bear fiscal exposure through infrastructure ownership, tax revenue disruption, social transfers and disaster recovery responsibilities. A public risk pool can formalize part of that exposure, enlarge the set of contractual protections, or facilitate private underwriting. Each intervention has a specific financial incidence. An institution funded through industry assessments reallocates resources within the insured population; one funded through general taxation redistributes across the wider economy. A reinsurance backstop can create a substantial contingent obligation while appearing modest in the primary premium statistics.
The international empirical problem is that the term insurance penetration often denotes written premium divided by gross domestic product, whereas disaster protection is frequently characterized by the fraction of economic loss insured. Public pools also report contracts ceded, participating insurers, insured residences, reinsurance limits or member governments. These are different observables. A household policy carrying the benefit of compulsory public reinsurance may count as privately distributed, publicly supported and partly privately retained at the same time. An analysis that forces these descriptions into an exclusive public/private market-share partition makes the institutional structure unintelligible.
The subject of this research is global catastrophe insurance market organization and the extent of public-pool participation, with distinct treatment of supply, market share and realized insured loss. Natural hazards are the central domain; catastrophe exposure associated with cyber conflict, infectious disease and other non-natural sources has distinct contractual and risk correlation problems and is outside the comparative case set. Sovereign insurance is included because public fiscal exposure is part of catastrophe finance, but its contractual unit is a government or qualifying public-sector entity, not a household. Disaster relief and contingent credit enter the discussion as complementary or competing instruments rather than as insurance premium.
The research asks how official institutions identify the supply of catastrophe cover; how ownership, legal obligations, cession and competition determine public participation; and how losses and protection gaps can be compared without mistaking different denominators for one phenomenon. The investigation proceeds by linking a documentary reconstruction with explicit measurement identities and transparent arithmetic. Its evidence supports descriptive and institutional inference. It does not estimate a global causal effect of government intervention.
Conceptual and Historical Foundations
The economic object of catastrophe insurance
A risk transfer contract exchanges a relatively certain present cost for a conditional future payment. The contractual payment depends on the insured interest, covered cause of loss, qualifying damage, period, deductible, limit, exclusions, aggregation clauses and applicable law. These attributes determine the portion of a catastrophe that enters the insurance system. The physical event is common to many contracts, but its insured consequences depend on heterogeneous property characteristics and terms.
Let X represent gross economic damage attributable to a defined covered interest and event. Under a simplified indemnity contract with deductible d, limit u and no further conditions, the payable amount is min(u, max(0, X − d)). The difference X minus that payment is retained by the economic owner, except where another contract or public assistance transfers it. The formula illustrates the effect of terms; it is not a representation of all insurance contracts. Percentage deductibles, replacement-cost clauses, coinsurance, eligibility rules, exclusions and repeated events alter the result.
An insurer's aggregate event loss is the sum of relevant claims before reinsurance. It depends on the joint distribution of those claims. If N exposures have identical variance v and common correlation r, the variance of their sum is N v[1 + (N−1)r]. With independent risks, aggregate variance grows linearly with N. Positive dependence causes it to grow more rapidly. Real catastrophe portfolios require spatial dependence, differing limits and nonlinear tail dependence rather than a single correlation parameter; the textbook expression is used here solely to illustrate why geographical accumulation matters. Diversification across hazards and territories is valuable to the extent that extreme losses do not coincide, while underwriting a large count of homes in a single high-hazard area does not mechanically remove catastrophe risk.
The commercial supply decision therefore reflects expected contractual loss plus the expenses and cost of capital required to maintain an enforceable promise over adverse outcomes. The insurer also faces model and parameter uncertainty, the risk of higher repair prices following a disaster and the practical capacity of claim adjustment. Correlated claims can generate a liquidity need long before all ultimate amounts are known. Public participation can offer financing at an additional layer without changing the primary relationship with the insured.
Formation of institutional solutions
The modern evidence illustrates a history of different policy responses rather than one global movement toward government insurance. France institutionalized the Cat Nat public-private structure in 1982, using existing property policies and publicly supported reinsurance (CCR, n.d., Cat Nat regime). The structure uses an established distribution channel to attach a specified category of natural hazard cover. In the Caribbean, a 2007 regional facility organized sovereign risk transfer; its 2014 restructuring provided segregated portfolios (CCRIF SPC, n.d., institutional history). The objective of the latter is rapid government liquidity rather than nationwide indemnity for all privately owned assets.
The difference in institutional origin persists in contemporary statistics. Where the intervention is built into property insurance, widespread ordinary property coverage can support broad catastrophe access. Where an intervention addresses fiscal liquidity, even universal participation of eligible states may coexist with sparse household property insurance. These arrangements supplement different positions in the loss-bearing chain.
Changes in the scope of hazards also matter. A property insurance market developed around fire may distribute cover differently for geological and hydrological risks. A scheme for earthquake risk cannot be presumed to cover every geophysical hazard or the flood consequences of any disaster. The appropriate definition follows the legal contract, not the generic classification of the event. Natural hazard categories include earthquake, tsunami, flood, tropical cyclone, severe convective storm, wildfire, drought and ground movement, each with variation in triggers and coverage.
The demand for insurance and public expectations
Risk aversion offers a reason to transfer large losses even where the premium exceeds the actuarial expected payment. Purchase nevertheless involves household liquidity, awareness, trust, loan requirements, expected assistance, and the distinction between the premium and total housing cost. A property owner who regards disaster as improbable during a short holding period may value financial protection differently from a lender with a long-term secured claim. Tenants and owners face different interests in buildings, contents and displacement. Businesses compare protection against direct damage and interruption with cash reserves and operating flexibility.
Wagner's (2022) flood insurance study uses variation associated with US premium reforms to examine demand and selection. Its published abstract reports a relationship between uptake and observable adaptation characteristics in the studied setting. That evidence cautions against treating all observed differences in purchase as private information about risk. The households with greater willingness to pay may have different risk-reduction investments, wealth, access to credit or knowledge. Extrapolation to other perils and countries requires data on these institutions.
Insurance also interacts with expected post-disaster government assistance. Where households expect unconditional compensation, the marginal benefit of paying premiums can fall. Where assistance is restricted or uncertain, insurance provides a legally specified claim more valuable than a discretionary budget allocation. The magnitude and direction of these effects depend on the eligibility of assistance, the credibility of government commitments, the timing of payments and the households' actual options. Expectations of relief are an empirical question rather than an assumed universal explanation of low participation.
Kousky's (2019) review of disaster insurance distinguishes evidence on recovery from evidence on investment in mitigation. The reported literature suggests a beneficial role in recovery while rigorous studies of preventive responses remain limited. For institutional evaluation this means that positive claims settlement, affordability and risk-reduction incentives require separate outcome measures. Their dependencies cannot be inferred merely from premium revenue or program enrollment.
The government as regulator, sponsor and risk bearer
The state can prescribe coverage terms, authorize a market institution, require industry participation, subsidize premiums, capitalize a pool, reinsure its obligations or compensate disaster victims after the event. These actions can occur in different legal entities. Public ownership is accordingly an insufficient criterion for public risk-bearing incidence. A private corporation can administer public coverage, while a public insurer may finance most of its tail risk through international reinsurance and private capital.
A public risk pool in this study is an organized, ex ante risk-financing arrangement established or supported by public authority that aggregates catastrophe exposures or provides predefined coverage or reinsurance to an identified participant population. This working definition encompasses multiple arrangements for comparison; it does not classify ordinary disaster relief as insurance. It requires specification of the liability, the applicable risk layer and the participants' obligations. A sovereign parametric arrangement satisfies this definition through a contingent policy promise even when the payment corresponds to modeled impact rather than assessed indemnity.
Public support may be justified where foreseeable private undersupply, affordability or coordination failures threaten recovery, economic continuity or fiscal stability. The rationale does not establish that any particular subsidy schedule is efficient. The appropriate question is whether the marginal benefits of the intervention exceed the fiscal costs, distortions and alternative uses of resources, with due attention to distribution. The OECD's (2026) catastrophic risk protection report provides a contemporary policy context for distinguishing protection problems from the choice of government-supported instrument.
Primary Institutional Documents and Comparative Research Design
Evidence and observation protocol
The study combines a structured comparative case analysis with published descriptive statistics. The selection is purposive: institutions are examined because they demonstrate distinct allocation mechanisms and varied types of measurable participation. The cases provide coverage of several major regions and of both highly developed and emerging insurance markets. They are not a random sample of countries, a census of national systems, or a statistically representative estimate of global public-pool share. Country-level laws and scheme documents are the principal evidence for coverage obligations; provider reporting establishes program-specific numbers; supervisory and international sources supply comparable context where their definitions permit it.
The temporal comparison uses observed 2024 and 2025 losses, dated historical country statistics and scheme arrangements documented through the research cut-off. A monetary figure retains its published year and denomination. A ratio built from an annual loss estimate is not combined with a fiscal-year policy count. Snapshot dates and reference periods are preserved even when source publication dates differ. Dynamic program websites are identified as undated institutional descriptions rather than assigned fictitious publication years.
The EIOPA (2025) natural catastrophe protection dashboard distinguishes modeled current exposure from historical loss and coverage data, with country and peril views. Its historical material spans 1980–2024. This demonstrates how estimates of future exposure and retrospective protection can be organized within a common interface without equating them. The present work examines the dashboard documentation, not the underlying country workbook; no unpublished table calculations from that workbook are claimed.
Evidence is evaluated according to relevance and institutional provenance. A statutory text is used for the legal obligation and parties to which it applies. An official program page can establish how the scheme describes its own operation, with the limitation that provider self-reporting is not independent evaluation. A regulator's monitoring summary offers a distinct perspective. Industry loss estimates have valuable cross-event breadth while remaining estimates subject to reporting methodology and revision. Journal abstracts support only the findings actually stated there; full article methods and replication datasets were not independently audited.
Four separate penetration measures
Primary market premium penetration is the ratio of written direct premiums to gross domestic product for the same jurisdiction and period. If catastrophe-specific premium is separately allocated, a corresponding disaster premium ratio can be computed. When bundled policies do not identify the catastrophe component, the ratio must be estimated through a disclosed allocation of premium; an entire multi-peril premium is an upper-bound proxy rather than an identified catastrophe premium.
Coverage take-up for a defined peril is the number of eligible exposure units that have qualifying coverage divided by the count of those eligible units at a specified date. The unit may be a household, dwelling, commercial establishment or other defined policy exposure. Policies and buildings are not inherently one-to-one. One residential association contract can insure multiple units; a household may have separate building and contents contracts. A count divided by a population denominator should therefore be described as a policy-count proxy unless the correspondence is verified.
Primary pool market share is the ratio of direct premium or direct covered exposure attributable to a public primary insurer to total direct premium or covered exposure for the same defined product market. A meaningful numerator and denominator must use the same accounting basis, time and currency. Gross written premiums can differ from earned premiums; net written premiums describe another level of retention. A pool that provides reinsurance instead of primary cover requires separate cession and risk-layer indicators. Ceded premium divided by direct premium is not the public insurer's direct market share.
Realized insured-loss coverage is aggregate insured loss divided by aggregate economic loss for matched events, loss classes, valuation date and geographic population. If economic loss E is positive and insured loss I is within the same loss perimeter, the uninsured gap is E−I and the uninsured proportion is 1−I/E. For multiple events, sum first and then divide; averaging individual shares weights small and large events equally and answers a different question. If insured and economic estimates use inconsistent definitions, the difference is only a diagnostic and must not be represented as exact uninsured damage.
Sovereign participation is calculated from a separate government universe: the number of active sovereign participants divided by the number eligible under the pool's institutional mandate. A second measure is total purchased limit compared with an explicitly estimated fiscal response need over a chosen risk horizon. The denominator for the latter is model-dependent and may concern emergency spending, public asset damage, recovery expenditures or revenue loss. Membership alone cannot describe such financial adequacy.
Data reconciliation and institutional classification
A comparative dataset needs one record per jurisdiction, peril, observation period and institutional layer. Key descriptors are legal program name, insurer or reinsurer identity, eligible lines, contract trigger, coverage cap, risk selection, funding source, compulsory or voluntary status at each level, carrier of the legal claims obligation, reinsurance attachment, measurement unit and reporting maturity. The population should be recorded in a way that permits reconstruction of each denominator.
| Observed field | Appropriate measurement | Principal hazard of interpretation |
|---|---|---|
| Direct premiums | Direct market premium share or premium-to-GDP | Double counts ceded premium if layers are added |
| Policies in force | Stock of active contracts at a stated date | Multiple contracts, building/contents overlap |
| Policy attachments | Fraction of qualifying contemporaneous contracts with rider | Conditional denominator, flow rather than stock |
| Ceded policies | Count of primary contracts whose specified risk is ceded | Same household may appear at both layers |
| Treaty limits | Contractual maximum risk transfer capacity | No direct equivalence to household limits or annual loss |
| Government guarantee | Legal residual undertaking and conditional valuation | Contingent exposure differs from current spending |
| Insured and economic loss | Ratio for identical event and loss perimeter | Estimate revisions and differing accounting boundaries |
| Sovereign members | Active qualifying participants | Membership differs from household coverage |
The Federal Emergency Management Agency (FEMA, n.d.) NFIP data documentation shows why transactional records require filtering by effective and termination dates and why renewals create new records. It also states that the released claims data cannot be attached to individual policies through a common identifier and do not give a payment-date series. These restrictions limit what a researcher can legitimately infer from publicly reported transactions. Counting all rows as insured households would overstate active policy exposure; grouping claims by occurrence year would not yield cash disbursements by calendar year.
For primary and reinsurance accounts the analogous rule is consolidation. A primary insurer's direct claim payment and a subsequent reimbursement from a reinsurer relate to the same underlying loss. The former satisfies the claimant's insurance right; the latter allocates financing between carriers. An adjusted picture of final risk-bearing requires netting and studying contractual recoveries. The combined primary direct claim and reinsurance recovery cannot be interpreted as additional aggregate economic damage indemnified.
A defensible public-sector aggregate must also identify whether the scheme has commercial premiums, compulsory levies, general budget transfers, donor grants or state credit. Public-sector resources differ in legal and economic incidence. Accounting for the entire gross premium of a privately administered policy as public financing would obscure the actual terms of the cession and each party's retention.
Catastrophe Protection Supply
Supply, solvency and the cost of retained risk
Private catastrophe insurance supply is determined by access to capital, the expected severity of insured claims, model confidence and the limits of diversification. A territorial risk can be offered at a higher price, at a reduced limit, with a larger deductible, with narrower wording, through a residual insurer, or not at all. Availability is therefore not binary. A quotation that excludes the main local peril provides limited catastrophe protection, while a policy with a high deductible transfers only part of the exposure.
Premium formation requires decomposition. For a simplified single-period contract portfolio, write P as expected loss plus acquisition and administration cost, reinsurance and capital cost, taxes and levies, and a margin allowing for operational and model uncertainty. Reinsurance premium is a gross financing cost and the expected recovery reduces retained expected loss; an actuarial pricing expression must avoid counting ceded expected losses twice. The relevant balance depends on whether the perspective is the primary carrier, the pool, a consolidated insurance sector or a government sponsor.
A pool can reduce costs by sharing model infrastructure, diversifying eligible exposures, accessing wider reinsurance markets, or receiving a public guarantee. It can also increase exposure to concentrated risks that incumbent insurers would otherwise limit. A subsidized price can be socially useful even when it does not cover the full economic cost of risk; the difference must be financed and made visible somewhere. A low tariff and a robust balance sheet are distinct attributes.
The ability to finance adverse years is evaluated using expected annual loss, the annual aggregate loss distribution, occurrence exceedance distributions, probable maximum losses at explicit return periods, and expected shortfall at specified confidence levels. A return period is an inverse modeled annual exceedance probability under a chosen probabilistic model; it does not promise a fixed recurrence interval. Nonstationarity, model revision and changes in exposure may invalidate a static interpretation. A severe event can also impair reinsurer recoverables or cause reinsurance prices to rise at renewal, changing the cost of future supply.
France and Spain: statutory links to private contracts
France illustrates a public-private system built around existing private insurance. In the consolidated French Insurance Code, Article L125-1 (France, 2026), relevant property damage contracts confer catastrophe cover under the statutory conditions. Where the policy already provides business interruption protection, the law also addresses its extension subject to the relevant contract. The legal classification of the disaster and qualifying damage remains consequential. An ordinary property policy count is thus a useful starting population, but the insured loss perimeter still depends on covered assets and terms.
CCR (n.d.) describes state-guaranteed catastrophe reinsurance available to participating insurers without an exclusive CCR monopoly. The same primary insured may benefit from a guarantee while its insurer retains part of the risk, and the reinsurer's state support enters at a later level. The primary insurer's retail premium is distinct from the reinsurance premium and the ultimate fiscal exposure.
The French nonmotor property surcharge rose from 12% to 20% of the relevant base damage premium effective January 1, 2025, according to CCR's (n.d.) surcharge history. The eight-percentage-point change represents a two-thirds relative increase in the specified surcharge rate. It is not a two-thirds increase in total household premiums: for an unchanged base premium of EUR 100, a EUR 12 surcharge becomes EUR 20, so the simplified sum increases from EUR 112 to EUR 120, approximately 7.14%. Taxes, other charges, underlying tariff changes and coverage modifications may alter the actual bill. The example separates an institutional financing reform from an observed market-wide price effect.
Spain distributes extraordinary-risk protection through a different legal arrangement. The consolidated Consorcio legal statute (Spain, 2004) specifies a compulsory surcharge for enumerated relevant insurance lines and conditions the Consorcio's indemnity obligations on the applicable coverage and solvency circumstances. Articles 7 and 8 also establish the relationship to the insured persons, objects and amounts under the underlying policy, subject to legal qualifications. This institutional structure makes the surcharge base and statutory indemnity perimeter essential indicators.
The two systems both attach broader catastrophe financing to existing insurance relationships, but their claims obligations and reinsurance arrangements differ. A comparison should identify in each case who indemnifies the claimant, whether a private carrier remains responsible, whether exceptional conditions activate the public institution, and who bears the final layer of loss. A national private gross premium total alone cannot resolve those differences.
United Kingdom and Australian reinsurance interventions
The UK government's description of Flood Re (n.d.) defines it as a joint public-industry reinsurance scheme seeking the availability and affordability of residential flood cover and transition toward risk-reflective pricing, with a planned termination in 2039. The intervention addresses a particular component of the retail contract. A household can purchase a policy from a private insurer while its flood exposure enters the scheme's reinsurance portfolio. Statistical assessment must identify the primary retail exposure, the cession and the retained risks independently.
The planned exit date matters economically. A transitional subsidy can preserve insurance relationships while risk mitigation and property adaptation improve the long-term affordability of risk-based premiums. If underlying exposure and vulnerability continue to rise, the cost of the transition increases. The relevant evaluation includes the availability of private rates at renewal, changed property-level vulnerability, household affordability and the dependence of underwriting on the reinsurance arrangement. An annual measure of cessions is informative about current use, not evidence by itself that the transition will be successful.
Australia provides a monitored example of a reinsurance intervention designed for a particular hazard. The ACCC's (2026) fifth insurance monitoring report identifies moderation of rate increases for medium- to high-cyclone-risk households and small businesses, a group representing roughly 2% of policies nationally in the stated summary. The regulator also describes continuing affordability concerns and some improvements in availability. Its findings distinguish the targeted population from the wider insurance market. The reported moderation does not identify a uniform national premium reduction, and the 2% is not the market share of the pool.
Catastrophe insurance that covers several hazards is subject to cost pressures outside a single intervention. The premium for an Australian combined property policy can reflect flood, storm, fire, theft, rebuilding inflation, expenses and capital as well as cyclone. An intervention that lowers the cost of one component may fail to reduce the full invoice if unrelated costs increase. Conversely, a national average can conceal significant benefits for a small highly exposed population. Estimating the direct effect requires geographically matched policy data and a counterfactual, not simply subtracting two national mean premiums.
New Zealand and Japan: public layers and private participation
New Zealand's Natural Hazards Commission (n.d.) specifies natural hazards cover associated with valid home fire insurance, generally capped at NZD 300,000 plus GST for the building per event, with limited coverage for designated land. Private insurance addresses covered building loss beyond the public cap subject to its own policy conditions. The structure differentiates eligibility, a primary statutory layer and supplementary private contractual risk transfer.
For a stylized residential building loss of NZD 500,000, ignoring GST, deductible effects and any eligibility adjustment, a first layer with a cap of NZD 300,000 leaves NZD 200,000 for the applicable excess cover or the property owner. This is a layer illustration only; actual entitlements depend on the insured sum, policy and statute. Counting the entire NZD 500,000 loss as either public or private would obscure the contractual allocation. Residential land requires its own valuation and limit, so it cannot be mechanically inserted into the building example.
The Commission's Natural Hazard Fund description records funding through levies, investment and risk transfer. From June 1, 2025 it reported NZD 10.3 billion of reinsurance protection including a NZD 225 million catastrophe bond, and it describes government financing of shortfalls in the form of a loan. The limit is a program-level financing quantity rather than the aggregate replacement value of New Zealand homes. A sovereign loan used to fulfill claims is initially financing; the eventual budget cost depends, among other things, on repayment and financing conditions. This illustrates why contingent guarantee exposure and permanent fiscal transfer require separate accounts.
Japan's Ministry of Finance (2026) describes a residential earthquake insurance system covering residential property and household goods, linked to fire insurance and supported by reinsurance involving the government. Its official outline shows a combined single-earthquake payout limit of JPY 12 trillion at the accessed version. The existence of a statutory allocation with substantial capacity does not determine household take-up, the value insured relative to replacement cost or the full consequences of losses beyond legal limits. The public role is best understood in conjunction with the institution's contract population and the terms of its risk financing.
Other forms of direct catastrophe supply
Türkiye's TCIP is a prominent compulsory earthquake insurance program. Its 2024 annual report documents the defined residential coverage denominator and policy volumes; these are evaluated separately in the market-share chapter. Compulsion affects both the pool's potential breadth and the enforcement tasks required of distribution institutions. A legal mandate becomes effective coverage through policy issuance, renewals, reliable building classifications and claims access. A lapse, inaccurate property record or inadequate sum insured can leave a material retained loss despite the mandate.
US flood insurance demonstrates the role of a national risk institution operating alongside private products and distribution. In this study the FEMA data guide is used for measurement rather than to infer nationwide public share. A precise US flood market share would require the NFIP numerator and comparable private standalone and embedded flood exposure or premium data at the same date. A private servicing agent's involvement is insufficient evidence that the corresponding risk has been transferred to a private balance sheet. Conversely, private flood contract sales must be recognized when defining the entire direct market.
These systems illustrate a general analytical distinction among creating a coverage obligation, providing a primary policy, supporting insurance distribution, and supplying capital at the reinsurance layer. Each can improve availability under particular conditions. A comparative result must specify which element of supply is observed, how broadly it reaches exposed assets, and the financial resources available in adverse conditions.
Insured Losses
Reading global loss estimates
International loss estimates measure the physical and financial consequences of documented disaster events within a reporting framework. The causal object is broader than the insured market: economic losses reflect damaged assets and other costs within the definition chosen by a database; insured losses reflect estimated claims under applicable policies. Data quality differs with the availability of official damage assessments, insurance reporting and event development. Estimates can change over time due to claims adjustment, inflation conversion, changes in event scope and delayed information.
Swiss Re Institute (2025), in sigma 1/2025, reports 2024 catastrophe loss estimates of USD 318 billion economic, USD 137 billion insured and USD 181 billion uninsured. Swiss Re Institute (2026), in sigma 1/2026, reports USD 107 billion insured losses for natural catastrophes in 2025. Its March 2026 report release gives USD 220 billion economic losses for 2025. The estimates are treated as published vintages rather than a harmonized fixed-price time series.
| Event year and vintage | Economic loss, USD bn | Insured loss, USD bn | Residual, USD bn | Insured share |
|---|---|---|---|---|
| 2024, 2025 reporting | 318 | 137 | 181 | 43.1% |
| 2025, 2026 reporting | 220 | 107 | 113 | 48.6% |
The insured proportions are 137/318 = 43.0818% and 107/220 = 48.6364%. The complementary fractions are approximately 56.9% and 51.4%. Rounding accounts for the difference between the latter calculation and published whole-number statements. The 5.55-percentage-point difference between these vintages is descriptive; the observation does not establish a structural improvement in global insurance uptake. The source series should be harmonized for constant prices, revisions and event perimeter before interpreting loss levels over time. In particular, a change in the locations, perils and composition of events can alter the loss share with no change in insurance contracts.
The 2025 forecast discussed in the 2025 publication is also distinguished from the observed 2025 result published the following year. It would be methodologically unsound to substitute a model trend or peak scenario for an actual loss vintage. A quiet year relative to the expected distribution can coexist with increasing long-run exposure.
Across OECD members, the OECD's (2026) summary reports 32% insured flood losses for 2000–2024 and identifies 24 of 38 OECD countries with less than half of natural hazard losses insured since 2000. The two statistics answer different questions: one is an aggregate flood-loss coverage ratio across a defined population and the other counts countries crossing an all-natural-hazard threshold. Neither describes the fraction of the global population covered by government pools. Countries with higher economic and asset values can dominate a worldwide loss-weighted ratio even where policy access in other regions is weak.
Loss decomposition and the exposure effect
For a fixed event catalog, aggregate economic damage can be represented as the sum over locations and exposed assets of asset value multiplied by an event-specific damage fraction. The insured portion applies participation, insured values, coverage definitions, deductibles and limits to the relevant assets. The relationship is nonlinear because policy caps and thresholds alter loss payments at different damage severities. Exposure can grow through construction and asset accumulation; vulnerability can change through adaptation, building codes and maintenance; hazard probabilities may shift with environmental conditions.
Insured loss therefore combines at least four forces: the value and geography of exposed assets, the physical intensity of the events, the vulnerability of those assets, and the contractual transfer of damage into claims. An increase in claims may result from broader risk transfer at unchanged physical damage, which can represent stronger financial protection. Alternatively, higher physical losses may overwhelm insurance despite unchanged policy counts. Written premium growth has analogous components, including higher prices for the same risk.
The issue can be demonstrated with a transparent hypothetical decomposition. In period A, economic loss is USD 100 million and insurance compensates 40%, producing USD 40 million insured loss. In period B, economic loss is USD 90 million and the compensated share is 60%, producing USD 54 million insured loss. Insured loss has risen 35% even though economic damage fell 10%; the uninsured residual fell from USD 60 million to USD 36 million. This result is purely arithmetic, not an empirical estimate. It illustrates why growing insured loss alone is an inadequate performance indicator for catastrophe markets.
A converse example arises when annual exposure rises but the event occurs in a lightly insured region. An aggregate gap can widen while nationwide enrollment increases. To infer a coverage effect one needs a matched panel of exposures and policy terms under comparable catastrophe conditions, or a catastrophe model with separately validated hazard, vulnerability and policy modules.
Reinsurance and fiscal risk in severe events
Consider a simplified indemnity portfolio with gross insured event losses L and a public excess-of-loss arrangement that pays min(U, max(0,L−A)), where A represents attachment and U the maximum recovery. Assume no other coverage layers or reinstatements. With A = USD 100 million and U = USD 300 million:
| Gross insured event loss, USD m | Public reinsurance recovery, USD m | Primary retained loss, USD m |
|---|---|---|
| 80 | 0 | 80 |
| 250 | 150 | 100 |
| 500 | 300 | 200 |
| 900 | 300 | 600 |
The public recovery adds no new underlying indemnity above gross insured claims. It moves funding between institutions. At USD 900 million of gross insured loss, the primary retention is USD 600 million, not USD 100 million, because the public layer has exhausted its limit. A scheme with an additional state guarantee or private retrocession would have a different allocation. The example shows why nominal limits, attachment, exhaustion and the identity of the final guarantor are needed to assess public risk penetration.
For a single event, an expected fiscal guarantee payment can be represented in stylized form as the expectation of the residual payable amount beyond available fund and contracted financing capacity, conditional on the guarantee terms. Actual fiscal cost also depends on whether assistance is repayable, financing rates, taxes, investment income and recovery from reinsurers. The difference between expected annual fiscal cost and severe-event contingent liability must be disclosed. A guarantee can have low observed annual spending and large adverse-state exposure. In a budget analysis the latter belongs in stress scenarios even if no payment has yet occurred.
Annual aggregate terms and event definitions complicate the example. Multiple events may exhaust a shared aggregate; reinstatement may require additional premium. A public pool can retain risks below private reinsurance attachment and above exhaustion. Correlation between the insurer's investment assets and the affected economy can reduce financing resources at the same time that liabilities grow. Counterparty concentration and collateral availability add further constraints.
The expected fiscal exposure cannot be estimated solely from annual claim history. Catastrophe models use a distribution of thousands of possible scenarios precisely because rare extremes are underrepresented in short samples. Risk financing should be examined under multiple hazard and capital scenarios, with sensitivity to inflation and changes in vulnerability. A single return-period loss estimate is not an exhaustive capital adequacy analysis.
Parametric payouts and basis risk
A parametric contract maps measured or modeled hazard impact to a payment. A typical capped payment function is min(U, c max(0,M−A)), where M is the contract index or modeled impact, c the chosen scaling fraction, A the attachment and U the limit. The function is illustrative; actual sovereign products may use more complex payout schedules and contractual calculation rules.
Suppose a sovereign purchases a policy with modeled-loss attachment USD 20 million, ceding fraction 50% and payout limit USD 40 million. If modeled impact is USD 60 million, the simplified payout is USD 20 million. If actual eligible response cost is USD 100 million, a substantial retained fiscal need remains. If modeled impact is USD 10 million while an actual disaster creates a USD 50 million funding need, the policy pays zero under this stylized contract. The mismatch between actual financing need and policy response is the relevant basis risk.
The reverse mismatch is possible: a payment may exceed an immediately measured response need. In that case the institution needs rules for appropriate use and accounting of funds. The value of parametric finance is the combination of timely disbursement and prearranged capacity, not the expectation of exact ex post indemnity. Comparing parametric payouts with estimates of insured property loss requires distinguishing the funded public spending from insured property claims and preventing double counting where funds support already insured assets.
A regional pool's diversity may reduce aggregate risk financing costs. The magnitude depends on joint exposure rather than member count. Two small islands in one cyclone path may suffer simultaneously. Drought can affect multiple countries over the same season. Portfolio construction must reflect meteorological dependence, varying fiscal exposures, and the terms and triggers of each certificate. A favorable diversification effect is therefore conditional on the risk structure and contract limits, not a universal discount.
Sensitivity and limitations of observed ratios
A loss ratio calculated from rounded data is not exact to four decimal places in economic substance. To illustrate sensitivity for 2025, hold the reported estimate of insured loss at USD 107 billion and vary economic loss by 10% around USD 220 billion. The implied insured share ranges from 107/242 = 44.2% to 107/198 = 54.0%. This is a deterministic perturbation, not a statistical confidence interval. It demonstrates how uncertainty in the economic denominator can materially affect the interpretation of a protection gap.
In a longitudinal analysis, common-currency conversion, local inflation, physical damage estimates, insurance settlement maturity, population changes and changes in peril attribution need explicit treatment. One measure of insured loss may be based on ultimate accrued claims while another reflects paid cash at a reporting date. Data at the level of individual policies and losses may be incomplete because of privacy or reporting constraints. A researcher should retain original published estimates and distinguish subsequent revisions from modeled transformations.
The historical insured-loss share may underrepresent uninsured economic consequences in locations with poor damage reporting. Insurance claims can be documented more precisely than losses outside the insurance system. Conversely, broad macroeconomic estimates can encompass costs that insurance policies do not ordinarily target. The gap should be decomposed into insurable property loss, uninsurable or deliberately retained categories, policy limitations and absent coverage where possible. The entire observed economic residual cannot automatically be represented as a commercially addressable market.
Integrated Scientific Synthesis and Future Research
Institutional interpretation of the comparative evidence
Across the examined arrangements, widening access to catastrophe cover and transferring final balance-sheet exposure are connected but distinct processes. The French structure combines statutory attachment with a public reinsurance mechanism. Spain links extraordinary-risk protection to a defined surcharge and statutory compensation rules. Japan and New Zealand combine household insurance relationships with specifically bounded cover and public financial participation. The UK and Australian cases target the reinsurance layer and affordability in identifiable segments. Türkiye documents the gap between compulsory design and realized residential participation. Sovereign pools arrange fiscal liquidity for governments under predefined triggers.
These comparisons answer the research questions in a qualified but substantive form. The supply of catastrophe insurance is documented through legal obligations, policy eligibility and risk-financing resources, not just premium totals. Public penetration depends on the institutional unit being measured: the share of direct policies or direct premiums, the share ceded, or the amount and probability of public assumption of risk. Insured loss is an event-weighted retrospective measure influenced by the composition of economic damage and contractual protection. Improving one quantity does not guarantee an improvement in the others, although the mechanisms interact.
An implication follows for the global market narrative. Where a scheme attaches cover to an existing contract, access depends heavily on the original retail market. A public insurer can create genuine financial inclusion but remains dependent on distribution and enforcement. A reinsurer can increase the capacity of private carriers and reduce tail exposure without acquiring a large retail premium base. A state guarantee can stabilize a program while making contingent fiscal risk substantial. Sovereign insurance can reduce post-disaster liquidity constraints while households remain underinsured. Each outcome belongs to a different evaluation account.
International reinsurance is potentially valuable because it redistributes concentrated local claims across a broader portfolio and investor base. Its cost and available capacity can change following severe disaster years. Public support can bridge a capacity shock, but where the public arrangement becomes the primary destination of severe risks the long-term financial consequences depend on premium adequacy, reserve accumulation, contractual cessions and claims performance. An evaluation must also distinguish insurance-market stability from total social resilience.
Financing, affordability and adaptation
Affordability is relational: a given premium is assessed against household income, housing costs, credit access, essential expenditure and the perceived benefits of coverage. A higher premium in a high-risk region may provide accurate information about expected costs while excluding households with constrained budgets. A policy intervention may reduce the financial burden through targeted assistance, subsidized reinsurance, cross-subsidy or mandatory broad risk sharing. The distributive incidence differs across these mechanisms.
A uniform premium or surcharge can simplify broad participation and express national solidarity. It can transfer funding from lower-risk to higher-risk locations within the eligible pool, depending on the true risk distribution and the underlying premium base. The incidence may be progressive or regressive depending on property values, incomes, housing tenure and tax sources. The analytic question is how the costs and benefits are distributed across risk and income groups, rather than whether the instrument is nominally public or private.
Risk-reflective pricing may encourage property protection or relocation where practical, but some households cannot relocate or afford adaptation. Premium support accompanied by capital investment in mitigation can address both affordability and underlying physical exposure, provided its distribution and effectiveness are evaluated. A distinction is necessary between lower expected damage from completed improvements and a reduction in charged premium due solely to financial transfers.
Evidence of insurance's contribution to recovery does not establish the size of mitigation incentives for every product. A comprehensive evaluation should observe property-level improvements, claims frequency and severity, take-up and continuity of insurance, and characteristics of recipients. A policy can benefit households while producing limited changes in measured hazard; such a program should be evaluated according to its purpose and costs.
Governance and accountability
Government-supported risk pools operate through legal and administrative mechanisms. Their credibility depends on clear eligibility, stable financing, transparent claim processes, supervision where applicable and defined responsibilities for extreme losses. The scheme administrator should disclose the financial relationship between premiums, public levies, reserves, recoverables, budget transfers, capital-market instruments and guarantees. A sovereign pool should disclose the difference between contractual limit and actual disbursement and identify the institutions responsible for using funds.
Public and private commitments require a common accounting language sufficient to reconcile aggregate risk bearing. A household policy may generate insurer claims, reserve changes, public reinsurance recoveries and debt financing. These flows should be traced through the funding chain rather than accumulated as separate societal losses. Reporting gross and net positions together allows readers to see both the extent of promised protection and the location of financial responsibility.
Recent policy discussions in Europe further demonstrate this division of roles. The joint EIOPA and ECB (2024) European catastrophe risk proposal presents complementary insurance and public disaster-risk elements. The proposal is an institutional option and is not counted as an operating European public pool. Its relevance lies in the explicit policy question of how to combine national structures, private coverage and common financial capacity.
Cross-border pooling adds governance concerns relating to the applicable legal seat, policyholders' rights, model versions, disputes, donor influence and the distribution of disaster funds. The unit of accountability is the actual insurer or reinsurer and the participating governments, not an undifferentiated continental balance sheet. International cooperation should be evaluated through contractual commitments, risk financing, transparency and performance after events.
Empirical research agenda
A future global dataset could measure primary public-market shares for a restricted set of comparable catastrophe products. The collection procedure should first establish national supervisory reporting categories, identify public primary insurers, distinguish commercial and statutory accounts, and map the relevant eligible property population. Reinsurance participation should be encoded separately with direct premium, ceded premium, limits and estimated retained losses. Sovereign data should include active participating entities, purchased layers, policy periods, donor premium support, payout records and independently estimated fiscal needs.
A second research direction is causal analysis of changes in market availability following pool introduction or altered cession terms. Suitable outcomes include policy counts, coverage of the relevant peril, real premium for standardized risk characteristics and renewal rates. Credible identification might exploit staggered implementation, threshold-based eligibility or well-documented tariff reforms, but only after evaluating anticipatory behavior, selection, parallel trends and concurrent reforms. Natural disasters affect demand and prices directly and must not be treated as an exogenous policy change without justification.
A comparison of highly exposed regions with less exposed ones can help identify whether a targeted reinsurance scheme changes availability. Geography also confounds the analysis through differing hazard probability, rebuilding cost, income, insurer capital and underwriting appetite. A suitable method would compare similar policies and property characteristics around the relevant implementation period, test alternative control groups, and report sensitivity to hazard and market cycles. The present documentary analysis establishes the need and potential design, not the resulting causal estimate.
A third research direction is fiscal-risk assessment. An explicit government guarantee can be valued under alternative hazard catalogs, exposure trends, risk reduction scenarios and market costs of reinsurance. Where government support is a loan, the expected fiscal loss should include repayment risk and financing cost rather than treating the gross claim payment as a permanent transfer. Where governments provide donor-backed premiums, a sustainable funding analysis should evaluate renewal without the subsidy. These calculations require program-specific contracts and balance-sheet data not assembled in the current study.
A fourth direction is the distribution of actual disaster recovery benefits. Researchers can compare claim settlement timing, restoration of housing and businesses, access to credit, household consumption and displacement. Sovereign liquidity should be traced through budget execution to the covered emergency activities and intended population. Distinguishing payouts, expenditures and realized recovery prevents conflation of the availability of money with its effectiveness.
Finally, changes in future hazards and exposure require consistent measures of modeled protection. A historical 25-year loss ratio is descriptive of the events in that period. A model that evaluates a future reference portfolio must define its assumptions, covered terms, adaptation state, exposure vintage and uncertainty. This is a practical research and data task, not an assertion that a single global forecast can be constructed from the available ratios.
Evidentiary boundaries
The reviewed materials support the institutional distinctions and dated descriptive calculations in this study. They do not supply a harmonized comprehensive worldwide census of direct public catastrophe premiums, policy counts and retained-risk shares. No global public-pool share estimate is therefore reported. Some primary source PDFs were accessible through extracted tables and text, but attempts to obtain page screenshots failed; the numeric citations are limited to the readable tables and checked definitions. The British scheme's official government description is used rather than inaccessible detailed provider pages. Journal findings are limited to the publisher abstracts examined.
No proprietary catastrophe model, confidential insurer portfolio, independent household-level dataset or full sovereign liability inventory was employed. The numerical scenarios identify accounting and contractual mechanisms under stated simplifying assumptions. They are separate from observed historical records. Policy language and legal descriptions are restricted to the cited dates and scope; local amendments and individual contracts require separate examination for any application beyond comparative scholarship.
The original contribution is the integrated interpretation of market supply, contractual participation and ultimate risk allocation across the stated evidence, including explicit treatment of denominator and financial-layer mismatches. It neither asserts statistical representativeness nor introduces a newly named theory. Its value lies in a reproducible account of the conditions under which public-pool measures can and cannot be compared, supported by institutional examples and transparent calculations.
Conclusion
Global catastrophe insurance markets are institutional systems that convert some natural hazard damage into financed contractual liabilities. Their ability to do so depends on the existence of eligible cover, the participation of exposed households and firms, the magnitude of insured limits, access to reinsurance and capital, and the financial credibility of public undertakings.
Public risk pools fulfill several distinct functions within these systems. Compulsory attachments can extend the peril scope of existing coverage; primary pools can issue or assume defined contracts; public reinsurance can finance particular layers while preserving private distribution; sovereign pools can prearrange emergency liquidity. These functions must be measured according to the relevant legal and financial unit. A policyholder, a reinsurance certificate, a retained claim amount, a sovereign limit and a premium-to-GDP ratio are different research objects.
The reviewed country statistics show the empirical consequences of this distinction. Türkiye's 56% reported compulsory earthquake insurance coverage in 2024 indicates substantial participation together with a material residual gap relative to the reported population. Japan's fiscal-year attachment ratio of 70.4% and household policy ratio of 35.4% establish that two legitimate penetration statistics can diverge markedly because they measure different contract populations. The comparison is informative precisely when the denominators are preserved. The global insured-loss estimates similarly distinguish a high-loss year from a higher insured proportion in a different year and event mix.
A credible comparative assessment consequently requires data at both the contract and risk-bearing layers. It should report the protected and eligible populations, primary and ceded premiums, policy limits, distribution of loss, public funding commitments and the allocation of residual exposure. Affordability, capacity and preventive investment are jointly significant; evidence for improvement in one dimension cannot stand in for the entire market.
The most consequential institutional implication is that public-pool penetration should be interpreted as a description of the position public arrangements occupy within the financing chain. The social purpose of catastrophe insurance is realized through the actual provision of enforceable and adequately financed protection and through its contribution to recovery. Measuring that purpose demands both market statistics and evidence of what protection delivers after a disaster. The limits of existing global data delimit the results here and establish a concrete agenda for more precise international research.
References
African Risk Capacity Group. (n.d.). How ARC Works. https://www.arc.int/how-arc-works
Australian Competition and Consumer Commission (ACCC). (2026). ACCC Insurance Monitoring Report 2026. June 25, 2026. https://www.accc.gov.au/about-us/publications/serial-publications/insurance-monitoring-reports/accc-insurance-monitoring-report-2026
Caisse Centrale de Réassurance (CCR). (n.d.). Garantie Catastrophes Naturelles. Institutional description of the Cat Nat regime. https://www.ccr.fr/portail-catastrophes-naturelles/garantie/
Caisse Centrale de Réassurance (CCR). (n.d.). Réassurance publique: Catastrophes naturelles. Institutional description and surcharge history. https://www.ccr.fr/decouvrir-ccr/activites/catastrophes-naturelles/
CCRIF SPC. (n.d.). About Us. Caribbean Catastrophe Risk Insurance Facility. https://www.ccrif.org/about-us
EIOPA. (2025). Dashboard on Insurance Protection Gap for Natural Catastrophes. Updated December 5, 2025. https://www.eiopa.europa.eu/tools-and-data/dashboard-insurance-protection-gap-natural-catastrophes_en
EIOPA and European Central Bank. (2024). EIOPA and ECB Propose European Approach to Reduce Economic Impact of Natural Catastrophes. December 18, 2024. https://www.eiopa.europa.eu/eiopa-and-ecb-propose-european-approach-reduce-economic-impact-natural-catastrophes-2024-12-18_en
Federal Emergency Management Agency (FEMA). (n.d.). FAQs About NFIP Data. FloodSmart for Agents. https://agents.floodsmart.gov/flood-maps-and-data/faqs-about-nfip-data
France. (2026). Code des assurances, chapter V, articles L125-1 to L125-7, consolidated version dated June 19, 2026. Légifrance. https://www.legifrance.gouv.fr/codes/section_lc/LEGITEXT000006073984/LEGISCTA000006157250/2026-06-19
General Insurance Rating Organization of Japan (GIROJ). (2025). Fiscal 2024 Earthquake Insurance Attachment Ratio. News release, August 26, 2025 (Japanese). https://www.giroj.or.jp/news/2025/20250826.html
General Insurance Rating Organization of Japan (GIROJ). (2025). Statistics Compiled by General Insurance Rating Organization of Japan, Fiscal 2024, Part 4: Earthquake Insurance, Table 7, p. 4-48. https://www.giroj.or.jp/publication/statistics/statistics_2024_4.pdf
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
Ministry of Finance, Japan. (2026). Outline of Japan’s Earthquake Insurance System. Ministry of Finance, Japan. https://www.mof.go.jp/english/policy/financial_system/earthquake_insurance/outline_of_earthquake_insurance.html
Natural Hazards Commission Toka Tū Ake. (n.d.). About Natural Hazards Cover. New Zealand. https://www.naturalhazards.govt.nz/insurance-and-claims/about-nhcover/
Natural Hazards Commission Toka Tū Ake. (n.d.). Natural Hazard Fund. New Zealand. https://www.naturalhazards.govt.nz/about-nhc/how-we-work/natural-hazard-fund/
OECD. (2026). Financial Protection Against Catastrophic Risks: Floods, Fires and Other Major Risks. OECD Publishing. DOI: 10.1787/32a3acd7-en. https://www.oecd.org/en/publications/financial-protection-against-catastrophic-risks_32a3acd7-en.html
Spain. (2004). Real Decreto Legislativo 7/2004, de 29 de octubre, texto refundido del Estatuto Legal del Consorcio de Compensación de Seguros. Boletín Oficial del Estado, consolidated statutory text. https://www.boe.es/eli/es/rdlg/2004/10/29/7
Swiss Re Institute. (2025). sigma 1/2025: Natural Catastrophes: Insured Losses on Trend to USD 145 Billion in 2025. April 29, 2025. https://www.swissre.com/institute/research/sigma-research/sigma-2025-01-natural-catastrophes-trend.html
Swiss Re Institute. (2026). sigma 1/2026: Natural Catastrophes in 2025: The Persistent Rise of Wildfire and Storm Risk. March 2026. https://www.swissre.com/institute/research/sigma-research/sigma-2026-01-natcat-2025-wildfire-storm-risk.html
Swiss Re Institute. (2026). Wildfires, Storms, Floods Contribute to Record 92% of Global Insured Losses in 2025. March 19, 2026. https://www.swissre.com/press-release/Wildfires-storms-floods-contribute-to-record-92-of-global-insured-losses-in-2025-says-Swiss-Re-Institute/7b39b1a5-b878-4a55-a5ff-bf5aa561a675
Turkish Catastrophe Insurance Pool (TCIP). (2025). TCIP Annual Report 2024, sections 2.1–2.3, pp. 28–29. https://www.dask.gov.tr/upload/TCIP%202024%20Annual%20Report.pdf
UK Government. (n.d.). About Us: Flood Re. GOV.UK. https://www.gov.uk/government/organisations/flood-re/about
Wagner, Katherine R. H.. (2022). Adaptation and Adverse Selection in Markets for Natural Disaster Insurance. American Economic Journal: Economic Policy, 14(3), 380–421. DOI: 10.1257/pol.20200378. https://www.aeaweb.org/articles?id=10.1257/pol.20200378
World Bank. (2017). Sovereign Catastrophe Risk Pools: A World Bank Technical Contribution to the G20. Executive summary, pp. 3–5. https://documents.worldbank.org/curated/en/603121502870773583/pdf/118676-WP-v1-PUBLIC.pdf
World Bank. (2025). Strengthening Disaster Resilience and Response in Latin America and the Caribbean. June 30, 2025. https://www.worldbank.org/en/results/2025/06/30/strengthening-disaster-resilience-and-response-in-lac