RESEARCH DOSSIER
Insurability and Its Boundaries
Abstract. Insurability is the conjunction of circumstances under which a contingent loss can be transformed into an enforceable, financially credible and mutually acceptable promise of compensation. Its boundaries are determined by loss frequency and severity, statistical information, dependence among exposures, incentives and selection, available capital, willingness and ability to pay, contractual design, governing law and public institutions. This monograph develops a comparative synthesis of classical insurance economics, actuarial aggregation, prudential supervision and evidence from catastrophe, climate, cyber and public-private protection arrangements. It distinguishes the technical feasibility of measuring a risk, the financial capacity to absorb its tail, the market feasibility of offering coverage, and the legal and social permissibility of transferring the loss. Explicit Bernoulli-pool calculations demonstrate why the diversification benefit of a growing pool can disappear under common shocks. Simple premium-feasibility inequalities and adverse-selection examples locate different failure mechanisms without claiming a new universal test. The analysis shows that insurability is conditional on coverage terms, portfolio structure, time, jurisdiction, institutional resources and the distribution of costs. Reinsurance, capital markets, preventive investment, contract limits and public support can enlarge specific feasible domains; each changes the allocation of residual losses and may create new incentives or fiscal obligations. A risk deemed uninsurable in one transaction may support narrower, layered or publicly coordinated protection elsewhere. The central conclusion is that the operative question concerns which losses, amounts, triggers, parties and time horizons can be financed credibly, and on what distributive terms. The synthesis provides research-ready distinctions rather than a newly instituted theory of insurability.
Keywords: Insurability; insurance economics; risk pooling; risk dependence; catastrophe risk; risk uncertainty; actuarial modeling; underwriting; adverse selection; moral hazard; capital adequacy; reinsurance; insurance affordability; climate risk; cyber risk; public-private insurance; prudential regulation; insurance contract law; insurance protection gap; systemic risk.
Introduction
The existence of a loss does not establish the existence of an insurance market for it. A firm may calculate expected damage from a flood and still be unable to find an insurer willing to promise compensation for the entire exposure. An insurer may offer technically sound coverage at a premium that the exposed household cannot afford. A jurisdiction may allow an insurance contract covering some liabilities while prohibiting the indemnification of particular sanctions. Each circumstance raises a different question about what is meant by insurability.
The classic inquiry associated with Berliner (1982) examines the criteria and dimensions of insurability. In the later discussion of large risks and their limits, Berliner (1985) explicitly treats judgments about risk size and insurability as partly relative to the risk carrier. Current policy analysis similarly distinguishes actuarial, market and societal conditions; the OECD's 2026 comparative assessment provides a timely cross-country reference. These strands warrant an examination that moves from a transaction to the insurance portfolio, and from private underwriting to institutional capacity.
The subject of this dossier is the general boundary of insurance as a mechanism of contingent financial protection. It covers private primary insurance, reinsurance and relevant public-private arrangements, including cases in which a government retains or shares losses. Particular insurance lines serve as examples for testing general propositions; detailed line-specific policy design belongs in the corresponding Science research. Ten approved Science relationships accompany this single canonical study: Foundations of Insurance, Risk Theory, Underwriting, Catastrophe Risk, Cyber Insurance, Climate Risk and Insurance, Systemic and Emerging Risks, Insurance Economics, Property Insurance, and Casualty and Liability Insurance.
The research proceeds through comparative classification, deductive examples, source-based institutional comparison and explicit analysis of causal channels. The mathematical examples are constructed by the author and use stated assumptions; they do not estimate the probability of any actual disaster. Institutional descriptions rely on published rules or the organizations' own accounts, and empirical claims are attributed to particular studies. Jurisdictional illustrations describe the sources and time frames stated, not universal legal rules. Every historical, legal and statistical proposition is interpreted in light of its original source and context.
The guiding question is precise: under what conditions can a specified financial loss be transferred by a specified contract to a specified risk carrier over a specified horizon, while maintaining a credible capacity to pay and terms acceptable to the contracting parties and legal order? This question places the contingent promise, rather than the bare hazard label, at the center of analysis.
Criteria and Traditions of Insurability
Concept, object, and levels of analysis
An insurance transaction promises a payment, service or indemnity conditional on a contractually defined event. The hazard may be fire, illness, business interruption, a third-party claim or a network outage; the insurable object is the contractually delimited economic consequence. The amount insured, deductibles, waiting periods, exclusions, triggers and period of coverage determine the extent of the transferred exposure. The same event can therefore be insurable up to a specified limit and unattractive or infeasible above it.
At the risk level, analysis concerns whether relevant loss events can be described and measured. At the portfolio level, it concerns dependence, diversification and concentration. At the insurer level, it concerns underwriting competence, capital, assets, liquidity and claims administration. At the market level, it concerns premiums, demand, competition and alternative capacity. At the legal and social level, it concerns enforceability, public policy, distributive consequences and the eligibility of losses for indemnification. These levels interact but cannot be substituted for one another. A well-specified risk distribution does not itself supply capital; available capital does not itself establish consumer willingness to buy; and market demand does not legalize a prohibited contract.
A useful working definition follows: a loss is practically insurable for a given arrangement when it admits a sufficiently explicit coverage specification, defensible risk assessment, workable incentive structure, adequate loss-absorbing resources, an acceptable financing price and lawful enforcement. Each component refers to a decision and an institutional setting. The definition is an analytical summary of recognized conditions, not an exclusive legal definition or a new scientific doctrine.
Intellectual lineage and competing criteria
Berliner's Limits of Insurability of Risks (1982) organized criteria including randomness, loss magnitude, information, premium and legal considerations. Berliner (1985) emphasized that the position of the risk carrier matters: a regional insurer may find a concentration beyond its capacity while a globally diversified reinsurer can accept a portion of it. This historical insight cautions against attaching a permanent binary label to a hazard without stating the carrier, limits and portfolio.
Demand for coverage has a separate intellectual genealogy. Mossin's (1968) analysis of rational insurance purchasing locates purchase decisions in individual risk preferences and prices. Akerlof's (1970) quality-uncertainty analysis established a general mechanism whereby privately held information can change the market composition. Rothschild and Stiglitz (1976) show in their competitive insurance-market analysis how hidden risk differences can affect contract menus and even the existence of a separating equilibrium in particular model settings. These works explain why actuarial pricing alone is an incomplete market description.
Ehrlich and Becker (1972) place market insurance alongside decisions to reduce loss magnitude or prevent loss events in their self-insurance and self-protection analysis. This distinction remains essential when evaluating catastrophe mitigation, cyber controls and building resilience. The empirical analysis of cyber losses by Biener, Eling and Wirfs (2015), based on 994 recorded cases, later applied classical insurability criteria to a newer peril marked by data limitations and dependence. The methodological continuity across these works is stronger than any claim that a single criterion can settle every modern insurance decision.
The OECD's government-protection framework (2026) explicitly groups insurability tests under actuarial, market and societal headings. That classification is especially useful for comparative work because it prevents a shortage of affordable policies from being read automatically as a statistical impossibility. It also permits different jurisdictions to reach different outcomes through law or policy while facing similar underlying physical hazards.
Losses, contracts, and the scope of a promise
Insurance covers defined losses and obligations rather than an unlimited category of misfortune. A property policy may exclude flood damage, a cyber policy may impose a waiting period, and a liability contract may distinguish defense costs from fines. The contract determines the transfer. If the insurer limits payment to the first layer of loss, the high-severity residual remains with the insured or another counterparty. If an event trigger is objectively observable but only imperfectly aligned with the insured's actual damage, a parametric arrangement exchanges claims-adjustment speed for basis risk.
This contract dependence is clearest in long-tail liability. The ultimate amount may depend on judicial decisions, changes in litigation behavior, inflation and settlement practice. The insurer can offer narrower periods, limits and exclusions to control uncertainty, while the buyer may prefer comprehensive protection. Contractual precision creates operational insurability only when the resulting promise remains commercially meaningful and legally effective. An exclusion so broad that it removes most of the material hazard may produce a technically manageable portfolio at the expense of the protective function.
The time horizon is also substantive. A one-year policy permits repricing or withdrawal when information changes, whereas a long-term fixed-price promise embeds uncertainty about future rates, technology, disease, legal standards and climate. Insurability evaluated for a one-year contract must not be assumed to establish feasibility for an indefinite guarantee. The difference matters for sustainability, consumer expectations and the stability of protection over time.
Classification of limiting conditions
The recurring constraints can be analyzed under six connected headings. Probability and severity describe what may happen and how much could be lost. Dependence determines whether a portfolio can diversify losses or is vulnerable to a common cause. Information and incentives determine how coverage changes who enters the pool and how they behave. Capital and operational capacity determine whether claims can be honored under stress. Price and affordability determine whether adequate protection can actually be transacted. Law and societal choices determine which transfers are permitted, required, subsidized or restricted. None is sufficient in isolation.
The conditions also vary in reversibility. Better data can reduce parameter uncertainty, but measurement alone cannot eliminate a floodplain. Reinsurance can move a tail between balance sheets, but it does not erase the aggregate loss to the economy. A subsidy can place an otherwise unaffordable policy within a household budget, but the financing burden appears elsewhere. A prohibited indemnification may remain outside the available domain regardless of risk modeling. Treating these mechanisms separately provides a basis for identifying remedies matched to the true constraint.
Probability, Severity and Uncertainty
Insurable randomness and predictable losses
Insurance is particularly valuable when financially significant outcomes are uncertain for individual participants and can be pooled at tolerable cost. The uncertainty need not mean that the event has no scientific explanation. Fire, illness and storms arise through causal processes; contractual loss remains contingent because occurrence, location, timing or severity is uncertain to the policyholder and insurer. What matters for pricing is an economically relevant distribution, not metaphysical indeterminacy.
A highly predictable recurring expense presents a different financial function. If an insured person expects to incur a fixed annual cost with near certainty, the coverage resembles prepaid financing plus administrative expenses. Insurance may still be embedded in a broad social or private benefit system for solidarity, billing or liquidity purposes, but its pure risk-transfer advantage is comparatively limited. At the other extreme, an event too rare to estimate from experience may support coverage using engineering models, exposure data, scenarios and explicit uncertainty loads, provided the carrier accepts their limits.
The distinction between individual risk and collective regularity is crucial. A population may generate a fairly stable expected annual claims total while each person faces uncertain timing and loss magnitude. This is the economic setting in which pooled protection can smooth individual outcomes. Conversely, a stable average of claims across years is weak reassurance if an extreme event can cause liabilities exceeding the insurer's resources. The distribution of aggregate losses, including its tail and dependence, is indispensable.
Estimation, model uncertainty, and nonstationarity
Expected loss may be written as E[L] when a probability model is specified. A claim that E[L] is measurable is conditional on the event definition, exposure data, chosen distribution, calibration and period. Sparse historical records, changes in building location or value, exposure migration and evolving digital systems can all change the relevance of past frequencies. Statistical error includes sampling uncertainty; model uncertainty concerns the adequacy of the functional form, dependence assumptions and omitted mechanisms.
Charpentier's (2008) study of climate-risk insurability emphasizes both geographical correlation and changing environmental conditions. The IPCC's (2022) assessment of impacts and adaptation limits establishes the wider scientific context in which changing physical exposures must be considered. Climate change can affect hazard intensity, exposure values, vulnerability and the relationships between loss drivers; an actuary must distinguish these rather than assign every loss trend to the same cause.
Nonstationarity does not automatically make a risk uninsurable. A carrier may shorten its underwriting horizon, revise exposure models, impose coverage limits, improve loss-prevention conditions or increase capital. Each response changes what is promised and at what price. Model updates require governance: uncertainty should be recorded, alternative scenarios stress-tested and material deviations monitored. The IAIS Insurance Core Principles describe the supervisory importance of ongoing integrated risk identification and management, not a guarantee that every model will forecast extreme losses accurately.
Tail severity, bounded contracts, and the limits of moments
Severity often determines whether an apparently modest-frequency hazard is financeable. A small probability multiplied by an enormous potential loss can produce a large expected payment; even a modest expected payment can coexist with a severe tail that threatens insolvency. Expected loss is an average, while a contractual promise must survive plausible deviations from that average. Reinsurance, capital and liquidity are required because realized claims arrive in finite, sometimes concentrated periods.
A stylized loss distribution illustrates the distinction. Suppose positive losses have a Pareto tail P(L>x)=(x_m/x)^a for x at least x_m. The mean exists only when a>1; the variance exists only when a>2. For 1<a<=2, actuarial pricing can use a finite mean but standard variance-based capital formulas are invalid without truncation or another approach. For a<=1, the uncapped idealized distribution has no finite mean. Actual insurance contracts generally place monetary bounds on payment, and a bound changes the mathematical distribution of the insurer's liability. This is a conceptual demonstration; it is not a claim that observed catastrophe losses literally follow one universal Pareto law.
Policy limits, annual aggregates and attachment points can contain exposures whose gross loss distribution is extremely uncertain. These tools allocate the excluded layer or tail elsewhere. Lowering the promised limit may make an insurer's layer financeable without making the underlying catastrophic damage affordable to society. It is therefore important to report the covered proportion and residual protection gap alongside any statement that a peril is insurable.
A basic loss model and its interpretation
For an exposure with occurrence indicator I taking value one with probability p and a fixed covered severity s when it occurs, define L=sI. Its mean is E[L]=ps and variance is Var(L)=s²p(1-p). The model assumes that severity and probability are known and fixed, the trigger has an unambiguous definition and no unmodeled event produces a different amount. Its purpose is to isolate the mechanics of loss transfer.
For p=0.01 and s=100. An insurance premium of exactly 100 is therefore not a defensible assertion about the market premium. If the event becomes more likely, expected cost changes; if severity becomes more concentrated or limits change, the covered loss changes. These are separate elements of an underwriting decision.
The model also clarifies the relationship between prediction and protection. Accurate estimation of ps improves the expected-cost component; it does not secure a claims-paying institution or assure that the buyer values coverage at the resulting premium. More detailed loss assessment can reduce some uncertainty while the principal boundary is still set by dependence or affordability.
Dependence and Portfolio Capacity
Independence and the diversification calculation
Consider N exposures, each with loss L_i=sI_i, the same p and s, and initially independent occurrence indicators. Aggregate covered loss is S=ΣL_i. Then E[S]=Nps, Var(S)=Ns²p(1-p), and the coefficient of variation is
[ CV(S)=\frac{\sqrt{\operatorname{Var}(S)}}{E[S]}=\sqrt{\frac{1-p}{Np}}. ]
The relative volatility of the aggregate falls as the pool grows. For N=10,000, p=0.01 and s=1,000,000 and the standard deviation is approximately $99,499. The relative standard deviation is approximately 9.95%. This illustrates the diversification mechanism under independence, not the solvency capital a real insurer would hold.
Increasing the count of insured objects can reduce relative fluctuation when the exposures are sufficiently distinct. Yet the improvement is conditional on dependence, identical distribution assumptions, exposure limits and changing loss frequencies. A pool of houses on one vulnerable coastline is not equivalent to a diversified population merely because it includes thousands of contracts.
A common-shock counterexample
Suppose the occurrence indicators have a common pairwise correlation ρ, with equal individual means and variances. Under this equicorrelation assumption,
[ \operatorname{Var}(S)=Ns^2p(1-p)[1+(N-1)\rho], ]
so
[ CV(S)=\sqrt{\frac{1-p}{Np}[1+(N-1)\rho]}. ]
For positive fixed ρ, the large-N limit of relative dispersion is √[ρ(1-p)/p], rather than zero. With the preceding N=10,000, p=0.01 and an illustrative feasible positive correlation of ρ=0.02, the aggregate standard deviation rises to approximately 1 million. Correlation amplifies claims volatility by a factor of approximately 14.18 relative to the independent pool. The numerical case can be realized using a shared random probability environment: conditional losses may be independent while unconditional probabilities move together. The example isolates a general mathematical limitation of diversification and does not infer an observed correlation for any actual peril.
This result is material to earthquakes, geographically concentrated storms, pandemics and cyber disruptions shared by many insured systems. Their causal mechanisms differ. A storm produces geographic and infrastructure dependence; a digital service outage may propagate through common suppliers or software; a pandemic may correlate health and economic disruption. Portfolio modeling must identify the relevant connection rather than assume all dependence can be represented by a single stable coefficient.
Capital, liquidity and claims-paying credibility
An insurer accepts losses that materialize after premiums have been collected. It must maintain resources for expected payments and adverse outcomes, including the possibility that several lines deteriorate together. Economic capital analysis examines stressed distributions and dependencies; liquidity analysis asks whether payments can be funded when due. A solvent balance sheet valued over a long horizon may still face operational liquidity stress during a sudden concentration of claims.
A simple, deliberately conservative probability inequality demonstrates the difference between mean funding and buffer funding. If S has finite variance σ², Cantelli's inequality states P(S−E[S]≥k)≤σ²/(σ²+k²) for k>0. Consequently, a buffer k at least σ√[(1−α)/α] is sufficient under this inequality to bound the one-sided exceedance probability by α. It is a distribution-free mathematical bound, not a recommended capital formula, supervisory standard or efficient pricing method. It shows that larger aggregate volatility increases the buffer required for a fixed risk tolerance even with unchanged expected claims.
In the European Union, Solvency II's risk-based regime combines quantitative solvency requirements, governance and risk management, and disclosure. The consolidated Solvency II Directive identifies categories of risk for the Solvency Capital Requirement, including underwriting, market, credit and operational risks. This legal reference is a dated 2024 consolidation and is used to describe the established framework rather than assert that no later amendments exist. The broader IAIS supervisory framework sets internationally relevant principles while implementation remains jurisdiction-specific.
Premium adequacy, reserve adequacy, capitalization, investment risk and governance therefore jointly determine whether an offer can be made credibly. Charging more for expected claims without addressing extreme accumulation may conceal the central financial obstacle. Conversely, withholding every exposure with a model imperfection would sacrifice protection where diversified, capped contracts can remain credible.
Reinsurance, securitization, and concentration
Reinsurance permits an insurer to cede portions of its liability. A proportional treaty shares covered losses; a nonproportional layer responds above a contractually defined attachment point and up to a limit. Reinsurance can diversify risk across markets and use the capital of specialized carriers, but it introduces counterparty and contract basis risk. A catastrophe bond may transfer a defined event layer to capital-market investors; trigger choice, basis risk, funding structure and pricing affect the realized protection.
Froot's (2001) examination of catastrophe-risk markets finds that high premiums and limited cover against extreme catastrophes cannot be understood from risk aversion alone; supply restrictions and market structure matter. The study evaluates transactions linking catastrophe risk to capital markets and provides a basis for asking why sufficient physical or financial capacity does not automatically become affordable coverage. Cummins (2006) in his analysis of government catastrophe insurance argues for attention to market incentives and potentially crowding out private capacity, illustrating a distinctive institutional interpretation.
Transfer across carriers does not remove a systemic aggregate loss. It can distribute the loss across capital providers, locations and time, sometimes reducing the probability that one institution fails. When the underlying shock affects investment markets and insured assets simultaneously, supposed diversification can weaken at precisely the time of need. Contractual disclosure of exclusions, reinstatements, collateral, attachment conditions and counterparty quality is essential to understanding the effective boundary of the transfer.
Incentive and Information Constraints
Private information and selection
Insurance pools depend on who chooses to buy coverage. If individuals know their own hazard better than the insurer and higher-risk persons buy more generous cover at a pooled premium, the observed claims cost can exceed the original population estimate. Repricing in response may cause lower-risk buyers to withdraw, changing the risk mix again. This adverse-selection mechanism is related to the broader quality-uncertainty problem analyzed by Akerlof (1970) and to the contract equilibria considered by Rothschild and Stiglitz (1976).
The mechanism is conditional, not inevitable. Risk classification, experience rating, broader compulsory participation, subsidies, waiting periods and contract menus can change selection incentives. Some interventions improve pooling while others impose privacy, fairness or administrative burdens. A market can have a positive level of coverage even without a single idealized competitive equilibrium. Empirical selection must be established from observed behavior and risk rather than assumed from the presence of insurance.
For a stylized illustration, suppose a population is half low-risk with expected claims of 250. The unconditional average is 150, expected claims per buyer become 100 per-buyer shortfall is an arithmetic consequence of the assumed purchase composition. It does not show that insurers universally cannot price heterogeneous risks; it shows how the relevant distribution is that of purchasers rather than the entire initially observed population.
Behavior after coverage and preventive incentives
Moral hazard concerns the effect of insurance on actions, precautions or claims behavior when those actions are imperfectly observed or contracted upon. It may arise because reducing the insured's share of a loss also changes the private return to prevention. A deductible or coinsurance restores some financial exposure, while inspections, contractual safety obligations and premium credits may encourage risk control. Moral hazard should not be used as a blanket accusation against policyholders; it describes a potential incentive structure requiring case-specific evidence.
Ehrlich and Becker's (1972) market-insurance and self-protection distinction is particularly important: market insurance can sometimes complement preventive effort, depending on how decisions and incentives interact. Insurer loss-control services, safer building requirements and cybersecurity assessments can create positive prevention incentives as well as financial protection. The Geneva Association's (2023) changing-risk-landscape study emphasizes the expanding role of risk-prediction and prevention services alongside indemnity.
Let the probability of a covered event be p(e), decreasing with prevention effort e, which has private cost c(e). With full indemnity, a simplified policyholder minimizing only own monetary cost may choose effort by minimizing c(e), whereas with retained loss s−b after insurance payment b, the objective becomes c(e)+p(e)(s−b). When the retained loss is positive, preventive effort can have private economic value. Real insurance may also affect income, legal duties, reputation, future premiums and nonfinancial harm, all of which can support prevention even with full monetary indemnity. The equation illustrates an incentive channel rather than predict an individual decision without further preferences and constraints.
Contract design, underwriting and data governance
Underwriting can modify feasibility through accurate classification, exposure limits, risk engineering, deductibles, waiting periods, conditions, exclusions and price differentiation. Yet narrower terms can create disputes or fail to meet the policyholder's protection needs. A viable contract balances identifiable triggers, claims administration, buyer understanding and insurer capacity. The technical achievement of modeling an exclusion is not the same as a socially acceptable allocation of risk.
Data-intensive underwriting may improve prediction while generating concerns about proxy discrimination, privacy, contestability and access. The scientific question is whether each variable improves relevant risk estimation, meets applicable legal requirements and has an appropriate causal or predictive basis. Strong governance includes documented model scope, validation, drift monitoring, review of disproportionate effects, human accountability and transparent contract communication. The IAIS Core Principles provide a supervisory setting for governance, conduct and risk management, while specific consumer and privacy requirements vary by jurisdiction.
Cyber underwriting offers a concrete risk-control context. The NIST Cybersecurity Framework 2.0 (2024) organizes cybersecurity risk governance and management outcomes. It is a risk-management reference rather than an actuarial guarantee: satisfying a control framework does not prove that a firm cannot experience a cyber incident or that catastrophic accumulation has been diversified. Insurers can use evidence of controls to make selection and mitigation more informed, but coverage terms and portfolio exposures remain separately important.
Adverse-selection and self-protection cases
The practical effect of information and incentives depends on the coverage market. In household flood insurance, risk maps, mortgage requirements, premiums and perceptions of disaster assistance can affect take-up. In cyber insurance, security controls and incident reporting can affect both estimation and incentives. In health protection, benefit design, cost sharing and collective participation shape different choices than those in property catastrophe insurance. These examples share broad economic mechanisms while retaining distinct institutions, welfare considerations and legal constraints.
An insurer can respond to uncertainty by excluding a peril, but a contractual exclusion should be assessed together with its information effects. If buyers misunderstand the exclusion, observed policy purchases may not reflect informed willingness to accept retained risk. A market with sophisticated risk classification and poor comprehension can price claims accurately while delivering weak protection. Editorial evaluation of insurability must therefore include the actual usefulness of the promise to the insured, not solely the likelihood that the insurer earns an underwriting profit.
Affordability and Legal Constraints
Price feasibility and the protection gap
A carrier's sufficient premium for a defined contract may be schematized as P_req=E[L_c]+C_admin+C_reinsurance+C_capital+C_other, where L_c is covered liability and the additional terms represent operating, risk-transfer, capital and other required costs, including appropriate margins. The buyer's maximum acceptable premium, P_max, depends on resources, preferences, perceived risk and alternative protection. A voluntary transaction can occur on these simplified terms only when P_req<=P_max and regulatory and contractual conditions are also satisfied. This is a statement of price feasibility, not a novel equilibrium theorem or a precise accounting specification. Cross-subsidies, government payments, taxes, financing and mandate design can change who faces which effective price.
The distinction between expected loss and affordable insurance is observable in policy discussions. The OECD (2026) identifies availability, affordability and take-up as different sources of inadequate financial protection. An offer may exist at an unaffordable price; a household may decline an affordable offer; an excluded location may have no offer at all. A population's protection gap can arise through all three channels and through limits or underinsurance among policyholders.
The EIOPA natural-catastrophe protection-gap dashboard combines estimates of hazards, losses and insurance coverage for European countries and distinguishes current, historical and country-specific views. The dashboard reports historical economic and insured losses for 1980–2024 and was updated in December 2025. It illustrates how protection-gap measurement requires denominator definitions, hazard-specific exposure and policy coverage data; a headline uninsured-loss share does not establish why coverage is absent.
A useful economic distinction concerns whether affordability is evaluated at household, firm, insurer or national budget level. A subsidy can make a premium affordable for a household while increasing fiscal cost. Lower insurer premiums without risk reduction or adequate public funding can shift the probability of shortfall to policyholders or taxpayers. Policy evaluation should report the final distribution of loss-bearing responsibilities.
Prudential and legal boundaries
An enforceable insurance contract exists within a legal system, and its content is shaped by statutes, regulatory rules and public policy. Jurisdictions vary in treatment of insurable interest, exclusions, indemnity against fines, punitive awards and deliberate unlawful acts. These matters require jurisdiction- and date-specific analysis; there is no single international rule automatically authorizing or prohibiting all such indemnities. OECD (2026) includes societal and legal permissibility among conditions of insurability and identifies regulatory fines and penalties as examples of potentially excluded indemnification.
Prudential regulation imposes a different type of constraint: a contract may be lawful, but the insurer may lack acceptable capital to support the exposure. EIOPA describes three pillars of Solvency II: quantitative requirements, governance including the Own Risk and Solvency Assessment, and supervisory reporting and disclosure. Internationally, the IAIS insurance supervisory principles form a reference for national frameworks, not a substitute for national implementation. Prudential capacity and contract legality must be assessed together.
Legal uncertainty itself can affect reserving and pricing. Interpretation of triggers, exclusion wording and damages may evolve through litigation; cross-border exposures may encounter multiple legal systems. The relevant question is which court, law and contractual provision govern the proposed transfer, and what payment the carrier could ultimately be required to make. A general statement that a given peril is insurable cannot resolve these issues.
Fairness, availability and household budgets
Risk-based premiums convey information about exposure and can encourage prevention, but the households with the highest estimated hazards may also face low incomes or limited relocation options. Insurance can be technically feasible and economically efficient for a carrier while leaving essential household assets uncovered. Social affordability concerns therefore require attention to income, housing, alternative support and the distribution of risk. The OECD's (2026) catastrophe-protection assessment considers both prevention incentives and public financial protection in evaluating available approaches.
Fairness questions include how to treat risks largely outside an individual's control, whether cross-subsidies should be transparent, and how to avoid subsidizing repeated exposure without preventive investment. These are choices among institutional objectives, not consequences determined by a loss distribution. An analytically serious assessment reports which population bears contributions, who receives protection, whether the system incentivizes safer behavior and what claims-paying guarantees exist.
For a household, insurance demand may also reflect trust. A low nominal premium can offer poor protection if exclusions are unclear or claims are unreliable. A high-quality policy may be underpurchased because consumers misperceive low-frequency hazards. Efforts to broaden insurability thus require credible institutions and intelligible contracts alongside underwriting expertise. Such demand and conduct considerations remain related to, but analytically distinct from, the quantitative capacity to insure correlated losses.
Comparative insurance arrangements
The United Kingdom's Flood Re program provides a concrete example of an industry-government reinsurance arrangement designed to support availability and affordability of household flood cover while managing a transition toward risk-reflective pricing. Its published scheme horizon extends to 2039. The arrangement does not imply that the physical hazard has been neutralized; it changes how an identified flood-risk layer is financed and how the transition is organized. Coverage eligibility and scheme design require attention to specific rules.
In the United States, the Treasury's Terrorism Risk Insurance Program documents the shared public-private compensation mechanism created by the Terrorism Risk Insurance Act and its legislative extensions. The Treasury page records the 2019 extension through December 31, 2027. Legislative proposals in 2026 should not be conflated with enacted law without further confirmation. This case illustrates government participation in the tail of a politically salient, potentially concentrated peril, subject to statutory terms and triggers.
Across both arrangements, the analytically relevant quantities are the private attachment and retention, premium and levy design, public contingent liability, claims administration and incentives for mitigation. The public mechanism may increase access or market continuity without lowering society's total disaster losses. Comparing institutional architecture is more informative than labeling any public backstop a solution to risk itself.
Changing Boundaries and Public Intervention
Catastrophe and climate adaptation
Catastrophes produce highly uneven losses in time and location. A single event can damage many insured properties, disrupt earnings and weaken local economic activity. Climate change adds the possibility that hazard distributions and exposures evolve during the contract period. Charpentier's (2008) climate insurability analysis links correlated large losses to higher capital needs and uncertainty. The IPCC's (2022) Working Group II assessment provides an authoritative scientific basis for treating adaptation and physical risk change as integral to financial decisions.
The OECD (2026) reports in Financial Protection Against Catastrophic Risks that fewer than half of natural-hazard losses since 2000 were insured in 24 of 38 OECD countries. This statistic describes a historical coverage outcome within a defined set of countries and hazards, not a timeless measure of uninsurability. The same report explores flood and wildfire protection and emphasizes that risk reduction is central to sustainable access. Better land-use policy, safer construction, infrastructure maintenance and adaptation can change expected losses and the tail that insurance must finance.
An effective policy response depends on the reason for a gap. Where models are weak, better exposure data and stress tests may improve risk understanding. Where risk is highly concentrated, spatial diversification or reinsurance may help. Where chronic high risk drives prices beyond budgets, adaptation and targeted financial support may be more sustainable than suppressing risk signals indefinitely. Where the hazard is already becoming predictable, policymakers may need to consider planned relocation, protective works or direct assistance in addition to insurance.
The insurer's annual contract and the household's long-lived investment horizon can diverge. Short renewals allow an insurer to react to new information, while a household cannot easily replace a building or move after repeated premium increases. This horizon mismatch creates a legitimate role for policies that improve long-run risk reduction and transparency. Insurance can finance recovery after shocks, but it is not an economic substitute for preventing avoidable losses.
Cyber accumulation and emerging exposures
Cyber losses vary from localized data incidents to events affecting shared infrastructure, software providers or network services. Historical claims data may be scarce or biased by reporting thresholds, while attack techniques and dependencies evolve. Biener, Eling and Wirfs (2015) identify interrelated losses and data scarcity as substantial challenges in their empirical insurability assessment. A small claims sample from one operational-risk database cannot uniquely determine the extreme-tail distribution of a global cyber catastrophe.
The Geneva Association's (2023) report on cyber risk accumulation analyzes how a single extreme incident can affect many insured entities and why standard actuarial experience may inadequately represent such events. Accumulation mapping, scenario analysis, clear definitions of cyber war and other exclusions, coverage aggregation controls and capital provision are relevant responses. Their adequacy depends on the actual portfolio and contracts, not on a generic declaration that cyber risk is either insurable or uninsurable.
The Geneva Association's (2023) changing risk-landscape report also highlights the interaction of climate, cyber and intangible risks. Cyber underwriting may combine operational controls guided by the NIST Cybersecurity Framework with limits, co-insurance and risk engineering. A prevention measure can reduce individual incident probability while shared cloud or software dependencies continue to make many insureds susceptible to one initiating event.
A material research challenge is to compare correlation assumptions across data sources and stress scenarios rather than estimate one supposedly universal cyber accumulation coefficient. Another is to distinguish direct event loss, contingent business interruption, liability and intangible harm, each of which depends on coverage wording. These are empirical and contractual tasks for future insurance research, rather than grounds for claiming a new authorial system.
Sovereign capacity, risk layering and public institutions
Public institutions can supplement private insurance when high-severity risks exceed a commercially acceptable layer or when access fails for defined groups. The World Bank's disaster-risk finance and insurance program describes the combination of reserves or contingency funds for smaller and more frequent shocks, contingent finance for intermediate funding needs, and risk transfer for larger events. These instruments serve different timing and cost purposes. A contingent loan creates funding capacity but usually retains a future repayment obligation; insurance transfers defined losses to a premium-funded carrier; a fiscal reserve pre-funds retained losses.
The OECD (2026) framework for government-supported protection identifies where public support may be justified and differentiates a generalized lack of available insurance from narrower affordability problems. It also stresses that a public insurance program and direct compensation mechanism have distinct effects. These distinctions permit evaluation of public roles without presuming that a state must insure every large risk or that every subsidy necessarily expands social welfare.
Risk layering should be evaluated against the timing and fiscal consequences of loss. Low-severity recurring expenditure may be more cheaply retained and budgeted. High-severity events may justify market transfer when capacity is available and priced adequately. Extreme aggregate events may require a public contingent funding arrangement, international assistance or social protection. Contractual and fiscal arrangements can coexist, but each has different beneficiaries and conditions for disbursement.
The possibility of government support also affects private behavior. Anticipated disaster assistance can change insurance purchase and investment in prevention; poorly structured premium support can mute risk signals. Conversely, a credible public backstop may allow insurers to offer coverage they could not sustain independently. Cummins' (2006) catastrophe-insurance policy critique provides one influential argument for minimizing market displacement; the OECD's 2026 comparative analysis surveys a broader range of public-private choices. Their different emphases can be assessed through explicit criteria: total fiscal exposure, private capacity mobilized, coverage distribution and incentives for loss reduction.
Integrated cross-case synthesis and research agenda
The cases yield several linked analytical findings. First, insurability is a property of a specified arrangement under particular conditions. This follows from contract limitation, carrier capacity and price feasibility, and is consistent with Berliner (1985). Second, the boundary can shift through improvements in knowledge or financial capacity while a physical hazard remains unchanged. Third, a shift that makes an insurer's layer feasible often transfers a residual loss to another actor. Fourth, actuarial feasibility, access and legal permissibility require separate evidence. Fifth, dynamic hazards and changing systems make review time and contract duration part of the assessment.
Three contrasting cases clarify the causal logic. For an ordinary diversified pool of small independent property losses, the central advantages are aggregation and predictable average costs. For a geographically concentrated flood portfolio, dependence and tail capital may dominate even if average loss is well estimated. For a cyber portfolio tied to a common cloud provider, uncertain accumulation and dependency mapping may dominate despite sophisticated client-level security controls. None of these statements sets a universal threshold; each identifies which component must be tested.
A robust future empirical study would identify the exact exposure and proposed contract, then document loss measurements, time and location, the dependence structure, tail assumptions and potential controls. It would independently evaluate the insurer's stress losses, capital and liquidity, available reinsurance, underwriting expenses and counterparty risk. It would then estimate premium affordability, demand and take-up, and record the jurisdictional contract and regulatory rules. The last stage would compare retained losses after the proposed transaction with plausible alternatives including mitigation, self-insurance, public funds and contingent finance. This is an application-oriented research sequence assembled from established disciplinary questions, not a claim to an original general theory.
Material uncertainties persist. Severe cyber losses and long-run climate exposures can defeat inference from short samples. Catastrophe-loss observations can be affected by reporting changes, urban development and inflation. Cross-national affordability measures use different income and coverage definitions. Litigation and regulatory amendments alter legal outcomes. These uncertainties establish requirements for transparent datasets, scenario sensitivity, and dated legal readback; they do not justify unsupported numerical forecasts.
Conclusion
Insurability concerns a precisely defined financial promise supported by risk information, contractual allocation, institutional capacity and law. Its boundaries emerge where statistical uncertainty, tail dependence, incentives, capital, affordability or public policy prevent an otherwise desired transfer. The constraints are linked but analytically distinct. An insurer can understand a peril and still lack the capacity to absorb it; a solvent insurer can offer protection that a household cannot buy; and a willing buyer and seller can face legal restrictions on the proposed indemnity.
The formal analysis establishes why the size of a pool is insufficient when exposures share a common shock. Under independent identical Bernoulli losses, the relative standard deviation falls at the rate N^{-1/2}. Under positive common correlation it approaches a positive limit. This finding explains why a concentrated catastrophe or digital-interruption portfolio may require unusually large buffers or a different risk-sharing architecture. Tail limits, retention and transfer can make selected layers financeable while leaving residual losses elsewhere.
The comparative analysis establishes the operational importance of information and incentives. Adverse selection can increase the average cost of the purchasing pool; coverage can affect prevention; risk engineering and contract terms may improve feasibility while creating coverage complexity. The premium needed to support a contract must be reconciled with what participants can and will pay. Legal and prudential requirements place further conditions on the promise, and those requirements are specific to jurisdiction and time.
Public institutions can alter the practical boundary through risk-sharing mechanisms, subsidies, contingency funding, capital arrangements and prevention. Flood Re in the United Kingdom, the U.S. terrorism backstop, and the World Bank's risk-layering approach illustrate distinct assignments of financing responsibility. Their existence demonstrates institutional adaptation to difficult risks; their evaluation requires attention to residual exposure, fiscal cost, private capacity and long-term mitigation.
The resulting synthesis supports a clear research conclusion: statements about a peril's insurability become scientifically meaningful when the proposed contract, insured population, portfolio dependence, risk carrier, price, legal framework and time horizon are made explicit. Changes in these elements can expand or contract feasible protection. This monograph supplies original comparative interpretation and mathematical examples within established insurance science. It introduces no newly named theory, doctrine or proprietary insurability principle. Its purpose is to clarify an internationally applicable analytical question and the evidence needed to answer it responsibly.
References
Akerlof, G. A. (1970). The market for “lemons”: Quality uncertainty and the market mechanism. The Quarterly Journal of Economics, 84(3), 488–500. https://academic.oup.com/qje/article-abstract/84/3/488/1896241
Berliner, B. (1982). Limits of Insurability of Risks. Prentice-Hall. https://books.google.com/books/about/Limits_of_Insurability_of_Risks.html?id=Kd3uAAAAMAAJ
Berliner, B. (1985). Large risks and limits of insurability. The Geneva Papers on Risk and Insurance, 10(37), 313–329. https://www.casact.org/abstract/large-risks-and-limits-insurability
Biener, C., Eling, M., & Wirfs, J. H. (2015). Insurability of cyber risk: An empirical analysis. The Geneva Papers on Risk and Insurance—Issues and Practice, 40(1), 131–158. https://ideas.repec.org/a/pal/gpprii/v40y2015i1p131-158.html
Charpentier, A. (2008). Insurability of climate risks. The Geneva Papers on Risk and Insurance—Issues and Practice, 33(1), 91–109. https://ideas.repec.org/a/pal/gpprii/v33y2008i1p91-109.html
Cummins, J. D. (2006). Should the government provide insurance for catastrophes? Federal Reserve Bank of St. Louis Review, 88(4), 337–380. https://www.fedinprint.org/item/fedlrv/24686
Ehrlich, I., & Becker, G. S. (1972). Market insurance, self-insurance, and self-protection. Journal of Political Economy, 80(4), 623–648. https://www.journals.uchicago.edu/doi/10.1086/259916
European Insurance and Occupational Pensions Authority (EIOPA). (n.d.). Solvency II. Institutional overview. https://www.eiopa.europa.eu/browse/regulation-and-policy/solvency-ii_en
European Insurance and Occupational Pensions Authority (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
European Parliament and Council of the European Union. (2009). Directive 2009/138/EC (Solvency II), consolidated text as of January 9, 2024. https://eur-lex.europa.eu/eli/dir/2009/138/2024-01-09/eng
Flood Re. (n.d.). About us. Government of the United Kingdom. https://www.gov.uk/government/organisations/flood-re/about
Froot, K. A. (2001). The market for catastrophe risk: A clinical examination. Journal of Financial Economics, 60(2–3), 529–571. https://www.sciencedirect.com/science/article/pii/S0304405X01000526
Intergovernmental Panel on Climate Change (IPCC). (2022). Climate Change 2022: Impacts, Adaptation and Vulnerability. Working Group II contribution to the Sixth Assessment Report. https://www.ipcc.ch/report/sixth-assessment-report-working-group-ii/22/
International Association of Insurance Supervisors (IAIS). (2024). Insurance Core Principles and Common Framework for the Supervision of Internationally Active Insurance Groups, updated December 2024. https://www.iais.org/activities-topics/standard-setting/icps-and-comframe/
Mossin, J. (1968). Aspects of rational insurance purchasing. Journal of Political Economy, 76(4, Part 1), 553–568. https://www.journals.uchicago.edu/doi/10.1086/259427
National Institute of Standards and Technology (NIST). (2024). The NIST Cybersecurity Framework (CSF) 2.0. NIST CSWP 29. https://csrc.nist.gov/pubs/cswp/29/the-nist-cybersecurity-framework-csf-20/final
Organisation for Economic Co-operation and Development (OECD). (2026). Financial Protection Against Catastrophic Risks: Floods, Fires and Other Major Risks. OECD Publishing, Paris. https://www.oecd.org/en/publications/financial-protection-against-catastrophic-risks_32a3acd7-en.html
Organisation for Economic Co-operation and Development (OECD). (2026). A framework for examining government-supported financial protection to respond to large-scale catastrophes. In Financial Protection Against Catastrophic Risks. https://www.oecd.org/en/publications/financial-protection-against-catastrophic-risks_32a3acd7-en/full-report/a-framework-for-examining-government-supported-financial-protection-to-respond-to-large-scale-catastrophes_3e3d1002.html
Pain, D. (2023). Cyber Risk Accumulation: Fully Tackling the Insurability Challenge. The Geneva Association. https://www.genevaassociation.org/publication/cyber/cyber-risk-accumulation-fully-tackling-insurability-challenge
Rothschild, M., & Stiglitz, J. E. (1976). Equilibrium in competitive insurance markets: An essay on the economics of imperfect information. The Quarterly Journal of Economics, 90(4), 629–649. https://academic.oup.com/qje/article-abstract/90/4/629/1886620
Schanz, K.-U. (2023). The Value of Insurance in a Changing Risk Landscape. The Geneva Association. https://www.genevaassociation.org/publication/financial-inclusion/value-insurance-changing-risk-landscape
United States Department of the Treasury. (n.d.). Terrorism Risk Insurance Program. Federal Insurance Office. https://home.treasury.gov/policy-issues/financial-markets-financial-institutions-and-fiscal-service/federal-insurance-office/terrorism-risk-insurance-program/
World Bank Group. (n.d.). Disaster Risk Finance and Insurance. https://www.worldbank.org/ext/en/topic/financial-sector/disaster-risk-finance-and-insurance