RESEARCH DOSSIER
Health Underwriting Risk Capital and Morbidity Shocks
Abstract. Health underwriting capital translates uncertainty in insured morbidity, disability, health-service expenditure, claims settlement, policyholder behavior, and extraordinary concentrations of illness or injury into an insurer's capacity to absorb adverse outcomes. The translation depends on the insurance contract, the insured population, the valuation horizon, the supervisory jurisdiction, the recognition of reinsurance, and the joint distribution of shocks. This monograph develops an integrated actuarial and comparative-regulatory account of that translation. It analyzes the European Union's non-similar-to-life and similar-to-life health classification, the explicit mass-accident, concentration, and pandemic components of health catastrophe risk, the United States' health risk-based capital supervisory mechanism, the International Association of Insurance Supervisors' Insurance Capital Standard, and Switzerland's total-balance-sheet solvency approach. It derives transparent incidence–continuance, cost-trend, and aggregation calculations, distinguishes expected claim liabilities from unexpected capital losses, and uses independently recomputable hypothetical portfolios to expose the sensitivity of capital to underwriting mix and dependence assumptions. The pandemic experience demonstrates why a population-level health emergency can increase healthcare financing needs while reducing contemporaneous utilization of some insured services. The synthesis establishes a practical scientific conclusion: health underwriting solvency is a joint problem of insured benefits, dynamic epidemiological and utilization states, and institutional financial transmission. Regulatory standard formulas create comparable decision rules; forward-looking assessment must additionally test the pathways by which biological, economic, and operational shocks become correlated insurer losses. All numerical portfolios in this work are hypothetical demonstrations, not estimates of particular companies or newly proposed regulatory calibrations.
Keywords: Health underwriting risk; morbidity shocks; health insurance; solvency capital; disability incidence; claim continuance; medical expense inflation; health catastrophe; pandemic risk; non-similar-to-life health; similar-to-life health; risk-based capital; Solvency II; Insurance Capital Standard; Swiss Solvency Test; tail dependence; reinsurance; own risk and solvency assessment.
Introduction
The underwriting of health risk creates financial promises whose settlement depends on the condition and behavior of human populations, medical-service systems, employment arrangements, and public institutions. A medical expense policy may transfer the frequency and price of care. Income protection may transfer the loss of earnings associated with disability. Critical illness coverage may pay on diagnosis, while long-term disability may create a stream of benefits extending across many years. Employer-sponsored coverage can bundle numerous dependent exposures, and reinsurance can move portions of these liabilities between legal entities. Each promise generates a different exposure to morbidity, medical inflation, claim duration, and catastrophic events.
Health underwriting risk capital is the financial capacity required by a specified solvency rule or internal risk assessment to absorb adverse changes arising from those health insurance obligations. Its numerical amount cannot be inferred from a disease-incidence rate alone. A useful analysis must first identify insured cash flows and contract boundaries, then estimate expected obligations, then construct adverse departures and trace them into the insurer's own funds or another legally defined capital metric. Capital is therefore attached to a measured balance-sheet exposure and a defined risk horizon, not to an abstract count of illnesses.
This problem has a mature legal expression in the European framework. Article 101 of the Solvency II Directive defines the Solvency Capital Requirement (SCR) through a one-year 99.5 percent value-at-risk standard for basic own funds, including health underwriting risk among the minimum covered risk categories (European Parliament and Council 2009). Article 105 describes the health underwriting component in terms of contract expenses, claim frequency, claim severity, provisioning, and epidemics under the basic capital requirement provisions. These provisions frame capital as unexpected financial deterioration arising from the insurer's obligations. They also provide a precise reason to separate an expected medical claim liability from a tail-risk charge.
Different institutional systems answer related questions through different capital constructs. The U.S. National Association of Insurance Commissioners (NAIC) treats risk-based capital as a supervisory intervention standard, explicitly distinct from a complete assessment of company solvency or a commercial ranking (NAIC 2026). The International Association of Insurance Supervisors (IAIS) adopted the Insurance Capital Standard in December 2024, primarily to support group-level comparability for internationally active insurance groups (IAIS 2024). The Swiss Solvency Test takes a risk-based total-balance-sheet approach (FINMA 2024). These systems can be compared by objectives, time horizon, eligible capital, treatment of risks, and intervention consequences. Their raw capital figures are not automatically interchangeable.
The research proceeds through six connected subjects: statutory classification, stochastic claim costs, disability continuance, catastrophe shock transmission, cross-risk dependence, and comparative market application. Primary legal and professional sources establish actual rules; published research and official expenditure accounts inform empirical mechanisms; algebra and explicitly hypothetical calculations test internal consistency. The cutoff for source-specific regulatory interpretation is November 14, 2026. In particular, the EU reform scheduled to apply from January 30, 2027 is treated prospectively, following EIOPA's reform implementation summary (EIOPA 2026). The analysis proposes no new named theory, doctrine, or regulatory standard.
Health Solvency Risk Categories
The insured obligation as the unit of capital analysis
Classification begins with the economic obligation borne by the insurer. Medical expense reimbursement, contracted healthcare service, income protection, and benefits triggered by an insured disability involve distinct stochastic mechanisms. For a fixed covered population, a claim process can be represented as the sum of insured benefits triggered by incidence events, multiplied where relevant by utilization, severity, benefit limits, contract exclusions, and the fraction retained after risk transfer. For recurring benefits, payment duration introduces a further random process. A capital calculation inherits the scope of the underlying contracts and therefore has to distinguish direct writers, reinsurers, healthcare service organizations, and administrators that only perform services for another risk bearer.
Let A denote assets on the applicable solvency valuation basis and L denote the corresponding insurance and other liabilities. A simplified measure of own funds is O=A−L, subject in practice to prescribed adjustments and eligibility rules. Let ΔO denote the one-year change after all required balance-sheet revaluations and the treatment of new business. An adverse capital event is a negative change in O relative to the prescribed initial or baseline position. In a simplified loss notation X=−ΔO, capital at confidence level α can be represented as VaRα(X), with attention to whether the regulation defines the risk measure over absolute losses or changes in available own funds. This abstraction explains the logic of Article 101; it does not replace the detailed SCR computation.
The IFRS 17 insurance-contract measurement principles combine current estimates of fulfillment cash flows with a risk adjustment for nonfinancial risk and, where applicable, a contractual service margin (IFRS Foundation 2023). IFRS 17 is an accounting standard rather than a statutory capital formula. Expected claim liabilities, accounting risk adjustments, solvency technical provisions, and the regulatory SCR have separate measurement purposes. Using the same cash flows in several reports does not make the resulting quantities equivalent. The IFRS 17 key-term definitions reinforce the distinction between an incurred-claims liability and a remaining-coverage liability (IFRS Foundation n.d.).
European Union: health risk classified by technical basis
The delegated standard formula separates health underwriting risk into non-similar-to-life techniques (NSLT), similar-to-life techniques (SLT), and health catastrophe risk, as required by Article 144 of Delegated Regulation 2015/35 (European Commission 2015). This is an allocation of insured obligations to prescribed risk modules. A product's marketing label by itself is insufficient to determine every applicable shock. The SLT classification recognizes obligations modeled through life-like techniques, while NSLT primarily addresses premium, reserve, and lapse variability. The catastrophe component is applicable to health obligations subject to its specified scope.
For NSLT business, the regulation combines premium-and-reserve risk with lapse risk through the NSLT health sub-module formula. The Article 146 premium-and-reserve calculation is explicitly written as
Here σ is the standard deviation prescribed through segment calculations and V is a regulatory volume measure, not an arbitrary premium total. Article 147 assigns premium and reserve exposure and a geographic adjustment according to specified volume-measure rules; Article 148 uses segment standard deviations and cross-segment coefficients in its volatility aggregation rules (European Commission 2015). The expression 3σV must not be treated as the output of an unconstrained normal approximation with user-selected inputs in an actual regulatory filing. Its elements have legal definitions and depend on the reporting version.
The SLT health module instead combines mortality, longevity, disability-morbidity, expense, revision, and lapse components, as listed in the SLT health sub-module provisions. The inclusion of both mortality and longevity reflects the possibility that opposing demographic movements can increase the present value of different health or annuity-type obligations. The inclusion of revision and lapse risk reflects contractual rights and changes to benefit amounts; these effects may be material even if observed morbidity rates are unchanged.
The overall health capital amount uses a square-root correlation aggregation of NSLT, SLT, and catastrophe charges. For illustrative notation, with module charges N, S, and C, and the three correlations given in the delegated regulation—0.50 for N–S and 0.25 for each relationship with C—the regulatory structure is
This relationship is a legally prescribed diversification calculation within the specified standard formula. Its correlation parameters are regulatory coefficients, not measured correlations for a particular insurer. Real insurer-specific loss dependence may differ substantially, which makes the distinction between a comparable prescribed requirement and an internal model of actual tail outcomes indispensable.
Capital across legal systems and time
In the United States, separate RBC formulas apply to life, property/casualty, and health organizations. The NAIC's health RBC explanation describes statutory capital thresholds, early supervisory action, and the importance of underwriting risk; it identifies continuing evaluation of the health H-2 underwriting framework as a 2026 work item (NAIC 2026). The ratio of total adjusted capital to Authorized Control Level RBC is designed for intervention, not direct comparison with a one-year 99.5 percent EU SCR figure. The fact that each regime uses risk-based language does not establish equal confidence levels, equal capital definitions, or equal policyholder-protection consequences.
The IAIS capital-standard overview identifies the ICS as a group-level risk-based capital measure for internationally active insurance groups (IAIS n.d.). Adoption in 2024 preceded phased national implementation and assessment. According to the IAIS implementation announcement, baseline self-assessments were scheduled for 2026 and detailed jurisdictional assessments were aimed at 2027. Its standards consequently provide a comparative supervisory frame without justifying the claim that every jurisdiction already applies identical capital rules.
Switzerland's SST measures changes in an insurer's economic balance sheet over one year and compares eligible risk-bearing capital with target capital. FINMA's SST supervisory description is an institutional comparison of approach rather than a license to substitute the EU health correlation matrix in Swiss calculations. The precise risk drivers must be evaluated under the applicable Swiss instruments and models.
Finally, the Solvency II review timeline fixes January 30, 2027 as the application date for the revised framework (EIOPA 2026). The legal description in this monograph refers to the rules operative at its November 2026 cutoff and identifies pending amendments separately. Applying future effective provisions to historical calculations would create a misleading legal reconstruction.
Health Expense and Morbidity
Incidence, utilization, severity, and contractual payment
A medical insurance portfolio's expected annual payment is determined jointly by the number of eligible lives, their health states, the probability of treatment, the volume of healthcare services consumed, the price or negotiated reimbursement for each service, and contractual cost sharing. One representation is
where H denotes a clinically and contractually relevant state, U utilization, P unit price, and B the share payable by the insurer after plan features. This decomposition is analytical: it prevents a change in service price from being mechanically described as a change in incidence, and a reduction in elective utilization from being confused with a decline in underlying disease prevalence. A change to the benefits contract can affect the probability of a claim or the insurer's portion of costs even if healthcare use is constant.
When claim count N and independently distributed individual severities Y have finite second moments, an elementary compound-sum model gives E[C]=E[N]E[Y] and Var(C)=E[N]Var(Y)+Var(N)(E[Y])². These identities require the stated independence assumption and appropriate conditional interpretation. They are useful for identifying claim-frequency and claim-size components; real health portfolios often violate the assumptions through common infections, correlated treatment pathways, clustered provider prices, and changing mix of insured lives. A richer model introduces shared covariates and conditional dependence.
Health underwriters face three related but distinct uncertainty categories. Process variation persists even with known parameters: the number and cost of claims are random. Parameter uncertainty arises because morbidity and treatment-cost estimates are based on finite data. Structural uncertainty arises when the data-generating process changes, perhaps because of new treatment technology, public health intervention, provider consolidation, climate-related exposures, or a new pathogen. A capital analysis based exclusively on a stable historical claims distribution can incorporate the first while systematically missing aspects of the latter two.
Expected liabilities and incurred-claims estimation
Reliable capital measurement starts with a defensible liability estimate. The U.S. Actuarial Standards Board's ASOP No. 5 on incurred health claims addresses claims incurred but not fully settled, including future payments on reported long-duration disability claims under the relevant definition (ASB 2017). The standard instructs actuaries to consider plan terms, medical and economic trends, reporting lags, business practices, claim administration, and relevant external events. The actuary's task is to estimate cash flows and explain their uncertainty for the assignment's actual purpose. It is conceptually insufficient to use a statutory capital factor to repair an unexamined deficiency in the underlying reserve.
Let I_t be the number of claims incurred in accident or service period t, R_{t,d} the fraction of aggregate incurred amount reported and settled by development period d, and C_{t,\infty} the ultimate cost. A simple development estimator divides the amount observed through d by an estimated completion factor, assuming development comparability. For disability benefits, a tabular estimate can instead project each open claimant using benefit amount, eligibility, recovery, mortality, and termination probabilities. The two methods address different time structures: short-tail medical reimbursement commonly depends on lag and development, while income protection may depend chiefly on years of post-onset continuance.
Under a prospective policy the present value of future benefit obligations may be written
where B denotes benefits, E expenses, P future contractually included premiums, v discount factors, and the information set at the valuation date defines the conditional expectation. This expression illustrates discounted net cash flows; a statutory best estimate must apply its specific contract boundaries, cash-flow inclusion criteria, valuation assumptions, and reinsurance rules. A change in morbidity assumptions can increase both future payments and solvency capital, but the two effects must be attributed separately.
A medical cost trend can be decomposed into general price inflation, medical input-price differences, utilization intensity, treatment-mix change, benefit changes, and population risk composition. These contributions can interact. If a claims severity projection uses annual multiplicative growth g, then over t years the factor is (1+g)^t. A persistent increase from g to g+δ has a larger proportional effect on distant liabilities than on payments due next quarter. The sensitivity of (1+g)^t to g is t(1+g)^{t−1}; discounting and claim survival determine the net balance-sheet effect. Long-term contracts therefore create interactions between morbidity, medical inflation, interest rates, and lapse behavior even when the immediate expected claims ratio appears stable.
Regulatory health expense shocks
Solvency II distinguishes a shock to the expense of administering insurance obligations from shocks to insured medical payments. For SLT health contracts, Article 157's expense stress specifies a 10 percent increase in the expenses entering technical provisions together with a one-percentage-point increase in expense inflation (European Commission 2015). The base of this calculation consists of the relevant servicing expenses and their future path; it must not be mistaken for a blanket 10 percent increase in every insured medical bill.
The medical expense disability-morbidity capital calculation, by contrast, uses the larger of adverse increases or decreases in medical payments under Article 155's paired medical shocks. The upward scenario changes the amount of medical payments by 5 percent and raises the medical-payment inflation rate by one percentage point; the downward scenario has corresponding decreases. The larger loss to basic own funds governs. Both directions are potentially relevant because a decline in medical payments can change reserve, revenue, or linked-contract economics in ways that make the net effect adverse. The calculation is defined for the specified SLT medical expense obligations, rather than for every health policy in every regime.
For NSLT premium and reserve exposures, the regulator defines volume by premiums and claims provisions with detailed boundary and geographical adjustments. The Article 147 premium and reserve bases include future premium measures and net claims provisions subject to specified conditions; they are not equivalent to a historical paid-claims ratio. A portfolio manager who uses earned premium as a shorthand must explicitly say that it is a simplifying proxy and cannot report the result as the statutory SCR.
The treatment of a legally qualifying health risk equalization system also matters. Article 149's equalization provisions specify adjustments for qualifying arrangements in NSLT risk assessment. An equalization transfer may alter the insurer's retained exposure, but the legal eligibility and formula must be established before credit is taken. The existence of government support or a risk-adjustment program in ordinary business language does not by itself prove regulatory recognition.
Adverse selection, access, and financial transmission
Medical claims capital is also shaped by institutional design. Restrictions on individual risk classification, open enrollment, guaranteed issue, community rating, risk adjustment, and public coverage interact with the composition of insured groups. If adverse selection increases the average conditional morbidity of an enrolled population, a pool can deteriorate even with no change in disease rates in the surrounding population. A morbidity shock should accordingly be evaluated conditional on the insured population, covered services, reimbursement contracts, and duration of the insurer's commitment.
Risk selection and access are distinct scientific questions from the calculation of financial solvency. More granular morbidity information can improve conditional prediction, but it can also induce classification instability, proxy discrimination, selection feedback, and privacy concerns. An insurer's capital model must be adequate for losses borne under the lawful coverage regime. That obligation is independent of whether the insurer is permitted to price separately for every predictor. Where regulatory risk-sharing mechanisms materially reduce exposure, the model should identify the legally enforceable payment stream and counterparty instead of inventing an implicit diversification credit.
The 2025–2034 CMS expenditure projection series reports projected U.S. health spending categories and source-of-funds composition (CMS 2026). National expenditure projections illuminate plausible changes in aggregate medical price and use; they are not direct forecasts of the net insured claims of a specific commercial policy. Applying a national growth rate without adjusting for insured age, geography, benefits, reimbursement, enrollment and public payment interactions confuses the macroeconomic expenditure aggregate with underwriting risk. Capital calibration needs an exposure-specific bridge from national information to contractual cash flows.
An empirical lesson from pandemic utilization
In the early COVID-19 period, lower use of some nonurgent medical services coincided with substantial pandemic-related healthcare expenditure. Hartman and colleagues (2021) show in their analysis of U.S. healthcare spending in 2020 that aggregate spending rose 9.7 percent to $4.1 trillion, largely because of federal expenditure, while private insurance spending declined 1.2 percent. Martin and colleagues (2022) document the subsequent 2021 spending and utilization rebound, when demand for previously delayed services contributed to greater use.
This evidence establishes a crucial methodological boundary. National healthcare spending, private insurance expenditure, insurance-claims liabilities, and insurer solvency capital respond to different institutional mechanisms. The same pandemic can lower current elective claims for one insurer, increase infection-related claims for another, reduce premiums through enrollment losses, and create later utilization catch-up. Claims incidence can change contemporaneously while latent health burden and future treatment demand change in different directions. Modeling pandemic capital as a single uniform positive multiplier on medical costs misses these observed channels.
Disability and Continuance
From morbidity onset to a long-duration liability
Disability and income protection exposures differ from annual medical expense insurance because payment duration may dominate initial event cost. The scientific problem begins with transitions among economically relevant states: active, temporarily disabled, permanently disabled, recovered, and deceased. Entry into disability is driven by incidence; remaining disabled depends on recovery and mortality; benefits depend on the contract's elimination period, waiting requirements, occupation tests, offsets, benefit escalation, and termination age. For some products reopening a closed claim also matters. The categories selected for modeling have to correspond to actual contract conditions.
A discrete-time multi-state process can be represented by a transition matrix P_t whose element p_{ij,t} is the probability of moving from state i to state j over interval t, conditional on the initial state and relevant covariates. If d_t is the number of disabled claimants at the start of year t and q^{rec}_t and q^{death}_t are mutually exclusive transition probabilities to recovery and death, then under a simple homogeneous state model
A broader model would use age, diagnosis, occupational class, disability duration, benefit type, inflation indexation, and economic conditions. The equation isolates two mechanisms: an increase in new disability incidence and a reduction in exits from existing disability. Both can increase reserves. Their effects need not be additive because the additional claimants themselves become exposed to the changed continuance probabilities.
The European SLT risk module includes an explicit disability-morbidity component. Article 154's separation of health morbidity shocks distinguishes medical expense from income protection obligations (European Commission 2015). Article 156's income-protection stress specifies increased disability and morbidity rates for the coming twelve months and later years, as well as conditional shocks to recovery or persistency rates. Its published text specifies a 35 percent increase in the first year and a 25 percent increase thereafter, with 20 percent changes to recovery or persistency rates under the stated thresholds. These parameters must be implemented with the exact applicable legal definitions; a simple one-rate multiplication is a teaching device rather than a complete supervisory scenario.
Derivation of a continuance factor
Consider an intentionally elementary annual disability model. There are n insured lives; a fraction q enters a qualifying disability state at the beginning of a specified future period; each new claimant receives annual benefit b in advance while disability continues. Annual probabilities r of recovery and d of death are constant, mutually exclusive, and independent of duration; the annual effective discount rate is i. The one-year continuance probability is p=1−r−d. The expected discounted benefit to a newly disabled claimant, starting immediately, is the annuity-due factor
The expected value of benefits resulting from the specified cohort's newly incurred disabilities is then nqb a_dis. This follows from the geometric series when p/(1+i) is strictly below one and the stated payments and transitions occur in that order. The derivation deliberately omits benefit expiration, age-related retirement, tax, expenses, existing claims, reinsurance, waiting periods, changes to transition rates, claim reopening, and stochastic interest rates.
For n=10,000, q=0.01, b=$20,000, r=0.15, d=0.02 and i=0.03, p=0.83 and a_dis=5.15. The hypothetical cohort produces 100 expected entrants and an expected discounted stream of $10.300 million. If q rises by 35 percent to 0.0135 with all other assumptions fixed, the expected amount becomes $13.905 million, an increase of $3.605 million. If instead recovery falls 20 percent from 0.15 to 0.12, p rises to 0.86, a_dis becomes approximately 6.058824, and the expected value becomes $12.117647 million. With both changes, it becomes approximately $16.358824 million, an increase of $6.058824 million over the initial amount.
| Hypothetical cohort assumption | New-disability incidence | Recovery probability | Annuity-due factor | Expected benefit PV |
|---|---|---|---|---|
| Baseline | 1.00% | 15.0% | 5.150000 | $10.300m |
| Incidence increase only | 1.35% | 15.0% | 5.150000 | $13.905m |
| Recovery deterioration only | 1.00% | 12.0% | 6.058824 | $12.118m |
| Combined changes | 1.35% | 12.0% | 6.058824 | $16.359m |
These are analytical cash-flow sensitivities. They do not calculate the actual SCR, its required shock basis, or its full loss in basic own funds. The example nevertheless reveals why the financial effect of lower recovery increases with long benefit duration, and why incidence and continuance stresses interact. With a low discount rate, duration sensitivity becomes more pronounced. A product subject to eligibility reassessment, policy limits, age caps, or benefit offsets may show a smaller effect, but only a contractual projection can establish its magnitude.
Claim-state observation and model error
Recovery is often observable only through administrative events, and apparent termination can include policy expiration, a change in benefit eligibility, mortality, or return to work. If these causes are pooled, the estimated recovery hazard can be systematically biased. A model with duration-dependent recovery must distinguish elapsed disability time and origin date, and compare experience across underwriting cohorts. Severe diagnoses may show low early recovery, while occupational or social changes can alter later patterns. A population's demographic composition also determines the weight assigned to prolonged benefit streams.
For health claims actuaries, the ASOP No. 5 guidance on long-duration claims is directly relevant to the estimation of future payments on reported claims. It addresses actuarial practice under U.S. scope rather than setting European regulatory capital factors. The analytic principle is transportable: data-definition discipline and projection assumptions must precede any meaningful stress test. An adverse scenario applied to an incorrect baseline creates a precise-looking but unreliable capital number.
The effects of pandemic illness, long-term impairment, and labor-market changes may appear at different times. Short-run medical claims can reflect acute hospitalization; disability costs may emerge through delayed diagnosis, functional impairment, behavioral health, or changes to work capacity. Empirical inference requires comparison cohorts and careful handling of clinical definitions and socioeconomic confounding. A rise in observed disability claims does not by itself establish the cause, and a temporary fall in claims administration activity does not establish lower underlying incidence. Reserving data, healthcare utilization, and labor-market data measure different stages of the economic process.
Currency, discounting, and financial interaction
Long-term disability promises also connect underwriting risk to financial risk. If benefit amounts are indexed to medical costs or consumer prices, adverse inflation changes the contractual payment path. If liabilities are long-dated, discount-rate changes affect present values. Benefit offsets tied to social insurance can move with law or eligibility. Each of these mechanisms should be modeled within its applicable capital module or internal scenario without counting the same financial change twice. A comprehensive stress may combine higher morbidity incidence with wage inflation, unemployment, asset-value deterioration, and impaired recovery.
This joint treatment is especially important for group income protection. A large employer can contribute significant concentration by occupation, geographic location, or common accident exposure. The employer's financial distress may simultaneously lower premium receipts or increase terminations. Reinsurance can transfer a portion of claim severity, while lapse or policy termination can alter the remaining contractual exposure. The realistic capital question is the net deterioration in the insurer's financial position after legally relevant contract and risk-transfer terms are applied, with counterparty and settlement risks remaining visible.
Health Catastrophe Events
Event concentration and systemic propagation
Ordinary underwriting variation arises from dispersed claims under a relatively stable exposure environment. Health catastrophe risk arises when an event or process generates an unusually large concentration of covered health outcomes. An industrial disaster can disable many employees in a common location. A major transport accident may generate acute medical and disability benefits across many contracts. An infectious disease outbreak can affect geographically dispersed insured persons through transmission networks. The underlying risk therefore depends on the physical or epidemiological event, its interaction with covered populations, and the insurer's distribution of contractual obligations.
The European delegated regulation implements this distinction explicitly. Article 160's catastrophe structure defines three components: mass accident, accident concentration, and pandemic. Their capital charges are combined by a square root of the sum of squares, subject to the regulation's definitions (European Commission 2015). Workers' compensation and group income protection receive particular treatment for accident concentration; the mass-accident and pandemic scopes exclude specified workers' compensation obligations. Classification and contract mapping are essential. Counting all employment-related disability exposures identically would distort the legally prescribed catastrophe calculation.
Article 161's mass-accident method considers country-specific exposure and the prescribed proportions of persons affected and benefit types. Article 162's concentration method centers on the largest accident risk concentration in each relevant country. The first examines exposure to a specified mass event at national scale; the second directs attention to a concentrated cluster of insured benefits. Both connect catastrophic loss to the spatial, employment, and contractual structure of the insurer's business, rather than to national event counts alone.
The pandemic scenario as a specified regulatory calculation
Article 163's pandemic capital expression defines a gross instantaneous loss amount by combining an income-protection exposure term and an insured medical-cost term:
E represents the stipulated income-protection pandemic exposure, N_c qualifying insured persons in country c, and M_c the prescribed expected average payment per insured person for the specified pandemic healthcare use. The expression is an input to the regulatory basic-own-funds loss calculation, not the declaration that gross L_p always equals final net SCR. Reinsurance recoveries, counterparty credit, and other legally applicable mechanics require independent treatment. The numerical coefficients are regulatory parameters, not frequencies estimated from the insurer's own pandemic data.
Consider a purely illustrative input set, E=$200 million, 50,000 covered people in one eligible country, and M=$120 per covered person. The first term equals $15,000 and the second equals $2.4 million, so gross L_p is $2.415 million. The example is intended only to verify the arithmetic and the interaction of the two components. In a regulatory filing M must be established under Article 163 and its annexed healthcare-use specifications; entering an independently chosen $120 does not make the resulting number legally valid. Nor does the calculation represent all economic costs of an actual pandemic, all types of policyholder illness, or every claim from the resulting health emergency.
The treatment of SLT obligations requires attention to benefit substance. EIOPA's pandemic applicability question and answer discusses how medical expense and income protection categories apply when an SLT line of business is not separately reported in these functional categories (EIOPA 2023). The source supports a granular mapping of underlying insured benefits. Its scope is a particular regulatory interpretation, and it cannot establish the eligibility of every individual contract without examination of the terms and current supervisory guidance.
Observed COVID-19 expenditure and insured loss
The historical COVID-19 episode exposes the limits of a one-directional cost narrative. A new pathogen can create intensive care demand, testing costs, quarantine-related income protection, mortality, and disability. The same period can reduce elective visits, defer preventive care, interrupt treatment, and suppress ordinary claims processing. Subsequent periods may include deferred treatment, changes in health status, policyholder behavior, provider prices, and premium or enrollment changes. A model that treats all these mechanisms as contemporaneous increases in one claims index fails to represent their temporal structure.
The 2020 national expenditure study by Hartman and colleagues (2021) documents the unusual split between public pandemic financing and declines in private insurer expenditure. The 2021 follow-up study by Martin and colleagues (2022) records rebound in healthcare use alongside a markedly different fiscal context. These are observations from the United States, not universal coefficients for pandemic claims capital. International application requires distinct health systems, provider reimbursement, contract coverage, public support, and insurance market structures.
The proper actuarial response is to distinguish event severity from insurer cost incidence. Hospital admission rates measure care need. Claims paid by the insurer measure covered and reimbursed care under a particular contract and financial arrangement. Capital measures the adverse change after expected liabilities and asset effects are accounted for. A pandemic can be grave in public health terms while producing heterogeneous insurer claims outcomes. Aggregate health expenditure can increase even when the private insurance component decreases. This difference is central to source criticism and to financial transmission modeling.
Scenario dimensions beyond disease prevalence
A robust health catastrophe scenario must specify an event pathway. For an infectious disease it may include attack rate, symptomatic fraction, hospitalization probability, service intensity, provider reimbursement, disability onset, claim duration, mortality, workforce absence, enrollment effects, operational disruption, and reinsurance response. An event tree can reveal where the insurer is actually exposed and where government or another party bears the expense. The branches can incorporate delayed healthcare use and uncertainty over benefit adjudication.
A stress should also consider clustering. Population-wide infection prevalence is a weak stand-alone proxy for concentration when the insured group is heavily represented among healthcare workers, emergency personnel, particular occupational settings, or regions experiencing high incidence. Geographic diversification works differently for infectious disease than for independent workplace accidents because an outbreak may affect many locations simultaneously. Multiple epidemics or recurring waves also create serial dependence, which may matter to a multi-year disability portfolio even if a one-year standard formula uses a single specified instantaneous event.
For accident catastrophes, a comparable scenario identifies the largest employment sites, travel concentrations, insured groups, and benefit limits. For medical expense catastrophes, it identifies the direct medical claims and potential follow-on treatment. Reinsurance attachments and event definitions should be included at the loss-construction stage, rather than applied as a generic percentage reduction after the aggregate catastrophe capital is estimated. The details determine whether a treaty responds to one event, several locations, multiple policy years, or only narrowly defined claims.
Capital and social loss
Insurer solvency and social protection overlap but are measured separately. A public health emergency can cause substantial mortality, household hardship, interrupted schooling, or public expenditure outside private insurance contracts. Health insurers' capital is designed to support the promises they legally bear. Evaluation of a national pandemic's social welfare loss is wider than underwriting capital. This distinction also explains why supervisory capital, public health preparedness, public programs, and insurer liquidity can require coordinated analysis while retaining separate measurement systems.
Risk Aggregation and Diversification
The square-root formula and its domain
Suppose an insurer has three already calculated health underwriting charges N, S, and C under the relevant legal basis. The standard Solvency II aggregation formula produces an outcome smaller than their arithmetic sum where cross-submodule diversification is recognized and the prescribed correlation matrix permits it. For a hypothetical example with N=$24 million, S=$18 million, C=$30 million and the legal Article 144 matrix, the result is
The unadjusted sum is $72 million, so the mechanical standard-formula aggregation benefit is approximately $18.502 million. This is an arithmetic statement for a hypothetical three-component vector. It is not an empirical estimate of diversification in a real insurer, and the assumed charges have not been derived from a real portfolio's statutory inputs. Reversing the order of operations or applying another undocumented correlation matrix would produce a different answer.
For illustration of sensitivity, replace the three off-diagonal correlations by 0.70 for N–S and 0.60 for each pair involving C, retaining the same three charges. The result becomes about $62.584 million, 16.99 percent above the original result. This is an internal stress test of the assumed dependence structure, never an authorized alteration of the legal standard formula. The illustrative stressed correlation matrix is positive definite: for these correlations its determinant is positive, and it can be used consistently in the quadratic calculation. The result demonstrates that diversification changes materially with dependence assumptions even when standalone risks do not change.
| Illustrative aggregation | Standalone charges N / S / C | Pairwise correlations N–S / N–C / S–C | Aggregate charge |
|---|---|---|---|
| Regulatory-matrix arithmetic | $24m / $18m / $30m | 0.50 / 0.25 / 0.25 | $53.498m |
| Stronger-dependence sensitivity | $24m / $18m / $30m | 0.70 / 0.60 / 0.60 | $62.584m |
| No diversification sum | $24m / $18m / $30m | Not applicable | $72.000m |
The formula makes its assumptions explicit, but a covariance-style aggregation is only one part of total insurer solvency. Health underwriting risk must be combined with market, credit, other underwriting, operational, and adjustment components in the applicable broader solvency architecture. The group structure and reinsurance counterparty exposures may change how diversification is recognized. The overall insurer capital cannot generally be inferred by adding the health SCR to a balance-sheet liability or by comparing the hypothetical result directly with NAIC Authorized Control Level RBC.
When dependence becomes the principal risk
Pearson correlation describes linear dependence and can be an inadequate summary of joint extreme events. Embrechts, McNeil, and Straumann (2002), in their analysis of correlation and dependence pitfalls, show why multivariate risk analysis must attend to distributional shape and dependence rather than treating correlation as a complete description. McNeil, Frey, and Embrechts (2015) provide a broader account of quantitative tail-risk modeling, including the statistical treatment of extreme loss distributions. Their analysis supports the scientific need to compare alternative multivariate specifications; it does not supply universal health catastrophe correlations.
In health insurance, one common factor may simultaneously increase acute medical claims, disability incidence, and claim duration. A recession may decrease employment-based enrollment, weaken premium receipts, and change the composition of insured risks. Provider-system congestion may both increase cost per case and defer routine treatment. Correlation measured during ordinary periods may understate the probability that these factors move together under a rare epidemic or catastrophic accident.
A rigorous internal loss distribution can be specified conditionally on a shared stress vector Z:
where Z contains common morbidity, price, utilization, financial and operational variables, while ε_j captures idiosyncratic risk. Conditional model assumptions determine whether independence remains after Z is specified. This equation is a standard probabilistic decomposition, introduced here as an analytical expression rather than a new theory. In a simulated stress, the insurer can estimate the frequency of large joint losses under an explicit scenario and inspect which common drivers dominate its upper tail.
The critical technical challenge is that rare health catastrophes supply limited insurer-specific observations. Selecting a copula family, tail-dependence coefficient, or pandemic event frequency may require expert judgment and external epidemiological evidence. Alternative assumptions should therefore be presented as conditional results. Precision in simulation output is different from precision in the historical evidence used to construct the distribution. The number of random draws does not eliminate parameter uncertainty.
Reinsurance and counterparty dependence
Reinsurance can reduce the retained amount of a morbidity or catastrophe shock if the contract covers the relevant loss events. Its effect depends on attachment and exhaustion points, per-event and aggregate limits, exclusions, reinstatement provisions, waiting periods, collateral, and the financial position of the reinsurer. The Solvency II SCR recognition of risk mitigation requires proper reflection of credit and other risks from recognized mitigation techniques. A gross claim reduction without corresponding counterparty and basis-risk analysis could overstate the insurer's residual resilience.
For an event loss X, a simplified excess-of-loss payment with attachment a and layer limit ℓ is R(X)=min(max(X−a,0),ℓ), subject to contract conditions. Retained loss is X−R(X) in a stylized single-event setting. The derivative of retained loss with respect to X is one below attachment, zero within a fully responding layer, and one again after exhaustion. Thus reinsurance changes the shape of the loss distribution rather than multiplying it by a uniform percentage. In a health catastrophe, correlated losses across insured groups or waves may exhaust annual aggregate covers even if any single event remains below a per-occurrence limit.
The insurer also faces counterparty default under extreme systemic shocks. If reinsurers are exposed to the same pandemic, liquidity disruption, or financial-market conditions, recovery certainty may decline when ceded claims increase. This creates a second-order dependence between underwriting severity and recovery credit quality. Reinsurance evaluations should distinguish gross loss, contractually ceded loss, collateralized recovery, timing mismatch, and residual risk; precise admissibility follows the applicable solvency framework.
ORSA and model validation
Solvency II's Article 45 own-risk assessment duties require firms to assess overall solvency needs, continuous compliance, and the materiality of differences between their risk profile and standard-formula assumptions (European Parliament and Council 2009). EIOPA's ORSA supervisory guidelines frame this as forward-looking assessment (EIOPA 2015). Health insurers therefore have a scientific reason to examine outbreak severity, disability persistence, medical inflation, reinsurance collectibility, operational resilience, and capital under alternative conditions beyond a single regulatory point estimate.
Model validation should separate data completeness, parameter estimation, conceptual specification, and outcomes monitoring. Medical claims triangles require consistent service and reporting dates. Disability data require robust recovery and termination coding. Catastrophe exposures require accurate concentrations and benefit mapping. Reinsurance data require actual treaty terms. A model may pass numerical reconciliation while failing a conceptual review if it attributes all post-pandemic cost growth to disease incidence or treats national expenditure as insurer claims.
Backtesting a one-year 99.5 percent capital measure is intrinsically difficult because exceptions are rare. A better validation program combines sensitivity analysis, scenario plausibility, multi-period diagnostics, peer comparison, and retrospective evaluation of intermediate claims drivers. Useful diagnostics include claim count by diagnosis, paid-to-incurred development, medical unit cost, utilization, disability continuance, net-to-gross loss, capital attribution, and stress coverage by contract. Adverse experience should lead to documented assumption revision without treating every historical outlier as evidence of a permanent new regime.
Health Market Application
Comparing legal capital measures without false equivalence
A practical cross-jurisdictional study begins with a common economic portfolio and translates its obligations into the categories and valuation rules of each jurisdiction. For example, consider a hypothetical insurer with annual renewable medical reimbursement, long-term income protection, a portfolio of group employees concentrated in two cities, and an excess-of-loss reinsurance treaty. The underlying insured events may be held constant for comparison, but the classification, balance-sheet measurement, eligible own funds, allowable mitigation, and supervisory capital thresholds will differ. A valid comparison therefore reconstructs each regulatory result from its own legal basis.
The EU approach identifies NSLT, SLT, and catastrophe charges; the U.S. statutory approach applies health RBC risk factors and supervisory intervention thresholds; the IAIS capital standard informs comparison for internationally active groups; the Swiss approach uses its economic balance-sheet measurement. The NAIC description of RBC limits and uses explicitly warns against treating RBC as a stand-alone measure of an insurer's full solvency. The ICS purpose and scope likewise place the IAIS measure in the international group-supervision context. A capital amount meaningful within one formula cannot simply be divided by another system's threshold and interpreted as a ranking of health insurers.
A careful comparison records the legal effective date and type of institution. U.S. health RBC applies to designated risk-bearing entities under relevant state implementation and reporting rules. Third-party administrators may process claims without assuming the underlying insurance risk. The NAIC 2026 adopted RBC modifications and its health capital working agenda demonstrate that current year-end instructions and proposed formula adjustments are version-sensitive (NAIC 2026). Statements concerning a proposed H-2 change should not be confused with a fully effective formula change. In the EU, a calculation as of November 2026 must likewise be distinguished from changes scheduled for January 2027.
An integrated case study of morbidity, capital, and financial response
Consider a generic risk-bearing insurer with annual medical premium income, an existing disability book, and geographic concentration of group income protection. The starting model estimates medical frequency and severity, open-claim reserves, disability onset and recovery, servicing expense, and the value of legally recognized reinsurance. An external shock has three stylized elements: a higher rate of work-disabling illness, a delayed increase in healthcare service utilization, and an increase in provider charges. The effects are separated by accounting period and by type of contractual benefit.
The first financial transmission occurs through incurred medical claims. A change in hospital admission frequency increases current-period utilization, but a simultaneous fall in elective treatments can offset some immediate claim volume. The second transmission occurs through income protection. New disability cases increase expected claims, while slower recovery extends benefit continuance and increases existing claim liabilities. The third is financial: rising servicing expenses, premium or enrollment shortfalls, reinsurance payments, and higher credit exposure affect own funds. These mechanisms can occur together yet be measured at different stages of the insurer's statement.
The illustrative disability calculation developed above isolates the incidence and recovery channels. Under its stated assumptions, combined higher incidence and lower recovery increase expected cohort payments from $10.300 million to approximately $16.359 million. Separately, the stylized EU health module aggregation shows that charges of $24 million, $18 million, and $30 million combine to approximately $53.498 million under Article 144's correlation matrix. The first calculation is a liability sensitivity based on hypothetical transition probabilities; the second is a module aggregation based on assumed standalone charges. They are not to be added as if they were independent components of an actual regulator-certified SCR. The disability stress may already contribute to an SLT standalone charge, and accounting deterioration may already be reflected in a change in own funds.
This distinction is a central control against double counting. Once expected liabilities are calculated using the stressed assumptions, an internal capital simulation should consistently measure additional unexpected changes relative to the same valuation basis. If the same disability shock is used to increase the best estimate and to create a standalone capital charge, documentation must show how these results fit in the prescribed one-year loss framework. Likewise, if catastrophic medical payments are included in an explicit health catastrophe module, an additional generic pandemic multiplier applied to all health claims can duplicate the shock. The correct procedure is contract-level identification followed by risk-specific and capital-framework-specific aggregation.
A comparable research table
| Analytical dimension | EU Solvency II | United States health RBC | IAIS ICS | Swiss SST |
|---|---|---|---|---|
| Primary regulatory use | One-year risk-sensitive insurer solvency requirement | Statutory capital trigger for progressive intervention | Global group-level capital comparability for internationally active groups | Risk-based economic solvency measurement and supervision |
| Health underwriting representation | NSLT, SLT, health catastrophe modules | Health-formula underwriting components including H-2; reporting-year specific | Quantitative group capital risk classification under adopted ICS documentation | Insurer-specific risk analysis within total-balance-sheet SST framework |
| Capital measurement | SCR calibrated to 99.5% one-year basic-own-funds VaR | Total adjusted capital measured against Authorized Control Level RBC | ICS capital adequacy as implemented by applicable jurisdiction | Risk-bearing capital compared with target capital |
| Main comparison risk | Mistaking legal standard-formula correlations for empirical tail dependence | Treating supervisory thresholds as EU-style SCR confidence levels | Assuming standard adoption means uniform local implementation | Assuming the Swiss model duplicates EU modules |
| Recommended additional analysis | ORSA and stress testing | Firm-specific stress testing alongside RBC reporting | Group aggregation and relevant implementation assessment | Scenario and model analysis consistent with SST rules |
This table summarizes legal purposes rather than providing complete formula instructions. It does not assume that all supervisory jurisdictions define health risk categories in the same way. The FINMA SST account, EIOPA's SCR calibration text, NAIC's RBC explanation, and IAIS's adopted ICS description establish the institutional scopes independently.
Model construction and documentary reproducibility
A reproducible underwriting-capital study records every step between original observations and the reported capital output. It begins with an exposure register containing insured lives, benefit definitions, inception and expiry dates, and risk-bearing legal entities. Claims observations then require event or service date, reporting date, paid amount, ultimate estimate, diagnosis or benefit category where lawful, disability start and termination, recovery, and contract flags. Data protection and minimization requirements determine how sensitive personal health information may be processed and retained.
The next step is to construct a baseline by claim type, with separate frequency, severity, duration, and utilization models. Estimates should retain statistical uncertainty, possible reporting changes, and benefit restrictions. A capital analysis should include the actual prescription of the applicable rules before adding internal stress scenarios. For EU NSLT risk this includes prescribed segment parameters and volume measures. For SLT risk it includes legally defined valuation shocks. For catastrophe risk it includes the prescribed benefits and concentration structure. For U.S. health RBC it includes the correct current reporting exhibits and factors. For a broader internal model it includes documented distributions and dependence assumptions.
The model must reconcile gross and net financial results. Reinsurance, risk equalization, management actions, and governmental transfers may affect retained losses if the relevant legal criteria are met. Contractual claims that may fail to trigger a reinsurance recovery need basis-risk treatment; future underwriting actions need timing and implementability assumptions. Observed claim data must be reconciled against ledger values and regulatory balance-sheet categories. These controls address the common failure in which technically precise modeling is performed on economically misclassified exposures.
A credible report includes assumptions, data vintages, legal sources, calculation version, sensitivity ranges, governance decisions, and residual uncertainties. The ASOP No. 5 actuarial guidance supports transparency about claims estimation methods and communications; EIOPA's ORSA guidance supports a forward-looking examination of an insurer's actual risk profile. These sources operate in different legal and professional scopes, but both emphasize the relevance of an analysis to the obligations being measured.
Stress testing beyond an isolated annual loss ratio
A scientifically useful test program varies medical incidence, service use, claim severity, benefit continuation, inflation, policyholder behavior, reinsurance response, investment conditions, and operational disruption. Each scenario should state which variables are shocked simultaneously and why. A high-morbidity scenario with lower elective use may produce a smaller current medical claims ratio yet a larger long-term disability liability. A medical-cost inflation scenario may increase present values even without more diagnoses. A concentrated workplace accident may be severe for group disability coverage but comparatively modest for a large diversified medical expense book. A pandemic scenario can combine the features of several of these cases.
For multi-year liabilities, the stress evaluation should trace the policy cohort through time and distinguish immediate losses, reserve remeasurements, subsequent premium adjustments, and later cash settlements. Solvency resources can also be affected by changes in asset values and required capital simultaneously. The insurer's solvency ratio therefore reflects both numerator and denominator effects. A fixed capital charge viewed in isolation can understate the effect of simultaneous own-funds deterioration.
A stress test does not become sound by assigning extreme values without empirical justification. Plausible ranges should draw from claims experience, official public health and expenditure records, validated models, expert review, and relevant regulatory prescriptions. Where evidence is thin, the report should show multiple conditional outcomes instead of manufacturing a definitive probability for a hypothetical catastrophe. This is particularly important for long-duration disability claims following new illnesses, for which a complete historical outcome distribution may not yet exist.
Reinsurance, governance, and capacity allocation
Underwriting capacity decisions combine expected performance, capital consumption, reinsurance availability, and the institution's authorized risk tolerance. If a risk-bearing insurer increases the proportion of concentrated employer disability exposure, its expected premium revenue can rise alongside correlation and catastrophe exposure. The appropriate evaluation measures marginal contribution to the total risk portfolio, not simply the product's standalone loss ratio. In a nonlinear capital framework, marginal capital depends on the pre-existing mix of risks and the correlation assumptions. Diversification can lower aggregate capital, but concentration can overwhelm nominal gains.
A portfolio manager might compare two proposed expansions that have equal expected medical claims but different geographic and occupational exposure. One creates many independent small accounts; the other adds a large employer in a location already heavily represented. Under a common accident or epidemic scenario, the second can contribute much more to tail loss. Conversely, stable reinsurance with suitable limits can change the retained distribution. These conclusions follow from conventional aggregation and contract analysis; they require no new named principle.
Supervisory intervention rules reinforce the need to study outcomes rather than capital labels alone. The NAIC's RBC regulatory-action overview describes progressively stronger responses to decreasing capital ratios. In the European Union, the SCR is part of a broader governance and supervisory architecture. In internationally active groups, the IAIS's adoption and phased implementation explanation makes clear that international standard-setting and legally operative national implementation are separate institutional stages. The practical output of capital research should therefore identify the actual rule, the affected regulated entity, and the consequences of breaching the applicable threshold.
Source limits and scientific boundaries
The documentary base is strongest for the wording of published regulatory formulas and the conceptual requirements of professional standards. The aggregate U.S. expenditure studies provide direct evidence for 2020–2021 financing and utilization patterns, with geographically specific scope. Financial loss distributions for novel, rare health catastrophes remain less directly observable; no empirical frequency estimate is claimed here. Publisher material for quantitative risk management supports general dependence modeling, not a specific health SCR parameter. The hypothetical tables are mathematical illustrations and cannot be used to infer a real firm's capital adequacy.
Several research boundaries follow. Disability duration requires longitudinal claimant data, while population morbidity uses broader incidence and severity information. National health spending is an institutional aggregate that can shift between private and public payers. U.S. capital triggers and the EU SCR have different regulatory functions. The 2027 EU reform has a known future application date relative to this manuscript's cutoff. Full empirical validation for a specific company would require confidential insurer data, current jurisdictional filing instructions, treaty information, and prospective model governance. The absence of such data is a limitation on company-specific numerical inference, rather than an obstacle to establishing the legal, mathematical, and comparative scientific results of this monograph.
Conclusion
Health underwriting risk capital is a measure of an insurer's vulnerability to adverse morbidity-related financial outcomes under a specified legal or analytical framework. Its object is the liability-bearing insurer's financial position, reconstructed from covered benefits, expected cash flows, one-year or other prescribed risk horizons, the distribution of shocks, and recognized risk transfer. The scientific connection among these elements can be established without claiming a new theory.
The comparative investigation yields six substantive findings. First, insured health outcomes enter capital calculations through contract and valuation classifications; diagnosis counts alone cannot determine required capital. Second, the separation of EU NSLT, SLT, and health catastrophe modules captures economically different routes by which a health risk becomes a financial loss, although the prescribed correlations are not evidence of actual portfolio tail dependence. Third, medical inflation and utilization can change independently of illness incidence; delayed treatment and shifting payment responsibility make pandemic claims trajectories heterogeneous. Fourth, long-duration disability exposure requires explicit incidence and continuance analysis, because a persistent recovery deterioration can substantially change present values. Fifth, reinsurance changes the distribution of retained losses in a contract-specific way while introducing credit and basis risks. Sixth, international comparison requires preservation of each regime's own capital definitions, supervisory functions, effective dates, and institutional scope.
The worked calculations show how these findings operate. A hypothetical disability cohort grows from $10.300 million to $16.359 million in expected discounted benefits under combined stated shocks; separate hypothetical health charges aggregate from a $72 million standalone sum to approximately $53.498 million under the EU prescribed matrix. Neither result is a real insurer assessment. The verified algebra illustrates the necessity of avoiding double counting and of distinguishing baseline liability deterioration from unexpected capital losses.
The central analytical conclusion is that health underwriting solvency must be evaluated as a coupled contractual, epidemiological, actuarial, and institutional process. A prescribed capital formula creates a consistent supervisory measure. The scientific task is to reconstruct the exposures within that measure, test the assumptions outside it, and disclose the pathways through which health events, claims obligations, financial markets, and risk-sharing institutions jointly determine an insurer's capacity to honor promised benefits.
References
Actuarial Standards Board. 2017. Actuarial Standard of Practice No. 5: Incurred Health and Disability Claims. Revised March 2017, effective September 1, 2017. https://www.actuarialstandardsboard.org/asops/incurred-health-and-disability-claims_186/
Centers for Medicare & Medicaid Services. 2026. National Health Expenditure Projections: 2025–2034. Office of the Actuary, National Health Expenditure Data, projected series. https://www.cms.gov/data-research/statistics-trends-and-reports/national-health-expenditure-data/projected
Embrechts, Paul, Alexander J. McNeil, and Daniel Straumann. 2002. “Correlation and Dependence in Risk Management: Properties and Pitfalls.” In Risk Management: Value at Risk and Beyond, edited by M. A. H. Dempster, 176–223. Cambridge University Press. https://doi.org/10.1017/CBO9780511615337.008
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35 of 10 October 2014 Supplementing Directive 2009/138/EC: Article 144, Health Underwriting Risk Module. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5764_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 145, NSLT Health Underwriting Risk Sub-module. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5765_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 146, NSLT Health Premium and Reserve Risk Sub-module. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5766_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 147, Volume Measure for NSLT Health Premium and Reserve Risk. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5767_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 148, Standard Deviation for NSLT Health Premium and Reserve Risk. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5768_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 149, Health Risk Equalisation Systems. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5769_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 151, SLT Health Underwriting Risk Sub-module. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5771_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 154, Health Disability-Morbidity Risk Sub-module. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5774_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 155, Medical Expense Disability-Morbidity Capital Requirement. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5775_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 156, Income Protection Disability-Morbidity Capital Requirement. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5776_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 157, Health Expense Risk Sub-module. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5777_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 160, Health Catastrophe Risk Sub-module. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5780_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 161, Mass Accident Risk Sub-module. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5781_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 162, Accident Concentration Risk Sub-module. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5782_en
European Commission. 2015. Commission Delegated Regulation (EU) 2015/35: Article 163, Pandemic Risk Sub-module. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5783_en
European Insurance and Occupational Pensions Authority. 2015. Guidelines on Own Risk and Solvency Assessment (ORSA). Published September 14, 2015. https://www.eiopa.europa.eu/publications/guidelines-own-risk-solvency-assessment-orsa_en
European Insurance and Occupational Pensions Authority. 2023. Question 2799: Pandemic Risk Sub-module Applicability to Similar-to-Life Health Obligations. Solvency II Q&A database. https://www.eiopa.europa.eu/qa-regulation/questions-and-answers-database/2799_en
European Insurance and Occupational Pensions Authority. 2026. Solvency II: Regulation and Policy. Includes implementation timeline for Directive (EU) 2025/2. https://www.eiopa.europa.eu/browse/regulation-and-policy/solvency-ii_en
European Insurance and Occupational Pensions Authority. 2026. “EIOPA Completes Solvency II Review Mandate with Final Guidelines and Draft Technical Standards Before Revised Framework Takes Effect Early Next Year.” July 15, 2026. https://www.eiopa.europa.eu/eiopa-completes-solvency-ii-review-mandate-final-guidelines-and-draft-technical-standards-revised-2026-07-15_en
European Parliament and Council. 2009. Directive 2009/138/EC on the Taking-Up and Pursuit of the Business of Insurance and Reinsurance (Solvency II): Article 101, Calculation of the SCR. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-2188_en
European Parliament and Council. 2009. Directive 2009/138/EC: Article 105, Calculation of the Basic Solvency Capital Requirement. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5974_en
European Parliament and Council. 2009. Directive 2009/138/EC: Article 45, Own Risk and Solvency Assessment. EIOPA Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-2128_en
FINMA (Swiss Financial Market Supervisory Authority). 2024. “Swiss Solvency Test (SST).” Supervisory guidance, updated January 22, 2024. https://www.finma.ch/en/supervision/insurers/cross-sectoral-tools/swiss-solvency-test-sst/
Hartman, Micah, Anne B. Martin, Benjamin Washington, Aaron Catlin, and the National Health Expenditure Accounts Team. 2021. “National Health Care Spending in 2020: Growth Driven by Federal Spending in Response to the COVID-19 Pandemic.” Health Affairs. Published December 15, 2021. https://www.healthaffairs.org/doi/full/10.1377/hlthaff.2021.01763
IFRS Foundation. 2023. IFRS 17 Insurance Contracts. Effective for annual reporting periods beginning on or after January 1, 2023. https://www.ifrs.org/issued-standards/list-of-standards/ifrs-17-insurance-contracts/
IFRS Foundation. n.d. “IFRS 17 Insurance Contracts—Key Terms.” Accessed November 14, 2026. https://www.ifrs.org/supporting-implementation/supporting-materials-by-ifrs-standards/ifrs-17/key-terms/
International Association of Insurance Supervisors. 2024. “IAIS Adopts Insurance Capital Standard and Other Enhancements to Its Global Standards to Promote a Resilient Insurance Sector.” December 5, 2024. https://www.iais.org/2024/12/iais-adopts-insurance-capital-standard-and-other-enhancements-to-its-global-standards-to-promote-a-resilient-insurance-sector/
International Association of Insurance Supervisors. n.d. “Insurance Capital Standard.” Standard-setting overview, accessed November 14, 2026. https://www.iais.org/activities-topics/standard-setting/insurance-capital-standard/
Martin, Anne B., Micah Hartman, Joseph Benson, Aaron Catlin, and the National Health Expenditure Accounts Team. 2022. “National Health Care Spending in 2021: Decline in Federal Spending Outweighs Greater Use of Health Care.” Health Affairs. Published December 14, 2022. https://www.healthaffairs.org/doi/full/10.1377/hlthaff.2022.01397
McNeil, Alexander J., Rüdiger Frey, and Paul Embrechts. 2015. Quantitative Risk Management: Concepts, Techniques and Tools. Revised edition. Princeton University Press, 720 pages, ISBN 9780691166278. https://press.princeton.edu/books/hardcover/9780691166278/quantitative-risk-management
National Association of Insurance Commissioners. 2026. “Risk-Based Capital.” Insurance Topics, updated June 30, 2026. https://content.naic.org/insurance-topics/risk-based-capital
National Association of Insurance Commissioners. 2026. “Capital Adequacy (E) Task Force: Adopted Modifications to Risk-Based Capital Formulas.” Updated November 2026. https://content.naic.org/capital_adequacy_task_force.htm
National Association of Insurance Commissioners. 2026. “Capital Adequacy (E) Task Force Working Agenda.” Revised November 2026. https://content.naic.org/cmte-e-capad-working-agenda