Insurance by BogdanovaThe Encyclopedia of Insurance and Risk Science

RESEARCH DOSSIER

Disability Income Insurance Reinsurance and Risk Concentration

Abstract. Disability income insurance substitutes contractually defined cash benefits for earnings lost when illness or injury restricts work. Reinsurance transfers specified components of the insurer's resulting financial obligation, while leaving the original insurer exposed to policyholder duties, contract interpretation, retained disability experience, payment timing, counterparty performance, and basis differences. This monograph investigates the joint determination of reinsurance design and risk concentration in long-duration disability income portfolios. The inquiry combines primary actuarial experience studies, international supervisory principles, accounting and legal instruments, an established empirical literature on disability incentives, and transparent actuarial calculations. It distinguishes incidence among exposed workers from claim termination among disabled lives; separates recovery and death; and treats occupation, industry, geography, diagnosis, benefit formula, and employment arrangements as interacting exposure dimensions. Proportional, excess-of-loss, aggregate, and modified-coverage structures are compared against distributional and operational stress. The analysis shows that reinsurance efficiency depends on aligning the contractual loss definition, financial attachment point, covered cohort, reporting period, and settlement timing with the actual drivers of aggregate liability. Concentrated occupational employment, common economic shocks, claim-duration deterioration, and correlated reinsurer exposures can reduce the diversification apparently achieved by spreading individual policies. A reproducible set of examples demonstrates correlation sensitivity, disability reserve duration sensitivity, recovery ceilings, and counterparty dependence without representing hypothetical values as observed market estimates. The resulting synthesis identifies the data and governance conditions under which reinsurance can improve net-risk capacity, liquidity, and capital resilience while preserving claimant entitlements. The conclusion sets out a source-grounded research agenda for joint incidence-duration estimation, cross-portfolio dependency, treaty-performance testing, and cross-jurisdiction legal comparability. No new actuarial theory or regulatory doctrine is proposed.

Keywords: Disability income insurance; long-term disability; group disability; individual disability; reinsurance; quota share; excess of loss; aggregate stop loss; claims incidence; claim duration; recovery; mortality; occupational concentration; geographic correlation; tail dependence; reinsurance recoverables; counterparty default; capital adequacy; IFRS 17; insurance enterprise risk management.

Introduction

Disability income insurance creates a financial promise whose economic content unfolds over time. An accident, illness, or other covered impairment can produce an onset date, a waiting period, a first eligible claim, periodic payments, partial work, recovery, recurrence, death, or exhaustion of contractual benefits. Liability also responds to the insured's occupation and earnings, employer benefit arrangements, statutory offsets, claim administration, investment conditions, and the market availability of reinsurance. A credible account of risk concentration must therefore analyze events and payment streams together. Counting disabled persons, insured claims, new disability awards, and outstanding monthly benefits as interchangeable observations obscures the object of insurance measurement.

The WHO classification of functioning (World Health Organization, 2001) situates functioning and disability within activities, participation, health conditions, and environmental context. Its definitions explain why an impairment does not uniquely determine entitlement under a private disability policy. A contract specifies covered earnings loss, occupational inability, proof requirements, and benefit duration. Public programs likewise adopt legal tests specific to their statutory purposes. Section 223 of the United States Social Security Act uses substantial gainful activity and a statutory duration threshold for Social Security Disability Insurance; a private own-occupation or residual-disability promise may operate differently. Reinsurance follows the underlying insured risks only insofar as the treaty and its legal interpretation provide.

This distinction is operationally decisive. A ceding insurer may report a stable count of new claimants while expected future benefits rise because recoveries slow, claimants stay disabled for longer, earnings-indexed benefits increase, or several occupational groups move adversely at once. A reinsurer may assume a percentage of reported claims and still leave the cedent carrying expenses, disputed claims, unreported exposures, benefit escalation, or recovery delays. Conversely, a treaty focused on severe portfolio losses may stabilize the cedent's tail outcome without purchasing coverage for routine morbidity fluctuations. Each structure must be assessed on its particular retained-risk distribution, cash-flow timing, enforceable terms, and cost.

The research method has four components. First, documentary interpretation reconstructs covered disability, payment, and reinsurance obligations from official experience-study methods, regulatory rules, and published accounting material. Second, actuarial analysis states transparent incidence, multi-state duration, present-value, and aggregate-dependency relationships. Third, contract comparison examines the consequences of proportional and nonproportional cessions, retention and limits, collateral, and counterparty security. Fourth, critical validation separates identifiable evidence from illustrative assumptions and documents the absence of insurer-specific treaty and individual claimant records. The primary statistical anchors are the 2015–2022 group incidence study (SOA Research Institute and LIMRA, 2025) and the 2015–2024 claim termination study (SOA Research Institute and LIMRA, 2026). Their different populations and risk sets are central to the interpretation.

The geographical scope is international, with close examination of US experience and legal materials because those sources give unusually explicit public evidence. International supervisory and accounting principles serve as comparative reference points; their appearance in this study does not imply universal implementation or identical national legal consequences. The empirical literature on public disability benefits supplies evidence about behavioral and institutional mechanisms, with transferability to private insurance assessed rather than presumed. All numerical portfolio examples below are hypothetical and reproducible; no private insurer dataset, confidential treaty, or independent causal experiment was available for this study.

Conceptual Foundations and Historical Development

Disability as insured earnings interruption

The insured event is a contractually specified interruption or impairment of earnings capacity. Covered disability may be defined in relation to the insured's own occupation, a suitable occupation, or any occupation; an impairment may qualify as total, partial, residual, recurrent, or presumptive depending on the contract. The observable claim is a request and adjudicated status under those provisions. The payment obligation is a time-indexed cash flow generated after applicable waiting periods, benefit conditions, offsets, and maximum limits. These distinctions separate the medical and functional event from administrative recognition and monetary settlement.

The WHO's ICF framework links health condition, activity, participation, and environment (WHO, 2001). Disability income contracts transform part of this multidimensional reality into a legally administrable earnings-risk promise. Work capacity depends partly on occupational duties, workplace accommodation, available alternative work, and labor-market circumstances. The same impairment can consequently produce different insured outcomes for a manual worker, an office worker, a highly specialized professional, or a worker with flexible accommodations. This is a causal reason to record occupation and job demands explicitly rather than treating diagnostic labels as sufficient measures of benefit exposure.

Individual disability income and employer-sponsored group long-term disability business also have different selection, contract, premium, and persistency mechanisms. Individually underwritten cover can concentrate high benefit amounts among particular professional occupations. Group plans can acquire broad populations through employment but remain dependent on employer selection, employer size, wage composition, and business-cycle dynamics. Self-funded employer arrangements administered by insurers must be separated from risk actually insured on an insurer's balance sheet. Public social insurance, workers' compensation, sickness absence benefits, and private income replacement interact through eligibility and offsets but cannot simply be pooled into one statistical class.

Development of experience measurement

Modern disability reserving distinguishes incidence from the time needed to terminate a recognized claim. The Society of Actuaries and American Academy of Actuaries' 2013 IDI valuation table documents the transition from earlier individual disability reserving bases; the SOA states that the NAIC approved the 2013 table in August 2016 as a statutory minimum morbidity basis. The joint work group's 2012 group LTD valuation table represents a separate group long-term disability line of development. These tables are valuation foundations with specified data histories, adjustment conventions, and national regulatory settings. They are not automatically the best estimate for every contemporary insured cohort or every other jurisdiction.

The SOA/LIMRA group LTD incidence results describe approximately 294 million life-years exposed and 1.2 million claims for 2015–2022, contributed by 19 companies. The report explicitly separates elimination periods, industry groups, case sizes, geographic areas, and diagnoses. The 2026 group LTD termination study covers 2015–2024 experience from 18 companies, exceeding 43 million life-months of disabled-life exposure with approximately 937,000 recoveries and 181,000 deaths. Those are study volume statistics, not population disability prevalence estimates, cross-country incidence rates, or an automatic basis for projecting unrelated contracts. They also do not constitute an estimate of correlation between occupational groups or insurers.

Recent individual-claim studies illuminate another essential dimension. Beal and Corcoran's 2024 diagnosis analysis separately studies recovery and mortality by diagnostic category, gender, occupation class, and onset age. Their 2025 expanded analysis extends the 2006–2014 individual disability claim database to additional elimination periods and benefit periods. These works support decomposition of termination mechanisms and demonstrate why one blended termination parameter can conceal materially different outcomes. They describe an earlier insured experience period, and their predictive relevance requires explicit reconciliation with current contract and claims mix.

Legal and economic development of the reinsurance relationship

Reinsurance is an indemnity relationship between insurer and reinsurer. The policyholder normally remains entitled to pursue the original insurer according to the insurance contract. The International Association of Insurance Supervisors (IAIS) explains in Insurance Core Principle 13 that risk transfer reduces selected insurance exposure while introducing credit, operational, and basis risks. The distinction between the economic reallocation of risk and the continuing legal position of the ceding insurer is central for disability benefits, which may remain payable over years.

Disability reinsurance developed within wider life, health, accident, and group-benefit reinsurance practice. Contemporary contracts may transfer a proportional part of premium and benefits, provide specific protection on unusually costly claims, or cap aggregate portfolio deterioration. Historical market labels cannot substitute for reading the treaty: the operative scope is the cession clause, eligibility of policies, defined covered claims, accounting basis, attachment and exhaustion, claims-control provisions, and duration of the reinsurer's responsibility. An apparently identical percentage can lead to different transfers if the agreement applies to newly incurred claims, paid claims, outstanding reserves, or a restricted cohort of claims at specified attachment dates.

The scientific problem of concentration

An insurer gains statistical stabilization by pooling sufficiently diversified risks. Concentration persists when insured lives share an employer, occupation, industry, geographic area, healthcare access constraints, income level, disease exposure, policy design, claims administrator, or a macroeconomic environment. A recession, pandemic, occupational hazard, or adjudication change may influence many policies together. Concentration can also occur outside underwriting: reliance on one reinsurer, a shared collateral custodian, a single claims data provider, or concentrated investment assets can create common failure channels.

Three dimensions require distinct measurement. Exposure concentration records how much earnings benefit or present-value liability sits within a segment. Dependency records the joint behavior of incidence, recovery, severity, and durations across segments. Counterparty concentration records the recoverable balance and settlement dependence on each reinsurance group. None can be inferred uniquely from the others. A geographically dispersed insured book may be concentrated in one industry; a well-spread insured book may become dependent on one reinsurer. This analytical distinction organizes the chapters that follow.

Disability Treaty Structure

The ceded object and its accounting basis

A disability treaty must establish which claims and liabilities are eligible for transfer. Identification begins with policy form, employer or individual status, issue and coverage dates, disability definition, occupational class, insured income, maximum monthly benefit, elimination period, benefit period, offsets, inflation provisions if applicable, and claim-incurrence or notification rules. The treaty must then specify how those policy obligations become ceded claims. Common variants use premiums and losses on a quota-share basis, claims exceeding a per-person retention, or aggregate losses exceeding a portfolio attachment. They may differ in accounting year, occurrence year, policy year, claims-made reporting, and run-off obligations.

Coverage of incidence does not imply identical coverage of subsequent deterioration. For a disability claimant admitted under a treaty, the reinsurer's obligations might follow all qualifying future payments, terminate when the treaty expires, continue subject to a claims-in-course clause, or be commuted by agreement. The insurer must separately establish the recoverable for outstanding claims, including incurred but not reported or admitted late claims when applicable. Contractual language governing offsets, recurrent disability and changes from own-occupation to any-occupation eligibility can materially shift projected recoveries without changing the headline cession rate.

A policy cash flow can be represented by a contractual payment function bi(t;wi,oi,di,zi)b_i(t;w_i,o_i,d_i,z_i), where wiw_i is eligible earnings, oio_i the occupation and occupational definition, did_i the disability and claim state, and ziz_i the relevant offsets, waiting periods, limits, and contractual conditions. Let Ii(t)I_i(t) indicate coverage and entitlement at time tt. Then gross benefits over a horizon TT equal

GT=∑i=1n∫0TIi(t)bi(t) dt.G_T=\sum_{i=1}^{n}\int_0^T I_i(t)b_i(t)\,dt.

The notation is continuous-time convenience; actual contracts commonly pay on discrete monthly schedules. A treaty's recoverable is a contract-specific function RT=R(G,cohort,terms,timing,limits)R_T=\mathcal{R}(G,\text{cohort},\text{terms},\text{timing},\text{limits}). The ceding insurer's net insured benefits, before other costs, are NT=GT−RTN_T=G_T-R_T if the recoverable is fully collectible on the stated settlement basis. An amount contractually recoverable and an amount actually collected at the valuation date are distinct observations.

Proportional cessions

In a pure quota-share illustration, the reinsurer receives a specified fraction of premium and reimburses the corresponding fraction of eligible claims. For qualifying losses GG and a constant share qq, the nominal ceded claims are R=qGR=qG and the retained claims are (1−q)G(1-q)G, subject to treaty exclusions and accounting adjustments. Such a structure can relieve expected and unexpected loss together. It does not automatically eliminate all underwriting, operational, or settlement risk; neither does it ensure that every benefit component is eligible under the same percentage.

Consider a wholly hypothetical block of 20,000 insured workers. If the annual onset probability is 0.005 and every newly admitted claim is represented by an undiscounted expected aggregate payment of $72,000, the simple expected new-claim count is 100 and the associated undiscounted expected ultimate benefits are $7.2 million. A 40% quota share would attribute $2.88 million to the reinsurer and $4.32 million to the cedent before premiums, expenses, discounts, and credit losses. This example deliberately collapses an entire claim-duration distribution into a fixed conditional mean; it demonstrates the arithmetic of cession rather than an actuarial estimate. In practice, the claim payment profile, policy term, offsets, reserving basis, and treaty coverage determine the values.

For proportional treaties, concentration analysis should assess whether a proportional cession covers the same employer and occupation mix as the underlying book. If high-earning occupations or certain employers are excluded, gross cession percentage overstates the relief on the portfolio's largest conditional benefits. A minimum per-life retention or maximum reinsurer share may produce nonlinear effects, particularly for individual disability contracts with large monthly benefits. Claims-control clauses, reports, audit rights, and access to diagnosis and occupation data can affect both price adequacy and actual recoverability.

Specific and aggregate excess-of-loss protection

Per-risk excess of loss indemnifies covered claims above a stated retention and below a treaty limit. Its suitability depends on the unit of loss. For one claim, the attachment may be defined as a fixed aggregate sum, a present value, or an amount of cumulative paid benefits; these bases have sharply different behavior for a long-term payment annuity. An insurer may retain the early months and cede later months, but the timing of attachment can itself become sensitive to recovery, inflation, and claim settlement practice.

An aggregate stop-loss form pays on aggregate qualifying claims beyond an attachment, up to a limit: R=min⁡{max⁡(G−A,0),L}R=\min\{\max(G-A,0),L\}, where AA is the stated attachment and LL the maximum recoverable under the modeled period and coverage definition. In a purely illustrative year with $9.36 million of eligible ultimate claim benefits, a $6 million attachment and a $4 million layer produce $3.36 million of reimbursement and a net of $6 million. If the same losses instead reach $11 million, reimbursement is capped at $4 million and the cedent retains $7 million. The outcomes do not include reinsurance premium, reimbursement lag, claim expenses, or credit impairment.

Loss aggregation must respect claim cohort and development. Paid-loss stop loss may respond slowly to deterioration in claim reserve estimates; incurred-loss cover may respond earlier but requires precise definitions of case reserve, IBNR, reserve adjustments, and audit. An occurrence-based disaster cover may be unsuitable for multiple unrelated disability onsets whose commonality is economic rather than one insured event. A clause limited to a listed peril can exclude substantial occupation-linked deterioration. A treaty intended for tail stabilization therefore needs event wording and loss aggregation rules matched to the hypothesized stress mechanism.

Treaty mechanisms for long-tailed liabilities

Long-duration disability claims present longevity-like payment persistence conditional on continued disability, alongside the competing probability of death. This creates several operational questions: does the treaty cover future payments after nonrenewal; may either party commute reserves; which reserve basis defines settlement; how are disputes over disability eligibility handled; and what happens when benefit offsets or public payments are later adjusted? The answers determine whether the cedent retains tail risk despite a seemingly substantial cession.

A contract can allocate reporting rights and claims-control authority without transferring the insurer's direct duties toward insureds. Administrative cooperation must preserve fair claim evaluation, appropriate handling of personal health data, and documented appeal rights. In the US employer-plan setting, the Department of Labor's ERISA disability claims guidance illustrates that legal claims procedures attach to qualifying plans, not to a reinsurer's internal reporting convenience. International comparison must use the governing law of the actual policy and treaty.

Contract selection and basis mismatch

Four tests make treaty comparisons more meaningful. The event test asks which disability triggers and diagnoses the reinsurance covers. The exposure test asks which policy years, insured cohorts, occupations, and employers are included. The cash-flow test asks which payments or reserve movements are eligible and when reimbursement occurs. The counterparty test asks whether the reinsurer can and will perform at precisely the times the cedent experiences elevated insured payments. Differences across these tests constitute basis risk; they must be documented independently of the percentage of apparent risk cession.

These tests are the application of established contract analysis and IAIS ICP 13 to disability benefits. They provide an analytic checklist for this study, not a newly asserted regulatory doctrine. Treaty arrangements should be evaluated under the actual provisions, including exclusions, aggregate caps, reinstatements, profit commissions, experience refunds, sliding-scale ceding commissions, collateral thresholds, and commutation options that can materially redistribute outcomes under stress.

Occupational and Geographic Correlation

Exposure units and the geography of earnings protection

An occupational code, an industrial classification, and a geographic location represent different information. Occupation identifies the tasks a worker performs and the potential effect of functional limitations on those tasks. Industry identifies the employer's productive activity and its exposure to financial, technological, regulatory, and employment shocks. Geography may describe residence, worksite, corporate headquarters, medical provider network, or the place whose law governs the insurance contract. When the insurer's recorded geography is only the employer's headquarters, a supposedly regional portfolio may actually span multiple work locations. Conversely, individually distributed policyholders may share an industrial employer concentrated in one metropolitan labor market.

A coherent exposure record therefore has a policyholder or group identifier, an employer-group link when applicable, occupational and industry classifications, location definition and observation date, insured monthly benefit, waiting period, claim eligibility and coverage period, and a versioned contract form. The exposure measure should be selected for the question. Number of covered lives describes frequency opportunity; monthly insured benefits describe conditional payment severity; projected gross claim present value describes longer-term economic exposure; and claim counts or paid benefits describe realized experience. Each is useful, provided its denominator is reported.

The US Bureau of Labor Statistics publishes industry-specific occupational injury rates (BLS, 2026 release of 2024 data). These rates illustrate occupational and industry heterogeneity in recorded workplace injuries and illnesses. They are not disability income insurance incidence rates. Work-related injury surveillance can exclude nonoccupational disease, can use other duration thresholds, and may have different insured populations. Treating either numerator as a substitute for group LTD claim onset would contaminate model calibration. Such figures can, however, serve as contextual covariates or external plausibility checks after their definitions are reconciled.

Incidence, termination, and benefit severity as dependent processes

Let Ng,tN_{g,t} be incident claims among exposed lives in segment gg during interval tt, Eg,tE_{g,t} the insured exposure measured in life-years, and λg,t\lambda_{g,t} its claim onset intensity. A simple conditional specification is Ng,t∣λg,t∼Poisson(Eg,tλg,t)N_{g,t}\mid\lambda_{g,t}\sim\mathrm{Poisson}(E_{g,t}\lambda_{g,t}). The Poisson assumption is a starting point for conditional independent counting; the observed marginal variance can substantially exceed the mean because of employer clusters, shared drivers, unobserved heterogeneity, or delayed reporting. Where the original records permit, a mixed-effects, frailty, or negative-binomial specification can represent such overdispersion. Model choice requires out-of-sample testing, not preference for a conveniently tractable distribution.

After onset, the claim enters a separate duration risk set. Let TT denote elapsed time to claim termination and JJ indicate reason, including recovery or death. Cause-specific recovery and death hazards can be written hr(u∣x,t)h_r(u\mid x,t) and hd(u∣x,t)h_d(u\mid x,t), conditional on elapsed duration uu, claimant characteristics xx, and relevant calendar or economic conditions tt. An observed administrative end, benefit exhaustion, or a change in covered status should be coded according to its actual mechanism. Treating benefit-period exhaustion as medical recovery changes the economic meaning of recovery estimates and can bias the prediction of subsequent claim activity.

The public 2015–2024 LTD termination experience (SOA Research Institute and LIMRA, 2026) documents recovery and mortality as separate outcomes among claimants. The individual disability diagnosis-specific recovery analysis (Beal & Corcoran, 2024) further shows why a portfolio's diagnosis and occupational composition matters to duration estimates. Aggregated public tables do not provide an identified estimate of cross-employer dependence, however. Estimating that parameter requires appropriately structured joint observations, time indexing, and disclosure of sampling limitations.

A claims portfolio can experience several forms of common deterioration. New claims may increase simultaneously; the same stock of disabled claimants may recover more slowly; the average eligible monthly benefit may change; or benefit offsets and adjudication requirements may alter payment per claim. These mechanisms have different treaty consequences. An aggregate excess layer can be sensitive to all four, whereas a specific per-claim layer may respond principally to claim severity or duration beyond a chosen attachment. Model stress should explicitly say which components change and why.

Covariance and diversification sensitivity

For net or gross segment losses L1,…,LmL_1,\ldots,L_m, the portfolio variance is

Var⁡(∑gLg)=∑gVar⁡(Lg)+2∑g<hCov⁡(Lg,Lh).\operatorname{Var}\left(\sum_g L_g\right)=\sum_g\operatorname{Var}(L_g)+2\sum_{g<h}\operatorname{Cov}(L_g,L_h).

An intuitive two-segment example holds each segment's annual loss standard deviation at $10 million. If the correlation between their losses is 0.10, the aggregate standard deviation is \sqrt{100+100+20}=$14.83 million; if it is 0.60, the standard deviation becomes \sqrt{100+100+120}=$17.89 million; at 0.90, it becomes \sqrt{100+100+180}=$19.49 million. The units inside the square root are squared millions of dollars. These numbers follow the covariance identity exactly; the correlations are hypothetical and are not estimated from the cited SOA studies. They demonstrate that an unchanged marginal risk profile can have materially different aggregate volatility when common drivers change.

Linear correlation describes only a fraction of the dependency relevant to extreme claims. A recessive economic shock, pandemic, major industrial closure, or regional health-system disruption may induce simultaneous increases in claims and decreases in recovery even when average-period correlations appear moderate. A multivariate scenario can vary incidence, duration, and wage benefits jointly, then apply the treaty's actual ceded-loss function before computing net outcomes. The insurer should not insert a single universal disability correlation coefficient across all geographies, groups, and time periods merely because the model requires one. Regional and occupational dependency may change with labor markets, age mix, and the cause of disability.

Common shocks and identifiable causal claims

A causal narrative for claim incidence needs explicit mechanisms and comparison populations. An increase in insured claim filings during a downturn might reflect deteriorating health, reduced ability to accommodate work limitations, changing application incentives, shifts in insured employment, changes in reporting, or delayed adjudication. A geographic association may be explained by shared industries or employer benefit design rather than location itself. Claim onset could be followed by differing return-to-work opportunities, which influence duration independently of medical recovery.

Kostøl and Mogstad's study of return-to-work incentives (2013) uses Norwegian administrative and policy variation to investigate labor-market responses among public disability insurance recipients. Its identified incentives matter for considering work capacity, benefit design, and economic behavior, while its institutional setting differs from private group LTD. The paper supplies evidence that disability durations can respond to policy incentives; it does not itself estimate a private reinsurer's aggregate losses or a cross-jurisdiction causal effect of reinsurance.

Identifiable evaluation of a private book would compare groups within common observation windows, preserve policy changes and adjudication dates, and distinguish exposure loss through employment termination from actual disability recovery. Correlation between employment shocks and claim outcomes can produce a real common risk even when the medical prevalence of a disorder is unchanged. Documentation of the conditioning set, selection into insured employment, and missing claims is essential to interpretation.

Concentration measures and their interpretation

Let wgw_g represent segment gg's share of an appropriately selected exposure measure, with ∑gwg=1\sum_g w_g=1. A Herfindahl-style concentration statistic H=∑gwg2H=\sum_gw_g^2 reports how heavily exposure is allocated across defined groups. It should be calculated separately for employers, industries, locations, occupations, and insured benefits, because these groupings overlap and answer different questions. An inverse value 1/H1/H can be interpreted as an effective number of equally weighted categories within that single partition, assuming the categorization is meaningful. It is not the number of independent risks and does not measure their correlations.

Occupation and geography should also be crossed. A portfolio can have dispersed addresses but most benefits in one specialized technology sector, or numerous employers in one regional health system. Joint cells reveal the structure while reducing the amount of experience in each cell. Statistical credibility and privacy protections become more consequential as granularity increases. When data are sparse, partial pooling and scenario analysis are preferable to implying precision from small cell rates. A concentration review should therefore report both exposure balance and the uncertainty surrounding the corresponding loss dependence.

Counterparty and Capital Exposure

Recoverables and the continuing insurance obligation

The insured person's contractual counterparty normally remains the direct insurer. Reinsurance replaces part of the insurer's underwriting loss with a receivable and a recovery process. The IFRS 17 reinsurance guidance (IFRS Foundation, 2018) emphasizes the separate accounting of reinsurance contracts held: the direct insurance promise and the reinsurance contract are separate contracts. Recording a gross liability and a reinsurance-held position expresses this separation more faithfully than assuming the gross obligation disappears whenever a treaty is signed.

The reinsurer's performance may be subject to reporting and documentation, reconciliation of covered losses, claims-control or audit clauses, withholding, retrocession, security, settlement periods, and disputes over the treaty interpretation. A solvency analysis should distinguish estimated treaty recoverables, amounts due and collectible, collateral actually available, losses already settled, and disputed or impaired balances. A reinsurance asset can reduce expected net ultimate claims while creating a short-term cash gap when the cedent must pay policyholders before collection.

IAIS ICP 13 requires supervisory attention to effective reinsurance arrangements and the risks remaining with the insurer, including credit, liquidity, operational, and basis risks. This applies especially to disability portfolios, because multi-year claim cash flows can create repeated dependence on the financial condition and claims-handling continuity of the same reinsurer. Counterparty analysis must consider related reinsurers and common ownership rather than relying solely on formal legal names.

Counterparty concentrations and contingent impairment

Suppose reinsurance recoverables are distributed across three economically independent counterparties with shares of 50%, 30%, and 20%. Their recoverable-share concentration is HR=0.52+0.32+0.22=0.38H_R=0.5^2+0.3^2+0.2^2=0.38, corresponding to 1/0.381/0.38, approximately 2.63, equally weighted counterparties in this descriptive index. If the two smaller entities are subsidiaries of one parent with common financial support, the effective economic concentration may be greater than this legal-entity calculation suggests. In any case the index omits default probabilities, exposure maturity, settlement timing, collateral, and correlations with disability claim stress.

A simplified conditional credit stress may assume an outstanding, unsecured $20 million recoverable and a 50% loss given default. Conditional impairment is then $10 million. It is not the expected credit loss unless combined with a properly calibrated default probability and other required assumptions, and it says nothing by itself about the solvency capital charge. A more complete simulation links recoverable size, insurer claim stress, reinsurer default state, collateral realizability, and the time of cash settlement. A large reinsurer loss under the same macroeconomic stress that worsens disability claims can produce wrong-way risk: higher claims coincide with weaker collectability.

Concentration can also accumulate through retrocession and common investment exposures. The cedent may observe three primary reinsurers while those firms pass portions of the risk to the same retrocession market or depend on overlapping collateral assets. Evidence of such dependence requires contractual and financial disclosures; an unobserved retrocession network must be treated as uncertainty rather than assumed interconnectedness. Reinsurance counterparties can also be diversified across jurisdictions while sharing one claims administrator, custodian, technology provider, or legal settlement arrangement.

Capital under solvency regimes

Capital benefit arises from the effect of reinsurance on the jurisdictionally applicable solvency calculation, subject to credit quality, effective transfer, diversification, and recognition rules. The European Union's Solvency II Directive (European Parliament & Council, 2009, consolidated text consulted) distinguishes underwriting, market, and counterparty-default components within the solvency requirement and imposes qualitative standards of governance and risk management. The exact treatment of disability benefits depends on their classification and legal context. A product commonly described as disability income may sit within health underwriting calculations under one framework and within other liability or reserve systems elsewhere.

The NAIC model-law collection provides the US reference texts for credit-for-reinsurance law and regulation, including models 785 and 786. US credit recognition depends on state law as adopted and amended, reinsurer status, security, and conditions applicable to the cedent; a list of model instruments cannot establish that each state has enacted identical rules. Cross-border assessments must name the supervisory and accounting regimes separately. In particular, capital relief under statutory solvency treatment cannot be inferred from the amount recognized under an IFRS financial statement.

A risk allocation problem can be stated as a comparison between the distribution of gross insurer obligations GG and retained obligations N=G−R(G)N=G-R(G) plus collection and financing costs. A reinsurance contract may decrease loss-tail volatility but increase premium outflow or financial counterparty exposure. A treaty can lower an underwriting charge while raising an associated counterparty charge or requiring collateral management. It is consequently necessary to assess net capital effects jointly, with the governing rules and calculation date explicitly identified.

IFRS 17 measurement and economic liquidity

IFRS 17 Insurance Contracts (IFRS Foundation, 2017, effective for annual reporting periods beginning on or after January 1, 2023) sets financial reporting requirements for insurance contracts issued and reinsurance contracts held. The measurement of reinsurance held is related to but distinct from the measurement of issued insurance contracts; distinct grouping and cash-flow requirements apply. A direct insurer must not conclude that recognizing a reinsurance asset is equivalent to eliminating its gross claims-payment duties.

Long-term disability benefits often require monthly disbursements, and a ceding insurer may finance them while awaiting treaty bordereaux reconciliation and collection. Liquidity stress therefore turns on the timing of receipts as well as ultimate loss recovery. The insurer should project monthly claim payments, premiums, reinsurance recoveries, disputed amounts, collateral calls, and usable liquid assets under consistent assumptions. A reinsurer whose reimbursement is correct on an ultimate basis but late during a shock can still create substantial short-term funding need.

An illustrative cohort with 1,200 accepted disability claims, a monthly insured benefit of $1,500, and 24 expected payable months has $43.2 million of undiscounted modeled future benefits. If the assumed mean paid duration rises to 30 months with all other quantities held fixed, projected gross benefits rise to $54 million, a difference of $10.8 million. Under a hypothetically comprehensive 40% quota share and full collectability, $4.32 million of that difference falls on the reinsurer and $6.48 million remains with the cedent. The calculation omits the changing active claimant population, competing exits, offsets, time value of money, premiums, and timing. It illustrates how duration deterioration transmits into a ceded recoverable and a retained cash-flow requirement, not an empirical expected reserve or capital estimate.

Contract and counterparty stress combinations

Stress testing should include at least four distinct outcomes: increased insured disability claims with full treaty performance; increased claim durations with contractual exclusions or layer exhaustion; increased claims coincident with delayed reimbursement; and high claims coincident with reinsurer impairment. The stress design should reflect legal-entity exposure and all material reinsurance features. For example, a strict annual aggregate cap can create a sudden change in retained marginal loss after the limit is exhausted. A per-person retention based on accumulated payments can shift cash requirements many months after a claim starts. A commutation provision may transform future periodic reimbursements into a settlement subject to valuation uncertainty.

Supervisory risk-transfer principles make these joint analyses a necessary part of assessing risk management, but the precise quantitative result must be supported by specific treaty terms and insured experience. Rating grades alone, absent recoverable concentrations and timing, are insufficient to characterize the risk. In an international context, insolvency priorities, collateral enforceability, and claims-payment mechanisms must be checked under the law actually governing the counterparties.

Integrated Findings and Research Agenda

The principal joint findings

The source-based analysis establishes that disability reinsurance capacity and risk concentration must be evaluated on the same financial, contractual, and population basis. A ceded percentage offers little information until the eligible claim population and timing of benefit payments are specified. Disability incidence and claim termination form different processes and must be modeled on distinct exposure denominators. Occupational, industry, employer, and geographic classifications reveal different dimensions of common loss. Correlations among insured groups influence aggregate tail exposure even when each group's own expected loss remains stable. Reinsurance transforms parts of these losses into recoverables that themselves require credit, liquidity, concentration, and legal assessment.

This synthesis follows the evidence rather than proposing an independent theoretical system. The actuarial experience studies establish how incidence and duration are observed, IAIS ICP 13 establishes the supervisory attention required for effective risk transfer, and IFRS 17 establishes distinct accounting treatment for reinsurance held. Their combination yields a practical inference: a treaty should be judged by the joint distribution of retained obligations and realizable collections throughout the disability benefit period, together with the governing capital framework.

Evidence-based research priorities

A high-value empirical program would combine linked group-level policy exposure, claim onset, monthly benefit payments, termination reason, employer and occupational classification, geography of worksite, and treaty recoveries. It would estimate jointly the distribution of incidence, claim duration, and benefit amounts; examine shared calendar shocks and industry exposures; and validate the tail of retained loss distributions against withheld periods. A multi-insurer consortium with standardized definitions could investigate dependencies that published aggregate studies do not identify. Research governance would require consent, privacy protections, transparent metadata, and disclosure of insurer participation and selection.

A separate comparative program should collect actual treaty wording and legally grounded case studies for quota-share, per-claim excess, paid-loss aggregate, incurred-loss aggregate, and claims-in-course structures. The investigation should track reported recoverables, cash settlement, reserve basis differences, dispute frequency, default or collateral events, and the time when layer limits bind. The comparison would distinguish contractual expected payment, actual collection, and insurance obligations that persist regardless of collection. Without such documentary evidence, assertions that one treaty form universally eliminates concentration are unsupported.

A third program should evaluate occupation and geographic exposure under employer closure, regional recession, disease shocks, and disability-management interventions, while explicitly separating health effects from labor-market effects. The Kostøl and Mogstad evidence (2013) suggests that work-incentive design can matter, but direct transfer to private cross-border LTD requires fresh identification. The purpose of new evidence is to estimate effect sizes and their conditions, rather than to infer private treaty outcomes from the existence of a public-program effect.

A fourth program concerns counterparty and capital interaction. It should combine legally grouped recoverables, collateral, payment lags, ceded future cash flows, and correlated credit and disability stress. Comparisons across regimes must state accounting and solvency distinctions and the effective date of legal provisions. Model outputs should show how a contract that reduces gross loss volatility can nonetheless produce concentrated collection dependence in an adverse scenario.

Implications for insurer governance and scholarly reporting

For insurer risk management, the principal measurable objects are the stock of covered benefits, the joint distribution of insured disability cash flows, each treaty's eligible claim function, projected receivables and settlement lags, and the insurer's obligations after a reinsurer impairment. Portfolio reporting should describe occupation, employer, industry, geography, and counterparties on consistent dates, but separately, to avoid mistaking diversification in one classification for diversification of the whole financial obligation.

For scientific reporting, every observed statistic should identify whether it measures worker exposure, onset claims, disabled-life months, exits, paid claims, or accounting liabilities. Every illustration should state its artificial assumptions. Quantitative claims should be auditable to the precise source and period. A transparent statement that joint correlation cannot be estimated from public summary tables is a meaningful research result: it identifies the data needed to resolve a decision-critical uncertainty without fabricating precision.

Conclusion

Disability income insurance reinsurance transfers contractually defined portions of earnings-replacement liabilities whose financial magnitude depends on claim incidence, ongoing benefit eligibility, duration, recovery, mortality, insured earnings, and payment timing. Long-term disability portfolios can remain concentrated through common employers, occupational tasks, industries, geographic labor markets, and shared economic conditions even when a large number of individual lives are insured. Valid concentration analysis therefore requires a multidimensional exposure record and a dependency model that respects the separate denominators and mechanisms of disability onset and claim termination.

The monograph's principal conclusion is that the effectiveness of a disability treaty must be evaluated as a joint relationship between gross insured cash flows, precisely eligible ceded losses, the timing and collectability of recoverables, and the reinsurer's own concentration and financial condition. Proportional and excess arrangements redistribute different portions of severity and duration risk; attachment, exclusions, benefit-period provisions, claims-in-course terms, aggregate limits, and settlement requirements determine the realized transfer. An apparent reduction in underwriting concentration can be accompanied by increased counterparty and liquidity concentration, particularly when disability stress and reinsurer impairment occur together.

The empirical base supports careful distinctions rather than universal numerical estimates. Primary SOA/LIMRA studies provide substantial evidence on the incidence and termination of specific disability insurance populations, and recognized actuarial methods explain their use and limits. IAIS principles, international financial reporting, national legal instruments, and independent scientific studies illuminate risk transfer, capital, claims rights, and work incentives. Publicly available summaries do not independently identify the complete dependence structure among ceded disability portfolios, nor do they reveal every treaty's counterparty and payment terms. The original contribution here is a fully specified comparative synthesis of these interacting objects, combined with explicitly reproducible arithmetic and a falsifiable program for further evidence collection. It makes no claim to introduce a new theory, doctrine, or empirical market estimate.

References