RESEARCH DOSSIER
Insurance Reserves and Contingent Liabilities
Abstract. Insurance creates legally and economically consequential commitments whose settlement amounts, timing, and sometimes triggering circumstances are uncertain. Estimating these commitments connects probability, claims administration, contract law, accounting recognition, supervisory valuation, asset-liability management, and capital policy. This monograph develops a unified analytical treatment of insurance reserves and contingent liabilities while maintaining the distinct measurement objectives of actuarial unpaid-claim estimates, IFRS 17 contract liabilities, IAS 37 contingent obligations, United States statutory accounting, and Solvency II technical provisions. It examines reported claims, incurred but not reported claims, future policy benefits, expenses, discounting, contractual options, reinsurance, risk adjustments, risk margins, and reserve deterioration. A reproducible set of synthetic numerical examples distinguishes undiscounted expected payments, their present value, booked liabilities, risk margins, solvency capital, and available liquidity. The study synthesizes primary standards with the research literature on stochastic claims reserving, emphasizes the evidentiary conditions needed for sound estimates, and presents a practical sequence for evaluating adequacy and the transparency of competing reserve measures. Its central analytical conclusion is that an insurer's future payment obligation is the common economic object, while the reported reserve is a framework-specific representation of that object. Sound governance requires reconciliation between representations rather than collapsing them into a single supposedly universal reserve number.
Keywords: Insurance reserves; technical provisions; contingent liabilities; IFRS 17; IAS 37; Solvency II; unpaid claims; incurred but not reported; future policy benefits; discounting; actuarial uncertainty; risk adjustment; solvency capital; reinsurance; reserve adequacy; claims development; insurance accounting; asset-liability management.
Introduction
A premium is received or promised today in exchange for the insurer's legally bounded commitment to provide coverage and settle covered events over time. As claims develop, the insurer gradually learns how much those commitments will cost. Reserving is the process of translating the resulting uncertain cash flows into estimates suitable for a specified decision. The estimate may be needed for internal management, statutory accounts, general-purpose financial statements, a solvency balance sheet, the pricing of risk transfer, or the assessment of whether assets can meet payments when due. Each use demands clarity about the insured events covered, the valuation date, the accounting boundary, the probability measure, the time horizon, the included expenses, and the desired confidence or prudence level.
The IFRS 17 accounting framework recognizes groups of insurance contracts using current fulfillment cash flows and a contractual service margin in circumstances where that margin applies (IFRS Foundation, 2023). The Solvency II valuation rule expresses technical provisions in terms of a best estimate and a risk margin, subject to its detailed statutory exceptions and implementing rules (European Union, 2009). The IAS 37 recognition criteria distinguish recognized provisions from disclosed contingent liabilities within that standard's scope (IFRS Foundation, 1998/2026). These are distinct normative questions. A cash outflow can be economically plausible without constituting an already recognized liability in every framework; the existence of an insurance obligation can also precede the realization of an individual insured loss.
This inquiry adopts a comparative methodology. First it identifies obligations and their contractual boundaries. Next it specifies the estimand, such as undiscounted ultimate losses or current fulfillment cash flows. It then evaluates models, reinsurance, reporting treatments, capital implications, and the tests by which an estimate can be challenged. It relies on publicly accessible primary-source standards and original research rather than confidential insurer datasets. All numerical cases are invented for exposition, with explicit assumptions; they are demonstrations of relationships, not estimates for a real balance sheet. The jurisdictional comparison concerns broad principles at November 14, 2026, and local legal and accounting application remains determined by the operative rules of the relevant territory.
The primary scientific question is how one can consistently recognize, measure, fund, validate, and explain the future cost of insurance promises when both expected outcomes and the rules for recording them differ. The answer requires separating three operations: identifying obligations, estimating their consequences, and determining what additional financial capacity protects against outcomes beyond the central estimate. These operations can be coordinated through transparent reconciliation, reproducible cash-flow models, and explicit governance of uncertainty.
Contingent Liabilities and Insurance Promises
The meaning of an insurance commitment
An insurance policy transfers specified uncertain risk in exchange for consideration. Contract formation can create duties to stand ready to provide coverage, to administer claims, and ultimately to settle qualifying losses. A claims event may be uncertain in occurrence, magnitude, or timing, but the contract itself generates a present coverage obligation when the applicable recognition conditions have been met. The IFRS 17 key-term definitions distinguish a liability for remaining coverage from a liability for incurred claims. The former concerns insurance services yet to be provided; the latter concerns amounts related to services already provided (IFRS Foundation, 2023). This difference matters for every attempt to equate reserves with known unpaid claims.
The word reserve has several meanings in practice. An actuary may call the expected unpaid claims an unpaid-claim estimate; management may book a reserve in an accounting ledger; a supervisor may call a broader valuation technical provisions; a life insurer may maintain a liability for future policyholder benefits. According to ASOP 43's terminology, an unpaid-claim estimate is an actuarial measurement, while a reserve in the strict sense is the amount recorded in a financial statement (Actuarial Standards Board, 2007/2011). Consequently, a booked reserve need not equal an actuary's central estimate and any comparison must disclose the target measure and purpose.
The underlying obligation can be analyzed through five linked dimensions. Its legal boundary specifies contracts, triggers, exclusions, limits, deductibles, and relevant claims law. Its temporal boundary specifies which coverages or events have occurred by the reporting date and which future services remain. Its cash-flow boundary includes claims, settlement expenses, relevant contract cash inflows, and amounts legally recoverable. Its information boundary specifies which events and data were known at the valuation and review dates. Its measurement boundary identifies the probability, discounting, and prudence conventions that will be used. These dimensions make explicit why two legitimately prepared liability values may differ.
Contingency under IAS 37 and insurance contract accounting
Under IAS 37's treatment of contingencies, a provision is a recognized liability of uncertain amount or timing when a present obligation from a past event, a probable outflow, and reliable estimation are established. A contingent liability may instead be a possible obligation, or an existing obligation that does not meet recognition requirements; disclosure is generally required unless the possibility of an outflow is remote (IFRS Foundation, 1998/2026). An uncertain future loss from ordinary business operations, without a present obligation at the reporting date, does not automatically justify a general IAS 37 provision.
Insurance contracts within IFRS 17 are governed by their own recognition and measurement model. An insurer's liability for remaining coverage therefore should not be reclassified as an IAS 37 contingent liability merely because a future insured event is uncertain. For a lawsuit arising outside the insurance contract measurement scope—for example, the insurer's separate dispute with a technology provider—IAS 37 may be relevant, subject to the specific applicable standards. For a reported claim under an insurance contract, the relevant insurer-side insurance accounting generally belongs to IFRS 17 rather than a duplicate IAS 37 provision. The same factual episode may give rise to different entities' obligations, but accounting requires identifying the reporting entity and the actual obligation.
The analytical significance is substantial. Contingency describes uncertainty about the existence or settlement of a possible obligation in a particular accounting rule. Insurance risk describes uncertainty transferred under a contract. Their domains intersect, but they answer different questions. A storm that has yet to occur may generate no incurred-claim liability; the insurer can nevertheless have a measurable obligation to provide remaining insurance coverage. After the storm occurs, the accounting relationship changes as obligations concerning past service and claims emerge.
Recognition boundaries and group effects
Recognition involves date, contract boundaries, and group accounting, not just an expected-loss calculation. IFRS 17's published overview identifies portfolios and groups, current fulfillment cash flows, and contractual service margin, with immediate loss recognition for loss-making groups (IFRS Foundation, 2023). Grouping affects aggregation and the recognition of future profit. An otherwise identical claim projection may produce a different financial statement result depending on whether it is part of remaining coverage, incurred claims, or an onerous group.
Reinsurance adds a second contract rather than erasing the direct promise. A primary insurer generally remains obliged to its policyholder unless novation or a legally effective transfer extinguishes that obligation. IAIS ICP 14 explains that reinsurance purchase normally does not derecognize the direct obligation and emphasizes coherent recognition of reinsurance rights (IAIS, 2024). The quality of reinsurance recoverables introduces counterparty and collection risk, even where the underlying direct claims are well measured. Gross expected losses and anticipated recoverables should be separately traceable before a net position is shown.
Institutional implications
These distinctions structure three questions for managers and supervisors. What does the insurer owe under the contract or other law? How much is the obligation worth according to a stated measurement framework? What assets, liquidity, and capital are available to discharge it under normal and stressed outcomes? Confusing these questions can conceal under-reserving, cause artificial reserve overstatements, or lead decision-makers to rely on funds that are economically unavailable when claims arrive.
Claims Reserves and Future Benefits
Reported, incurred, and developing obligations
For property-casualty coverage, reserve components are often organized according to whether losses have occurred and whether they have been reported. Reported but unsettled (RBNS) claims require individual or case-level estimates of future payments and expenses. Incurred but not reported (IBNR) claims cover events that have occurred but have not yet appeared in the claims system; broader actuarial usage sometimes includes future development on known claims, also described as incurred but not enough reported or IBNER. Case reserves and actuarial development estimates must be reconciled to prevent double counting. ASOP 43's scope and guidance encompass claims adjustment expense and require appropriate definition of the intended measure and gross/net treatment (Actuarial Standards Board, 2007/2011).
A compact accounting identity clarifies the economics. Let U represent the estimated ultimate cost of a cohort of insured events, P payments already made, and R the estimated remaining unpaid amount. Then R = U − P, when costs are measured on the same basis, including the same claims expenses and recoverables. Let C be current reported case outstanding. A development-based estimate of additional unpaid cost is D = R − C. It is possible for D to be negative when case estimates are collectively conservative or when model assumptions differ; that is a signal for investigation, not automatic proof of an error.
Consider one synthetic accident-year cohort, in thousands of currency units. Paid losses equal 600, aggregate case outstanding equals 400, and reported incurred losses therefore equal 1,000. If an incurred-development factor of 1.20 is appropriate, projected ultimate incurred cost is 1,200. Expected unpaid is 1,200 − 600 = 600. Of this, 400 is case outstanding and 200 is additional development. The chain-ladder calculation is 1,000 × 1.20 = 1,200. It remains conditional on the factor and historical stability; it is not a probabilistic guarantee.
A second synthetic cohort illustrates prior-expectation weighting. Reported incurred equals 850, paid equals 520, the estimated emergence proportion is 0.75, and the prior expected ultimate is 1,100. Under a conventional Bornhuetter–Ferguson construction, ultimate = reported incurred + (1 − 0.75) × 1,100 = 1,125; unpaid = 1,125 − 520 = 605. A corresponding pure development estimate would be 850 / 0.75 ≈ 1,133.33 and unpaid ≈ 613.33. The 8.33 difference in unpaid cost arises from the prior ultimate assumption, not from an alternative liability-recognition rule. England and Verrall's stochastic review places the Bornhuetter–Ferguson method within a wider statistical and Bayesian reserving context (England and Verrall, 2002).
Why claim data develop
Claims evolve as notifications arrive, coverage investigations proceed, legal liability becomes clearer, injuries mature, repair costs change, and cases are settled or reopened. Long-tail liability business can be exposed to judicial precedent, social inflation, medical cost changes, and policy-limit erosion for many years. Property catastrophe claims may be reported quickly yet undergo extended adjustment or litigation. The selection of accident, policy, report, or calendar-year segmentation therefore materially changes development patterns and the meaning of the statistics estimated from them.
Claims triangles compress repeated valuations of cohorts. The usual chain-ladder method assumes that observed development relationships provide a meaningful basis for projecting later emergence. This assumption is threatened when an insurer changes claim handling, enters a new line, alters limits or attachment points, or receives a large nonrecurring catastrophe. Reserving should then integrate exposure, underwriting and claim-level information with aggregate patterns. Mack's distribution-free analysis derives uncertainty results for chain-ladder reserve estimates under specified conditions rather than assuming the central estimate is certain (Mack, 1993).
Life insurance, disability, and annuities
Life and long-duration benefits require a different temporal view. Future payments can arise from death, survival, disability, morbidity, annuitization, surrender, and contractual guarantees, conditioned on policies remaining in force and coverage boundaries. The relevant cash flows may include future premiums, benefits, operating expenses, investment-linked amounts, and the effect of policyholder options where the accounting and solvency framework requires them. Persistency, mortality improvement, longevity, expense inflation, and interest-rate assumptions interact. A small change in discount rates or longevity assumptions can materially affect a long-duration liability, especially when the portfolio includes guaranteed annuity payments.
Consider a highly simplified fixed benefit of 1,000 due in ten years, payable with probability 0.80. Under a constant effective annual discount rate of 3%, its probability-weighted present value is 0.80 × 1,000 / (1.03)^10 ≈ 595.28. Under a 5% rate it is ≈ 491.13. This comparison omits mortality timing details, expenses, premiums, guarantees, and risk adjustment. Its purpose is to show the sensitivity of the expected present value to discount rates. It does not represent a complete life insurance reserve.
Insurance liability valuation is therefore an integration of actuarial events and financial cash-flow timing. Reporting frameworks specify which portfolio boundaries, margins, option features, and discount curves are appropriate. ISAP 4's IFRS 17 guidance addresses actuarial services relating to that accounting standard while retaining a distinction between professional guidance and the accounting standard itself (International Actuarial Association, 2019).
Expenses, recoverables, and negative adjustments
An unpaid estimate that includes indemnity but excludes defense, legal services, claim handling, or settlement expense may fail to represent the obligations relevant for reporting. The appropriate expense boundary follows the underlying contract and applicable measurement rules. Similarly, expected salvage, subrogation, deductibles, and reinsurance recoveries can change a net outcome without changing the gross claim itself. Recoverables have their own payment timing and credit risk. If a reinsurer is distressed precisely when a catastrophe produces unusually high claims, treating reinsurance recovery as risk-free can understate joint loss exposure.
For a simple synthetic example, gross unpaid claims are 500, expected reinsurance recoveries before credit adjustment are 180, and a current estimate of recoverable shortfall is 15. The expected recoverable after the adjustment is 165; indicative net economic exposure is 500 − 165 = 335. The calculation is a management illustration, not an instruction to net reported IFRS 17 liabilities and reinsurance balances in the primary balance sheet. Presentation follows the applicable accounting framework. Gross/net reconciliation is necessary because a nominal reinsurance credit and a collectible asset are different economic quantities.
Discounting and Estimation Uncertainty
Present-value measurement
Reserves often concern payments extending over multiple years. The expected present value of a stream of future net payment cash flows can be expressed as
[ BE_0 = \sum_{t=1}^{T}\frac{\mathbb{E}[C_t]}{(1+r_t)^t}, ]
where C_t is the relevant period-t outflow net of cash inflows allowed by the measure, r_t represents the effective discount rate for the specified maturity, and all cash flows, expectations, currency, expenses, and discount assumptions share a consistent valuation basis. For a more general term structure, the appropriate discount factors are maturity-specific and need not have the simple power form above. Financial options, dependencies between discount rates and cash flows, and scenario-contingent benefits may demand stochastic valuation rather than discounting an unconditional average with one rate.
Suppose a synthetic cohort is expected to require claim settlements of 300 at the end of year one, 400 at year two, and 300 at year three, in thousands. Undiscounted expected payments are 1,000. At a constant annual effective 4% rate, the expected present value is 300/1.04 + 400/(1.04)^2 + 300/(1.04)^3 ≈ 924.98. At 2%, it is approximately 961.28. The 36.30 increase as rates decline illustrates liability duration exposure. Neither amount includes a risk margin or compensates for adverse claim development. A further caution is that nominal inflation assumptions in the cash flows must be consistent with nominal discount rates; combining real claims payments with nominal discounting would misstate the amount.
Solvency II Article 77 requires the best estimate to use probability-weighted expected future cash flows and an applicable risk-free term structure, as distinct from the risk margin (European Union, 2009). IFRS 17's valuation description specifies risk-adjusted current fulfillment cash flows and the treatment of changes in estimates within the accounting model (IFRS Foundation, 2023). Similar present-value arithmetic thus serves different reporting architectures.
Discount-rate selection and consistency
The discount curve can reflect jurisdictional prescription, currency, liquidity characteristics, asset-liability considerations, and the cash flows' financial risk. A yield on the insurer's actual assets is not automatically the rate for valuing its insurance obligations. Using optimistic asset returns to reduce projected claims liabilities would obscure asset-side investment risk. Conversely, a risk-free term structure does not settle how to recognize nonfinancial insurance risk. Each framework specifies whether such risk is measured in expected cash flows, a separate risk adjustment, or another liability component.
The revised ASOP 20 on cash-flow analysis was adopted in December 2025 and took effect June 1, 2026. Its scope covers analysis of property-casualty cash flows, including discounting, and emphasizes purpose, assumptions, timing, and disclosure (Actuarial Standards Board, 2025/2026). Earlier ASOP 20 versions must not be silently treated as the current US actuarial guidance. The practice standard governs relevant actuarial work; accounting or regulatory permissibility of discounting depends additionally on the applicable reporting requirements.
Three distinct forms of reserve uncertainty
Process uncertainty is the inherent randomness of future claims even when parameters and model structure are known. Parameter uncertainty concerns the limited precision with which frequency, severity, development, mortality, or other parameters can be estimated. Model uncertainty concerns which mathematical representation, segmentation, exposure assumptions, or scenario specification is appropriate. These categories are analytically useful because adding more observations can improve parameter estimates without eliminating process variation or resolving structural regime change.
The Mack (1993) prediction-error method evaluates variability associated with a chain-ladder setting under its assumptions. England and Verrall (2002) review stochastic extensions, smoothing, Bayesian approaches, and predictive distributions. A full predictive distribution can inform quantiles, expected shortfall, stress testing, or management ranges, provided the tails, correlations, and assumptions are scientifically defensible. It would be incorrect to relabel the output of one simulation as a universal confidence interval independent of model limitations.
A common source of false precision arises when long-tail portfolios are projected through one set of historical development factors with no representation of future legal, economic, or operational change. A probability distribution conditional on that unchanged model cannot alone quantify structural uncertainty beyond the model. Governance must incorporate scenario design and expert assessment, explicitly labeled, rather than conceal those judgments in the statistical output.
Dependence and nonlinearity
Insurance obligations are often dependent. A common catastrophe affects property, interruption, liability, and reinsurance recoverables. Economic inflation may simultaneously increase claim severity and affect discount rates. Pandemic or longevity changes can alter claims across many insured lives. Dependence becomes decisive for joint adverse outcomes even when expected claims by business line are comparatively stable. A portfolio reserve is an aggregate expected liability; its high-percentile distribution cannot generally be obtained by adding independent high percentiles without modeling cross-line dependence.
For example, if two portfolios have marginal reserve standard deviations of 100 and 80 in common units, the aggregate standard deviation under a simple variance approximation is sqrt(100² + 80² + 2ρ × 100 × 80). At zero correlation this equals about 128.06; at correlation 0.75 it is approximately 168.52. These numbers assume the variances and correlation are appropriate and stable. Their purpose is to expose how dependency changes uncertainty, not to prescribe a regulatory capital measure. Correlation itself can rise sharply under stress, so normal-state correlations may fail in a catastrophe or financial crisis.
Reserving versus Risk Capital
Expected obligation and adverse deviation
A reserve liability recognizes or measures an insurer's obligations according to a defined framework. Capital absorbs losses or deviations that may exceed the booked valuation and other protected resources. Cash and liquid investments fund actual payments. Their functions are connected but their amounts cannot be treated as synonyms. A management balance sheet can be solvent at a valuation date yet face a liquidity shortage when claims require immediate settlement; similarly, sufficient cash for near-term claims provides no proof of long-term capital adequacy.
IAIS ICP 14's capital distinction explains that technical provisions contain a margin over the current estimate for risk and that regulatory capital requirements are a separate component of the solvency assessment (IAIS, 2024). In Solvency II, a risk margin is included within technical provisions while capital requirements are computed separately; the same uncertain future outcomes influence both calculations, but the measurement objectives, time horizons, and prescribed methodologies differ. A risk margin cannot simply be called 'extra capital held inside reserves.' It is a liability valuation component.
Consider a hypothetical insurer with an undiscounted central claim estimate of 1,000; a present value of expected payments of 925; and an illustrative liability risk margin of 75 under a stated internal convention. The corresponding illustrative liability measure is 1,000 (= 925 + 75). Assume separately that the management capital requirement for adverse outcomes is 230, computed on its own prescribed basis. It would be incorrect to call 1,230 'the claims reserve.' The first 1,000 represents the assumed liability measure; the separate 230 is a capital requirement. Actual available capital also depends on admissible assets, all other liabilities, eligibility criteria, and the relevant solvency balance sheet. These numbers deliberately do not claim equivalence to an official Solvency II or IFRS 17 formula.
Why capital should not mask inadequate reserves
Suppose the central expected ultimate claim estimate is 1,200 but management books a claim reserve corresponding to only 1,050 without a justified adjustment or differing basis. Reporting an ample solvency buffer does not cure the undisclosed 150 estimation gap. A solvency calculation built on understated technical provisions may also overstate eligible capital or distort its own risk requirement. Adequacy assessment must begin with a disciplined review of the underlying liability estimate. The ASOP 43 reasonableness criteria address methodology, assumptions, uncertainty, and the communication of limitations (Actuarial Standards Board, 2007/2011).
An equally important error is to add an unspecified 'prudence cushion' to a best estimate and then apply another margin without identifying whether the risks overlap. Conservative assumptions can generate hidden reserves, obscure loss development, or impede comparison between reporting periods. The defensible sequence is to describe the central estimate, identify each additional risk adjustment or margin with its objective, and reconcile this valuation to capital and liquidity assessments.
The asset side and liquidity timing
Insurance finance connects the term structure of liabilities with the availability and behavior of assets. A liability increase caused by a fall in discount rates may coincide with asset valuation gains for some bond portfolios, but the hedge is incomplete when durations, optionality, currencies, or credit exposures differ. Inflation-linked claim costs may rise when nominal fixed-income assets cannot fully respond. A reinsurer default can reduce recovery values exactly when gross claims are unusually large. The economic balance sheet must therefore evaluate common scenarios for liabilities, assets, and counterparties.
The operational liquidity question is narrower and more immediate: what payments could fall due before enough assets can be sold or cash transferred? Catastrophe claims can generate accelerated disbursements and collateral demands; surrender behavior can generate liquidity pressure in relevant life products. Reserve adequacy and payment liquidity should be reconciled through projected cash flows under ordinary, adverse, and severe but plausible scenarios. Neither a present-value reserve nor a capital adequacy ratio directly substitutes for such a cash-flow forecast.
Reinsurance and capital relief
Reinsurance can reduce some retained risk and the associated potential capital need. It may also create basis risk, exclusions, aggregate limits, reinstatement payments, settlement disputes, and concentrations of counterparty exposure. IAIS ICP 14's reinsurance treatment links recognition of direct obligations and reinsurance assets to the economic substance of each agreement (IAIS, 2024). An analysis that simply subtracts nominal limits from gross reserves and treats the difference as available capital ignores timing and collectibility. For valid risk transfer, the retained loss distribution and projected reinsurance recoveries must be modeled coherently.
Reporting and Supervisory Measurement
IFRS 17: current insurance contract measurement
For entities applying IFRS 17, IFRS 17's accounting principles require insurance contracts to be identified and grouped, fulfillment cash flows measured with current assumptions, and contractual service margins recognized over the relevant service period. The accounting model differentiates liabilities for remaining coverage and for incurred claims, as reflected in the IFRS Foundation's defined terms (IFRS Foundation, 2023). The risk adjustment compensates the entity for bearing nonfinancial risk. An appropriate short-duration group may qualify for the premium allocation approach, while qualifying contracts with direct participation features may invoke the variable fee approach. Applicability and details depend on IFRS 17 requirements, not a universal insurance-sector choice among arbitrary reserve methods.
A claims estimate for already incurred losses is associated with the liability for incurred claims; the liability for remaining coverage concerns the broader promise to provide services and may incorporate unearned profit in the form of a contractual service margin. The same label 'unpaid claims reserve' cannot describe both components without erasing a fundamental accounting distinction. Contractual service margin is a deferred unearned-profit component; it is not an additional provision against claims inflation or catastrophe tail risk. Likewise, the recognition of losses for onerous groups follows IFRS 17's specific loss recognition rules rather than a mechanical application of the IAS 37 model to all insurance contracts.
The insurer's economic analysis must follow the IFRS 17 contract boundary, policyholder options, recognized groups, discounting and risk adjustment requirements. An analyst wishing to compare the reported liability with an actuarial ultimate-loss estimate must explicitly bridge gross versus net, incurred versus future coverage, expense inclusion, present-value effects, risk adjustment, and contractual service margin. Without this reconciliation, apparent reserve strengthening or release could simply reflect differences in definitions.
IAS 37: separate provisions and disclosed contingencies
IAS 37's official summary requires an appropriate present obligation, probability of outflow, and reliable estimate for recognition of a provision and describes disclosure of contingent liabilities under the relevant thresholds (IFRS Foundation, 1998/2026). Its direct use concerns matters within IAS 37's scope, including appropriate non-insurance-contract legal obligations. For insurer litigation outside insurance contract measurement, the analyst should establish whether a present obligation exists and whether another more specific accounting standard applies. For insured events already encompassed by an IFRS 17 insurance contract, the existence of an uncertain final payout does not by itself convert the insurer's obligation into an unrecognized IAS 37 contingency.
Solvency II: technical provisions and risk margin
Under Solvency II's technical provisions rules, technical provisions cover insurance and reinsurance obligations, with valuation principles framed in Articles 76–86 (European Union, 2009). Article 77 establishes the best-estimate-plus-risk-margin approach subject to the detailed statutory framework and exceptions. The expected present value uses an appropriate risk-free term structure; reinsurance recoverables are generally calculated separately under Article 81. This differs from labeling a financial reporting contractual service margin as a solvency risk margin. Solvency II also prescribes a separate regulatory capital framework. Applicable transitional provisions, delegated regulations, and phased amendments must be checked as of the jurisdiction's valuation date; a later consolidated or future-effective text cannot be assumed operative retroactively.
A best estimate is prospective, probability weighted, and dependent on contract boundaries, obligations, and credible assumptions. Its independence from the separate margin makes the measurement more transparent. It also imposes a discipline: future cash flows and discount curves must be internally consistent, and assets or recovery expectations cannot be selected to manufacture a desired liability level.
International supervision: IAIS ICP 14
The IAIS valuation principle articulates internationally relevant supervisory expectations for asset and liability valuation and for explaining differences between regulatory and general-purpose financial statements (IAIS, 2024). It addresses current estimates, margins over current estimates, recognition, reinsurance, valuation consistency, and the role of regulatory capital. ICPs are international supervisory standards and guidance, implemented through jurisdictional frameworks; they should not be presented as automatically operative legislation in every country. Their scientific contribution to cross-border comparison is a common vocabulary for assessing whether different national regimes protect policyholders in economically coherent ways.
The International Actuarial Association's ISAP 4 practice guidance similarly provides professional guidance for IFRS 17-related actuarial services (International Actuarial Association, 2019). A local standard setter or actuarial association may adopt or adapt practice standards with local requirements. Careful authorship distinguishes a technical recommendation, a supervisory expectation, an accounting standard, and legally binding national implementation.
United States statutory accounting and related practice
United States insurer statutory accounting follows state-specific law and NAIC-based statutory accounting principles rather than a uniform assumption that IFRS 17 applies to all U.S. insurance reporting. The NAIC statutory accounting program explains the role of the Accounting Practices and Procedures Manual and Statements of Statutory Accounting Principles in regulatory financial reporting (NAIC, 2026). The NAIC statutory issue-paper index identifies SSAP No. 55, Unpaid Claims, Losses and Loss Adjustment Expenses (NAIC, 2026). The precise form of recognition, discounting, valuation, admissibility, and disclosures depends on the relevant state, product and statutory guidance.
U.S. actuarial work involving property-casualty unpaid claims is addressed by ASOP 43; data responsibilities by ASOP 23; modeling practices by ASOP 56; and current property-casualty cash-flow and discounting considerations by ASOP 20 (Actuarial Standards Board, 2007/2011, 2016, 2019, 2025/2026). These standards guide professional services where applicable; they do not independently replace the reporting law or financial statement accounting basis.
Comparative measurement matrix
| Analytical question | IFRS 17 insurance contract accounts | IAS 37 provisions and contingencies | Solvency II prudential valuation | U.S. P/C statutory and actuarial context |
|---|---|---|---|---|
| Primary unit | Recognized insurance contract groups | Obligation or possible obligation within IAS 37 scope | Insurance/reinsurance obligations and prescribed technical provisions | Statutory reporting line and booked reserve; related actuarial unpaid estimate |
| Recognition emphasis | Contract recognition and remaining/past service | Present obligation and recognition threshold; disclosure for qualifying contingencies | Obligations valued under prudential rules | Jurisdiction- and product-specific statutory rules |
| Main measurement architecture | Fulfillment cash flows, risk adjustment, contractual service margin as applicable | Best estimate of obligation for provision; contingent liabilities generally disclosed, not recognized | Best estimate plus risk margin, subject to applicable exceptions | Booked loss and loss-adjustment-expense reserves under applicable statutory rules |
| Timing | Current fulfillment cash flows, contract boundaries and services | Settlement timing of recognized provision | Expected discounted cash flows with regulatory term structure | Discounting governed by statutory rules; separate professional discounting guidance |
| Risk component | Risk adjustment for nonfinancial risk | Risk and uncertainty considered in provision measurement | Explicit prudential risk margin, alongside separate capital framework | Actuarial estimates and booked reserves may include varying purposes and margins |
| Reinsurance | Separate accounting for reinsurance contracts held | Separate recognition principles as applicable | Gross best estimate with reinsurance recoverables evaluated separately | Gross/net reporting and statutory recoverability governed by relevant rules |
The matrix summarizes purposes, not one-to-one numeric conversion formulas. IFRS 17, Solvency II, IAS 37, and national statutory systems have different recognition conditions, levels of aggregation, discounting and disclosure requirements. A cross-regime comparison that ignores these differences can misidentify a measurement change as a change in real insurance risk.
Disclosure, transparency, and reserve movements
A credible reserve report should explain both the level of obligations and changes since the preceding valuation. Movement analysis may identify paid claims, new incurred claims, remeasurement, discount-rate changes, inflation assumptions, acquisitions and divestitures, exchange-rate movements, ceded recoveries, risk adjustment changes, and claims administration changes. The presentation categories differ among reporting frameworks. A reserve release after favorable emergence may be economically warranted, while a sudden reduction without traceable experience or assumption support creates a governance concern. The reader needs to know which portfolio, model, and reporting basis produced the movement.
Model Validation and Reserve Adequacy
From raw records to defendable estimates
Reserve adequacy begins with a testable chain of evidence. Contracts define the coverage. Policy and exposure records identify the insured obligations. Claims systems record occurrence, notification, payments, expenses, recovery, settlement, and reopening events. Actuarial calculations derive expected ultimate claims and the unpaid component. Accounting rules determine recognized amounts, grouping, and presentation. Asset-liability and capital systems connect the result to solvency and liquidity. The links between systems must be reproducible at the relevant valuation date.
The ASOP 23 data-quality standard requires attention to data appropriateness, review, reliance, and disclosure in actuarial work (Actuarial Standards Board, 2016). A useful controls program compares policy counts with premium and exposure data, reconciles cumulative paid amounts with the general ledger, tests negative outstanding balances and duplicate claim identifiers, checks catastrophe coding, and documents changes in recovery practices. Data validation is distinct from auditing every underlying transaction; the review should say which checks were performed and which inherited limitations remain.
The ASOP 56 modeling standard supports scrutiny of model purpose, assumptions, validation, documentation, and governance (Actuarial Standards Board, 2019). Validation should compare multiple appropriate estimators and expose reasons for material disagreement. Reconciliation of case-reserve emergence, development-triangle methods, exposure-based priors, and claim-level severity models is more informative than agreement among models that all depend on the same flawed dataset. Structural breaks should be evaluated through explicit calendar-year, judicial, inflationary, and operational scenarios.
Back-testing and reserve development
The central empirical validation exercise compares estimates made at an earlier date with subsequent claims emergence and settlement. For a cohort valued at t, estimated ultimate loss Ût can be compared with revised Û(t+k) or final ultimate loss when sufficiently mature. The development difference Δ_(t,k) = Û_(t+k) − Û_t measures estimation movement under a stable cohort and cost definition. Positive movement indicates unfavorable development relative to the earlier estimate; negative movement indicates favorable development. The interpretation must separate information genuinely unavailable at the earlier review date from evidence of data errors, unsupported assumptions, or reporting bias.
A hypothetical three-year evaluation might show prior ultimate estimates of 1,000, 1,100, and 1,200 for three cohorts, later revised to 1,080, 1,060, and 1,320. Differences are +80, −40, +120; total unfavorable development is +160 against earlier aggregate estimates of 3,300, or about 4.85%. This is not automatically proof that each original estimate was outside a defensible reasonable range. A sound review examines calendar-year effects, product mix, exposure growth, claim reopening, settlements, and whether subsequent observations were reasonably foreseeable at the initial review date.
Stress testing: a reproducible sensitivity
Return to the synthetic expected payment stream 300, 400, 300 at years one, two, and three. At 4% the expected present value is about 924.98. Suppose all forecast nominal cash flows increase 10% while the discount rate remains unchanged. The present value increases mechanically by 10% to about 1,017.48. If, independently, the rate decreases from 4% to 2% with unchanged cash flows, present value becomes about 961.28. Combining both changes yields about 1,057.41. These are deterministic sensitivities, and the last number assumes the two changes can be applied simultaneously without an additional interaction in claim timing or amounts. Real stress models should consider correlations and behavioral responses.
A model validation exercise should also test reporting delays, loss-adjustment expense inflation, policy-limit changes, judicial outcomes, tail assumptions, pandemic or catastrophe concentration, and reinsurance nonpayment. Some impacts will fall outside historical reserving distributions and require explicit scenario narratives. Where a data-driven model cannot support a quantitative estimate, a transparent scenario can still expose the potential direction and mechanism of risk; it should not be labeled a calibrated probability.
Independent challenge and feedback
A complete internal scientific review should ask whether the estimate's objective is correctly specified, the data can be reconciled, the method is appropriate, the uncertainty measures are compatible with the model, and every material assumption can be explained. It should test whether risk margins and capital are counted in the right place, whether reinsurance recoveries are collectible, and whether the liability calculation responds sensibly to changes in claim severity, discount rates, and duration. Management approval should record unresolved differences between actuarial central estimates and booked reserves; any additional margin should be identified and justified.
Governance requires an audit trail of data extracts, models, versions, assumptions, judgments, approvals, and corrections. It also requires a mechanism for learning from adverse development. Comparing actual settlement data with earlier predictions can identify problems in claim notification practices, assumptions, or incentives. An independent review can challenge these findings, but it cannot replace the original author's responsibility to document calculations and sources. Periodic validation is especially important when the portfolio or macroeconomic conditions change faster than the underlying historical data.
Analytical synthesis and boundary conditions
The research establishes a structured conclusion from existing standards and actuarial literature. The same economic portfolio may be described by a gross expected unpaid-claims estimate, a net cash-flow forecast, a financial reporting insurance contract liability, a prudential technical provision, and a set of contingent obligations requiring separate disclosure. These measurements differ through identifiable transformations: recognition boundary, insured event timing, discounting, expenses, reinsurance, confidence or margin convention, contract aggregation, and reported presentation. A reconciliation should name each transformation and its supporting evidence rather than assert that competing totals must match.
Three boundary cases expose the importance of this approach. First, a future insured catastrophe is relevant to remaining coverage while no particular claims from that future event have been incurred. Second, a severe already-incurred but unreported liability loss can contribute materially to an incurred-claims reserve even when the insurer lacks an individual claim file. Third, litigation against an insurer outside the applicable insurance contract measurement scope may be evaluated under IAS 37 or another governing standard, rather than being silently absorbed into the claims reserve. These cases demonstrate why 'future possible payments' is too broad a definition for any one reported reserve line.
The work's original contribution is its comparative analytic synthesis, transparent case arithmetic, and explanation of how distinct reserve representations can be reconciled. It introduces no separately named theory, doctrine, or proprietary reserving model. The findings are conditional on the specified standards, the hypothetical examples, the verified public sources, and the limited ability of any single method to quantify structural shifts. Application to real insurers requires the relevant contracts, valuation data, legal regime, current effective standards, and independent assessment of assumptions.
Conclusion
Insurance reserves measure uncertain consequences of enforceable insurance promises through conventions designed for actuarial assessment, financial reporting, prudential supervision, and operational decision-making. An incurred claim, remaining coverage obligation, disclosed contingent liability, technical provision, capital requirement, and liquidity resource represent different economic or legal elements. Their reconciliation begins with the obligation's boundary and continues through credible projected cash flows, discounting, risk adjustments, expenses, reinsurance, and the intended reporting purpose.
The formal examples show that expected ultimate claims differ from unpaid claims, undiscounted costs differ from present values, and margins added for distinct objectives should not be confused with solvency capital. Historical claims development and stochastic reserving methods can inform central estimates and uncertainty, but their validity depends on data quality, model conditions, claims operations, and the treatment of changing external conditions. International accounting and supervisory standards clarify these responsibilities through different institutional architectures; comparisons should preserve the authority, applicability, and effective date of each.
The monograph therefore endorses a verifiable research and governance practice: maintain explicit recognition and cash-flow boundaries; test several appropriate valuation perspectives; report expected values and additional margins separately; analyze adverse scenarios and liquidity; reconcile gross, net, and cross-framework amounts; and preserve source-to-claim traceability. On this basis, reserve adequacy becomes an accountable scientific and financial judgment about obligations whose ultimate settlement remains uncertain. The analysis is complete for its stated theoretical and comparative scope, with empirical numerical cases clearly identified as synthetic.
References
Actuarial Standards Board. 2007, last revised 2011. Actuarial Standard of Practice No. 43: Property/Casualty Unpaid Claim Estimates. https://www.actuarialstandardsboard.org/asops/propertycasualty-unpaid-claim-estimates/
Actuarial Standards Board. 2016. Actuarial Standard of Practice No. 23: Data Quality. https://www.actuarialstandardsboard.org/asops/data-quality/
Actuarial Standards Board. 2019. Actuarial Standard of Practice No. 56: Modeling. https://www.actuarialstandardsboard.org/asops/modeling/
Actuarial Standards Board. 2025, effective June 1, 2026. Actuarial Standard of Practice No. 20: Analysis of Property/Casualty Cash Flows, Including Discounting. https://www.actuarialstandardsboard.org/asops/asop-no-20-analysis-of-property-casualty-cash-flows-including-discounting/
England, Peter D., and Richard J. Verrall. 2002. “Stochastic Claims Reserving in General Insurance.” British Actuarial Journal 8 (3): 443–518. https://doi.org/10.1017/S1357321700003809
European Union. 2009. Directive 2009/138/EC (Solvency II), Articles 76–86, as amended and applicable. EUR-Lex original directive and later amendments. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32009L0138
European Insurance and Occupational Pensions Authority. Current online rulebook accessed November 14, 2026. Solvency II Single Rulebook, Article 77: Calculation of Technical Provisions. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-2160_en
IFRS Foundation. 1998, adopted by IASB 2001; official summary accessed November 14, 2026. IAS 37: Provisions, Contingent Liabilities and Contingent Assets. https://www.ifrs.org/issued-standards/list-of-standards/ias-37-provisions-contingent-liabilities-and-contingent-assets/
IFRS Foundation. 2017, amended 2020; effective for annual reporting periods beginning on or after January 1, 2023. IFRS 17: Insurance Contracts. https://www.ifrs.org/issued-standards/list-of-standards/ifrs-17-insurance-contracts/
IFRS Foundation. Accessed November 14, 2026. IFRS 17 Insurance Contracts—Key Terms. https://www.ifrs.org/supporting-implementation/supporting-materials-by-ifrs-standards/ifrs-17/key-terms/
International Actuarial Association. 2019. International Standard of Actuarial Practice 4: IFRS 17 Insurance Contracts. https://actuaries.org/actuarial_practices/isap-4-ifrs-17-insurance-contracts/
International Association of Insurance Supervisors. 2024, amended December 2024. Insurance Core Principles and ComFrame: ICP 14, Valuation. https://www.iais.org/icp-online-tool/
Mack, Thomas. 1993. “Distribution-Free Calculation of the Standard Error of Chain Ladder Reserve Estimates.” ASTIN Bulletin 23 (2): 213–225. https://www.casact.org/abstract/distribution-free-calculation-standard-error-chain-ladder-reserve-estimates
National Association of Insurance Commissioners. 2026. Statutory Accounting Principles (E) Working Group: Mandate and Accounting Practices and Procedures Manual. https://content.naic.org/committees/e/statutory-accounting-principles-wg
National Association of Insurance Commissioners. 2026. Appendix E: Statutory Issue Papers Index, including SSAP No. 55—Unpaid Claims, Losses and Loss Adjustment Expenses. https://content.naic.org/sap_app_updates/app_e_issues.htm
Scientific authorship and comparative synthesis: Angela Bogdanova (Artificial Sapiens). Published by Aisentica Development. This document is a completed internal research manuscript held for a separately authorized batch release. It carries no first-publication date and has not been publicly published at its reserved canonical address.