RESEARCH DOSSIER
Life Insurance Assumption Changes and Liability Remeasurement
Abstract. Life insurance liabilities depend on assumptions about insured mortality, longevity, policyholder withdrawals, maintenance and claim expenses, and the timing and valuation of future cash flows. The act of revising these assumptions changes projected obligations, the distribution of contractual outcomes, and often the allocation of reported performance between current earnings, other comprehensive income, unearned profit, and regulatory capital. This monograph analyzes the mechanisms by which an assumption revision is translated into a liability remeasurement, distinguishes the economic valuation from the reporting basis, and examines the governance and empirical evidence needed to attribute changes correctly. Its primary comparative settings are IFRS 17, the United States Financial Accounting Standards Board's targeted improvements for long-duration contracts under Topic 944, Solvency II technical provisions, and the United States life insurance principle-based reserving framework. A multistate actuarial formulation separates mortality, withdrawal, and contract cash-flow probabilities; a reproducible portfolio illustration quantifies sequential assumption effects; and a separate contractual service margin example makes the distinction between remeasurement and immediate earnings recognition explicit. Recent insurer mortality-improvement research is evaluated alongside official standards and empirical work on surrender incentives. The findings establish that the interpretation of any reserve change depends jointly on the cash-flow model, the effective measurement basis, the information date, and the financial-statement or solvency location where that change is recorded. These are analytical conclusions within existing actuarial and accounting frameworks; no new theory or empirical result is claimed.
Keywords: Life insurance; liability remeasurement; actuarial assumptions; mortality; longevity; lapse; surrender; expenses; discount rates; IFRS 17; contractual service margin; U.S. GAAP; long-duration targeted improvements; Solvency II; VM-20; principle-based reserves; assumption governance; asset-liability management; sensitivity analysis.
Introduction
Life insurance reserve revisions are a material expression of how present information changes the valuation of obligations issued in the past. Contracts can remain legally unchanged while the current actuarial estimate of their future benefits, expenses, premiums, and surrender cash flows changes substantially. Because life policies may last decades, small changes in projected transitions or discount curves can produce large changes in present value. At the same time, financial reporting standards prescribe different recognition destinations for some of those changes. The economic value of a revised stream of expected cash flows, the accounting carrying amount, and the statutory reserve are therefore separate measurements that must be reconciled rather than treated as interchangeable numbers.
The accounting dimension acquired greater international consistency with the 2023 effective application of IFRS 17 Insurance Contracts, which requires current estimates of fulfilment cash flows and recognizes insurance service profit over coverage periods. United States generally accepted accounting principles underwent a distinct change through FASB Accounting Standards Update 2018-12, which introduced regular review of cash-flow assumptions and current discount-rate updates for specified long-duration liabilities. Solvency II instead requires a current probability-weighted best estimate plus a risk margin under Directive 2009/138/EC, Article 77. The 2026 NAIC Valuation Manual supplies a further, prudentially oriented U.S. valuation setting, with risk-factor-specific prescribed procedures and margins.
These frameworks share long-duration insurance as a subject but answer different institutional questions. IFRS 17 seeks financial reporting of insurance service, finance results, and remaining unearned profit. The U.S. GAAP long-duration model addresses the measurement and reporting of specified insurance liabilities and related balances. Solvency II evaluates technical provisions in a prudential balance sheet, while U.S. principle-based reserving establishes regulatory minimums under an applicable state implementation. The same updated mortality experience is relevant to each, but it need not generate equal numerical liabilities or identical profit effects. Recognizing these different objects is the organizing condition of rigorous comparison.
The research uses a documentary and mathematical method. It compares original standards and standard-setting material, reviews published professional and empirical research, derives the consequences of selected assumption changes in a transparent cash-flow representation, and subjects hypothetical calculations to reproducibility checks. Its geographic examples are explicit about jurisdiction and scope. Publicly available source materials are distinguished from subscription-only full texts, institution-level surveys, and the original insurer-level data that would be necessary to conduct empirical reserve attribution. Monetary illustrations are constructed for this monograph and represent no real insurer.
The study answers three questions. First, what constitutes an assumption change for a defined life-contract population, and how can its liability effect be identified? Second, how do governance and reporting frameworks alter the measurement and financial presentation of otherwise similar updated experience? Third, how do assumption changes influence earnings, solvency, and liquidity, including the interactions that make simple one-variable stories misleading? The six chapters proceed from contractual and actuarial foundations through evidentiary selection, accounting comparison, institutional effects, numerical testing, and integrated findings.
Foundations, Contract Boundaries, and Measurement
The object of remeasurement
A life insurance liability is an obligation associated with contractual rights, contingencies, and future policy cash flows as evaluated under a specified basis. A prospective actuarial valuation begins with the benefit and premium clauses, the insured population, the remaining contract term, and a defined valuation date. The estimate is conditional on information available at that date. An assumption change is a revision to a prospective estimate of an uncertain future variable in the model. An experience variance records what actually occurred relative to the previous expectation. A method change modifies the mathematical mechanism that converts information into estimates. A portfolio change alters the contracts or risks being measured. These distinctions determine which part of a reported movement can reasonably be called assumption-driven.
Suppose a policy pays a death benefit when the insured dies in force, a surrender benefit when the policyholder withdraws, and collects premiums while the contract remains active. A model that includes the expected present value of all three streams cannot apply a single mortality rate independently of surrender probabilities. Surrender changes the number of exposed insureds, while mortality and contract duration influence the population available to surrender. Premium affordability, conversion options, partial withdrawals, and guarantees can introduce additional states. The document describing a product is therefore a principal source of model scope: a cash-flow projection must follow the policy's enforceable features and the relevant boundary of modeled obligations.
For a stylized discrete-time portfolio, let represent expected in-force exposures at the beginning of policy year , the conditional death probability during that period, and the conditional probability of withdrawal before the mortality period within the year. A simple ordering convention gives
Here and are expected deaths and withdrawals under the stated within-year event order. A model with continuous hazards, simultaneous competing risks, different surrender settlement dates, or contractual death benefits triggered before surrender will use different equations. The ordering is an explicit modeling choice, not a universal mortality identity. For death benefits , surrender amounts , premium inflows , attributable expenses , and annual one-period rates , the elementary economic net present-value measure is
Premiums and expenses in that expression are aggregate expected cash flows by year; the signs establish benefits, surrender payments, and expenses as outflows and premiums as inflows. The expression omits risk adjustment, financial options, reinsurance, detailed tax, actual-versus-expected settlement, and reporting-specific boundaries. It serves as a transparent reference measure, not a shortcut to an IFRS 17, Topic 944, Solvency II, or VM-20 reserve.
Sources of a reserve movement
A comparison of successive reporting dates combines several processes. Expected passage of time unwinds discounting and reduces remaining coverage. Recorded deaths, premiums, surrenders, and claims settle contractual obligations and produce experience variances. New policies and derecognition alter the population. Updated assumptions modify the projection for the still-existing contracts. New curves change the present value of those revised projections. For an intelligible attribution, analysts must first roll the old valuation forward to the new measurement date using its original assumptions, reconcile observed events, and then introduce one or more updates on a comparable contract population. Direct subtraction between two published balance-sheet amounts usually mixes all of these effects.
For a positive future net liability , the local response to a discount-rate shift is approximately , where is an appropriate modified duration and a small parallel rate shift. A life liability with substantial future premiums may have nonstandard net duration or even change sign as premium and benefit streams offset. Duration is accordingly a property of a defined net cash-flow measure, not of “life insurance” in the abstract. Convexity, changes in policyholder behavior, and nonparallel curve movements also change the response. Discount shocks and surrender shocks must be modeled consistently when surrender value is economically sensitive to market rates.
Mortality and longevity are opposite benefit exposures
An increase in death probabilities usually increases the expected early death payments on a pure term insurance contract, conditional on premiums and lapse effects. In an immediate lifetime annuity, higher death probabilities typically shorten expected payment duration, lowering annuity benefit value. Contracts combining death benefits, survival benefits, savings accounts, or minimum guarantees can have mixed sensitivity. It follows that labeling a mortality assumption “more conservative” without identifying the benefit pattern can obscure the sign of its effect. Mortality improvement is also defined relative to a baseline; a reduction in annual mortality probabilities improves survival but may increase liability for annuities and decrease the death-claim component of term coverage.
This distinction matters especially for late-2026 information. The Society of Actuaries Research Institute's 2025 Mortality Improvement Survey Report, published in October 2026 and based on 47 company-country respondents, found substantial variation in methods and post-pandemic responses rather than a single industry-wide direction. The 2026 individual life mortality scale used updated insured-population evidence and was accepted by the NAIC Life Actuarial Task Force for its specified U.S. valuation applications. By contrast, the RPEC 2026 mortality update addresses retirement-plan and population evidence and did not issue a new retirement mortality-improvement scale for 2026. These are different observed populations and different applications. Substitution of one for another requires a justified transfer analysis.
Expense and withdrawal assumptions
Projected expenses include the servicing of in-force contracts, claims handling, and contract-specific obligations. A revision to wage inflation, technology unit costs, staffing capacity, or policy administration scale can raise or lower maintenance expenses even if the number of active policies remains fixed. When lapses increase, some future per-policy expenses disappear while fixed overhead must be spread across a smaller population. Projected expense per active policy can therefore rise while total projected expenses fall. The direction must be determined from the expense function and the contract dynamics rather than from a single lapse percentage.
Surrender assumptions are likewise product-dependent. An increase in lapse rates may reduce future death-benefit and premium exposures but accelerate cash payments when guaranteed surrender values are large. A policy that is deeply valuable to a policyholder may be retained by an individual whose health has deteriorated, producing selective persistence and an insured cohort unlike the average lapsed population. For savings-oriented products, higher competing yields may make guaranteed surrender values attractive. These effects link lapses with market rates, policyholder circumstances, and mortality selection.
Measurement date and the unit of account
The actuarial estimate must have an explicit information cut-off. If a health outcome is observed in one year but only reported after a reporting date, a later dataset may support revising the next prospective estimate without automatically converting the original reporting date's judgment into an error. Subsequent information may reveal a prior error, an ordinary experience development, or a legitimately new estimate; the accounting conclusion depends on contemporaneous evidence and the applicable guidance. The IFRS Foundation's IAS 8 framework distinguishes estimate changes from error correction and accounting-policy changes. A reserve study should record the event date, exposure period, reporting date, data extraction, model release, approval date, and accounting recognition date so that those distinct questions remain auditable.
Assumption-Update Governance and Evidence
Experience selection and measurement bias
Actuarial assumption selection combines statistical estimation with knowledge of the population under contract. The observation base includes policy identifiers, insured attributes, underwriting category, issue age, duration, benefit amount, exposure, premium mode, claim status, surrender status, and contract modifications. For mortality, an actual-to-expected ratio is generally calculated as observed deaths or claims divided by expected deaths or claims after adjustment for the study's exposure convention. Ratios weighted by number of lives and those weighted by sum assured answer different questions. If a large face-amount claim changes a claims-weighted ratio, the inference about mortality frequency must be separated from the effect of severity weighting.
The U.S. ASOP No. 23 on Data Quality requires professional consideration of data relevance, sufficiency, internal consistency, limitations, and the appropriateness of reliance on supplied information. In a reserve context, a mortality assumption is therefore inseparable from denominator construction. Policies removed by prior lapses cannot remain in the same exposure denominator as currently in-force business. Claims incurred but reported late require an observation-period convention. Underwriting changes and business acquisition can alter the composition of insured lives even if underlying mortality at the individual level is stable.
A credible estimate must describe sampling and transfer. A company may have thousands of policy-years but too few deaths in an older-age cell for a stable granular estimate. Conversely, a company-specific sample may be statistically stable but inapplicable to recently underwritten products with revised eligibility rules. The ASOP No. 25 credibility guidance addresses selection and application of credibility procedures. For illustration, if observed mortality is 1.2 times a defined reference rate and a credibility factor is 0.4, a linear blend gives times the reference rate. Such a blend demonstrates arithmetic only. A legal minimum, prescribed mortality table, or a more appropriate credibility methodology may supersede it, and age-specific evidence cannot be represented by a single portfolio multiplier without loss of information.
Governance of changes
Assumption governance is a repeatable documentary process with identifiable decisions. A useful audit trail establishes the approved prior assumption, its effective use dates, the proposed update, source data and adjustment decisions, model compatibility, expected quantitative effect, sensitivity, independent challenge, management approval, and downstream accounting and regulatory interpretation. A material decision should include counterfactual valuations on identical contract snapshots, a reconciliation of reasons for change, and an explanation of whether the update arises from new information, a changed method, management policy, or a prescribed regulatory requirement.
For U.S. actuarial modeling, ASOP No. 56 on Modeling calls attention to model purpose, appropriateness and consistency of assumptions, limitations, monitoring, and documentation. The related ASOP No. 41 communications standard addresses disclosure and explanation of material assumptions, methods, reliance, and uncertainty. These professional standards do not determine the balance-sheet location of each movement; they strengthen the evidence and communication that allow accounting and prudential rules to be applied correctly.
One central control is version integrity. The same dataset, cash-flow engine and reporting interpretation should produce the same result when rerun under identical inputs, absent a genuinely stochastic implementation with controlled random seeds and numerical convergence tolerances. Model validation must compare an updated result against a frozen prior baseline, inspect significant policy cells, and test that movements attributed to mortality, lapse, expense or discounting disappear when the relevant assumption is restored. An unexplained residual is useful diagnostic evidence, but it should not automatically be forced into “other assumptions”: it may reflect a model correction, the interaction of assumptions, rounding or differences in event sequencing.
Mortality updates after the pandemic
The interpretation of recent mortality experience requires distinguishing broad trends from the insured risks actually held. The October 2026 SOA company survey describes heterogeneous company practice, with improvement assumptions varying by product, function, age and projection horizon. The September 2026 AG38/VM20 life improvement recommendation applies a specific insured-population-informed methodology and contains age- and sex-specific design. The October 2026 RPEC retirement-population assessment addresses a different population and found that results for ages 65 and above had broadly returned near an MP-2021-based benchmark. A company that uses these publications should identify whether its own book is individual life insurance, payout annuity, insured pension, or another population and must avoid assuming that published population-level inference determines individual contract valuation.
An improved life table can be revised for a new valuation date while the contractual death benefit remains constant. The updated reserve depends on the pattern of future age-specific benefits and premium continuation. The claim that an entire insurer's net liability rises or falls from improved mortality alone requires product weights, reserve basis, and competing risk effects. The proper empirical claim is narrower: if the relevant death probabilities decrease and all other variables, premiums, and benefit definitions remain fixed, the death-claim component on pure term coverage normally decreases. The actual net liability may move differently because lower deaths extend the period during which expenses and premiums arise.
Lapse studies, economic conditions, and adverse selection
An assumption update for surrender requires more than a historical observed lapse proportion. When cash values are guaranteed, the opportunity cost of keeping the policy depends on replacement yields, tax consequences, surrender charges, access to credit, and the perceived value of embedded benefits. Kubitza, Grochola, and Gründl (2025) report empirical evidence that higher interest rates increase surrenders on savings-oriented life products and use a calibrated model to examine liquidity and asset-sales consequences. Their conclusion addresses an identified product setting and their empirical and calibrated-model results should not be applied uniformly to protection-only term insurance. Earlier literature on voluntary termination in the U.S. market also illustrates the need to treat lapses as behavior conditional on contract and economic features rather than as an exogenous constant.
A practical dynamic lapse model allows surrender probabilities to depend on duration, premium-payment status, surrender-value moneyness, interest-rate spreads, policyholder age, guarantees, and observable economic conditions. It may also require constraints where experience is thin. The Solvency II policyholder-behavior rule expressly requires retrospective and prospective analysis of contractual-option exercise, economic conditions, and management actions. It permits independence assumptions only where supported by empirical evidence. For a model already sensitive to higher interest rates, a shock that raises the discount curve while retaining an unchanged surrender rate can be internally inconsistent if the product's surrender option becomes more attractive.
Expenses, inflation, and allocation
Expense updates should distinguish controllable unit expenses, inflation assumptions, overhead allocation, and costs contractually linked to benefits. Under Solvency II, Delegated Regulation 2015/35, Article 31 identifies relevant expense categories and demands realistic, objective overhead allocation. A company that anticipates a technology-based reduction in per-policy expense must support implementation timing, operational investment, and the future volume of active business. Assuming a unit-cost reduction across all future years while simultaneously forecasting rising near-term transformation costs would omit part of the projected cash-flow story.
The difference between nominal and real expense rates is similarly consequential. If expenses in year are forecast as , where denotes expense inflation, a one-percentage-point change in has a compounded effect on distant cash flows. Discounting all future expenses at a nominal market rate while retaining constant real expenses without escalating nominal cost can materially understate the liability. The Solvency II Directive's Article 78 expressly calls for expenses and inflation to enter technical provisions. The information date and historical version of this provision should be identified in any jurisdiction-specific replication.
Separating material revisions from ordinary noise
Statistical significance and financial materiality are different concepts. A mortality factor change can be supported by a statistically precise study but have limited reserve impact if the block is small, fully reinsured, or short-duration. A change that is statistically uncertain can be financially important for a large annuity portfolio. Source credibility, model uncertainty, and balance-sheet sensitivity should therefore be reported separately. A decision may reasonably retain an assumption while describing its range, or it may adopt a measured revision with explicit uncertainty. Governance should avoid silently converting a sensitivity test into the selected base assumption.
For each assumption the record should identify the sign and magnitude of the expected liability effect, the most relevant competing variables, and the reporting metric affected. Mortality improvement for death cover, expense inflation for servicing, lapse shifts for surrender benefits, and discount movement for the long end of the term structure are different stress axes. Reporting only the percentage by which an input changed does not reveal the reserve implication. A change in a probability from 0.2% to 0.3% is a 50% relative increase but only a 0.1-percentage-point absolute change; the relevant cost is the affected present value of payments, not the headline percentage.
Reporting Framework Differences
IFRS 17: current fulfilment flows and future service
IFRS 17 measures groups of in-scope insurance contracts using fulfilment cash flows and, for relevant profitable groups, a contractual service margin representing unearned profit. Fulfilment cash flows comprise current estimates of future cash flows adjusted for the time value of money and financial risk and an explicit nonfinancial risk adjustment. The IFRS 17 accounting model recognizes insurance service results separately from insurance finance income or expenses, and a group becoming loss-making is recognized as such immediately. Its contracts are organized by the standard's recognition and grouping rules; changes to one policy cannot simply be netted against profitable unrelated contracts at the insurer's discretion.
The timing and character of the revised estimate determine the effect on the contractual service margin. For contracts under the general measurement model, changes in estimates of future cash flows and the nonfinancial risk adjustment relating to future service normally adjust the CSM, subject to the standard's requirements and limitations. Changes relating to incurred claims or current and past service enter the insurance service result as appropriate. Insurance finance effects arising from the time value of money and financial risks have a different recognition pathway. The publisher's IFRS 17 key-term materials help distinguish fulfilment cash flows, CSM and insurance contract liability components. The complete treatment depends on whether the group uses the general measurement model, variable fee approach, or eligible premium allocation approach; broad statements about all IFRS 17 contracts must identify those measurement differences.
A deterioration in expected future mortality on a profitable group of traditional term policies may increase fulfilment cash flows. When the deterioration relates to future insurance service, the CSM ordinarily absorbs the adverse change to the extent permitted. The insurer has less unearned profit remaining, and that loss of future profitability may not appear immediately as an equal amount of current-period insurance service expense. Once the available margin is exhausted, further adverse future-service revisions can make the group onerous and cause immediate loss recognition. Conversely, favorable experience changes can create or restore future-service profit subject to the particular loss-component and CSM mechanics. A published aggregate liability increase must thus be read with the separate change in the CSM, relevant risk adjustment, and finance components.
The distinction is reinforced by the fact that a market discount-curve movement changes the value of future cash flows without being equivalent to a new nonfinancial risk estimate. Under the general measurement model, aspects of finance measurement can enter profit or loss or, under the permitted accounting-policy election, be disaggregated so that some effect is recorded in other comprehensive income. Financial assumptions also interact with participation features, options and guarantees. The selected accounting-policy approach and the nature of the contract determine the display of the movement. An observer who interprets every liability change as an immediate insurance service loss is therefore likely to misread both earnings and future profitability.
IFRS 17 and the accounting estimate distinction
The nature of a revised assumption should be documented even when a standard has a special measurement requirement. A change based on improved experience can be a new accounting estimate, while failure to apply data already available or a misapplication of an accounting requirement may raise a prior-period error question. The IAS 8 treatment of estimation provides the general context, but it is IFRS 17 that governs the specific measurement and allocation for the insurance contract. Applying a generic prospective-estimate rule without reading the insurance-specific CSM and finance provisions would be incomplete. The explanatory materials used here summarize the public official standard and do not substitute for a licensed full-text implementation study in each jurisdiction.
U.S. GAAP LDTI: differentiated income and OCI effects
The U.S. accounting route is materially different. Under FASB's Topic 944 targeted improvements, specified traditional and limited-payment long-duration contracts require cash-flow assumptions to be reviewed and updated, if changed, at least annually. The discount-rate assumption is updated at each reporting date using an upper-medium-grade, low-credit-risk fixed-income instrument yield. For the affected liability for future policy benefits, the estimate change from cash-flow assumption updating is recognized in net income while the discount-rate updating effect is recorded in other comprehensive income. The standard also changes the approach to deferred acquisition cost amortization, market risk benefits, and rollforward disclosures. The measurement must be applied to the relevant ASC 944 contract category rather than generalized to every insurer liability.
The cash-flow remeasurement has a specific technical structure. The update revises net premium calculations using actual historical and newly expected future cash flows, applies the revised assumptions on the appropriate retrospective basis for deriving the revised liability, and distinguishes the resulting remeasurement gain or loss from the discount-rate update. The resulting net income effect cannot be reproduced by simply discounting a revised future outflow vector at a new market rate and booking the entire difference as current expense. The standard specifies an original-rate basis for elements of the cash-flow assumption remeasurement, and a separate current-rate effect on the carrying value. Cohort-specific cash flows, the net premium ratio cap, and the designated benefit accounting are important operational constraints.
The original 2018 publication's transition dates were subsequently deferred. FASB ASU 2020-11 amended effective dates: the model became effective for public business entities meeting the specified non-small-reporting-company conditions for fiscal years beginning after December 15, 2022, and for other affected entities for fiscal years beginning after December 15, 2024, with related interim-period provisions. The relevant reporting entity and filing period must therefore be identified before comparing old and new U.S. disclosures. Statements that all U.S. life insurers adopted LDTI in one calendar year disregard the staggered schedule and the distinction between generally accepted accounting principles and statutory reporting.
The contrast with IFRS 17 is consequential. A negative revision to expected future mortality under LDTI may affect current-period net income through a liability remeasurement, whereas a future-service change under IFRS 17 may reduce the unearned contractual service margin before generating an immediate loss. Both frameworks provide information about changed expectations, but a cross-border comparison of current earnings must control for their different reporting mechanisms and contract grouping. Even an identical economic cash-flow model would not make the two accounting results directly interchangeable.
Solvency II: current best estimate, risk margin, and available own funds
European prudential valuation has a different object. Under Solvency II Article 77, technical provisions ordinarily consist of a best estimate and a risk margin. The best estimate is a probability-weighted present value of all future cash flows necessary to settle obligations, discounted using the relevant risk-free term structure and reflecting credible current information. The risk margin is a separate prudential component with its own prescribed calculation. The technical provision is consequently not the IFRS 17 CSM-bearing liability and does not reproduce its immediate income or deferred-profit allocation.
The general assumption conditions under Delegated Regulation Article 22 require assumptions to be justified, linked to portfolio characteristics, and used consistently within relevant groups. Article 26 on policyholder behavior requires past-behavior analysis and prospective assessment of option exercise in its economic context, while Article 31 on expenses specifies a wide range of expense cash flows and realistic allocation of overhead. These provisions make the relationship between mortality, surrender, expenses, and management actions a substantive part of the best-estimate valuation.
An increase in the best estimate, other things equal, reduces the excess of asset values over technical provisions, and therefore may reduce eligible own funds, although matching asset movements, changes in risk margin, taxation, deferred adjustments, and other balance-sheet effects can modify the result. A reserve increase does not, by itself, specify the change in the Solvency Capital Requirement, which is calculated using separate risk modules or approved internal modeling methods. Consequently the solvency ratio, broadly eligible own funds divided by SCR, can move because of changes in numerator, denominator, or both. The particular undertaking, valuation date, applicable transitional measures, and national supervisory implementations are essential to an actual ratio analysis.
The legal text itself has a version history. EIOPA's rulebook is a documentary access point; for binding interpretation the authentic EU legislation and its effective amendments remain authoritative. This study cites the older consolidated Solvency II Directive provisions to document the treatment of expenses, inflation, options and surrender, while recognizing that later legislative changes may have future effective dates and may require separate period-by-period evaluation. A researcher must neither retroactively apply a future effective rule to 2026 figures nor assume that a historic consolidated version captures every national implementation detail.
U.S. statutory reserves: prescribed procedures and prudence
The United States regulatory measure should be separated from U.S. GAAP financial statements. The NAIC 2026 Valuation Manual, particularly VM-20 for covered life products, combines prescribed principles and reserve components with modeled deterministic or stochastic elements and risk-factor assumption governance. Applicable state insurance law determines operative requirements. VM-20 addresses mortality, persistency, expenses, margins, material risks, and documentation. The actuarial framework includes insurer experience where relevant and sufficiently credible, industry data where appropriate, and margins that can increase modeled reserves in response to uncertainty.
The ASOP No. 52 guidance elaborates how actuaries should consider assumptions, credibility, materiality, scenario-dependent policyholder behavior, and uncertainty for covered principle-based reserves. It also makes clear that selected future mortality improvement may be restricted by applicable valuation rules. It would be inaccurate to import a newly published mortality improvement scale into VM-20 solely because it seems statistically reasonable, without checking the valuation-year permissions and prescribed approach. Conversely, that limitation on certain statutory practices should not be assumed to govern IFRS 17 current expectations or the company's U.S. GAAP model.
A U.S. statutory reserve and a U.S. GAAP liability may thus diverge for the same contract cohort in the same year. The statutory number answers a prudential requirement and may include adverse deviation margins or regulatory floors. The GAAP number answers a different financial-reporting measurement and its recognition destination. This difference is fundamental when investigating dividend capacity, ratings discussions, and the relationship between earnings surprises and regulatory capital.
Comparative matrix
| Question | IFRS 17 general measurement model | U.S. GAAP long-duration targeted improvements | Solvency II technical provisions | U.S. VM-20 life reserves |
|---|---|---|---|---|
| Central measurement | Current fulfilment cash flows with risk adjustment and, where applicable, CSM | ASC 944 liability for future policy benefits under prescribed group and net-premium mechanics | Current best estimate plus prudential risk margin | Prescribed minimum reserve including relevant principle-based components |
| Mortality / lapse / expense updating | Current estimates; future-service changes may adjust CSM | Review and update if changed at least annually for specified liabilities | Current credible and realistic assumptions | Current valuation-year requirements, prudent estimates and margins where prescribed |
| Discount movement | Insurance finance result, subject to relevant measurement and OCI election | Required current-rate update; specified effect in OCI | Current relevant risk-free curve in best estimate | Valuation-manual-prescribed modeling and investment assumptions |
| Immediate earnings implication | Depends on future versus current or past service, CSM, group profitability, and finance presentation | Cash-flow assumption remeasurement enters net income; discount update OCI | Prudential own-funds and technical-provisions implications | Statutory financial position and regulatory minimums; GAAP income linkage is indirect |
| Major comparability obstacle | CSM, risk adjustment, group, measurement variant | Net-premium ratio and original/current discount mechanics | Risk margin, own funds, SCR, supervisory measures | Floors, margins, scenario tests, state adoption |
This is a functional comparison of covered cases, not a replacement for the complete recognition and measurement rules. The general measurement model does not exhaust IFRS 17 variants, and Topic 944's long-duration changes do not govern every U.S. insurance product. In a comparative empirical study, the accounting basis must be specified before a reserve number can be interpreted.
Profit, Solvency, and Liquidity Effects
Economic versus reported change
An unfavorable revision to projected future net cash flows usually raises a gross liability measure, but its immediate effect on profits depends on accounting mechanics. A useful analytical decomposition distinguishes (i) the change in estimated remaining fulfilment cash flows; (ii) the allocation of that change to unearned service profit or an immediate insurance result; (iii) changes in financing from discounting and financial risk; (iv) the independent behavior of assets and reinsurance; and (v) regulatory capital consequences. Treating all five as a single “reserve hit” conceals precisely the mechanisms that management and users need to understand.
The difference also has consequences for time-series analysis. Consider two insurers with identical expected mortality deterioration in equally sized term-risk portfolios. One reports on an IFRS 17 basis with a positive contractual service margin. The other reports on the covered U.S. long-duration GAAP model. Their reported current earnings responses may differ substantially even if the underlying revised discounted death-claim estimate is the same. If one also has a different rate curve, guarantee profile, reinsurance structure, or in-force premium stream, comparison of aggregate net reserves will differ further. An observed cross-company correlation between assumption updates and profits is therefore not a causal estimate of the economic cost of mortality change without these controls.
Discount-rate risk and financial matching
Life liabilities often have longer duration than available assets and may contain policyholder options that respond nonlinearly to interest rates. Increasing discount rates lowers the present value of fixed future positive liabilities in a simple deterministic calculation, yet the same higher-rate environment can raise surrenders and cause earlier cash payments on products with attractive guaranteed surrender values. Such withdrawals can require asset sales at market values below book amounts. On the other hand, a fall in rates may increase the valuation of guaranteed future benefits while assets rise less than liabilities if duration is mismatched. Interest-rate changes should consequently be understood through both valuation and liquidity.
The empirical and model-based study Life Insurance Convexity (2025) develops this interaction for savings-oriented life products: guaranteed surrender options become more attractive in higher-rate environments, and the modeled response can entail asset sales and broader market consequences. The study's market and model-specific estimates are evidence for a mechanism, rather than universal numerical conversion factors for unrelated contracts or jurisdictions. Its publication also illustrates why a duration estimate that fixes surrender behavior while moving the curve can miss a major component of the actual balance-sheet response.
For a conventional fixed benefit with no optional behavior, one may estimate a present-value decline using discount duration. For a surrender-sensitive contract, the effective liability derivative also contains the induced change in the underlying expected cash flows. Formally, if , then the total derivative is
This equation is an application of ordinary multivariable differentiation. Its interpretation depends on the functional model: lapse behavior may depend on age, account value, surrender-charge schedule and many other variables. Neither derivative is necessarily constant. For current cash-flow models, scenario-consistent rather than independently shocked inputs are therefore essential to realistic stress evaluation.
Profit emergence and future service
An important IFRS 17 effect of assumption revisions is their redistribution of expected profitability across periods. The CSM exists to recognize unearned profit as insurance service is provided. An adverse future-service assumption revision that reduces positive CSM weakens the expected profit remaining to emerge, even when the immediate insurance service result is partly insulated. This creates a genuine economic implication: the revised policy block is expected to generate a lower future margin. Users who focus only on one quarter's reported net income can overlook it. At a sufficiently adverse revision, the loss-making test triggers a different immediate profit effect.
The U.S. LDTI approach was designed to update specified cash-flow assumptions in a more timely fashion than prior locked-assumption practice. Its liability remeasurement gain or loss enters earnings for the affected category; the discount-rate-related change is presented in OCI. The contrast means that the same assumption revision can have different temporal recognition paths. One should compare updated expected future claims and service margins as well as current earnings, and should avoid presenting an IFRS CSM reduction as if it were numerically identical to a U.S. GAAP current-period charge.
Solvency resilience and management actions
A prudential balance sheet registers a different set of consequences. The Solvency II best estimate responds to updated mortality, lapse, expenses and discount assumptions, with risk margin and capital requirement effects considered separately. A lower value of technical provisions can improve own funds if assets and other elements remain constant, but that favorable movement may coexist with worsening liquidity or heightened lapse stress. A higher valuation of obligations may reduce own funds even if a regulator's risk measure declines. Both variables enter the solvency ratio and its interpretation.
Management responses can alter the projection itself: discretionary crediting, premium-rate changes where legally allowed, operational cost programs, hedging, and reinsurance all influence future cash flows. Only actions permitted by law and contract, and reasonably supportable in the model, can be treated as available. For Solvency II, the general rule on realistic assumptions and policyholder-behavior provisions require consistency; there is no license to assume perfect future management action merely to offset an unfavorable stress. Changes to credited rates may themselves change surrender behavior. The time required to implement a management decision is economically material.
Risk transfer and reinsurance dependence
Reinsurance can materially alter the net effect of life mortality revisions. Quota-share treaties or mortality excess protection may offset part of expected death-claim increases, but the recovery estimate depends on treaty terms, exhaustion conditions, creditworthiness, and the timing of recoveries. A gross liability revision is not the net exposure to the ceding insurer. Ceded cash flows may also be measured under separate accounting requirements, so direct netting within a primary insurance group is not always permitted in presentation. The analytical record should reconcile gross projected benefit change, eligible ceded recovery change, credit effects, and any timing mismatch.
Reinsurance does not remove every risk dimension. A block may be materially reinsured for mortality yet remain exposed to maintenance expenses and mass lapses. Under a stress involving simultaneous high surrenders and wider credit spreads, counterparties and collateral arrangements can become more important. Comparing assumption changes without mapping the applicable treaty can lead to incorrect attribution of apparent reserve volatility to the underlying insured lives alone.
Policyholder protection and distributional effects
Assumption revisions can have consequences for policyholders even though their contractual rights are unchanged by an actuarial estimate. Higher prudential reserve requirements may support financial resilience while reducing the insurer's distributable capital or prompting changes in new-business pricing. Surrender-sensitive products can expose policyholders to opportunity costs, tax consequences, or option exercise decisions during market disruption. Premium changes on contracts with permitted adjustable elements can alter persistency and selection. These effects are conditional on the actual contract and national consumer-protection framework. A cross-country monograph can identify these mechanisms but cannot assert that all jurisdictions permit the same policy adjustments.
There is a further distributional issue within insurance cohorts. An improved aggregate mortality assumption could conceal deteriorating results for a subgroup of insureds and favorable results for another. If the model uses only a portfolio-average correction, reserves may be understated for the higher-risk cohort and overstated for another. The company must evaluate whether the grouping used for experience estimation is also appropriate for valuation and whether different national rules allow or prescribe different reserve groupings. The practical consequences include capital allocation, product design, reinsurance and the interpretation of reported insurance service margins.
Comparative Quantitative Evaluation
Principles of a reproducible illustrative calculation
A numerical assumption bridge is informative when the contractual cash-flow timing, exposure convention, accounting scope, and order of changes are explicit. This section constructs its own simplified calculations. They are neither samples of confidential insurer data nor computations of formal IFRS 17 fulfilment cash flows, U.S. GAAP Topic 944 liabilities, Solvency II technical provisions, or VM-20 reserves. The examples isolate mechanics and demonstrate why the economic bridge and the reporting entry must be considered separately. Currency units are U.S. dollars, displayed to the nearest dollar unless stated otherwise.
Consider a synthetic portfolio of 10,000 identical one-year policies, each with a $100,000 death benefit. A contract may surrender at the beginning of the year, then an insured who remains in force may die during the year. Claims, administrative expenses, and premiums settle at the end of the year. An annual premium of $400 is paid for every contract, regardless of whether the contract subsequently surrenders during the period, and a $75 per-contract administrative charge is incurred by the insurer. Both are deliberately simplified conventions; they avoid making a separate surrender benefit or refund assumption. The one-year end-of-period expected net outflow under this stylized product is
The present value is . The baseline assumptions are , , , and . Expected deaths are , expected death claims $3,760,000, expenses $750,000, and premiums $4,000,000. Net expected year-end outflow is $510,000 and its present value is approximately $495,146.
Now update the death probability to 0.0045, the surrender probability to 0.08, unit expenses to $80, and the annual discount rate to 0.035. With all changes applied, expected death claims are , expenses are $800,000, premium inflows remain $4,000,000, and net outflow is $940,000. The revised present value is approximately $908,213. The change of approximately $413,067 is a change in the example's net present value under the stated conventions, not an immediate accounting loss.
Sequential attribution and interaction
A transparent sequential bridge first changes mortality, then surrender, then unit expenses, and finally discounting, retaining previous revisions at each step. The values are:
| Illustrative valuation state | Expected net year-end outflow | Present value |
|---|---|---|
| Initial assumptions | $510,000 | $495,146 |
| Mortality updated to 0.0045 | $980,000 | $951,456 |
| Surrender then updated to 0.08 | $890,000 | $864,078 |
| Unit expenses then updated to $80 | $940,000 | $912,621 |
| Discount then updated to 3.5% | $940,000 | $908,213 |
The corresponding sequential increments, subject to rounding, are +$456,311 for mortality, −$87,379 for surrender, +$48,544 for expenses, and −$4,409 for discounting. Their sum equals approximately +$413,067. The signs arise from the conventions selected: increased surrender prevents some simulated deaths and carries no surrender payment, while the assumed premiums are retained. With a guaranteed surrender benefit, surrendered policies may generate a separate and substantial payment; if premiums cease after surrender, their loss can add to the liability. The sign of surrender sensitivity is consequently not generalizable from this example.
Sequential attribution is path-dependent when cash flows depend jointly on multiple assumptions. In the equation above, mortality and lapse appear as the product . If surrender is updated before mortality, the attributed effect of the mortality revision uses a smaller exposed population, and more of the interaction is assigned to surrender. The total change remains the same; the labeling of individual contributions changes. A governance report can either disclose its ordered bridge or calculate a symmetric average across orders. In the case of two assumption dimensions, an equal-order allocation gives each variable half of the interaction term. This is a standard algebraic attribution practice, not a new theory. It prevents a chosen update sequence from being mistaken for evidence of a uniquely identifiable causal decomposition.
Discount duration as a separate stress
For a single positive payment of $10 million in ten years, with no optionality or further cash flows, its present value at 3% is , approximately $7,440,939. At 4%, the present value is approximately $6,755,642, a decrease of about $685,297. The example has large discount sensitivity because of the long payment horizon. It demonstrates mechanical present-value sensitivity; it says nothing about the real insurer's interest-rate hedges, assets, contractual options, locked-versus-current accounting rates, or corresponding reserve floors. The observed net capital result of an interest-rate movement may be quite different.
If contractual benefits are payable at multiple times, a duration calculation can weight future cash flows by present value, but premiums are negative cash flows, while surrender and mortality react to the same economic environment. This introduces economic interaction and may make the net-liability duration unstable when the net present value is near zero. For that reason a reported 'reserve duration' should specify whether it describes gross benefits, net insurance cash flows, current IFRS fulfilment cash flows, or another balance. Sensitivity to the full discount curve is preferable to a single parallel-rate statistic for a long-term guarantee portfolio.
Contractual service margin absorption example
Assume a profitable IFRS 17 group under the general measurement model has a positive CSM of $3 million immediately before a qualifying future-service change, and that all other mechanics are fixed for illustration. An unfavorable increase of $2.4 million in the fulfilment cash flows relating to future service reduces the CSM to $0.6 million. The example has a genuine $2.4 million adverse revision in expected future service cash flows, but the illustrative change is absorbed into unearned profit rather than booked automatically as an equivalent immediate insurance service loss. A further unfavorable future-service change of $1 million, with no offsetting effects, would exhaust the remaining $0.6 million and leave an illustrative $0.4 million excess requiring onerous-group loss treatment.
Actual IFRS 17 calculations apply specific rules for the CSM rollforward, any loss component, financial versus nonfinancial changes, contract measurement variants, and the relevant discount basis. The calculation intentionally isolates a single future-service CSM mechanism. It must not be used as a complete journal-entry template. It does clarify a comparative point: a reporting analyst must track both immediate profit-or-loss recognition and the reduction in expected future profitability. In contrast, an otherwise comparable expected cash-flow assumption revision for a covered U.S. Topic 944 future-policy-benefit liability would follow the prescribed net-premium and current-income remeasurement mechanics rather than this CSM demonstration.
An insurer-level study design
An empirical study capable of generalizing these mechanisms would assemble at least six layers of information. The contract layer identifies product features, issue dates, premium schedules, surrender terms, guarantees and reinsurance. The exposure layer identifies in-force populations, age, underwriting and coverage. The experience layer links deaths, surrenders, premiums and expenses to the appropriate periods. The assumption layer records each approved version, effective date, supporting study and judgmental overlay. The valuation layer includes model version, curve, valuation date, scenario set and reserve basis. The financial layer reconciles gross and net liability rollforwards, CSM or earnings entries, capital requirements, assets, and the recognition period.
The first empirical comparison would hold the contract and model version fixed while revaluing old and revised assumptions at the same date. A second comparison would use the actual prescribed reporting basis to identify recognition destinations. A third would compare local sensitivity derivatives with joint scenario shocks, including lapse–rate interactions, reinsurance, and expenses. Finally, a cross-insurer analysis would normalize business mix, financial reporting frameworks, calendar dates and the interpretation of assumptions, with documented treatment of outliers and missing data. Such research would require access to insurer records and legally permitted disclosure of data. This monograph offers the design and interpretation, not an invented empirical sample or a claim that the methods have already been fitted to proprietary observations.
Statistical uncertainty and falsification
A reserve-impact narrative should be evaluated against alternative explanations. A higher reported mortality ratio could arise from a true change in conditional mortality, changed underwriting selection, altered exposure calculation, reporting delays, unusually large claims, or a different insured-age mix. A decline in total surrenders could arise from higher customer persistence or from a shrinking block with fewer policies eligible to surrender. An apparently favorable expense trend could reflect deferral or reclassification of expenses rather than lower costs of administering current obligations. An apparent discount-rate effect could contain changes in credit spreads or the measurement curve itself. Each hypothesis has a different counterfactual test.
Quantitative reports should therefore give the uncertainty around estimated assumptions and reserve effects rather than presenting a point estimate as a physically observed liability. Model risk grows when a long projection horizon exceeds the observed experience. Combining highly correlated stress variables can produce more realistic adverse states than summing independently stressed effects. The effect of an assumption revision is also sensitive to whether it applies to contracts remaining in force and whether reported historical experience was already used in the prior model. Scientific reproducibility requires the complete input version, event order, formula, scenario definition, and rounding rule.
Integrated Scientific Findings
Four observations from the comparative evidence
The first finding is that a liability revision has at least two meanings. Its economic meaning is the movement of a well-defined valuation of remaining obligations under changed expectations. Its reporting meaning is the movement of a carrying amount or reserve under particular recognition and prudence requirements. The two can be aligned in some simplified cases but remain conceptually distinct. The legal contract determines the future payment rights; actuarial assumptions determine an expected valuation; accounting and regulatory rules determine the reportable measure. An explanation that begins with 'reserves increased because mortality worsened' is incomplete until these links are specified.
The second finding is that one assumption cannot always be changed independently in a meaningful economic counterfactual. Mortality, lapse and expenses are linked through the active policy population, while market rates influence both discount factors and surrender incentives. The original illustrative computation shows an explicitly favorable lapse effect under its artificial timing and no-surrender-payment conditions; the product-specific research on interest-sensitive surrender behavior demonstrates why that sign can reverse in practice. A valid assumption bridge therefore separates mechanical conditional sensitivities from joint economic scenarios and documents the choice of decomposition.
The third finding is that profitability and capital effects cannot be inferred from the sign of the present-value change alone. For profitable IFRS 17 general-model groups, some unfavorable future-service updates reduce the CSM before current loss recognition. U.S. Topic 944 targeted improvements require a differentiated income versus OCI presentation for specified liability updates. Solvency II changes the best estimate and potentially eligible own funds, while the risk margin and solvency capital requirement must be evaluated separately. The IFRS 17 measurement principles, FASB long-duration amendments, and Solvency II valuation rule thus produce distinct institutional pathways for a related actuarial input.
The fourth finding is that new evidence demands evaluation for transferability. A 2026 mortality-improvement publication is newer than an insurer's existing selected table, but recency alone does not identify the insured population, contract cash-flow exposure, horizon, or applicable valuation-year permission. The 2026 insured-life scale and 2026 retirement mortality update serve different populations and regulatory purposes. A transparent selection memo must record why one source is relevant to the modeled policies, how it is adjusted, and whether binding requirements constrain its adoption.
Implications for comparison and accountability
A scientifically credible assumption-change report should identify six directly testable items: the exact affected contract group; the old and revised assumption and their information dates; the controlled impact on projected cash flows; the reporting basis and allocation of the movement; the asset, reinsurance and capital interactions; and the governance evidence supporting selection. These items follow from the documented differences in original standards, multistate cash-flow algebra, and the observed importance of surrender incentives. They are analytical reporting criteria arising from existing knowledge, not a newly established proprietary system or doctrine.
The scope of the finding is international but not universal in its empirical detail. IFRS 17 adoption varies by applicable jurisdiction and entity; the Solvency II assessment discussed here is European; the U.S. GAAP discussion concerns the particular long-duration contracts subject to the FASB amendments; and VM-20 is a U.S. regulatory valuation mechanism subject to state adoption and valuation-year rules. The study does not estimate a single global reserve shift attributable to post-pandemic mortality, interest-rate changes or expense inflation. It establishes the methods and interpretive conditions that such estimates would need to satisfy.
Research limitations
Primary-source access was strongest for the publicly readable IFRS Foundation summary, FASB issued update text, EIOPA rulebook provisions, NAIC Valuation Manual, Actuarial Standards Board texts, and the Society of Actuaries' research descriptions. The journal articles on surrender behavior were assessed from publisher-accessible bibliographic and abstract information and selected openly visible explanatory material; their subscription-only underlying full texts and microdata were not independently replicated. This distinction limits claims about econometric identification, external validity, coefficient precision, and the numerical scale of any general effect. No proprietary life-insurer cash-flow model, claims data, policy form archive, audited financial statement sample, or independent external peer review was used.
The numerical evaluations explicitly hold many real-world features fixed. They exclude explicit mortality-dependent premiums, surrender benefits, policy dividends, stochastic guarantees, deferred taxes, reinsurance receivables, regulatory floors, and multi-year reporting rollforwards. They nevertheless establish valid mathematical consequences of stated assumptions and can be recalculated without access to a vendor actuarial system. Future empirical work should audit the informational provenance of updated assumptions, compare insurer-specific versus reference data, and reconcile reserve bridges to both public financial results and prudential disclosures. Generalization beyond this dossier requires corresponding evidence, not only additional numerical precision.
Conclusion
Life insurance assumption changes alter the estimated cost and timing of long-duration contractual obligations. Mortality, withdrawal behavior, expenses and market rates interact through survival, persistence, policy options and the discounting of future cash flows. The act of remeasurement therefore requires a defined contract population, an explicit observation and valuation date, a consistent cash-flow model, and an auditable account of why each assumption changed.
The comparison establishes that liability remeasurement and immediate profit recognition occupy distinct analytical levels. IFRS 17 allocates certain future-service changes through the contractual service margin, while the FASB's U.S. long-duration improvements use differentiated earnings and OCI treatment for specified assumption and discount-rate effects. Solvency II measures a current best estimate plus risk margin, with implications for own funds and potentially capital requirements, and U.S. VM-20 operates under prescribed regulatory reserve and margin procedures. The same economic revision may therefore have different reported effects across frameworks, periods and products.
The independently reproducible examples demonstrate both the significance of mortality and expense sensitivity and the interaction-dependent sign of lapse changes. Their purpose is explanatory: no real insurer observations or newly created theoretical principles are claimed. The scientific result is a structured interpretation of existing standards and evidence showing how an assumption revision becomes an economic valuation change, a recognized accounting movement, and a prudential or liquidity consequence. Each interpretation becomes reliable when the corresponding contractual, evidentiary and reporting basis is preserved.
References
Actuarial Standards Board (2010). Actuarial Standard of Practice No. 41, Actuarial Communications. Adopted December 2010; effective May 1, 2011.
Actuarial Standards Board (2013). Actuarial Standard of Practice No. 25, Credibility Procedures. Adopted December 2013; effective May 1, 2014.
Actuarial Standards Board (2016). Actuarial Standard of Practice No. 23, Data Quality. Adopted December 2016.
Actuarial Standards Board (2017). Actuarial Standard of Practice No. 52, Principle-Based Reserves for Life Products under the NAIC Valuation Manual. Adopted September 2017; effective December 31, 2017. Document No. 189.
Actuarial Standards Board (2019). Actuarial Standard of Practice No. 56, Modeling. Adopted December 2019; effective October 1, 2020. Document No. 195.
Clark, Kara (2026). 2025 Mortality Improvement Survey Report. Society of Actuaries Research Institute. October 15, 2026.
European Commission (2015, as amended). Commission Delegated Regulation (EU) 2015/35, Article 22: General Provisions on Assumptions. EIOPA Solvency II Single Rulebook.
European Commission (2015, as amended). Commission Delegated Regulation (EU) 2015/35, Article 26: Policyholder Behaviour. EIOPA Solvency II Single Rulebook.
European Commission (2015, as amended). Commission Delegated Regulation (EU) 2015/35, Article 31: Expenses. EIOPA Solvency II Single Rulebook.
European Parliament and Council (2009, as amended). Directive 2009/138/EC (Solvency II), Article 77: Calculation of Technical Provisions. European Insurance and Occupational Pensions Authority, consolidated rulebook.
European Parliament and Council (2009; consolidated version June 30, 2021). Directive 2009/138/EC on the Taking-up and Pursuit of the Business of Insurance and Reinsurance (Solvency II), Articles 78–79. EUR-Lex.
Financial Accounting Standards Board (2018). Accounting Standards Update No. 2018-12 — Financial Services—Insurance (Topic 944): Targeted Improvements to the Accounting for Long-Duration Contracts. August 2018.
Financial Accounting Standards Board (2020). Accounting Standards Update No. 2020-11 — Financial Services—Insurance (Topic 944): Effective Date and Early Application. November 2020.
IFRS Foundation (undated, consulted November 2026). IFRS 17 Insurance Contracts—Key Terms. Official IFRS Foundation supporting implementation material.
International Accounting Standards Board (2017, amended 2020). IFRS 17 Insurance Contracts. IFRS Foundation, official standard overview and historical amendments.
International Accounting Standards Board (current consolidated standard overview). IAS 8 Basis of Preparation of Financial Statements — Accounting Policies, Changes in Accounting Estimates and Errors. IFRS Foundation.
Jiang, Shi-jie (2010). “Voluntary Termination of Life Insurance Policies: Evidence from the U.S. Market.” North American Actuarial Journal 14(4): 369–380. https://doi.org/10.1080/10920277.2010.10597596.
Kubitza, Christian, Nicolaus Grochola, and Helmut Gründl (2025). “Life Insurance Convexity.” Journal of Banking & Finance 178, article 107502. https://doi.org/10.1016/j.jbankfin.2025.107502.
National Association of Insurance Commissioners (2025). Valuation Manual, January 1, 2026 Edition; particularly VM-20, Requirements for Principle-Based Reserves for Life Products.
Society of Actuaries Research Institute, Mortality Improvement Life Working Group (2026). Individual Life Insurance Mortality Improvement Scale—For Use with AG38/VM20—2026. September 22, 2026.
Society of Actuaries Research Institute, Retirement Plans Experience Committee (2026). RPEC 2026 Mortality Improvement Update. October 22, 2026.