RESEARCH DOSSIER
Gender-Neutral Annuity Pricing and Distributional Effects
Abstract. Gender-neutral annuity pricing establishes the same contractual premium or life-contingent payment for otherwise comparable buyers without using sex to differentiate the individual quotation. Because the present value of a lifetime income promise depends on survival, a common quote can redistribute expected benefit value across groups with different mortality. Its ultimate economic incidence also depends on participation, retirement capital, product guarantees, financing institutions, competitive behavior, and legally permitted rating characteristics. This monograph reconstructs the European Union’s Test-Achats legal transition, the distinct treatment of occupational and statutory benefits, and the United States employment-pension equality cases. It derives the actuarial accounting of unisex pooling, illustrates its sensitivities with transparent synthetic calculations, and critically interprets empirical work on UK and German annuity markets. The analysis integrates mortality selection, consumer access, aggregate pension inequality, insurer solvency, and household welfare. Its central finding is that equal treatment at the point of contract has a determinate legal meaning while its financial and social effects require explicit market and demographic counterfactuals. The author offers an original comparative synthesis and reproducible hypothetical calculations; the work asserts neither a novel actuarial theory nor new causal findings from market data.
Keywords: Gender-neutral annuities; unisex pricing; individual equality; Test-Achats; annuity contracts; life-contingent payments; mortality selection; adverse selection; expected transfers; pension gender gap; occupational pensions; actuarial valuation; longevity risk; retirement welfare; insurance regulation; distributional incidence.
Introduction
A lifetime annuity converts a stock of capital into a stream of payments contingent on survival, sometimes combined with guaranteed periods, survivor rights, refunds, or investment participation. The expected present value of those payments depends on the insured life, benefit schedule, discount rate, contractual options, and mortality assumptions. In a market allowing sex-specific quotations, an insurer may offer different annual payouts for an otherwise identical premium where its actuarial assessment of survival differs. Under a gender-neutral or unisex rule, the insurer must quote the same contractual amount to otherwise comparable individual applicants even when their expected survival probabilities differ.
This is simultaneously a problem of legal equality, actuarial balance, market composition, and distributional welfare. A uniform payment is observable and enforceable under a defined legal rule; the economic transfer associated with it is a modeled difference between the promised payment’s expected value and its price. The consequences for retirement living standards require additional information on accumulated pension wealth, health, household composition, participation and product choice. A legal measure of equal treatment, an actuarial present-value calculation and a consumer welfare comparison must each retain their own evidentiary basis.
The equilibrium implications are important. Finkelstein, Poterba, and Rothschild’s analysis of annuity market regulation models suppliers and consumers responding to unisex pricing and shows why fixed-menu comparisons can overstate eventual redistribution. The problem is therefore not exhausted by identifying the group with longer average life expectancy. It requires identifying the actual insured pool, the source of the premiums, the conditions of guaranteed payments, the adjustments available to issuers, and the legal authority that constrains pricing.
The methods applied here are documentary source comparison, actuarial derivation, synthetic counterfactual sensitivity and source-critical interpretation. Official directives, court judgments, pension and insurance supervision instruments, actuarial experience releases and peer-reviewed findings are treated according to their actual scope. Academic research accessible as a publisher abstract is cited for the finding reported there, without claiming to have inspected restricted full articles. All numerical examples are fictional, independently recomputable from their stated assumptions, and explicitly separated from market estimation.
Research Foundations and Institutional Genealogy
The contract is the unit of comparison
An immediate single-life annuity can promise level monthly payments until death, but retirement-income contracts vary substantially. Deferred annuities introduce the probability of surviving to commencement; period-certain provisions guarantee a minimum set of payments; return-of-premium provisions change the cash flow on early death; joint-life contracts depend on two survivals and the survivor continuation fraction. Investment-linked and escalating annuities further change the amount paid at successive dates. A valid comparison of unisex quotations must therefore hold these contractual dimensions constant along with premium, issue age and all permitted underwriting characteristics.
For a given contract and discount basis, the actuarial present value is an expectation over possible payment dates and survival states. If individuals face identical contract benefits but possess different conditional survival distributions, their expected value differs. This arithmetic feature neither establishes an individualized right to pay a group-specific price nor determines a social judgment about fairness. The legal entitlement follows the relevant law. The actuarial calculation describes possible financial redistribution within a particular insured portfolio.
The terminology requires care. European Union legislation addresses sex as a rating factor; academic articles often call the resulting rule gender-neutral. Historical mortality data typically use binary sex classifications recorded in administrative systems. These variables do not, by themselves, characterize every gender identity, medical circumstance, or individual risk. An applied analysis must state how classifications were collected, what is missing, and why the chosen data relate to the contractual and legal population. A comparative calculation based on female and male historical tables remains an illustration of those two data-defined groups; it must not be silently extended to all persons.
European legal genealogy
Council Directive 2004/113/EC applied the principle of equal treatment between men and women to access to and supply of goods and services, including insurance and related financial services. Article 5(1) required the use of sex as a factor in calculations of premiums and benefits not to result in differences in individuals’ premiums and benefits for new contracts after the specified implementation date. Article 5(2) originally permitted Member States to allow proportionate differences where sex was a determining factor in risk assessment grounded in relevant and accurate actuarial and statistical data.
The 2011 Test-Achats judgment declared Article 5(2) invalid with effect from December 21, 2012. The Court considered an indefinite possibility of differentiating individual insurance premiums and benefits inconsistent with the equal-treatment objective and the Charter framework. The ruling determined a legal result about the derogation, not a scientific result concerning the equality of male and female mortality. It is accordingly incorrect to infer that insurers must estimate identical survival rates for all applicants to comply with the rule.
The European Commission’s 2012 implementation guidelines explained the temporal application to new contracts and distinguished transactions creating a new agreement from changes occurring automatically under an existing contract. They also clarified that aggregate risk calculations, technical provisions and reinsurance may continue to use sex-related data where that use does not lead to prohibited differences in an individual's premium or benefits. This distinction recognizes that actuarial measurement and legally permissible customer differentiation serve separate functions.
The scope distinction for retirement provision is equally significant. Private voluntary individual annuities are not institutionally identical to employment-based occupational schemes or statutory pensions. The Commission’s guidance discusses employment-related arrangements in relation to the separate equal-treatment regime, while Directive 2006/54/EC supplies employment and occupational social security rules, including defined exceptions concerning actuarial factors. The classification of each product and national transposition must be checked before attributing the individual-insurance rule to a particular plan.
The Court also addressed sex-based life-expectancy calculations for a statutory work-accident lump sum in X v. Sosiaali- ja terveysministeriö. The 2014 decision applied the statutory social-security equality framework to the compensation formula at issue. It provides an additional illustration of how average group life expectancy can conflict with an individual equality rule, while retaining its distinct benefit type and legal authority.
United States employment pension cases
The United States Supreme Court decided the closely related employment-pension contribution issue in Los Angeles v. Manhart, 435 U.S. 702 (1978). A municipal retirement plan collected larger pension contributions from female employees to finance the same monthly retirement benefit, relying on their group’s greater average life expectancy. The Court found this practice inconsistent with Title VII's requirement to treat individual employees equally. The decision distinguished general statistical longevity from an employer's statutory obligation to individual workers.
The Court addressed the benefit side in Arizona Governing Committee v. Norris, 463 U.S. 1073 (1983), concerning deferred compensation and annuity offerings arranged through an employer. Women were offered lower monthly payments for the same contributions. Title VII obligations extended to that employer-sponsored arrangement, and remedial questions about amounts accumulated before and after the judgment were expressly considered. A benefit supplied by an outside insurer did not erase the employer’s responsibilities.
Manhart and Norris supply strong examples of the legal distinction between individual equality and actuarial group averages. They establish propositions within U.S. employment law, with specific institutional and temporal conditions. They do not establish a universal federal prohibition on sex-related quotations in every privately purchased U.S. retail annuity. Such an analysis requires separate attention to the applicable state insurance framework, the contract's nature and the legal period.
Mortality evidence and its provenance
An annuitant mortality table is an estimate derived from exposure and death experience in a defined group. It may differ materially from a general population table because annuitants are selected by health, income, employment history, retirement resources and purchasing decisions. A table designed for reserve valuation also may contain margins or projection conventions that make it inappropriate as a pure expected-value mortality estimate.
The Society of Actuaries’ individual annuity reserving materials document one institutional source of annuitant-specific bases and their valuation purposes. The Society’s 2005–2008 payout-annuity experience report identifies historical insured annuitant experience as opposed to contemporary national population survival. The vintage, data coverage, selection and table purpose must be disclosed when such records enter a calculation.
Finkelstein and Poterba’s UK annuity market evidence found mortality relationships with certain annuity contract characteristics while reporting no substantive mortality association with annuity size in their sample. That finding is more specific than the blanket claim that high-value annuities invariably come from longer-lived policyholders. Contract attributes, issue timing, purchase selection and particular market conditions all matter.
Hierarchy of evidence
The most appropriate legal evidence for an EU obligation is the directive and relevant Court judgment in its original form. Implementing guidance documents the administrative interpretation and particular contract-date questions. The controlling rule for an employment plan or statutory benefit comes from its own legislation and jurisprudence. Insurance supervisors and accounting standard setters address prudential obligations and measurement, and do not themselves establish the equality entitlement. Empirical papers answer questions about observed or modeled behavior under specific data and assumptions. OECD pension indicators provide population comparisons; they are not automatically estimates of policy effects. Maintaining this hierarchy protects the analysis from importing conclusions across institutions and legal systems.
Unisex Rate and Legal Duty
New contracts and the effective date
The Test-Achats result makes December 21, 2012 a critical date for new contracts within Article 5’s scope. A comparison of insurer tariffs before and after that date must first distinguish new policies from automatic renewals, unchanged option exercises and contract amendments. The Commission’s new-contract interpretation addresses arrangements where additional consent effectively creates a new transaction. An empirical observation dated after the transition cannot automatically be treated as a newly underwritten contract subject to the same treatment.
A legal tariff comparison should hold the insured payment schedule, issue age, premium, guarantee, expense basis, relevant permitted health information and contract conditions constant. It then tests whether changing only the protected sex classification changes an individual premium or benefit. This matched-quote audit establishes an observable feature of compliance. It does not calculate the welfare effects of compliance, the profitability of the insurer, or the adequacy of the resulting pension.
Internal risk modeling and individual quotes
An insurer with a unisex outward schedule may retain sex-disaggregated experience data for estimating aggregate liabilities, modeling risks and financing the product. The Commission’s aggregate-pricing distinction is important: internal mortality discrimination in the statistical sense of distinguishing risk groups cannot be allowed to translate into prohibited price or benefit discrimination at the individual contract level. The insurer therefore needs a defensible separation between back-office assumptions and quote production.
A risk pool containing a greater share of long-lived annuitants requires more assets per unit of promised annual payment. A legally uniform tariff is financially sound only when its structure accounts for actual pool composition, expenses, permitted capital charges, relevant options and future expected mortality. If a business reuses historical assumptions after the composition of buyers has changed, the absence of a sex-specific quotation will not prevent aggregate mispricing. Prudential review and equality-compliance review are thus complementary controls.
The legal audit should also test product access. Differential steering, unadvertised product restrictions or selective availability may alter the effective offers received, even where the algorithm calculating a selected standard annuity is uniform. An adequate record includes the quote, product options actually displayed, referral and underwriting rules, manual overrides, contract acceptance and the sales channel. The researcher must distinguish evidence of discriminatory access from an observed association that may have another explanation.
Occupation, employment, and public benefits
A pension scheme should first be classified by whether the benefit is an insurer's independently contracted product, an occupational social-security entitlement associated with employment, or a statutory benefit. These distinctions govern the relevant protected persons, exceptions, remedies and implementation rules. Directive 2006/54/EC expressly considers actuarial factors in occupational social security, subject to the relevant provision, and cannot be reduced to a simple extension of the consumer-insurance tariff rule. A plan using an insurance company as an administrator or risk bearer does not thereby make every plan term a freestanding retail insurance transaction.
The United States offers a further institutional comparator. Under Title VII, the employer's connection to an annuity offering was decisive in Norris, whereas the EU consumer-insurance rule focuses on individual insurance contractual premiums and benefits. In statutory schemes, policy goals such as redistribution, contribution history and coverage adequacy shape benefits under public law. The same descriptive phrase, gender-neutral pension, can therefore refer to legally different instruments and redistributive mechanisms.
Controls for legal and actuarial audit
An auditable control register should contain the legal instrument, jurisdiction, effective date, exact contract form, issue and amendment chronology, policyholder and annuitant roles, benefit schedule, underwriting variables, internal assumptions, and the observed quote. It should preserve historical products without rewriting their records as if they had been sold under contemporary law. The audit should describe the risk classification permitted for internal valuation and prohibit its accidental propagation to the outward rate calculation where the applicable law excludes it.
The findings should be expressed at the correct level: a particular quote may comply with a defined uniformity test; a given product portfolio may be adequately financed under a stated survival projection; a jurisdiction may show measured differences in participation or benefit distribution. Each proposition demands its own data and inference. Treating one as proof of all three is the central analytical error in many simplified discussions of unisex annuity pricing.
Expected Transfer by Survival Group
Mathematical representation of a lifetime promise
Let a contract pay one unit continuously per year while a purchaser is alive. For a buyer from population group g, with survival probability S_g(t) from contract commencement through time t and time-specific discount factor D(t), its actuarial present value per unit annual income is
A_g = ∫₀^∞ D(t) S_g(t) dt.
For a discrete monthly or annual annuity, the integral is replaced by a sum across actual payment dates, incorporating payment-in-advance or payment-in-arrears conventions and survival to each contractually relevant date. In a deferred annuity, the survival event during deferral is included. If there is a certain-payment term or beneficiary refund, the death-contingent and guaranteed cash flows must be valued separately.
Let P be a common single premium, with expenses and loadings temporarily set aside. A fully group-specific, actuarially balanced annual payment is P/A_g. If the insurer must instead quote one benefit B to all otherwise comparable purchasers, and the portfolio shares w_g represent shares of equal premiums within this cell, the common annuity factor is A_U = Σ_g w_g A_g. The balanced benefit is B = P/A_U. A group's expected value at the common rate is B A_g, and its expected present-value position relative to premium is T_g = B A_g − P. The condition Σ_g w_g T_g = 0 follows directly from the definition of A_U.
The identity describes an expected transfer within a fixed cohort under precisely defined assumptions. A positive T_g does not mean the insurer has already transferred that amount to the policyholder in cash. Realized payments depend on survival; the expected value is a risk-weighted valuation of alternative future outcomes. The negative value for another group likewise represents an expected discounted cash-flow relationship, not an allegation that its members will receive less than they paid in every eventuality.
For differing premium amounts, the pool factor must weight liabilities according to amounts payable and premiums received. When contract options, issue ages or underwriting categories differ, a single average of survival probabilities is not enough. An issuer may maintain distinct legally permitted risk cells, with a unisex quotation within each relevant comparison. The relevant accounting is the present value of all obligations, including guarantees, expenses, capital charges and the actual distribution of promised benefits.
A reproducible two-group synthetic illustration
Consider two fictional groups M and F. The labels correspond to the conventional classifications of historical actuarial tables; the rates below are hypothetical mathematical inputs and are not estimates of male or female mortality at a particular retirement age. Each person pays P = 1,000 currency units. There are no expenses, commissions, taxes, mortality-improvement adjustments, profit requirements, premium refunds, survivor benefits or payment guarantees. Benefits are paid continuously from issue until death. Assume a constant force of interest δ = 0.03 a year and constant mortality hazards h_M = 0.06 and h_F = 0.04. Then S_g(t) = exp(−h_g t), D(t) = exp(−δt), and the factor reduces exactly to A_g = 1/(δ + h_g).
The factors are A_M = 11.111111 and A_F = 14.285714. Their arithmetic mean is 12.698413 if half the new purchasers come from each group. The common actuarially balanced annual income is 1,000/12.698413 = 78.75 units. If separate group pricing were permitted under exactly these conditions, the corresponding incomes would be 90.00 and 70.00. At the common income of 78.75, the expected discounted payouts equal 875 for M and 1,125 for F, implying T_M = −125 and T_F = +125. The weighted expected benefit equals the original 1,000-unit premium.
| Synthetic measure | Group M | Group F | Equal-weight portfolio |
|---|---|---|---|
| Annual force of mortality | 0.060000 | 0.040000 | Mixed |
| Continuous annuity factor | 11.111111 | 14.285714 | 12.698413 |
| Separate-pricing annual income | 90.00 | 70.00 | Not applicable |
| Common-price annual income | 78.75 | 78.75 | 78.75 |
| Expected discounted payout at common price | 875.00 | 1,125.00 | 1,000.00 |
| Expected discounted payout less premium | −125.00 | +125.00 | 0.00 |
This is a demonstration of an identity, not an empirical finding about actual insurance buyers. Constant mortality hazards are chosen to make the integrals independently checkable. Real human old-age mortality varies strongly with age, cohort, period, socioeconomic circumstances and selection into the insured portfolio. The illustration deliberately omits features needed for a commercial quote.
The direction of the hypothetical transfer follows from the difference between each annuity factor and the pool factor. If one group has higher survival over the payment horizon under the chosen discount convention, it receives a higher actuarial present value under a common payment. The magnitude depends on the distance between survival curves and on the share of premiums and benefit commitments represented by each group. There is no general basis for attaching this synthetic transfer to the actual pension gender gap or regarding it as an observed welfare change.
Endogenous participation changes the adequate quote
The 50:50 split is an assumption about new purchasers, not a fixed characteristic of the surrounding population. Suppose the share of new F purchasers increases to 65% and M falls to 35%, with premiums and group survival assumptions unchanged. The new common factor is A_U = 0.35(11.111111) + 0.65(14.285714) = 13.174603. The corresponding level annual benefit is about 75.903614 units per 1,000 premium. A larger proportion of longer-lived purchasers creates more expected claims for every unit of common annual income.
If the issuer continued to offer 78.75 after the portfolio changed, expected discounted benefits would be 78.75 × 13.174603 = 1,037.50 per policy. The 37.50 excess over premium is a synthetic aggregate shortfall, before operating costs or capital charges. A properly updated tariff reduces the new common payment, changes other permitted terms or reflects a loss to the issuer. This example makes composition risk visible without estimating any actual insurer's behavior.
Participation can respond to the quote itself. People expecting shorter lives may find a less favorable pooled annuity more expensive relative to alternatives, while relatively longer-lived buyers may find it attractive. Income, health, retirement timing, perceived life expectancy and capacity to choose between drawdown and lifetime income all mediate the response. In a compulsory annuitization system, participation effects differ because the decision to purchase is less discretionary; in voluntary markets, the purchaser population may be strongly selected.
The German evidence of von Gaudecker and Weber’s quasi-experimental unisex pricing study reports unisex annuity payout rates often close to the preexisting female rate in the examined subsidized market. The interpretation emphasizes selection incentives and insurer responses. Because that study concerns a particular institutional arrangement and earlier reform, its results cannot be treated as a universal coefficient describing the whole European insurance market after Test-Achats. It establishes the empirical relevance of purchaser selection and anticipated portfolio composition.
Guarantees change the survival exposure
A level life annuity with payments guaranteed for ten years has different mortality sensitivity from an annuity stopping immediately at death. For the same continuous-payment hypothetical setting, the first ten years are paid independently of survival, and payments afterward depend on survival. Its factor is
A_g(10) = [1 − exp(−0.03 × 10)]/0.03 + exp[−(0.03 + h_g) × 10]/(0.03 + h_g).
The factors are approximately 13.156833 for M and 15.733468 for F. The difference between the factors, about 2.576635, is smaller than the no-guarantee difference of about 3.174603. The first decade’s unconditional payment entitlement makes that part of the promise insensitive to mortality. The additional guarantee also raises the insurer's expected payout factor for both groups, lowering the common annual income affordable for a fixed premium. Thus a narrower survival-factor difference does not by itself imply a cheaper or more suitable product.
A refund-of-premium annuity has a different payment path after death; a joint-life annuity requires two conditional survival processes and their dependence; an escalating annuity requires projected increases in the benefit amount; a variable annuity requires investment and payout-unit dynamics. A calculation for the pure single-life annuity should not be reused for these contracts without reconstructing the actual cash flows. The importance of differing contract attributes is reinforced by Finkelstein and Poterba’s selection analysis of UK annuities, whose results linked mortality to features other than simply the size of an annuity in their sample.
From static redistribution to market equilibrium
A static transfer estimate fixes the portfolio and contract menu, revalues obligations under a common price and compares that result with the permitted sex-specific counterfactual. An equilibrium assessment allows buyers and suppliers to adjust. The supplier may offer greater guarantees, different commencement dates or altered expense charges; buyers may shift retirement resources among annuity and drawdown products, or postpone purchase. These responses can affect the size, and occasionally the distributional ordering, of changes in expected value.
Finkelstein, Poterba, and Rothschild (2009) analyze such endogenous responses in a calibrated UK annuity setting. Their contract-menu equilibrium result indicates that endogenous contracts can offset up to half the redistribution generated by a benchmark model with an exogenous contract set. This magnitude applies to the assumptions and calibration of that study and must not be transplanted as a general forecast. The deeper methodological point is that pricing law can modify both prices and the choices through which prices matter.
A complete empirical estimate should distinguish the immediate quote effect, the expected-value transfer for existing purchase choices, the changes in who buys and how much they annuitize, and the welfare effects of all those changes. All four objects can be legitimate research results. Confusing them makes otherwise sound evidence appear contradictory.
Distributional Access and Fairness
Contractual equality and pension capital
Where the governing law requires unisex insurance quotations, two otherwise comparable applicants receive the same annual payout for the same premium. This creates a clear contractual entitlement. But the annual income that each buyer can purchase also depends on the premium amount supplied from retirement savings. Equal conversion factors do not equalize pension wealth, years employed, past wages, caregiving interruption, investment returns or household resources.
The OECD’s 2025 gender pension-gap evidence reports an average gender pension gap of approximately 23% across OECD countries in 2024, compared with about 28% in 2007. The same chapter reports an average expected duration after leaving the labor market of 22.8 years for women and 18.6 for men in 2024. These are cross-national statistics with defined sample and measurement conventions. They support the proposition that accumulation and retirement horizons differ across populations; they do not identify an annuity-policy effect by themselves.
A woman receiving a greater expected present value per unit premium under a pooled tariff may still have a smaller total annual pension than a man who accumulated much more capital. Conversely, a shorter-lived individual may receive less expected lifetime annuity value while having greater wealth or access to other benefits. An equitable interpretation cannot infer either individual's welfare solely from their group’s factor or pension amount. Retirement spending, nonpension income and family support can alter the relevant comparison.
Within-group and socioeconomic redistribution
Variation within historically sex-classified groups may be substantial. Life expectancy differs with earnings, education, occupation, disability, access to health care and inherited social conditions. If longer-lived high-income buyers purchase more annuity income, a pooled quote may channel actuarial expected value toward people who already command substantial pension assets. A distributional assessment must therefore ask who enters the insured pool, what their premiums represent, and how actual longevity correlates with economic circumstances.
A useful empirical structure is a cross-tabulation of projected present value by issue age, source-classified sex, health-risk category, income or wealth group, annuitized amount and benefit design, together with uncertainty intervals and data provenance. It should separately report participation rates to avoid treating the market's selected buyers as a representative population. Product access matters particularly for people with limited savings: a pricing advantage per unit capital has limited practical effect when little capital is available to annuitize.
The OECD's comparative pension indicators also caution that pension differences among recipients should not be confused with household poverty, because resources may be shared and some individuals have no measured pension income. Thus a comprehensive study considers the distribution of ownership, pension access, retirement income and household consumption, rather than using one gap as a proxy for all of them.
Annuity welfare and the value of alternative protections
Annuities insure against surviving longer than one's resources. Their consumption-smoothing value depends on other pensions, insurance, family arrangements, bequest motives, financial liquidity and the risks that remain uninsurable. Davidoff, Brown, and Diamond’s analysis of annuities and welfare shows that the desirability of annuitization depends on market completeness and the household's economic environment. A favorable actuarial money's-worth ratio is one relevant feature, not a complete welfare ranking.
A retiree with substantial uninsured care costs might rationally value liquid savings alongside lifetime income. A household supporting a surviving spouse may value joint-life protection even when the first-life annual payment is smaller. A person expecting a shorter remaining life may choose a death guarantee or refund option. These decisions interact with unisex pricing, and the resulting distributional effects cannot be captured by a single bare life-annuity factor.
The OECD Pensions Outlook 2026 examines retirement saving adequacy, investment and pension design in contemporary systems. It provides institutional context for differences in who is able to annuitize and the importance of meaningful retirement choices. The report is not treated here as an estimate of the causal impact of Test-Achats or a worldwide measure of unisex annuity market equilibrium.
Financing incidence and consumer protection
A modeled expected transfer between two classes of purchaser is not necessarily a payment financed exclusively by those classes. Actual obligations may be financed through premiums, employer contributions, occupational funds, insurer capital, shareholder returns, public transfers, participating surpluses or reinsurance. To determine who bears the incidence, each study should specify the risk-bearing institution and flow of capital at every stage. A group-level actuarial subsidy calculated at policy issuance does not necessarily identify a taxpayer subsidy or a cross-subsidy between all people of different sexes.
The insurer's internal solvency assessment has a distinct role. Under the Solvency II technical-provisions rule, best estimates reflect probability-weighted future cash flows and are supplemented by a risk margin. EIOPA’s assumptions requirements concern justified and appropriate assumptions in calculating technical provisions. These standards support maintaining an accurate projection of heterogeneous mortality even where the outward individual quotation is uniform.
IFRS 17 insurance accounting separately measures groups of insurance contracts using fulfillment cash flows, a risk adjustment and contractual service margin as applicable. Neither the financial-statement liability nor the insurer's capital requirement is equivalent to the consumer's price or actuarial transfer. Keeping the accounting, prudential, conduct and welfare concepts distinct allows them to be studied together accurately.
Disclosure and advice complete the analysis. The buyer should understand when payments start, whether they are guaranteed, what happens at death, how survivor payments change and what liquidity is surrendered. Comparisons should show benefits using genuinely comparable product terms, rather than conflating a higher initial payout with greater lifetime security. The sex-neutral rating rule can be fully respected while consumers still need protections concerning access, suitability and the understandable presentation of contractual risks.
Robustness, Independent Validation, and Source-Critical Analysis
Identify the estimand before measuring it
A strong study begins by stating exactly which question its data can answer. In a legal-compliance study, the estimand is the difference between individual quotations for two otherwise identical eligible applicants when a protected characteristic changes. In an actuarial analysis, it is the expected present value of the contractual cash flows under defined survival and discount assumptions. In a market-incidence study, it is the difference in prices, purchase rates and benefits attributable to the policy relative to a plausible counterfactual. In household welfare analysis, it is the change in lifetime utility or retirement security after accounting for resources and the person's preferences.
These quantities must not be merged into one undifferentiated “effect of unisex pricing.” A matched-quotation study can establish an equal price but not estimate aggregate welfare. An observed change in average payout can reflect the reform, changing interest rates, cohort mortality, product mix, charges or changing purchaser composition. A transfer calculation holding the number of contracts constant measures only the direct cash-flow component. Researchers need separate documentation for each causal link in a proposed explanation.
Research design around a legal reform
A before-and-after comparison around December 21, 2012 should record the new-contract definition, policy issue dates and amendments; coverage under the relevant legal instrument; interest-rate term structure; actuarial mortality basis; insurer capital constraints; expenses; commission arrangements; and payment guarantees. A simple mean quote before and after a reform is weak evidence if all those conditions were also moving. An event-study or difference-in-differences design can improve identification only where the comparison group, timing and parallel-trends assumptions are defensible.
The German von Gaudecker–Weber annuity evidence covers an earlier subsidized private-pension setting. Its institutional features are useful for understanding insurer expectations and applicant selection, but it cannot be treated as a controlled comparison for the later EU-wide Article 5 transition. Finkelstein, Poterba, and Rothschild’s calibrated annuity equilibrium analysis supplies a structural model, not an observed causal effect for every country after 2012. The UK contract-characteristic study informs the selection mechanism without identifying the reform’s population-wide effect. Their methods and temporal scopes should remain visible when their conclusions are integrated.
For a robust market study, researchers could preserve matched insurer quotations for several product designs and rating cells; reconstruct the market share of those cells before and after the policy; measure the extent of annuitized capital, not just the number of purchasers; and estimate contract-specific present values on a consistent discount basis. When an untreated comparison market is unavailable, research should present partial-identification bounds and descriptive patterns rather than a falsely precise causal claim. Results should be accompanied by assumption and sample-selection diagnostics.
Mortality basis, population selection, and time
A population mortality table captures the persons observed in a national statistical system, while an insured annuitant mortality table captures an institutionally selected subset. A period table represents mortality rates by age and calendar period; a cohort projection attempts to account for future mortality improvement. Both the baseline and future trajectory of survival matter to an annuity sold today. A statistical association between recorded sex and survival may change with age, health, income, migration and the selection process. The sample’s occupational, socioeconomic and regional composition therefore requires explicit reporting.
The SOA’s individual-annuity valuation documentation and historical payout-annuity study illustrate different uses of mortality data. Reserve tables may embed prudent margins and are not necessarily unbiased forecasts of cohort survival. Experience studies have specified periods and records; their observations should not be transported to another country or modern contract cohort without an explicit adjustment and validation procedure.
A reliable model audit tests alternative survival bases, mortality-improvement scenarios, interest-rate curves, policyholder selection weights and contract guarantees. It also reports sensitivity to unusually old ages, where small exposure counts can yield uncertain mortality estimates. Results should be examined with and without socioeconomic stratification, and missing or nonbinary administrative categories should be documented rather than forced silently into historic binary classifications. A high-quality model explains which population its estimates represent, and which populations remain outside the observed evidence.
Independent arithmetic validation
The synthetic illustration is intentionally elementary enough to verify by two independent methods. First, integrate exp[−(δ+h)t] from zero to infinity to obtain 1/(δ+h). Second, substitute the reported parameters into the identity A_g(δ+h_g)=1. Both yield 11.111111 for M and 14.285714 for F. Their 50:50 average is 12.698412698, so 1,000 divided by the mean is exactly 78.75. The identities B A_M = 875 and B A_F = 1,125 confirm that the average expected payout equals the premium.
Changing the weights to 35% and 65% produces A_U = 13.174603175 and B = 75.903614458. Under the old payout of 78.75, the new pool expectation is 1,037.50, a synthetic 37.50 gap. A reconciliation using individual expected values weighted by 0.35 and 0.65 leads to the same total. Therefore the reported effect depends on portfolio weights, with no need to assume a change in the mortality of either fictional group.
The ten-year guaranteed-payment factor can be independently reconstructed as the sum of a ten-year certain annuity, [1−exp(−0.3)]/0.03, and a post-ten-year life-contingent tail, exp[−(0.03+h)10]/(0.03+h). It yields A_M(10)=13.156833308 and A_F(10)=15.733468412. The death-contingent difference is smaller than in the pure life annuity. The calculation does not model investment-linked annuity units, premium refunds or second-life continuation, each of which requires its own contingent-payment structure.
A reader can reproduce all numerical results with a calculator using only the formulas and assumptions printed above. No proprietary insurer data, purchased mortality table or stochastic simulation is needed for the illustration. Conversely, no conclusion about a real insurer's exact tariff or national population transfer can be obtained from these synthetic inputs.
Legal classification sensitivity
Legal conclusions are also sensitive to classification. A consumer retail annuity, occupational pension conversion and statutory accident benefit may involve the same elementary life-contingent mathematics but different equality duties. The EU Article 5 framework, the employment and occupational framework, and the U.S. employer-annuity precedent show why the exact contractual legal relationship matters.
Effective dates likewise determine which rule applies. The Test-Achats judgment removed a derogation from December 21, 2012; Norris contains important remedial timing distinctions for past and future contributions; later national transposition may define details beyond a pan-European summary. The evidence here supports the identified framework and salient court holdings. It does not purport to supply a legal opinion on any unexamined national contract.
Financial reporting and prudential checks
The internal liability estimate must follow the actual insured portfolio and the contract’s promised payments even when the outward quotation is unisex. Under EIOPA’s Solvency II valuation provisions, technical provisions include the best estimate and a risk margin. The associated assumption governance requirements reinforce data and methodology discipline. The insurer should reconcile product cash-flow projections with demographic experience, lapse or surrender behavior where relevant, capital support and valuation uncertainty.
The IFRS 17 measurement framework introduces its own accounting groupings, risk adjustment and contractual service margin. Accounting liabilities and prudential technical provisions have different objectives and should be separately reconciled to product cash-flow economics. A uniform market quote does not imply homogeneous reserve risk. Equally, a prudent mortality assumption for reserving does not automatically justify a different individual retail quote in a jurisdiction applying the unisex rule.
Evidence access and falsification
The legal findings are tied to original official texts and judgments. The OECD statements are aggregate statistics. The SOA documents indicate the existence, period and designated purpose of actuarial data, rather than supplying a newly estimated population survival curve. Research-paper conclusions are attributed to their accessible publisher abstracts or identified university working-paper records where full-text scrutiny could not be documented. The synthetic calculations are independently checkable and deliberately isolated from these empirical claims.
Several propositions are testable and potentially falsifiable: matched retail quotations should not differ solely by protected sex classification where the EU Article 5 duty applies; a portfolio with a larger concentration of long-lived purchasers requires more premium per unit lifetime payment when other variables remain fixed; observable contract design and participation changes can cause distributional incidence to diverge from static pricing comparisons. Actual observations may support or reject these propositions in a defined market. This monograph does not claim that such tests have been newly executed with confidential or market-wide microdata.
Integrated Analytical Synthesis and Scientific Agenda
Joint interpretation of law, cash flows, and welfare
The evidence supports four connected results. First, an equality rule establishes the permitted relation between protected characteristics and an individual contractual quote. Second, the present-value calculation captures the expected cost of a lifetime promise given a survival distribution. Third, insurers and buyers respond to market rules, changing the population and contracts to which the price applies. Fourth, the retirement consequences depend on accumulated resources and the distribution of longevity, liquidity needs and household circumstances.
The first proposition is established by legal sources within their own institutional scope. The second is a mathematical identity once cash flows and probabilities are specified. The third is a finding from established market-equilibrium research and specific historical empirical investigations. The fourth requires household-level measures beyond a quote. Together they support an original analytical synthesis: fairness in annuity pricing needs simultaneous evaluation of the legal entitlement, the actuarial pool, the economic response and the social distribution of realized retirement resources. This synthesis organizes established scholarship rather than claiming an independently invented doctrine.
A common quote can create legal equality while a heterogeneous risk pool generates expected-value transfers. The transfers can be modified by consumer choice, guarantee design and insurer pricing. Consequently, the existence of an equal quote neither proves nor disproves that every individual has received an equal actuarial expected value. Nor does inequality in expected values establish that the regulatory rule has failed in its legal objective. The answer to each question follows from its clearly defined standard of evaluation.
A replicable empirical agenda
A comprehensive evaluation should begin with source provenance. For every jurisdiction, archive legislation, case law, operative guidance and national implementation applicable to the contract and issuance date. Identify whether the arrangement is consumer insurance, occupational social security, or statutory benefit. Document contractual participants and who finances the benefit.
Next, obtain contemporaneous rates for identical issue ages, premiums and terms, preserving both quoted and purchased products. Record which customer characteristics were used in the quote, what advice was presented, whether the insurer offered particular guarantee options and whether an applicant was denied access. Compute contract-level discounted expected payouts with multiple clearly identified annuitant survival bases, reporting mortality projection and discounting assumptions.
Then reconstruct the distribution of annuitized wealth and purchasing participation across demographic and economic strata. A market-incidence analysis should permit enrollment and product composition to adjust. A welfare comparison should integrate household assets, other pensions, needs for liquidity, bequest preferences and risks of outliving resources. Separate accounting for insurer capital, reinsurance, occupational financing and public support is necessary to identify who ultimately bears each cost.
Pre-specified robustness tests should include alternative demographic bases, shifts in the proportion of longer-lived purchasers, benefit guarantees, varying interest rates, changing portfolio charges, health risk and missing records. Studies should distinguish descriptive differences from identified causal effects; uncertainty intervals must reflect sampling, parameter estimation and model selection. Publishing methods alongside underlying permitted data and reproducible calculations is essential to ongoing scientific evaluation.
Open empirical questions
The evidence raises several tractable questions. How rapidly does annuity participation respond to an altered unisex payment schedule in voluntary markets? Does a greater use of guarantees narrow sex-classified present-value differences while raising charges for the pool? Which wealth and health groups actually receive larger expected values after controlling for premium size and survival? How do occupational scheme rules, retail insurance rules and statutory benefit rules distribute obligations among distinct institutions? Does sex-classified annuitant experience remain a sufficient actuarial descriptor after socioeconomic, health and contract-choice variables are observed? Which consumer explanations allow purchasers to understand the distinction between an equal annual quote and unequal expected payment durations?
These questions require specific samples and causal or descriptive methods chosen in advance. Their resolution could change a jurisdiction-specific welfare assessment without modifying the basic law or actuarial identities established here. No new proprietary model or policy doctrine is introduced by listing the research problems.
Conclusions
Gender-neutral annuity pricing gives otherwise comparable applicants the same individual contractual quotation where the governing law imposes that duty. The CJEU’s Test-Achats judgment rendered the former Article 5(2) exception invalid from December 21, 2012, while Commission guidance allowed internally appropriate use of demographic data for reserving and aggregate risk assessment. Occupational and statutory pension arrangements must be considered under their own legal instruments. The U.S. Manhart and Norris decisions demonstrate related employment-equality principles in specifically employer-connected retirement arrangements.
Actuarial redistribution follows transparently from fixed-cash-flow pooling: at a common premium and annual benefit, groups with different survival expectations receive different discounted expected benefits. The hypothetical illustration produces positions of −125 and +125 per 1,000 premium under its initial assumptions. When the proportion of longer-lived purchasers rises, the pool's balanced annual payout falls, even though its common-pricing rule has not changed. Guarantees reduce the mortality-contingent share of some benefits while altering their cost and the consumers' reasons to buy them.
Evidence from UK equilibrium modeling and German annuity pricing indicates that endogenous contract choice, participation and supplier behavior are central to assessing the ultimate effect of a reform. OECD pension indicators show why unequal accumulated retirement wealth and duration of retirement matter for distributional outcomes. Reliable evaluation must measure the rule’s actual contract coverage, the present-value transfer, market and product adjustments, financial incidence, and household welfare on separate evidentiary bases.
The resulting scientific conclusion is that equal contractual pricing and heterogeneous expected economic value are simultaneously possible and institutionally coherent. Their joint significance is determined through contractual detail, lawful pricing, demographic selection, financing and retirement resources. This monograph establishes the relevant comparative evidence, reproducible actuarial relationships, source-critical boundaries and an empirical validation agenda. It makes no claim to an original actuarial theory or to policy effects not verified by the cited evidence.
References
Council of the European Union. 2004. “Council Directive 2004/113/EC of 13 December 2004 Implementing the Principle of Equal Treatment between Men and Women in the Access to and Supply of Goods and Services.” Official Journal of the European Union L 373: 37–43. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32004L0113
Court of Justice of the European Union. 2011. “Association belge des Consommateurs Test-Achats ASBL and Others v Conseil des ministres.” Case C-236/09, judgment of March 1, 2011, ECLI:EU:C:2011:100. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=celex%3A62009CJ0236
European Commission. 2012. “Guidelines on the Application of Council Directive 2004/113/EC to Insurance, in the Light of the Judgment of the Court of Justice of the European Union in Case C-236/09 (Test-Achats).” Official Journal of the European Union C 11, January 13: 1–11. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=oj%3AJOC_2012_011_R_0001_01
European Parliament and Council of the European Union. 2006, consolidated June 19, 2026. “Directive 2006/54/EC on the Implementation of the Principle of Equal Opportunities and Equal Treatment of Men and Women in Matters of Employment and Occupation (Recast).” https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A02006L0054-20260619
Court of Justice of the European Union. 2014. “X v Sosiaali- ja terveysministeriö.” Case C-318/13, judgment of September 3, 2014, ECLI:EU:C:2014:2133. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A62013CJ0318
Supreme Court of the United States. 1978. “Los Angeles Department of Water and Power v. Manhart.” 435 U.S. 702, April 25, 1978. https://www.law.cornell.edu/supremecourt/text/435/702
Supreme Court of the United States. 1983. “Arizona Governing Committee for Tax Deferred Annuity and Deferred Compensation Plans v. Norris.” 463 U.S. 1073, July 6, 1983. https://www.law.cornell.edu/supremecourt/text/463/1073
Society of Actuaries. 2011. “2012 Individual Annuity Reserving Report & Table.” Payout Annuity Table Team. https://www.soa.org/resources/experience-studies/2011/2012-ind-annuity-reserving-rpt/
Society of Actuaries. 2014. “2005–08 Individual Payout Annuity Experience Report and Excel Files.” https://www.soa.org/resources/experience-studies/2014/research-soa-individ-payout-annuity-exp-report/
Finkelstein, Amy, James Poterba, and Casey Rothschild. 2009. “Redistribution by Insurance Market Regulation: Analyzing a Ban on Gender-Based Retirement Annuities.” Journal of Financial Economics 91 (1): 38–58. https://www.sciencedirect.com/science/article/pii/S0304-405X%2808%2900153-0
von Gaudecker, Hans-Martin, and Carsten Weber. 2006. “Mandatory Unisex Policies and Annuity Pricing: Quasi-Experimental Evidence from Germany.” Working Paper 07-07, Sonderforschungsbereich 504, University of Mannheim. https://madoc.bib.uni-mannheim.de/2560/
Finkelstein, Amy, and James Poterba. 2004. “Adverse Selection in Insurance Markets: Policyholder Evidence from the U.K. Annuity Market.” Journal of Political Economy 112 (1): 183–208. https://doi.org/10.1086/379936
Davidoff, Thomas, Jeffrey R. Brown, and Peter A. Diamond. 2005. “Annuities and Individual Welfare.” American Economic Review 95 (5): 1573–1590. https://www.aeaweb.org/articles?id=10.1257%2F000282805775014281
Organisation for Economic Co-operation and Development. 2025. “Gender Pension Gap.” In Pensions at a Glance 2025: OECD and G20 Indicators. Paris: OECD Publishing. https://www.oecd.org/en/publications/pensions-at-a-glance-2025_e40274c1-en/full-report/gender-pension-gap_90ed13b5.html
Organisation for Economic Co-operation and Development. 2026. OECD Pensions Outlook 2026: Are We Saving Enough for Retirement? Paris: OECD Publishing. https://www.oecd.org/en/publications/oecd-pensions-outlook-2026_a51aa775-en.html
European Insurance and Occupational Pensions Authority. n.d. “Calculation of Technical Provisions.” Solvency II Single Rulebook, Directive 2009/138/EC, Article 77. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-2160_en
European Insurance and Occupational Pensions Authority. n.d. “General Provisions—Assumptions Underlying the Calculation of Technical Provisions.” Solvency II Single Rulebook, Commission Delegated Regulation (EU) 2015/35, Article 22. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-2435_en
International Accounting Standards Board. 2017, as amended. IFRS 17 Insurance Contracts. IFRS Foundation; effective for annual periods beginning on or after January 1, 2023. https://www.ifrs.org/issued-standards/list-of-standards/ifrs-17-insurance-contracts/