Insurance by BogdanovaThe Encyclopedia of Insurance and Risk Science

RESEARCH DOSSIER

Pension Benefit and Funding Models

Abstract. Pension valuation connects an enforceable or substantively established promise of retirement income to a schedule of contingent future payments, an attribution of those payments to service, and a financing mechanism capable of delivering them. This monograph derives the principal benefit and funding relationships for occupational defined benefit, defined contribution, and collective retirement arrangements. It separates accrued benefit measurement, projected benefit measurement, actuarial cost allocation, financial-reporting liabilities, regulatory technical provisions, and the contributions required to finance a plan. Demographic, salary, inflation, and investment assumptions are examined as a coherent valuation system rather than interchangeable numerical inputs. A reproducible final-salary example demonstrates how projected-unit-credit obligations respond to salary growth and discount rates; further examples distinguish contribution accumulation from benefit guarantees and illustrate funding ratio sensitivity. International comparison examines IFRS accounting, U.S. actuarial standards and federal funding law, the United Kingdom's funding regime, European occupational pension regulation, and comparative OECD evidence. The central analytical finding is that the same pension cash-flow promise supports several legitimate measurements because each measurement answers a different institutional question. The reliability of pension finance therefore depends on exact specification of the promised benefit, attribution rule, valuation purpose, asset recognition, funding policy, and allocation of residual risk. The work offers original source-based analysis and numerical synthesis without proposing a new proprietary pension theory or reporting unobserved empirical estimates.

Keywords: pension actuarial valuation; defined benefit; defined contribution; projected unit credit; accrued benefit; pension funding; actuarial cost methods; mortality; salary projection; duration; asset-liability management; pension accounting; regulatory technical provisions; sponsor covenant; intergenerational risk sharing; retirement income

Introduction

A pension is a financial arrangement extending work-related rights, contributions, and investment resources across decades. Its scientific analysis begins with three questions: what benefits are promised or targeted; under which demographic and economic conditions will payments occur; and how are present resources and future contributions connected to those payments? Every pension valuation is an answer to these questions under a specified legal contract, population, time horizon, and measurement purpose.

The distinction between defined contribution (DC) and defined benefit (DB) arrangements is fundamental. An employer making fixed contributions without a further legal or constructive obligation generally has a different exposure from a sponsor promising a formula-based pension. The IFRS Foundation's IAS 19 classification of pension plans expresses this distinction by reference to the substance of the employer's obligations. It also establishes the projected unit credit method for measuring defined benefit costs and obligations for the reporting entity (IFRS Foundation, IAS 19). The legal promise and the measurement convention jointly determine the meaning of a reported liability.

Scale reinforces the need for rigorous concepts. The OECD's 2025 pension assets evidence reports assets in pension plans managed by pension providers equivalent to roughly 95% of aggregate OECD GDP at the end of 2024. This aggregate is a measure of assets accumulated in pension providers and does not mean that every national pension system is 95% funded or that aggregate assets cover its accrued liabilities. The distinction between an assets-to-GDP measure and an assets-to-liabilities funding ratio will recur throughout the analysis (OECD, 2025).

This Research Dossier concerns pension plan accrual, actuarial measurement, and funding, principally occupational and collective commitments. It treats individual life-annuity pricing as an input to pension present value, rather than reproducing the broader science of annuity product design. Public pay-as-you-go systems are considered where their financing logic clarifies differences from asset-backed occupational funding. Geographic examples identify their respective regimes and are not presented as universal rules.

Methodologically, the monograph develops deterministic discounted cash-flow identities, actuarial expectation formulas, algebraic funding recursions, hypothetical calculations, scenario analysis, and source-critical legal and reporting comparisons. The numerical cases are constructed examples, not observations of an actual pension scheme. All nominal amounts are assumed to be in one currency unless otherwise stated. The factual and normative boundaries are taken from the primary reporting and actuarial standards, legislation, public regulators, and identified research literature cited in the relevant chapters.

Pension Plan Cash Flows

Benefit promises as contingent payment processes

Let the valuation date be time zero and let Bt(ω)B_t(\omega) denote the gross benefit payable at future time tt under state ω\omega. Payment depends on retirement, death, disability, survivor eligibility, service, salary, indexation, optional commutation, and plan amendment rules. If v(0,t)v(0,t) is an explicitly defined discount factor, the present value of promised benefits is

L0=E0[∑t=1Tv(0,t)Bt(ω)].L_0=\mathbb{E}_0\left[\sum_{t=1}^{T}v(0,t)B_t(\omega)\right].

The expectation is taken under specified demographic and other valuation assumptions. If discount factors themselves are stochastic and correlated with cash flows, the expectation must be taken over their joint distribution, or an appropriate valuation functional must be substituted. A deterministic scenario using expected benefits and fixed discount factors is a tractable special case; its apparent precision cannot replace disclosure of model risk.

For a retired member aged xx receiving bb annually at the end of each year while alive, with survival probabilities tpx{}_tp_x and annual effective discount rate ii, the conventional life-annuity present-value factor is

ax=∑t=1ω−xtpx(1+i)−t,Lx=bax.a_x=\sum_{t=1}^{\omega-x}{}_tp_x(1+i)^{-t},\qquad L_x=b a_x.

Here ω\omega is a limiting age assumed by the mortality model. The result must be adapted for monthly frequency, benefits paid in advance, contingent survivor pensions, guaranteed payment periods, taxes if relevant to the obligation, and escalation. A benefit formula is incomplete without these payment conventions. Discrete mathematics is used here to exhibit the structure; real actuarial systems often use monthly cash-flow engines and multiple-decrement tables.

For an active member, the payment event also involves survival to retirement, future service, salary, retirement choice and potential withdrawal benefits. A simple retirement-only projected pension for member jj, valued at time zero, may be written as

PVj=R−xpxactive v(0,R−x) bj(R) aR,PV_j={}_{R-x}p_x^{\mathrm{active}}\,v(0,R-x)\,b_j(R)\,a_R,

provided the active survival term incorporates the relevant pre-retirement decrements and the annuity factor is consistent with the post-retirement mortality model. This simplified expression excludes death and withdrawal benefits, which must be valued separately when promised. The total projected benefit present value equals the sum of all member-specific benefit components and administrative outflows included in the stated valuation scope.

DB, DC, and collective designs

In a final-salary DB plan, the normal retirement pension may be specified as b=αnSRb=\alpha n S_R, where α\alpha is the annual accrual percentage, nn is credited service and SRS_R is final or final-average pensionable salary. Other DB designs promise a flat dollar amount per year of service, a career-average revalued earnings amount, or a cash-balance credit with prescribed interest. Vested rights, early retirement reductions, salary caps, conditional indexation, and survivor benefits alter the cash-flow schedule.

In a DC arrangement, a participant's account evolves as

Wt+1=(Wt+Ct−Ft)(1+Rt+1)−Dt+1,W_{t+1}=(W_t+C_t-F_t)(1+R_{t+1})-D_{t+1},

where CtC_t is contribution paid before the return period, FtF_t is a fee paid before that period, Rt+1R_{t+1} is the realized portfolio return, and Dt+1D_{t+1} is any distribution after investment returns. The sequence of cash-flow timing is part of the definition. Reordering cash flows changes the balance; the equation is an accounting recurrence, not a return forecast. Retirement income is determined by the account's realization and the payout rule. The PBGC's definitions of DB and DC plans correctly emphasize the promised benefit versus account-based entitlement (PBGC, Glossary).

Collective defined contribution (CDC) and some hybrid arrangements share asset and longevity experience among members while adjusting benefits according to plan rules. The literature on intergenerational sharing in collective DC investigates redistribution and welfare within a fully funded model without an external guarantor (Chen, Kanagawa, and Zhang, 2023). Such models are analytically distinct from a guaranteed DB promise. Before comparing funding ratios, the observer must determine which benefits can be reduced and by whose decision.

Public finance and occupational funding

Asset-backed funding accumulates segregated resources or qualifying assets to discharge promised benefits. Pay-as-you-go finance uses current contributions or fiscal revenue to pay contemporaneous benefits, potentially alongside reserve assets. A country's pay-as-you-go pension may generate substantial implicit obligations while holding comparatively small earmarked assets. Therefore, a zero or low asset reserve in a pay-as-you-go system cannot be interpreted as an occupational DB funding deficit without first defining the corresponding accounting unit and financing law. The OECD Pensions Outlook 2024 analyzes the design and risks of asset-backed pensions, including retirement income and longevity risk (OECD, 2024).

The measurement perimeter

Five different amounts require explicit labels: the present value of all expected projected benefits, the obligation attributed to service already rendered, the accounting liability recognized by the sponsor, the regulatory or funding technical provisions, and the present market or recognized value of assets. A financial statement may combine these amounts in prescribed ways. A contribution schedule can be governed by additional law, plan rules, amortization periods, or sponsor agreements. These amounts can differ correctly because they answer different questions. Equating them creates false diagnoses of pension solvency or accounting understatement.

Defined Benefit Accrual

Service attribution

Accrual connects total projected pension payments to past and current service. Suppose an active employee is age xx with ss credited years already rendered, retirement age RR, and total projected credited service N=s+(R−x)N=s+(R-x) under continuous future employment. For a level accrual final-salary promise bR=αNSRb_R=\alpha N S_R, the portion attributed to prior service under a simple projected-unit-credit allocation is

bRpast=αsSR,bRcurrent year=αSR.b_R^{\mathrm{past}}=\alpha s S_R,\qquad b_R^{\mathrm{current\ year}}=\alpha S_R.

If KxK_x represents the present-value factor from retirement pension amount to the valuation date, then the simplified attributed obligation and normal cost are

LxPUC=αsSRKx,NCxPUC=αSRKx.L_x^{\mathrm{PUC}}=\alpha s S_R K_x,\qquad NC_x^{\mathrm{PUC}}=\alpha S_R K_x.

These expressions assume uniform allocation of benefit credits by service year, constant accrual percentage, and consistent projection and discount conventions. They are illustrations of the mathematical attribution rule; actual IAS 19 attribution includes detailed treatment of benefit formulas, later service, and cases requiring straight-line attribution. The IFRS Foundation's IAS 19 projected unit credit method is the appropriate accounting starting point for such analysis (IFRS Foundation, IAS 19).

A critical point is that a salary-linked benefit earned through past service remains sensitive to assumptions about future pay. The past-service service quantity ss is fixed, while projected SRS_R is not. Consequently, a projected benefit obligation for past service can rise when expected future salary increases, even when the employee earns no additional credited year between two measurement dates. Changes can also result from the passage of time, demographic experience, discount curves, or amendments.

Vested, accrued, projected, and payable amounts

A vested benefit is an entitlement that survives a specified termination event under the plan's conditions; its cash value still depends on payment age and mortality. An accrued benefit is the amount attributed to service to date under a specified rule. A projected benefit incorporates future salary or service assumptions as the particular definition permits. A payable benefit is the cash amount under the occurrence of an eligibility event. These categories can overlap without being equivalent.

The same caution applies to the phrase pension obligation. The ASOP No. 4 measurement standard treats measurements in relation to their purpose and methods rather than presuming that there is one universal actuarial liability (Actuarial Standards Board, 2021; effective 2023). Accrued liability under an actuarial cost method is a cost-allocation construct defined by that method. It need not be the amount required to buy out every promised benefit at commercial annuity prices.

Unit credit and entry age normal

The projected unit credit (PUC) method allocates projected benefits by earned service, recognizing actuarial present value of each allocated unit. The entry age normal (EAN) approach allocates the actuarial present value of projected benefits over the expected career by a level monetary amount or a level percentage of pay, depending on its selected variant. EAN normal cost is designed to produce a systematic contribution pattern; the residual present value assigned to prior service is the actuarial accrued liability under EAN. Unit credit and entry age normal can produce different accrued liability and normal cost for the same projected cash flows, even when mortality and discount assumptions coincide.

Let PVFBxPVFB_x be the present value of projected future benefits at valuation age xx, and let PVFNCxPVFNC_x denote the present value of future normal costs on an EAN schedule. Then

AALxEAN=PVFBx−PVFNCx.AAL_x^{\mathrm{EAN}}=PVFB_x-PVFNC_x.

This is an identity under the chosen EAN allocation, not a separate payment promise. Suppose level-percentage-of-pay normal cost is cStc S_t. The rate cc is determined by requiring the present value of such career allocations to equal projected benefits at the entry-age basis. The resulting cc depends on entry-age assumptions, the projected salary path, survival, and benefit design. Entry-age normal is therefore not automatically numerically interchangeable with accounting PUC. The differences and appropriate uses of cost and contribution allocation are covered in ASOP No. 4 (Actuarial Standards Board, 2021).

Reconciliation through experience and changes

A pension obligation changes for multiple reasons. A useful projected roll-forward starts with opening liability, adds attributed new service, interest accretion and benefit or assumption changes, and subtracts benefit payments, with additional adjustments for settlements, currency translation or other defined events. For accounting under IAS 19, terminology and routing among service cost, net interest, and remeasurements follow the specific standard. The roll-forward is a reconciliation framework; a funding valuation uses its own measure and should not silently reproduce an accounting presentation.

Benefit improvements and past-service changes must also be distinguished from demographic or financial assumption revisions. A new benefit formula changes the contractual cash flows. A new discount rate changes their valuation. Treating an assumption revision as a benefit granted by the employer would obscure the allocation of economic responsibility.

Demographic and Salary Assumptions

Mortality and longevity improvement

Mortality is a term structure, not a single death probability. For member aged xx in calendar year tt, period mortality qx,tq_{x,t} may be combined with projection assumptions to produce future cohort survival probabilities. Benefit cash flows extending from employment through retirement depend on the appropriate joint assumptions concerning mortality, disability, retirement, and eligible survivors. A life pension is exposed to survival longer than assumed; a death benefit may respond differently.

Longevity forecasts require explicit treatment of extrapolation risk. The published Cairns, Blake, and Dowd mortality model examines a two-factor stochastic mortality structure, longevity risk, and parameter uncertainty, with particular analysis of the United Kingdom (Cairns, Blake, and Dowd, 2006). That article establishes the relevance of stochastic mortality for long-duration obligations; it does not license transplantation of its historical calibration to current pension populations without new estimation and validation.

Differences in occupation, socioeconomic position, retirement status, and benefit size can materially affect pensioner mortality. Small plans may have insufficient deaths to estimate subgroup mortality independently. Credibility-weighted use of external experience and explicit model judgment are then necessary. The revised ASOP No. 27 assumption standard addresses population characteristics, available experience, external tables, future change and consistency of assumptions (Actuarial Standards Board, 2023; effective 2025).

Retirement, turnover, disability, and survival dependencies

A projected pension liability depends on retirement dates: early retirement can increase payment duration and alter the benefit formula, while deferred retirement can increase accrued service and salary. Employee withdrawal may trigger a deferred vested pension or a cash refund. Disability transitions may involve service accrual, early commencement or different mortality. Modeling each decrement separately and summing probabilities without a coherent transition system can generate impossible totals. A multi-state or multiple-decrement framework requires each state transition to exhaust the appropriate probability mass.

Plan behavior is not necessarily invariant to incentives. A generous unreduced early retirement option changes the distribution of retirement dates. Lump-sum elections respond to interest rates and plan rules where options exist. Assumptions should be tested both for empirical credibility and for responsiveness to changes in incentives, consistent with the risk-sensitive treatment in ASOP No. 51 (Actuarial Standards Board, 2017; effective 2018).

Salary, inflation, revaluation, and benefit escalation

For a basic deterministic salary projection, SR=S0(1+g)R−xS_R=S_0(1+g)^{R-x}, where gg is effective annual salary growth. A realistic salary scale may decompose general inflation, productivity, career progression, promotion and merit adjustments, with correlations and changes over time. Pensionable salary might exclude bonuses or be capped. A career-average revalued earnings arrangement requires a rule for each historical year's credit and its subsequent revaluation. Consequently, a single salary-growth factor cannot generally represent every scheme.

Inflation appears in at least four distinct channels: salary indexation before retirement, revaluation of accrued deferred rights, indexation of pensions in payment, and nominal versus real discounting. If benefits and discounting are consistently specified, nominal and real projections can be equivalent under a deterministic inflation assumption. If a nominal discount rate is ini_n and inflation is π\pi, the exact corresponding real rate is (1+in)/(1+π)−1(1+i_n)/(1+\pi)-1; the approximation in−πi_n-\pi is increasingly inaccurate as rates rise. Conditional indexation creates state dependence that a deterministic real rate cannot fully capture.

Economic assumptions should be mutually consistent and consistent with the measurement mandate. The updated ASOP No. 27 framework combines economic and demographic assumption selection, requires attention to the combined effect of assumptions, and took effect for the specified reports and measurement dates on or after January 1, 2025 (Actuarial Standards Board, 2023). The former separate ASOP No. 35 was repealed for work performed on or after that date (Actuarial Standards Board, 2024). A current pension analysis should not cite ASOP No. 35 as the independent operative assumption-selection standard.

Discount rate: purpose determines the basis

The discount rate is one of the most consequential inputs. An accounting measure may require a market-based discount curve; a statutory funding measure may prescribe segmented rates or impose smoothing; an economic risk valuation may use market-consistent liability replication; a pension budget may incorporate anticipated long-term returns for another specified purpose. These bases answer distinct questions.

Under IAS 19 discount rate requirements, the reporting obligation is measured using the prescribed market reference framework for high-quality corporate bond yields, with a government bond fallback in currencies lacking a deep relevant corporate bond market, together with maturity and currency consistency. The IFRS guidance does not instruct entities to discount IAS 19 liabilities at their forecast equity portfolio returns (IFRS Foundation, IAS 19). U.S. funding standards may apply a different prescribed rate methodology. Selecting a higher discount rate solely to improve a funding ratio is not an actuarial validation argument.

Model validation and assumption governance

Reasonableness involves data relevance, selection authority, reproducibility, sensitivity, and disclosure. A robust pension assumptions report records census date and completeness, age and service reconciliation, historical salary and pension anomalies, mortality table source, projection period, retirement decrement structure, yield curve or prescribed rate basis, expense treatment, and the changes since prior valuation. Back-testing can assess experience against earlier expected events, while prospective stress testing identifies exposures to plausible future states.

Validation must distinguish parameter uncertainty from model uncertainty. Parameter uncertainty concerns estimates within a chosen model, such as a mortality improvement coefficient. Model uncertainty concerns whether the chosen structure, including retirement behavior and cohort effects, is suitable at all. Where empirical evidence is unavailable, a transparent scenario range is more informative than a fabricated precision interval.

Projected Benefit Obligations

A fully specified final-salary example

Consider an illustrative final-salary defined benefit plan with one active member. At the valuation date the member is age 45, has 15 completed years of service, and receives an annual pensionable salary of 60,000 currency units. Normal retirement is at age 65, with another 20 years of assumed service. The formula provides an annual pension of 1.5% of final salary per credited year, paid at the end of each year of retirement. Projected salary growth is 3% effective per year, the deterministic discount rate is 4%, and the simplified annuity consists of exactly 20 annual payments beginning one year after retirement. All survival, retirement, turnover, tax, expenses, indexation, plan-option, and post-retirement inflation contingencies are deliberately suppressed. Thus this is a deterministic time-value and attribution illustration, not an actuarially complete or IAS 19-compliant valuation of an actual plan.

The final salary projection is

S65=60,000(1.03)20=108,366.67.S_{65}=60{,}000(1.03)^{20}=108{,}366.67.

The annual pension attributable to the 15 years already rendered equals $0.015\times15\times108{,}366.67=24{,}382.50. The total normal-retirement pension after 35 years would be \0.015\times35\times108{,}366.67=56{,}892.50$. The 20-payment annuity-immediate factor at 4% is

a20‾∣0.04=1−1.04−200.04=13.59032634.a_{\overline{20}|0.04}=\frac{1-1.04^{-20}}{0.04}=13.59032634.

The present-value factor for each annual pension amount, with 20 years of deferral and first annual payment at time 21, is

K=1.04−20a20‾∣0.04=6.20244754.K=1.04^{-20}a_{\overline{20}|0.04}=6.20244754.

Consequently, the simplified projected-unit-credit present value of the 15-year past-service portion is $24{,}382.50\times6.20244754=151{,}231.19. The illustrative service cost of one additional unit-credit year, holding all other assumptions and timing fixed, is \1{,}625.50\times6.20244754=10{,}082.08. The present value of the full 35-year projected pension is \56{,}892.50\times6.20244754=352{,}872.77$. These three numbers are internally coherent because each uses exactly the same retirement salary, payout schedule and discount basis.

The illustration exposes two frequently conflated dimensions. Service attribution changes the number of benefit units belonging to past service; projected salary changes the size of each unit. The difference between 151,231.19 and 352,872.77 represents the allocation of a projected total benefit across completed and assumed future service, not an actual shortfall in plan assets. Under real employee data, the valuation would incorporate probabilities of remaining active, eligibility conditions, payments on death or disability, survivor benefits and the precise IAS 19 benefit-attribution rules.

Discount-rate sensitivity of the example

Keeping annual salary growth at 3%, the accrued 15-year projected payment stream has the following deterministic values:

Effective annual discount rate Deferred annuity factor at time zero Past-service present value
3% 8.23729711 200,845.91
4% 6.20244754 151,231.19
5% 4.69687601 114,521.59

The result is pronounced because payments are remote. The value at 3% is roughly 32.8% higher than at 4%, whereas the value at 5% is approximately 24.3% lower. This asymmetry is consistent with convexity: discount-rate reductions raise a fixed distant cash-flow value more than equally sized rate increases reduce it, to first order modified duration plus a positive second-order effect. These percentages are specific to the invented 20-year deferral and 20-payment structure; they are not generalized pension industry sensitivities.

Salary-growth sensitivity

Keeping the discount rate at 4% yields a second sensitivity table:

Annual salary growth to age 65 Projected final salary Past-service present value
2% 89,156.84 124,422.90
3% 108,366.67 151,231.19
4% 131,467.39 183,469.41

Because the benefit is proportional to projected final salary, the corresponding present value is also proportional to the salary projection. The example assumes an unbounded final-salary formula. Real plans may use averaging periods, contribution caps, caps on pensionable pay, service ceilings or benefit floors that make the relationship piecewise rather than perfectly proportional. In a career-average revalued benefit arrangement, salary growth affects each annual accrual separately, with potentially different revaluation rules.

Projected obligation, accrued obligation, and net pension position

Suppose the same hypothetical plan has recognized assets worth 130,000 on the same valuation basis and date. Comparing 130,000 with the illustrative past-service obligation of 151,231.19 gives a ratio of approximately 85.96% and an arithmetic deficiency of 21,231.19. Both figures are meaningful only after declaring that this comparison uses the simplified 4% PUC obligation and a hypothetical asset amount. A plan's statutory funding ratio, IAS 19 net defined benefit liability, insurance buyout position and economic solvency may differ materially.

The IFRS Foundation's IAS 19 benefit measurement rules require an entity to determine the present value of its defined benefit obligation and deduct the fair value of plan assets, subject to the recognition framework applicable to a surplus (IFRS Foundation, IAS 19). Distinctly, the IAS 26 retirement plan reporting standard governs the reporting of retirement benefit plans and addresses net assets available for benefits alongside actuarial present value of promised retirement benefits, distinguishing vested and nonvested amounts (IFRS Foundation, IAS 26). The sponsor's accounts and the plan's own accounts therefore have different reporting perspectives.

Why accounting obligations differ from funding valuations

A reporting liability is defined by a reporting framework, measurement date, treatment of the sponsor's obligation and prescribed accounting assumptions. A funding target is defined by the plan's legal funding regime and the relevant actuarial contribution methodology. A settlement or insurance buyout cost reflects additional items such as pricing margins, administrative expenses, market conditions and the actual ability to complete the transaction. One cannot infer a commercial settlement price by multiplying an IAS 19 obligation by an arbitrary percentage. Nor does an accounting surplus necessarily constitute cash that an employer can freely withdraw.

In U.S. private single-employer funding, Internal Revenue Code section 430 provides statutory definitions of the funding target, target normal cost, funding shortfall and required contribution components. Its segment-rate and other statutory provisions may produce a target different from an employer's financial-reporting amount (United States, 26 U.S.C. § 430). In the United Kingdom, the Pensions Act 2004 funding objective requires sufficient and appropriate assets to cover technical provisions for schemes within its relevant scope, with technical provisions calculated under the regulatory framework and consistent with the scheme's funding and investment strategy (United Kingdom, Pensions Act 2004, section 222). Neither figure can simply be substituted for the other.

Measurement when the plan is underfunded

The next question after measuring a liability is what action follows. An underfunded position has at least four dimensions: the contractual ability to alter benefits, the availability and liquidity of existing assets, the legal and economic strength of the contributing employer or guarantor, and the time allowed to repair the shortfall. A contribution requirement is an intertemporal policy rule. It does not follow mechanically from dividing an accounting deficit by the number of years remaining until retirement. The funding chapter derives that relationship in a controlled form.

Funding Methods and Contribution Policy

The plan's fund balance identity

Write AtA_t for plan assets immediately before the period's contributions, CtC_t for contributions received at the start of the period, Bt+1B_{t+1} for benefits paid at the end, Et+1E_{t+1} for expenses paid at the end, and Rt+1R_{t+1} for the realized portfolio return. The corresponding balance equation is

At+1=(At+Ct)(1+Rt+1)−Bt+1−Et+1.A_{t+1}=(A_t+C_t)(1+R_{t+1})-B_{t+1}-E_{t+1}.

The identity establishes what the investment portfolio actually funds. It also clarifies the distinction between realized investment return and an actuarial assumed return. An actuarial model may specify a forecast distribution for Rt+1R_{t+1}; no forecast changes the asset value realized at the end of the year. When cash contributions arrive during the year, their timing must be included explicitly, rather than being treated as if all payments arrive at the opening date.

Let LtL_t be an actuarial funding liability measured on a consistent basis. Define the arithmetic unfunded actuarial liability Ut=Lt−AtU_t=L_t-A_t and funding ratio Ft=At/LtF_t=A_t/L_t when Lt>0L_t>0. These are state variables and diagnostics, not the contribution policy itself. An obligation estimate is not a segregated asset; a target funding amount is not a legal guarantee that the employer can supply the missing contributions.

Normal cost and amortization

A conventional actuarially determined contribution can be summarized as

Cttarget=NCt+AMt+EXt−CRt,C_t^{\mathrm{target}}=NC_t+AM_t+EX_t-CR_t,

where NCtNC_t is normal cost for benefits attributed to the period; AMtAM_t is a scheduled amortization payment on a defined unfunded amount; EXtEX_t represents expenses included in the contribution policy; and CRtCR_t represents specifically authorized credits, such as excess asset or prefunding credits, under the governing framework. This is an accounting of possible components, not a universal statutory formula. The legal required contribution may impose floors, caps, timing rules, elections and separate amortization bases.

For an illustrative fixed unfunded liability U0U_0 amortized by nn equal year-end payments with effective rate iai_a, the amortizing payment is

AM=U0an‾∣ia=U0ia1−(1+ia)−n.AM=\frac{U_0}{a_{\overline{n}|i_a}}=U_0\frac{i_a}{1-(1+i_a)^{-n}}.

If U0=100,000U_0=100{,}000, the policy chooses n=10n=10 and ia=4%i_a=4\%, then AM≈12,329.09AM\approx12{,}329.09 per year, assuming no newly arising gains or losses, no benefit payments affecting the specified closed amortization base, and timely year-end payments. The equation illustrates the present-value identity, not the U.S. regulatory schedule. An open amortization period that resets indefinitely does not, by construction, guarantee repayment of the original base on a fixed date.

Funding method is a policy selection

A funding method links projected costs and existing assets to the contribution time path. Under PUC, attributed accrued liability can increase markedly when a final-salary career approaches retirement, even if service-based normal cost is conceptually uniform per year. Under EAN, actuarial normal cost can be designed as a level percentage of pay from entry age, thereby allocating expected cost differently over the career. Differences in accrued liability follow from this allocation; they are not necessarily a disagreement about the benefit formula.

U.S. ASOP No. 4 cost allocation guidance addresses both actuarial cost methods and contribution allocation procedures, including considerations that differ according to the measurement's purpose (Actuarial Standards Board, 2021). The same plan may legitimately use projected unit credit for IAS 19 accounting and a separate prescribed or selected actuarial funding method. Presenting a method without its purpose makes the liability ambiguous.

Statutory minimum contributions in the United States

For private U.S. single-employer defined benefit plans within the specified section 430 regime, the minimum funding statutory structure connects target normal cost, funding shortfall amortization and potentially waiver amortization charges, with particular rules for excess assets, prefunding balances, timing, at-risk status and applicable interest rates. The actual required contribution is determined under the current statutory and regulatory provisions, including special rules and elections that may apply. A simple five- or ten-year amortization illustration cannot replace those requirements (United States, 26 U.S.C. § 430).

The Pension Benefit Guaranty Corporation adds a contingent protection institution for covered U.S. private-sector DB arrangements. Its insurance coverage explanation specifies covered plan types and legal limits, and distinguishes the single-employer and multiemployer insurance programs (PBGC, 2026). PBGC protection changes the consequences of some employer or plan failures, but does not remove all benefit risk and does not cover ordinary defined contribution accounts. Plan coverage and benefit guarantees require their own legal eligibility determination.

Statutory funding in the United Kingdom

The United Kingdom applies a distinct institutionally integrated funding regime to relevant occupational DB schemes. The statutory funding objective is expressed through technical provisions. The Pensions Regulator's defined benefit funding code came into force on November 12, 2024, and applies to activities concerning valuations with effective dates on or after September 22, 2024 (The Pensions Regulator, 2024). The applicability date of the valuation is therefore part of the legal factual description.

The regulator's technical provisions guidance explains trustee responsibility for assumption selection, actuarial advice and consistency with the funding and investment strategy. Its recovery plan requirements connect a technical-provisions deficit to a recovery period and the employer's reasonable affordability (The Pensions Regulator, 2024). Accordingly, the shortfall is analyzed in relation to the employer covenant and long-term strategy, rather than solely by assigning a mathematical amortization rate.

European occupational pension supervision

Directive (EU) 2016/2341, commonly known as IORP II, establishes an EU framework concerning institutions for occupational retirement provision. Its consolidated occupational pensions directive contains requirements concerning technical provisions, financing, governance and, within the specified activities, solvency margins and supervisory powers (European Parliament and Council, 2016, consolidated 2025). National implementation and institutional scope must be checked before attributing a particular funding rule to a named pension scheme. The EU directive cannot be read as prescribing one universal cross-border discount rate for all occupational plans.

Asset smoothing, funding stability, and delayed recognition

A funding policy may use an actuarial asset value that smooths short-term market movements. This can stabilize contributions, but it also introduces a difference between recognized funding assets and the current realizable market value of the portfolio. Disclosures must therefore identify the amount and method of smoothing. A stable published contribution rate can coexist with volatile economic funded status; stability in the chosen reported metric is not proof that risk has disappeared.

A similar distinction applies to liability smoothing or prescribed interest-rate stabilization. If rising market yields reduce an economic liability but a statutory discount curve incorporates a multiyear average, the reported funding target may adjust more slowly. Conversely, a market shock can produce immediate economic deterioration before all statutory effects appear. A fair comparison across reporting systems must align valuation dates, maturity profiles, asset concepts and actuarial conventions.

Open and closed plans

Closed plans are especially sensitive to the remaining runoff of liability cash flows and to the ability of a shrinking sponsor or member base to support fixed administrative costs. An open plan admits new members or continued accruals, but entry of new participants is not equivalent to new assets unless contributions meet the corresponding incremental cost. Using expected future new members to subsidize an existing deficit requires explicit authorization and a demonstrable policy; indefinitely assuming a growing contribution base can hide structural underfunding.

The employer's financial capacity, pension funding law, and participants' rights determine the feasible contribution path. Actuarial valuation supplies evidence for those decisions. Sustainable funding requires that assumed investment returns, required contributions and permissible benefit adjustments be jointly consistent with the actual institution.

Asset and Liability Sensitivity

Duration and convexity of pension liabilities

For fixed deterministic benefit amounts CFtCF_t and a flat annual yield ii, let

L(i)=∑t=1TCFt(1+i)t.L(i)=\sum_{t=1}^{T}\frac{CF_t}{(1+i)^t}.

The Macaulay duration DMD_M of this liability is the discounted-payment weighted average time, while modified duration Dmod=DM/(1+i)D_{\mathrm{mod}}=D_M/(1+i) is the local sensitivity to the effective yield:

DM=∑tt CFt(1+i)−tL(i),ΔLL≈−DmodΔi+12C(Δi)2.D_M=\frac{\sum_t t\,CF_t(1+i)^{-t}}{L(i)},\qquad \frac{\Delta L}{L}\approx-D_{\mathrm{mod}}\Delta i+\tfrac12\mathcal C(\Delta i)^2.

Here C\mathcal C denotes the appropriate discrete-yield convexity measure, and the approximation omits higher-order terms and benefit changes. A pension promise with salary-linkage, inflation-linked benefits, or interest-sensitive retirement elections may have additional cash-flow responsiveness that fixed-cash-flow duration does not capture.

For the constructed 20-year deferred, 20-payment level annuity at 4%, the liability's Macaulay duration is approximately 29.21 years, yielding modified duration of about 28.09. The exact sensitivity tables in the preceding chapter are preferable to a first-order approximation for a full percentage-point shock. An investment portfolio holding short-duration fixed income against this long-duration liability will ordinarily leave material rate exposure even if its current market value equals the liability value.

Funding-ratio response to interest rates

Let F=A/LF=A/L and suppose a small parallel yield change produces asset and liability modified durations DAD_A and DLD_L, respectively, for fixed cash-flow approximations. Then, to first order,

ΔFF≈(DL−DA)Δi.\frac{\Delta F}{F}\approx(D_L-D_A)\Delta i.

If liability duration is 16 and asset duration is 5, a 1 percentage-point yield decline implies an approximate −11%-11\% relative change in the funding ratio from duration effects alone: (−0.01)(16−5)=−0.11(-0.01)(16-5)=-0.11. If the initial ratio is 90%, the first-order estimate becomes approximately 80.1% (0.90 multiplied by 0.89). This does not say that every 90%-funded plan reaches 80.1% after a rate fall. It is a controlled hypothetical with parallel yield changes, constant spreads and fixed cash flows; convexity and real-world asset composition can alter the response substantially.

A plan can also be market-value funded at 100% and still have a substantial mismatch in key-rate durations, inflation exposures, credit spreads or longevity. Full funding is a level condition at a date. Immunization is a sensitivity condition across changes in underlying drivers. The two should be assessed separately.

Longevity, inflation, salary, and option risk

Longevity improvement raises expected periods of pension payment for otherwise unchanged annual amounts. Inflation surprises change purchasing power and nominal cash flows where benefits have explicit or conditional indexation. Above-expected salary growth raises final-salary benefits while generating changes in contribution bases. Option exercise affects retirement timing and commutation. A single deterministic stress can isolate one risk, but risks are often correlated: a downturn can alter investment returns, salary growth, employment, employer covenant and discount-rate markets simultaneously.

The ASOP No. 51 risk framework addresses assessment and disclosure of risk associated with measuring pension obligations and contributions (Actuarial Standards Board, 2017). In substantive application, stress design should correspond to the decisions the pension sponsor and fiduciaries can take: contributions, investment strategy, benefit options, or risk transfer. Scenario counts alone are not evidence of a validated risk distribution.

Stochastic asset-liability simulation

A general simulation draws joint states for market yields, asset returns, inflation, salaries, retirement, and mortality. At each time step, the asset recurrence updates contributions, investment outcomes and payments; the liability model updates projected cash flows and discount conventions. Appropriate outputs include the distribution of terminal funding ratios, expected additional contributions, probability of exhausting liquid assets under a defined rule, downside benefit outcomes, and the distribution of benefit adjustment or sponsor calls.

Simulation validity requires compatible processes, calibration periods, dependency assumptions, valuation versus real-world measure distinctions, and credible tail behavior. A Monte Carlo estimate of a probability is conditional on the model and estimated inputs. Stochastic simulation cannot turn incomplete member records, unspecified contractual rights or an arbitrary discount curve into reliable pension obligations. The OECD Pensions Outlook 2024 emphasizes the importance of asset-backed retirement system design, investment options and protection against longevity risk (OECD, 2024).

Liquidity and the timing of funding

A plan may have a reported funding surplus and still face liquidity stress if benefits due promptly exceed cash and assets readily realizable without material loss. Conversely, an underfunded plan can meet all near-term payments while requiring substantial later contributions. Valuation should therefore accompany a payment ladder, liquidity policy and realistic contribution timetable. The asset-liability problem is simultaneous: the value, duration, liquidity, credit quality and currency of assets must be assessed against the timing and state-contingency of benefits.

Hedging can reduce some interest-rate and inflation mismatches, but hedge effectiveness depends on collateral, basis risk, instrument maturity, counterparty exposure and liquidity in stressed markets. Buying long-duration bonds does not mechanically hedge salary growth or longevity changes. Pension risk transfer through insured buy-in or buyout transactions can alter the identity of the risk bearer and the definition of assets and obligations; it requires contract-specific and regime-specific analysis beyond this mathematical funding dossier.

Sponsor covenant and beneficiary security

The sponsor's covenant is the economic capacity and legal exposure supporting future contributions under the governing rules. Funding ratios by themselves do not capture it. Two plans with identical asset and liability values may offer very different benefit security if one has a strong, enforceable employer contribution obligation and another depends on a distressed sponsor. The U.K. regulator's integrated funding and covenant framework expressly situates scheme valuation within funding strategy, risk management and employer covenant oversight (The Pensions Regulator, 2024).

Benefit guarantee institutions change the allocation of losses at failure. For qualifying U.S. private DB plans, PBGC benefit protection depends on program, plan coverage and statutory guarantee rules (PBGC, 2026). Such protection is a distinct institutional claim and should not be added indiscriminately to plan assets or assumed to guarantee all promised benefits in every situation.

Collective risk sharing and fairness over generations

In collective arrangements, a funding shortfall may lead to benefit adjustments, changed contribution rates, altered indexation or a reallocation of investment risks across age cohorts. The scientific question includes who holds the right to decide and how shocks are assigned. Research by Chen, Kanagawa, and Zhang (2023) studies a stylized fully funded CDC setting, whereas a separate stochastic interest-rate CDC study investigates investment and benefit adjustment under explicit model assumptions (Chen, Kanagawa, and Zhang, 2023; Wang, Wang, and Zhang, 2023). Such findings depend on their objectives, market assumptions and benefit adjustment mechanics; they cannot be automatically generalized to every national pension reform.

An increase in average modeled participant welfare may coexist with an unfavorable outcome for a particular cohort or a member with an unusual career. Group-level optimality therefore does not settle the distributional question. Evaluation should report cohort-level benefit outcomes, the rule for adjusting benefits, and transition consequences for existing members. This is an analytical extension of established pension risk-sharing questions, not a proposed independent theory.

Comparative interpretation of pension adequacy

The OECD's gross replacement-rate analysis reports modeled future gross pension replacement rates across OECD and G20 economies, defined as pension income from mandatory arrangements relative to earnings under specified careers and retirement rules (OECD, 2025). A replacement rate is an adequacy indicator under its modeling assumptions. It cannot be interpreted as a funded percentage: a system with a high modeled replacement rate may have material fiscal exposure, and a highly funded account arrangement may provide variable retirement income depending on investment and annuitization outcomes.

Likewise, the OECD's net replacement-rate comparisons incorporate taxes and social security contributions in comparing retirement income to earnings (OECD, 2025). Net and gross replacement rates answer different distributional questions from actuarial liabilities. Sound cross-national analysis requires each indicator's denominator, model worker, inclusion of mandatory and voluntary schemes, and tax treatment to be stated.

Integrated interpretation and methodological boundaries

The preceding analysis establishes a hierarchy of inquiries. Contractual and constructive obligations define cash-flow rights. Demographic and economic assumptions describe their contingent evolution. Attribution determines which portion is assigned to periods of employment. Valuation rules convert cash flows into a measurement appropriate to a stated purpose. The funding rule determines how assets and contributions meet that measurement over time. Risk allocation determines who bears outcomes departing from expectation. Governance, accounting and law make these layers operational.

This hierarchy provides an original synthesis of recognized approaches; it creates no new actuarial standard or named conceptual system. Its practical consequence is rigorous comparability. A pension funded ratio becomes analytically interpretable only with a named liability basis, asset basis, valuation date, plan population and specified treatment of future service. An actuarial funding policy becomes interpretable only with a stated contribution rule, amortization structure, benefit adjustment authority and sponsor capacity. Exact definitions are more valuable than unsupported precision.

Conclusion

Pension benefit and funding models express long-duration obligations through several interconnected operations: benefit definition, probabilistic cash-flow projection, attribution to periods of service, discounting, asset valuation and contribution scheduling. Defined benefit and defined contribution arrangements allocate investment, longevity, salary and sponsor risks differently; collective designs can redistribute these risks among participants under formal adjustment rules. The relevant scientific object is therefore the complete pension commitment and financing mechanism, including its legal powers and its assumptions.

The mathematics demonstrates that projected unit credit translates past service into a present-value obligation while preserving exposure to future salary and discount-rate conditions. Entry age normal and other cost methods allocate projected cost over service by different schedules, so identical promises can produce different normal costs and accrued liabilities. The deterministic numerical example, calculated from stated assumptions, shows that a nominally fixed benefit formula can yield substantially different projected obligations across alternative economic assumptions. Its carefully limited construction is a check on arithmetic and causal interpretation, rather than an empirical estimate.

International evidence reinforces the distinction between measurements. IAS 19 serves employer accounting; IAS 26 addresses plan reporting; U.S. professional standards guide pension actuarial practice; U.S. federal law establishes specific minimum funding requirements; the United Kingdom bases occupational DB funding on technical provisions and long-term strategy; and EU occupational pension law provides a further regulatory framework. These measurements can legitimately differ because their definitions, institutional responsibilities and purposes differ. They should always be named before conclusions are drawn from reported balances or deficits.

Funding sufficiency requires more than a favorable point-in-time ratio. A plan must align the actual benefit payment schedule with assets, contributions, the employer or guarantor's capacity, liquidity, risk exposure and participants' legal rights. Duration matching reduces selected market sensitivities but does not eliminate longevity, salary, covenant or governance risk. Scenario modeling clarifies possible outcomes only when the underlying data, assumptions and risk dependencies are defensible.

The research concludes that pension security is best evaluated by tracing each material obligation from benefit clause through demographic projection and valuation method to the legally feasible funding and loss-allocation mechanism. This conclusion is a source-based analytical synthesis within established pension science. It carries no claim to a newly invented actuarial theory and no claim of externally peer-reviewed validation. The manuscript's hypothetical examples and jurisdictional comparison remain subject to the stated model and legal boundaries.

References