Insurance by BogdanovaThe Encyclopedia of Insurance and Risk Science

RESEARCH DOSSIER

Life Contingent Benefit Valuation

Abstract. Life-contingent benefit valuation converts contractual payments conditional on death, survival, or combinations of life events into time-indexed random present values. This monograph derives the central discrete and continuous formulas for term insurance, whole-life insurance, deferred benefits, pure endowments, endowment insurance, and survival annuities; distinguishes their expectations from their distributions; and examines the assumptions connecting individual lifetime models, payment conventions, interest rates, and accounting or solvency purposes. Classical actuarial literature establishes an intellectual lineage from life-table annuity calculations to modern stochastic life-contingency analysis, while contemporary mortality experience and forecasting research show why estimates require explicit population, calendar-time, and selection qualifications. A reproducible three-year numerical case computes expected present values, second moments, variance, net premium equivalence, and two distinct directions of sensitivity. International comparison distinguishes elementary actuarial present values from the broader measurement systems of IFRS 17, United States principle-based valuation, and European Solvency II. The central analytical finding is that a life-contingent value is defined by the joint specification of an insured event, benefit entitlement, settlement time, survival or death law, and discounting convention; changing any one of these changes the estimand. This is an original synthesis and derivational exposition of established actuarial mathematics, without a claim to a new theory or valuation doctrine.

Keywords: Life contingencies; actuarial present value; term insurance; whole-life insurance; pure endowment; endowment insurance; survival benefit; annuity; mortality; force of mortality; life table; select mortality; mortality improvement; discount factor; interest-rate sensitivity; prospective valuation; equivalence principle; IFRS 17; Solvency II; principle-based reserving; actuarial model validation.

Introduction

A life-contingent benefit is a contractual payment whose occurrence, amount, or time depends on the survival or death of one or more identified lives. It is measured by locating an entitlement in time, assigning a probability to its triggering state, and translating the resulting payment into a value at a specified valuation date. That three-part description becomes rigorous only when the covered life, benefit definition, observation period, settlement convention, mortality basis, and discounting basis are fixed. The same insured event can give rise to materially different present values under a death-at-year-end contract, an immediate-upon-death contract, an annually payable annuity, or an escalating survivor benefit.

Life-contingent valuation is an established branch of actuarial mathematics developed across centuries of mortality investigation, probability theory, financial mathematics, and insurance practice. Edmond Halley's 1693 Breslau mortality and annuity study used death observations in an attempt to price life annuities; Benjamin Gompertz's 1825 law of human mortality advanced a functional representation of age-related mortality. Their historical contribution should be interpreted through the assumptions and population data available to them rather than treated as directly usable mortality bases for contemporary insurance contracts. Dickson, Hardy, and Waters (2020) provide a modern comprehensive reference for actuarial mathematics of life contingencies.

The monograph proceeds by deriving the value of each benefit from its associated random cash flow. Mathematical identities follow from stated assumptions. References to professional standards establish documented procedural expectations; references to accounting and prudential frameworks describe particular institutional uses of cash-flow estimates. The examples use invented mortality and interest assumptions for reproducibility. They represent neither current market premiums nor empirical estimates for an insured population. The boundary of this dossier is valuation of life-contingent benefits as a foundational scientific object; comprehensive premium structures, insurer expenses, statutory reserves, and capital requirements receive contextual discussion because their complete treatment belongs to related research dossiers.

Life Contingencies and Cash Flow States

The future lifetime as the valuation state variable

Let a life aged exactly x at valuation time have nonnegative future lifetime T_x. The survival function is S_x(t) = P(T_x > t), ordinarily denoted by {}t p_x; the death probability over (t,t+s] is {}t p_x - {}{t+s} p_x. For annual intervals write q{x+k} for the conditional probability of death during the next year given survival to exact age x+k, and p_{x+k}=1-q_{x+k}. When these probabilities describe one coherent population and duration basis,

kpx=∏j=0k−1(1−qx+j),P(k<Tx≤k+1)=kpx qx+k.{}_k p_x=\prod_{j=0}^{k-1}(1-q_{x+j}),\qquad P(k<T_x\leq k+1)={}_k p_x\,q_{x+k}.

The empty product for k=0 equals one. If death is assumed to occur at most once and the lifetime distribution is proper, the death probabilities across all future annual intervals sum to one. At a finite limiting table age, terminal-age conventions must be documented; blindly imposing a terminal q equal to one can materially alter high-age values.

A life table typically uses l_x as a conventional count of survivors at age x, d_x=l_x-l_{x+1} as annual decrements, and q_x=d_x/l_x when l_x is positive. It follows that {}k p_x=l{x+k}/l_x for a consistent table. The l_x radix, often chosen for computational convenience, is not the number of persons insured. Table quantities are modeling or data summaries whose population, exposure, and calendar period must be established. The Society of Actuaries' 2015 Valuation Basic Tables illustrate differentiated insured-life mortality tables by sex, smoking classification, and age convention, as well as documented subsequent corrections and updates. An analyst must select the appropriate table version before calculating contractual values.

For differentiable survival functions with positive survival, the force of mortality is μ_{x+t}=−d log S_x(t)/dt, giving S_x(t)=exp(−∫0^t μ{x+s} ds). The force is an instantaneous conditional death intensity. It is not itself a probability and can exceed one per unit of time without violating probability axioms. A constant force μ yields exponential survival exp(−μt); an age-varying Gompertz-type hazard produces a materially different duration profile, even when short-horizon mortality probabilities are calibrated similarly. A continuous hazard also allows death benefits paid at the instant of death to be valued without relying on an end-of-year approximation.

Contractual event states and payment clocks

Define the benefit-payment random variable by describing mutually exclusive events in which money changes hands. Death during the covered interval may generate a death benefit B_D(t); being alive at a specified maturity date may generate a survival benefit B_S(n); remaining alive at periodic annuity dates may generate a sequence of payments b_k. A survival payment at a fixed future time is legally different from an inheritance, a refundable premium, an unconditional maturity benefit, or a cash surrender value. Valuation follows the contract actually described.

A payment conditioned on death can be measured at the time of death, at the end of the policy year of death, after a contractual notification or claims-assessment delay, or at another date. The time of death is therefore an event-time variable while the settlement time may be a distinct random variable. If a fixed delay δ applies after death, payment is discounted to T_x+δ; if the delay depends on claims administration or jurisdiction, the model must explicitly represent that dependence. For policies requiring active coverage, a lapse or conversion state may end one type of benefit or create another. A pure survival/death model assumes away these additional decrements for the purpose of elementary derivation.

A convenient valuation object is the discounted payment stream Z=Σ_j D(τ_j) C_j I_j, where C_j denotes a payment amount, τ_j the payment time, I_j its eligibility indicator, and D(t) the discount factor to the valuation date. The events need not be independent of benefits, financial outcomes, or each other. The assumption that allows the elementary formula E[D(τ) C I]=D(τ) C P(I=1) is that D(τ) and C are deterministic at the specified payment time. General models retain their stochastic dependence and value the expectation of the entire product.

Measurement date and discount conventions

For deterministic annual effective interest i, v=(1+i)^{−1}; a payment of one unit at time k has present value v^k. A general deterministic term structure has discount factors D(t)>0, frequently written D(t)=exp(−∫_0^t r(s)ds) for a specified continuously compounded forward-rate curve r(s). The distinction between annual effective rate, nominal rate convertible m times, continuously compounded rate, and spot yield curve is essential. For example, applying an annual effective 4% rate to a payment at year three gives (1.04)^{−3}, whereas discounting with 4% continuously compounded gives exp(−0.12), a different number.

If benefit amounts are indexed to prices or earnings, the indexation rule and nominal or real discount basis must be mutually consistent. If a benefit is denominated in one currency and the liability is reported in another, exchange conversion, discounting, and contractual currency must also be specified. Regulatory values may require prescribed curves rather than the insurer's preferred investment yield. Hence identical projected mortality does not by itself establish a unique accounting or solvency carrying value.

Term and Whole-Life Benefits

Annual term insurance

Consider a level death benefit B paid at the end of the policy year of death if death occurs during the next n years. Its present-value random variable is Z_T=Bv^{K_x+1} if K_x<n and zero otherwise, where K_x=floor(T_x) for deaths between integral policy anniversaries. Thus the n-year term insurance actuarial present value is

Ax:n‾∣1=∑k=0n−1vk+1kpxqx+k,EPV⁡(ZT)=BAx:n‾∣1.A_{x:\overline{n}|}^{1}=\sum_{k=0}^{n-1}v^{k+1}{}_k p_x q_{x+k},\qquad \operatorname{EPV}(Z_T)=B A_{x:\overline{n}|}^{1}.

The superscript merely distinguishes the term death benefit from other endowment conventions in traditional actuarial notation. The derivation requires a coherent mortality sequence and a specified end-of-death-year payment. A time-varying death benefit B_{k+1} replaces B in each summand. If eligibility ends at an earlier of death and a contract termination event, the death probability term must be replaced by the applicable joint or competing-risk probability.

For a deferred n-year term death benefit beginning at duration m, only deaths in policy years m+1 through m+n are covered:

m∣Ax:n‾∣1=∑k=mm+n−1vk+1kpxqx+k=mpxvmAx+m:n‾∣1,{}_{m|}A^{1}_{x:\overline{n}|}=\sum_{k=m}^{m+n-1}v^{k+1}{}_k p_x q_{x+k}={}_m p_x v^m A^{1}_{x+m:\overline{n}|},

where the final factorization assumes a consistent conditional attained-age mortality basis. This identity is valuable for decomposing later-life coverage, but its right-hand side ceases to be adequate when duration selection or underwriting class changes independently of age. In a select table, the conditional model must be selected by attained age and time since underwriting.

Whole-life insurance and convergence

For a whole-life death benefit payable at the end of the year of death, the time horizon extends over the remaining lifetime:

Ax=∑k=0∞vk+1kpxqx+k.A_x=\sum_{k=0}^{\infty}v^{k+1}{}_k p_xq_{x+k}.

For i≥0 and B finite, the benefit present value lies between zero and B, and convergence follows by comparison with the total probability mass of death events. At i=0 and a proper lifetime distribution, A_x=1 for a unit whole-life death benefit. If a model admits a positive probability of surviving indefinitely, the sum at zero interest equals the probability of eventual death, which must be distinguished from unity. With strictly positive interest, a death deferred far into the future has a lower present value, even though the payment amount is unchanged.

Whole-life valuation is particularly sensitive to long-run mortality improvement, age extrapolation, and discounting assumptions, because the remaining term of coverage is often decades. Using a period mortality table as though it were a projected cohort table imports a strong assumption: every future attained-age mortality rate equals a rate observed for that age during the table's period. Empirical forecasting research rejects treating that assumption as an unexamined identity. Lee and Carter's (1992) mortality forecasting analysis represents a landmark stochastic treatment of historical changes in age-specific death rates. Neither a period table nor a particular forecasting model automatically supplies insurer-specific selection, socioeconomic composition, or future scenario uncertainty.

Immediate payment and fractional-year treatment

For death benefits payable immediately upon death, with deterministic discount function D(t), benefit B(t), and force of mortality μ_{x+t}, the actuarial present value becomes

A‾x=∫0∞D(t)B(t)tpxμx+t dt.\overline A_x=\int_0^\infty D(t)B(t){}_t p_x\mu_{x+t}\,dt.

The density {}t p_x μ{x+t} represents the probability of death near t per unit time. For a fixed benefit and constant force μ>0 together with constant continuous discount rate δ≥0, the integral is B μ/(μ+δ). This is an exact special case that provides a useful independent unit test for continuous-life software. At δ=0, the result equals B for a proper exponential lifetime. For a finite n-year term the value is B μ[1−exp(−(μ+δ)n)]/(μ+δ). Any tabular approximation should be judged against the contract's true payment time and the materiality of the approximation.

It is common for a year-end death payment to have a smaller value than an otherwise identical immediate payment when discount factors decline with time, provided all other assumptions are identical. This comparison relies on ordering of the actual payment times; it can reverse or lose its simple interpretation under unusual financial curves, contingent settlement rules, or index-linked benefit amounts. In practice, claim-payment timing, not just mortality timing, determines valuation accuracy.

Survival Benefits and Endowments

Pure endowment as the elemental survival benefit

A pure endowment promises B at a fixed future date n only if the covered life is then alive. The random present value is Z_S=Bv^n I(T_x>n), and the actuarial present value is

nEx=vnnpx,EPV⁡(ZS)=BnEx.{}_nE_x=v^n{}_n p_x,\qquad \operatorname{EPV}(Z_S)=B{}_nE_x.

This simple expression is the survival counterpart of term death insurance. It makes two dimensions transparent: mortality determines the probability of payment, while discounting determines the price of the deterministic timing conditional on payment. Unlike a death benefit, a pure endowment can become more expensive as longevity improves. Its value becomes smaller as the nominal discount rate rises when B is fixed and positive. These two sensitivities can offset one another in changing economic and demographic environments.

A pure endowment differs from a guaranteed bond payment because the former is extinguished by death under the assumed contract. It also differs from a return-of-premium death benefit, which is contingent on a separate event and may have a separate amount and payment date. Terms such as maturity benefit or guaranteed benefit are not enough: a valuation model must ascertain whether the cash flow is unconditionally guaranteed, conditional on survival, or conditional on maintaining the policy in force.

Endowment insurance and the partition of outcomes

An n-year endowment insurance paying B on death during the term at the end of the year of death, or B at time n on survival, partitions future events into disjoint possibilities. The two-part random present value is

ZE=BvKx+1I(Tx≤n)+BvnI(Tx>n),Z_E=Bv^{K_x+1}I(T_x\leq n)+Bv^n I(T_x>n),

under the convention that a death exactly at an anniversary is allocated consistently to one branch. Its expectation equals

Ax:n‾∣=Ax:n‾∣1+nEx.A_{x:\overline n|}=A^{1}_{x:\overline n|}+{}_nE_x.

The equality is linearity of expectation over the two disjoint contingencies, not an assertion about risk pooling, premiums, or reserves. If the death and maturity benefits differ, multiply each element by its corresponding benefit level. If the endowment adds bonuses, maturity options, policyholder participation, or surrender rights, further state-contingent cash flows enter the total. A contract offering multiple payments under the same realized lifetime may require nonzero covariance terms in its second moment even when the expected values remain additive.

Life annuities as repeated survival benefits

A temporary life annuity-due pays one unit at time k for k=0,...,n−1 if the life is alive at that payment date. Its expected present value is

a¨x:n‾∣=∑k=0n−1vkkpx.\ddot a_{x:\overline n|}=\sum_{k=0}^{n-1}v^k{}_k p_x.

The initial payment at time zero is certain if the contract is in force and the annuitant is alive at valuation. A corresponding immediate annuity pays at year ends conditional on survival to those dates: a_{x:\overline n|}=Σ_{k=1}^n v^k {}k p_x. Confusing the two patterns omits or misplaces one payment and therefore changes the net premium or annuity purchase value. For a lifetime annuity-due, extend the sum to infinity, assuming convergence. For an m-year deferred temporary annuity-due beginning at time m, use Σ{k=m}^{m+n−1}v^k {}_k p_x.

There is a familiar annual whole-life identity, A_x+d\ddot a_x=1, where d=i/(1+i), when the death benefit is one unit payable at the end of the year of death, the annuity-due pays one unit each year while alive, interest is a constant effective annual rate, and eventual death occurs with probability one. To see the identity for an individual dying in year K_x+1, note that v^{K_x+1}+d Σ_{k=0}^{K_x}v^k=1. Taking expectations gives the result. This proof demonstrates why matched timing and benefit conventions matter: the identity does not apply unchanged to an immediate-death benefit and a due-annuity or to arbitrary varying discount rates.

For a finite n-year contract the corresponding decomposition is A^{1}_{x:\overline n|}+{}nE_x+d\ddot a{x:\overline n|}=1 with the same annual effective interest assumption. It follows pathwise by treating death before n and survival to n as exhaustive scenarios. These relations are valuable cross-checks for implementation because they link independently calculated benefits under a single convention. Their diagnostic power disappears if different mortality versions or payment clocks are mixed.

Joint life and benefit state extensions

Payments contingent on two lives need a joint survival law. For example, a benefit payable only while both lives survive uses P(T_x>t,T_y>t), and a last-survivor annuity uses P(T_x>t or T_y>t). Under independence, the first probability factors into individual survival probabilities; independence is an empirical assumption and should not be silently adopted for spouses or household lives. The general identity P(A or B)=P(A)+P(B)−P(A and B) makes the dependence explicit. Competing-decrement models similarly need joint or cause-specific transition information to value disability, lapse, conversion, and death benefits without counting mutually exclusive exits more than once.

Present Value Random Variables

Expectation, distribution, and second moments

The actuarial present value E[Z] summarizes a random cash-flow value by its probability-weighted average. It does not by itself measure dispersion, liquidity pressure, capital required, or the market price of transfer. For a level n-year term death benefit paid at year end, the second moment is

E[ZT2]=B2∑k=0n−1v2(k+1)kpxqx+k,E[Z_T^2]=B^2\sum_{k=0}^{n-1}v^{2(k+1)}{}_k p_xq_{x+k},

and Var(Z_T)=E[Z_T²]−E[Z_T]². For a pure endowment, E[Z_S²]=B²v^{2n}{}n p_x. For the endowment benefit with mutually exclusive death or maturity payments, the second moment is the sum of the second moments of the two branches. For a life annuity, payments at different dates can co-occur, so the second moment includes cross-products. In particular, E[I(T_x>j)I(T_x>k)]={}{max(j,k)}p_x; multiplying separate survival probabilities would generally misstate annuity variance.

A useful abstract representation is Z=Σ_j X_j for discounted contingent payments X_j. Then E[Z]=Σ_j E[X_j], whereas Var(Z)=Σ_j Var(X_j)+2Σ_{j<k}Cov(X_j,X_k). The covariance terms express common lives, common economic conditions, shared mortality trends, and other dependencies. For an individual life, many future survival indicators are nested: survival to year ten implies survival to year five, and the two payments are correlated. For an insurer's portfolio, individual conditional independence still allows correlation through systematic mortality or financial factors.

An individual endowment's benefit amount may be almost certain to be paid if it pays either at death or at maturity, yet its present value remains uncertain because the timing differs. If the discount rate were zero and the two benefits were exactly equal, the endowment present value would be constant B under the exhaustive-event convention, giving zero variance. A positive discount rate reintroduces timing risk. This simple limiting case is a stringent check on formulas and on explanations of the source of risk.

Death, survival, and the direction of mortality exposure

For the same remaining period and fixed positive benefits, higher mortality typically increases a term death-benefit present value, subject to the timing and interest assumptions, and decreases a pure endowment present value. For endowment insurance the net effect depends on whether an earlier death benefit is larger or more heavily discounted than the maturity payment. It is therefore invalid to state that a higher mortality assumption always raises or always lowers all life-contingent liabilities. A change in the hazard at an early duration influences every later survival probability and may change the weighted distribution of death dates. The sign of a product's sensitivity must be established from its cash-flow design.

Mortality table comparisons require additional care when a revised table changes exposure definitions, underwriting selection, smoker categories, or the age basis. For example, a table built on age-nearest-birthday exposures cannot simply be re-labeled age-last-birthday and used as if the ages were identical. Observed claims experience is also conditioned by policyholder behavior and the continuation of contracts. An increase in lapses can change the risk composition of the remaining insured population and thereby affect observed insured mortality.

From individual expected value to portfolio risk

For N statistically independent and identically distributed policies with present values Z_i, the total S_N=Σ_i Z_i has mean N E[Z_1] and variance N Var(Z_1). Its coefficient of variation is reduced by a factor N^{-1/2} if the mean is positive and the individual distribution has finite variance. That scaling describes diversification of independent idiosyncratic risk. Under shared mortality improvement uncertainty, a common catastrophe, or interest-rate dependence, covariance terms remain and may prevent concentration risk from decreasing at the same rate. Financial reporting and solvency requirements respond to particular purposes and cannot be derived merely by attaching a universal loading to the individual expected value.

The actuarial literature's modern treatment of stochastic lifetime processes and dependent contingent values is surveyed structurally in Dickson, Hardy, and Waters (2020), whose life-contingency textbook includes advanced survival models, embedded options, and stochastic longevity models. The publisher record establishes the edition and scope; the present derivations are independently reconstructed rather than presented as quotations from inaccessible pages.

Assumptions and Equivalence

Mortality data, estimation, and projection

Selecting mortality for valuation begins with defining the reference population, whether general population, insured lives, pensioners, annuitants, or a medically underwritten subpopulation. The estimator must be appropriate to the exposure metric, with clear distinctions between deaths and exposure-to-risk, age definitions, cohort effects, and calendar-period effects. The Society of Actuaries maintains individual life mortality experience studies that document changes in experience sources and publication years. As of the 2026 materials shown in its current research index, the page includes a September 2026 mortality improvement scale for specified U.S. valuation use. That availability neither supplies an automatically universal assumption nor erases the need to match the scale's applicability to a particular contract and valuation rule.

Period mortality and cohort mortality answer different questions. A period table assembles age-specific rates prevailing in a selected calendar period. A cohort projection follows an actual or hypothetical birth cohort as calendar time advances. Mortality improvement factors can convert a base-table rate q_{x,0} to a projected q_{x,t} under specified age- and time-specific assumptions; for example q_{x,t}=q_{x,0}∏{s=1}^t(1−m{x,s}) when m values are defined as multiplicative annual reductions and the resulting probabilities remain in [0,1]. The product is a modeling convention requiring coherent rates. Lee and Carter (1992) provided a stochastic mortality forecast framework using age-specific effects and an aggregate mortality index; uncertainty in future mortality dynamics must be distinguished from the random death outcome of one insured individual conditional on a fixed table.

Mortality assumptions should be challenged against data quality, credibility, selection duration, exceptional shocks, and the purpose of the valuation. The Actuarial Standards Board's ASOP No. 23 on Data Quality addresses selecting, reviewing, using, and communicating about actuarial data. Its ASOP No. 25 on Credibility Procedures describes methods for using subject experience and other relevant experience; a chosen blending weight is a professional statistical decision and cannot substitute for evaluating population comparability. Neither U.S. standard is automatically a binding professional requirement in every jurisdiction. They are used here as explicit examples of published practice guidance.

Equivalence for a simplified level-premium contract

The elementary equivalence principle selects a premium whose expected present value equals the expected present value of the stipulated benefits under a defined basis. Let P be a level premium payable annually in advance at durations 0,...,n−1 while the insured is alive and coverage remains in force. Ignoring expenses, taxes, profit loadings, lapse, and other contract cash flows, the benefit-equivalence equation is

Pa¨x:n‾∣=EPV⁡(Zbenefit),P=EPV⁡(Zbenefit)a¨x:n‾∣.P\ddot a_{x:\overline n|}=\operatorname{EPV}(Z_{\mathrm{benefit}}),\qquad P=\frac{\operatorname{EPV}(Z_{\mathrm{benefit}})}{\ddot a_{x:\overline n|}}.

The denominator is the expected present value of a unit premium stream, not its guaranteed nominal count. A benefit value of 50,000 and a due-annuity factor of 2.5 produces a net annual premium of 20,000 in the simplified mathematical example, regardless of commercial expense or capital charges. The calculation does not determine a retail premium or insurer reserve. It is an identity within a limited model. If premiums are paid in arrears, at monthly intervals, for fewer years than coverage, or waived after disability, the expected premium stream changes.

The prospective reserve at duration t under a specified state and basis is a conditional expected present value of future benefit and expense outflows minus future premium inflows, using the valuation assumptions applicable at t. This is conceptually different from a premium established at issue: assumptions may have changed, the contract may have accumulated options and guarantees, and reporting rules can require distinct margins or risk adjustments. A premium computed by net equivalence at issue is not a comprehensive recognition or measurement model for a current insurance liability.

Multiple valuation purposes in international institutions

An elementary actuarial present value measures a specifically defined discounted contingent cash flow. IFRS 17 uses a more comprehensive measurement of groups of insurance contracts: the IFRS Foundation's IFRS 17 overview describes estimates of future cash flows, time value and financial risk, a risk adjustment for non-financial risk, and a contractual service margin. Its effective date for annual reporting periods beginning on or after January 1, 2023, is a matter of accounting-standard adoption; national implementation and company-specific accounting judgments require separate confirmation. The basic life insurance formula is a building block for projected claims, and it is not the complete IFRS 17 liability. The International Actuarial Association's ISAP 4 on IFRS 17 is an international model standard relevant to actuarial work in that domain, subject to how a jurisdiction or professional association adopts or references it.

European Solvency II's Article 77 best-estimate rule describes probability-weighted future cash flows discounted using the relevant risk-free term structure and adds a risk margin in technical provisions. This is an institutional valuation framework, not a universal formula for the premium of an individual term policy. The applicable law, amendments, reporting date, and any permitted adjustments must be checked for the actual undertaking. In the United States, the NAIC's principle-based reserving description and 2026 Valuation Manual index locate life insurance reserve requirements within state regulatory arrangements that include prescribed requirements and scenario-based calculations. The manual's VM-20 provisions include assumptions and methods beyond a standalone one-life expected benefit value.

Comparative interpretation therefore requires naming the estimand: a net single benefit premium, a commercial product price, a current liability measured under accounting rules, and a solvency technical provision can all legitimately differ because they incorporate different payment sets, recognition boundaries, assumptions, and measurement requirements. Financial and demographic inputs may have superficially similar units while playing different regulatory roles. Treating the four values as interchangeable obscures the actual economic and legal obligations they address.

Model and method validation

Independent verification should examine mortality probabilities between zero and one, nonincreasing survival values, matching date and currency units, positive discount factors, benefit-state exclusivity, and preservation of identity relations. For actuarial models used in U.S. professional contexts, the Board's ASOP No. 56 on Modeling emphasizes model structure, purpose, assumptions, and validation; the revised ASOP No. 7 cash-flow analysis is effective for actuarial work performed on or after June 1, 2026. These standards speak to professional applications and do not turn every pedagogical formula into a mandated regulatory cash-flow engine. An adequate research check therefore combines derivation, independently reproduced arithmetic, numerical boundary tests, and explicitly qualified comparison with institutional sources.

Sensitivity and Illustrations

Reproducible three-year mortality and endowment calculation

Consider a hypothetical insured life at age x, under three annual conditional death probabilities q_x=0.01, q_{x+1}=0.02, and q_{x+2}=0.03. Set effective annual interest to i=0.04, hence v=1/1.04. The policy pays 100,000 at the end of the year of death if death occurs in the next three years and 50,000 at the end of year three if the life survives. All amounts are in one arbitrary currency unit. There are no premiums, administrative expenses, lapses, taxes, or payments after year three.

Survival to the first, second, and third anniversaries is respectively 0.990000, 0.970200, and 0.941094. The unconditional probabilities of death in years one, two, and three are 0.010000, 0.019800, and 0.029106. Their sum, 0.058906, and terminal survival 0.941094 add to one. The values can be independently reconstructed from the following complete distribution:

Outcome Probability Payment time Nominal benefit
Death in first policy year 0.010000 1 100,000
Death in second policy year 0.019800 2 100,000
Death in third policy year 0.029106 3 100,000
Survival through maturity 0.941094 3 50,000

The term death benefit has expected present value

100000(0.011.04+0.01981.042+0.0291061.043)=5,379.672565.100000\left(\frac{0.01}{1.04}+\frac{0.0198}{1.04^2}+\frac{0.029106}{1.04^3}\right)=5{,}379.672565.

The survival benefit has expected present value 50,000 × 0.941094 / (1.04)^3 = 41,831.456958. Adding the disjoint branches yields the combined benefit expected present value of 47,211.129523. At time zero a single net benefit premium under this deliberately restricted basis would equal that sum. The result follows directly from the specified numerical data; it is not quoted from a mortality study or presented as a market quote.

For the combined benefit, the second raw moment is

E[ZE2]=1000002∑k=02v2(k+1)kpxqx+k+500002v63px=2,351,136,442.678391.E[Z_E^2]=100000^2\sum_{k=0}^{2}v^{2(k+1)}{}_kp_xq_{x+k}+50000^2v^6{}_3p_x=2{,}351{,}136{,}442.678391.

Thus Var(Z_E)=122,245,691.820753 and its standard deviation is 11,056.477369 in the same currency units as the present value. These quantities describe single-contract benefit randomness given fixed mortality and interest assumptions; they exclude uncertainty about the assumptions themselves. Independent implementation checks should reproduce both E[Z] and E[Z²] from the four mutually exclusive states.

If the same benefit is financed with premiums P paid in advance at times zero, one, and two while the insured is alive, the associated temporary annuity-due factor is 1+0.99/1.04+0.9702/(1.04)^2 = 2.848927515. Net equivalence gives P=47,211.129523/2.848927515=16,571.544653 per annual premium. The equality is restricted to the stipulated stream of benefits; an insurer cannot treat it as a complete gross premium without adding all contract cash flows and any chosen commercial loadings.

Interest-rate sensitivity under unchanged mortality

Holding the mortality assumptions and all nominal benefits fixed, substitute a flat annual effective interest rate of 3%, 4%, or 5%. This isolates discounting effects while preserving the contractual timing and death probabilities.

Annual rate Death-benefit EPV Survival-benefit EPV Combined EPV
3% 5,500.825000 43,061.716238 48,562.541239
4% 5,379.672565 41,831.456958 47,211.129523
5% 5,262.585034 40,647.619048 45,910.204082

The combined benefit value decreases as the interest rate increases because all positive cash flows are discounted over positive future durations. For each term v^k B p, the derivative with respect to i is −k B p (1+i)^{−k−1}; the negative derivative follows for i>−1 whenever the relevant benefit and probability are nonnegative. A duration-weighted sensitivity can be defined by the derivative of the expected present value, but any resulting summary depends on the entire future cash-flow distribution. For variable yield curves, sensitivity is a vector or function of curve-node changes rather than a unique response to a single uniform interest rate.

Mortality sensitivity and the changing balance of benefits

Now hold the annual effective discount rate at 4% and uniformly scale each of the three one-year mortality probabilities by 0.75 or 1.25. These are stylized instantaneous assumption stresses, not empirical mortality-improvement forecasts. Each stressed q remains in [0,1], and survival probabilities are recomputed multiplicatively rather than multiplying the final three-year death probability by the same factor.

Mortality multiplier Death-benefit EPV Survival-benefit EPV Combined EPV
0.75 4,053.048302 42,476.966938 46,530.015240
1.00 5,379.672565 41,831.456958 47,211.129523
1.25 6,694.184252 41,191.958817 47,886.143069

The death-benefit component rises with the assumed mortality intensity while the survival component falls. The combined result increases because the earlier death benefit is twice the maturity benefit in this particular contract. If the benefit levels and payment conventions changed, so could the sign of the total sensitivity. The stress also demonstrates why mortality effects cannot be assessed by looking only at the last-year survival probability: a change in q_x alters all later survival-contingent quantities.

Analytical special cases and reproducibility checks

Several identities provide independent tests without external numerical data. At zero interest an n-year endowment with equal death and maturity benefit B has present value exactly B if the two events exhaust all outcomes and the benefit is paid once. For the present unequal-benefit example at i=0, expected value is 100,000 × 0.058906 + 50,000 × 0.941094 = 52,945.30. If every annual q is zero, the death component vanishes and the sole benefit is the discounted maturity payment. If q_x=1, death occurs in the first covered year and the expected benefit value is B_D/(1+i) under the year-end convention. These limiting calculations test event allocation and the power of the discount factor.

For the whole-life annual identity, simulate or enumerate a fixed death year K, verify v^{K+1}+dΣ_{j=0}^{K}v^j=1, and only then average across a mortality distribution. For a life annuity, verify that its second moment includes covariance of survival indicators. For immediate death payment under constant μ and δ, compare numerical integration with the exact result Bμ/(μ+δ). Unit tests that reproduce only the central expected value may miss an off-by-one timing error shared by both the primary implementation and its comparison function; event-level tests are stronger.

Integrated interpretation, limitations, and research significance

The foregoing derivations support a direct interpretation: a life-contingent benefit value is the expectation of a contractual payment process translated into a specified valuation numeraire and date. Its mathematics is exact relative to the chosen lifetime distribution, interest convention, and contract representation; its use in practice depends on statistical evidence, mortality projection, financial calibration, legal contract terms, and the reporting purpose. The meaning of the number is preserved by documenting these inputs and the transitions between them. This is an analytical synthesis of established methods, not a newly named theory or principle.

The study does not estimate population parameters, fit a hazard model to raw experience data, compare product profitability across insurers, construct a stochastic yield-curve model, or independently validate a regulator's technical provision for a real company. The experimental inputs are explicitly hypothetical. Institution-specific valuation rules are discussed at the level supported by their official publications; actual jurisdictional applicability and contractual details require further work when a real valuation is performed. Long-horizon uncertainties include changes in mortality trends, pandemic effects, migration, selection, health behavior, financial regimes, lapse, and benefit utilization. A single present value cannot report all of these uncertainties; the distributional and scenario framework must be chosen for the decision.

Conclusion

Life-contingent benefit valuation is a coherent application of probability, survival analysis, contract interpretation, and discounting to cash flows triggered by death or survival. Term insurance sums discounted death-event probabilities over a covered interval; whole-life insurance extends those probabilities across remaining life; pure endowments value survival to a fixed date; and endowment insurance combines disjoint death and survival outcomes. Life annuities aggregate repeated survival payments and therefore require attention to nested survival events in variance calculations. Their expected present values follow directly from the defined cash-flow states, while second moments, dependence, and sensitivity analysis explain properties hidden by the mean.

The numerical demonstration shows that even a simple three-year product can have different and opposing mortality sensitivities in its death and survival components, and that premium equivalence depends on the exact schedule of contingent premium payments. International institutional comparison confirms that elementary actuarial present value is a component of broader reporting and solvency measurement rather than a substitute for IFRS 17 fulfilment cash flows, European Solvency II technical provisions, or U.S. principle-based reserving. Reliable research therefore records the underlying mortality population, projection period, benefit condition, payment date, discounting basis, contractual eligibility, uncertainty, and valuation purpose. The derivations and source-checked comparisons provide a finished scientific basis for subsequent specialized research into life pricing, pensions, insurance reserves, and stochastic asset-liability valuation.

References