RESEARCH DOSSIER
Mortality Tables and the Origins of Actuarial Valuation
Abstract. Mortality tables made human survival legible as an age-dependent probabilistic quantity and thereby enabled systematic valuation of promises payable upon death or continued life. This monograph reconstructs the connected but historically distinct developments of mortality observation, population inference, life-contingency mathematics, annuity financing, life-assurance pricing, and prospective valuation of an insurer's liabilities. It examines London Bills of Mortality and John Graunt's 1662 analysis; the Dutch life-annuity problems addressed by Jan de Witt and Johannes Hudde in 1671; Edmond Halley's Breslau analysis of 1693; mathematical elaborations by Abraham de Moivre and Thomas Simpson; Dutch, French, German, and Swedish demographic contributions; Richard Price's work on reversionary payments; and the establishment of sustained actuarial practice at the Equitable Society, particularly William Morgan's valuation as of January 1776. The study reconstructs the cash-flow identity that connects survival probabilities and interest to annuity, assurance, joint-life, and reserve values, then exposes the assumptions needed to translate recorded deaths into financial prices. Original analytical synthesis demonstrates that the history is best understood as the convergence of three forms of credibility: credibility of observations, credibility of calculations, and credibility of the institution obligated to pay. That interpretation is a historical argument grounded in documentary comparison, not a newly proposed mathematical theory or doctrine. The monograph treats contested priority claims, deficient data, migration, selective populations, period-versus-cohort interpretation, the interaction of mortality and interest, and the danger of reading modern actuarial terminology backward into early texts. It ends by distinguishing the historically verifiable achievement of actuarial valuation from claims of exact prediction and by identifying the durable epistemic demands carried into contemporary actuarial practice.
Keywords: Mortality tables; history of insurance; actuarial valuation; life contingencies; John Graunt; Edmond Halley; Jan de Witt; Johannes Hudde; Abraham de Moivre; Thomas Simpson; Richard Price; William Morgan; Equitable Life; annuities; demographic statistics; present value; life insurance reserves; historical source criticism.
Introduction
A life-contingent promise has two essential financial dimensions: an amount whose payment depends on survival or death, and a time when payment may become due. Its present value requires a distribution of those possible payment times and a rule for discounting them. Historical actors could sell annuities, promise funeral support, and insure lives before possessing a reliable, systematically constructed probability of death at each age. The emergence of actuarial valuation was a process in which existing contractual practices, empirical descriptions of populations, mathematical methods, and an institutional demand for accountability became connected.
The central scientific question is how a documentary count of births, burials, and ages became usable evidence for an uncertain financial obligation. The question has several answers because the transformation involved multiple distinct operations. Observers first had to classify records, specify a population, and infer the distribution of ages at death. Mathematicians had to attach a probability to survival over a term. Financial analysts needed to translate future contingent payments into present monetary values. Insurance organizations then had to apply the calculations repeatedly to a portfolio, monitor experience, and decide what assets should remain available against promised payments. The historical record provides evidence of these operations at different dates, places, and levels of maturity.
The chronology therefore begins well before Halley's celebrated 1693 work and continues well after its publication. The published record of Graunt's 1662 observations, the Dutch 1671 annuity calculations, and Halley's 1693 scientific paper document contributions of different kinds. Their order alone settles neither a universal inventor nor an unbroken chain of institutional adoption. The task is to identify precisely what each document establishes and what remains inferential.
This research dossier concerns mortality measurement and valuation rather than a general history of life insurance companies, private saving, public pensions, or insurance regulation. Those subjects enter only where they explain the demand for mortality information or the adoption of valuation methods. Throughout, “mortality table” denotes a specified schedule connecting age to survival and death, while “actuarial valuation” denotes the computation of monetary values of future uncertain benefits and contributions under identified assumptions. The terms are used analytically; their modern technical precision cannot be imputed to every eighteenth-century author.
Historical Questions, Sources, and Method
The study combines historical bibliography, selective inspection of digitized early text, provenance-checked catalogs, the original scientific article and manuscript record, archival collection descriptions, and contemporary historical scholarship. Priority is given to documents whose dates and titles can be independently identified: Graunt's edition in the Wellcome catalog, the text of a later Graunt edition reproduced in C. H. Hull's historical collection, the Royal Society's Halley manuscript record and journal DOI, the National Library of Medicine catalog for de Moivre's 1725 volume, the Wellcome record of Simpson's 1742 book, institutional evidence concerning Equitable Life, and official Swedish demographic archives. This hierarchy distinguishes the existence and dating of a source from an independent verification of every numerical table it contains.
A historical claim is classified by its evidentiary reach. An identified original title page or archive catalog directly establishes an edition's existence, its date, and often its intended subject. Surviving original prose can establish an author's explicit method. A modern peer-reviewed interpretation is evidence for a historiographic argument and must be attributed as such. Retrospective institutional histories can corroborate chronology, although priority language such as “first” requires a clearly delimited comparison class. Mathematical identities can be checked by derivation without pretending they reproduce an unpublished notebook or missing calculation.
The source-language problem is consequential. Early English “bills,” “expectations,” and “years' purchase,” Dutch “lyfrenten,” and French “rentes viagères” developed in their own legal and economic settings. The English phrase “life expectancy” may refer to an arithmetic mean of years remaining, an expectation based on an assumed schedule, or a contemporary description of survival prospects. Those senses are separated here rather than silently merged. Public archives can establish that a paper or ledger survives; their existence does not imply that this study has inspected every folio in a restricted collection.
A notable recent contribution is Geoffrey Poitras's study of joint-life valuation (2025). Its examination of correspondence and attribution complicates the customary exclusive credit given to de Witt for the 1671 Dutch work. The article supplies a useful methodological benchmark: reconstruct the exchange of problems and calculations among correspondents, recognize missing letters, and avoid treating signature and sole mathematical authorship as equivalent facts. This monograph adopts that standard of attribution and documents remaining uncertainties rather than assigning unsupported singular priorities.
The research is a critical synthesis, not an exhaustive census of surviving European archives. No fresh archival transcription, manuscript collation of all editions, original eighteenth-century dataset, or modern demographic estimation has been presented as performed. The reproducible illustrations below are intentionally synthetic and explicitly separate derived mathematical implications from historical premium quotations.
Before the Life Table: Contracts, Registrations, and Political Arithmetic
Life-contingent finance existed as a practical arrangement before probability mathematics could offer its now-familiar notation. A life annuity exchanged a present sum for payments while a nominated person remained alive; the duration of payments was uncertain, but the contract could nevertheless be priced through convention, bargaining, political need, or broad beliefs about age. A death-contingent promise reverses the survival condition. Such contracts could be administratively meaningful without the modern differentiation of mortality intensity, survival probability, discount rate, and expense loading.
Annuity issuance also addressed a different problem from private life assurance. Early public issuers sought funds and incurred future annuity obligations, while life offices later had to calibrate premiums across applicants and match future benefit promises with assets. Their economic motivations affected what questions they asked: public finance could ask what an annuity ought to cost, while a life office had to ask whether premium income and accumulated assets would suffice for all future benefits and expenses. Comparable formulas eventually served both purposes, but the original institutional incentives were distinct.
The history of the Dutch life-annuity tradition establishes a seventeenth-century government-finance setting for explicitly mathematical debate. Poitras (2025) goes further by placing joint-life contracts in a longer northern European record and warning that earlier issuers sometimes displayed knowledge of nominee age even before a formal age-table pricing method existed. It follows that the assertion “early annuities ignored age” needs a date, place, issuing institution, and transactional source. A uniform story of ignorance followed by discovery would erase heterogeneity and surviving practices.
Parish registers supplied a different enabling condition. Burials and christenings were recorded for religious, civic, and local administration. Their subsequent use to reason about population converted records made for one institutional purpose into statistical material for another. This reuse was powerful because births and deaths recur, but hazardous because coverage, classification, place of residence, and completeness were not designed for actuarial inference. Political arithmetic developed in that gap: the population was made visible through numbers that had to be interpreted rather than merely added.
John Graunt and the London Bills of Mortality
John Graunt's 1662 Natural and Political Observations belongs to the documentary origin of quantitative mortality analysis. The Wellcome Collection identifies the original edition, while a later text can be studied in Hull's reproduction of Graunt's Observations. The distinction between those editions matters: the fifth edition of 1676 was enlarged, and the later printed text includes materials absent from the first version. Statements based on it must therefore be attributed to the reproduced edition when the difference may matter.
Graunt analyzed recurring reports of burials, christenings, causes of death, and differences between places or periods. The scientific act was one of abstraction from the reporting system. Rather than treating a weekly bill as a complete explanation of a recent local event, he asked what longer runs of numbers might reveal about persistence, seasonal variation, urban population, disease, and the approximate distribution of lives. He also confronted inconsistencies in labels and numbers. That work helped establish that mortality was patterned enough to warrant statistical reasoning despite individual uncertainty.
The source bears both promise and limits. London burial counts were counts of recorded events, not an unbiased census of all residents at risk. Plague, registration practices, changes in parish coverage, infant classification, and migration could affect the apparent relationship of deaths to population. Graunt's estimates of survival at successive ages relied on inference in a period when ages at death were not consistently reported in the way a modern life table requires. His achievement can be recognized without attributing to him an adequately observed cohort table or the complete mathematics of life-contingent present value.
Authorship itself has a documented historiographic history. The Wellcome entry records association with William Petty, and Hull explicitly distinguishes Graunt's work from some older attributions. This is a reason to cite the edition and the authorship evidence rather than treating a modern web transcription as if it were an untouched seventeenth-century file. By the same logic, Graunt's priority in statistical observation does not settle priority in annuity valuation: observations and contract prices are different intellectual objects.
The central conceptual transition was to regard mortality as a repeatable aggregate pattern. A person dies once, at an uncertain age. A set of sufficiently comparable persons produces frequencies that can be discussed over many ages. This transition made possible the actuarial idea that a financial obligation can be valued by weighting its possible payment times, provided the frequencies are suitable for the lives insured or subsidized. Whether the data actually meet that condition is a separate empirical question.
From Recorded Deaths to a Survival Schedule
A life table is an organized relation among the ages reached by a hypothetical or observed set of lives. In modern notation let l_x be the number alive at exact age x from an initial radix l_0. If d_x denotes deaths between ages x and x+1, then d_x = l_x - l_{x+1}; the conditional probability of death over that interval is q_x = d_x / l_x, and the one-year survival probability is p_x = 1 - q_x. Survival for k additional complete years is {}kp_x = l{x+k}/l_x when the table is internally consistent.
These identities are bookkeeping equalities once l_x is specified. They do not establish that any particular historical observations identify l_x. For a complete cohort, individuals can in principle be followed from birth to death; in a period life table, age-specific mortality observed across different cohorts in a period is assembled into a synthetic schedule. Graunt's retrospective burials, the Breslau age information used by Halley, annuitant lists, and national parish registers have distinct relationships to those constructions. Describing them all as a direct observation of one cohort obscures the inferential step.
A life table may be derived from age-specific deaths and reliable populations exposed at each age. If counts of deaths are available without matching exposures, additional assumptions about population stationarity, the distribution of births, migration, and changing mortality become necessary. A locality that receives many young adult migrants cannot be interpreted exactly as a closed population, even if its burial reporting is excellent. Likewise, births observed in one year and burials across many earlier birth cohorts cannot automatically be combined as if they pertained to one closed cohort.
The expected complete future lifetime at age x is often written e_x = integral from zero to infinity of {}_tp_x dt, provided the integral exists. Its discrete curtate analogue sums probabilities of surviving complete future years. These expectations summarize duration; neither alone supplies the present value of a stream when interest is positive and payment timing matters. A payment conditional on survival in year twenty has a different time value from an equal payment conditional on survival next year. This difference is fundamental to the transition from demographic description to actuarial finance.
Historical source criticism should ask four questions whenever a purported life table appears: what event generated the records, what population was exposed, how ages and periods were constructed, and what financial use the resulting schedule could support. Without those answers, a beautiful age column can convey false precision. The problem persists in modern experience studies, although data availability and estimation methods have transformed.
De Witt, Hudde, and the Dutch Annuity Problem
Jan de Witt's 1671 Waerdye van Lyf-renten naer proportie van Los-renten is a crucial document because it sought to value government life annuities in relation to redeemable debt. The original-edition holding described in NN Group's history of life annuities locates the treatise in The Hague in 1671. Its economic problem was an assessment of alternative forms of public borrowing, with mortality and the ages of annuitants affecting the cost borne by the issuer.
Poitras (2025), in early joint-life annuity valuation, reconstructs the surviving exchanges among de Witt, Johannes Hudde, and Christiaan Huygens. The paper records that some correspondence is missing and that the mathematical contribution conventionally credited solely to the signatory de Witt requires closer attention to Hudde's involvement. The appropriate conclusion is specific: a 1671 publication attributed to de Witt survives; collaboration and correspondence support a material role for Hudde; the exact partition of mathematical authorship cannot be fully recovered from the surviving documents. This qualified description strengthens the history by connecting intellectual credit to available evidence.
The Dutch problem is particularly revealing because the calculation was not merely an academic experiment. A government deciding between a payment lasting for the lifetime of a nominee and a redeemable debt instrument faced a real fiscal choice. Even a carefully calculated expected value could depend on the assumed mortality schedule and the rate of interest, and thus was open to policy pressure. The figure selected for public finance was an analytical conclusion under assumptions, not a natural constant of a life annuity.
Joint-life and last-survivor obligations generated further complexity. For a last-survivor annuity, payments continue until the final survivor dies; using one person's expected lifespan cannot determine the payment period. The de Witt–Hudde correspondence is historically relevant here because it discusses these multi-life commitments well before the mature algebraic notation of the eighteenth century. Poitras traces the problem through later geometric and approximation methods, showing that nineteenth- or twentieth-century actuarial formulas cannot simply be projected backward as literal seventeenth-century procedures.
The Dutch evidence therefore changes the usual textbook genealogy. A history beginning with Graunt and proceeding directly to Halley loses the already existing connection between probabilistic life valuation and state finance. The more defensible sequence recognizes parallel paths: English work on bills and population regularities, Dutch financing calculations on life-contingent contracts, and subsequent attempts to connect usable mortality schedules with repeatable annuity pricing.
Halley's Breslau Paper and Life-Contingency Pricing
Edmond Halley's 1693 paper, An Estimate of the Degrees of the Mortality of Mankind, is a central published demonstration of an age-dependent mortality schedule linked to annuity valuation. The Royal Society preserves a corresponding 1693 manuscript record, and the journal article is identifiable by its original publication DOI. Its title itself joins the measurement of mortality from Breslau births and funerals to an attempt to ascertain the price of life annuities.
The Breslau context matters. Its recorded information was suitable for an exercise in constructing age-dependent survival estimates, but that suitability involved assumptions about the observed town and the mapping from registered deaths to persons living at each age. An apparently stable distribution of deaths is not independent evidence that the population was stationary or closed. Halley's calculations therefore exemplified both the great insight of an explicit survival schedule and the limits of historical exposure measurement.
The decisive mathematical connection is straightforward. Suppose an annuity pays one unit at the end of each year while the purchaser survives. For someone age x, the expected discounted payment in year k is v^k {}_kp_x, where v=(1+i)^{-1} and i is the chosen effective annual interest rate. Summing over possible years produces a life-annuity value. The principle transforms a column of survival counts into a price schedule that can vary with the purchaser's age and the yield used for discounting. Halley's importance lies in demonstrating such a connection in a published systematic treatment, while retaining the different earlier Dutch contribution.
One historical hazard is to translate Halley's work into modern language so fluently that its observational setting disappears. The journal article did not possess contemporary census exposures, stochastic interest-rate calibration, current mortality improvement projections, or detailed insurer experience studies. Calling it an inaugural actuarial milestone is warranted in a defined documentary sense. Calling it the beginning of all risk pricing, the first contemplation of life expectancy, or proof that contemporary reserve methods were already present would extend beyond the cited source.
The broader scientific importance follows from the formulation of an actionable question: given recorded survival by age, what sum is reasonable to pay now for uncertain future benefits? By making the timing of mortality a calculable component of a financial transaction, Halley's work strengthened the bridge from observations to financial valuation. Whether practical sellers immediately changed prices is another historical proposition requiring transaction-level evidence.
The Mathematics of Single-Life Annuities
The mathematical representation of life-contingent values permits a precise reconstruction of the problem early writers were attempting to solve. Consider a notional cohort of persons who have just reached age x. Write l_x for the number surviving to that exact age under a specified mortality schedule. Let d_x=l_x-l_{x+1} denote deaths in the next year and q_x=d_x/l_x the conditional probability of dying before age x+1. Under the table, the conditional probability that an individual age x survives k complete years is {}kp_x=l{x+k}/l_x. The numerator and denominator must refer to consistent definitions of entry, age, calendar time, and population. These symbols are modern explanatory notation, and their use here does not assign modern notation to Graunt, de Witt, or Halley.
An immediate annuity of one unit payable at each year-end while the annuitant is alive has expected present value a_x = Σ_{k=1}^{ω-x} v^k {}kp_x, where v=1/(1+i) and ω is the limiting age of the chosen table. The corresponding annuity-due, with a payment at inception and thereafter at the start of each year while alive, has value ä_x = Σ{k=0}^{ω-x-1} v^k {}_kp_x. With identical assumptions and payment periods, ä_x=1+a_x for a whole-life annuity whose terminal boundary is handled consistently. Changing the commencement date, annual interest rate, payment frequency, guarantee period, or mortality basis alters the value even when the stated annual benefit remains the same.
The mathematical value is the sum of weighted, discounted payments. Each term combines an amount, an indicator that payment occurs, and a discount factor. A table of ages may therefore be used for more than describing how long people live; it becomes a computational instrument for allocating present capital to obligations due at unknown times. Halley's 1693 published estimate connects the two tasks in an identifiable historical case. The Dutch financing problem of 1671, described in the history of life annuities, shows that the problem of attaching a price to a promised stream of lifetime income also arose in public finance.
Expected present value is conditional on assumptions, not a guaranteed realization. In a portfolio of many similar but dependent lives, the average may differ materially from its expectation because of mortality shocks or shared demographic exposures. If a city experiences epidemic deaths, both the estimated distribution and the ability to meet contemporaneous claims may change. A mathematically consistent annuity price also depends on who selects the contract. Persons expecting longer lives have a stronger incentive to purchase life annuities, so a population table can understate the cost of the selected purchasing group. This conclusion follows from the direction of the benefit, not from asserting that eighteenth-century sellers always measured selection quantitatively.
An annuity factor alone cannot tell a government or society what premium to charge. The full transaction may require expenses, administrative loadings, investment risk, uncertainty margins, taxes, adverse selection allowances, and the legal form of the promise. Early actuarial history is more comprehensible when the calculated expected value and the market price are kept distinct. The calculated value supplies a comparable basis for discussion of fairness and solvency; the price also reflects institutional objectives and contractual realities.
The relationship between the interest rate and mortality schedule is especially important. A higher discount rate lowers the present value of later payments, everything else held constant. Lower assumed mortality at older ages tends to raise the expected duration and value of a life annuity. These effects interact because improvements in survival at distant durations receive more discounting than improvements near the valuation date. Historical authors often knew that both life duration and interest were relevant while lacking the data and conventions needed to standardize every component.
Insurance Payable on Death and Prospective Reserves
A life assurance benefit reverses the survival condition governing a life annuity. For an insurance policy paying one unit at the end of the policy year in which death occurs, the present value is A_x=Σ_{k=0}^{ω-x-1}v^{k+1} {}kp_x q{x+k}. A temporary k-year contract truncates the sum; a deferred contract changes the commencement of covered years. The benefit might also be payable immediately upon death, in which case timing within each year matters. Such differences produce genuine financial distinctions even when the death probability is unchanged.
An insurer issuing a level-premium whole-life contract receives future premiums as long as the policy remains in force and must pay the sum assured at death according to the policy conditions. Under a simplified net-premium convention that excludes expenses and other decrements, an equivalence premium P satisfies P ä_x = B A_x, where B is the sum assured. This is an equality of expected present values at issue, conditional on the same chosen table and interest assumptions; it does not imply that the actual premium of a historical institution equaled this quantity.
The mathematical origin of a prospective reserve lies in asking the same valuation question after the policy has run for some years. For a survivor then aged x+t, and under a continuing policy with future level net premiums, the prospective net liability is V_t=B A_{x+t}−P ä_{x+t}, using the future-death-benefit and future-premium factors appropriate to the contract. The time index matters. A reserve is a balance of prospective future obligations and receipts at a particular valuation date, while a premium is determined for the initial transaction. A portfolio valuation aggregates such contract-specific values and then applies the reporting, prudential, or institutional conventions in force.
In ordinary financial language, the insurer's position cannot be assessed by counting cash premiums alone. A recently issued contract may bring substantial cash but also an obligation extending over decades. Conversely, an old policy may have fewer expected future premiums and a relatively imminent sum assured. The need to distinguish earned cash from the financial burden of promises helps explain why historic life offices eventually required systematic valuations. The archives of the Institute and Faculty of Actuaries preserve material relevant to those practices, although an archive collection's existence is not itself a proof about every past reserve decision.
A retrospective reserve is another accounting representation: premiums accumulated with interest less the accumulated value of past benefit outgo, under specified assumptions and on an appropriate per-survivor basis. In an internally consistent idealized net-premium model, retrospective and prospective reserve expressions can agree. Historical financial statements, dividends, and distributions can depart from that ideal because of expenses, heterogeneous contracts, imperfect investments, dividends, and the organization's actual risk experience. Both calculations therefore need clearly defined financial and mortality bases before they can be interpreted.
The distinction between paying for the risk of a single life and valuing a portfolio of enforceable promises is fundamental to the history. It permits separate appraisal of mathematical innovations such as de Moivre's formulas, published demographic evidence such as Halley's table, and operational changes at Equitable under William Morgan. These contributions were cumulative in function even when authors and institutions differed sharply in chronology, aims, and available information.
Joint Lives, Last Survivors, and Reversions
Many historical annuities and estate arrangements depended on the lives of two or more people. A joint-life annuity can require both persons to survive; a last-survivor annuity continues while at least one survives. A reversionary benefit may begin only after one death and continue during the survival of another person. A single-life table is necessary for some calculations, but it may be insufficient if dependencies between lives matter.
For two lives X and Y, let S_X(k) and S_Y(k) denote their individual probabilities of surviving k years. If their future lifetimes are assumed independent, the probability that both survive is S_X(k)S_Y(k), and the probability that at least one survives is S_X(k)+S_Y(k)−S_X(k)S_Y(k). A joint-life annuity-immediate is then a sum of discounted probabilities that both persons remain alive at payment dates. A last-survivor annuity uses the probability that either remains alive. Independence is an explicit simplifying assumption, never an observed fact derived from two marginal life tables.
Shared households, geographic exposures, infectious disease, wars, and other common disturbances may correlate deaths. In such circumstances the product of marginal survival probabilities misrepresents joint survival. The direction and size of the error depend on the contract: for a benefit payable on the first death, positively associated lifetimes can have a different effect from a benefit payable only at the second death. This is why joint-life valuation has both mathematical and evidentiary challenges. A sophisticated formula does not compensate for a false dependence assumption.
Historical financial arrangements made these problems financially pressing. Widows' pensions, reversions, joint annuities, and inheritance arrangements all depended on sequences of mortality events, often within families or selected social strata. The survival distribution of working adults might be very different from that of a citywide cohort containing infants and the poor. As Poitras's historical reassessment demonstrates, the joint-life annuity tradition also changes how early intellectual credit is assigned: Dutch financial mathematics belongs within a larger body of valuation problems than a single-life annuity example alone would suggest.
An analytically accurate account of early life contingencies should therefore reconstruct each contract's trigger. Does payment continue while all specified people live, while one lives, or after one death? Is payment annual, immediately due upon an event, or deferred? What age and marital-status data were available? Which lives entered the institution's experience? The words annuity and insurance cover substantially different obligations unless the conditions are made explicit.
De Moivre, Simpson, and Computational Simplification
The eighteenth-century development of life-contingency mathematics involved an effort to reduce computational complexity. A mortality table provides the proportions that survive to successive ages, but every contract can require a large discounted sum. Calculation by hand was costly, and computational simplifications had practical institutional value. Abraham de Moivre's Annuities upon Lives (1725) stands within this tradition. His earlier Doctrine of Chances (1718) belongs to the development of probability more generally; distinguishing the titles and dates helps prevent a claim about his annuity treatise from being attributed to the wrong work.
One simplifying mortality convention commonly associated with de Moivre is a linear decline in surviving lives over a restricted age range, equivalent in an idealized case to a uniform distribution of future lifetimes before a limiting age. If l_x is proportional to ω−x, then the conditional one-year probability of death rises as the limiting age approaches. This yields tractable sums and approximations. Its conceptual appeal is transparency: it permits a closed-form or easily tabulated relationship between attained age and remaining contingent payment years. Its weakness is equally clear: real mortality does not follow a uniform distribution over all ages, and the empirical suitability of a linear survivorship assumption varies sharply by period, population, and age.
Thomas Simpson's 1742 treatise on annuities documents a further stage of systematic mathematical treatment of survivorship and reversions. The bibliographic dating of the work is independently supportable through the surviving published volume. Its existence and subject do not independently establish the correctness of every table or priority claim embedded in later retrospective accounts. The significance of such treatises is the expansion of calculational techniques available for an increasing variety of contracts.
One historical interpretive problem is to mistake elegant mathematics for reliable mortality forecasting. An exact formula under a simplifying distribution is exactly correct within that distribution. Whether the distribution describes purchasers of lifetime annuities is a different question. Mathematical simplification can aid reproducibility and provide common valuation conventions, but it also conceals errors if the assumptions disappear from view. The professional consequences of mathematical development depended on the ability of the institution to test assumptions and reconcile actual experience to predictions.
The body of work associated with de Moivre and Simpson is important for actuarial valuation because it made technical relationships portable across specific financial transactions. A method that can be communicated, checked, and reused has greater institutional value than a one-off approximation. Yet one should not infer from the publication of a formula that all contemporary insurers used it. Evidence of adoption requires surviving policy records, correspondence, premium tables, account books, or explicit statements by practitioners.
Continental Evidence: The Netherlands and France
The continental history enlarges the scale of the inquiry beyond London. The Dutch case connects annuity valuation to public finance and the obligations of political authorities. A life annuity bought from a public treasury exchanges immediate funds for contingent future expenditures, and therefore poses the same fundamental question of expected discounted liability that arises for an insurer. The surviving institutional account of Dutch life annuities in 1671 establishes a concrete setting for de Witt and Hudde's investigations. Recent historiography cautions that surviving documents and later attributions do not settle every question of who first formulated an intermediate method.
State-issued annuities differ in material ways from later mutual life assurance. Public capacity to raise taxes, political credibility, currency conditions, and choices of borrowing instruments affect the contractual transaction. A government might choose annuity terms because of fiscal urgency rather than because the offered value matches a modern fair-value calculation. The analytical use of a present-value comparison is to identify the implicit cost of funds and variation across annuitants; it does not establish that eighteenth-century public finance already operated under a modern balance-sheet standard.
France provides another line of demographic and actuarial investigation. Antoine Deparcieux's Essai sur les probabilités de la durée de la vie humaine appeared in 1746, including investigations into life durations and their application to annuities. The French National Institute for Demographic Studies preserves an institutional edition and history. Its scholarly republication makes the work more accessible while also identifying it as an eighteenth-century text interpreted through later editorial work. A modern catalog or institutional introduction can identify the original and its historical context; quotations about methods must correspond to the particular edition inspected.
The importance of Deparcieux's mortality investigations rests partly on attention to the empirical origin of survival figures. Annuity lives and members of religious communities may have different characteristics from general city populations. Comparing these observed groups raises questions about who was counted, how deaths were established, and whether continued observation or selection affected the apparent survival rate. Such questions are more consequential for valuation than a superficial contest over which nation produced the earliest printed table.
The continental examples demonstrate that the mathematical problem emerged at the intersection of administrative obligations, investable capital, and the availability of population records. A chronology that assigns a single center of invention would hide the diversity of observed populations and institutional motivations. A comparative reconstruction can instead specify which documentary contribution concerns data, which concerns a formula, which concerns the financial contract, and which concerns the durable organizational practice of valuation.
German Demography and Swedish Statistical Institutions
The scientific infrastructure of mortality tables developed alongside wider investigations into population size, births, deaths, and the regularities of civic life. Johann Peter Süssmilch's demographic writings in the German-speaking world treated population observation as a subject of systematic inquiry with theological and administrative dimensions. Jacqueline Hecht's study of Süssmilch reconstructs the significance and context of that intellectual project. Demographic regularity could be discussed independently of the immediate act of pricing a commercial life insurance policy.
Sweden developed a particularly influential institutional system for collecting demographic statistics. Statistics Sweden traces official statistical history to the establishment of Tabellverket in 1749; its historical overview describes that institutional beginning. The historical data have also been documented and made accessible through Tabellverket research resources. The institutional history is relevant because systematic registrations can extend the observation base beyond the city bills or a selected annuitant cohort.
National statistical systems do not automatically solve the actuarial question. Recorded annual deaths must still be connected to people at risk and analyzed by age, sex, period, and cohort. Definitions, coverage, territorial boundaries, and registration compliance can change. A large population table may possess stronger statistical precision yet remain unsuitable for the experience of a particular insurer without allowance for selection, socioeconomic composition, or temporal mortality change. The modern language of exposure and selection makes those issues explicit.
The German and Swedish examples also suggest that actuarial measurement benefited from investments whose primary objectives were administrative and intellectual. A state compiling demographic information may improve the feasibility of mortality comparisons used by multiple institutions; the practical use of these figures in a specific annuity market, however, still requires a documented chain of adoption. This distinction separates historical infrastructure from the narrower question of valuation technique.
International comparison should avoid treating historical countries as interchangeable statistical units. Cities had different burial customs and record coverage. Public offices had different incentives and administrative capacity. Migrations, epidemics, and conflicts altered populations in ways visible neither in a single year's figures nor in a modernized historical table stripped of its documentary context. The comparative conclusion is that demographic knowledge and life-contingency mathematics developed in mutually reinforcing but institutionally distinct settings.
Richard Price, Northampton, and the Choice of Mortality Basis
Richard Price's work constitutes a major chapter in the movement from mortality observation to financial and social applications of life-contingency mathematics. His Observations on Reversionary Payments, first published in 1771 and subsequently revised, dealt with annuities, survivorship, widows' provisions, and the management of contingent payments. A surviving 1771 edition catalog and a digitized 1783 edition illustrate why edition identification matters. An assertion found in a late revised edition cannot automatically be assigned to the book as first printed.
The Northampton table became influential in historical actuarial calculations and debates because it offered age-specific mortality assumptions that could be applied to annuities and assurance. An underlying methodological issue is whether recorded burials and a proposed population structure correctly describe survival conditional on reaching a specified age. If a table is inferred from a nonstationary town, assumptions about births, migration, and the relationship between recorded deaths and persons living can distort implied mortality. Historical tables must therefore be assessed for both arithmetic consistency and the demographic conditions of their construction.
A life office's choice between a town-based table and a table of selected assured or annuitant lives entails a measurable transfer of value. If the assumed chance of dying within a year exceeds the true chance for the insured portfolio, a given pure death-benefit premium can be excessive, all else equal; if the assumption understates longevity among pensioners, a life-annuity valuation can be deficient. The direction of the bias changes with the benefit trigger. No table can be labeled universally conservative without specifying the contractual cash flow and the populations compared.
Price's work also provides a bridge between technical argument and public policy. Financial provision for widows and families requires solvency across generations as well as an equitable division of contributions. The difference between a benefit's expected present value and the financial safety of a scheme becomes salient when an organization promises payments over decades. Evaluating the work historically requires separate attention to intended application, available records, and later institutional reinterpretations.
Numerical detail in original mortality tables is vulnerable to edition differences, transcription errors, age labels, and omitted definitions. This research does not reproduce Price's full Northampton column as if independently collated from every original impression. Instead, it identifies the relevant published work and analyzes the valuation question under modern explicit notation. This practice preserves the historical argument while avoiding spurious quantitative certainty.
Equitable Life and the Transition to Institutional Valuation
The Society for Equitable Assurances on Lives and Survivorships, commonly called Equitable Life, occupies an important place in the history of life assurance because it made systematic, age-sensitive contractual valuation an enduring institutional concern. Its survival in historical actuarial archives provides more than a ceremonial origin story: accounts, policies, actuarial correspondence, and later professional investigations can illuminate how calculations shaped premium administration, assets, liabilities, and distributions. The actuarial historical collections identify relevant holdings, though every specific business decision requires its own documentary evidence.
An organization selling long-duration contracts faces a timing problem at its inception. Early premiums may arrive rapidly, while many of the largest benefits remain payable years later. Comparing current cash with current claims therefore understates the obligations. Even if average expected deaths are estimated correctly, the institution may require reserves, asset management, attention to expenses, and mechanisms for reconciling the accumulation of funds with benefits already promised.
An equitable premium scheme depends on the insured's attained age, expected mortality, benefit amount, premium duration, and assumed interest accumulation. Administrative costs and risk margins also matter. The historical advance lies in institutional use of consistent valuation rules over a portfolio. A founder's intentions, a printed premium table, and an actual financial valuation are distinct types of evidence. For that reason, Equitable's role should be described by the observable development of business practice and the surviving accounts rather than by assigning every element of contemporary actuarial science to the society at its founding.
The mathematical program at Equitable also drew upon James Dodson (c. 1710–1757). The historical account of Morgan and Equitable describes Dodson's work in 1756 on age-rated life assurance premiums and anticipated cash flows, and identifies Equitable's 1762 establishment as related to that program. Dodson's role establishes an important link between the probability mathematics of de Moivre and the subsequent organizational practice of level-premium life assurance. A model of historical origins that passes directly from Halley to Morgan omits the transformation of mortality analysis into a proposed premium structure and the distinct development of an institution willing to apply it. The same account identifies Richard Price's advisory involvement from 1768 and his concern, by 1774, about the society's financial position. In that setting, the prospective valuation undertaken by Morgan responds to a preexisting governance question about what level of surplus, if any, could responsibly be distributed.
An institution can have a carefully calculated price at issue and still fail to maintain an adequate financial position. Interest yields may fall, longevity may exceed assumptions, expenses may grow, or the size and age composition of its insured population may shift. A valuation therefore requires a date, a specified portfolio, a liability basis, and an inventory of assets or resources supporting it. The transition from mathematical table-making to repeated whole-office valuation reflects the expansion of the unit of analysis from an individual payment stream to a corporate promise.
Historical descriptions often identify Equitable as the first life office associated with professional actuarial methods, but the definition of professional application needs care. Specialized calculation predates Equitable; the office did not create mortality itself; actuarial practice expanded through many people and institutions. The distinctive historically documented problem was the management of a durable portfolio of life-contingent obligations under changing conditions. Its significance becomes clearer when institutional routine is compared with the earlier isolated scholarly and public-finance calculations.
William Morgan and the Prospect of a Valued Balance Sheet
William Morgan (1750–1833) is closely associated with the development of practical actuarial valuation at Equitable. The University of St Andrews' biographical history of Morgan describes his appointment to the society in 1775 and the valuation of obligations as of January 1, 1776. The historical literature includes a dedicated 1933 actuarial biography, which provides a further identifiable scholarly source. These records support a bounded claim about a notable early whole-office valuation; they do not warrant a universal assertion that no public or private institution anywhere had previously estimated its liabilities.
A valuation requires selecting a basis for prospective claims and contributions and applying it to a membership whose ages, amounts insured, contract dates, and remaining payment obligations are known. It aggregates the future values of each relevant promise. It can then inform a comparison between estimated liabilities and supporting resources. Such an exercise demands an information system: the table of mortality, the actuarial calculation method, a register of contracts, a reliable policyholder census, and accounting data must agree on the same valuation date.
A key historical consequence is the distinction between premium equity and distributable surplus. A society receiving more premiums than it pays in claims during one year may still have accumulated an insufficient fund relative to its obligations. Equally, an actuarial valuation showing resources in excess of estimated liabilities may enable distributions only subject to governing rules and sensitivity to assumptions. In this sense the balance sheet must represent future uncertainty rather than merely the cash history of the preceding year.
Morgan's work demonstrates the practical requirements of such an approach. The documentary priority that can be defended concerns his role, timing, and distinctive application inside Equitable, not the invention of all expectation mathematics. The evidence needs careful dating: appointment in 1775 and a valuation measured as of the beginning of 1776 describe related events without being interchangeable dates. Recording the valuation's effective date is critical because every asset, survivor count, and future cash flow must be understood relative to it.
This episode also illustrates how technical expertise can become an institutional function. A valuation expert does not merely produce a number: the task involves testing whether assumptions, portfolio records, reserves, premiums, and investment expectations are mutually coherent. Once repeated, such work supports institutional memory, auditing, comparisons across periods, and a continuing obligation to explain financial judgments. The professional development of actuarial practice thus depended as much on stable records and governance as on advances in mathematical notation.
Actuarial Professionalization and International Diffusion
The emergence of a recognized actuarial profession occurred after many of the mathematical innovations described here. The Institute of Actuaries was founded in 1848, an anniversary chronology confirmed by the Institute and Faculty of Actuaries. The Faculty of Actuaries in Scotland followed in 1856. Professional organizations institutionalized communication, shared methods, training, and scrutiny of work already performed across life offices and financial institutions.
Professional identity is historically distinct from the earlier adjective actuarial and the individual carrying out insurance calculations. A job title, professional society, and formal valuation method can arise at different dates. The history of the profession should therefore examine not just founding charters but the development of examinations, professional journals, interoffice tables, standard notation, mortality investigations, and expectations of accountability. These mechanisms enabled comparability across practitioners and progressively refined the relationship between data and valuation judgments.
The development became transatlantic. The American actuarial profession's institutional history includes the founding of the Actuarial Society of America in 1889, recorded by the Society of Actuaries' history. Local insurance laws, mortality experience, capital markets, policy designs, and reporting requirements shaped subsequent adaptations. A formula imported from another jurisdiction does not by itself establish a locally reliable mortality table; professional diffusion requires both mathematical communication and empirical adjustment.
Over time, valuation came to encompass multiple dimensions of financial risk. Mortality remained central, but longevity trends, disability, lapse, policyholder options, investment returns, expense inflation, and catastrophic events affected many portfolios. Modern professional practice also requires distinctions among statutory reserving, financial reporting, economic capital, pricing and internal risk management. The historical subject of this dossier is the earlier foundation of life-contingency valuation; the later expansion demonstrates its consequences without implying that contemporary regulatory frameworks existed in the seventeenth century.
There is a historiographic temptation to make a professional founding date the culminating proof of science's maturity. Professional associations can improve standards and accountability, but their existence is not independent evidence that every member's work meets the same empirical or ethical quality. The credible historical development is the creation of institutions capable of examining, discussing, and revising the assumptions underlying life-contingent promises. On this reading, professionalization formalizes a practice whose strongest origins were distributed among record-keepers, mathematicians, public financiers, and insurance offices.
Source Criticism and Historical Priority
The expression “origin of actuarial valuation” can refer to at least five different things: the first observed age pattern of mortality; a mathematically coherent calculation for a life-contingent benefit; a public publication that made such methods available; practical pricing by an annuity issuer or insurance office; or systematic valuation of a continuing institution's outstanding obligations. Different documentary comparisons can therefore produce different “firsts.” An accurate account states which comparison is being made rather than treating all five accomplishments as one event.
Graunt's identifiable 1662 publication belongs principally to the systematic examination of the London mortality bills and the reconstruction of demographic patterns. De Witt and Hudde's work belongs to life-annuity valuation in a public-finance setting. Halley's 1693 contribution lies in a published systematic mortality analysis explicitly connected to annuity prices. De Moivre and Simpson belong to mathematical development and computational methods; Deparcieux to empirical and theoretical mortality research in France; Price to the elaboration and public application of life-contingency reasoning; Morgan to sustained office valuation. The sequence is interpretive, and individual contributions overlap rather than fitting into completely separate boxes.
Priority claims are particularly sensitive to the survival of documents. A first edition may be lost; a manuscript can precede publication; a technique can circulate in correspondence; a transaction may be undertaken without explicit mathematical proof. Poitras's 2025 joint-life annuity analysis demonstrates that the available Dutch record and its modern interpretation can change the allocation of conceptual credit. Responsible historical synthesis should preserve the distinction between “first surviving identified example,” “first published treatment in this comparison,” and “first institution documented to repeat a procedure.”
Bibliographic criticism also requires tracking editions. An eighteenth-century volume may be reissued with tables, appendices, corrected calculations, or new arguments. A later reproduced text of Graunt and a first-edition catalog are related yet separate evidentiary objects. Similarly, a modern institutional summary of Deparcieux is not his original title page, and a biographical discussion of Morgan is not the actual valuation ledger. The references below make those access boundaries visible.
Three additional historical hazards deserve attention. Retrospective sources often attach prestigious titles to earlier writers that were not the actors' own. Modern nation-states may be projected onto jurisdictions with different political boundaries. Finally, public praise of a scientific contribution may be interpreted as evidence of commercial adoption despite a lack of policy records. The evidentiary solution is to narrow each substantive assertion to its source type and resist extrapolation beyond the documented transactions and writings.
The resulting chronology is genuinely plural and intellectually coherent. Life-contingency valuation was assembled from observations, calculations, and durable institutional administration. This conclusion does not need an invented single founder. It explains why separate historical traditions can each make defensible contributions without contradicting the mathematical structure of the present-value problem.
Data Quality, Selection, and Population Change
Mortality tables are constructed from observations through explicit assumptions about what the observations represent. A death count alone is not a mortality rate. The rate or conditional probability requires a denominator, which may be a measured exposure, a cohort entering an age interval, or a population estimate based on further assumptions. Parish burials, published annual bills, modern census counts, and life office records differ in the populations covered and the way deaths and ages are documented.
Age heaping is one common source of error. When persons report or record ages with a preference for round numbers, deaths or survivors may appear concentrated at ages ending in zero or five. Age misstatement at older ages can exaggerate or diminish apparent survival. Delayed registration, missing burials, unrecognized infant deaths, incomplete geographic coverage, and classification changes generate other distortions. These are not merely defects in historical description: they can directly affect valuations of benefits payable at particular ages.
Population movement also matters. In an open city, the distribution of deaths by age depends both on the mortality of residents and on in-migration and out-migration. A prosperous town may attract working-age adults; epidemics may alter both migration and recorded deaths; births may fluctuate substantially from decade to decade. An inference from deaths at age to the number alive at that age is therefore unreliable without assumptions about the population's stationarity and composition. The analytical limitation applies to the early London and Breslau contexts in different empirical ways.
A period mortality table uses conditions observed during a specified calendar interval to describe age-specific mortality. A cohort table follows or projects mortality as a birth cohort ages, potentially allowing mortality to improve or deteriorate over future calendar years. Historical published tables were often based on observations made within a period and applied to future financial promises. Their apparent accuracy depends on whether future conditions resemble the measured period. A contemporary actuarial application to pensioners may require explicit future longevity improvements, a problem absent from some early fixed-table formulations.
Selection alters the interpretation of every inferred survival curve. A person purchasing a life annuity may possess health information or longevity expectations different from the general population. A life assured under medical underwriting may initially have mortality below that of an unselected population. The same historical rate can therefore have opposite valuation implications for a life-annuity seller and a death-benefit insurer, depending on differences between the table and the covered population. Even within one organization, newly selected entrants and long-duration members may not share the same mortality profile.
Statistical reliability also varies by sample size and age. At advanced ages, sparse exposure can produce unstable observed rates. Some method of combining, smoothing, or graduating ages can help produce a coherent table; modern work on mortality graduation methods illustrates the continuing methodological concern. Smoothing, however, introduces a model and constraints that should be documented and tested against data rather than treated as a replacement for evidence. An apparently smooth curve can still systematically misestimate the survival of a subgroup.
A reproducible mortality study should disclose the observation period, territory, inclusion rules, age definitions, deaths, exposure or reconstruction method, adjustment for migration where relevant, treatment of sparse cells, assumptions for open-ended ages, and a source-to-table transformation trail. A historical reconstruction may not be able to recover every item. In that case, limitations should be stated and the resulting precision constrained accordingly. This is a central lesson of the early actuarial record: financial precision in a spreadsheet cannot exceed the credibility of the mortality basis and promised cash flows on which it relies.
Sensitivity Analysis and Reproducible Illustrations
The following calculations are deliberately hypothetical. They are provided to make the mathematical distinction between observed survival, contract timing, premium equivalence, and valuation reproducible. None of the inputs is represented as an eighteenth-century mortality table or the financial accounts of a named institution. Consider 10,000 persons alive at exact age 60; 9,900 at age 61; 9,750 at age 62; and 9,500 at age 63. Assume each annual death occurs for benefit-payment purposes at the end of the corresponding policy year and that effective annual interest is 4 percent. These conventions define a three-year illustration; the years after age 63 are intentionally outside its scope.
| Exact age | Notional lives surviving | Deaths in following year | Conditional one-year death probability |
|---|---|---|---|
| 60 | 10,000 | 100 | 0.010000 |
| 61 | 9,900 | 150 | 0.015152 |
| 62 | 9,750 | 250 | 0.025641 |
| 63 | 9,500 | — | — |
The probability, assessed at age 60, of dying in the first three policy years is respectively 0.010, 0.015, and 0.025. Notice that the year-two and year-three unconditional probabilities are not equal to q_61 and q_62: each includes the probability of surviving to the beginning of that year. The three-year probability of surviving all the way to age 63 is 0.950; 0.050 die within the modeled term. These probabilities sum to one, providing an immediate internal reconciliation.
For an annuity paying one unit at each year-end if alive, a_60:3 = 0.990/1.04 + 0.975/(1.04)^2 + 0.950/(1.04)^3 = 2.697912, rounded to six decimal places. For a three-year temporary insurance paying one unit at the end of the year of death, A_60:3 = 0.010/1.04 + 0.015/(1.04)^2 + 0.025/(1.04)^3 = 0.045709. The former value is dominated by survival over the payment years, whereas the latter is a discounted weighted probability of death over those same years. They describe different contractual events, and neither can be substituted for the other simply because both use the same table.
If level premiums for this temporary insurance are payable at the beginnings of years one, two, and three while the insured remains alive, the premium-annuity factor is ä_60:3 = 1 + 0.990/1.04 + 0.975/(1.04)^2 = 2.853365. The net equivalence premium for a one-unit death benefit is consequently 0.045708636777 / 2.853365384615 = 0.016019 per unit of sum assured per year, with the exact numerator and denominator evaluated before rounding. An actual market premium may exceed this amount because the illustration excludes expenses, acquisition costs, profit, prudential provisions, capital costs, and other contract terms.
A simple interest-rate sensitivity can be checked without a historical table. For a positive future annuity payment, each discounted component has derivative −k(1+i)^{−k−1} {}_kp_x with respect to i, so the derivative of the finite annuity factor is negative. If the payment probability is zero the corresponding term contributes nothing; otherwise increasing interest reduces its value. Mortality improvements generally increase the survival-weighted annuity value if all relevant future survival probabilities rise, a statement requiring no assumption that every death-benefit contract moves in the same direction by the same amount.
A valuation sensitivity should vary the assumptions that drive the particular portfolio, not a generic isolated rate. A small interest-rate change can have a large effect on long-duration annuities because distant payments receive substantial weights; a change in mortality at young ages may be decisive for term death protection; improvement at old ages can materially increase pension liabilities. Dependencies and selection can further affect portfolio tail risk without changing the mathematical expectation calculated from a simple marginal table.
The example shows why the documentary sources must be read with care. A historical table might give integer deaths or survivors without an explicit measure of exposed lives, or might describe a different age interval. Applying the above modern formulas mechanically would then generate numbers whose rounding is exact relative to inputs yet whose historical interpretation is unreliable. Reproducible arithmetic and credible empirical inference are complementary requirements; the first never certifies the second.
Interpretation: The Three Conditions of Valuation
The comparative historical record supports an analytical conclusion about the circumstances under which life-contingency calculations become usable valuations. First, the population evidence must plausibly describe the lives to which the contract applies. Second, the calculation must connect that evidence with specified benefit triggers, time intervals, and financial discounting. Third, an institution must be able to maintain records and honor the resulting contractual obligations over time. These are conditions that can be extracted from the documented cases, not a newly established doctrine or a proposed universal actuarial model.
The first condition is visible in Graunt, Halley, Deparcieux, and the growth of administrative statistics. Their work illustrates differing ways to organize records and infer survival. The importance of a population denominator, the treatment of age and time, and the choice of observed group are as historically significant as the printed mortality column itself. The second condition appears especially clearly in the Dutch annuity calculations and Halley's demonstration that the future survival of an annuitant could be combined with discounting to establish a present price. Subsequent treatises made the algebra and computation more systematic.
The third condition is highlighted by Equitable's institutional use of valuation and Morgan's work. A quoted price at issue addresses one transaction; continuing assurance demands a view of all remaining obligations under the contracts already issued. The organization must know who is alive, which benefits are still payable, when premiums remain due, what investments support the liabilities, and which assumptions make that comparison possible. The technical and administrative tasks are mutually dependent.
These conditions also clarify historical priority. A mathematician can make a decisive contribution to valuation before a professional office exists. A government can issue annuities before it can infer age-specific mortality from rich demographic records. A firm can collect many premiums while lacking a sufficiently rigorous assessment of long-term liabilities. The history is an accumulation of distinct capacities whose combination creates a durable practice.
This interpretation is consistent with the visible source sequence rather than being a claim of newly discovered archival facts. Graunt's identified 1662 publication, the Dutch annuity valuation history, Halley's 1693 paper, and Morgan's documented Equitable role anchor the synthesis. The resulting explanatory pattern is valuable precisely because it does not assert a fictitious universal moment in which statistical, mathematical, and institutional knowledge all appeared simultaneously.
The conditions carry a further implication for economic history. Governments, mutual societies, and private offices might adopt formally comparable life-contingency calculations while valuing different objects under different obligations and political constraints. Calculation is therefore embedded in law, administration, and the credibility of the promisor. The historical emergence of actuarial valuation is also a history of making future financial promises inspectable in the present.
Present-Day Significance and Limits of Historical Transfer
The foundational equations of expected present value continue to express a common component of contemporary life insurance and annuity valuations. The definition of the covered population, the time origin of the calculation, the conditions under which a claim arises, and the interest or discount basis remain indispensable. This mathematical continuity explains why the early contributions can be read productively in modern actuarial notation. It does not make a seventeenth-century mortality table a contemporary risk model.
Contemporary actuaries have access to larger and more complex administrative and insurance datasets, population registers, cause-of-death classifications, and stochastic methods. Those resources create analytical choices: periods and cohorts, selection adjustments, experience investigations, longevity improvement, dependence among lives, and the treatment of model uncertainty. The continuing mortality graduation literature is one example of modern technical effort to translate irregular observed rates into coherent age-dependent schedules. The fundamental obligation remains to identify which aspects of the resulting schedule are observed and which are modeled.
Financial valuation has also become more explicitly purpose-dependent. A quotation to a prospective customer, a statutory reserve, a financial statement, an internal economic valuation, and capital adequacy analysis can employ different conventions under different rules. Their values need not coincide. This variety reinforces the lesson of historical source criticism: a number called the “value” of a life-contingent obligation is interpretable only in relation to a date, object, cash-flow definition, and assumptions.
The direct lesson for the reading of history is methodological. A modern researcher should be able to reproduce a calculation, but reproduction begins by selecting the correct edition, identifying the observed population, and preserving the original limitations of the records. Mathematical recasting can explain a historical author's insight while remaining explicit about where the modern representation exceeds the documentary source. This is especially important when claiming precedence for a formula, a valuation, or the profession.
The direct lesson for financial institutions is one of accountability. Life-contingent promises may remain enforceable long after the assumptions governing their original price have changed. An organization that records only premium income or current-year claims cannot assess the present burden of those promises. Its management requires repeated valuation, careful assumption selection, credible data, controls over distributions and investment, and a clear distinction between expectation and realized outcomes. In this way the historical development speaks to the durable requirements of responsible risk finance.
Conclusion
The origins of actuarial valuation are located in a sequence of documented inquiries and practices rather than a single instantaneous invention. John Graunt's examination of mortality records helped establish the empirical problem. Dutch investigations by de Witt and Hudde showed the significance of life-contingent prices for public finance. Edmond Halley's Breslau analysis made age-dependent mortality and the valuation of annuities part of a published systematic scientific argument. De Moivre, Simpson, Deparcieux, Süssmilch, and later demographic statisticians expanded the mathematical and evidentiary resources available for calculation. Richard Price connected survivorship and financial provision to broader schemes of reversionary benefits. Equitable and William Morgan made the evaluation of a continuing insurer's obligations a distinctive institutional practice.
The evidence establishes several precise distinctions. Mortality observation is an empirical task. A survival schedule is an inference from observations under demographic assumptions. Present-value calculation is a mathematical relationship between survival, benefit conditions, time, and discounting. A charged premium also reflects an institution's costs and market or public objectives. A reserve is measured at a specific date for obligations that remain outstanding. Professionalization made these practices more communicable and subject to collective scrutiny, but developed after the major early mathematical and institutional milestones.
The original analytical conclusion is that credibility in actuarial valuation depends jointly on the fitness of mortality evidence, transparency of the computation, and organizational capacity to apply the result to real promises. Each part can be historically present without the others. This explains why different places and periods made indispensable contributions and why a mathematically exact premium can still be economically misleading. It also explains why practical actuarial valuation required continuing records and periodic reassessment of liabilities, not merely a printed table of lives.
The limits are explicit. Some early works survive chiefly through digitized editions and bibliographic descriptions; archive holding records do not substitute for reading every original folio; accounts of the earliest actuarial priority remain open to revision as surviving Dutch correspondence and other archives are reassessed. The hypothetical calculations presented here are modern explanatory reconstructions, not reported transactions from the seventeenth or eighteenth centuries. Within those evidentiary limits, the historical sequence demonstrates the emergence of a durable capacity to assign inspectable present financial values to uncertain life-contingent obligations.
References
De Moivre, Abraham. (1718). The Doctrine of Chances. London. Mathematical Association of America historical bibliographic presentation: https://old.maa.org/press/periodicals/convergence/mathematical-treasure-abraham-de-moivres-doctrine-of-chances.
De Moivre, Abraham. (1725). Annuities upon Lives: Or, the Valuation of Annuities upon Any Number of Lives; as Also of Reversions. London. U.S. National Library of Medicine historical catalog: https://www.ncbi.nlm.nih.gov/nlmcatalog/192031.
Deparcieux, Antoine. (1746). Essai sur les probabilités de la durée de la vie humaine. Paris. Modern scholarly edition and editorial information: Institut national d'études démographiques, https://www.ined.fr/fr/publications/editions/classiques/essai-sur-les-probabilites-de-la-duree-de-la-vie-humaine/.
Graunt, John. (1662). Natural and Political Observations Mentioned in a Following Index, and Made upon the Bills of Mortality. London: Thomas Roycroft. First-edition bibliographic record, Wellcome Collection: https://wellcomecollection.org/works/ear22f7e.
Graunt, John. (1676 [fifth edition]). Natural and Political Observations Made upon the Bills of Mortality. In Charles Henry Hull's later historical edition of The Economic Writings of Sir William Petty, online text, https://oll-resources.s3.us-east-2.amazonaws.com/oll3/store/titles/1680/Petty_0605-02.html. Textual witness to a later edition, distinguished from the first edition.
Halley, Edmond. (1693). An Estimate of the Degrees of the Mortality of Mankind, Drawn from Curious Tables of the Births and Funerals at the City of Breslaw; with an Attempt to Ascertain the Price of Annuities upon Lives. Philosophical Transactions of the Royal Society. https://doi.org/10.1098/rstl.1693.0007.
Halley, Edmond. (1693). Paper concerning the Breslau Bills of Mortality. Royal Society, Making Science manuscript catalog, https://makingscience.royalsociety.org/items/clp_21_45/paper-of-the-breslau-bills-of-mortality-by-edmond-halley.
Hecht, Jacqueline. (1987). Johann Peter Süssmilch: A German prophet in foreign countries. Population Studies, 41(1), 31–58. https://doi.org/10.1080/0032472031000142526.
Institute and Faculty of Actuaries. (2023). IFoA celebrates its 175th anniversary: 1848 founding of the Institute of Actuaries. https://actuaries.org.uk/news-and-media-releases/news-articles/2023/july/7-jul-23-ifoa-celebrates-its-175th-anniversary/.
Institute and Faculty of Actuaries. (n.d., consulted 2026). Historical collections and archives: repository and collection descriptions for the history of actuarial practice. https://actuaries.org.uk/insight/library-services/historical-collections-and-archives/.
MacTutor History of Mathematics Archive, University of St Andrews. (n.d., consulted 2026). William Morgan: Biographical account of his 1775 Equitable appointment and early valuation. https://mathshistory.st-andrews.ac.uk/Biographies/Morgan_William/.
NN Group. (n.d., consulted 2026). Life annuities in the seventeenth century: Jan de Witt, Johannes Hudde, and Dutch annuity valuation. https://www.nn-group.com/our-company/our-history/life-annuities-in-the-17th-century.
Poitras, Geoffrey. (2025). More on the Origin of Financial Economics: Early Contributions to Joint Life Annuity Valuation. Journal of the History of Economic Thought, 47(3), 339–362. https://doi.org/10.1017/S1053837224000300.
Price, Richard. (1771). Observations on Reversionary Payments; on Schemes for Providing Annuities for Widows, and for Persons in Old Age; on the Method of Calculating the Values of Assurances on Lives; and on the National Debt. London. Wellcome Collection bibliographic entry: https://wellcomecollection.org/works/vh4datuu.
Price, Richard. (1783). Observations on Reversionary Payments, revised fourth edition. Digitized catalog and scan, Google Books: https://books.google.com/books/about/Observations_on_Reversionary_Payments.html?id=31RHAAAAYAAJ.
Simpson, Thomas. (1742). The Doctrine of Annuities and Reversions, Deduced from General and Evident Principles. London. Wellcome Collection copy and catalog record: https://content.www.wellcomecollection.org/works/bz2tuf57.
Society of Actuaries. (n.d., consulted 2026). Historical background: The Actuarial Society of America, founded in 1889, and its successors. https://www.soa.org/about/historical-background/.
Statistics Sweden. (n.d., consulted 2026). Statistics Sweden's history: the founding of Tabellverket in 1749. https://www.scb.se/statistics-swedens-history/.
Umeå University, Centre for Demographic and Ageing Research. (2025). Tabellverket research resources and historical demographic tables. https://www.umu.se/en/centre-for-demographic-and-ageing-research/doi/tabellverket_v01/.
“William Morgan, F.R.S., 1750–1833.” (1933). Journal of the Institute of Actuaries, 64(2), 364–365. Historical biographical note. https://doi.org/10.1017/S0020268100033035.
Institute and Faculty of Actuaries. (2026). “Graduating Mortality Base Tables: Theory and Practice Part 1.” British Actuarial Journal, 31, e5. Sessional meeting discussion following a September 10, 2025 webinar presented by Matt Fletcher and Piero Cocevar. https://doi.org/10.1017/S135732172510038X.