Total Fertility Rate

The total fertility rate (TFR) is the average number of live births a woman would have if she experienced the age-specific fertility rates of a particular period throughout her reproductive life and survived to its end. Expressed in births per woman, it describes a hypothetical, or synthetic, generation rather than the actual lifetime childbearing of women observed in that period. It is a central measure in demography, allowing comparisons of fertility across populations with different age structures. (un.org)

TFR is used to study changing patterns of childbearing, assess generational replacement, and construct population projections. The United Nations estimated global TFR at 2.25 births per woman in 2024, compared with 3.31 in 1990. Interpreting such figures requires distinguishing period rates from completed cohort fertility, changes in birth timing from changes in eventual family size, and fertility from the other components of population change. Replacement fertility is approximately 2.1 births per woman in low-mortality populations, but the precise threshold varies, and reaching it does not immediately stabilize population size. (un.org)

Definition, Scope, and Related Measures

The total fertility rate (TFR) is the average number of live births a hypothetical woman would have over her reproductive lifetime if she survived through that lifetime and experienced, at every age, the age-specific fertility rates observed in a specified period. Usually calculated for a calendar year, it is a synthetic lifetime total: the fertility of women of different ages during that year is combined into one measure expressed in births per woman. The assumption of unchanged rates is a condition of interpretation, not a claim that future rates will remain constant. Likewise, survival through reproductive life is assumed rather than weighted by the actual probability of dying before or during those years. (digitallibrary.un.org)

In demography, fertility means realized childbearing, whereas fecundity denotes biological reproductive capacity. TFR therefore measures births, not the physiological ability to reproduce. The birth counts underlying its age-specific rates are live births, rather than pregnancies, deliveries, or children still alive at a later date. A live-born child remains included even if the child subsequently dies; pregnancies ending without a live birth do not enter the numerator. This distinction also separates TFR from measures of surviving offspring or generational replacement that explicitly incorporate mortality. (un.org)

The crude birth rate relates live births during a year to the entire midyear population, conventionally per 1,000 people. Its denominator includes men and women of every age, making it sensitive to the population’s age and sex composition. The general fertility rate instead relates annual live births to women of reproductive age, conventionally per 1,000 women. The specified age range matters: United States National Center for Health Statistics publications, for example, use ages 15–44 for this measure. Restricting the denominator removes the direct influence of the proportion of men, children, and older people in the total population, but the result still depends on how women are distributed across reproductive ages. (archive.cdc.gov)

TFR differs from both measures because it combines separate age-specific rates rather than dividing total births by one aggregate population denominator. It consequently avoids the direct age-composition effect within the reproductive-age population. Two populations with identical fertility rates at each age can have the same TFR but different general fertility rates if one has a larger proportion of women at ages when childbearing is more frequent. The distinction was illustrated in a 1984 National Center for Health Statistics report: Maine’s general fertility rate for 1980 was 64.8 births per 1,000 women aged 15–44, an annual frequency rather than a lifetime total. (humanfertility.org)

A TFR of 1.6 births per woman therefore does not mean that women currently living in the population have already had an average of 1.6 children. Such an observed average would depend on which ages were included and how much of their reproductive lives those women had completed. Nor does 1.6 predict any individual woman’s eventual number of children: it describes the lifetime total implied by a population-level period schedule. For example, a TFR calculated for 2020 combines the rates experienced by younger and older women in 2020, not the successive rates that women born in 2020 will encounter as they age. A measure following one birth cohort through its reproductive life answers a different question. (humanfertility.org)

Calculation and Mathematical Interpretation

An age-specific fertility rate (ASFR) relates live births to the time women spend in a specified age interval. For an interval beginning at age $x$ and spanning $n_x$ years, let $B_x$ denote births to women in that interval during the reference period and $E_x$ their corresponding female person-years of exposure. The annual rate is $f_x=B_x/E_x$, measured in births per woman-year. The denominator includes all women in the age interval, not only mothers or married women. Exposure measures time actually spent within the interval: a woman observed for six months contributes half a woman-year. For a single calendar year, the midyear female population is commonly used to approximate annual exposure, with each woman counted representing approximately one woman-year. Births and exposure must refer to the same age interval and population. (millenniumindicators.un.org)

The grouped-age calculation multiplies each annual rate by the number of years represented by its age interval, then adds the contributions:

Here $n_x$ is the interval width in years, not the length of the statistical observation period. Multiplication converts births per woman-year into births per woman over that age interval. When published ASFRs, denoted $r_x$, are expressed per 1,000 women rather than per woman, $f_x=r_x/1000$, giving

For the conventional seven five-year groups covering ages 15–49, this becomes $5\sum_x r_x/1000$. With single-year ages, the interval width is one and the rates are simply summed after any necessary conversion from the per-1,000 scale. (pmc.ncbi.nlm.nih.gov)

Consider a hypothetical annual schedule, constructed solely to illustrate the calculation:

The rates sum to 376, so the hypothetical TFR is

For ages 25–29, an annual rate of 110 per 1,000 becomes $0.110$ births per woman-year; applying that rate for five years contributes $0.550$ births per woman. The factor of five therefore represents passage through five years of age, not five years of observed births. Although every input comes from one annual schedule, the calculation accumulates contributions across the entire specified reproductive-age span. It does not divide the annual number of births by the total number of women aged 15–49. (pmc.ncbi.nlm.nih.gov)

Age coverage is part of the measure’s specification, rather than a universal biological boundary. The Human Fertility Database notes that some statistical agencies calculate TFR over ages 15–49, while others use 15–44; its own calculations use a wider range, with endpoint categories covering age 12 and younger and age 55 and older. Births below age 15 and above age 49 can therefore enter one published measure but fall outside another. Holding all shared-age rates unchanged, adding omitted age-specific contributions can only increase the calculated total, although the size of the difference depends on fertility at those ages. Comparisons consequently require checking the lower and upper age limits and the treatment of births outside them, rather than assuming that all figures labelled “TFR” cover identical ages. The United Nations separately defines adolescent birth rates for ages 10–14 and 15–19, making childbearing below the conventional starting age explicitly measurable. (humanfertility.org)

Period Fertility, Cohort Fertility, and Birth Timing

Period TFR combines the fertility rates of women of different ages—and therefore different birth cohorts—during one calendar period. A rate for women aged 25 in 2020 and a rate for women aged 35 in 2020 describe different generations, not successive stages in one generation’s life. Completed cohort fertility, by contrast, sums the age-specific fertility rates experienced by women born in the same year as they pass through their reproductive lives. It measures that cohort’s average lifetime childbearing rather than a synthetic lifetime constructed from one period’s rates. Its principal limitation is the waiting time: the Human Fertility Database normally reports completed measures for cohorts that have reached age 50, with complementary measures through age 40 distinguished from full reproductive-life totals. (humanfertility.org)

The distinction separates fertility quantum, the amount of childbearing, from tempo, its timing. When successive cohorts postpone births, fertility rates at younger ages fall before the corresponding births occur at older ages. Period TFR can consequently decline even if those cohorts eventually have the same average number of children. If postponement subsequently stops, this downward timing effect can diminish, allowing period TFR to rise without an increase in completed family size. Earlier childbearing can produce the opposite effect, temporarily concentrating births within a period. These mechanisms do not mean that all postponed births are eventually recovered: changes in timing and lifetime birth numbers can occur together. A fertility tempo effect specifically denotes the contribution of changing birth timing to period fertility measures. (knowledgecommons.popcouncil.org)

Several complementary measures distinguish these processes. Mean age at childbearing summarizes timing; in the Human Fertility Database it is calculated by weighting ages with age-specific fertility rates, rather than simply averaging mothers’ ages in registered births. Birth-order-specific rates separate first, second, and subsequent births, allowing changes in entry into parenthood to be distinguished from changes in additional childbearing. Completed cohort childlessness measures the proportion remaining at parity zero, where parity denotes the number of children already born. A parity progression ratio measures the probability of having another child conditional on having reached a specified parity: progression from one to two children concerns mothers of one child, not all women. Thus, a falling period first-birth rate can reflect delayed parenthood, whereas rising completed cohort childlessness indicates a change in eventual entry into parenthood. (humanfertility.org)

In 1998, John Bongaarts and Griffith Feeney proposed adjusting period TFR to remove the estimated effect of changing birth timing. Their method operates separately by birth order:

where $\mathrm{TFR}_i(t)$ is the observed period contribution of births of order $i$, and $r_i(t)$ is the annual change in the period mean age at that birth order, expressed in years of age per calendar year. Positive $r_i$ raises the adjusted contribution above the observed one; negative $r_i$ lowers it. For a hypothetical illustration, if all birth orders have a mean-age increase of $0.10$ years annually and observed TFR is $1.80$, the adjustment gives $1.80/(1-0.10)=2.00$ births per woman. This calculation attributes the difference to timing under the model, not to directly observed additional births. (knowledgecommons.popcouncil.org)

The adjustment assumes that timing shifts operate similarly across ages and cohorts within each birth order, with the normalized age-specific fertility schedule retaining its shape as it shifts. Age-specific differences in postponement or changes in the schedule’s dispersion can therefore affect its interpretation. The adjusted value is a counterfactual period measure: an estimate of fertility in that period without its modeled timing change. It is neither an observed completed cohort total nor an unconditional forecast of future family size. Subsequent births, cohort-specific behavior, and future changes in quantum remain outside what the adjustment alone establishes. (demogr.mpg.de)

Data Sources, Estimation, and Measurement Quality

Direct estimation uses live births recorded through civil registration and corresponding estimates of female population exposure. National statistical offices compile births by the mother’s age and residence, then relate them to women of the same ages in the same population and period. Exposure is measured in woman-years, often approximated by midyear population estimates. Numerators and denominators must use compatible geographic boundaries and residence definitions: births occurring in a city’s hospitals are not necessarily births to that city’s residents. Maternal age must also be classified consistently, including the distinction between age at the birth and age reached during the calendar year. The Human Fertility Database explicitly checks registration coverage, timeliness, maternal-age categories, and the availability of live-birth rather than total-birth counts. (humanfertility.org)

Where registration is incomplete, household surveys provide retrospective reproductive histories. In the Demographic and Health Surveys (DHS), women report births and their dates, allowing births and exposure to be allocated to the mother’s age when each event occurred. The standard DHS TFR covers the three years preceding the survey, rather than a single calendar year; its survey-year label therefore does not identify an annual fertility rate. Population censuses may instead ask about children ever born, births during the preceding 12 months, or the date of the most recent live birth. Lifetime birth totals are not themselves a period TFR, but they can support indirect estimation using demographic models and consistency relationships when recent-birth information or registration is deficient. Such estimates depend on assumptions in addition to the reported counts. (dhsprogram.com)

Errors affect both sides of the fertility-rate calculation. Birth underregistration and omitted survey births generally depress measured fertility; misdated births can move events across reference-period boundaries, creating artificial changes over time. Maternal-age misstatement reallocates births or exposure between age groups. Population undercounts, outdated population estimates, and incomplete accounting for migration can distort denominators even when birth registration is complete. Revisions to population estimates can consequently revise historical rates without changing recorded births. For example, an April 2026 US National Center for Health Statistics report recalculated rates for 2015 using revised population estimates and noted that they could differ from those originally published. DHS methodological investigations likewise distinguish omitted births from displaced birth dates, since both can produce misleading retrospective trends. (humanfertility.org)

Survey estimates also carry sampling uncertainty: precision depends on the number of interviewed women and births, survey weights, and the clustered sample design. Confidence intervals describe sampling variability, not the full effects of recall errors or omissions. Complete registration avoids survey sampling error but does not eliminate random variation in birth occurrence. In small populations, a few additional births can substantially alter an age-specific rate and thus the TFR, particularly where exposure in individual age groups is small. Multi-year estimates increase the number of observed events but describe a longer reference period and may conceal short-term changes. Registration-based comparisons therefore distinguish random fluctuations from recording errors and sustained changes in fertility. (dhsprogram.com)

International databases serve different purposes. The World Population Prospects series of the United Nations Population Division combines registration, census, and survey evidence into internationally consistent estimates and projections. Its 2024 revision, released on 11 July 2024, incorporated 1,910 censuses conducted during 1950–2023 and 3,189 nationally representative surveys; historical estimates cover 1950–2023, whereas 2024–2100 belongs to the projection period in that edition. These model-based series must be distinguished from individual reported observations. The Human Fertility Database, by contrast, applies uniform calculations to official birth counts and population exposures in populations with virtually complete registration and reliable denominators; it does not correct inputs for incomplete coverage. A reproducible comparison identifies the reference year or survey interval, whether the figure is reported, adjusted, or projected, and the database revision or download date. (population.un.org)

Replacement Fertility and Generational Reproduction

Replacement fertility is the fertility level at which successive generations replace themselves under specified fertility and mortality schedules, excluding migration. Its exact criterion is one daughter per newborn girl, on average, allowing for female mortality before and during childbearing. This is a generational relationship, not immediate equality between annual births and deaths: existing age structure can sustain growth or decline after fertility reaches replacement. (unstats.un.org)

The familiar threshold of approximately 2.1 births per woman, rather than exactly two, reflects both the sex ratio at birth and female survival. With 105 boys born for every 100 girls, the female share of births is $100/205 \approx 0.4878$. Even if every female survived throughout reproductive life, an average of $1/0.4878 = 2.05$ births would therefore be required to produce one daughter per woman. Mortality before and during the reproductive ages raises the requirement further. The United Nations report Fertility Levels and Trends as Assessed in the 2012 Revision of World Population Prospects described approximately 2.1 as the replacement level for most countries with low or moderate mortality and a birth sex ratio near 105 boys per 100 girls. The additional fraction is thus not exclusively an allowance for childhood deaths. (un.org)

The gross reproduction rate (GRR) counts daughters per woman under a specified fertility schedule, assuming survival throughout reproductive life; the net reproduction rate (NRR) additionally allows for female mortality. For age intervals of width $n_x$, let $f_x$ denote annual births of both sexes per woman, $p_x$ the female proportion of those births, and $s_x$ female survival from birth to the corresponding childbearing age. Then

Grouped-age calculations approximate survival within each interval. An NRR of one indicates exact replacement; values below or above one indicate less or more than replacement under unchanged schedules and without migration. Both are ordinarily period measures for a hypothetical cohort, not observed lifetime outcomes. (unstats.un.org)

A hypothetical calculation illustrates mortality’s effect. If the female birth share is constant at $p$, and $\bar{s}$ is survival averaged over ages using fertility as weights, then $\mathrm{NRR}=\mathrm{TFR}\,p\,\bar{s}$; consequently, replacement requires $\mathrm{TFR}_{R}=1/(p\bar{s})$. With $p=100/205$ and $\bar{s}=0.98$, the threshold is approximately 2.09 births per woman. If weighted survival were instead 0.80, it would rise to approximately 2.56. These are illustrative assumptions, not country estimates. The relevant survival measure includes deaths from birth through childbearing ages, rather than only deaths associated with childbirth; it also depends on when births occur within reproductive life. (unstats.un.org)

Accordingly, 2.1 is a useful rounded benchmark, not a universal demographic constant. In its fertility assumptions published on October 29, 2015, the United Kingdom’s Office for National Statistics specified 2.075 children per woman for long-term natural replacement. Comparisons across countries or historical periods require the corresponding female mortality schedule and sex ratio at birth: higher female mortality or a smaller female share of births raises the fertility needed for replacement. (ons.gov.uk)

Historical Change and International Patterns

Fertility decline forms part of the demographic transition, the historical shift from high mortality and high fertility toward longer lives and smaller families. Declining mortality, particularly among children, commonly precedes declining childbearing: population growth accelerates while more people survive but fertility remains high, then slows as fertility falls. Many European countries began this transition during the nineteenth century, whereas most countries in less developed regions entered it during the twentieth century. The interval between mortality and fertility decline, and the subsequent speed of fertility change, have varied substantially; the transition does not imply a uniform national timetable. (un.org)

The United Nations World Population Prospects 2024 revision places global total fertility at 3.31 births per woman in 1990, 2.93 in 1994, and approximately 2.25 in 2024—a reduction of roughly one birth per woman since 1990. These figures belong to one revision rather than a series assembled from successive editions. The 2024 figure is the revision’s contemporary assessment, although 2024 formally begins its projection period; historical estimates extend through 2023. The accompanying summary reported that more than half of countries and areas had fertility below 2.1 births per woman in 2024. (un.org)

Common reference years reveal contrasting regional trajectories. Between 1994 and 2024, fertility fell from 6.06 to 4.26 births per woman in sub-Saharan Africa, from 3.92 to 2.24 in Central and Southern Asia, and from 1.98 to 1.34 in Eastern and South-Eastern Asia. Latin America and the Caribbean moved from 2.98 to 1.80, crossing below 2.1 in 2014. The combined UN grouping of Europe, Northern America, Australia and New Zealand changed less, from 1.63 to 1.48, having already reached low fertility. These aggregates conceal substantial national variation: the same revision placed 2024 fertility at 5.94 in Niger and 3.17 in Kenya. (digitallibrary.un.org)

Below-replacement fertility therefore encompasses populations with different histories, not a single regional pattern. For comparison, 2.1 births per woman is an approximate low-mortality replacement benchmark, while “very low” fertility here denotes fewer than 1.4; the latter is a descriptive cutoff, not a universal demographic standard. The UN’s 2024 summary identified China, the Republic of Korea, Italy and Spain among countries below 1.4. Italy and Spain had already recorded approximately 1.2 births per woman in 1994, showing that very low fertility was established in parts of southern Europe decades before 2024. (un.org)

East Asian trajectories also differ in timing and depth. In the UN’s consistently dated 1994–2024 comparison, Japan declined from 1.47 to 1.22 births per woman, the Republic of Korea from 1.67 to 0.73, and China from 1.63 to 1.01. Thus Japan’s longer history of low fertility did not translate into the lowest period rate in this comparison: Korea’s 2024 level was substantially lower. These are period indicators, rather than the completed family sizes of women born in particular years. (digitallibrary.un.org)

Mechanisms and Socioeconomic Correlates

Social circumstances affect births through proximate determinants of fertility: biological and behavioural mechanisms operating between social conditions and reproduction. John Bongaarts’s 1978 framework identified four especially important determinants of differences between populations: exposure to intercourse, originally approximated by the proportion married; contraceptive use and effectiveness; induced abortion; and postpartum infecundability. These mechanisms affect different stages of reproduction: contraception reduces the probability of conception, abortion interrupts a pregnancy before a live birth, and breastfeeding can prolong the temporary suppression of ovulation following childbirth. Postpartum abstinence further lengthens the interval before renewed exposure. Bongaarts’s 2015 revision explicitly addressed sexual activity outside marriage and overlapping periods of contraceptive use and postpartum infecundability, which otherwise risk being counted twice. (jstor.org)

Education and urbanization are broader correlates rather than equivalent physiological mechanisms. Schooling can influence age at union formation, knowledge and use of contraception, employment opportunities, and participation in reproductive decisions. A United Nations analysis published in the Population Bulletin in 2002 found that women’s education and child survival together accounted statistically for almost four-fifths of cross-national TFR variation among 57 developing countries with available data. This was an aggregate association, not an estimate of the causal effect of either factor. The same analysis found lower fertility among urban residents but a weaker relationship between national urbanization levels and fertility, partly reflecting inconsistent definitions of urban settlements. Such findings do not establish a universal sequence in which schooling, urban residence, or improved survival produces a fixed fertility response. (digitallibrary.un.org)

In low-fertility settings, employment and housing influence whether prospective parents can establish an independent household and accommodate children. OECD analysis reported in 2024, using country-level data for 2002–2019, found positive associations between TFR and both men’s and women’s employment, but negative associations with unemployment and housing expenditure. Job insecurity can complicate commitments involving long-term expenditure; expensive housing can make additional living space difficult to obtain. The OECD explicitly classified these regression results as associations rather than evidence of causation, and noted that substantial fertility variation remained unexplained. (oecd.org)

The compatibility of employment and care also changes the costs of childrearing. These include direct expenditure and the opportunity costs of interrupted earnings, reduced working hours, and foregone career progression. Affordable childcare matters through its availability and suitability for parents’ working schedules, not simply its nominal existence. The OECD’s 2025 examination of Korea illustrates the distinction between educational attainment and the ability to combine work with motherhood: employment among women with a university degree or higher was 69%, compared with an OECD average of 83%, and many highly educated Korean women left employment upon motherhood. Employment protection, working practices, childcare arrangements, and household responsibilities therefore interact rather than acting as interchangeable determinants. (oecd.org)

Partnership formation and social norms further mediate reproductive intentions. In 2020, just over 40% of OECD births occurred outside marriage, compared with less than 3% in Korea, Japan, and Türkiye; delayed marriage consequently has different implications across settings. Unequal care responsibilities can also produce disagreement between partners about additional children. Intended family size is therefore distinct from realized births: intentions may change, partners may disagree, and economic or partnership circumstances may prevent their fulfilment. Neither a stated preference nor a national TFR identifies an individual’s eventual family size. (oecd.org)

Differences Within Populations and Aggregation

Fertility comparisons within a country commonly distinguish administrative regions, urban and rural residence, educational attainment, and socioeconomic position. Each subgroup requires its own age-specific birth counts and corresponding female person-years of exposure: births to university-educated women aged 25–29, for example, must be divided by exposure among women of that age and educational category, not by all women aged 25–29. Reference periods, age coverage, and classification rules must match. In retrospective surveys, residence or socioeconomic characteristics recorded at interview need not describe circumstances when earlier births occurred. Such comparisons identify differences, but do not by themselves establish causal effects of schooling, residence, or wealth. (dhsprogram.com)

National totals can conceal substantial geographical variation. India’s National Family Health Survey of 2019–2021 reported a national TFR of 2.0 births per woman, while state estimates ranged from 3.0 in Bihar to 1.1 in Sikkim; the urban estimate was 1.6. These are survey-based period measures, not completed family sizes. Disaggregation also reduces the observations supporting each estimate. The DHS statistical guide therefore flags fertility rates based on 125–249 unweighted person-years of exposure and suppresses those based on fewer than 125, illustrating why a detailed subgroup table may contain unavailable or imprecise estimates. (pib.gov.in)

Birth order provides another decomposition. If $B_{x,i}$ denotes births of order $i$ to women in age interval $x$, and $E_x$ denotes exposure among all women in that interval, the order-specific contribution is $TFR_i=\sum_x n_xB_{x,i}/E_x$, and $TFR=\sum_iTFR_i$. This separates changes in first births from changes in second and higher-order births. These unconditional contributions differ from parity progression ratios: for a completed cohort, the latter measure the proportion of women reaching a given parity who subsequently have another child. A second-birth contribution uses all women’s exposure, whereas a conditional second-birth rate uses exposure among women with exactly one previous live birth. Consequently, low second-birth contributions can reflect both fewer women entering motherhood and lower progression among mothers. (humanfertility.org)

Migration changes both population exposure and the composition of the women being observed. Birth numerators and exposure denominators must apply consistent residence and migrant-status definitions. Moreover, a migrant period TFR combines women arriving at different ages and with different durations of residence; births concentrated shortly after arrival can raise it without implying an equally high completed family size. A single annual figure therefore cannot describe an entire migration-linked reproductive history. The Human Fertility Database includes migration when estimating female population exposures, but its reconstruction of age–parity distributions assumes that migrants’ parity structure matches that of the general population—a separate assumption that may not hold. (demographic-research.org)

Aggregation across countries or subgroups must occur at the age-specific-rate level. For country $c$, let $f_{x,c}=B_{x,c}/E_{x,c}$. Applying the standard births-to-exposure definition to pooled observations gives

and hence

The weights depend on female exposure at each age, not on countries’ total populations or their birth shares. Because these weights generally vary across ages, the pooled TFR is not generally either the arithmetic mean or a single population-weighted mean of national TFRs. This follows directly from pooling the numerator and denominator before summing rates. (un.org)

A hypothetical example demonstrates the difference. Suppose country A has annual fertility rates of $0.20$ and $0$ in two five-year age groups, while country B has $0$ and $0.20$; rates at all other reproductive ages are zero. Both national TFRs equal $5(0.20)=1.0$. If A supplies 80% of pooled female exposure in the first group but only 20% in the second, both pooled rates equal $0.16$. The pooled TFR is therefore $5(0.16+0.16)=1.6$, exceeding both national values despite their identical TFRs.

Population Dynamics, Projections, and Planning

In cohort-component projections, fertility generates births within a population whose size and age composition are themselves changing. Each birth cohort is advanced through time using age- and sex-specific survival assumptions, with immigrants added and emigrants removed. Births are calculated by applying projected age-specific fertility rates to the corresponding female population exposures: $B_t=\sum_x f_{x,t}E_{x,t}$, where $E_{x,t}$ denotes female person-years at age $x$. Newborns are divided by sex and subjected to infant survival assumptions. Consequently, identical TFRs can produce different birth totals because populations contain different numbers of women at childbearing ages; different age patterns of fertility can also matter even when their totals coincide. Migration affects both population size directly and subsequent births by changing reproductive-age exposures. (www2.census.gov)

Population momentum explains why reaching replacement fertility does not immediately stop population growth. In a youthful population, large cohorts entering reproductive ages can produce more births than the deaths occurring among relatively small older cohorts, even when each generation only replaces itself. The United Nations’ World Population Prospects 2024 attributed approximately 79 per cent of projected global growth between 2024 and 2054—about 1.4 billion people—to momentum embedded in the existing age structure. Conversely, a population with small cohorts entering childbearing ages and numerous older people may continue shrinking despite an increase in fertility: higher rates do not necessarily offset fewer potential mothers and more deaths. (un.org)

Sustained fertility decline changes the balance between age groups. Smaller birth cohorts initially reduce child dependency; later, as they enter working ages and larger preceding cohorts grow older, they contribute to population ageing and higher old-age dependency ratios. These ratios compare conventionally defined age groups, not actual dependants and earners: employment, retirement, unpaid care and productivity determine how demographic change translates into economic support requirements. In the 2024 UN projections, countries whose populations had already peaked were expected to experience a fall in the share aged under 20 from 21 per cent in 2024 to 14 per cent in 2054. (un.org)

For planning, the timing of cohort movement matters as much as the eventual population total. Fewer births affect maternity and early-childhood services first, school-entry cohorts several years later, and potential labour-market entrants roughly two decades later. School enrolment additionally depends on attendance and progression, while workforce size depends on participation and migration. A temporarily larger working-age share can create conditions for a demographic dividend, but does not automatically generate economic growth: education, health and productive employment influence whether that opportunity becomes an economic gain. (un.org)

Alternative fertility assumptions reveal how strongly projections depend on future reproductive behaviour. In the 2024 UN revision, high- and low-fertility scenarios place fertility approximately 0.5 births per woman above or below the medium trajectory. Persistent differences compound because they alter both the first generation of births and the number of potential parents in subsequent generations. These sensitivity scenarios are distinct from probabilistic prediction intervals. The UN also reports 80 and 95 per cent intervals derived from simulated demographic trajectories, using Bayesian models for fertility. Such intervals describe uncertainty conditional on the model and its assumptions; they are neither guarantees nor the same as high–low scenario bounds. A constant-fertility scenario likewise asks what would follow if fertility remained fixed, rather than asserting that reproductive behaviour will remain unchanged. (population.un.org)

Policy Uses, Evaluation, and Reproductive Autonomy

TFR is used to monitor demographic trends and assess the context in which population and family policies operate, but increasing births is not synonymous with improving family welfare. The OECD Family Database places fertility rates alongside indicators of desired and actual family size, childlessness, parental employment, leave entitlements, and childcare support. The OECD’s 2024 policy review distinguishes objectives including reducing child poverty, supporting child development, advancing gender equality, reconciling employment with care, and enabling people to have children when they wish. Consequently, a programme can achieve important objectives without raising TFR: the review reports that Poland’s 500+ child-benefit programme substantially reduced extreme child poverty without achieving its fertility objective. (oecd.org)

Policy instruments operate through different constraints. Paid parental leave replaces earnings during time away from employment; childcare provision reduces care costs and facilitates continued employment; cash benefits and tax concessions reduce direct childrearing expenses. Housing assistance addresses the cost of obtaining suitable accommodation. Hungary’s package described in the OECD’s 2024 review combined cash and tax benefits, paid leave, and housing grants and subsidised loans. These mechanisms do not establish that an observed national TFR change was caused by the package. Separately, voluntary family-planning services provide contraception, information, and counselling to support decisions about whether and when to become pregnant; their rights-based purpose is informed choice rather than compliance with a numerical birth target. (oecd.org)

Evaluation therefore requires a credible comparison with what would have happened without a policy. Relevant distinctions include announcement and implementation dates, eligibility and actual uptake, and pre-existing differences between recipients and non-recipients. Employment-linked eligibility, for example, can select people with different earnings and family-building trajectories. A natural experiment exploiting an eligibility cutoff can strengthen causal inference, provided that other changes or strategic behaviour do not invalidate the comparison. Gordon B. Dahl, Katrine V. Løken, Magne Mogstad, and Kari Vea Salvanes examined Norwegian reforms expanding paid maternity leave from 18 to 35 weeks. Their study, published in 2016, compared births around reform cutoffs and found little effect on completed fertility. That result concerns extensions of an existing entitlement in a particular institutional setting, not every possible leave policy. (econstor.eu)

Evaluation must also distinguish advancing births from increasing lifetime family size. A benefit can bring an already intended birth forward, temporarily raising period TFR without adding a birth over the reproductive life course. Conversely, postponement following a benefit restriction can lower current fertility without establishing an equivalent permanent reduction. Long follow-up, birth-order-specific outcomes, and completed cohort fertility help separate these effects. The OECD review documents substantial variation across policies and settings, including short-lived increases following birth allowances in Swiss cantons; a short-term rise is therefore insufficient evidence of a lasting increase in family size. (oecd.org)

UNFPA’s State of World Population 2025, subtitled The Pursuit of Reproductive Agency in a Changing World, treats reproductive autonomy as distinct from aggregate fertility targets. Its framework concerns whether people can decide freely whether, when, and with whom to have children, supported by the conditions necessary to exercise that choice. In its survey with YouGov across 14 countries, nearly 20% of reproductive-age adults expected to be unable to achieve their desired number of children, while 39% reported financial constraints affecting or potentially affecting that goal. These are findings from the surveyed countries, not a census of global intentions. TFR alone records neither unwanted births nor unrealised desired births, and cannot establish whether choices involved informed consent or coercion; reproductive well-being consequently requires evidence about intentions, service access, and decision-making power as well as birth rates. (unfpa.org)

References

1. Methods Protocolhumanfertility.org
2. The World's Women 2000: Trends and Statisticsdigitallibrary.un.org
3. The World Population Situation in 1970un.org
4. Rate - Health, United Statescdc.gov
5. Principles of Epidemiology: Lesson 3, Section 4archive.cdc.gov
6. Birth and Fertility Rates for States: United States, 1980cdc.gov
7. HFD Frequently Asked Questionshumanfertility.org
8. Indicator 3.7.2 - E-Handbook on SDG Indicators - UN Statistics Wikimillenniumindicators.un.org
9. Indicator 3.7.2 - Adolescent birth rate (aged 10–14 years; aged 15–19 years) per 1,000 women in that age group - U.K. Indicators For The Sustainable Development Goalssdgdata.gov.uk
10. Calculating fertility and childhood mortality rates from survey data using the DHS.rates R packagepmc.ncbi.nlm.nih.gov
11. Comparisons of DHS Estimatespreview.dhsprogram.com
12. HFD Data Availabilityhumanfertility.org
13. HFD Overviewhumanfertility.org
14. Human Fertility Database: Expanding research opportunitieshumanfertility.org
15. On the quantum and tempo of fertilityknowledgecommons.popcouncil.org
16. A Cohort Model of Fertility Postponementpmc.ncbi.nlm.nih.gov
17. How Big Are Educational and Racial Fertility Differentials in the U.S.?pmc.ncbi.nlm.nih.gov
18. Tempo Effects in the Fertility Decline in Eastern Europe: Evidence from Bulgaria, the Czech Republic, Hungary, Poland, and Russiademogr.mpg.de
19. Demographic translation and tempo effects: An accelerated failure time perspectivedemogr.mpg.de
20. A sensitivity analysis of the Bongaarts-Feeney method for adjusting bias in observed period total fertility ratespubmed.ncbi.nlm.nih.gov
21. Natality - Birth Records Documentationwonder.cdc.gov
22. Guide to DHS Statisticsdhsprogram.com
23. The DHS Program - Indicator Explorerdhsprogram.com
24. Fertility: 1999 Zimbabwe Demographic and Health Surveydhsprogram.com
25. United Nations Statistics Division - Demographic and Social Statisticsunstats.un.org
26. A Decade of Changes in State Fertility Rates for Women Age 30 and Older: United States, 2015–2024cdc.gov
27. Evidence of Omission and Displacement in DHS Birth Historiesdhsprogram.com
28. Vital and Health Statistics, Series 21, No. 57cdc.gov
29. Fertility Rate: Stats of the Statesportugues.cdc.gov
30. Table 3: Notes on Selected Derived Measures of Natalityunstats.un.org
31. Fertility assumptions — Office for National Statisticsons.gov.uk
32. World Population Prospects 2024: Summary of Resultsun.org
33. Demographic Upheaval: The world will struggle with population growth, aging, migration, and urbanizationelibrary.imf.org
34. World Population Prospects 2024: Summary of Resultspopulation.un.org
35. World Fertility 2024un.org
36. World Fertility 2024digitallibrary.un.org
37. A Framework for Analyzing the Proximate Determinants of Fertilityjstor.org
38. Modeling the fertility impact of the proximate determinants: Time for a tune-updemographic-research.org
39. Modeling the fertility impact of the proximate determinants: Time for a tune-updemographic-research.org
40. Population Bulletin of the United Nations, Nos. 46/47digitallibrary.un.org
41. World population monitoring 2002: reproductive rights and reproductive healthdocuments.un.org
42. Guide to DHS statisticsdhsprogram.com
43. Fertility Differentials: Socioeconomic and Cultural Factorsdhsprogram.com
44. Union Health Ministry releases NFHS-5 Phase II Findingspib.gov.in
45. NFHS-5 Key Insightsindia.unfpa.org
46. Guide to DHS Statisticsdhsprogram.com
47. The fertility of recent migrants to England and Walesdemographic-research.org
48. United Nations Division for Sustainable Development—Total Fertility Rateun.org
49. Methodology, Assumptions, and Inputs for the 2023 National Population Projectionswww2.census.gov
50. Methodology and Assumptions for the Population Projections of the United States: 1999 to 2100cdn.www.census.gov
51. World Population Prospects 2024: Summary of Resultsun.org
52. World Population Prospects 2024: Methodology of the United Nations Population Estimates and Projectionspopulation.un.org
53. Data Portal: Total population by sex—projection variantspopulation.un.org
54. Explainer: What is UNFPA and why does it matter?eeca.unfpa.org
55. What Is the Case for Paid Maternity Leave?econstor.eu
56. State of World Population 2025unfpa.org
57. State of World Population Report 2025: The Real Fertility Crisisunfpa.org