Skip to main content
v2026.11,986 entries · CC-BY 4.0

Population Projections: Models, Formulas & Guide

A comprehensive methodological guide to population projections: cohort-component modeling, demographic equations, Leslie matrices, and code implementations.

Written and maintained by CASRAI Editorial Board

Published

On this page

Definition in One Paragraph

A population projection is a formal mathematical simulation that calculates the future size, age structure, and sex distribution of a defined population by applying explicit, conditional assumptions regarding fertility, mortality, and net migration to an empirical baseline population. Unlike backward-looking demographic estimates that measure past or current population stocks using census enumerations and administrative registries, or speculative economic forecasts that attempt to predict behavioral reactions to macroeconomic shocks, a population projection demonstrates the necessary demographic consequences that unfold over time if specified demographic rates persist or transition along predefined trajectories. In demographic research and sociological analysis, projections provide the foundational quantitative baseline required for public finance planning, dependency ratio evaluations, school and healthcare infrastructure allocation, and long-range environmental impact assessments.

Demographers distinguish estimates, projections, and forecasts by operational certainty:

  • Demographic Estimate: An empirical assessment of past or current population stocks from census and vital statistics.
  • Population Projection: A conditional simulation of future population trajectories under explicit rate assumptions.
  • Population Forecast: A projection variant designated by an authority as the most probable future trajectory.

Applying established demographic research methodologies distinguishes behavioral trends from demographic momentum.

Formula with Every Symbol Explained

The standard framework for population projection is the cohort-component method, originally formulated by P.K. Whelpton in 1928 and formalized in matrix algebra by P.H. Leslie in 1945. The method projects population change by advancing individual age-sex cohorts through discrete time intervals under three mathematical components: the balancing equation, cohort survivorship, and the Leslie matrix.

The Demographic Balancing Equation

At the aggregate population level, total population size at time t + n is governed by the demographic balancing equation:

P(t + n) = P(t) + B(t, t + n) − D(t, t + n) + NM(t, t + n)

Where P(t + n) is the total resident population at the end of the interval; P(t) is the baseline population; B(t, t + n) is total live births; D(t, t + n) is total deaths (frequently reconciled using methods for excess mortality estimation during mortality shocks); and NM(t, t + n) is net migration: gross immigrant arrivals (I) minus gross emigrant departures (E).

Age-Specific Cohort Transitions and Survivorship

In discrete-time models with a uniform projection step of n years, individuals in age cohort [x, x + n) at time t transition into age cohort [x + n, x + 2n) at time t + n according to the cohort transition formula:

N(x + n, t + n) = N(x, t) × S(x, t) + NM(x, t)

Where N(x + n, t + n) is the population in age interval x + n to x + 2n at time t + n; N(x, t) is the population in age interval x to x + n at time t; S(x, t) is the cohort survival probability computed from life table person-years lived as S(x, t) = L(x + n) / L(x); and NM(x, t) is net migration allocated to age class x + n.

Fertility Generation and Sex Allocation

The newborn entry cohort (ages 0 to n at time t + n) is generated by applying age-specific fertility rates to the female reproductive population:

B(t, t + n) = n × Σ [ W(x, t + n/2) × ASFR(x) ]

Where W(x, t + n/2) is the mean female population in reproductive age class x (spanning ages 15–49), approximated as [W(x, t) + W(x, t + n)] / 2; ASFR(x) is the age-specific fertility rate for female cohort x; and births are divided by the biological sex ratio at birth (SRB ≈ 1.05, yielding female proportion p_f ≈ 0.4878 and male proportion p_m ≈ 0.5122). Surviving female infants aged 0 to n at time t + n equal total female births multiplied by newborn survival: N(0, t + n) = B_female × [L(0) / (n × l_0)].

The Leslie Matrix Formulation

For age-structured female populations across uniform age classes of length n, the cohort-component algorithm is expressed as a matrix-vector product n(t + 1) = L × n(t) + m(t), where n(t) is the population vector, m(t) is net migration, and L is the square projection matrix:

    [ F_0   F_1   F_2  ...  F_{k-1} ]
    [ s_0    0     0   ...     0    ]
L = [  0    s_1    0   ...     0    ]
    [  0     0    s_2  ...     0    ]
    [  0     0     0   ...  s_{k-2} 0 ]

Row 1 contains effective fertility coefficients (F_x), combining birth generation with infant survival. The subdiagonal contains cohort survival probabilities (s_x). By the Perron-Frobenius theorem, dominant real eigenvalue λ_1 gives the long-run growth factor per period, while its right eigenvector defines the stable age distribution.

Worked Example with Real Numbers and the Arithmetic Shown

To demonstrate the explicit numerical arithmetic of discrete-time cohort-component projection, consider a three-cohort female population evaluated over a 15-year projection interval (n = 15 years): Cohort 1 [0–14], Cohort 2 [15–29], and Cohort 3 [30–44].

Baseline Parameters and Rates

The baseline female population vector (in thousands) at time t = 0 is: w_1(0) = 40.00 (Cohort 1), w_2(0) = 50.00 (Cohort 2), and w_3(0) = 45.00 (Cohort 3), totaling 135.00 thousand.

Transition rates over the 15-year interval are:

  • Effective Fertility (F): F_1 = 0.000; F_2 = 0.720; F_3 = 0.480 surviving female offspring per woman.
  • Cohort Survival (s): s_1 = 0.970 (survival from Cohort 1 to 2); s_2 = 0.960 (survival from Cohort 2 to 3).
  • Net Migration Additions (M): M_1 = +2.50; M_2 = +4.00; M_3 = +1.50 thousand.

Step 1: Computing Natural Transitions via Leslie Matrix

Applying the matrix-vector multiplication w_nat(15) = L × w(0):

[ w_1(15)_nat ]   [ 0.000  0.720  0.480 ]   [ 40.00 ]
[ w_2(15)_nat ] = [ 0.970  0.000  0.000 ] × [ 50.00 ]
[ w_3(15)_nat ]   [ 0.000  0.960  0.000 ]   [ 45.00 ]

Evaluating each row:

  • Cohort 1 (Surviving Births): (0.000 × 40.00) + (0.720 × 50.00) + (0.480 × 45.00) = 0.00 + 36.00 + 21.60 = 57.60 thousand
  • Cohort 2 (Survivors from Cohort 1): (0.970 × 40.00) + (0.000 × 50.00) + (0.000 × 45.00) = 38.80 + 0.00 + 0.00 = 38.80 thousand
  • Cohort 3 (Survivors from Cohort 2): (0.000 × 40.00) + (0.960 × 50.00) + (0.000 × 45.00) = 0.00 + 48.00 + 0.00 = 48.00 thousand
  • Subtotal Natural Projected Population: 57.60 + 38.80 + 48.00 = 144.40 thousand

Step 2: Incorporating Net Migration

Adding age-allocated migrant counts to each natural cohort (w_final = w_nat + M):

  • Cohort 1 Final: 57.60 + 2.50 = 60.10 thousand
  • Cohort 2 Final: 38.80 + 4.00 = 42.80 thousand
  • Cohort 3 Final: 48.00 + 1.50 = 49.50 thousand
  • Total Projected Female Population at t = 15: 60.10 + 42.80 + 49.50 = 152.40 thousand

The total population expands by 17.40 thousand (+12.89%), decomposed into 9.40 thousand from natural increase and 8.00 thousand from net migration.

Interpretation and Reading Guidance

Interpreting population projections requires evaluating structural composition, dependency ratios, and age transitions alongside total counts. In demographic research, balancing quantitative and qualitative demographic data ensures that projected figures are contextualized by community infrastructure, familial structures, and local economic opportunities.

Core Structural Demographic Indicators

  • Child Dependency Ratio (CDR): Youth (ages 0–14) divided by working-age adults (ages 15–44). Baseline: 40.00 / 95.00 = 0.4211 (42.1 per 100). At t = 15, larger entering birth cohorts raise dependency: 60.10 / 92.30 = 0.6511 (65.1 per 100).
  • Old-Age Dependency Ratio (OADR): Elders (ages 65+) divided by working-age adults (ages 15–64). Elevated values indicate escalating fiscal pressures on pension and healthcare systems.
  • Rate of Natural Increase (RNI): Crude birth rate minus crude death rate, indicating biological expansion.
  • Median Population Age: The exact age dividing the population into two equal halves. Upward shifts denote structural population aging.

Reading Guidance and Demographic Thresholds

Demographic Indicator Formula / Definition Typical Range Reading Guidance & Interpretation
Total Fertility Rate (TFR) Σ [ ASFR(x) × n ] for ages 15–49 0.8 to 5.5 children/woman TFR < 2.1 denotes sub-replacement fertility; TFR < 1.3 indicates lowest-low fertility with rapid structural aging.
Total Dependency Ratio (TDR) [ Pop(<15) + Pop(65+) ] / Pop(15–64) × 100 40 to 90 per 100 workers TDR < 50 indicates a favorable demographic dividend; TDR > 70 reflects high societal dependency burdens.
Old-Age Dependency Ratio (OADR) Pop(65+) / Pop(15–64) × 100 8 to 55 per 100 workers OADR > 30 indicates super-aged societies requiring pension adjustments and expanded geriatric care infrastructure.
Intrinsic Growth Rate (r) ln(λ_1) / n −1.5% to +3.0% annually Negative values denote long-term contraction; values > 2.0% indicate rapid population doubling in under 35 years.
Sex Ratio at Birth (SRB) Male live births / Female live births 1.03 to 1.07 (baseline 1.05) Ratios > 1.08 suggest gender-biased prenatal sex selection; 1.05 reflects standard biological baseline equilibrium.

The data table below compares the structural specifications, projection horizons, and component estimation models used by leading international statistical agencies.

Assumptions and Common Mistakes

Analysts must critically evaluate underlying assumptions and avoid standard methodological pitfalls across long horizons.

Core Methodological Assumptions

  • Assumption of Rate Stability or Smooth Convergence: Models assume vital rates evolve along smooth mathematical trajectories. In reality, external disruptions (economic crises, conflicts, pandemics) produce sudden discontinuities.
  • Baseline Accuracy and Coverage: Models assume baseline census enumerations are complete and unbiased. Researchers designing survey frames should use rigorous probabilistic sampling methods to prevent undercounting mobile sub-populations.
  • Component Independence: Models treat fertility, mortality, and migration as independent processes, though immigrant arrivals often exhibit distinctive age-specific fertility profiles.

Five Common Mistakes in Population Projections

  1. Treating Projections as Unconditional Prophecies: Confusing conditional mathematical simulations with guaranteed forecasts.
  2. Ignoring Sub-National Internal Migration: Applying national vital rates to municipal projections while ignoring volatile inter-regional migration.
  3. Assuming a Closed Population: Omitting migration in low-fertility nations where net immigration sustains labor supply.
  4. Holding Component Rates Indefinitely Constant: Extrapolating static vital rates without incorporating mortality reductions or fertility shifts.
  5. Omitting Uncertainty Quantifications: Publishing single point projections without reporting scenario variants or Bayesian probabilistic bands.

How to Report It (APA Style Example)

When reporting population projections in academic manuscripts, follow American Psychological Association (APA 7th edition) standards by specifying baseline years, horizons, core models, and scenario variants.

APA Narrative Reporting Example

To evaluate regional healthcare infrastructure demands through 2040, an age-structured cohort-component population projection was executed across 5-year discrete intervals from a 2025 census baseline (N = 1,450,200). Baseline mortality parameters were derived from national life tables, assuming life expectancy at birth advances from 79.4 years in 2025 to 82.1 years by 2040. Age-specific fertility rates were parameterized assuming a gradual convergence of the Total Fertility Rate from 1.62 to 1.55 children per woman. Under the principal medium-variant scenario, annual net migration was modeled at a constant net inflow of +4,500 individuals per annum. By 2040, the total population is projected to reach 1,532,800 (95% CI [1,478,000, 1,594,300]), with the old-age dependency ratio expanding from 24.3 to 34.8 elders per 100 working-age adults.

Table 1. Projected Population and Dependency Indicators by Scenario Variant (2025–2040)
Scenario Variant Projected Population 2040 Absolute Change TFR Assumption Net Migration / Year Old-Age Dependency (OADR)
Low Growth Variant 1,462,100 +11,900 1.35 +2,000 36.9
Medium (Principal) Variant 1,532,800 +82,600 1.55 +4,500 34.8
High Growth Variant 1,610,400 +160,200 1.80 +7,000 32.5
Zero Net Migration Variant 1,418,600 −31,600 1.55 0 37.4

How to Calculate It in Excel, R, or Python

Below are verified computational implementations in Excel, R, and Python replicating the 3-cohort Leslie Matrix projection model from the worked example above.

Implementation in Microsoft Excel

To execute the 3-cohort projection model in Microsoft Excel:

  1. Input Base Population: In cells B2:B4, enter 40.0, 50.0, and 45.0.
  2. Input Leslie Matrix (L): In range D2:F4, enter Row 1: 0.00, 0.72, 0.48; Row 2: 0.97, 0.00, 0.00; Row 3: 0.00, 0.96, 0.00.
  3. Input Net Migration: In cells H2:H4, enter 2.5, 4.0, and 1.5.
  4. Matrix Multiplication: In cell J2, enter array formula =MMULT(D2:F4, B2:B4), outputting natural totals 57.60, 38.80, and 48.00.
  5. Add Net Migration: In cell K2, enter =J2+H2 and fill down to K4, yielding 60.10, 42.80, and 49.50.
  6. Sum Total Population: In cell K5, enter =SUM(K2:K4), returning 152.40.

Implementation in R

In R, demographic projections leverage matrix multiplication. Analysts evaluating survey data can integrate these models with packages for survey analysis in R.

# Cohort-Component Population Projection via Leslie Matrix in R
w0 <- c(Age_0_14 = 40.0, Age_15_29 = 50.0, Age_30_44 = 45.0)
L <- matrix(c(0.00, 0.72, 0.48, 0.97, 0.00, 0.00, 0.00, 0.96, 0.00), nrow = 3, byrow = TRUE)
M <- c(Age_0_14 = 2.5, Age_15_29 = 4.0, Age_30_44 = 1.5)

# Natural projection and migration adjustment
natural_projected <- as.vector(L %*% w0)
final_projected <- natural_projected + M
names(final_projected) <- names(w0)

cat("Final Projected Total (t=15):", sum(final_projected), "thousandn")
print(round(final_projected, 2))

# Dominant Eigenvalue
eig <- eigen(L)
dominant_lambda <- Re(eig$values[1])
cat("Dominant Eigenvalue:", round(dominant_lambda, 4), "n")
annual_r <- log(dominant_lambda) / 15
cat("Intrinsic Annual Growth Rate (r):", round(annual_r * 100, 3), "%n")

Implementation in Python

Below is a clean Python implementation executing cohort-component matrix multiplication, migration adjustments, and dependency metrics:

def project_cohort_component(base_pop, leslie_matrix, net_migration):
    k = len(base_pop)
    natural_proj = [sum(leslie_matrix[i][j] * base_pop[j] for j in range(k)) for i in range(k)]
    final_proj = [natural_proj[i] + net_migration[i] for i in range(k)]
    return natural_proj, final_proj, sum(base_pop), sum(final_proj)

base_pop = [40.0, 50.0, 45.0]
leslie_L = [[0.00, 0.72, 0.48], [0.97, 0.00, 0.00], [0.00, 0.96, 0.00]]
migration_M = [2.5, 4.0, 1.5]

nat_res, final_res, total_0, total_15 = project_cohort_component(base_pop, leslie_L, migration_M)
print(f"Base Total: {total_0:.2f}k | Final Total: {total_15:.2f}k")
for idx, (nat, fin) in enumerate(zip(nat_res, final_res), 1):
    print(f"  Cohort {idx}: Natural={nat:.2f}k, Final={fin:.2f}k")
cdr_base = base_pop[0] / sum(base_pop[1:])
cdr_proj = final_res[0] / sum(final_res[1:])
print(f"Base CDR: {cdr_base:.4f} | Projected CDR: {cdr_proj:.4f}")

Frequently Asked Questions

What is the difference between a population projection and a population forecast?

A population projection is an objective calculation showing how a population changes under explicit mathematical assumptions of fertility, mortality, and migration, making no claim of inevitability. A population forecast designates a specific projection scenario as the single most probable future outcome. Statistical agencies publish multiple projection variants to provide scenario bounds rather than deterministic forecasts.

Why is the cohort-component method preferred over exponential curve fitting?

Exponential models treat population as an undifferentiated aggregate scalar, ignoring age structure. The cohort-component method models age-specific biological realities: fertility concentrates within ages 15–49, mortality rises exponentially at older ages, and migration concentrates among young working adults. Disaggregating cohorts captures demographic momentum and dependency shifts.

How does demographic momentum influence population projections?

Demographic momentum is the tendency of a population to continue growing after fertility falls to replacement levels (2.1 children per woman). When past high fertility creates a large youth cohort entering childbearing ages, their collective births maintain total growth for several decades until age distributions stabilize.

What is the difference between deterministic and probabilistic population projections?

Deterministic projections model alternative trajectories using predefined scenario variants (e.g., low, medium, and high fertility), leaving variant probabilities unquantified. Probabilistic projections, implemented by the United Nations Population Division, use Bayesian time-series models to produce thousands of stochastic paths, yielding explicit statistical confidence intervals (such as 80% and 95% predictive bands).

How do demographers estimate net migration when administrative records are incomplete?

When direct migration records are missing, demographers use the residual method via the demographic balancing equation: net migration equals total intercensal change minus natural increase (NM = [P(t + n) − P(t)] − [B − D]). Forward and reverse survival techniques allocate residual migration into age-sex cohorts.

How far into the future can population projections remain reasonably accurate?

Projection accuracy deteriorates significantly beyond a 20- to 25-year horizon. Projections spanning 10 to 15 years maintain high reliability for adult cohorts because those individuals are already alive at baseline. Beyond 25 years, projections depend increasingly on unborn cohorts, compounding errors in fertility and migration assumptions.

Authoritative Standards and Primary Sources

Data: Comparative Specifications of Official National and International Population Projection Frameworks

Original compilation comparing the methodological parameters, projection horizons, baseline census years, component modeling frameworks, and uncertainty specifications across major national and international demographic statistical agencies.

Period covered: 2022/2100. Coverage: Global, United States, United Kingdom, European Union, Canada, Australia.

Comparative Specifications of Official National and International Population Projection Frameworks
Statistical AgencyProjection Horizon (years)Baseline Census Year (year)Core Projection MethodologyUncertainty FormulationFertility Modeling ApproachMortality Modeling ApproachMigration Component Treatment
United Nations Population Division (UN DESA)762024Discrete-time single-year cohort-component model (CCMPP)Probabilistic Bayesian hierarchical models (80% and 95% predictive intervals)Bayesian hierarchical model on Total Fertility Rate with double-logistic trajectoryBayesian random walk model for life expectancy gains at birth (e0)Deterministic medium-term convergence with historical administrative flow extrapolation
U.S. Census Bureau (NP Projections)772023Single-year cohort-component model by sex, race, Hispanic origin, and nativityDeterministic multi-scenario variants (Main, High Imm, Low Imm, Zero Imm)Parameterized double-exponential cohort fertility rate splinesExtrapolated cause-of-death mortality rate models with sex differentialsEconometric and legal visa category modeling across distinct immigration scenarios
UK Office for National Statistics (ONS)502022Single-year cohort-component model across 4 constituent UK nationsPrincipal projection plus 8 standard deterministic sensitivity variantsLong-term completed family size convergence by age groupAnnual percentage reduction rates in age-specific mortality by cohortExpert panel calibration combined with historical moving-average net inflows
Eurostat (EUROPOP)772023Cohort-component framework across EU-27 member states and EFTA nationsDeterministic baseline scenario with sensitivity variantsLong-term convergence model toward benchmark European fertility levelsModified Lee-Carter mortality extrapolation model with common EU convergenceDynamic gravity and partial-equilibrium socio-economic convergence models
Statistics Canada502023Multiregional cohort-component model by province, territory, age, and sexDeterministic low, medium (M1-M5), and high growth scenariosCohort-specific logarithmic regression and target fertility assumptionsLi-Lee multi-population coherent mortality forecasting modelTwo-tier modeling: national targets allocated via historical interprovincial matrices
Australian Bureau of Statistics (ABS)492022Cohort-component model across states, territories, capital cities, and regionsDeterministic Series A (high), Series B (medium), and Series C (low)Age-specific fertility rate trajectories calibrated against TFR assumptions (1.45 to 1.75)Short-term and long-term decline rate assumptions in age-specific death ratesNet Overseas Migration (NOM) targets with regional inter-state origin-destination matrices

Follow CASRAI

Research-administration guidance, standards updates and independent tool reviews.

Ask CASRAI · free to try

Ask about Population Projections: Models, Formulas & Guide

2 free questions a day, no account.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →