Population Projections

Core Idea

Population projections calculate future population size, age structure, and composition by applying assumed future rates of fertility, mortality, and migration to a current population. The cohort-component method — the standard approach — advances each age-sex group forward in time by applying age-specific survival rates, adds newborns using age-specific fertility rates, and adjusts for net migration. Projections are not predictions; they are conditional statements — "if fertility, mortality, and migration follow these assumed paths, then the population will be X." The UN produces projections under multiple variants (low, medium, high, constant-fertility) to bracket uncertainty. Projection accuracy depends critically on assumptions about future fertility, which is the most uncertain and consequential component.

How It's Best Learned

Build a simplified cohort-component projection by hand: start with a population distributed across 5-year age groups, apply survival ratios and fertility rates for one projection interval (5 years), and observe how the age structure transforms. Then vary the fertility assumption and see how dramatically the projected population diverges after just a few intervals.

Common Misconceptions

• Population projections are not forecasts or predictions — they are scenarios conditional on explicit assumptions. The UN's "medium variant" is widely treated as a forecast, but it is the midpoint of a range, not a best guess about what will happen.
• Long-range projections (50+ years) are inherently uncertain, especially regarding fertility. Small differences in TFR assumptions compound dramatically over decades, producing enormous divergence between high and low variants.

Explainer

From life tables and fertility measures, you have the tools to describe current mortality and fertility patterns. Population projections extend those tools into the future by asking: if current or assumed future rates persist, what will the population look like in 10, 50, or 100 years? The answer is computed through the cohort-component method, which is essentially a bookkeeping exercise applied to the demographic balancing equation, disaggregated by age and sex.

Start with the current population arranged in 5-year age-sex groups (e.g., males 0-4, males 5-9, ... , females 0-4, females 5-9, ...). To advance one 5-year interval: apply age-specific survival ratios (derived from the projected life table) to move each cohort forward — the males aged 0-4 become the males aged 5-9, reduced by mortality. Apply age-specific fertility rates to women in the childbearing ages to compute the number of births, split by sex using the assumed sex ratio at birth, and survive these newborns through the first age interval to produce the 0-4 age group. Add net migrants by age and sex. Repeat for each interval. The output is a complete age-sex distribution at each future time point.

The method is mechanically straightforward, but the difficulty lies entirely in the assumptions. Three components must be projected: future mortality (how fast will life expectancy improve?), future fertility (will TFR remain below replacement, rise, or fall further?), and future migration (how large and age-distributed will net flows be?). Of these, fertility is by far the most consequential and the most uncertain. A difference of 0.5 in TFR — say, 1.6 versus 2.1 — seems small but compounds across generations. Under 2.1, each generation roughly replaces itself; under 1.6, each generation is about 24% smaller than the last. Over 50 years (two generations), this produces dramatically different populations. Mortality improvements, while important for quality of life, typically add years at older ages and have a much smaller effect on total population size.

The United Nations Population Division produces the most widely cited global projections, updated biennially. They publish multiple variants: the medium variant assumes fertility converges toward replacement in most countries; the high variant assumes it converges 0.5 children higher; the low variant, 0.5 children lower. The constant-fertility variant shows what happens if current rates persist unchanged. These variants bracket a wide range of possible futures. The medium variant is commonly treated in media and policy as a forecast, but it is more accurately understood as the central scenario in a range of plausible outcomes. Probabilistic projections, which assign probability distributions to future fertility and mortality, offer a more honest representation of uncertainty — the 95% prediction interval for world population in 2100 spans roughly 9 to 12 billion, a range of 3 billion people.