How Population Projections Actually Work (and How Much to Trust Them)

Numbers like “9.66 billion people by 2050” get quoted as if someone had seen the future. Nobody has. A population projection is conditional arithmetic: if fertility, mortality and migration follow assumed paths, then the population follows. The assumptions are the forecast. Knowing how the machine works tells you exactly how much weight each number can bear — and the answer differs enormously between 2035 and 2100.

The cohort-component method

Virtually every serious projection, including the UN's, uses the same engine. Start with the population counted by age and sex — the same structure shown in the pyramids on this site's country pages. Then, year by year: age everyone forward, apply age-specific death rates, apply age-specific birth rates to women of childbearing age, add or subtract migrants. Repeat to 2100. The method itself is almost mechanical; everything interesting lives in the assumed rates.

Why near-term projections are strong

The next two decades of demography are largely already determined, because the people involved are already alive. Everyone who will be over 25 in 2050 has been born; the potential parents of the 2030s are in school today. Short of catastrophe, aging forward is the most predictable process in social science — which is why counts like “working-age population in 2040” are treated as near-facts by planners, and why this site presents 2030 figures with some confidence.

Where the uncertainty lives

Fertility. Mortality changes slowly and migration nets out at world level, but a difference of half a child per woman, compounded over three generations, moves the 2100 total by billions. That is why the UN publishes variants: the medium path — the one used across this site — reaches about 10.18 billion in 2100 after peaking near 10.28 billion in the 2080s, while its published low and high fertility variants land billions apart by century's end. The medium variant is best read as a central scenario, not a prediction: useful for direction and rough magnitude, unreliable for precision at long range.

The track record

For horizons of ten to twenty years, UN world projections have historically landed close to the eventual counts — the momentum described above does most of the work. The misses concentrate where assumptions had to do the heavy lifting: the pace of fertility decline in specific countries has repeatedly surprised in both directions, most recently falling faster than assumed across much of East Asia. Successive revisions have nudged the projected peak earlier and lower. The lesson is not that projections are wrong but that they are living documents — each revision folds in the newest data, which is why this site rebuilds automatically when the sources publish.

How to read a projection like a demographer

Three habits: check the horizon (2035 is close to arithmetic; 2100 is scenario-space); check the variant (a “could reach” headline usually quotes the high one); and check what is already locked in (statements about people already born are the reliable core). Applied to the numbers on this site: the 2050 figure of 9,664,378,587 deserves real confidence; the 2100 figure deserves a healthy “roughly, if trends hold.”