Demographics: The Population Curve
Why almost every developed country is below replacement fertility, why aging is already locked in, and why migration now decides whether many rich countries still grow.
(3.31 in 1990; about 5 in the 1960s)
(up from a record low of 0.72 in 2023; Taiwan's 2025 rate was 0.695)
(2025: 30% and 20%)
A note on data freshness. Demographic indicators lag. National statistics agencies typically publish yearly figures a few months to a year after the year ends, and the World Bank harmonizes them across countries with a further delay. The charts on this page use the latest World Bank releases, which do not all end in the same year. Age-structure series - the working-age and over-65 shares - run through 2025. Fertility and life expectancy lag a further year and end at 2024, because they rest on vital-registration data that takes longer to collect and harmonize. The text uses newer national figures where they exist, and the UN's estimate for the world: 2.25 in 2024, where the World Bank series in the chart below shows 2.19. For slow-moving forces like fertility and aging, this lag rarely changes the picture. The trend is what tells the story, and the trend has been consistent for decades.
The raw picture
Fertility has fallen in every major region of the world over the past sixty years. That is not a forecast - it is a measurement. In the 1960s, the average woman on Earth had about five children. In 2024, she had just over two. The world is now close to replacement fertility, the level at which a population - absent migration - neither grows nor shrinks over time, and two-thirds of humanity lives in countries already below it.
Why is fertility falling? Several causes converge, in a similar order wherever the transition has happened. Women's education and labor-force participation rise, contraception becomes reliable, and people move from countryside to city, where children stop being an economic asset and become a cost. The material standard for what counts as "ready to have a child" rises too. Once expectations about family size shift, they reinforce the other causes in ways that are very hard to reverse. Every country that has industrialized and urbanized has seen the same pattern. The variation is the speed.
Below replacement is not, by itself, a crisis. Many demographers argue some moderate decline is desirable - lower resource consumption, higher capital per worker, a narrower environmental footprint. The crisis is the timing mismatch. Children take twenty years or more to enter the labor force. By the time a country notices it has too few workers, the cohorts that would have been those workers went unborn two or three decades earlier. Few policy levers close that gap on a useful timescale; the fastest is migration, discussed below.
How countries compare
"Below replacement" is a single number that hides enormous variation. Each country in the table below tells a different version of the same demographic transition - some early, some late, some sharp, some gentle. Replacement fertility for developed countries is approximately 2.1 (slightly higher in countries with higher mortality).
Developed-economy fertility has been below replacement for decades, and even the countries that held up longest are slipping. France went from about 2.0 in 2010 to 1.56 in 2025, and Sweden from 1.98 to 1.43 in 2024. The countries that hold up best - Israel (near 2.9), the US, France - share some combination of higher migration, stronger family-policy infrastructure, or specific cultural factors that other countries cannot easily import. Most of the developed world is in some version of Japan's or Italy's situation, and several major countries (Korea, China, Spain) are sharper.
Aging is mathematically locked in
Once fertility has been below replacement for twenty-plus years, aging is no longer a forecast - it is arithmetic. The cohorts that would have been working-age twenty years from now were not born twenty years ago. The main options at that point are to wait decades for the cohorts to pass, import workers, keep people working longer, or accept a smaller economy.
Japan started this conversation first. It went below replacement in 1974, hit 10% over-65 in 1985, 20% in 2005, and 30% in 2025. By 2050, projections place it near 38%. That is not a model artifact - the people who will be 65+ in 2050 are alive today, the youngest of them now in their early forties. The arithmetic does what it does.
South Korea's curve is the same shape as Japan's, twenty years delayed and steeper. Italy is on Japan's path with about a fifteen-year lag. Germany sits on a softer version because of its relatively higher migration. China, having gone below replacement later but harder, is now compressing into one decade what Japan spread over three. The United States, back below replacement only since 2008 and still receiving net migration, if much less of it, is the gentlest case in this group - which is to say its situation looks like Italy's, just delayed by twenty-five years.
The labor-force consequences are concrete. The old-age dependency ratio - people aged 65 and over per hundred people of working age - tells the pension and healthcare-financing story. In 1960 the average across high-income countries was about 13. Today it is about 32. By 2050 it is projected at about 45, and above 70 in Japan, Korea, and Italy; Germany's reaches about 54. Pension systems built on the 1970s ratio of 4-5 workers per retiree do not survive a future with 1.5 workers per retiree without major adjustment.
Migration as the load-bearing variable
Once aging is locked in, migration is the fastest way to add workers. Everything else - pronatalist incentives, automation, raising retirement age, accepting smaller economies - takes decades, runs into political limits, or doesn't fully substitute for the missing tax base that pays the pensions and the elder-care bills. Pronatalist policies, where they have been seriously tried, move fertility by maybe 0.1 to 0.2 - useful at the margin, nowhere near enough on their own. Automation substitutes for some labor but not for the social-insurance contributions that retired workers depend on.
That makes migration policy far more consequential than it was fifty years ago. A society that admits few migrants accepts a shrinking workforce, slower overall growth, and a heavier pension burden on each worker. A society that admits many takes on the costs and political strains of integration, and the fiscal effect depends on who arrives and whether they find work. There is no costless option.
The numbers are large. The UN projects immigration to be the main source of population growth through 2054 in 52 countries and areas, including the United States, Canada, and Australia; in Italy and Germany, it delayed the point at which population peaked. In the US, net international migration peaked at 2.7 million in 2024 and fell to 1.3 million in 2025, and the Census Bureau expects about 321,000 in 2026 if current trends continue. Germany's fell from 1.46 million in 2022 to 235,000 in 2025.
The paths from here
Dated September 2026, with net migration to the United States, Britain, and Germany sharply lower in 2025 and South Korea's births rising for a third year. There is no costless way through this. Every aging society has to make some combination of the choices below. The combinations differ; the menu does not.
Migration as the valve
Sustained net inflows to maintain working-age population. Most advanced economies are already doing this with varying levels of acknowledgment.
Will it happen? Already happening, but the direction has split. Japan's foreign population is growing at a record pace, while net migration to the United States, Britain, and Germany fell sharply in 2025. Lower inflows mean slower growth in the workforce and in total output. Supporters of lower immigration argue that housing, wages, and public services come first, and expect higher participation and automation to fill part of the gap. Where deaths already exceed births, a lasting fall means a shrinking population.
Pronatalism
Government incentives for families: child allowances, parental leave, childcare subsidies, housing support, sometimes direct cash bonuses for births.
Will it happen? Tried widely (Hungary, France, Korea, Japan, Singapore, Russia), usually with small positive effects - on most estimates 0.1 to 0.2 fertility points. France's long run near 2.0, which ended in the mid-2010s, is partly attributed to long-running family policy. Hungary spends heavily on families; its rate rose from 1.23 in 2011 to 1.61 in 2021, then fell to 1.31 in 2025. Korea's rise since 2024 is too recent to judge. The UN rates a return to 2.1 within 30 years as highly unlikely for the 24 ultra-low-fertility countries whose populations have already peaked.
Raise retirement age
Push the official retirement age up, which directly improves the working-age-to-retiree ratio and the pension-system arithmetic.
Will it happen? Already happening across the OECD, slowly and not in a straight line. Denmark legislated in 2025 to raise its pension age to 70 by 2040; France, after mass protests in 2023, suspended its planned rise until January 2028. Most advanced economies are likely to keep drifting the retirement age upward by 1-2 years per decade. Of the pension fixes, it is the most widely used.
Productivity / automation
Each remaining worker becomes more productive, partially substituting for the missing workers.
Will it happen? Yes for some sectors, no for others. White-collar tasks (admin, basic analysis, customer service) automate well. Healthcare, elder care, hospitality, agriculture, and construction substitute much less easily. Net effect is helpful but uneven, and the sectors most under labor pressure (elder care above all) are the ones that resist automation hardest. AI may shift this balance; how much is genuinely unknown.
Accept smaller economies
Some societies will simply choose, explicitly or not, to be smaller. Lower growth, lower immigration, preserved cultural cohesion, eventual decline in global weight.
Will it happen? Partly happening in Japan, whose population is shrinking even as it admits more foreign workers; Italy's holds steady only through immigration. Korea may end up there by default. Supporters see a smaller, older, richer, and more cohesive society as a coherent national choice. Critics point to the cost in geopolitical weight and economic dynamism: such a country may count for less abroad, whatever its quality of life at home.
Benefit cuts
The unspoken sixth lever. When contributions no longer cover promises, pension benefits get reduced - sometimes overtly, more often through inflation erosion or eligibility changes that look small individually and compound dramatically.
Will it happen? Gradually, in many places. The main US Social Security trust fund is projected to run out in late 2032, after which incoming revenue would cover only about 78% of scheduled benefits unless Congress acts. Many other OECD countries have already begun similar slow erosions through formula changes. This is the path that gets taken when politicians can't agree on any of the others.
Most advanced economies will combine all of them: ongoing migration sufficient to slow but not stop working-age decline, pronatalist policies that nibble at the edges, retirement age drifting upward by a year or two per decade, modest productivity gains, and gradual benefit erosion. A few countries (Japan, perhaps Italy and Korea) will end up at the smaller-economies end of the menu. The variable is not whether these adjustments happen - they will - but how each country distributes the costs.
Where analysts disagree
The framing above is the mainstream demographic reading, but it's not the only credible one. Lant Pritchett, Daron Acemoglu and Pascual Restrepo, and a growing productivity-led literature hold versions of the arguments below, and the data does not rule them out. A reader who only ever hears "demographic collapse" is getting an incomplete picture.
Fertility could rebound
Fertility has rebounded before - the postwar baby boom is the most-cited example - and recent declines have not all run one way. Czechia rose from 1.43 in 2011 to 1.83 in 2021 before falling back. South Korea's births have risen for 25 straight months, lifting its rate from 0.72 in 2023 to 0.80 in 2025.
What would drive it: If economic and housing conditions for young families improve, fertility could plausibly rebound from current lows. Ten or fifteen years of data is not enough to declare any current rate the new permanent floor.
Aging is not the same as decline
A smaller, older, better-educated, more productive population can be richer than a larger, younger, less-educated one. Per-capita GDP is what determines living standards, not headcount.
Held by: economists in the productivity-led literature, including Daron Acemoglu and Pascual Restrepo. Across countries, those two found no negative link between aging and growth in GDP per person; faster-aging countries have if anything grown faster, which they tie to quicker adoption of automation. The aging-is-decline framing implicitly assumes more humans is automatically better, which is at least debatable.
Migration is a much bigger lever than projections assume
Historical migration responses to economic gradients have been very large - tens of millions of people over decades. Projections that assume current migration rates often understate the substitution potential.
The catch: Whether developed societies politically choose to use the lever at the scale demographically required is a different question. But the upper bound is much higher than the conversation usually treats it.
The global picture is redistribution, not collapse
At the global level the demographic story is not "decline" - it is redistribution. Sub-Saharan Africa is projected to add more than 800 million people by 2050. South Asia continues to grow. The global working-age population is projected to keep expanding beyond 2050.
Why it matters: "Collapse" framing is regional and selective. Whether the resulting people end up as global labor force, migrants, or stay home matters enormously, but the species-level picture is far from a population crash.
Lower fertility might be optimal
The UN's central projection - a peak of about 10.3 billion people in the mid-2080s - assumes fertility keeps falling only gradually. Lower-fertility scenarios mean less climate stress, lower resource consumption, easier capital-per-worker math, and arguably a more sustainable global trajectory.
The framing point: The "we must raise fertility" message carries an implicit assumption - more humans is always better - that is not self-evident. It is contested by environmental economists, some demographers, and a growing segment of the public who treat smaller family size as a positive choice.
None of this dismisses the concerns earlier in this piece. It says the timing, severity, and distribution of consequences depend on policy and economic responses that have not yet happened. A reading that fits both: aging is real and locked in for the next twenty to thirty years, and its economic consequences look manageable if countries find the right combination of migration, productivity, and retirement-age adjustments. At the global level, the numbers describe a redistribution of people toward Africa and South Asia rather than a collapse: the UN expects world population to keep growing until the mid-2080s.
What this means for you
Demographics is the slowest-moving large force in political economy. It will not change much next year. But the direction over your working life is fixed in ways you can position around without panic.
If you're choosing a career
Sectors that need warm bodies - elder care, healthcare, hospitality, agriculture, construction, and skilled trades - are likely to face lasting labor shortages and, accordingly, rising relative wages. Sectors most exposed to AI and automation substitution (mid-level office work, basic analysis, routine knowledge work) may see the opposite. The combination of aging labor markets and improving AI is likely to compress the value of degrees that mainly signal competent paperwork-handling and raise the value of work that physically requires a person.
If you're planning where to live
Aging societies have features that may be desirable (low crime, strong social services, good healthcare for those who can access it, quiet) and undesirable (high tax burden, slow innovation, healthcare strain, less dynamism). Younger societies are the inverse. There is no universal answer; choose deliberately. If you are choosing a country to build a long career or business in, weigh the demographic curve as seriously as you weigh the tax rate. Twenty-year decisions need twenty-year data.
If you're thinking about retirement
Public pension systems in aging countries are mathematically strained. Plan as if benefits will be 15-25% lower than currently promised, or as if the retirement age will be 2-4 years higher by the time you reach it. Personal savings, employer retirement matches, and broad asset-class investing matter more for your generation than they did for your parents'. This is not pessimism - it is the same arithmetic the pension actuaries publish.
If you're thinking about family
Whatever you decide is yours alone. Do not let fertility-policy debates frame what should be a personal life choice. Most policy interventions that aim to raise fertility are weakly effective; the strong driver of family size is what you and your partner actually want. If a society's pronatalist messaging starts to feel like pressure, that is a separate problem from your own preferences, and worth recognizing as such.
If you're thinking globally
This century's biggest population stories are in Africa, India, and Indonesia. Anyone planning a long-term career, business, or investment positioning that ignores those regions is operating with a 1990s map. Conversely, anyone betting on the continued central economic gravity of aging societies (Western Europe, Japan, Korea) without factoring in the demographic drag is making the opposite mistake. The geography of growth is not what it was, and it is moving fast.
The mechanics behind this
Demographics interacts with almost every other topic on this site. Three fundamentals that are most useful as background:
