Why this series starts with mechanics, not headlines
Population change gets reported as a mood: a country is “aging,” a city is “booming,” a region faces a “migrant crisis.” Those words describe real pressure, but none of them is a mechanism. This article is the opening piece in a series on demography, migration, and urban futures, and its job is narrower than a forecast — it is to lay out, from scratch, how the three load-bearing pieces of population science actually work: the documented pattern of demographic transition, the accounting method that turns today’s population into tomorrow’s projection, and the empirically grounded mechanics of why people move. Everything downstream — aging budgets, housing shortages, climate displacement — follows from these three pieces rather than from intuition about them.
Three registers get kept separate throughout: verified fact (a rate that has been measured and published), method (a defined procedure, such as how the United Nations actually builds a projection), and analysis or scenario (a reasoned but conditional extension of the above). Predictions, where they appear, carry a horizon, stated assumptions, and a condition under which they would be wrong.
The demographic transition is a pattern, not a law
Demographic transition theory describes an empirically documented sequence: a population moves from a regime of high mortality and high fertility, through a period where mortality falls first and fertility keeps lagging behind it, into a new regime where both are low. Dudley Kirk’s often-cited synthesis calls it “one of the best documented generalizations in the social sciences,” built from historical vital-registration data across dozens of countries rather than derived from first principles [3]. The theory does not claim every country repeats an identical schedule, and it does not, by itself, predict the size of the lag between falling mortality and falling fertility — that lag is set locally by public-health investment, women’s education and labor-force participation, the cost of raising children, and access to contraception. What is documented is the ordering: mortality decline (driven by control of infectious disease, clean water, and improved nutrition) precedes fertility decline in essentially every national case on record, producing a temporary bulge of rapid population growth in the interval between the two curves [3].
That interval matters more than the theory’s name suggests, because it is where population momentum is built. A large cohort born during high-fertility, falling-mortality years does not disappear when fertility subsequently falls — it ages forward as a permanent bulge that will, decades later, become a large elderly cohort supported by whatever smaller cohort follows it. This is not a prediction; it is arithmetic on cohorts already alive.
The current global position, as measured rather than modeled, is stark: the UN’s 2024 revision recorded global fertility at 2.25 births per woman, down from 3.31 in 1990, with 131 of 237 countries and areas — 55 percent, representing 68 percent of the world’s population — already below the replacement threshold of roughly 2.1 births per woman [2]. Close to a fifth of all countries and areas, including China, Italy, South Korea, and Spain, are recorded at “ultra-low” fertility below 1.4 [2]. These are measured figures from vital registration and survey data, not projections — the projection element enters only when this measured starting point gets carried forward, which is the next mechanism.
How a population projection is actually built
“The UN projects population X will fall to Y by 2100” is often read as a single forecasted number, as though a model simply extrapolates a trend line. That is not what happens, and the difference matters for how much confidence the number deserves.
The UN Population Division builds its projections using the cohort-component method: rather than extrapolating total population, it starts from the population by single year of age and sex in a base year, and ages every cohort forward one year at a time by applying age-specific and sex-specific survival probabilities (mortality), age-specific fertility rates applied to women of reproductive age (which generates the next year’s births), and net international migration assumptions applied by age [1]. Formally, for an age group
where
This is why the method captures population momentum as a structural fact rather than a modeling assumption: even a country whose fertility rate dropped to exactly replacement level today would keep growing for decades, because its current age structure already contains an outsized number of women about to enter childbearing age. The momentum is visible directly in the age pyramid; it requires no forecasting assumption to exist [1].
Three of the UN’s own qualifications matter for reading any projection responsibly. First, the near-term numbers (five to fifteen years out) are far more reliable than end-of-century numbers, because the people who will be of working age in 2040 are already born and counted — only mortality and migration assumptions remain uncertain, not fertility. Second, migration assumptions are the least constrained input: they are set from recent observed trends and policy judgment, not from a comparable biological regularity to fertility or mortality. Third, the UN publishes a full probabilistic range (low, medium, high variants) precisely because a single point estimate for 2100 embeds decades of compounding assumption error; the medium variant is a central scenario, not a claim of certainty [1]. The 2024 revision projects global fertility converging toward replacement level around 2036, with fertility contributing little further to world population growth after that point, growth from 2024 onward increasingly driven by population momentum among cohorts already alive rather than by any assumed future change in fertility rates [2].
Why people move: push, pull, and what actually filters between them
Migration mechanics have a documented backbone that predates modern data by half a century. Everett Lee’s 1966 framework separates the decision to migrate into four interacting factor sets: conditions at the place of origin, conditions at the destination, a set of intervening obstacles between them, and the personal characteristics of the person deciding [4]. The framework’s durability comes from resisting single-cause explanations: a wage gap between two places is a necessary condition for economic migration, not a sufficient one, because distance, cost, legal barriers, information, and family circumstance all filter whether that gap actually produces movement [4].
A second mechanism, documented later but equally load-bearing, is cumulative causation: migration streams are not static once started, because each act of migration changes the social conditions under which the next migration decision gets made. Massey and colleagues formalize this as migrant networks reducing the cost and risk of moving for people connected to those who already moved — an existing community at the destination lowers the effective “intervening obstacle” for the next migrant from the same origin, which is why migration flows between specific origin-destination pairs tend to grow and persist rather than track wage differentials alone [5]. This explains a pattern that a pure push-pull wage model cannot: migration streams that continue, or even accelerate, after the original wage gap that started them has narrowed.
Framed this way, most migration is not a single undifferentiated flow but several distinguishable regimes operating on different mechanics: labor migration responding to wage and employment differentials filtered by networks and policy; family reunification migration responding to established networks rather than current wage gaps; displacement from conflict or disaster responding to acute push factors with weak filtering by cost; and — the subject of this series’ later pieces — climate-linked internal migration, addressed below. Collapsing these into one number obscures which mechanism is actually operating in a given flow.
Aging, cities, and housing: where the mechanics land
The transition and projection mechanics above are not abstract once population momentum reaches the older end of the age pyramid. The OECD recorded the old-age dependency ratio (people 65+ per 100 people aged 20–64) rising from 19 percent in 1980 to 31 percent in 2023 across its member area, a measured trend built directly from the age-structure accounting described above, and projects it reaching roughly 52 percent by 2060 under current fertility, mortality, and migration assumptions — a projection, not a certainty, since it depends on those same three inputs staying near their current trajectories [8]. The OECD’s own analysis separately notes that in more than a quarter of member countries the working-age population itself is set to shrink by more than 30 percent by 2060, a direct consequence of the below-replacement fertility recorded above working through the cohort-component recursion over multiple decades [8].
Cities absorb the migration side of this system directly. UN-Habitat’s 2022 World Cities Report records that urban areas already hold 55 percent of the world’s population, a measured figure, and its own scenario work projects growth to roughly 68 percent by 2050, concentrated mainly in Africa and parts of Asia — a scenario built from current urbanization-rate trajectories rather than a fixed law of urban growth [7]. The same report is explicit that this is not simply a story of rural people arriving in cities faster: urban growth in the projection is driven jointly by natural increase within existing urban populations, reclassification of previously rural settlements as they grow, and net rural-to-urban migration, with the relative weight of each varying sharply by region [7].
The housing mechanism connecting migration to urban pressure is mostly a matter of timing mismatch rather than aggregate scarcity: housing stock is built on multi-year to multi-decade cycles, while migration and natural-increase inflows can shift on a scale of months. A city whose net inflow accelerates faster than its permitting and construction pipeline responds will show rising costs and crowding regardless of whether its long-run housing supply is theoretically adequate. UN-Habitat frames responsive urban and territorial planning explicitly as one of eight required pathways to a sustainable urban future precisely because this timing mismatch, left unmanaged, is what converts an ordinary migration inflow into a housing crisis [7].
Climate mobility: mostly internal, mostly about existing pressure
Climate-linked mobility is frequently discussed as a future wave of cross-border “climate refugees,” but the best-documented modeling to date describes something narrower and already measurable in its early effects: internal displacement toward less-exposed regions within the same country. The World Bank’s Groundswell modeling, covering Latin America, Sub-Saharan Africa, South Asia, East Asia and the Pacific, the Middle East and North Africa, and Eastern Europe and Central Asia, projects that up to 216 million people could be displaced within their own countries by 2050 under a pessimistic scenario combining water scarcity, declining crop productivity, and sea-level rise, with these are explicitly scenario projections built on assumed emissions and development pathways, not fixed forecasts [6]. The same report’s central finding about mitigation is a conditional analytical claim rather than a prediction: coordinated climate and development action could reduce projected internal climate migration by as much as 80 percent relative to the pessimistic pathway, meaning the 216 million figure describes a policy-contingent upper bound, not a fixed destiny [6].
Mechanistically, this form of mobility routes through the same push-pull and intervening-obstacle framework described above rather than constituting a separate phenomenon: declining agricultural productivity and water scarcity are push factors at origin, urban labor markets and existing family or ethnic networks in receiving cities are pull factors and network-based obstacle-reducers, and the same cumulative-causation dynamic applies once an initial displaced group establishes itself at a destination [4, 5]. Because the displacement in Groundswell’s modeling is overwhelmingly internal rather than cross-border, its practical landing point is the same urban housing and infrastructure system already under the timing-mismatch pressure described in the previous section — climate mobility does not create a new kind of urban pressure so much as it adds another inflow variable to a housing and planning system that already has to absorb ordinary rural-to-urban and natural-increase growth [7, 6].
What is fact, what is method, and what is still open
Pulling the registers apart one more time: it is a measured fact that global fertility has fallen from 3.31 to 2.25 births per woman since 1990 and that most of the world’s population now lives below replacement fertility [2]. It is documented pattern, not law, that mortality decline precedes fertility decline in the transition sequence [3]. It is defined method, not extrapolation, that population projections are built by aging measured cohorts forward through survival, fertility, and migration assumptions one year at a time [1]. It is empirically grounded mechanism, not a single-cause story, that migration responds to push and pull factors filtered through intervening obstacles and amplified by migrant networks once a stream exists [4, 5]. And it is conditional scenario, explicitly bounded by assumed policy pathways, that up to 216 million people could face internal climate displacement by 2050, a figure the source itself treats as reducible by up to 80 percent under different collective action [6].
The disconfirmation conditions worth stating plainly: the OECD’s 2060 dependency-ratio figure would be wrong if fertility, mortality, or net migration in member states diverges materially from current trends over the next three decades [8]; the UN’s own near-term population figures are far more robust than its end-of-century figures because the cohorts driving the next fifteen years are already born [1]; and the Groundswell displacement figures are explicitly an upper bound under a specific emissions and development pathway, not a forecast independent of policy choices made between now and 2050 [6]. None of this collapses into a single number worth quoting without its method attached — which is the reason this series opens with the mechanics rather than with the headline.
This opening piece has deliberately stayed at the level of mechanism rather than moving to country-level or city-level cases, because the later pieces in this series depend on readers holding these distinctions rather than reaching for a single average. A national fertility rate quoted without its transition stage tells a reader little about whether a country is entering, in the middle of, or decades past its period of rapid population momentum. A projected population figure quoted without its variant (low, medium, high) and its time horizon tells a reader little about how much of that figure is measured cohort arithmetic and how much is an assumption about migration or fertility three decades out. And a migration figure quoted without separating labor flows, family reunification, displacement, and climate-linked internal movement collapses four different mechanisms, responsive to four different sets of policy levers, into one number that answers none of the practical questions a city or a national government actually needs to act on. Subsequent articles in this series will apply these same three registers — fact, method, and scenario — to specific regions, specific cities, and specific policy interventions, rather than introducing new mechanics from scratch each time.
