Global Fertility Has Fallen Below Replacement, Though Population Keeps Growing

The world’s total fertility rate is now below the level needed to replace its population, according to economist Jesús Fernández-Villaverde. He put the global rate at about 2.17 or 2.18 children per woman, against a replacement rate of roughly 2.2. That does not mean the world’s population is already shrinking. Demographic momentum—the fact that parents and grandparents are still alive even as each generation has fewer children—can keep population growing for decades after fertility falls below replacement.

Fernández-Villaverde illustrated momentum with a family in which two parents have one child. The family grows from two people to three; it does not shrink until the parents die. At the global level, he argued, many older generations have not yet passed away, even as births have fallen. He estimated that world population would continue to rise until around 2055, based on the roughly 30-year period that demographic momentum has historically lasted.

That estimate depends on fertility not turning around. “We don’t know,” Fernández-Villaverde said of the limits of demographic forecasting. People could change their behavior, he stressed, but based on what he had observed, the decline did not appear to be stopping. He argued that people underestimate how much a sustained fall below replacement could reshape societies, economies and geopolitics.

The numbers Russ Roberts and Fernández-Villaverde discussed suggest why the shift can be difficult to see. Roberts cited about 138 million births worldwide in 2012; Fernández-Villaverde confirmed that figure and put the current annual number at about 124 million. Roberts also pointed to China, where deaths now exceed births by roughly 3.3 million a year. A population can keep growing for a time while those changes are under way.

The total fertility rate, or TFR, is a period measure. It combines the birth rates of women at different ages and describes how many children a woman would have if those age-specific rates continued throughout her childbearing years. It is not a count of the children women currently in those years will ultimately have. A completed fertility rate counts how many children women have actually had by the end of their childbearing years. In 2026, for example, researchers can look at women who are 46 and count the children they have had. The measure comes with a lag: the completed fertility of women born around 2000 will not be known until they have passed their childbearing years. Fernández-Villaverde said that historically the two measures have not been very different, though the distinction matters when fertility is changing quickly.

The familiar replacement figure of 2.1 children per woman is a useful starting point, not a universal threshold. In a simplified example, 100 women need to have about 210 children to produce 100 daughters in the next generation. The calculation allows for a birth ratio with around 5 percent more boys than girls and for some girls not reaching childbearing age. Where more girls are born missing from the population because of selective abortion, or more die before reaching childbearing age, the fertility rate needed for replacement is higher. Fernández-Villaverde said it can be around 2.3 or 2.4 in some countries, and as high as 2.5 in parts of sub-Saharan Africa. On his account, the world replacement rate is about 2.2.

The world is already below the level Fernández-Villaverde considers necessary for long-run population stability, even though its population is still growing. He expects momentum to give way to a peak and subsequent decline. His estimate of 2055 is a projection; the population arithmetic he emphasized does not describe a brief dip followed by automatic stabilization.

The striking change is not only that rich countries have fewer children than they once did. Fertility has fallen even further in many middle-income countries, reversing the familiar relationship between national income and family size.

The Barro–Becker model, which Fernández-Villaverde said shaped his own training, helps explain why fertility fell as countries grew richer: parents may have fewer children while investing more in each child. That pattern still describes the decline in rich countries. What it does not explain, he argued, is that middle-income countries have often experienced steeper declines. If sub-Saharan Africa is set aside, he said, richer countries now tend to have higher fertility than poorer ones.

Mexico’s total fertility rate is about 1.51, compared with 1.57 in the United States, Fernández-Villaverde said. The rate among non-Hispanic white women in the United States is about 1.55—also higher than Mexico’s. Colombia, he added, has lower fertility than Japan, despite Japan’s reputation as a country struggling with population decline. Thailand, Sri Lanka and Chile are among the middle-income countries where fertility has fallen sharply.

Guatemala is a striking case. Its TFR was about 3.7 in 2007; last year it was probably around 1.8, Fernández-Villaverde said. If the current trajectory continues, Guatemala could fall below the United States before 2030. Looking back to 2005, he said, he would have been willing to bet heavily against that outcome—and would have lost.

The reversal also appears within countries. Fernández-Villaverde said three children are not unusual among people in high-income professions, while many people at lower income levels have one. He joked that three children can be the “ultimate status symbol” for a finance professional. He described the decline as concentrated among people with lower or middle incomes and education levels, both within countries and across them.

He also pointed to changing patterns among racial groups in the United States. From the country’s founding until around early 2024, he said, African-American fertility was higher than fertility among non-Hispanic white women. Since 2024, the latter has been higher, while African-American fertility has fallen rapidly. He connected the earlier pattern to the African-American share of the population remaining roughly constant despite successive waves of immigration from Europe, Latin America and Asia. These comparisons illustrate how changes in fertility can alter the composition of a population as well as its overall size.

Low fertility is not confined to middle-income countries. Fernández-Villaverde estimated rates of about 1.2 in Spain and Italy, 1.3 in Germany, 1.4 in the United Kingdom, 1.5 to 1.7 in most Scandinavian countries, and 1.6 to 1.7 in France. None, he said, is above 1.8.

The reversal matters because it makes a simple income-based explanation inadequate. Richer countries have fewer children than they once did, but the steepest declines are not confined to the rich. Nor does the pattern within countries match the old expectation that poorer families would have more children than richer ones. The shift raises a broader question: what is changing in how people form families, and why is it happening so quickly?

Fernández-Villaverde argued that population forecasts from the United Nations deserve closer scrutiny. He said past birth counts in UN data can differ substantially from figures published by national statistical agencies. He cited Colombia, where he said the UN overestimated births by about 60 percent, and Denmark, where children are registered at birth and official records are comprehensive.

He also questioned the scale and pace of forecast revisions. Fernández-Villaverde said the UN had reduced its population forecast by about 250 million every two years, removing roughly 1 billion people from its forecast over seven years. A billion, he emphasized, is a change in the projected total population, not merely a change in the number of births in one year.

His other concern is that the UN’s fertility projections build in a return toward the mean. In other words, the statistical method assumes that fertility will recover rather than continue falling at its recent rate. That may prove correct, he said, but it is an assumption embedded in the forecast.

For Fernández-Villaverde, the problem is not that forecasters make assumptions. It is that assumptions with major consequences should be visible. If a forecast expects fertility to recover, readers should be told why, rather than having to search through technical appendices to find the mechanism. He objected to an assumption with substantial effects being tucked away in a footnote.

Those assumptions matter beyond population totals. Fernández-Villaverde noted that climate models use UN population estimates. The population path used in those models therefore depends in part on the UN’s forecasts.

The explanation Fernández-Villaverde offered for falling fertility is explicitly speculative: modern life, he and his co-author propose, makes a third child expensive and childlessness easier. The point is not that parents formally calculate the costs and benefits of each child. It is that ordinary constraints can lead one family to stop at two rather than have three, or at one rather than have two.

In earlier generations, he said, a middle-class family in a small American town could expect children to finish high school, take local jobs and begin adult life relatively early. Today, parents who want their children to have a middle-class life may feel responsible for college or further professional education. Children require a longer period of financial and practical support, increasing the perceived cost of a larger family.

Longer education also delays careers and families. Fernández-Villaverde used the academic path as an example: college through age 22, further training and a PhD that may last until 30, and perhaps a first job around 30 or 31. By then, he argued, some of the childbearing years have passed. The same timing pressures apply, in his account, to people who pursue law, an MBA or other extended training.

The cost is not only the expense of education. Parents may have to remain closely involved for much longer, while adults spend more years establishing themselves before they feel settled enough to start a family. Housing adds another constraint. Fernández-Villaverde argued that zoning decisions in the United States have made affordable starter homes harder to find. A couple that delays marriage or children while saving for a down payment may lose time. He offered the everyday example of partners waiting six months to save another $25,000. A delay like that can affect when a family starts and how many children it has.

Career and housing costs do not, by themselves, explain why fertility has also fallen among people who do not follow a long professional training path. For that, Fernández-Villaverde pointed to changing gender norms. In many societies, he said, women once had fewer viable options outside marriage. A woman who was single in a small American town in the 1950s, or in Mexico in 1990, faced different social constraints from a woman today. Being single is now more acceptable and more practical in many places.

He treated that shift as a gain in freedom, not simply a demographic problem. A woman who no longer feels compelled to marry to have a tolerable life may choose not to marry or have children. Fernández-Villaverde used the example of a young woman walking her dog alone in Cambridge, Massachusetts, at 4 a.m., with a reasonable expectation that she would be safe. He said that was not the case in much of the world 25 or 30 years earlier. A society in which a woman has more choice about marriage, he argued, is better in that respect—even if fewer women then marry and have children.

Roberts raised declining religiosity, smartphones and social media as possible contributors to the change. Fernández-Villaverde’s “modernity” argument is a conjecture about how social institutions, career paths, housing and gender relations may combine, not a settled account of the decline.

The arithmetic of family size helps explain how relatively small changes in decisions can produce a large shift in the average. A person can have zero, one, two or three children; no one has 1.4. If two children becomes the default among people who have families, the overall average can still be far lower than two. Some women will have no children, some will have one, and relatively few will have three or more. Fernández-Villaverde argued that a society in which most parents stop at two can end up with a TFR around 1.4.

That makes the decision to have one more child especially consequential. Moving from two to three is not a small adjustment for a family. Fernández-Villaverde described the choice as a conversation in which one partner lists the practical difficulties of a third child—not a calculation written out as an equation. In his account, modern life need not push most people to choose childlessness. It can instead shift some from three to two, and others from two to one.

Social networks may amplify that change. If a person’s friends are childless, evenings and weekends may be organized around activities that do not involve children. If friends have children, social life may center on school or sports. Shifts in family choices can change what a peer group treats as normal, making it easier for more people to stop at one or two.

At the same time, Fernández-Villaverde cautioned that not every delay or decision not to marry reflects a preference. Some people may want to marry or have children but be held back by housing costs or other policy choices. Roberts noted that the age of first marriage in the United States has risen from the early twenties to close to 30. Fernández-Villaverde agreed that economists need to distinguish choices people should be free to make from choices constrained by economic policy.

Fernández-Villaverde said the broad decline also applies to the Muslim world, while emphasizing variation within countries and religious groups. In Israel, he said, fertility among Jewish mothers has remained roughly constant while fertility among Muslim mothers has fallen quickly. Non-religious Muslim women in Israel, he added, are already below replacement. He specified that he was referring to Israel within its 1949 frontiers.

For sub-Saharan Africa, he was less confident in the underlying data. He described UN fertility figures for the region as unreliable and said the evidence is better in some countries, including Kenya and Congo. In Congo, he said, fertility fell from 5.1 to 3.5 over about a decade. A simple extrapolation would put the country below replacement around 2040, but he stressed that this was a naive extrapolation. Nairobi is already below replacement, he said; Dakar and its metropolitan area are probably there too. He noted that fertility declines often appear first in big cities and then spread to the rest of a country.

His conjecture is that fertility is falling rapidly across sub-Saharan Africa, but that the decline is lagging behind the rest of the world because of structural features of the region’s growth model. He suggested that much of Africa could be below replacement around 2050, including countries at far lower income levels than those where the transition has already occurred.

A declining population could bring benefits as well as costs. Some Latin American cities grew rapidly during the 1960s, 1970s and 1980s, leaving crowded neighborhoods and shantytowns. Slower growth could give governments more room to improve those neighborhoods and invest in roads and other infrastructure.

The harder problem, in Fernández-Villaverde’s account, is that governments have made long-term commitments to older people that will become more difficult to sustain as the population ages. He used Spain as an example. From 2007 to the time of the discussion, spending on retirees—principally pensions and the Spanish equivalent of Medicare—had continued to grow. Other government spending had not grown and had declined somewhat, even as the country’s population increased by eight million, in part because of immigration.

Fernández-Villaverde pointed to poor highway conditions, problems with high-speed rail and limited capacity for other public services as consequences of that squeeze. He expects the pressure to sharpen as retirees and people nearing retirement become a larger voting bloc. In his view, voters will tend to protect pensions and health care over education, infrastructure maintenance or defense.

He argued that countries such as France, Germany and Spain will not be fiscally sustainable by 2040 under current arrangements. He was less certain about how the pressure would resolve, but suggested that public frustration and radical political proposals may partly reflect the condition of public finances.

Roberts complicated the political argument. People may care about their children and resist making them bear the cost of supporting the older generation, he said. The political system, though, is not necessarily nimble enough to adjust gradually. Fernández-Villaverde maintained that fiscal choices will become harder while acknowledging that he did not know precisely how the strain would play out.

Public debt adds to the pressure. Fernández-Villaverde noted that debt is measured relative to GDP; if a shrinking population contributes to weak or negative GDP growth, countries with high debt-to-GDP ratios may find that debt harder to service. He also raised national defense as a concern. A wealthy country can still have limited capacity to fund military forces, he said, pointing to Iceland as a country unable to finance an aircraft-carrier task force.

Environmental effects are less straightforward than a smaller population might suggest. Fernández-Villaverde said that global CO₂ emissions per person had risen only about 9 percent since 1990 and had fallen since 2011. In his account, population growth explains most of the increase in emissions since 1990. A lower population trajectory would therefore change emissions scenarios, and he argued that accurate population estimates matter because climate models use UN projections.

But environmental protection also depends on politics and income. When governments face pressure to pay pensions and other obligations, voters may be less willing to fund conservation or restrict the sale of natural resources. Fernández-Villaverde described the environment as a “luxury good”: people tend to care more about it when their immediate material needs are met. A government needing revenue might find it harder to persuade voters to protect a forest rather than sell its timber to help pay public obligations. A smaller population could reduce emissions, but fiscal pressure could weaken the political will to protect land and resources. Fernández-Villaverde said he has mixed feelings about the environmental consequences and argued that economic growth is an important support for environmental protection.

Fewer births mean fewer people entering the pool from which scientists, artists, athletes and inventors emerge. Roberts invoked Julian Simon’s argument that human creativity is a central resource. Fernández-Villaverde accepted the basic logic: with fewer people, there is a lower chance of producing an exceptional person.

He used sports to make the point. A country selecting its best 22 soccer players will, other things equal, have a smaller talent pool if far fewer children are born each year. The same reasoning, he said, applies to top scientists and artists. Small populations can still produce exceptional performers—he cited the number of classical musicians from Iceland—but developing talent from a smaller group may require more deliberate effort.

The scale of a possible decline is also easy to underestimate. A fertility rate below replacement does not mean that a population falls by a little and then levels off. If behavior stays unchanged, Fernández-Villaverde argued, the decline continues. He illustrated the point with a scenario for Thailand: from a population of about 62 million, it could fall to 600,000 in 200 years at the current fertility rate. In 400 years, he said, it would fall below the minimum population needed to maintain enough genetic variation. He presented this as a scenario, not a forecast.

Societies could respond through political and cultural change rather than a smooth economic adjustment. Roberts argued that the consequences would provoke a reaction. Fernández-Villaverde agreed that they might, and offered Pennsylvania’s Amish population as an example: its growth has made Amish voters increasingly important in statewide elections. He joked that, if Pennsylvania remains a swing state and the trend continues, Amish voters could become pivotal in a presidential election in 2060.

He also said fertility among religious people has not fallen in the same way as among the broader population. If religious identity persists across generations, societies could become more religious over time, potentially raising fertility. But he emphasized that a religious parent’s children need not remain religious. The possibility of a more religious United States or world pushing fertility back toward replacement is one possible response, not a prediction.

For Fernández-Villaverde, the immediate mechanism is less dramatic than a sudden decision to reject family life. If modern conditions make three children difficult, some families stop at two; if the same pressures make two difficult, some stop at one. Because childbearing choices come in whole numbers, relatively small changes in how many people make each choice can pull the average down quickly.