The Global Birth Crash Is Happening Faster Than Demographers Expected

Globally, humans are having a lot fewer babies. Take for example, in 1950 the lowest fertility level anywhere on Earth belonged to Latvia, at a shade under 2.1 births per woman. Today the record sits with Macau, which logged a near-zero rate of 0.47 in 2025. A report published this month by the American Enterprise Institute argues that no working theory of childbearing can account for how far the floor on human fertility has dropped, or for why it dropped in Iran and Nepal and Bogotá at more or less the same time. Demographers spent decades expecting birth rates to level off at around two per woman, the point where a population holds steady; instead, they have sailed straight below that number.

By the report’s reckoning, something like 71 percent of humanity was already living in sub-replacement societies – ones where birthrates drop below the level required to maintain current population levels – in 2023, spread across 130 of the 237 countries and territories in the UN Population Division dataset the analysis rests on.

Replacement level is usually quoted as 2.1 births per woman. That figure is a convenience rather than a constant. It has to clear 2.0 because slightly more boys are born than girls – roughly 105 to 100 without intervention – and it has to clear a little more because not every girl survives to childbearing age. Where infant mortality runs high, or where sex selection skews the ratio, the real bar sits higher: Nigeria’s works out at about 2.6, for example. Run that arithmetic across the whole dataset, the authors say, and the true global replacement level for 2023 comes to roughly 2.18, against a measured world rate of 2.25. A margin of 3 percent, and Eberstadt and Norrick reckon the 2.25 is probably too generous.

That said, it’s important to note, this is think-tank report, not peer reviewed, and several of the numbers are the authors’ own derivations rather than published statistics: China’s 0.9 births per woman, for example, is reverse-engineered from a crude birth rate based on stated assumptions about how many women are of childbearing age. Further, the replacement-level calculations behind that 2.18, by the authors’ own footnote, were generated with the help of an AI assistant. Neither of which makes the figures wrong. It does mean some of them are estimates, not as actual measurements, and the body text does not always say which is which.

Nepal and Iran Were Not Supposed to Do This

For most of the postwar era, standard explanations tied falling birth rates to development: income, literacy, urbanization, national family planning programs. Yet Iran, an Islamic theocracy, is sub-replacement. Nepal and Myanmar sit among the poorest countries the UN classifies, and they are sub-replacement. Brazil never ran a national family planning program and got there as well.

Spin the globe and the pattern holds regardless of what else is going on; Seoul reported 0.63 in 2025, Taiwan 0.695, Singapore 0.87. Bogotá came in at 0.8, below Hong Kong. Puerto Rico slipped to 0.89 in 2023, making it the first US territory under one birth per woman, and Manila, capital of Asia’s largest Catholic country, sat at 1.4. The United States, which people tend to think of as the fertile outlier among rich nations, has just posted 1.57.

What People Say They Want

So what’s left standing, once the material explanations have been cleared away? The report’s answer is that people are simply choosing this. It leans on a finding that has held up better than anything else on offer: the single best statistical predictor of how many children a society ends up having is how many children women in it say they want. That gets you about 90 percent of the way, which may not be everything, but is darn close. Eberstadt and Norrick take some comfort in this fact, noting that the numbers affirm that “we humans are neither rabbits nor robots”. Then they hand the mystery back, because a good predictor is not a mechanism and nothing in the data explains why stated preferences should have drifted so far downward at once across cultures that share almost nothing else. “The rules, such as they are, are being rewritten in front of our eyes.”

One statistical caveat sits in the report’s endnotes deserves more attention than it gets. Nearly every figure quoted here is a period rate: a snapshot of the family a woman would end up with if she spent her entire reproductive life under a single year’s conditions. Push childbearing later and that snapshot dips even where completed family size holds up, a distortion demographers have modelled in some detail. The authors define the distinction carefully, concede that timing shifts can distort the measure but argue that the distortion does not much matter.

Israel is offered as the lone affluent democracy where fertility rose over a generation, from 2.9 to 3.0, powered by rising Jewish fertility. The report’s own chart then puts the 2025 national figure at 2.8, with secular Jewish Israelis at 1.9 and Muslim fertility falling faster than Jewish fertility climbed.

And not everyone reads these numbers the same way. The UN Population Fund argued last year that the real crisis is not too few births but too little reproductive agency, with people worldwide unable to have the number of children they say they want, which is the same statistic pointed in the opposite direction. South Korea, meanwhile, the country this report uses as its floor, has reported a rise: births in the opening quarter of 2026 were up 14.8 percent on the year, the largest quarterly jump since 1981.

Back Down to Pre-Industrial Birth Totals

Prolonged sub-replacement has an arithmetic consequence that arrives decades late. Over 50 countries and places are now net-mortality societies, where deaths outnumber births year after year: most of Europe, effectively all of East Asia, plus Thailand, Cuba and Uruguay. Italy registered 355,000 births in 2025 against 833,000 in 1862, when the country had barely finished unifying. Japan’s 2025 total of 671,236 came in below its 1873 count, the first year anyone was counting.

Whether humanity has already crossed the line is, as the report concedes, something we will only know looking backward. The figures from sub-Saharan Africa (the last high-fertility region, and also the least well counted) will take years to firm up. The interesting question is no longer where the floor sits. It is what a species does once it works out there may not be one.

FAQ