Fertility Rate Explorer

Global fertility has collapsed. In the 1960s, the world average was 5 children per woman. By 2023, it had fallen to 2.3 — below the long-term replacement rate of 2.1. More than half of all countries now sit below replacement. This is unprecedented in human history.

This is not a developed-world story. Bangladesh, Vietnam, Brazil, and Iran have all undergone demographic transitions in a single generation. What took Sweden 150 years has happened in parts of Asia and Latin America in 30.

If people don't want kids, falling fertility wouldn't be a puzzle. But Americans consistently say they want about 2.5 children — and consistently have about 1.7. The gap between desired and actual fertility has widened since the 1990s.

Never-married status is the single strongest predictor of lower fertility in the US (standardized coefficient: −0.293). Married women average about 2.2 children; never-married women average 0.7. And the share of never-married adults has tripled since 1970.

On the left: how the composition of marital status has shifted over time. On the right: how many children each group averages.

Secularization removes pro-natalist norms. The share of Americans identifying as "None" (no religion) has risen from 5% in 1972 to over 25% today. Protestants — once the majority — have fallen below 40%. And the non-religious have fewer children: 1.5 vs 2.1 for Protestants.

As women gain education and career opportunities, the opportunity cost of having children rises. Female education is the strongest international predictor of lower fertility — explaining 80% of the variance in a random forest model. Within the US, College+ women average 1.6 children vs 2.3 for those without a high school diploma.

Some theorists predicted a "J-curve": fertility falls with early development, then recovers at very high income levels. The data shows no such recovery. The relationship between GDP and fertility is monotonically negative — richer countries have fewer children, period.

Once a country drops below 1.5 TFR, approximately 95% never recover. The fertility trap is self-reinforcing: smaller cohorts produce fewer absolute births, which shifts cultural norms toward smaller families, which keeps fertility low. Early intervention (before crossing 1.5) is far more cost-effective than attempting recovery.

Is the fertility decline just women having children later — or actually having fewer? Cohort analysis settles it: this is primarily a reduction in completed fertility, not merely a delay. Women born in the 1960s had ~2.0 children by age 45; women born in the 1950s had ~1.9. And childlessness rates have risen from 7% (1930s cohort) to 18% (1960s cohort).

Two complementary models: an OLS regression on 74,045 US GSS respondents identifies individual-level predictors, while a random forest on 265 countries identifies international-level drivers.

What happens to a country's fertility if its drivers change? Select a country and adjust the sliders to see how changes in education, urbanization, female labor participation, and income would shift projected TFR. The shaded bands show uncertainty ranges (68% and 95%) based on the model's residual error, widening over time. This is a sensitivity analysis — "if X changed by Y, TFR would shift by Z" — not a time-series forecast.