Married Before Twenty: The Fertility Transition That Stopped at the First Birth

Executive Summary. Bangladesh finished the half of its fertility transition that shows up in the headline number and left the other half untouched. Total fertility was 2.138 births per woman in 2024, down from 6.831 in 1971, a fall of 68.7%, and only 1.09 times the peer median of 1.962. On family size, Bangladesh has converged with the region. The adolescent birth rate has not: 70.8 births per 1,000 teenage women in 2024, the highest of the six countries, 2.8 times the peer median of 25.6 and 45.2 births above it. India records 13.0, Thailand 25.0, Indonesia 25.6, Vietnam 33.9, Pakistan 39.9. The two explanations usually offered do not survive the comparison. Girls are enrolled in secondary school at 1.143 times the male rate, the highest of the four countries reporting. Contraceptive prevalence is 62.7%, second of the four reporting and above the median of 52.5%. What is left is the timing of the first birth, which the cross-country data used here does not measure.

The number that did not converge

In 2024 Bangladesh recorded 70.8 births per 1,000 teenage women, the highest of the six countries compared here. The rest of the group: Pakistan 39.9, Vietnam 33.9, Indonesia 25.6, Thailand 25.0, India 13.0. The peer median is 25.6, so the Bangladeshi rate sits 2.8 times above it, a gap of 45.2 births per thousand.

Set that beside the aggregate and the oddity sharpens. Total fertility in Bangladesh was 2.138 births per woman, second-highest of the six behind Pakistan's 3.547, and 1.09 times the peer median of 1.962. Indonesia is at 2.115, India 1.962, Vietnam 1.895, Thailand 1.204. On the number demographers use to describe a country's family size, Bangladesh is unremarkable, a rounding error above the group median. The same population, measured one age band lower, is a clear outlier. A country of 173.6 million people is running two fertility regimes at once.

Source: World Bank World Development Indicators, adolescent fertility rate (SP.ADO.TFRT), 2024.

Two curves, one country

The teenage rate has fallen a long way, and any honest account starts there. In 1990 it stood at 178.3 per 1,000. By 2024 it was 70.8, a decline of 107.5 births per 1,000, or 60.3%, to about four tenths of its 1990 level. That is not a country standing still, and nothing in this essay should be read as saying otherwise.

The comparators were falling over the same period, and the series compared here does not carry their 1990 levels as stated figures, so how much of today's gap is a difference in starting point rather than a difference in speed cannot be quantified. What the data does settle is the endpoint. Thirty-four years on, Bangladesh is still 2.8 times the peer median and still first of six.

Source: World Bank WDI, adolescent fertility rate (SP.ADO.TFRT), Bangladesh against comparators, 1990-2024.

Now the other curve. Total fertility fell from 6.831 births per woman in 1971 to 2.138 in 2024, a drop of 4.693 births, or 68.7%, leaving the rate at 0.313 times its starting level. Nearly seven children per woman became barely more than two inside two generations. Whatever machinery a state and a society need to move a demographic aggregate at that speed, Bangladesh evidently had it.

Source: World Bank WDI, fertility rate, total (SP.DYN.TFRT.IN), Bangladesh against comparators, 1971-2024.

A total fertility rate can fall two ways. Women can stop earlier, cutting off the third, fourth and fifth births. Or they can start later, pushing the first birth out of the teenage years and into the twenties. The two Bangladeshi series are consistent with the first and not the second: near-peer family size, far-above-peer teenage childbearing. The transition worked on the last birth and left the first one where it was.

That inference deserves a flag. These indicators are rates, not a decomposition: the age-specific rate for teenagers and the aggregate for all women, not the birth-order composition that would prove the point directly. A table of age at first birth by cohort would settle it in one page. The data used here cannot, and reading two aggregates as if they were a decomposition is exactly the move a critic should press on.

The explanations that do not fit

Two answers usually arrive before the question is finished. Neither survives contact with this group of countries.

The first is schooling. Girls who leave school early marry early and give birth early, so the teenage rate is a symptom of girls being out of classrooms. Bangladesh is the wrong country for that argument. The secondary enrolment gender parity index was 1.143 in 2019, meaning the female gross enrolment rate ran 1.143 times the male rate, the highest of the four countries reporting: India 1.019, Thailand 0.997, Pakistan 0.868. Indonesia and Vietnam do not report in that year, so this is a four-country statement, not a regional one. Bangladesh is the country among those four where secondary enrolment tilts furthest toward girls, and it has the highest teenage birth rate of the six.

The obvious rejoinder is that gross enrolment counts registration, not attendance, includes over-age students, and says nothing about whether a girl enrolled in one grade is still there the next year. That is fair, and it is why the enrolment finding is used here narrowly: it does not prove girls are in school, but it rules out the version of the story in which they are absent from the rolls in the first place.

The second answer is contraception. Prevalence of any method among women of reproductive age was 62.7% in 2019, second of the four countries reporting, behind Thailand at 72.97% and ahead of Indonesia at 52.5%, which is also the peer median, and Pakistan at 34%. This is not a country where family planning failed to arrive.

The caveat here is larger than the finding. The indicator pools all women of reproductive age, a population dominated by married women in their twenties and thirties. A national figure of 62.7% is fully compatible with near-universal method use among older married women and very little among married adolescents. The comparable series carries no age-disaggregated prevalence, so the aggregate cannot be read as evidence about what a seventeen-year-old can obtain, or decide. It rules out the crudest supply story and nothing beyond that.

Strike both explanations and what remains is timing: when unions form, and how quickly a first birth follows. No measure of age at marriage or age at first birth exists in this cross-country set, so the honest form of the finding is negative rather than positive. The two mechanisms with comparable international data do not track the gap, and the mechanism that would explain it is not in the comparison at all.

Where those births land

Whatever produces them, teenage births in Bangladesh meet a delivery system that reaches fewer births than any comparator that reports. Skilled health staff attended 59% of births in 2019, the lowest of the four countries reporting, against a median of 94.7%: Thailand 99.1%, Indonesia 94.7%, Pakistan 73.7%. Bangladesh sits at 0.623 of the peer median, 35.7 points below it. India and Vietnam do not report in that year.

That pairing is the operational finding: the highest teenage birth rate of the six meets the least attended delivery of the four countries reporting. If teenage deliveries carry above-average obstetric risk, as obstetric practice assumes and as none of these indicators can confirm, then the riskiest births in Bangladesh are also the ones least likely to happen in front of someone trained to manage them.

The outcome indicators are messier, and the mess belongs in the essay rather than in a critic's reply. Low birthweight ran at 23.05% of births in 2020, second-highest of the five countries reporting and 2.278 times the median of 10.12%: India 27.44%, Thailand 10.3%, Indonesia 9.9%, Vietnam 6.3%. Maternal mortality was 115 per 100,000 live births in 2023, third-highest of the six and 1.438 times the median of 80: Pakistan 155, Indonesia 140, India 80, Vietnam 48, Thailand 34.

Neither pattern tracks teenage childbearing cleanly. India has the lowest teenage birth rate of the six, 13.0, and the highest low-birthweight share of the five reporting, 27.44%. Indonesia sits at the peer median for teenage births, 25.6, and records maternal mortality of 140, above Bangladesh's 115. Maternal undernutrition, anaemia, birth spacing, how many births are weighed at all, and how each country's civil registration handles a stillbirth all sit between adolescent childbearing and these outcomes. Across six countries these are associations, and loose ones. They do not identify a cause, and no causal claim is made here.

One thing income does not explain is the attendance gap. Pakistan runs the group's highest total fertility at 3.547 and the group's highest maternal mortality at 155, and still recorded skilled attendance at 73.7% against Bangladesh's 59%. Whatever is keeping Bangladeshi deliveries away from trained staff, a comparator with worse fertility and worse maternal outcomes has already cleared it.

A note on scale, because the temptation is to turn the rate into a headline count. The size of the teenage female cohort is not in this data set, so that multiplication is not performed here. The rate is the claim.

What would change this conclusion

The essay reads the gap as a timing problem: births arriving early because unions form early, rather than because girls are out of school or methods are unavailable. That is falsifiable. If a nationally representative survey shows the median age at first marriage for Bangladeshi women is already above twenty while the teenage birth rate stays near 70.8, the births are not concentrated where this argument places them and the argument fails. Equally, if coming World Bank releases show the ratio to the peer median falling well below 2.761 with no change in age at first union, something else is moving the number and the reading here is wrong.

Three moves.

1. Publish age at first birth and birth order as standing annual statistics. Internationally comparable data describes the shape of Bangladeshi fertility only through two aggregate rates, which is what this essay was left to work with. Owner: Bangladesh Bureau of Statistics, through the SVRS. Success signal: an annual published series for median age at first birth and for the share of births that are first births to women under twenty.
2. Attach a skilled attendant to the adolescent delivery specifically. Attendance at 59% is the lowest of the four countries reporting, and it is the indicator in this set that a health system most directly controls. Owner: Health Services Division, through upazila health complexes and the community clinic network. Success signal: skilled attendance in the World Bank series above Pakistan's 73.7%, which would move Bangladesh out of last place among the countries reporting.
3. Disaggregate contraceptive prevalence by age and marital status. The national 62.7% is compatible with almost any pattern among teenagers, so policy aimed at that group cannot be evaluated against it. Owner: Bangladesh Bureau of Statistics with the national demographic and health survey programme. Success signal: a published prevalence figure for married women under twenty, which the aggregate currently hides.

The counterargument

The strongest objection is that the comparison is unfair to a country that started far behind. Bangladesh was at 178.3 per 1,000 in 1990, and a rate that has since fallen 60.3% is a rate in motion. Judging it on today's level penalises history rather than policy. A related point follows: adolescent fertility reflects marriage norms with a generational lag, so the 2024 figure reports on decisions taken when today's mothers were themselves children.

Both points have force, and because comparator levels for 1990 are not among the stated figures here, the starting-point argument cannot be dismissed with numbers. Two things narrow it anyway.

First, the same society moved the aggregate fast. Total fertility fell 68.7% in the period covered, from 6.831 to 2.138. Whatever combination of family planning, female education, rural service delivery and changing expectations produced that, it worked, and it did not reach the first birth. A country that can cut completed family size by two thirds and leave teenage childbearing at 2.8 times the peer median has not run out of capacity. It has aimed elsewhere.

Second, the lag argument is a prediction, not an explanation, and it is checkable in the next few data releases. It says the ratio to the peer median should be visibly closing. It stands at 2.761. If it is still near that level when this decade ends, the lag will have been invoked for another ten years without ever being tested.

---

Data sources: World Bank World Development Indicators, retrieved from the BDPolicyLab data lake, 2026-08-10. Comparators are India, Pakistan, Indonesia, Thailand and Vietnam; "peer median" is the median of the comparators reporting in the stated year, excluding Bangladesh. Fertility figures are 2024, maternal mortality 2023, low birthweight 2020, and contraceptive prevalence, skilled birth attendance and secondary enrolment parity 2019, being the latest years with comparable coverage. Contraceptive prevalence, skilled attendance and enrolment parity have four reporting countries in those years and low birthweight has five, so comparisons on those indicators are stated as such rather than as regional claims.

Sources

1. World Bank WDI, Adolescent fertility rate (births per 1,000 women ages 15-19), SP.ADO.TFRT: https://data.worldbank.org/indicator/SP.ADO.TFRT
2. World Bank WDI, Fertility rate, total (births per woman), SP.DYN.TFRT.IN: https://data.worldbank.org/indicator/SP.DYN.TFRT.IN
3. World Bank WDI, Contraceptive prevalence, any method (% of women ages 15-49), SP.DYN.CONU.ZS: https://data.worldbank.org/indicator/SP.DYN.CONU.ZS
4. World Bank WDI, Maternal mortality ratio (modelled estimate, per 100,000 live births), SH.STA.MMRT: https://data.worldbank.org/indicator/SH.STA.MMRT
5. World Bank WDI, Births attended by skilled health staff (% of total), SH.STA.BRTC.ZS: https://data.worldbank.org/indicator/SH.STA.BRTC.ZS
6. World Bank WDI, Low-birthweight babies (% of births), SH.STA.BRTW.ZS: https://data.worldbank.org/indicator/SH.STA.BRTW.ZS
7. World Bank WDI, School enrollment, secondary (gross), gender parity index (GPI), SE.ENR.SECO.FM.ZS: https://data.worldbank.org/indicator/SE.ENR.SECO.FM.ZS