Teenage pregnancy rates remain high across many parts of the developing world: In Africa, on average, about one in ten girls between the ages of 15 and 19 has already given birth. These early pregnancies often come with serious consequences for young mothers and their children. They are linked to lower education levels, poorer health outcomes, and reduced economic opportunities.
Scientists, development agencies and NGOs have long heralded education as a powerful tool to reduce early childbearing. Education may directly influence women’s reproductive behaviour, but it can also improve their employment and income-generating opportunities, leading them to postpone pregnancy.
But does access to basic education for young girls result in such successes uniformly across population groups?
We are economists who conducted a study to explore the effect of primary school education on fertility and its related outcomes in Burundi. A bold education reform took place in that country in 2005: the government abolished formal school fees for primary education. As a result, many children who had been excluded from school by cost were able to get a basic education.
The free primary education policy displays a natural experiment for researchers interested in the effects of education. Because the reform applied only to children young enough to be in school, we could compare girls who were eligible for free schooling with those who were just too old to be eligible (but similar in other ways). This allowed us to track the policy’s direct and causal effects.
Indeed, we see that Burundi’s free primary education policy increased educational attainment of women by 1.22 years on average. Our findings also provide new, robust evidence that education can reduce downstream effects, as we see teenage childbearing reducing by as much as 6.9 percentage points. In other words, while about 37% of teenage women who did not benefit from free primary education had given birth before the age of 20, only 30% of those eligible for free primary education had done so.
Importantly, and new in our findings, education conferred the greatest benefit to girls from the poorest segment of society. Our study thereby underscores an important lesson for policymakers: education policies can be highly effective, but not necessarily for everyone in the same way.
A natural experiment in Burundi
We used nationally representative data from Burundi’s Demographic and Health Surveys to establish the effects of education. We compared women born between 1987 and 1991 to those born between 1992 and 1996 – aged 14-18 and 9-13 respectively when the free school policy took effect. We applied modern econometric techniques to identify the increase in years of schooling induced by the policy. We then examined the effect of this increase in schooling on girls’ outcomes, including teenage pregnancy, literacy, and the likelihood of working for cash income, among other outcomes.
The results were striking. Girls who had been young enough to benefit from free schooling gained, on average, 1.22 more years of education thanks to the programme. That corresponds to a 34% increase in the years of education compared to similar women who missed out on the policy. Crucially, this increase occurred across the board – both poor and wealthier women gained more education.
But there was a twist: only young women from poor backgrounds seemed to reap broader benefits from that extra schooling.
For girls from very low-income households, one additional year of schooling reduced the likelihood of becoming a teenage mother by nearly 7 percentage points.
It also decreased their desired number of children and boosted their literacy and chances of working for a cash income outside their own home. These are all powerful indicators of women gaining autonomy and making more informed reproductive choices.
While girls from wealthier households experienced an increase in education too, this additional schooling showed no measurable effect on fertility, literacy, or employment outcomes for them. Thus, we did not find any statistically significant impact of increased schooling for these girls.
In other words, the free primary education programme in Burundi increased the number of years of education for girls in general but the downstream effects of that education appear to have materialised only for the very poor.
Why does household wealth matter?
Why would women from the relatively wealthier families not benefit equally from more education?
One reason could be that somewhat wealthier households had already ensured higher levels of education for their daughters, even before school fees were abolished in Burundi. The education reform thus made less of a difference in their lives. Very poor families, on the other hand, were far more likely to be constrained by the costs of primary education. When that barrier was removed, their daughters could finally access schooling, and this had transformative effects also for sexual and reproductive health.
For the most disadvantaged, education is more likely to open up new economic opportunities. We found that policy-induced education increased their likelihood of working outside of their own household for a cash income, which raises the opportunity cost of early childbearing. The classic economic theory by Nobel prize laureate Gary Becker and Jacob Mincer suggests that when women have better employment prospects, they are more likely to postpone childbirth. And they invest more in their children but tend to have fewer of them. This is precisely what we observed in our data.
Education also seems to empower women by increasing their knowledge and capacity to access information. We found that literacy rates among poor women rose significantly with each added year of schooling. Another prominent theory in the literature on education is that educated women are more likely to understand and use contraception, make informed reproductive decisions, and challenge traditional gender norms.
Rethinking one-size-fits-all policies
Our study underscores an important lesson for policymakers: education policies can be highly effective, but not necessarily for everyone in the same way.
When evaluating the success of reforms like free primary education, we must go beyond average effects. Aggregated data can mask substantial differences between population groups. If we had only looked at average outcomes, we might have concluded that free schooling had little effect on teenage childbearing. But by disaggregating our data by household wealth, we see a different and far more hopeful picture. Free schooling has powerful effects – if we know where to look.
Frederik Wild received funding from the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), EXC 2052/1 – 390713894 for the research paper referenced in the article. The views expressed in this article are solely my own and do not represent those of my employer or affiliated organizations.
David Stadelmann received funding from Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), EXC 2052/1 – 390713894.
By Frederik Wild, Postdoctoral Researcher, University of Heidelberg And
David Stadelmann, Chair of Economic Policy and Economic Development, Bayreuth University
