Anyone who works on development knows this stylized fact: as income rises, fertility falls. In the latest available data, each doubling of GDP per capita is associated with about 0.6 fewer children per woman (Figure 1, on the right). The link was steeper in 1990 (on the left), around 0.8 fewer children per doubling, but the basic pattern has stayed the same.
Figure 1: Children per woman versus income across countries, 1990 and 2023.
And it’s not just an artefact of looking across countries: the latest published DHS data for Kenya shows that the total fertility rate falls from 5.3 children in the poorest wealth quintile to 2.7 in the richest. This, of course, (mostly) generalizes. Tom Vogl’s 2016 paper uses DHS data from 48 developing countries to show that higher household wealth is associated with fewer children within countries—though one exception is West-African households pre-1995, where the relationship was positive. All this data (and associated notion of the opportunity cost of women’s time) led to a policy mantra that strengthening women’s economic status would reduce fertility in regions where fertility was high, producing a ‘demographic dividend’—the growth boost a country gets when the working age share of its population increases relative to its share of children and the elderly.
A new crop of studies is complicating this consensus (and it may have taken this long because RCTs of programs that raised income or wealth rarely looked at fertility—so the question fell between silos). Lars Berge et al.’s new paper from Tanzania shows that a successful entrepreneurship training raised the income of young women “on the doorstep of adulthood” by about 27% and cut the share of women below the poverty line by 4 percentage points. These women also have more children: overall fertility increased by about 20%.
Figure 2: Impact on the number of children at first follow-up, with 95% confidence intervals (Donald et al. 2024)
And it’s not just Tanzania. Donald et al. (2024) pool six experiments of programs that raised women's income or household wealth across five other African countries—through business trainings or land titling. These were programs that did not have any specific empowerment or reproductive health messaging that might affect fertility (whereas Bandiera et al. 2020, where marriage and childbearing fell among adolescent girls, does have such messaging). The business trainings raised profits of female business owners by about 30%—and made women 22% more likely to have a child in Ethiopia and roughly twice as likely in Togo. Land titling increased the number of children about 5% in Benin and 7.5% in Ghana. And the results keep coming. A recent working paper by Kotsadam et al. finds that women offered a formal wage job in garment and shoe factories in Ethiopia earned about 10% more over the study period and had 5% more children in the long term, with childlessness falling by almost half.
So how to reconcile these findings with the stylized fact from cross-sectional data that women in higher income households or countries have fewer children?
First, consider one potential factor for why more income results in more births in the above studies: job flexibility. In Donald et al., the women were entrepreneurs or farmers, not formal wage workers—so they had more scope to combine work with more children. In Berge et al., the fertility effects of entrepreneurship training were entirely driven by self-employed women (vs. those who went into salaried work). And in Kotsadam et al., fertility rose after women had accumulated resources and left the formal factory jobs, which were less compatible with childbearing. In other settings, this incompatibility shows up as young women delaying marriage and having fewer children after being offered jobs at factories or call centers (Jensen 2012; Heath & Mobarak 2015).
These results are consistent with children being normal goods in contexts where new income opportunities for women accommodate child-rearing. But then over the longer run, under structural transformation, the income-fertility relationship may turn negative as jobs for women are accompanied by labor market shifts, such as a pivot from informal, home-based work to formal firm work, among other changes.
Another important feature is missing safety nets. Donald et al.’s results come from women without a son who may be at risk of dispossession if their husband dies. Evidence points to women investing in old-age security once they can afford to do so as a key driver of the fertility response. The null finding from Rwanda (Figure 2) supports this: fertility didn't rise there, the one program that directly strengthened women's own long-term property rights. Likewise, it did not rise among already-wealthy women in a second Ethiopia training. Consistent with the safety nets channel, Rossi & Godard (2022) find that extending social pensions in Namibia reduced fertility.
Third, consider that income is highly correlated with education. The income–fertility relationship can thus reflect education effects rather than income (Murtin 2013, Vogl 2016 from earlier). One well-documented pathway through which education decreases fertility is delayed marriage (Breierova & Duflo 2004; Duflo, Dupas & Kremer 2015)—and delaying marriage is a powerful channel: many of the job programs or opportunities that reduced fertility, like those in Jensen (2012) and Heath & Mobarak (2015), work through it. Education can also shorten the reproductive window (not just through the marriage effect), change who women marry, and influence the number of children women want (the classic quantity-quality tradeoff).
Coming back to our earlier results: Donald et al. and Kotsadam et al. study already-married women, so the marriage-delay window there is closed. So what seems to matter is whether family formation can still be delayed—which connects to debates about falling fertility in the rich(er) world, and how much the drop in couple formation (“the relationship recession”) is to blame.
Summing up what to take from this discussion leaves us with some conclusions but also many questions. Interventions that delay marriage (through keeping girls in school or girls’ livelihoods clubs) are still a good bet for policymakers looking to reduce teen pregnancy and raise future earnings. But strengthening women's economic status, particularly among women who want children (whether for long-term security or other reasons), may raise fertility, contrary to received wisdom. This new group of studies also points to several open questions. Does increasing women’s income actually work as a solution to boost fertility in settings where low fertility is a concern? Does it matter whether the income gain is woman-specific (as in these RCTs) or part of a broader, economy-wide trend, which may have a tighter feedback loop with social norms? And what does the relationship between self-employed work and fertility imply for how these patterns might change as jobs formalize and shift towards wage employment in emerging economies? Causation may even run the other way: where desired fertility stays high, that itself may slow the shift out of self-employment, with implications for structural transformation. And last but not least, these results (again) demonstrate the value of careful and causal research to enrich and challenge what descriptive data say.
Join the Conversation