So I started this blog post as a comparison of two of my favorite Pink Floyd songs. But then I had the good fortune to pick up a really interesting new paper by Kjetil Bjorvatn, Denise Ferris, Selim Gulesci, Arne Nasgowitz, Vincent Somville, and Lore Vandewalle. Bjorvatn and co. are comparing the impact of free childcare versus an cash transfer in Uganda. Time v money where it matters.
The setup is neat. Bjorvatn and co. take a bunch of households in communities that have at least one childcare center across Uganda and winnow it down to households that have one kid from 3 to 5 years old. They then randomize the households into a cash transfer (for the woman) equal to the cost of a year of childcare, a year of full-time free daycare or both.
Let’s start with the use of childcare. It’s important to note that 82 percent of the control group has kids in any childcare. But far fewer (34 percent) have their kid in full-day childcare. The interventions boost this. The childcare subsidy bumps the full-time care by 48 percentage points (i.e. more than double the control group). And, interestingly, the cash transfer bumps it by 7 percentage points – so there is some unfunded demand out there. Combining the two gets you no more than the childcare subsidy.
So, what happened to moms? The childcare intervention leads to a significant increase in her self-employment revenue -- these go up by 46 percent relative to the control mean. But this happens without an increase in the probability that she works or in the hours that she puts in. This is reminiscent of the neat paper by Delecourt and Fitzpatrick which I blogged about last year. They show that kids in the workplace have a big, negative impact on female entrepreneurs’ profits (through a mechanism any parent can relate to).
On the cash transfer side of things we do see a significant increase in the probability that she engages in self employment (by 19 percentage points), and also a significant increase in hours. Earnings follow, with revenues from self-employment up by a similar amount to the childcare intervention. There is a small decline in wage income, but looking at total revenues, the childcare and cash interventions yield roughly the same increase. Now, the childcare plus cash intervention yields effects that are equivalent to either alone (although the point estimate is bigger) and this pattern seems to run through almost all of the results – so let’s just say there are no meaningful complementarities and leave this arm out of the discussion that follows.
What about dad? The cash transfer has no significant impacts. Childcare, on the other hand leads to a 9 percentage point increase (on a control mean of 27 percent) in his engagement in wage labor as well as an increase in his wage-labor hours. Wage income goes up by about a third. There are no significant increases in his enterprise income. Part of the mechanism for the boost in wage labor might be his time. As Bjorvatn and co. point out, men in Uganda spend a significant amount of time on care (and yes, it’s significantly less than women, like in every country).
Taken together, these add up to significant increases in household income as a whole. When Bjorvatn and co. look at treatment impacts, there are significant increases in total household revenue from the childcare intervention (driven by self-employment) and for the cash transfer (significant at 10 percent and driven also by self-employment). The cash transfer shows significant impacts on total household labor supply. The childcare intervention does not.
Bjorvatn and co. also take a look at heterogeneity. Two really interesting results here. First, the childcare subsidy has no impact on enterprise income when the mother has another kid younger than the 3-5 year old targeted for the care subsidy. So, it seems that it’s about care for all of the younger kids, not just one (this is in line with the Delecourt and Fitzpatrick results). Second, fathers are present in 68 percent of the households. It turns out the effects on revenue for self-employment (as well as some labor supply results) are concentrated on single mothers – there are no significant impacts of the childcare intervention on revenues (or labor supply) for women with a partner.
What about the kids? The childcare intervention shows significant impacts on an index of kids’ cognitive and non-cognitive skills – an increase of around 0.15 SD. Significant sub-components include emergent literacy, motor development, and emergent numeracy (this only at 10 percent). There are no significant impacts for the cash transfer but its important to note that the positive coefficients are not significantly different from those for childcare.
So that’s income and child development. Are these folks better off? Let’s start with mom’s happiness. Here money leads the way. The cash transfer bumps up happiness by 20 percent and life satisfaction by around 16 percent. It also significantly reduces perceived stress. Childcare has positive impacts on happiness and life satisfaction, but these are about half of those of the cash transfer. While women seem to be happier after getting the cash transfer, there is also a downside – women are 6 percentage points more likely to report experiencing physical violence from their partner (childcare is a well estimated zero for this outcome). Finally, consumption for the household overall is significantly higher (driven by non-food consumption) with the childcare intervention increasing it by 16 percent and the cash transfer increasing it by 18 percent.
These results are exciting – they show promise for childcare interventions not only for the mothers, but in terms of both child and father benefits too. There is more work brewing in this space from my colleagues in the Africa Gender Innovation Lab which will broaden this argument. And maybe all of this will help get a productive policy discussion going.
Finally, dear reader, a post-script. This is my last post for Development Impact. It’s been a great eleven years. For all of you who have had kind words for me along the way, thank you.