One of the few problems with an article published this spring in the Journal of Population Economics, “Can’t buy mommy’s love? Universal childcare and children’s long-term cognitive development,” is the mismatch between its title and its subtitle. The article is a rigorous analysis of what happens when expansion of state-funded preschools “crowds out” family care. The bottom line is that introducing quality care for three-year-olds improves educational outcomes—but that is hardly everything we want to know about successful child development, and certainly not everything that a mother’s love produces.

Nonetheless, good educational outcomes are good for individuals, families, and national economies. The authors, Christina Felfe, Natalia Nollenberger, and Núria Rodríguez-Planas, made a compelling case that universal preschool has been beneficial in Spain.

Here’s how they did it: they reasoned that if shifting three-year-old children from family-based care to publicly funded preschools mattered twelve years down the line, the change should be reflected in standardized test scores among 15-year-olds. So far so good, but how do you separate changes that resulted from the expansion of state-funded preschool from change that would have happened anyway? After all, twelve years provides ample opportunities for all sorts of factors unrelated to preschool to influence standardized test scores.

The authors used data from Spain, a country that had experienced profound expansion of public preschools over the period they studied, but at an uneven pace across various states within the country. They then tested whether the change in test scores over time differed between the first states to offer more slots to three-year-olds and states that were slower to adopt the new programs. In other words, did test scores go up more twelve years later in areas where many three-year-olds were newly enrolled than in states where changes in enrollment were initially small? This is what economists call a difference-in-difference approach: Felfe et al. didn’t simply test for differences in test scores over time, but rather whether the trend in test scores differed according to how quickly states responded to the new funding. They also controlled for other differences between states besides how quickly they expanded three-year-old school programs.

Reading test scores in fact went up more in the states where more children enrolled in public school at age three. That by itself is a fairly compelling case for the merits of universal preschool. But the part I found even more compelling was that improvements among girls and socioeconomically disadvantaged children drove the overall results. It is hard to close educational gaps between kids from privileged backgrounds and those from disadvantaged backgrounds. Family background tends to matter more than school quality in determining children’s educational outcomes, especially in industrialized countries. And yet this evidence from Spain shows that schools can in fact help close the gap. I think we all want to believe that schools can matter, to believe that poor kids given good schooling can overcome the disadvantages associated with their background. It doesn’t work that way as often as we’d like. But if good schooling starts at age three, it seems it can contribute to greater equality.

Introducing state-funded three-year-old programs did not improve standardized math test scores. The authors explain that the programs introduced in Spain targeted emotional and social competencies that undergird literacy development. The competencies needed for math develop at later ages. Children affected by the programs were also about 50 percent less likely to repeat a grade in primary school (but not secondary school). In other words, the positive effects the researchers document were not consistent across all educational outcomes, but all the significant effects were good ones.

I wish these results were generalizable beyond Spain: it would be great to have a recipe for better reading skills and less grade repetition that also reduced socioeconomic inequality. Universal preschool might produce similar outcomes elsewhere, but the reason we can’t be sure is that reduced child care costs often increase the labor force participation of mothers with young children. In Spain, however, government funding of three-year-old programs did not have a significant effect on women’s paid work, and that affects how we should interpret the study’s findings.

It might seem surprising that Spanish women don’t work more when their children were taken care of from 9 a.m. to 5 p.m. for free, but the siesta splits the workday into 9 a.m. to 2 p.m. and 5 p.m. to 8 p.m. If women had taken up full-time work in response to universal preschool, their kids would have spent very few waking hours in maternal care. But they didn’t. Three-year-olds got a play-based curriculum focused on the social and emotional competencies that build toward reading skills, and only about one-third of them had mothers working for pay (both before and after state funding of three-year-old programs). That means that two-thirds of them had mothers juggling fewer responsibilities during the day and who might have given them more attention in the evenings. I’m sure I would read to my children more in the evenings if I had days to do things like grocery shop and pay bills.

In a country where women did take up paid work during preschool hours, children might not get as much direct attention in the evenings. That might be OK: preschoolers can learn a lot from “helping” in the kitchen, and they are already being read to during the day. On the other hand, the net positive effect in Spain might have resulted from the combination of what actually went on in school and the way it reduced the demands on maternal time. That means I’m not sure how much this excellent research from Spain contributes to the fervent political debate concerning the extent to which governments should provide sufficient, affordable childcare.