To Work or Stay Home? An Economist Digs into the Research
My wife, Katie, and I both finished graduate school two years ago and are now employed in our respective fields; I am an economist, and my wife is a lawyer. We also recently welcomed our first child—a beautiful baby boy! Amidst all the excitement there lurks a question: who will take care of the baby when our parental leave is over?
As an expectant father, I have been tuning in to the wild debate that the world is having about parenting, and as a social scientist, I’ve generally been shocked by how much poor thinking and bad information is out there. There’s everything from books insinuating that the rise in female labor force participation is the culprit of the rise in mental health problems among teenagers, to Instagram posts shared at the speed of light claiming that “the data” says that a mother staying home has no effect on her kids so it doesn’t matter what you do. As someone who has spent years writing and reading academic research, I don’t have a lot of patience with these sort of extreme claims unsupported by actual research, so I decided to get into the research myself to figure out who was right.
Specifically, I wanted to know what the research had to say about the effects of different childcare arrangements on children, both in the short term and long term. Most of the academic research compares maternal care and the child being cared for by another adult (whether that be at daycare or in a non-institutional setting), so comparing these two childcare arrangements will be my focus.
The takeaway from all this research is not that a mother staying home to care for her children has a huge effect and everybody should be doing it or else you’ll ruin your kid’s life, and the takeaway is also not that it doesn’t matter at all—even though it feels like those are the two main narratives you will hear about this topic. The reality is more nuanced. The existing research makes clear that having a mother primarily care for a child does matter but not on a life-changing scale for most kids. What is also clear is that there are simply a lot of questions that we cannot answer given the current state of the research, and it’s important to be transparent about that. I hope you’ll join me on this journey, both because the topic at hand is important and because learning to think critically about data and research is a vital skill to be able to understand the complicated world in which we live.
Causality is tricky
Understanding the causal effect of the mother caring for her children (as opposed to some other arrangement) turns out to be a very tricky thing to do. A researcher’s first impulse may be to simply collect data on many families, including whether or not the mother cared for the children, and then see how the kids turned out, but this would lead to misleading results. Mothers who stay home to care for their children tend to be different in many ways from mothers who return to work quickly and choose alternative childcare arrangements. So if we were to see in the data that the children of mothers who fill the primary caregiver role are different from children of mothers who did not, we actually don’t know if the reason the children turned out differently is that they were cared for by their mother full time or if differences exist because their family had a different socioeconomic, educational, or cultural background. It is possible to control for many of these factors, but it is simply never possible to control for all of them.
The gold standard to deal with these unobserved factors and understand causality is to run randomized controlled trials. Due to ethical concerns, we, of course, would never do a fully randomized controlled trial where mothers are assigned to either stay home or go back to work and then researchers evaluate differences between the children. Thankfully, there are actually policy changes that have occurred that create some quasi-random variation in how long mothers care for their children at home versus work outside the home. We call these natural experiments. What kinds of policies affect how long a mom stays home with her children? Parental leave policies.
Parental leave and natural experiments
When parental leave policies are introduced, they are often introduced such that the mother benefits from the new policy only if her baby was born after a certain date. For example, when Norway expanded its family leave policy in the 1970s, mothers who gave birth to their baby on or after July 1, 1977, were eligible for four months of paid leave, whereas mothers who gave birth before that date were only eligible for four months of unpaid leave. Getting paid makes taking leave a lot more feasible for most families, so as a result, a woman who gave birth on July 2 ended up being much more likely to stay at home with her new baby for at least four months compared to a very similar mother who gave birth on June 28, all because of something that (from their perspective) was essentially random.
Assuming there are no other policy changes that were implemented on July 1, we can be confident that differences in the outcomes of children who were born before versus after July 1 are the effect of the maternity leave policy (and thus the effect of a mother staying home longer with her children) rather than other factors, because the policy changed how long mothers stayed home with their kids in a way that was uncorrelated with other factors that affect kids’ outcomes.
What is the effect of a mother working in the first few months of a baby’s life?
There is only one economic study looking at the effect of going from no paid leave to some amount of leave, and its findings are very clear. Economists Pedro Carneiro, Katrine Loken, and Kjell Salvanes analyzed the effect of the policy change in Norway described above. In their analysis, they find that “the increased time spent with the child led to a 2 percentage point decline in high school dropout rates and a 5 percent increase in wages at age 30.” They also found that the leave policy increased the likelihood that the kids would go to college. The effects here are really clear and speak to the importance of a mother being able to be with her child during the first few months of its life. While the US generally has poor parental leave policies (federal law provides for only 12 weeks guaranteed unpaid leave and only for most, not all, workers and circumstances), these findings suggest that it is probably worth it to take leave for your child, even if the leave is unpaid.
The effects here are really clear and speak to the importance of a mother being able to be with her child during the first few months of life.
If we just consider dollars and cents, if you as a mother earn the median US wage of $54,000 per year, then if you were to take four months of unpaid leave, you will sacrifice about $18,000 of income, but the researchers find that you are getting your child 5% higher wages at age 30. Assuming your child then earns the median wage and gains that extra 5%, that works out to be an extra $2,700 per year for them. In less than seven years of your child in the labor force, that investment has already paid off.
What is the effect of a mother working after the first few months of her child’s life?
While the last question had a very clear and straightforward answer, this question is much more complicated. A great deal of research has been done, and the findings do not paint a tidy story. A whole bunch of researchers have found that an increase in existing maternity leave (for example, increasing paid leave from 6 months to 12 months or even from 12 months to 24 months) has no effect on any measured outcomes like educational attainment, test scores, or behavioral development. But then a bunch of other researchers have found that increasing paid leave does have an effect on important outcomes. Which researchers are right?
Two ways to find out
Without getting too technical, all of the studies that look at the effect of maternity leave on the baby’s life use one of two econometric models to estimate the effect: a difference-in-differences model or a regression discontinuity model. Difference-in-differences models are generally reliable, but recent work has shown that difference-in-differences models may give incorrect results in some settings. These problems with difference-in-differences models were not well understood when these studies were written, and so it is possible that some of their findings may be incorrect. On the other hand, regression discontinuity models are better understood, and the conditions required for them to give correct results can be more convincingly tested and evaluated. When we only look at papers that used a regression discontinuity model, a clearer picture emerges.
There are three studies that use a regression discontinuity model (the more trustworthy method in my opinion), and of these three studies, two of them find that a mother staying home can positively affect the child and one finds no effect. (Here are the three studies: Rasmussen, 2010; Danzer and Lavy, 2018; Ginja, Jans, and Karimi, 2020.) The two studies that do find that there is an effect are measuring the effect of substantial increases in paid maternity leave—one policy changed leave from 12 months to 24 months, and the other policy varied by baby but was on the same large scale. On the other hand, the one study that finds no effect is studying the effect of only an increase from 14 weeks to 20 weeks of maternity leave. Thus, it seems reasonable to suppose that this study may have only found no effect because the treatment size was so small (it’s statistically difficult to measure an effect of such a small increase in maternity leave).
What exactly do these studies find? The Rasmussen study finds no effect of a six-week increase in maternity leave on long-term educational outcomes. On the other hand, the Danzer and Lavy study finds that a 12-month increase in maternity leave improved science and reading test scores for the sons of highly-educated women (women with a tertiary degree) at age 15, though they find that increased maternity leave may have decreased reading test scores for sons of women without tertiary education. Lastly, the Ginja, Jans, and Karimi study finds that a large increase in the length of maternity leave increases the likelihood that boys will be in college by age 24 by 2 percentage points (a change in probability from 37.5 percent to 39.5 percent).
A mother staying home to care for her children certainly does affect the child’s educational outcomes, but these effects seem to only exist for boys, and the effects are relatively small.
So when we restrict our view to just the studies that use the more trustworthy method, it seems that a mother staying home to care for her children certainly does affect the child’s educational outcomes, but these effects seem to only exist for boys, and the effects are relatively small.
Other research to note
I want to highlight a few other studies that don’t use my preferred method but still deserve mention. Sayour (2019) finds that a three-month increase in maternal care improves a baby’s emotional disorder scores but has no other effect on short-term cognitive, non-cognitive, or health outcomes. Fabel (2021) finds that increasing maternity leave from two to six months led to about two fewer hospital admissions over the course of the baby’s life, where the baseline is about 121 hospital admissions per person, and this is driven especially by hospital admissions due to mental and behavioral disorders. Again, not a large effect but also not nothing.
It is clear to me that [Emily Oster’s] conclusion that there is no effect whatsoever [of a mother choosing to be the primary caregiver] was incomplete.
I also want to note that my conclusions from the research literature are different from the conclusions Emily Oster comes to in her well-known book Cribsheet. She concludes that there is no effect whatsoever of a mother staying home to care for her child. I’m a big fan of Oster’s work injecting more science into these debates, and her book was how I originally approached this topic. That said, I think there are two reasons for the difference in our conclusions. One is that her book was published in 2020, and a few of the papers that really clarified the picture were published in 2020 and 2021. Second, I think she may not have done as close a reading of the research literature as I did; even without the papers published in 2020 and 2021, it is clear to me that her conclusion that there is no effect whatsoever was incomplete.
What this research doesn’t tell us
Now that we’ve talked about what this research does tell us, it’s worth stepping back and taking stock of what this research doesn’t tell us. First, there are a lot of important outcomes that this research doesn’t have a lot to tell us about. We have learned about the effect a mother can have on her child’s academic success, but this research doesn’t tell us anything about the effect that a mother staying home has on the quality of the relationship between the parent and child. It also doesn’t tell us much about the effect the mother might have on a child’s happiness later in life (though the research on hospital admissions in the Fabel (2021) study at least looks at something similar to it). There may be some effect or none at all. We simply don’t know. This is, in my opinion, a huge hole in the research literature.
As parents, we care about our kids’ academic success, but the most important thing is that we want our children to be happy, and we want to have happy, fulfilling relationships with our children. The research can’t really say anything about whether a mother staying home with her children affects these outcomes.
This research doesn’t tell us anything about the effect that a mother staying home has on the quality of the relationship between the parent and child . . . [or] the effect the mother might have on a child’s happiness later in life. . . . This is, in my opinion, a huge hole in the research literature.
Why has the research failed us in this way? The most likely explanation is this: the research that really does a good job of isolating causal effects tends to have been done by economists, and economists tend to be very wary about working with outcomes that are hard to measure objectively—like happiness and relationship quality.
The second thing that this research doesn’t tell us is how these effect sizes change over the course of a child’s life. The papers discussed above all look at the effect of a mother staying home during the first few years of the baby’s life, because that’s what the maternity leave policies affected. We don’t know if a mother would have a bigger effect, a smaller effect, or the same size effect on a child if she stays home with them from ages three to five or longer, because there simply haven’t been natural experiments that have occurred that would allow us to answer this convincingly. Because of this, we don’t know what total effect a mother staying home would have on her child.
A brief note on daycare
Before I close, I want to talk briefly about daycare. When we talk about a mother staying home to care for her kids due to increased maternity leave, the mother is replacing some other type of care, and while for most of the studies cited above the alternative was non-institutional care (most likely friends, relatives, or a nanny), for many people the alternative is daycare.
Evaluating a child placed in daycare against a child whose mother is at home with them can be very nuanced. A lot of research has been done evaluating daycare; some researchers have found that daycare can have positive effects, particularly for kids of low socioeconomic status, and other researchers have found that daycare can have negative effects.
The most notable study in the latter group was done by Baker, Gruber, and Milligan (2019), who studied the effect of the introduction of universal daycare in Quebec. This study is cited in pretty much every anti-daycare article you’ll ever see, because they found that the introduction of universal daycare led to “worse health, lower life satisfaction, and higher crime rates later in life” for the affected cohorts of children. This is certainly alarming, but in light of the other studies that find that daycare can have positive effects on children, to me what it really highlights is the fact that the quality of daycare varies widely, and clearly the universal daycare in Quebec was of particularly bad quality. Delving into what makes for good daycare is beyond the scope of this article, so I’ll limit myself to just saying that sometimes it can be good, sometimes it can be bad, and you’ll want to evaluate the quality of a daycare yourself if you are considering it.
The research literature on the effect that a mother has in staying home to care for her children is large, complicated, and incomplete, and sometimes it doesn’t make sense. A healthy way to interpret the evidence is this: moms who stay home with their kids do make a difference, but if you decide that working outside the home is the right thing for you, you are by no means ruining your kid’s life.
A healthy way to interpret the evidence is this: moms who stay home with their kids do make a difference, but if you decide that working outside the home is the right thing for you, you are by no means ruining your kid’s life.
My wife, Katie, and I still have not figured out what we are going to do for our little boy’s care. The evidence I reviewed here is only one of many factors in making that important decision. Every family situation is different, and I hope this article will be helpful as you consider the costs and benefits of different care arrangements for your children.
Editing credential to Bethany Bartholomew