Mini Post #3: Elder Daughter Syndrome
Mini Post #3: Elder Daughter Syndrome
How does having a baby affect the time use of mothers - and other daughters?
I’ve decided to try something new: once a week, to write a shorter, narrower post focusing on one specific paper. Last week, I wrote about how Italian families that were wealthy in the 15th century are more likely to be wealthy in the present. This is the tab with all the previous (and current) posts.
Onto the actual post: how does having a baby affect the time use, work hours, and pay of mothers - and of other family members?
Most people, unlike what the media will tell you, want to get married and have children. But like any parent will tell you, having children is really hard - especially if you already have other kids. This is doubly so during infancy, as parents (or paid caretakers) have to spend even more time around the baby, with few childcare options available. Traditionally, this “burden” of caretaking has fallen on the mother.
A major academic question, then, is whether or not this tendency has a (presumably negative) impact on women’s labor market outcomes, and why. A large chunk of the gender wage gap (i.e., the pay difference between men and women) can be attributed to a gap between mothers and men, with no comparable gap between fathers and childless women. For instance, Danish women with successful IVF treatments have lower lifetime earnings compared to women whose treatment does not succeed (one can assume that IDF success rates aren’t very related to competency).
This can have a number of possible explanations: biological factors (after all, childbirth is a major medical event), different preferences for spending times with children between mothers and fathers, gender norms, comparative advantages of women versus men in childcare (this could boost any of the other four factors), and discrimination from employers.
To clear up this question, this week’s relevant paper is “Inside the black box of child penalties: Unpaid work and household structure" (2019) by Sandra Aguilar-Gomez, Eva Arceo-Gomez, and Elia De la Cruz Toledo. Using monthly data from Mexican labor and time use surveys spread out over 15 years, the authors try to ascertain if there is a wage and labor force participation penalty after having children, and whether it affects other family members - so that it can rule out other potential explanations.
The data, coming from the Encuesta Nacional de Ocupación y Empleo (National Survey of Employment and Occupation or ENOE) by Mexican Insistute of Statistics1, has monthly measurements of how much time is used in paid or unpaid work of various kinds by all individuals over the age of 12. Regarding paid work, the variables of interest are labor force participation, time spent working, and earnings. Unpaid (domestic) work is defined as childcare or care of other family members; going shopping or doing other administrative tasks; taking household members to work, school, or appointment; home repairs or maintenance; housework or “chores”; or community services of various kinds.
Mexico is an interesting case because most studies of time use and gender discrepancies in it are from developed countries, which have high purchasing power, ample opportunities for professional childcare, and relatively egalitarian gender norms. In contrast, developing countries have worse family-related infrastructure, and tend to be more socially conservative - Latin America is, after all, more sexist than Sweden or Denmark. This means that, if the factors are social, they should be more easily noticeable by researchers.
Looking at the average family over the 15 years of the study, it’s evident that women spend more time in unpaid work than men, spend less time working, and are less likely to be in the labor force. Similarly, women earn less in paid work than men. This trend is much stronger for mothers than non mothers, but there is no similarly large discrepancy between fathers and non-fathers - though there are differences in paid labor, but not in unpaid labor, surprisingly enough.
Two factors that play an impact and are Mexico-specific are the presence (or absence) of maternity leave, and the possibility of accessing a public daycare. In Mexico, women have 12 weeks of paid maternity leave paid full wages, but around 60% of women work in the informal sector - meaning they have significantly reduced protections. Similarly, mothers working formal sector jobs have access to state-run daycares, but these have very scarce vacancies - out of roughly 2,000,000 newborns that could need daycare, they can take up to 240,000.
To estimate the impact of childbirth on employment and time use, the study uses what is called an event study. An event study is, more or less, a methodology to compare the same individual before and after a specific occurrence, called an event. The effect of the event on variables of interest is estimated by comparing people (or firms) who had the event happen to them with those who didn’t, and also people who had it happen at different points in time. The comparison is across time before and after the event - for instance, changes in income before and after getting fired from a job for both fired and non-fired people. Including other individuals is done so that there can actually be a comparison; including periods before and after helps cancel out the effect of prior trends, such as a bad economy.
Looking at a time window from one year before childbirth to 15 months after, the study measures the difference between men and women in time spent in domestic and paid work. This is tempered by controls for earnings (as a way to include the impact of, say, education), as well as impacts across age groups. One final feature are fixed effects for the individual, time and location - fixed effects are impacts common to all people who share a characteristic in the sample, so that, say, a recession, or a uniquely bad region, don’t distort the general results. The study also includes variables for gender across ages, so that the impact can have nuance across the lifecycles - perhaps the impact is only for younger moms. The big assumption is that, before and after childbirth, the only big change in the determinants of time use and income is childbirth itself.
So what are the results? Unsurprisingly, that having kids affects Mexican women’s labor force participation (negatively) both in terms of whether they have jobs or not (the extensive margin), and how many hours they work (the intensive margin). This effect does not exist for men. There is a small recovery 3 months after birth, but it is not complete, and itstabilizes from there on - meaning that Mexico’s maternity leave policy does have an impact on female labor force participation, at least.
There is also a large, positive effect for women’s unpaid work burden - after childbirth, women do around 10 more hours of unpaid work. Interestingly enough, and as seen in the figure above, this effect is also there for women who don’t leave the workforce after childbirth, and there is no significant difference between the two groups vis-a-vis changes in men’s unpaid work burden, meaning that women who work are taking on childcare on top of their existing responsabilities, rather that comparatively more of them being taken up by their husbands.
The most interesting finding: looking at other household members who aren’t the father or mother of the child, one finds a large increase in the amount of hours dedicated to unpaid work, but a very small downtick in the hours spent in paid work. This is true for both other people in the household under 18 (i.e. other children) and over 45 (assumed to be grandparents or elder relatives). This increase is exclusive to women: related men in the household, such as older brothers or grandfathers, have no statistical increase in their unpaid work burden. When a baby arrives, girls from 12 to 18 have to take on an additional four hours of childcare a week, while female aunts or grandmothers have to take on 13 hours - but none for male cohabiting relatives.
These results, then, help point out that three of the five explanations listed above don’t make sense given the facts. The biological explanation can’t be true if other women, but not other men, in the house have similar changes in patterns of behavior - they aren’t getting pregnant too. Plus, as other research finds, mothers who adopt have similar behavioral shocks - meaning that it’s not “biology”. Secondly, it can’t be anti-pregnancy discrimination, for the same reason: other women in the house aren’t getting pregnant too, almost definitionally for grandmothers. Lastly, the comparative advantage of women versus men in childcare can’t really make sense if the effect is present across ages: perhaps adults do have different opportunity costs, but it’s unclear why brothers, or grandfathers, would have significant differences. This means that the drivers of the decline in female paid work, and the increase in female unpaid work, without corresponding changes in men’s (un)paid work has to come from either systematically different preferences, or from discriminatory gender norms.
To finish up, some links:
The original paper by Aguilar-Gomez, Arceo-Gomez, and De la Cruz
A similar paper doing an event study of (un)successful IVF treatments
A “The Atlantic” article about “eldest daughter syndrome”
A previous blog post about the gender wage gap, and one about Claudia Goldin’s Nobel Prize (her research is foundational for the subject)