This is how women anticipate the employment effects of motherhood

Women’s lives have changed drastically over the past century. Women have surpassed men in earning college degrees, and they are having children later than ever, accumulating increasing amounts of on-the-job experience. Yet, since about 1990, convergence in labour force participation between men and women has stalled, even as societies have made progress on workplace gender discrimination (Blau and Kahn 2017). Why has this stalling occurred, especially as women are more prepared for the labour market than ever?

In a new paper, we propose a novel explanation for this phenomenon – that when they are making their key human capital decisions, women do not anticipate the effect that children will have on their careers (Kuziemko et al. 2018). We conclude with some speculations as to why and focus on a key trend – that the employment costs of motherhood have risen unexpectedly, and especially so for educated mothers.

Children continue to pose a considerable cost to careers, and that cost is disproportionately borne by women. Recent studies by Angelov et al. (2016) and Kleven et al. (forthcoming) using rich administrative data from Sweden and Denmark, respectively, provide compelling evidence that women’s careers diverge from that of men’s only after the birth of the first child. Using several nationally representative longitudinal datasets in the US and the UK, we replicate these findings, and estimate that in the five to ten years after birth, women’s employment rates decline by about 25-40 percentage points and show little evidence of recovery relative to the counterfactual no-baby trend (Figure 1 for the US). For men, we find little to no analogous effect.

Figure 1 Event-study analysis of how probability of employment changes after motherhood in the US

Our focus, however, is not primarily on these child penalties, but instead on why women continue to invest so heavily in their human capital development and yet drop out of the labour force at such high rates upon motherhood. We provide several pieces of evidence that suggest a potential role of the lack of anticipation of the career costs of motherhood, both in the short run and in the long run.

The surprise

If women are not fully anticipating the employment effects of motherhood, we would expect that the birth of the first child serves as an information shock, causing them to update their beliefs regarding their ability to maintain both family and work commitments. Consistent with this idea, we find that women become more traditional in their attitudes about gender roles regarding work and family after the birth of their first child. For example, after becoming a mother, a woman is more likely to agree with statements such as “A family is harmed when the mother works outside the home,” and disagree with statements such as “Both a husband and a wife should contribute to household income.” Using an index of gender role attitudes that we construct using the available questions in each dataset, we find that women’s attitudes become significantly more ‘anti-work’ in the first two years post-baby and exhibit no recovery in the five years after birth (see Figure 2 for the UK). In fact, in the UK, women’s attitude change post-baby is more than the average difference in gender-role attitudes between men and women. Interestingly, college-educated women appear most caught off-guard by motherhood, as they experience a larger ‘anti-work’ shift in their views than non-college-educated women. We find little evidence of a significant shift in men’s attitudes.

Figure 2 Event-study analysis of how gender-role attitudes change after motherhood in the UK

More directly related to our hypothesis, we look at survey questions from the PSID that ask parents if they think that parenthood was harder than they expected. Consistent with the finding above that educated women are most surprised and update their beliefs the most, we find that women agree with this statement much more than men, and, moreover, that college-educated women agree much more than non-college-educated women.

Why are college-educated women so caught off guard by the employment costs of motherhood, especially since motherhood is such a common occurrence? We hypothesise that expectations regarding the employment costs of motherhood play a major role – specifically, that women who foresaw a career requiring a college degree were exactly those women who believed that balancing work and raising children would be easier than it turned out to be. We specify a model with this dynamic at play, where women inherit an imperfect prediction of employment costs from their mothers and use this prediction to decide whether or not to attend college. The model shows that women who inherit the lowest predictions of costs attend college, whereas women with the highest predictions do not. Upon motherhood, true employment costs are revealed, and these educated women are negatively surprised by how difficult it is to combine work and family. Our model also demonstrates that our empirical results can only be reconciled by an unexpected increase in the employment costs of motherhood across generations.

The longer view

How have expectations about career and family changed over time? To examine longer-run trends in career expectations, we look at what high school seniors thought they would be doing in their 30s and compare these expectations to what their cohorts ultimately did (Figure 3). We focus on high school seniors because they are at an age when they are making key decisions about attending college. As late as 1968, over 60% of female high school seniors expected to be homemakers at age 35. This expectation fell drastically over the ensuing decade, and by 1978, roughly 10% of high school seniors expected to be homemakers at age 30. Today, virtually no girl in her senior year of high school expects to be a homemaker when she turns 30.

Figure 3 Expectations and realizations of future home-making among US high school seniors

When we compare these expectations to reality, however, an interesting trend emerges. Whereas the class of 1968 overestimated the probability that they would be homemakers, the class of 1978 underestimated that probability. In particular, while over 60% of girls in the class of 1968 expected to be homemakers at age 35, only about a quarter of them ultimately were. (Note that all outcomes in this graph are plotted according to the year that a woman graduated from high school.) On the other hand, about 10% of the class of 1978 expected to be homemakers at age 30, but ultimately twice as many of them were. Since then, every cohort of graduating seniors has underestimated the likelihood that they would become homemakers.

Why has this reversal in expectations versus reality occurred? We argue that the employment costs of motherhood play a key role and, specifically, that across generations, there has been an unexpected increase in these employment costs. Several papers have emphasised the decline of costs of motherhood as a key driver of the rise in female labour force participation over much of the 20thcentury. For example, electrification that led to the widespread adoption of labour-saving household appliances (Greenwood et al. 2005) and medical innovations such as reduced morbidity associated with childbirth and infant formula (Albanesi and Olivetti 2016) enabled women to better reconcile work and family responsibilities, allowing them to enter the labour force in large numbers.

More recent data, however, suggest that these trends may have played out and employment costs of motherhood may, in fact, have risen for more recent cohorts of women. Time-use surveys document that time spent on childcare rose drastically around 1990, and especially so for college-educated mothers (Ramey and Ramey 2010). Figure 4 plots these time-use estimates, as well as other indicators that suggest that the employment costs of motherhood at different child ages have risen for more recent cohorts of new mothers.

Figure 4 Various proxies for ‘employment costs of motherhood’

Childcare costs over the past 50 years have roughly followed a U-shape. For example, breastfeeding rates fell in the 1950s and 1960s (in the US) but then began rising again in the 1970s. Other proxies for employment costs of motherhood, such as the dollar cost of childcare, have also been on the rise in the recent few decades.

Conclusion

Researchers have puzzled over the stalling of the gender gap in labour force participation, especially given concurrent trends that suggest that women have been investing increasingly in human capital accumulation. We propose a novel explanation for this puzzle: that when women are making key decisions about human capital investment, they underestimate just how costly child-rearing will be to their careers. We use several pieces of evidence to provide support for our hypothesis. Most notably, women (but not men) become more ‘anti-work’ in their attitudes about gender norms regarding work and family upon becoming mothers, and educated mothers are most surprised by the employment costs of parenthood. Over the longer run, cohorts of graduating female seniors once overestimated the probability that they would stay at home and raise children, while more recent cohorts of graduating seniors now increasingly underestimate that probability. We provide some speculative evidence that rising costs of childcare, after decades of decline, may be responsible for this lack of anticipation on women’s part. To address the remaining gender gaps in the labour market, there is a pressing need for future work to examine the underlying causes of this apparent increase in the employment costs of motherhood.