Does postponing fertility lower gender inequality?
This year our research will be displayed in the poster session of the European Population Conference, one of the largest gatherings of demographers and social sciences researchers in Europe. Below we include a summary of the paper (and our poster:) )
Our research focuses on how changes in fertility decisions had impacted gender inequality in the labor market. Particularly, our research asks whether fertility timing affects the adjusted gender wage gap, which is a proxy measure for labor market discrimination. To answer this question we collected individual level databases from over 50 countries covering a period of 40 years. From each database, we obtained one measure of the adjusted gender wage gap, which we later regressed on the mean age at first birth in the same year (side note: collecting this variable was a challenge on its own). The estimated coefficient shows that postponing fertility by a year reduces the adjusted gender wage gap by 2 p.p. of women's average wages (or around 12% of the unexplained wage difference).
But... wait. Fertility timing is likely endogenous. In countries where the gender wage gap is larger, the opportunity costs of childbearing might be lower, and one would expect women to have more children. To pin the direction of causality, we instrument fertility timing using four sets of instruments. Our preferred instrument is variation in the time when the pill was admitted in the country. Since the admission depends mostly on technical procedures, this measure is exogeneous (whereas how it is sold -OTC or prescription - need not be). Moreover, pill has been shown to causally affect fertility decisions among cohorts exposed at its introduction. Our other instruments include the lenght of the military conscription, the years of compulsory education and the fertility of the parents' generation. The IV estimates prove to be remarkably close to our initial estimate.
So, how can we explain this result? A possible mechanism is given by statistical discrimination theory. Imagine that employers are rational and they set wages equal to the productivity of the workers. These employers have certain expectations concerning fertility and the costs associated to it. They expect that workers will become parents during their tenure with a certain probability, moreover they expect that the costs related to childbearing (and rearing) are not equally distributed within the couple. They expect that on average the cost will be larger for women. Finally, employees would like to communicate the distribution of costs, but credible communication is not possible. Given these features, employers will discount the fertility costs in their wage offers, resulting in a gender wage gap that is unrelated to productivity. More importantly, if women postpone fertility, the probability that a young worker will give birth during their tenure falls, which is then linked to a fall in gender inequality as shown by our empirical specifications.
The model highlights the importance of two elements: probability of childbearing and differences in costs associated to childbearing. We then brought the model to the data and compare it to our empirical estimates. We find that our empirical model predicts values close to those implied by our theory, though point estimates tend to be higher. This suggests that while statistical discrimination is part of the story, there might be more elements that should be explored further.