Does The Child Earnings Penalty Actually Exist?

The prevailing consensus in the economics literature is that women suffer a significant earnings penalty after they have children. More precisely, women who have children end up with lower earnings than women who do not, holding all else equal.

Recent research using Nordic administrative data have called this consensus into question.

  1. Reconciling Estimates of the Long-Term Earnings Effect of Fertility (Norway)
  2. Is There Really a Child Penalty in the Long Run? New Evidence from IVF Treatments (Denmark)
  3. The Economics of Infertility: Evidence from Reproductive Medicine (Sweden)
  4. The Labor Market Returns to Delaying Pregnancy (Sweden)

In a typical study on this question, the authors compare the earnings trajectories of women who become mothers at a certain point to women who do not become mothers at that point. These studies generally show large short-term and long-term earnings penalties associated with becoming a mother.

This approach shows that women tend to see their earnings levels flatten out around the time that they have kids. This could be because having a kid causes women’s earnings levels to flatten out due to responsibilities that come with having a kid (the “child penalty” scenario). Or it could be because women choose to have kids around the time that their earnings levels are already flattening out. If it is the latter, then the “child penalty” does not really exist.

These recent Nordic studies take a different approach to estimating the size of the child penalty. Rather than looking at women who have children at a certain point and comparing them to women who don’t, they look at women who are undergoing IVF fertility treatments and compare women who succeed at getting pregnant through IVF to women who fail at getting pregnant through IVF.

Using this approach, the authors find that the size of the child earnings penalty is much smaller than conventional estimates.

However, when it comes to the separate estimates for mothers and partners, the event-IV model paints a different picture than the event-study model. For mothers, it estimates only a small longrun negative impact of children on earnings of 3 percent. While we cannot reject a zero effect on earnings, we strongly reject the event-study estimates

[W]e find that the birth of the first child does not lower lifetime female earnings. If anything, we calculate that motherhood leads to a small rise in lifetime female earnings cycle of 2-3 percent. Based on these findings, we conclude that there is no evidence of a sizable long-run child penalty.

At the same time, we find no protective long-run impact on labor-market incomes. This finding is at odds with a large literature that has documented a negative effect of child-bearing on female earnings, which hence may be expected to imply a large positive effect of infertility on earnings. The lack of a “child penalty” in our estimates likely reflects the fact that our analysis focuses on women who reveal a preference for having a child, and is consistent with a small body of evidence using IVF-births. Among women who want a child, our results suggest that remaining childless does not have a long-run protective effect on income. Women who have a successful first assisted conception experience a short-run large drop in income, but catch up with women whose conception was unsuccessful within a year after childbirth.

People who are seeking IVF are, of course, different from the overall population in certain respects. For one thing, their pregnancies and subsequent births are necessarily planned. They also tend to be older. Americans reading this post will also object that they also need to be able to afford IVF, but that is less true in these countries due to their national health insurance programs. For instance, in Norway, the national health insurance covers three rounds of IVF.

The last study in the list above is especially interesting because, in addition to looking at people who succeed or fail at IVF, it also looks at people who get pregnant while on long-acting reversible contraceptives (LARCs). Just as IVF provides a good way to look at people with planned pregnancies, LARC provides a good way to look at people with unplanned pregnancies.

This points towards the conclusion that people with unplanned pregnancies and births face a child penalty while people with planned pregnancies do not. This also bolsters the general conclusion that people tend to plan their pregnancies around the time that their earnings trajectory is already flattening out, causing most research on the magnitude of the child earnings penalty to be overstated.

This presentation of things matches with lots of little threads you see in the discourse on family-formation and fertility but adds some useful caveats and clarifications:

  1. Many people do pursue having kids as a “capstone” event to their life but the particular thing they are capping off is hitting their labor market potential.
  2. Many people are at least implicitly prioritizing work (i.e. hitting their labor market potential) over family formation and delaying accordingly, consistent with the “workism” discourse.
  3. The concern that having a kid too early (i.e. before hitting one’s labor market potential) will affect one’s career and lifetime earnings is based in reality.
  4. Absent unplanned pregnancies or a change in norms about the importance of earnings, the age of first birth will tend to converge later in the fertility window because that is when people have had a chance to move beyond their entry-level jobs.
  5. Given (4), average lifetime births will decline because, among other things, it is not possible to have very many kids when age of first birth is pushed so far back in the fertility window.