Thanks for doing this! I was curious so I dug a bit more into the Federal Reserve paper. I think you are accurately reporting their conclusion, but the conclusion feels somewhat shaky to me.
First, they note that, without controls, people with children are generally more productive than the childless, not less:
They then introduce a number of different controls. The âThe total productivity cost of having one, two, or three preteens was 9.5%, 22%, and 33%â statistic you report is one model, and it may be worth noting that the 9.5% number (= exp(â0.1003) â 1) is not significant at p = 0.1. Amusingly, the impact of getting tenure dwarfs the impact of children, causing an estimated 50% decrease in productivity:
They then introduce a difference-in-differences estimation âto remove any estimation bias related to time-invariant unobserved heterogeneityâ. They say of this modelâs results:
The estimate is negative, relatively small, and statistically not significant, implying that pregnancy, childbearing, and motherhood have no large effect on the motherâs research productivity in the first three years after the birth of her first child⌠The only statistically significant effect of fatherhood relates to tenure in column (6) of Table 9. Untenured men become substantially more productive after having a first child. Their output increases by 20%
Iâm not sure why they chose to report their first model in the summary but not the second one. The cynical explanation is that they did this because only the first model returned significant results, although then Iâm left wondering why they included the second model at all.
Anyway, my personal summary of this paper is something like âdepending on how you slice the data, you can conclude anything from âhaving kids is a mild costâ to âhaving kids is a mild benefitâ, and it feels hard to confidently say much more.â
(To be clear: I donât think you are citing this paper incorrectly, I just think the paper itself is reporting its results in a confusing way.)
Thanks for doing this! I was curious so I dug a bit more into the Federal Reserve paper. I think you are accurately reporting their conclusion, but the conclusion feels somewhat shaky to me.
First, they note that, without controls, people with children are generally more productive than the childless, not less:
They then introduce a number of different controls. The âThe total productivity cost of having one, two, or three preteens was 9.5%, 22%, and 33%â statistic you report is one model, and it may be worth noting that the 9.5% number (= exp(â0.1003) â 1) is not significant at p = 0.1. Amusingly, the impact of getting tenure dwarfs the impact of children, causing an estimated 50% decrease in productivity:
They then introduce a difference-in-differences estimation âto remove any estimation bias related to time-invariant unobserved heterogeneityâ. They say of this modelâs results:
Iâm not sure why they chose to report their first model in the summary but not the second one. The cynical explanation is that they did this because only the first model returned significant results, although then Iâm left wondering why they included the second model at all.
Anyway, my personal summary of this paper is something like âdepending on how you slice the data, you can conclude anything from âhaving kids is a mild costâ to âhaving kids is a mild benefitâ, and it feels hard to confidently say much more.â
(To be clear: I donât think you are citing this paper incorrectly, I just think the paper itself is reporting its results in a confusing way.)