Stressed populations yield fewer males?

When a country experiences a dramatic, faltering economy, there is an equally dramatic, faltering effect on the birth rate. Specifically, there are more female births than male births, according to new research from the University of California, Berkeley. Although this phenomenon has long been observed in herd animals, this is the first time it’s been shown to happen in humans, too.

Here is the full article. Here is a press release from Berkeley. Here is Ralph Catalano’s [the researcher] home page. Here is the article itself, but the whole thing will cost you $19.00.

By the way, here is another puzzle: more boys are born in southern Europe than in northern Europe. More females are born in Mexico. Could this all correspond to stress? Here is another study, it says that stressed out parents are more likely to have girls.

My take: I can believe the result, but I haven’t much raised my priors. Note that the spike in East German female births comes in 1991, when the East German economy collapses. OK, we have found some related results elsewhere, and for other animals, that counts for a good bit in my eyes. Nonetheless I am reminded of Brad DeLong’s excellent article Are all Economic Hypotheses False?”, with Kevin Lang. If you spend enough time looking, and there is publication bias, some results will pop up, whether they are true or not. The cynical would say that behind this article, as described in the press releases, there is really only a single data point, namely 1991. N.B.: I haven’t paid the $19 to read the article itself, final judgment is reserved for those who pay and read.

Addendum: Jon Klick read the piece and sent me the following:

You’re exactly right that you’re really only focusing on one data point to identify the hypothesized effect. One wonders why he didn’t include the GDP numbers directly rather than simply including a “collapse” variable. Doing so would have allowed for more variation. If the claim is endogeneity (isn’t it always), then that’s not really solved through the collapse variable since it would be endogenous too. Also, isn’t it kind of funny how he uses a 55 year pre-period relative to an 8 year post period. If the ARIMA structure went through any significant changes around the 1991 period, you would be implicitly weighting your early data (and their ARIMA parameters) more heavily than the potentially new parameters. Thus, the 1991 change could be due to these structural parameters rather than anything GDP related; impossible to sort the effect in this design.

…[it] potentially suffers from huge omitted variable bias. That said, there are intuitive reasons to believe the thrust. Apparently, boys are more expensive in every way during the early years of life (e.g., they have more health problems which cause financial and physical drains on parents . . . this shows up in the evidence that mothers of boys die earlier and the like). Presumably then, the reproductive system has developed such that boys are spontaneously aborted when their births are likely to be most troublesome (e.g., when economic and psychological stresses are particularly acute).