13 Mar
Correlation

Association, Aggregation, and Paradoxes: On the Positive Correlation Between Fertility and Women’s Employment

Can we use cross-country, macro-level correlations between fertility rates and women’s employment rates to study the extent to which women combine work and family? I tend to think this is not very fruitful. Today, the journal Demographic Research published my note on a recent macro-level article.

In my note, titled Association, Aggregation, and Paradoxes: On the Positive Correlation Between Fertility and Women’s Employment, I respond to a recent article by Brehm and Engelhardt. Their article revisits the cross-country correlation between total fertility rates (TFR) and female labour force participation rates (FLFP). The interesting thing about this correlation is that it turned from from negative to positive after 1985. My disagreement with their (otherwise excellent) article is that the pre-1985 negative correlation is taken as support for the hypothesis that for women having young children and being employed are (partially) incompatible, implying that the correlation turning positive contradicts that hypothesis regarding the later period.

My note provides three comments on why this cross-country correlation is not informative to critically test hypotheses on the degree to which women combine motherhood and employment:

  1. The macro-level correlation across countries turned positive due to decreasing fertility in southern European countries, but this was hardly associated with more female labour force participation. This is not in line with the notion that within countries higher fertility was associated with more employment.
  2. There is a whole literature on aggregation paradoxes, that dictate that correlations on different levels of aggregation can have opposite signs. So, a positive correlation at the aggregate country-level is not informative regarding a correlation at the individual level
  3. In my own study in Journal of Marriage and Family I used individual-level data to find that mothers were still (substantially) less likely to be employed than women without children. Moreover, in various countries the individual-level association between motherhood and employment did not change at all in the period that the country-level correlation turned positive.

The original article by Brehm & Engelhardt and my response are available online from the Demographic Research website. Those who follow my research will recognise some arguments that were developed in my dissertation (Family Policy Outcomes).

On a final note, the editorial team of Demographic Research has been incredibly efficient in processing this note, and seem very committed to facilitate academic debate in their journal. The whole process (from submitting to publishing) took just a couple of days, and given that the original paper was published only a week ago, this makes for a timely discussion.

Leave a Reply