When economists estimate the return to education, they usually limit their sample to workers – or even full-time workers.  This is a serious error.  If you were evaluating the return on business loans, you wouldn’t limit your sample to firms that remained in business.  The right measure, for students and borrowers, is unconditional: If part of your sample shows zero payoff from prior investment, include their zeroes in your sample.  After this adjustment, your estimated return to education will markedly fall – especially for groups with lower employment rates.*

This arithmetic has an awkward implication.  Since males continue to have higher labor force participation than
women, conventional estimates seriously overstate women’s return to
education.  The more traditional the society, the greater the overstatement.  This labor force participation graph ballparks the size of the effect.


[Blue=men, Pink=women, Black=combined; source: Wikipedia]

You could object that rising female education and rising female labor force participation were connected.  Fair enough.  But education was far from the only factor.  Female college grads in 1950 were foreseeably a lot more likely to spend their lives as housewives than they would be today.  Why?  Because values and aspirations were more traditional.  As a result, the unconditional female return to education in 1950 was low.

Why do I bring this up?  Because I just read a passage where a sociologist makes this point to chide a pair of eminent economists.  From David Labaree‘s critique of Goldin-Katz in Someone Has to Fail:

Consider one fact about the growth of educational opportunity that is a particular point of pride for [Goldin and Katz]: unlike the European model in the late nineteenth and early twentieth centuries, the American high school system educated girls at a high rate.  In fact, girls graduated from high school at a substantially higher rate than boys through the twentieth century, and this was especially true in the early part of the century, when girls had a graduation rate that was 39 percent higher than boys.  But during the same period, most women were not entering the workforce.  In 1910 the female proportion of the U.S. labor force was 18 percent and by 1940 it had only risen to 25 percent, when female graduation rates reached 60 percent.  So a large portion of the investment in high school education was going to students who were not part of the formal economy and thus could not be considered as a contribution to gains in economic productivity.

Why is Labaree so willing to point out facts that makes the typical economist squirm?  Here’s my guess: Economists live by cost-benefit analysis.  Praising investments with low returns makes our skin crawl.  Sociologists, in contrast, are pluralists.  If female education has a low return, a sociologist can still praise it with a clean conscience.  So despite economists’ superior quantitative skills, it is sociologists who find it psychologically easier to do the math.

* This is not true in a simple model with zero tuition and age-invariant employment rates, but it is true in the real world.