Male and Female Earnings Trends

The statistics on education and earnings presented by Becker are dramatic, but also puzzling, at least superfically. Why should the sex ratio of either education or earnings change over a relatively short period of time (30 or 40 years)? It is fairly easy to explain the growth over this period in the percentage of women who work full time in the market rather than in the household—improvements in contraception, a fall in the marriage rate (though that is a function in part of women's higher market earnings potential), a reduction in the demand for children (also, however, in part a function of that higher potential), the shift from a manufacturing to a service economy (and the growing automation of manufacturing), and the growth in household labor saving appliances—all these things have contributed to increased female participation in the labor force, but it is not obvious that they would increase the ratio of full-time female earnings to full-time male earnings.

It's not as if there's been a relative increase in the number of jobs for which women are better suited than men. Women are not as well suited to perform jobs requiring upper-body strength as men are, but men can perform virtually all service jobs as well as women can. So an increase in demand for service workers should draw men as well as women into such jobs, leaving the gender wage ratio unchanged. Similarly, one would expect an increase in the returns to education to affect men the same way it would affect women, so that relative graduation rates would not change and therefore would not affect relative earnings.

One factor in the increased ratio of female to male earnings is undoubtedly that until quite recently most women who worked full time were unable, unless they were unmarried, or married but childless, to spend as many years working full time as they are able to do today. They would have to take years off from full-time employment to take care of their children, and so would be investing less in their human capital than male workers and therefore earning less. And they would tend to cluster in full-time jobs that involve short work days, notably teaching, and so more of their compensation (relative to men's) would take the form of leisure, as distinct from pecuniary income, than men's compensation would. Law, and especially medicine, fields that require protracted education and long hours—and are compensated accordingly—would be unattractive professions for most women.

Another factor is discrimination. Women were largely excluded from the major professions until the 1960s (in part at least because of expectations that they would become full-time practitioners), and their educational opportunities were limited until then as well—many elite colleges and professional schools did not admit women. Beginning in the 1970s, antidiscrimination laws corrected but also overcorrected sex discrimination, by placing pressure on employers to hire and compensate women at higher rates than justified by labor costs. For, in order to avoid accusations of discrimination, employers began bending over backwards to hire and retain women, even ones who were slightly less qualified than men. And the laws forbade employers to charge higher health insurance or life insurance premiums to female employees, even though they tend to use more medical services than men, and live longer, and so cost more to health and life insurers.

But all this leaves unexplained why women would be graduating at higher rates from colleges and from graduate and professional schools than men. One possibility is differences between men and women in variance in IQ—the issue that got Larry Summers into trouble when he was president of Harvard. Suppose as he conjectured (with some evidence) that men and women have the same IQ but the distribution of male IQs is flatter than that of women—a higher proportion of men than of women have very high and very low IQs. As graduation from most college and most graduate or professional programs requires a normal or high but not very high IQ, the greater male variance in IQ would tend to truncate male but not female graduation (and hence enrollment) rates: low-IQ males would be underrepresented in higher education, but high-IQ males would be overrepresented in just a few programs, such as high-energy physics, and so would not balance the males who were not admitted or dropped out at high rates. Males would continue to be overrepresented in jobs involving upper-body strength, but these tend not to require a high level of education.

Cultural factors may also be at work, especially in the black community, where academic performance is disparaged among young men but not young women. For example, 42 percent of black women who graduate from high school go on to college, compared to only 37 percent of black males; and just 35 percent of black male college students graduate within six years, compared to 45 percent of black female college students. This implies that 19 percent of black women who graduate from high school are graduating from college within six years compared to only 13 percent of black males. The overall situation is actually worse, because only 48 percent of black males graduate from high school, compared to 59 percent of black females (implying that the college graduation rate for black females is almost twice that for black males); and the disparity is almost as great for Hispanics. Blacks and Hispanics constitute a sizable fraction of the U.S. population. Nevertheless, there is a gap between white male and white female graduation rates as well; the high school graduation rate for white males, for example, is 74 percent, compared to 79 percent for white females, and the college graduation rate is 43 percent for white males compared to 57 percent for white females.