income ratios over time and the changing representation of women in various professions.

Comparability

In order to determine the existence or magnitude of sex discrimination by employers, as distinguished from other sex discrimination which may have occurred in education, or domestic differences between the sexes which affect job choices, we must compare women and men who have similar education, skills, job experience and other relevant characteristics.

Many statistics on male-female income differences do not do this but simply make gross comparisons of women and men in general as groups, either without regard to comparability or with very limited attempts to hold the differing variables constant. Thus a study in Britain found that women as a group earned 17 percent less per hour than men when both worked fulltime. However, this same study found that the pay differential was not due to women and men being paid different wages or salaries for doing the same job, but that women took lower-paying jobs more often than men, especially after returning to the labor force after having children. As young beginning workers, British women's incomes were 91 percent of that of British men but, as mothers, their incomes were just 67 percent of that of men who were fathers. Mothers' incomes declined as a percentage of male incomes more or less steadily until about a dozen years after giving birth, when it began to rise again, though never getting back to where it was before a child was born, perhaps indicating the permanent income loss due to interrupted careers.3 7

In the United States, a study of graduates of the University of Michigan Law School found a similar pattern:

The gap in pay between women and men was relatively small at the outset of their careers, but 15 years later, women graduates earned only 60 percent as much as men. Some of this difference reflected choices which workers had made, including the propensity of women lawyers to work shorter hours.3 8

76 Economie Facts and Fallacies

Attempting to control simultaneously for part-time versus full-time employment and for the effect of children and domestic responsibilities, another study found "that the gender pay gap is 5 percent for part-time workers age 21-35 without children, under 3 percent for full-time workers age 21-35 without children, and that there is nopay gap for full-time workers age 21-35 living alone."3 9 All these gaps represent the upper limit of the effect of employer discrimination plus whatever other factors might favor men, such as differences in education and in occupations involving physical strength or dangers. That the income gaps between women and men are so small without even taking these other factors into account suggests that employer discrimination by itself has far less influence on the gross income gaps between the sexes than gross statistics might suggest.

While comparing truly comparable individuals is something that must also be done when trying to determine the existence or effect of discrimination against ethnic or racial groups, achieving similar comparability between women and men is more challenging. That is

more than blacks with less education, just as among whites, even if not to the same extent— the same factors can have opposite effects on women and men: As we have already noted, marriage and parenthood tend to lead to increased incomes for men and reduced incomes for women.

Nor are single women and single men comparable, when "single" people include people who have been married for years, and have then been divorced. That is because the long-term negative economic effects of marriage on women, such as the interruption or even cessation of full-time employment, do not disappear with a change in marital status and re­ entering the labor force with less experience than a man of the same age. By the same token, the beneficial economic effects of marriage for men do not disappear completely after a divorce, since past seniority and increased job skills continue to make the man a higher earner than a man who was never married or a woman who was never married.

To get comparability among people of both sexes who have not had their incomes affected by marriage means comparing women and men who were

"never married" rather than simply "single." For purposes of separating out the effects of marriage, and the asymmetrical domestic arrangements which marriage often creates, from the effects of employer discrimination, the most comparable women and men are those who were never married. If male-female income gaps remain large among women and men who were never married, then obviously the economic effects of marriage do not explain sex differences in income. But if that differential changes substantially according to marital status and parenthood, then the employer is correspondingly less of a factor in income differences between women and men.

What are the facts?

As already noted, comparing never-married women and men who are past the child-bearing years and who both work full-time in the twenty-first century shows women of this description earning more than men of the same description. As far back as 1969, academic women who had never married earned more than academic men who had never married, while married academic women without children earned less and married academic women with children earned still less.4 0 For women in general— that is, not just academic women— those single women who had worked continuously since high school were in 1971 earning slighdy more than men of the same description.4 1 All this was before affirmative action was defined as "under-representation" in a 1971 Executive Order which went into effect in 1972, and so represents what was happening under competitive labor market pressures before any major government intervention to advance women.

Both trends over time and studies of women at given times show the same pattern of a negative correlation between marital responsibilities (including children) and women's educational levels and career advancement. Whether in the earlier or the later era, women who were married and had children lagged furthest behind men in income, career advancement, or even working at all. A study published in 1956 showed that most women with Ph.D.s from Radcliffe were not working full-time and those who were averaged fewer children than those who worked parttime, intermittendy, or not at all.4 2

in the general population remained relatively unchanged at about 60 percent during the decades of the 1960s and 1970s, and then began to rise significandy from the early 1980s, reaching 70 percent in 1990 and 77 percent in 2004.4 3 Finer breakdowns of the data to compare women and men of comparable ages, education, and employment histories showed the sexes much closer in earnings. Merely comparing full-time, year-around workers showed women's pay to be 81 percent of men's pay in 2005.4 4 Parttime workers not only earn less total pay, they are also paid less per hour and are less likely to be promoted.4 5 There have been, and continue to be, more women than men who are part-time workers.4 6

Very substantial income differences between women and men in a particular field can co-exist with little or no income differences between women and men who are comparable within that field. For example, a study published in The New England Journal ofMedicine found:

In 1990, young male physicians earned 41 percent more per year than young female physicians. . . . However, after adjusting for differences in specialty, practice setting, and other characteristics, no earnings difference was evident.47

The young male physicians in this study worked over 500 hours a year more than the young female physicians.4 8

In general, it makes a difference whether the male-female income gap is compiled on an annual basis, a monthly basis or as per-hour earnings. Since women tend to work fewer hours than men, the largest gap tends to be in annual income and the smallest in per hour pay. For example, the U.S. Department of Labor has reported weekly male-female earnings differences, showing women's earnings to be 76.5 percent of that of men in 1999, for example, but women's hourly earnings were 83.8 percent of that of men that same year. Comparing women and men who were comparable in occupation, industry and other variables, the per-hour difference shrank to 6.2 cents.4 9

While there are factors tending to reduce the male-female income gap over time, there are other factors tending to widen it. For example, as job experience becomes more important in the economy and more highly rewarded, this tends to widen the gap between men's incomes and women's incomes, since women of a given age tend to have less job experience than men of the same age. Shifts in demand from one industry to another and from one occupation to another also affect male-female income gaps, since women and men continue to form different proportions of the workforce in different industries and occupations.5 0 Moreover, these proportions have changed over time: Nearly half of female college graduates in 1960 became teachers, while less than 10 percent did so by 1990.5 1 The net effect of conflicting trends on male-female earnings differentials makes explanations of their income differences far from easy.

Although the male-female income gap has generally been declining over time since the 1960s, within a given lifetime that gap tends to widen. That is, young women tend to have earnings closer to the earnings of young men than older women's earnings compared to older men. What happens in between is that women's labor force participation rates are affected by domestic, and especially child-rearing, responsibilities. A study in the

American Economic Review showed:

In March 2001, at ages 25-44, the prime period for career development, 34 percent of women with children under the age of six were out of the labor force, compared to 16 percent of women without children. Thirty percent of employed mothers worked part-time, compared to 11 percent of women with no children. Among men, however, the presence of children is associated with an increase in work involvement. Only 4 percent of men with children under the age of six are out of the labor force, and among employed fathers only 2 percent work part-time.52

In short, having children has major effects on labor force participation rates— and opposite effects on women and men. Although younger women are of course more likely to have children, it is older women whose job qualifications have been most affected by the accumulated differences in job experience from that of men of similar ages. This difference is reflected in the wider male-female earnings gap in the older years. All this is

particularly relevant to the notion of a "glass ceiling" restricting how high women are allowed to rise, especially in top management positions. These are positions usually reached after many years of experience. Empirically, the gaps between women and men are huge in both representation among highlevel executives and in incomes at the executive levels, but shrink dramatically when comparing women and men of comparable experience, including continuous experience with a given company.

For example, only about 2.4 percent of top-level management positions

executives was at least 45%." Part of the reason for the compensation gap was that women were more likely to be executives in smaller corporations, whose executives tend to be paid less than executives in the largest corporations— and part of the reason for women being executives in smaller corporations is that this reflects their lesser experience. Taking these and other differences into account shrinks the male-female compensation gap considerably:

Women in the sample were much younger, and had much less seniority in their company, than men. Part of the effect of age and seniority on the gender gap seems to be reflected in the size of companies women managed. All in all, we find that the unexplained gender compensation gap for top executives was less than 5% after one accounts for all observable differences between men and women.53

Despite the complexities revealed by a closer examination of statistical data, lawsuits continue to be filed, claiming discrimination, based on purely numerical differences in the economic situations of women and men. As the New York Times reported in 2007:

In the lawsuit, the lead plaintiff— a former assistant store manager who was upset about not being made a store manager— asserts that Costco discriminated against women in promotions because 13 percent of the company's store managers were women, while nearly half of its employees were women.5 4

This lawsuit was by no means unique. A similar claim was made against Wal-Mart in 2004, likewise based on statistical disparities. Back in 1973, the Equal Employment Opportunity Commission filed a sex discrimination