The latest US Department of Labor data show that women working full-time make 81 percent of full-time men's wages. But this figure is both inaccurate and misleading. This statistic looks only at raw averages and does not take into account factors such as education, skills, and hours worked. After controlling for other factors, the gender pay gap practically disappears. Legislation to close the gender "wage gap" is misguided: in reality, there is no gap to close. — Diana Furchgott-Roth, "Sorry, Elizabeth Warren, Women Already Have Equal Pay," Economics21, The Manhattan Institute for Policy Research, July 27, 2016

"We believe in equal pay for equal work." That was all Sen. Elizabeth Warren (D-Mass.) said about the gender pay gap during her keynote address to July's Democratic National Convention. But it was enough to provoke a response from economist Diana Furchgott-Roth, a senior fellow at the free-market Manhattan Institute.

That's hardly surprising. Furchgott-Roth has spent two decades issuing one version or another of one basic claim: "there is no gap to close between men's and women's wages."

Publishing article after article claiming that there is no gender pay gap, however, doesn't make it so. Here's why.

No Statistical Artifact

To begin with, the gender pay gap is no statistical artifact. The most common measure of the gender pay gap compares the median earnings (wages and salaries) of full-time working women over the year to the median earnings for men. That ratio does not compare the earnings of men and women doing the same job, but rather the earnings of all men and women who work full time.

In 2014, the latest year for which data are available, men's median earnings for the year were $50,383, while women's median earnings were $39,621, or 78.6% of men's. That's where the figure that women earn 79 cents for each dollar a man earns comes from. The National Committee on Pay Equity inaugurated the tradition of using this ratio to determine the date on which "Equal Pay Day" falls each year. This year, it fell on April 12, 2016, the date by which women would have earned enough to make up the $10,762 gap between their pay and men's in 2015. (Furchgott-Roth's figure for the gender pay gap, 81% in 2015, is calculated in the same way but compares the median weekly earnings of full-time wage and salary workers.)

Whether women earn 79 cents or 81 cents for every dollar a man earns, the gender pay gap is long standing. In 1963, the year the Equal Pay Act became law, a full-time working woman (earning the median pay for women) got 59 cents for each dollar a full-time working man received (at the median pay for men). By the first Equal Pay Day in 1996, women earned 74 cents for a dollar of men's earnings; now the figure is up to 79 cents. The gender pay gap, however, is no longer narrowing as fast it did earlier. During the 1980s, the gap declined by more than one-quarter (28.7%), as women's earnings improved from 60 cents for every dollar a man earned to 72 cents; during the 1990s, by just 6%, as women's earnings increase from 72 cents to 73 cents for every dollar of men's earnings; in the last ten years (2004–2014), by 7.4%, as women's earnings increased from 77 cents to 79 cents for every dollar of men's earnings.

The gender pay gap is also pervasive. Regardless of her education, her occupation, her race, or her age, a full-time working woman (getting the median wage for women of that group) is paid less than a full-time working man (getting the median wage for men of that group).

•  Women earn less than men at every educational level. In 2015, the median weekly earnings of women without a high school diploma were 80% of their male counterparts' earnings, 77% for women with (only) a high school diploma, 75% for women with some college, 75% for women who were college graduates, and 74% for women with an advanced degree.

•  Women earn less than men in all but five of the 800 detailed occupations tracked by the Bureau of Labor Statistics (for which there is comparable data). Women in female-dominated occupations — from maids to secretaries to registered nurses — earn less than men do in those same jobs, as do women in male-dominated jobs — from truck drivers to retail supervisors to janitors. The same is true for women in elite jobs such as physicians, surgeons, and financial managers.

•  Women of all racial/ethnic groups are paid less than white men and less than men of the same race/ethnicity. In 2015, the median weekly earnings of white women working full time were 80.8% of those for white men. The weekly earnings of black women were 89% of the earnings of black men; the earnings of Hispanic women, 90% of the earnings of Hispanic men. Meanwhile, the weekly earnings of black and Hispanic women were just 62% and 67%, respectively, of the weekly earnings of white men.

•  Women workers of all ages are paid less than their male counterparts. Older women, however, face the largest pay gap as they are penalized for leaving the workforce more often than men for childbirth, childcare, and eldercare. In 2014, the annual median wage of women ages 18-24 who worked full time was 88% of the median wage of full-time male workers of the same age group, but 81% for women ages 35-44, and just 68% for women over 55.

Making the Unequal Look Equal

But those differences, no matter how widespread or long lasting, don't impress economist Furchgott-Roth. In her version of reality, those differences disappear once the pay gap is adjusted for gender differences in hours worked, education, experience, and choice of industry and occupation. But each of these adjustments is problematic or makes less of a difference than Furchgott-Roth and other pay gap deniers suggest.

The deniers complain that earnings differentials calculated for full-time workers, including anyone who works 35 or more hours a week, mask the fact that men work more hours (in the money economy) than women. In fact, men are almost twice as likely as women work more than 40 hours a week. But that problem can be corrected by using hourly earnings to measure the gender pay gap. In 2014, hourly earnings of full- and part-time women wage-and-salary workers were 84.6% of men's. While smaller, that gap is still quite substantial and persists at all levels of education and for all racial/ethnic groups.

Nor will making adjustments for gender differences in education and experience, two traditional measures of labor-market skills, make the gender pay gap disappear. Adjustments for education explain much less of today's gender pay gap than they did in the early 1980s. Since then, more women have graduated from college than men, and by 1999 the average full-time working woman had more years of education than her male counterpart. Gender differences in years of experience are also far smaller than they were in the past. In 1981, men had, on average, 6.8 more years of full-time labor market experience than women, but the experience gap was just 1.4 years in 2011. In their detailed study of the sources of the gender pay gap, economists Francine Blau and Lawrence Kahn estimate that, taken together, differences in education (which favor women) and differences in experience (which favor men) explained 8.2% of the gender pay gap in 2011, or just 2 cents of the 23 cent gap.

There is little disagreement that differences between women and men in terms of the industries they work in and the jobs they hold have a profound effect on the gender pay gap. Blau and Kahn, for instance, estimate that industry and occupation accounted for fully one-half (49.5%) of the gender pay gap in 2010.

But just how women ended up in particular industries and occupations and not in others is a matter of sharp debate. For gender pay gap skeptics, this is a matter of individual choice. "Women gravitate toward jobs with fewer risks, more comfortable conditions, regular hours, more personal fulfillment, and greater flexibility," argues Carrie Lukas, executive director of the Independent Women's Forum. Women, she concludes, are "willing to trade higher pay for other desirable job characteristics." But the story Lukas tells is not the empirical reality faced by most women. To begin with, women's jobs do not possess the other desirable characteristics she says compensate women for accepting lower pay. In their study of the characteristics of men's and women's jobs in 27 countries including the United States, sociologists Haya Stier and Meir Yaish found that on average the jobs held by women offered less autonomy or time flexibility and that their working conditions were more stressful and exhausting than those of men, a condition that was surely exacerbated by women bearing an inordinate share of domestic labor. (Women's jobs did require less physical labor than men's jobs.)

If individual choices of women don't explain what crowds many women into lower paying jobs, then what is responsible for gender segregation by occupation and industry? Gender discrimination that disadvantages women in the labor market and devalues their work is the more plausible answer. If you doubt that women's work is undervalued, political scientist Ellen Frankel Paul would ask you to consider this example: zookeepers — a traditionally male job — earn more than workers caring for children — a traditionally female job. The evidence that the sorting of genders into industries and occupations is shaped "by discrimination, societal norms and other forces beyond women's control," as economists Jessica Schieder and Elisa Gould argue, is compelling. For instance, it is well documented that women in better-paying male-dominated jobs have faced hostile work environments. A 2008 study found that "52% of highly qualified females working for SET (science, technology, and engineering) companies quit their jobs, driven out by hostile work environments and extreme job pressures." And gender discrimination plays a role in who gets hired in the first place. In two studies, when participants reviewed resumes that were identical except for the names, the ones with male names were more likely to be offered a job. According to another study, after five top US symphony orchestras switched to blind auditions, women were 50% more likely to get past the first round. But gender norms already direct women and men toward different jobs long before they enter in the labor market. For instance, Schieder and Gould report that women arrive at college far less likely to major in engineering, computer sciences, or physics than men, even though those fields promise lucrative job opportunities.

Most low-paying jobs, on the other hand, are female-dominated. In their 2009 study, sociologists Asaf Levanon, Paula England, and Paul Allison reported that occupations with a higher percentage of women workers generally paid less than those with a lower percentage of women, even when correcting for education and skill demands. On top of that, they found evidence that when more women enter a job category, employers start paying less. For example, as jobs in parks and camps went from being male-dominated to female-dominated, between 1950 and 2000, the median hourly wages (corrected for inflation) fell by more than half.

Finally, the adjustments favored by Furchgott-Roth and other gender-gap skeptics are not enough to statistically eliminate the gender pay gap. For instance, one research study, commissioned by the Department of Labor during the George W. Bush administration, estimated a wage gap between 4.8 and 7.1 percentage points after making adjustments for other gender differences. In the Blau and Kahn study the remaining gender gap in 2010 was 8.4 percentage points when fully adjusting for differences in education, experience, region, race, unionization, industry and occupation. Those gender pay gaps, which assume that differences in occupation and industry are not evidence of ongoing gender discrimination, are much smaller than the unadjusted gap, but still substantial. For Blau and Kahn, the unexplained portion of the gender pay gap, "suggests, though it does not prove, that labor market discrimination continues to contribute to the gender wage gap." The unexplained gender pay gap (the portion still left over after statistically adjusting for occupation, industry, or worker qualifications) has actually worsened since the late 1980s (from 7.6 cents for each dollar a man made in 1989 to 8.4 cents in 2010). In 2010, over one-third (38%) of the gender pay gap remained unexplained. If we include the portion of the gap due to gender differences in occupation and industry, a whopping 87.5%, or 18 cents of the 21 cents of the unadjusted gender gap in their study, can be interpreted as a product of continued discrimination.

Truly Equal Pay

One important step to reduce continued labor market gender discrimination would be to pass the Paycheck Fairness Act. The law would require employers to show that wage differentials are based on factors other than gender, and would strike a blow against pay secrecy by banning retaliation against employees who reveal their own wages to other employees.

But much more needs to be done to combat workplace gender discrimination. More family-friendly policies are needed. The United States is the only advanced country that does not guarantee paid maternity leave. Comparable-worth policies are needed to promote pay equity. Those policies would ensure that jobs having the same value to employers would be paid the same whether performed by women or men. Also, in order to short-circuit historical gender pay discrimination, newly passed comparable-worth legislation in Massachusetts bars employers from asking job applicants how much they earned in previous jobs. In addition, raising the minimum wage would boost the earnings of workers in low-income jobs, the vast majority of which are female-dominated. Unionization in female-dominated occupations would also reduce the gender pay gap, as it has done among public employees.

For Furchgott-Roth and the gender-pay-gap skeptics the pay gap disappears by statistical manipulation. These policies, on the other hand, are ways to make it go away for real.