Is the Gender Wage Gap Real?

The most-cited figure (“women earn 77 cents on the dollar”) comes from comparing the median annual earnings of all men and all women who work full time, year round. The comparison takes no account of occupation, industry, experience, or how many hours beyond full time they work. This is a real gap, but it measures aggregate earnings differences, not pay discrimination within identical jobs.

The “77 cents” figure was the Census Bureau’s ratio for full-time, year-round workers around 2010 to 2012. It left part-time workers out. The same measure stood at 83.9 cents in 2025, up from a revised 80.6 in 2024. The Census Bureau calls it the first significant rise in the ratio since 2016, after two years of significant falls. The Bureau of Labor Statistics (BLS) weekly measure for full-time workers was 82.1 cents in 2025.

Each of the four measures above correctly describes aggregate earnings, but all are misleading if used to describe pay discrimination within comparable jobs. The raw gap combines several factors: differences in occupation and industry, average hours worked, years of work experience, and employer and sector. Some of these differences, such as occupation and industry, may reflect discrimination, but conflating them with “same job, different pay” misrepresents what the data shows.

The appropriate question is how much of the gap remains after controlling for these factors: the adjusted gap.

Download this conclusion: four measures of one gap

This section as one image: where the 77 cents came from, and the four current figures with what each one counts. They are 85 cents for the hourly pay of all workers in 2024, 83.9 for the annual pay of full-time, year-round workers in 2025, 82.1 for the weekly pay of full-time workers in 2025, and 77.5 for the annual pay of everyone with earnings in 2025.

Download the card (PNG, 1,080 × 1,350)

Blau and Kahn, two Cornell economists, found in 2010 data that taking account of measurable differences between workers shrank the gap but did not close it. Economists decompose the wage gap into the portion explained by those differences (such as education, experience, occupation and industry) and an unexplained residual. The unexplained portion is often called the “adjusted gap.” It captures a mix of potential discrimination and unmeasured factors.

The unexplained residual gap does not equal discrimination. Blau and Kahn call it suggestive but not conclusive. It overstates discrimination if men have more of the skills the data cannot see, or worse working conditions. It understates discrimination if experience, occupation or industry were themselves shaped by it. Separating these requires other methods, such as experiments with fictitious job applicants (Experiments on Discrimination).

In Blau and Kahn’s study, differences in occupation and industry accounted for 51% of the gap in 2010, more than any other measured factor; children were not among them. Economists call this occupational sorting: women and men working in different occupations and industries. Women are overrepresented in education, healthcare support, administrative services, and social work; men are overrepresented in engineering, construction, securities sales and financial advising, and technology.

Sorting is not the whole of it. In every occupation in the table where the survey publishes both figures, women’s median pay is below men’s, from 79.8% of it among registered nurses to 94.5% among software developers.

Levanon, England and Allison studied US census data from 1950 to 2000 and found a “devaluation” pattern. When the share of women in an occupation rose, the occupation’s pay a decade later fell relative to others, even with controls for education and experience. The authors read this as employers placing a lower value on work done by women. They could not rule out, though, that employers first cut the skill content of jobs as women entered.

So “choice of occupation” and “labor market discrimination” are not cleanly separable: what an occupation pays partly depends on who does it. Earlier studies, Paula England’s among them, found that occupations with a higher share of women generally pay less, even with controls for education and skill demands, the same authors note.

Children account for much of the gap. The motherhood penalty, which economists measure as a “child penalty”, is how far women’s earnings fall behind men’s after their first child is born.

The motherhood penalty is driven by several mechanisms:

• career interruptions and reduced experience accumulation;
• reduction in hours worked;
• switching to lower-paying but more flexible employers or roles;
• possible employer discrimination against mothers (the “maternal wall”).

The relative contributions of these mechanisms vary by country. In Denmark the three parts of the penalty are roughly equal (table above). In the United States and the United Kingdom, Kleven and colleagues find that fewer mothers in work is a key driver.

Claudia Goldin won the 2023 Nobel Prize in economics for her work on women in the labor market. She argues that the gap “would be considerably reduced and might vanish altogether if firms did not have an incentive to disproportionately reward individuals who labored long hours and worked particular hours.”

In business and law, she finds, long hours bring a more than proportional rise in pay, and for business-school graduates any time out is heavily penalized. Children are the main reason women’s hours change: among Chicago business-school graduates, mothers worked 24 percent fewer hours a week than men.

Audit studies find some direct employer discrimination against mothers. In one, equally qualified résumés that signal parenthood drew fewer callbacks for women but not for men (Correll et al., 2007, American Journal of Sociology). This “maternal wall” effect is separate from the hours/flexibility channel.

The most direct evidence on gender discrimination comes from experiments. In audit studies, researchers send employers matched applications that differ only in the applicant’s sex; natural experiments use real changes, such as blind auditions. Because audit-study applicants are equally qualified by design, a difference in how employers treat them cannot come from a difference in qualifications.

Audit studies find discrimination is real, and it varies by setting. A 2014 survey of sixty-seven experiments conducted since 2000, four on gender, found it against women in some jobs and against men applying to female-dominated ones. A 2025 study that pooled 37 American audit studies, covering 243,202 fictitious job applications, found no statistically significant gender discrimination in hiring overall. In it, the direction of bias depended on whether an occupation was mostly male or mostly female.

The audit studies measure callbacks and job offers, not the pay people go on to earn. Blau and Kahn note that experimental findings generally cannot be easily translated into a share of the pay gap.

The gap varies widely between countries of the Organisation for Economic Co-operation and Development (OECD). In the latest figures it runs from 27.8% in South Korea to below zero in Luxembourg, where women’s median earnings were higher than men’s.

Child penalties, the largest single driver, differ sharply between countries (Motherhood Penalty table): smallest in Denmark and Sweden, largest in Germany and Austria. The researchers who measured them discuss two explanations, family policy and gender norms.

On family policy, they cite evidence from Austria that parental leave and childcare had little or no long-run effect on child penalties. Sweden, with longer and more generous parental leave than Denmark, still has a 7.8% gap.

On gender norms, penalties in the six countries the researchers compared are larger where more people think mothers of young or school-age children should stay at home. The authors say, though, that conclusive evidence on norms is hard to get.

The child penalty appears in all six countries Kleven and colleagues compared, measured in population-wide records in Denmark, Sweden and Austria and in surveys elsewhere. In Denmark, ten years after the first birth, women’s earnings were 19.4% behind men’s.

The sociologist Michelle Budig finds that lost experience explains about a third of the motherhood wage penalty. She also finds that the argument that women trade earnings for family-friendly jobs “is not well supported by the statistical analyses.” Goldin locates much of the gap in how firms reward long and particular hours (Motherhood Penalty).

Occupational devaluation is documented (Occupation Sorting). So the “choice of occupation” argument does not fully explain the gap.

Experiments find employers treating equally qualified applicants differently: science faculty rating a female student less hireable, high-price restaurants turning women away, employers calling mothers back about half as often as childless women. Blau and Kahn conclude that discrimination “cannot be discounted”, so the 8.4% gap left unexplained in 2010 may include some.

About 62% of the 2010 gap is explained by measurable differences between workers (Adjusted Gap). Women aged 25 to 34 earned 95 cents for each dollar men their age earned in 2024, on the Pew Research Center’s hourly measure. Among never-married full-time workers, women earned 91.7% of what men earned in 2024, by the Bureau of Labor Statistics’ weekly figures. The gap opens with childbearing: women’s and men’s earnings move in parallel until the first child and diverge after it.

Goldin offers an account of the unexplained gap that does not rest on discrimination. Firms pay more to people who will work long and particular hours, and less to those who want flexibility, a trade economists call a compensating differential. Her remedy is to change “how jobs are structured and remunerated to enhance temporal flexibility,” which, she writes, “does not (necessarily) have to involve government intervention.”

Men take more dangerous jobs: 91.9% of US workers killed on the job in 2024 were men. Men working full time also work longer days, 8.2 hours on the days they worked in 2025 against 7.9 for women working full time.

Blau and Kahn note that jobs with more variable earnings are expected to pay extra for the risk. Women’s greater average aversion to risk, they add, could lower their pay if it steers them away from such jobs. A comparison of medians captures none of these differences in danger, hours and risk. Blau and Kahn, however, judge traits such as risk aversion a small to moderate part of the gap.

The evidence on both sides agrees on several points:

• the raw gap is real;
• occupation and industry account for about half of it;
• about 8% remained after the fullest controls in 2010;
• children are the largest single driver;
• discrimination appears in some settings, not all.

The dispute is over interpretation: whether the occupational structure reflects discrimination (the devaluation view) or preferences meeting a neutral market (the compensating-differentials view). Each has evidence behind it.

Yes, the gender wage gap is real, but most of it is not a gap in pay for the same job. In Blau and Kahn’s study of 2010 data, measured differences between workers, chiefly occupation and industry, accounted for about 62% of the gap. With them controlled, women earned 8.4% less than men.

Children account for about half of parents’ earnings gap in Kleven’s estimate, and for more than two-thirds of the overall US earnings gap in Cortés and Pan’s. The unexplained part is some mix of discrimination, which experiments find in some settings, and differences the data cannot see.

Download the conclusion

The verdict above as one image. It gives the full-time pay ratios for 2025: 82.1 cents on the weekly measure and 83.9 on the annual one. Its bars show Blau and Kahn’s 2010 gap: 20.7% before any adjustment and 17.9% with education, experience, race and location. With occupation, industry and union status added it was 8.4%. The card is 1,080 by 1,350 pixels, and anyone may download it and post it.

Download the card (PNG, 1,080 × 1,350)