The gender gap myth

Women earn less than men on average. The myth begins when that statistic is translated as “women are paid less for the same work.”

Those are different quantities.

A gender pay gap usually compares average or median earnings across large groups of people. One number mixes occupations, firms, job levels, working time, career paths, responsibility, and working conditions. The statistic is real. It does not, by itself, identify why the gap exists or how much of it is discrimination.

In 2023, the median full-time working woman across the OECD earned about 11% less than the median full-time working man.¹ That is a real gap. It is also an unadjusted measure: it does not make the man and woman identical in job, firm, experience, schedule, or responsibility.

One number combines several different gaps

Eurostat makes this especially clear. Using the 2022 Structure of Earnings Survey, the EU’s average unadjusted hourly pay gap was 13.2%. Women were also paid for 11% fewer hours per month on average, while the employment-rate gap was 13.1%. The resulting overall earnings gap — a measure that incorporates hourly pay, paid hours, and employment — was 32.8%.²

These percentages are not meant to be added together. The point is simpler: “women earn less” is not one economic mechanism.

Part of the difference comes from whether a person is employed at all. Part comes from paid hours. Part comes from occupation, employer, and job level. Part comes from doing different work within the same firm. Only after those questions do we reach the narrower question: are men and women paid differently for work of equal value?

Eurostat explicitly warns that its standard unadjusted gender pay gap measures a concept broader than equal pay for equal work. The gap can reflect worker characteristics as well as sectoral and occupational segregation.³

So “a woman gets 87 cents for the same dollar of male work” is not a translation of the statistic. It is an interpretation added to it.

An hour is a unit of time, not a unit of work

Moving from annual earnings to hourly wages improves the comparison. It still does not produce a same-work comparison.

One employee may own a budget, a team, and the outcome of a project while another performs a bounded set of tasks. One may be available evenings and weekends while another works only a fixed window. One may work at a firm that pays a large wage premium while another works at a lower-paying firm. Even the same job title does not guarantee the same economic function.

This matters most where the market pays disproportionately for more than the number of hours — for particular hours and temporal availability. Claudia Goldin showed that in a number of high-paying occupations, long and inflexible hours carry nonlinear rewards. Her argument is that the gender pay gap could be substantially reduced, and in some sectors might nearly disappear, if firms had less reason to disproportionately reward workers who supply long hours and particular hours.⁵

Two sixty-minute hours can therefore have different market values. An employer is not buying time in the abstract. It is buying responsibility, availability, substitutability, exposure to failure, and the ability to solve a problem when the problem actually occurs.

Compare genuinely similar work and the picture changes sharply

The most useful evidence for this dispute comes from an OECD study using linked employer–employee data across 16 countries. It compares women and men with similar skills and decomposes the wage gap into differences between firms and differences within firms.⁴

About 77% of the gap between similarly skilled women and men was associated with differences in pay within firms. But “within the same firm” still did not mean “same work, different pay.” The within-firm component itself contains two very different things: differences in tasks and responsibilities, and differences in pay for work of equal value.

The detailed decomposition is the key result: on average, about eight-ninths of the wage gap between similarly skilled women and men reflected differences in the work they did — tasks and responsibilities. About one-ninth reflected differences in pay for work of equal value, a category in which the OECD groups bargaining and discrimination.⁴

That does not mean “discrimination equals one-ninth.” The residual does not cleanly separate employer discrimination from bargaining and other differences in firm wage premia. The OECD also warns that where working-time information is missing, the bargaining/discrimination component may be overstated.⁴

But the result is enough to reject the popular substitution. Most of the remaining gap even among similarly skilled workers is not a different price for identical work. It is different work.

“Same qualifications” still does not mean same work

A degree is not a job. A job title is not responsibility. Responsibility is not output.

Two engineers with similar education can sit at different levels, own systems of different scale, and carry different consequences when something fails. Two doctors can have similar basic credentials while one works a fixed shift and the other remains available overnight. Two lawyers can have similar tenure while one handles standard documents and another owns an entire transaction before the client.

Controls for education and experience are useful. They do not magically create “same work.” That is why the OECD decomposition separately identifies tasks and responsibilities.

Salary negotiation is part of the gap too

There is another inconvenient point: the price of labour is not set by the employer alone. Workers participate in setting it.

In a large natural field experiment involving nearly 2,500 job seekers, Andreas Leibbrandt and John List found that when job advertisements did not explicitly say wages were negotiable, men were more likely than women to initiate salary negotiation. When negotiability was stated explicitly, the gender difference disappeared and even tended to reverse.⁶

That is a real mechanism. If one group more often asks for more, bargains, changes employers for a raise, or selects into settings where wage rules are ambiguous, average group pay can diverge without any instruction to “pay women less.”

David Card, Ana Rute Cardoso, and Patrick Kline found another part of this mechanism in Portuguese administrative data. Women were less likely to work at firms paying high wage premiums and received about 90% of the firm-specific pay premium earned by men. Sorting and bargaining together accounted for about one-fifth of the cross-sectional gender wage gap in their data.⁷

So the adjusted residual cannot simply be renamed discrimination. It can contain negotiation, sorting across employers, firm-specific wage setting, and discrimination at the same time.

Discrimination exists. That is exactly why it should be measured separately

The opposite claim is wrong too. Saying that the raw gender gap is not a measure of discrimination does not mean discrimination never occurs.

A Danish reform requiring some firms to publish gender-disaggregated wage statistics produced a measurable effect. In the study by Morten Bennedsen and co-authors, the gender pay gap at affected firms fell by roughly 2 percentage points, or 13% relative to the pre-reform mean. A substantial part of the reduction came from slower wage growth for male employees.⁸

This is what evidence for a mechanism looks like: a policy change, a comparison group, and a measured effect. That is much stronger than taking the average earnings difference between millions of different workers and declaring all of it discrimination.

Male earnings are counted. The price paid for them usually is not

Public discussion contains a recurring asymmetry.

If a man earns more, the difference enters the gender gap. If he earns more partly because he supplies more paid hours, accepts a less flexible schedule, remains available at inconvenient times, or takes on more expensive responsibility, those conditions are easily relabelled “personal choices” and removed from the gender discussion.

Economically, they are part of the same contract. You cannot count the premium honestly while pretending the condition that earns the premium has no cost.

Eurostat measures a substantial gap in paid hours.² Goldin shows the disproportionate price of temporal availability in parts of the high-wage labour market.⁵ The OECD finds that most of the gap between similarly skilled men and women is associated with different work, tasks, and responsibilities.⁴

That is the missing half of the argument: higher male earnings can be not only an advantage men receive, but also the price attached to a labour role men are more often expected to take and are more often willing to accept.

Where the myth actually is

The myth is not that a gender pay gap exists. It exists and can be measured.

The myth is the substitution of three different claims:

• an average earnings gap is presented as underpayment for identical work;
• differences in hours, tasks, responsibility, employers, and availability disappear from the explanation;
• whatever remains after statistical controls is automatically called discrimination even when the research itself groups bargaining and discrimination together.

The useful question is not “is there a gap?” There is.

The useful question is: how much of the gap comes from men and women doing different work, how much from selling their labour differently, and how much from women being paid less for genuinely equal work because they are women?

Only the last quantity measures the claim implied by the slogan about “unequal pay for the same work.”

An earnings gap is a statistic. Underpayment for equal work is a causal claim. One cannot be substituted for the other.

Sources

1. OECD, Gender Equality in a Changing World, 2025.
2. Eurostat, Gender pay gaps in the European Union — 2025 edition, revision, Structure of Earnings Survey 2022 data.
3. Eurostat, metadata for Gender pay gap in unadjusted form, updated 27 February 2026.
4. OECD, Is it where you work, what you do, or what you get? Unpacking the gender wage gap and its evolution over the life-course, The Role of Firms in Wage Inequality, 2021.
5. Claudia Goldin, A Grand Gender Convergence: Its Last Chapter, American Economic Review, 2014.
6. Andreas Leibbrandt and John A. List, Do Women Avoid Salary Negotiations? Evidence from a Large-Scale Natural Field Experiment, Management Science, 2015.
7. David Card, Ana Rute Cardoso, and Patrick Kline, Bargaining, Sorting, and the Gender Wage Gap: Quantifying the Impact of Firms on the Relative Pay of Women, Quarterly Journal of Economics, 2016.
8. Morten Bennedsen, Elena Simintzi, Margarita Tsoutsoura, and Daniel Wolfenzon, Do Firms Respond to Gender Pay Gap Transparency?, Journal of Finance, 2022.