Europe’s published gender pay gap understates what employers will find inside their own workforces
17 September 2026
From June 2027, every employer in the EU with 250 or more workers has to publish its gender pay gap every year, and to justify, fix or jointly assess any gap of 5% or more between women and men doing work of equal value. The number most people will benchmark against is the national gap Eurostat publishes. It is the wrong benchmark. Inside the same sector, the gap is wider in 22 of 27 Member States, and it is wider still in the large employers that report first.
Explore every country in the interactive dashboard · read the working paper (PDF)
The unadjusted gap for the EU-27 fell from 15.8% in 2010 to 11.1% in 2024. It narrowed in 24 of 27 countries and widened in Croatia and Slovenia. In 2024 it ranged from -0.8% in Luxembourg to 18.8% in Estonia.
That is the figure behind most headlines, and it is computed well: mean gross hourly earnings of men and women in enterprises with ten or more employees, across the whole economy except public administration and defence (Eurostat 2026). What it cannot show is where the gap sits.
A national gap mixes two things: where women and men work, and how they are paid once they are there. Splitting the 2022 gap across the 17 economic sections of the Structure of Earnings Survey separates the two. The split is symmetric, so the two parts add up to the published gap exactly and do not depend on taking men’s or women’s pay as the yardstick.
In 22 countries the “where” part is negative: women are concentrated in sectors that pay more per hour than the average. In Italy, 22% of women in the survey work in education against 5% of men, and 18% in health and social work against 6%. That lowers the Italian headline by 10.6 points, and leaves a gap of 14.4% between women and men working in the same sector.
The countries where the headline hides the most are Luxembourg, Italy, Cyprus, Portugal, Romania. Part of the explanation is well documented: where fewer women work, those who do are often the better paid, which keeps the average gap low (Olivetti and Petrongolo 2008). For an employer, the reason matters less than the consequence. The Directive looks inside each employer and compares women and men doing the same work or work of equal value. That comparison is far closer to the within-sector view than to the national average, and there the gap is wider than the headline suggests.
The result does not hinge on the method. Taking men’s pay as the reference, the within-sector gap exceeds the published one in 22 countries; taking women’s, in 19.
Sectors are broad. Where Eurostat publishes complete figures by occupation (6 countries) or by sector and occupation together (10 countries), the gap between women and men in the same cell stays between 8.1% and 16.8%:
| Country | Published gap | Within occupation | Within sector and occupation |
|---|---|---|---|
| Austria | 18.7% | 15.6% | 14.5% |
| Bulgaria | 12.6% | 13.7% | |
| Czechia | 18.4% | 15.9% | |
| Denmark | 13.8% | 11.6% | 8.1% |
| Finland | 16.0% | 9.5% | |
| Germany | 17.7% | 16.4% | 13.2% |
| Greece | 13.5% | 14.5% | 11.9% |
| Hungary | 17.5% | 16.8% | |
| Italy | 3.8% | 14.1% | |
| Romania | 1.5% | 13.1% | 12.0% |
| Slovenia | 6.8% | 14.2% |
Where the published gap is small, finer cells confirm a wide gap between women and men doing similar work. Where it is large, as in Austria and Germany, part of it reflects occupations, and the gap inside cells sits a few points below the headline without disappearing. Elsewhere the cells are suppressed for confidentiality, and the analysis does not guess.
Across the EU, 51% of employees in enterprises with ten or more staff work in one of 250 or more, from 29% in Estonia to 64% in Finland. Those employers report every year from June 2027. Employers of 150 to 249 report from 2027 every three years, and those of 100 to 149 from 2031; the survey groups all enterprises of 50 to 249 together, so the share of workers they cover cannot be separated.
The first wave is also where the gap is widest. In Germany, Spain, Finland, France, Poland, Romania, the only countries with complete figures by size, the gap in enterprises of 250 or more is larger than in smaller ones: 19.9% against 16.1% in Germany, 14.0% against 0.8% in Poland.
Put the two measures together. Germany and Netherlands combine a within-sector gap above the EU level with an above-average share of employees in annual reporters: the first reports will show wide gaps, and many of them. In Estonia, Latvia, Cyprus, Czechia, Hungary, Slovakia, Austria, Lithuania, the gap inside sectors is also above the EU level, but more workers sit in mid-sized firms. There, the three-yearly reports of employers with 150 to 249 workers, and the right of every worker to ask for pay information (Article 7), will carry more of the weight.
The private sector carries the wider gap. In 20 of the 21 countries that publish both, the gap is smaller in enterprises under public control; Slovenia is the exception. The Directive applies to public employers too, but the pressure will be greater in private ones.
The gap opens with age. In the median country it is 4.4% for workers under 25 and peaks at 13.8% at 45-54. Gaps build through careers, not starting salaries, which is why the Directive’s requirement to make pay progression criteria accessible (Article 6) matters as much as the reporting itself.
Data. Eurostat’s unadjusted gender pay gap series (2010-2024) and the Structure of Earnings Survey 2022, downloaded on 2026-09-17 through the Eurostat API and kept as snapshots in the repository. Scope as in the official gap: enterprises with ten or more employees, NACE sections B to S excluding O.
Decomposition. For each country, the gap in mean hourly earnings is split into a between-cell and a within-cell part with symmetric weights (Oaxaca 1973; Blinder 1973; Reimers 1983): employment shares and wages of women and men are averaged, so the parts sum to the gap exactly. A country is used only if the cells cover at least 95% of each sex’s employees and rebuild the published 2022 gap within 0.5 points. All 27 countries pass with sector cells (the largest difference is 0.42 points). The EU figure is the employee-weighted mean of the national ones, which is how Eurostat compiles it; pooling EU earnings instead would give 13.2%.
Limits. Cells are much coarser than an employer’s categories of workers, so the within part is a guide to what employers will find, not an estimate of discrimination (Blau and Kahn 2017). The 2022 survey predates the Directive, excludes public administration and enterprises with fewer than ten employees, and measures enterprises, not legal employers.
The full method, robustness checks and country tables are in the working paper. Data, code and an interactive dashboard: github.com/D0M3N1C0X/where-pay-transparency-bites.