Adjusted vs. Unadjusted Pay Gap: Why Article 9 Requires Both
Most organisations preparing for EUPTD Article 9 reporting focus on the adjusted pay gap. The adjusted gap is easier to understand, easier to defend, and produces smaller numbers. It is also only half of what Article 9 requires. Organisations that report only the adjusted gap are not compliant and are also missing the more important diagnostic signal.
The two measures defined
The unadjusted gender pay gap compares average pay across all men and all women in the organisation, without controlling for any other factor. It answers a single question: what is the average difference in what men and women earn here?
The adjusted gender pay gap controls for job group, seniority, working time, and other legitimate pay-setting factors. It compares pay between men and women who are in the same job group, at the same level, working the same hours. It answers a different question: conditional on all observable factors being equal, is there still a pay difference by gender?
Why both are required under Article 9
Article 9(1)(b) of EUPTD 2023/970 requires employers to report the gender pay gap and the gender pay gap in complementary or variable components of pay. The Recitals and the accompanying Commission guidance make clear that this means both the unadjusted gap (referred to as the "raw" gap) and the adjusted gap within job categories.
The reason both are required is that they answer different questions and expose different problems. An organisation can have a near-zero adjusted gap and a large unadjusted gap simultaneously. This is the most common pattern in organisations with gender-segregated job hierarchies: women and men in the same roles are paid equally, but women are clustered in lower-paying roles. The adjusted gap misses this entirely. The unadjusted gap makes it visible.
A zero adjusted gap does not mean pay equity. It means you do not pay men and women differently for the same work. It says nothing about whether the same work is available to both on equal terms.
The four outcome combinations and what they mean
| Adjusted gap | Unadjusted gap | Diagnosis |
|---|---|---|
| Low | Low | Good position. Equal pay for equal work, and relatively balanced representation across levels. |
| Low | High | Most common pattern. Equal pay within roles, but women concentrated in lower-paying roles or grades. Root cause is in recruitment, promotion, or job architecture, not direct pay discrimination. |
| High | Low | Rare. Pay differences within comparable roles that are masked at the aggregate level. Usually indicates a small number of high-value male outliers or classification errors. |
| High | High | Direct pay discrimination compounded by structural segregation. Highest compliance risk and highest enforcement exposure. |
The most common Article 9 preparation mistake
Organisations run the adjusted gap regression, find a number below 5%, declare themselves compliant, and stop. Three problems with this approach.
First, Article 9 explicitly requires both measures. A report containing only the adjusted gap is non-compliant regardless of what the number shows.
Second, the adjusted gap is only as good as the job grouping it controls for. If your job groups do not satisfy Article 4 (gender-neutral classification criteria), your adjusted gap is controlling for a potentially discriminatory variable, which inflates the apparent fairness of the result.
Third, the unadjusted gap is what the public reporting requirement makes visible. Article 9(2) requires employers with 250 or more workers to publish pay gap information. What gets published is the unadjusted gap. A low adjusted gap combined with a high unadjusted gap will generate questions from employees, works councils, and journalists that a compliance certificate cannot answer.
What the regression actually requires
Running an adjusted pay gap analysis correctly requires a dataset that includes: individual-level pay data (base salary, fixed allowances, variable pay separately); job group assignment (Article 4 compliant); grade or seniority level; working hours or FTE; and any other factor your organisation uses as a legitimate pay-setting input (tenure, geographic location, shift premium).
The regression model regresses log hourly pay on a gender dummy variable and all legitimate controls. The coefficient on the gender variable is the adjusted gap. If that coefficient is statistically significant after controlling for all legitimate factors, you have unexplained pay differences that require investigation under Article 10.
Run your pay equity baseline
We conduct both adjusted and unadjusted pay gap analysis using your actual payroll data, mapped to EUPTD Article 9 reporting requirements. Output is an evidence package you can take to your works council and your legal team.
Book the pay equity baseline