A gender pay analysis can provide insights into potential salary inequalities in an organisation. There are different ways to conduct a gender pay analysis. A gender pay gap reported by governments or in the media usually refers to what is known as the unadjusted gender pay gap. The unadjusted gender pay gap is the difference in average pay for women and men without accounting for the potential influence of other variables. In the Netherlands, the unadjusted gender pay gap was 13% in 20221 which means that on average women in the Netherlands earned 13% less than men or put differently, women earned on average about €0.87 per euro earned by men. An unadjusted gender pay gap of 13% made 14 November 2022 Equal Pay Day – the day when women effectively stopped getting paid for the remainder of 2022.
As figure 1 shows, in 2022, the average unadjusted gender pay gap in the EU was 12.7% with Estonia having the highest gender pay gap (21.3%) and Luxembourg having the lowest gender pay gap (-0.7%). A gender pay gap of -0.7% actually means that on average women were paid 0.7% more than men.

While the unadjusted gender pay gap provides important information, it is important to also calculate the adjusted gender pay gap which is the difference in average pay for men and women accounting for the potential influence of other variables such as age, years in service, and education level. The adjusted gender pay gap separates the impact of different employee characteristics and shows whether an organisation still has a pay gap after accounting for these characteristics.
Eurostats recommends including the following explanatory variables in an adjusted gender pay analysis: age, education level, occupation, years in service in the current organisation, employment contract (temporary or permanent), and working time (part-time or full-time)2. Besides age and years in service, also age squared and years in service squared are included because the relation between salary and age and between salary and years in services can be non-linear. This means that while salary typically increases with age and years in service, at some point salary will not increase anymore and show a plateau or even reversed U curve. Education level should be classified according to the ISCED classification3 into four groups: ISCED 0, 1 and 2, ISCED 3 and 4, ISCED 5 and 6, and ISCED 7 and 8. Occupation refers to the kind of role and responsibility of the employee and should be classified according to the ISCO-08 classification4 at a 2-digit level. It is possible to group both education level and occupation according to a different classification system; outcomes of a gender gap analysis with a different classification are equally valid, just not comparable to the analyses conducted by Eurostats or other organisations that use Eurostats’ classification system.
When conducting an adjusted gender pay analysis, it is common to encounter several problems, most importantly relating to the quality of the available data. As a result, it may not be possible to include all explanatory variables recommended by Eurostats. For instance, organisations may not record employees’ level of education or may not record it for all employees which may make it impossible to include this variable in the analysis. Additionally, it is possible that all or almost all employees have an indefinite employment contract. If that is the case, it may not be necessary to include employment contract as an explanatory variable in the analysis. Another potential issues is that it may be difficult to categorise employees according to the ISCO-08 classification or employers may not agree to this classification. In that case, employees can be grouped based on other objective criteria that take into account the role and responsibility of employees.
The dependent variable in an adjusted gender pay analysis is salary. This is the variable that we want to predict. It is important to include total compensation including base salary, but also any potential additional pay such as bonuses and shares. Alternatively, it is possible to conduct multiple regression models for base salary and other pay separately. When conducting an adjusted gender pay analysis, it is common practice to use the natural log of the salary. Transforming salary using the natural log makes sure that the salary data is normally distributed and makes it easier to interpret the results since the estimated coefficient gives the approximate percentage change in salary from a one-unit change in the explanatory variable.
The adjusted gender pay gap is calculated by using linear regression analysis to predict salary:

The estimated coefficient b1 tells us the adjusted gender pay gap when all other variables are accounted for. When using the natural log for salary, the coefficient on the male indicator gives the approximate percentage gender pay gap if all other employee characteristics are constant. A regression analysis also provides a value for the fit of the model which indicates how well the explanatory variables included in the analysis explain the variance in the dependent variable.
A fictitious example of a gender pay analysis
In a fictitious organisation with 9.603 employees, the average annual salary of male employees is €70.000 and the average annual salary of female employees is €63.000. The unadjusted gender pay gap can be easily calculated by subtracting the average salary of female employees from the average salary of male employees and dividing it by the average salary of male employees. The unadjusted gender pay gap in this example is 10% (€7.000). This means that women earn €0.90 for every euro that men earn or put differently, while men get paid for a 5-day work week, women effectively work every other Friday without getting paid.
Assuming that in this fictitious organisation both women and men get an annual salary increase of 3%, this means that over a 10 year period, the total earnings of a woman will be more than €80.000 less than the total earnings of a man.

In this fictitious example, a regression analysis including age, age squared, occupation, years in service, years in service squared, and working time reveals an adjusted gender pay gap of 5%. Figure 2 shows a visualisation of the breakdown of these explanatory variables. A 5% adjusted gender pay gap means that on average, a woman earns 5% less than a man with the same age, occupation, years in service and working time. Looking again at the 10 year difference in earnings between women and men, half of this difference is due to gender which means that a woman with a starting salary of €63.000 will lose out on more than €40.000 simply because she is a woman5. The difference in earnings may be even higher for individual employees in this fictitious example, because any potential bias in promotion and annual salary increases or bonus payments are not taken into account when calculating the average 10 year difference in earnings. It seems likely that an organisation with a 5% adjusted gender pay gap does not practice equal opportunity which can result in women being less likely to get promoted or taking a longer time to receive a promotion, all of which can result in bigger earning losses for individual employees.

Summary and conclusion
While the unadjusted gender pay gap provides important information, it is necessary to also calculate the adjusted gender pay gap which separates the impact of different employee characteristics and shows whether an organisation still has a pay gap after accounting for these characteristics.
Conducing a gender pay analysis is complicated and takes a lot of time. It is important to be aware of potential quality issues in the available data and handle them appropriately in order to obtain a valid outcome.
While a gender pay analysis provides valuable information into an organisation’s equality practices, it is only part of the story. Besides regular gender pay analyses, it is important for organisations to continuously monitor other equality metrics such as the number of women and men across the different layers of the organisation. For instance, an organisation that does not have a gender pay gap, but only has 20% women in its executive management still has room for improvement.
In addition to continuously monitoring specific equality metrics, it is important for any organisation that conducts or commissions a gender pay analysis to have a clear action plan. This action plan needs to outline how the organisation will ensure that the gender pay gap will decrease over time. Making equal pay a key performance indicator for senior management can help in achieving the objectives of the action plan.
Even though by law women and men must be paid equally for equal work, the reality is different. Because of this, some countries have started requiring organisations to conduct annual gender pay analyses and to report the outcomes publicly6. Hopefully, more countries will follow in the future. Before this is the case, any organisation that conducts regular gender pay analyses and transparently reports the outcomes should be applauded for leading the way.
- Gender pay gap statistics ↩︎
- A decomposition of the unadjusted gender pay gap using Structure of Earnings Survey data ↩︎
- International Standard Classification of Education (ISCED) ↩︎
- International Standard Classification of Occupations (ISCO) – 08 ↩︎
- The exact difference in mean or median pay between women and men can be calculated using the coefficients of the regression model. ↩︎
- Gender pay gap reporting: guidance for employers ↩︎