Using the category of worker variable in your compensation model

When organizations define categories of worker by grade, it often also drives pay. This raises some common questions: does it make sense to use grade as an objective factor in the compensation model, or is that double counting the variable? And if it is not, why does the estimated adjusted pay gap change depending on whether grade is included in the model or not?

You will find answers to these questions in the following sections:

1. Key points

2. The two roles of grade

3. Why grade is not double-counted

4. Why the gap changes when grade is removed

5. What triggers a joint pay assessment

This is an note/information message.

This article assumes that your categories of worker are defined by grade alone, and that you are deciding whether grade should also be included in the compensation model.

Key points

• Include grade in the model if grade is part of how your organization sets pay.

• Including grade is not double counting: within a category of worker defined by grade, grade explains none of the pay gap and is never used as a justification.

• Removing grade changes the estimated pay gap per category because it distorts every other factor in the model, and those distorted factors influence the result.

• A low adjusted gap does not mean no joint pay assessment: the 5% Article 10 threshold applies to the unadjusted gap per worker category.

The two roles of grade

Under the EU Pay Transparency Directive (EUPTD), grade can play two different roles:

• Category of worker (Article 4 / Article 9), which establishes that employees perform equal work or work of equal value. This is the legal predicate.

• Model factors (Article 10), which provide objective, gender-neutral reasons for a pay difference in a category of worker.

This is where the double-counting concern comes from, and it is a valid one. If grade played both roles at once, the reasoning would be circular: employees would be considered to perform work of equal value because they share a grade, and the pay difference between them would then be explained by their grade. Such a justification would not hold up in a joint pay assessment with workers' representatives.

However, this is not how the compensation model uses grade, as explained in the following section.

Why grade is not double-counted

The system does not double-count grade. The examples below show why.

This is an note/information message.

To keep our examples easy to follow, they add euro amounts together. In PayAnalytics, the model works on log pay: factors multiply rather than add, and the gap is expressed as a percentage. The same principle applies, and holds even more precisely, because in log terms it is the percentage gap that stays the same.

Anna and Ben are both in Grade 5. Anna has 4 years of tenure and Ben has 8. The model calculates expected pay as follows:

Expected pay = €40,000 base + Grade 5 effect + €1,000 per year of tenure

In the first scenario, the Grade 5 effect is €20,000, as shown in the following table:

The gap between them is €2,000 − (−€1,000) = €3,000 in Ben's favor.

In the second scenario, the Grade 5 effect is doubled to €40,000, as shown in the following table:

The gap between them is −€18,000 − (−€21,000) = €3,000 in Ben's favor – exactly the same result as in the first scenario.

The value assigned to grade does not affect the result. Because Anna and Ben are both in Grade 5, the Grade 5 effect is applied equally to both employees and cancels out when their pay is compared.

In terms of Article 10, within a category of worker defined by grade, grade explains zero percent of the gap and is never offered as a justification. It is included in the model only to prevent the other factors from being credited with pay differences that they do not explain.

This means that the justification relies entirely on the other factors used, such as tenure, location, job family, and performance. Including grade in the model gives you no legal advantage. It only keeps the estimates for the other factors accurate.

Why the gap changes when grade is removed

Removing grade from the model does not only delete one number; it also changes all the others.

Without grade, the remaining factors absorb part of its effect. For example, employees with longer tenure tend to be in higher grades. As a result, the model attributes a larger share of the pay difference to tenure than is justified.

In the third scenario, Anna and Ben have the same actual salaries and the same grade, but there is no grade term in the model. All the other values in the model change as a result. The base amount amounts to €55,000, and tenure is now worth €3,000 a year instead of €1,000. The model calculates expected pay as follows:

Expected pay = €55,000 base + €3,000 per year of tenure

The results are shown in the following table:

The gap between them is −€9,000 − (−€4,000) = €5,000 in Anna's favor.

Nothing about Anna and Ben has changed, but the gap has reversed direction. It moved from €3,000 in Ben's favor to €5,000 in Anna's favor, only because grade was removed from the model and tenure absorbed its effect.

How much the gap for a given grade moves depends on how differently men and women in that grade are distributed across the other factors, such as tenure, job family, and location. It does not depend on the number of employees in the grade.

This is an attention message.

This change can be large enough to affect which grades exceed your pay gap target, and therefore where you allocate your remediation budget.

What triggers a joint pay assessment

Under Article 10 of the EUPTD, the 5% threshold applies to the unadjusted pay gap reported per category of worker under Article 9, not to the estimated adjusted gap. The adjusted pay gap is the evidence you use to justify a pay gap; it does not trigger the process.

The Article 10 process consists of the following steps:

1. A category of worker reports an unadjusted pay gap of 5% or more.

2. The employer justifies it on objective, gender-neutral criteria. This is where the adjusted analysis is used. If the gap can be justified, the process stops here.

3. If it cannot be justified, the employer remediates within six months.

4. If it remains unjustified after that, the employer carries out a joint pay assessment with workers' representatives.

For example, a category of worker with a 14% unadjusted pay gap and a 3% adjusted pay gap is still within the scope of Article 10. It enters the process at step 2, with strong evidence to justify the gap. An internal target that you set on the adjusted gap is a management tool, not the statutory test.

This is a tip message.

Article 10 requires the criteria for justifying pay gaps to be established jointly with workers' representatives, so your compensation model, including which factors it uses and why, may become part of that discussion. We recommend documenting the reasoning behind your model factors in advance.