Concepts and calculations

This article brings together how the pay-equity figures are calculated, using one worked example throughout. It is written so that you can explain the numbers to management, union representatives and an auditor. Each Salary Mapping page also has its own “How it is calculated” section in its help article.

The worked example. The role Case Officer (same location, same equal-work group) has seven employees:

Women: Anna SEK 30,000 (full time), Bea SEK 24,000 (80 %) and Cilla SEK 36,000 (full time).
Men: David SEK 33,000, Erik SEK 35,000, Filip SEK 37,000 and Gustav SEK 43,000 (all full time).

1. Which pay is compared. Monthly base salary. Variable pay (bonus, commission and the like) is added only when you tick “Include variable pay”: this year’s entries are summed and divided by 12, so a SEK 12,000 bonus this year becomes SEK 1,000/month. Fixed allowances are added to the base salary only if the organisation has chosen the pay basis “Base salary + fixed allowances” under Settings › System parameters (default: base salary only); the choice applies to Overview, Classification, Group Insights and Individual Analysis, is shown as “Pay basis: …” on each page and is saved in the year’s snapshot. One-time payments are never included. An employee without a valid base salary (missing, 0 or invalid) is left out, also when variable pay is included, and the pages show how many were left out. An employee is included in the analysis unless “Include in analysis” is explicitly set to no.

Exception: the EU report. Variable pay above is “year to date” (the current calendar year divided by 12). The EU report’s art. 9 items instead count variable components over the report’s reference year (by default the previous full calendar year) and include allowances as complementary pay; see EU Pay Transparency Report. With the EU pay concept per pay component switched on, the organisation’s choice per component decides instead what is basic pay, variable pay or not counted (point 12). The pay basis above (System parameters) never affects the EU report, and the choice per component never affects the Swedish analysis.

2. Full-time equivalent. Every comparison uses full-time pay: salary ÷ employment rate. Bea: 24,000 ÷ 0.80 = SEK 30,000. A value such as “80” is read as 80 %. A missing, zero or invalid employment rate counts as full time. The pay shown for an individual, and any proposed raise, is always the actual amount.

3. Mean and median. The mean is the sum divided by the count. The median is the middle value when the salaries are sorted; with an even count it is the average of the two middle values.

Women (full-time pay 30,000, 30,000 and 36,000): mean 32,000, median 30,000.
Men (33,000, 35,000, 37,000 and 43,000): mean 37,000, median (35,000 + 37,000) ÷ 2 = 36,000.

4. Women’s pay as a percentage of men’s, and the pay gap. Women’s pay as a % of men’s = women’s mean ÷ men’s mean × 100, rounded to one decimal: 32,000 ÷ 37,000 = 86.5 %. The pay gap = (men’s pay − women’s pay) ÷ men’s pay × 100. Mean: (37,000 − 32,000) ÷ 37,000 = 13.5 %. Median: (36,000 − 30,000) ÷ 36,000 = 16.7 %. A positive gap means women earn less, a negative one that they earn more.

5. The 5 % rule. The EU report, the annual report, the approval warning in the salary review and the pay-equity guardrail mark a group when the gap is at least 5 % in mean or in median pay, in either direction. The decision uses the unrounded gap:

Women SEK 47,300, men SEK 50,000: 5.4 %. Shown as 95 % but marked.
Women SEK 47,520, men SEK 50,000: 4.96 %. Also shown as 95 % but not marked.
Women SEK 52,600, men SEK 50,000: −5.2 % (in women’s favour). Marked.

The 5 % line is fixed. The system parameter “Pay Gap Warning Threshold” does not change it (it only sets the colours in Salary Simulation).

6. Anonymity threshold (minimum group size). At least 3, or the higher value you have set. It counts people with a usable salary, not employees. Women’s average is shown only when at least that many women have a salary, likewise for men, and a comparison between the genders only when both clear the threshold. A group’s average is also hidden if someone’s pay could be worked out by subtraction: 1 woman and 5 men with the men’s average shown gives her pay = 6 × the group average − 5 × the men’s average. Anyone allowed to see individual salaries sees every figure in the internal analysis. Everything that leaves the organisation, such as the EU report and its exports, is always anonymity-protected; there the threshold is 5 unless you have saved a value of your own.

7. Female-dominated. A role (or group) is female-dominated when the share of women is at least the threshold, normally 60 % (can be set between 50 and 70 %). So 3 women out of 5 = 60 % counts as female-dominated. Case Officer in the example (3 of 7 = 43 %) is not. The share is counted among those included in the analysis who have a salary, with the same rule on the Overview, in Classification, in Group Insights and in Individual Analysis.

8. Group Insights: role against role. Each female-dominated role is compared with higher-paid roles that are not female-dominated, both in the same group and, at the same location, in groups ranked lower (lower demands). The gap = (comparator role’s median − female-dominated role’s median) ÷ female-dominated role’s median × 100. Example: Assistant Nurse with a median of SEK 30,000 against Operations Technician with a median of SEK 34,500: (34,500 − 30,000) ÷ 30,000 = 15.0 %, and the female-dominated role is at 30,000 ÷ 34,500 = 87 % of the comparator. Note the base: here the gap is measured against the lower-paid role, while the pay gap in point 4 is measured against men’s pay (the same two salaries give 13.0 % there). The whole difference is reported as unexplained; no flat deductions for experience, tenure or performance are made.

9. Individual Analysis: person against target. Within each combination of location, equal-work group and role:

A woman is compared with the men’s median in the role, and a man with the women’s median.
If the role is female-dominated, everyone in it, men included, also gets a structural target: the highest median among higher-paid roles that are not female-dominated, in the same group at the same location.
The target is the higher of the two. Anyone whose full-time pay is below their target is flagged.
Adjustment per month = (target − the person’s full-time pay) × employment rate.

Example: Anna (30,000) has the target 36,000 (the men’s median) and needs SEK 6,000/month. Bea (full-time pay 30,000, 80 %) needs (36,000 − 30,000) × 0.8 = SEK 4,800/month as an actual raise. David (33,000) is compared with the women’s median of 30,000 and is not flagged. A role where fewer people than the minimum group size have a salary is not analysed in Individual Analysis.

Justified. A group or category counts as justified only when every person Individual Analysis flags in it has an objective justification (Experience, Performance or Market). “Unjustified” never counts. The same rule applies in the EU report, in the salary-review warning and on the start page, where only people who are flagged right now are counted. Small roles. Anyone allowed to see individual salaries also gets roles of 1–2 people audited in Individual Analysis, so their pay can be flagged. For everyone else, and in anything shared onwards, the anonymity threshold applies.

10. Job-evaluation points. Each group is scored 1–5 on the four basic criteria (skills, effort, responsibility, working conditions) and on your own. The score = the sum of weight ÷ 100 × value, on the same 1–5 scale. Weights are whole percentages adding up to 100; in the Sweden and EU/EEA profiles each basic criterion has at least 10 %. Example with weights 40/20/30/10 and values 4/2/3/5: 0.4 × 4 + 0.2 × 2 + 0.3 × 3 + 0.1 × 5 = 3.4. A criterion with sub-criteria gets the weighted mean of its sub-criteria to one decimal, and exactly x.x5 rounds up (3.55 becomes 3.6).

11. Pay quartiles. Everyone in the analysis is ranked by full-time total pay (base + variable) and split into four bands as equal in size as possible; a remainder goes to the lowest bands (10 people give 3, 3, 2 and 2). The share of women and men is reported per band. People on exactly the same pay across a band boundary are split proportionally. A band with fewer people than the anonymity threshold is shown as protected.

12. Variable pay components. The share receiving variable pay = number of women (men) with variable pay ÷ all women (men) in the analysis. The variable-pay gap is calculated only among those who actually receive it, with the formula from point 4. Example: three women with SEK 1,000, 1,500 and 2,000/month (mean 1,500) and four men with SEK 1,500, 2,000, 2,500 and 3,000/month (mean 2,250) give (2,250 − 1,500) ÷ 2,250 = 33.3 %. The EU report’s quartiles and variable-pay section always include variable pay; its per-category table is based on base salary. Pay components in the EU report. When the EU pay concept per pay component is switched on (Settings › System parameters), the organisation’s choice for each component decides where it counts: “Basic pay” is added to the base salary (the mean and median gap per category and basic pay per category of workers), “Variable (complementary) pay” counts in the variable components (the gap, the share receiving it and variable pay per category), and both are part of the quartiles’ total pay and hourly pay; “Not included” counts nowhere. Each component is turned into a monthly amount over the reference year: variable pay per type and one-off payments dated that year divided by 12, allowances as the average of the year’s months, and benefits by their periodicity (monthly as is, annual divided by 12, a one-off benefit divided by 12 in the year it applies, a percentage of the base salary) for the months the assignment was in force. Benefits without a monetary value are not counted, and the report shows how many there are. Example: three women with a base salary of SEK 30,000 and an allowance of SEK 3,000/month, three men with a base salary of SEK 30,000 and a SEK 36,000 annual bonus. With allowances as variable pay the base-pay gap is 0 % and the variable components are 3,000 against 3,000/month; with allowances as basic pay, 33,000 is compared with 30,000 in basic pay and no woman receives variable pay.

13. Hourly pay (preparation). Hourly pay = (base + variable pay per month) × 12 ÷ (employment rate × full-time weekly hours × 52). The weekly hours come from the employee’s legal entity, else the organisation’s default, else 40. Example: SEK 30,000 + 1,000/month at 80 % and 40 hours: 372,000 ÷ 1,664 = SEK 223.56/hour. With a 37-hour full-time week it is 372,000 ÷ 1,539.2 = SEK 241.68/hour. Employees without a recorded employment rate are left out.

14. Cost of correction. Annual amount = adjustment per month × 12. The plan is spread over 1–5 years (default 34/33/33 %). For each year you see that year’s new raise and the total cost including earlier years’ raises, because a raise stays. Anna: SEK 72,000/year on 34/33/33 gives new raises of 24,480, 23,760 and 23,760 and a total cost of 24,480, 48,240 and 72,000, and then 72,000 every year. The amounts are salary, without employer social contributions or pension.

15. Currency and employer. Salaries in different currencies are never added together and never converted with an exchange rate. An analysis uses the organisation’s default currency (or the largest group) and shows how many were left out. When the annual evidence is kept per legal employer, each employer is calculated separately, on its own employees in its own currency; employees without a legal entity (or with an unknown one) count towards the headquarters.