6. Pay and statistical measures

In the context of equal pay for work of equal value in the national statistical databases, occupational classifications, as discussed in the previous section, are the basis for determining which jobs, or occupations, are comparable and can possibly be considered work of equal value. In the following section, the problem of occupational classification is ignored to instead focus on pay differences themselves. An analysis is presented of current pay levels within and, above all, between occupations based on existing classifications and assumptions of equal occupational value, as presented by Harriman et al. (2023).

In the following section, we clarify slight differences in the meaning of pay in different national databases and present some of the implications of different pay measures for pay comparisons. We also discuss the difference between using mean or median averages in pay comparisons, as both are used and also required by the EU Pay Transparency Directive.

Pay refers to remuneration for work done, but there are different ways of defining this remuneration, i.e. different pay measures used in statistics. The remuneration (pay) for work can be linked to different units of time: for example, an hour or a month. Pay can also be fixed or variable. Fixed pay is the same month by month or hour by hour, while variable pay can change upwards or downwards, for example as a result of piece work and commission pay. Remuneration, and therefore measures of pay, can also vary according to what it encompasses, i.e. whether or not increments or bonuses are included. Increments can also be fixed or variable: a fixed increment, for example a shift premium, is the same every month, while a variable increment, such as an overtime increment, depends on the number of overtime hours worked, and can vary every month (Neu Morén & Eriksson Lindwall, 2013). Income types such as bonuses and increments have been found to particularly benefit men (Rubery & Grimshaw, 2015). They are often not included in official statistics on pay, so in the case that bonuses apply in one of two comparable occupations their impact on the pay gap is unknown.

The Swedish statistics on pay structure allow for a choice of comparisons based on both basic and monthly pay. In addition to basic pay, monthly pay includes income such as fixed and variable increments. The National Mediation Office uses monthly pay in its analyses of the gender pay gap. All pay in the Swedish statistics is converted to the full-time equivalent to facilitate comparisons for monthly units, i.e. pay for part-time employees is converted to full-time equivalence to enable calculation of the pay gap. Statistics Finland does the opposite: when comparing full-time and part-time pay, full-time pay is converted to hourly pay. However, the converted levels of pay are not readily available in Statistics Finland’s database, and individuals working part time are not included in statistics based on monthly pay – which makes these statistics somewhat misleading for occupations with many part-time workers.

Monthly pay is the main concept used in pay statistics provided by Statistics Norway (SSB). Monthly pay in the Norwegian statistics include contractual monthly pay (i.e. fixed pay), irregular increments linked to specific tasks or working hours, and bonuses (monetary benefits that are usually not linked to specific tasks, e.g. profit sharing). Overtime pay is not included in monthly pay statistics.

The Finnish monthly pay measure includes more separate components than the Swedish and Norwegian statistics; see Table 5. There are thus variations between how the three Nordic statistics authorities define monthly pay.

As an example from the statistics available through Statistics Norway, a user who wants to know more about monthly pay by occupation (Table 11418) can access information on pay such as: monthly pay, agreed monthly pay, variable increments, bonuses and overtime pay, as well as information on age, and contracted working hours per week. These variables reported in the Norwegian statistics with respect to monthly pay are not available in the same way, or cannot be separated from or combined with monthly pay, in the Swedish or Finnish statistics.

Table 6 shows the average pay for the female-dominated occupational group Hotel receptionists (STYRK code 4224) and the male-dominated occupational group Heavy truck and lorry drivers (STYRK code 8332).

When comparing the monthly pay of the occupational groups, it can be seen that hotel receptionists are paid about 82% that of heavy truck and lorry drivers. The pay gap between these two occupational groups is thus 18%, corresponding to an average pay difference of NOK 7,600 per month. The pay gap for agreed monthly pay is equally large, 18%, and thus seems to be the ‘decisive’ factor in the size of the pay gap for these occupational groups. Table 6 also shows that women’s pay is lower than men’s in both occupational groups.

When overtime pay is taken into account (which is not included as part of monthly pay in Norway), the pay gap widens between the occupational groups in Table 6. When monthly pay and overtime pay are combined, the pay gap is 23% between the occupational groups. Overtime pay serves as compensation for additional hours worked and is not part of contracted monthly pay, thus leading to different levels of pay income for employees in each occupational group. In this case, the corresponding average difference in pay is NOK 10,070.

Whether the normal working hours for hotel receptionists and heavy truck and lorry drivers are the same is unknown. Any differences in normal working hours can increase or decrease a pay gap: in the case of equal pay but different normal working hours, there is an actual pay difference because the compensation per hour worked is different. In this case, the basis for calculating overtime is also different, and since overtime is compensated with above-regular pay, the overtime effect is greater than if the basis for the calculation is the same.

The concept of the gender pay gap is often used to describe the differences in pay between women and men on an average basis, i.e. mean values. Mean values are calculated by dividing the sum of all values by the number of values: the mean pay is thus the sum of all pay divided by the number of people in the group. The median pay is the middle pay level among a group of pay levels sorted in ascending order, from lowest to highest. The National Mediation Office, which is responsible for statistics on the pay structure in Sweden, states that ‘the median value can be useful if a distribution is skewed, with many people earning significantly more or less than the rest of the group’ (www.mi.se). A report on the pay gap in Norway (Grini & Fløtre, 2023) states that the pay gap is reduced if median rather than mean pay is used as a measure. One aspect addressed by the report is that there are more women in the lowest paid group, as well as noticeably more men in the highest paid group. The Nordic Council of Ministers’ report Increasing Income Inequality in the Nordics (Aaberge et al., 2018) uses median values throughout but also points out that the use of related values, such as percentiles (differences in pay between those who, for example, are in the 25% pay percentile – rather than 50%, as in the median – from the lowest or highest earners) may be needed to detect differences that are not picked up through the use of median values alone.

The choice of whether to use mean or median values is a judgement call. If the purpose of analysis is to determine the difference in pay for ‘most’ people across a subset of occupations, ignoring the fact that a number of individuals (usually men) will have significantly higher pay within this group, the median is the more suitable metric. If, on the other hand, the comparison of all pay within an occupation is most relevant, the mean pay is most applicable. For the purposes of the calculations in this report, the latter position is taken and the mean average is used. Another reason is that the mean is the measure most often used in other contexts. However, the authors believe that both mean and median values can, or even should, be used when analysing pay gaps for work of equal value.

It is also not the case that pay gaps calculated on the basis of the median are always smaller than pay gaps calculated on the basis of the mean; the pay dispersion within occupations also plays a role. Here, an example is presented based on data from Swedish pay statistics. Figure 3 shows two pairwise comparisons of the pay situation across occupations.

The first pair compares mean and median pay in the male-dominated occupation of financial analysts and investment advisors, etc. (reference value 100%) with mean and median pay in the female-dominated occupation of public relations professionals. The pay gap is significantly larger when the mean is used. This result thus follows the ‘expected’ pattern of smaller pay gaps when using the median for comparison. This happens in the case of a large pay dispersion, especially in the higher paid occupation – which is often the case.

The second pair compares mean and median pay in the male-dominated occupation of software and system developers (reference value 100%) with mean and median pay in the female-dominated occupation of bank clerks. In this case, the pay gap between the occupations is larger when based on median rather than mean pay. This is because the difference between mean and median pay is small for software and system developers, while the median pay for bank clerks is significantly lower than the mean.