In other languages
  • English
Create a book

Salary calculator

From Statistics Explained

This article introduces the Salary calculator, a simple tool to provide the interested user with estimates of average salaries and adjusted gender pay gaps, broken down as far as possible by gender, age, education, profession, job experience, type of contract, NACE activity sector and enterprise size, for the European Union (EU) and Member States. Its operation is based on microdata obtained from the EU Structure of earnings survey (SES) 2010.

Figure 1: The tool

The tool

Figure 2: Input menu and numerical results
Figure 3: Example of graphical output salaries
Figure 4: Example of graphical output GPG
Map 1: Unadjusted GPG in the EU-27 in %, 2010

The purpose of the Salary calculator is to provide the interested user estimates of average salaries and adjusted gender pay gaps based on the microdata from the 2010 Structure of earnings survey.

It is a simple interactive application designed to give users an easy method to access regression-based information on the impact of personal, job and enterprise characteristics on an individual's wages. Using results from regression analysis for this purpose has several advantages over the use of cross-tables. Apart from providing a much more detailed picture of gross salaries and an adjusted gender pay gap for specific groups in the population than available in currently published tables, it enables the user to easily see the impact of individual factors on wages and the pay gap, all else being equal. This particular tool at hand also gives a quick visual impression of the estimated wages and adjusted gender pay gaps in the EU for any combination of characteristics chosen.[1]

How to use the Salary calculator

The Salary calculator can be accessed here.

  • When saving the Excel sheet, please do not change the name (it will not work if you do). After opening the sheet and allowing macros, click the line saying compare your salary (see Figure 1).
  • A menu with a number of drop down lists will appear (see Figure 2); you can now select for each category the item you are interested in. More information on these categories is available in the methodology section of this page. Please note that some industries and occupations (e.g. fishing industry, armed forced) are not included at all in the data. Some are just not available for individual countries (e.g. the public sector). In this case, no indication for the average salary or the adjusted gender pay gap for these particular industries and occupations can be given. Once you're done, press ok.
  • The rounded point estimates for hourly and monthly wages 2010 will appear, based on the input data (see bottom third of figure 2). For hourly wages below 10 euro, results are rounded to one decimal. The adjusted gender pay gap for 2010 shown is calculated with reference to the male wage, following the calculation done for the unadjusted gap.
  • In order to view the bar charts comparing the values of all Member States for the particular combination of characteristics chosen, press the buttions saying "Show salary graph" or "Show GPG graph". In order to get new results, repeat the steps - the graphs will only be adapted to a new selection after pressing "OK" and the relevant graph button. Please note that the calculator may not show graphs when using earlier versions than Excel 2010 - we apologise for this inconvenience.

Interpreting the results

The most important fact to remember when using the tool is that results are based on regression models, and are thus not directly comparable to descriptive statistics of the SES or any other (national) wage statistics. Regressions are used to calculate the average contribution of each particular characteristic to the hourly gross wage, while controlling for all other variables; final wage estimates (for hourly gross wages) are calculated by adding up all coefficients from all chosen characteristics. Therefore, wage estimates for groups of persons who do not actually exist can nevertheless be calculated (e.g. it is possible to combine the occupation "skilled fishery worker", with the attributes "female", "tertiary education", "65 years old", "senior legislator"). In these cases, estimates are possibly meaningless. However, for all in-sample sub-groups, the tool provides relevant information on average wages as well as adjusted gender pay gaps.


Regression estimates

The SES data on which the Salary calculator is characterised by very detailed information on individual earnings, as well as the match between individual and enterprise level information. In order to take into account unobserved enterprise-level characteristics, an enterprise-level random effects estimator is used.

<math>\qquad \qquad y_{ij}= x'_{ij}\beta + \mu_j + \varepsilon_{ij}</math>

where y is the natural logarithm of hourly (gross) wages of an individual i working in enterprise j. Wages do not include bonuses or irregular payments. The vector x of explanatory variables consiste of personal characteristis, job characteristics and enterprise characteristics (see table below); interactions of several variables with the gender dummy were used where statistically significant. μ is the enterprise level random effect and ε the error term.

The analysis is constrained by the different effects personal and job characteristics have on wages in the different Member States, as well as the differences in sample sizes and coverage. Therefore, explanatory variables on a relatively high aggregation level which deliver satisfactory results in all Member States have been selected into the regression despite the availability of much more detailed information.

Type of variable Values Notes
gross hourly wages natural logarithm Dependent variable; the lowest and highest 0.5% of wages were excluded from the sample.
Personal and job characteristics        
Gender male (base), female Interactions between female and age, age squared, education and occupation are included.
Age age, age squared Individuals aged 23 - 65 are included. Proxy for experience; the age squared term is necessary to capture changing returns to experience.
Education ISCED level 1+2 (base), 3+4, 5+6  
Occupation 1-digit ISCO-08 code Base category is ISCO code 2 (professionals); ISCO code 6 (skilled agricultural and fishery workers) are not available for all Member States.
Job experience in years, years squared Only the experience in the current job is taken into account in this variable.
Type of contract fixed term, permanent Apprentices were excluded from the sample.
Enterprise characteristics                                             
Industry NACE rev. 2 sections Information for section "Public administration and defence; compulsory social security" is not available for all Member States.
Enterprise size 1-9; 10-49; 50-249; 250-499; 500-999; 1000+ Information for enterprises with less than 10 employees is not available for all MS.

In line with the relevant scientific literature, persons below the age of 23, those working less than 16 hours as well as apprentices were excluded from the analysis, as well as any cases with incomplete information in the variables of interest. The individuals with the lowest and highest 0.5 % of hourly wage are excluded as well in order to avoid a bias in the results due to outliers. Some industries and occupations (e.g. fishing industry, armed forced) are not included at all in the data. Some are just not available for individual countries (e.g. the public sector). In this case, no indication for the average salary or the adjusted gender pay gap for these particular industries and occupations can be given.

For the SES 2010, provision of data on the public sector and firms with less than 10 employees was optional. Many Member States collected this data anyway, so we show the information where it is available.

Data and to some extent wages for those below the age of 23 and those above the age of 65 are influenced strongly by the legal and institutional situation in the Member States, and therefore hard to compare. Excluding these age groups is in line with the relevant econometric literature.

There are no cut-offs for tenure in the current firm, but if the amount of years entered exceeds "age-14" it will not be accepted, as it implies the individual started to work at age 13 or younger. Individuals working less than 16 or more than 60 hours per week were excluded in the regression analysis.

The gender pay pap estimates are based on the coefficients of the dummy for female, and interaction effects of female with age, age squared, educational levels, and occupation dummies. Coefficients which are at 95% level not significantly different from zero are set to zero for the calculation of salary levels and GPG. The complete regression outputs are available upon request.

The adjusted gender pay gap

The adjusted gender pay gap is a concept which differs from the unadjusted gender pay gap (GPG), one of the structural indicators used to monitor the European Strategy for Growth and Jobs. Figure 4 shows the GPG in the EU in 2010. The GPG measure the relative difference in the average gross hourly earnings of women and men within the economy as a whole:

<math> GPG\ in\ \% = 100* \left ( \frac{gross\ hourly\ wages_{males}-gross\ hourly\ wages_{females}}{gross\ hourly\ wages_{males}} \right) </math>

The adjusted gender pay gap on the other hand is the remaining and unexplained part of the pay gap between men and women once observed personal, job and firm characteristics have been controlled for. This adjusted GPG differs according to the data set, specification and regression model used; it is noteworthy that it cannot be interpreted straightforward as solely being due to discrimination. Other factors which influence the size of the adjusted GPG are unobserved characteristics of the individual, job or firm. It is however also important to note that by accounting for the effect of factors such as the number of hours worked per week, the education and experience of the individual, occupation and industry, parts of the gap may be explained, but still reflect the result of discrimination (e.g. if access to higher education is made more difficult for girls, some occupations are not accessible for women, etc.). In addition, estimates of the adjusted gender pay gap are influenced by the self-selection of women with specific characteristics into employment, often depending on the legal framework and existing labour market institutions. Econometric methods to control for these effects exist, but are difficult to apply to the SES due to the limited amount of information on personal characteristics and family situations. A study attempting to apply these methods nevertheless using auxilliary data for the SES 2006 is available here A general discussion of the differences between adjusted and unadjusted gender pay gap, literature reviews as well as implications for policy by Eurofound can be found here.

See also


  1. Other examples of salary calculators based on regression analysis are available for Switzerland and Luxembourg; calculators based on cross tables are available for the EU (GPG only) and Belgium