STATISTICS AUSTRIA

Directorate Social Statistics
Living Conditions, Social Protection

Guglgasse 13

A-1110 Vienna

AUSTRIA

The Austrian Structure of Earnings Survey (SES) has been conducted every four years since 2002.The survey is based on Council Regulation (EC) No 530/1999 concerning structural statistics on earnings and on labour costs and Commission Regulation (EC) No 1738/2005. The survey covers all mandatory variables. Since 2010, the optional variable 'Citizenship' has also been available.

This quality report refers to the SES for the reference year 2018. It is based on the Commission Regulation (EC) No 698/2006 of 5 May 2006 Implementing Council Regulation (EC) No 530/1999 as regards quality evaluation of structural statistics on labour costs and earnings.

Statistical classification of economic activities in the European Community (NACE Rev. 2), International Standard Classification of Occupations (ISCO-08), International Standard Classification of Education (ISCED11), Nomenclature of Units for Territorial Statistics (NUTS).

The concept and the definitions are set out in the Council Regulation (EC) No 530/1999 of 9 March 1999 concerning structural statistics on earnings and on labour costs, the Commission Regulation (EC) No 1738/2005 of 21 October 2005 amending Regulation (EC) No 1916/2000 as regards the definition and transmission of information on the structure of earnings and the Structure of Earnings Survey 2018 Implementing Arrangements.

Enterprises or local units and employees in these units, as defined in Council Regulation (EEC) No 696/93.

Employees in enterprises with at least 10 employees in the sections B to S (excluding O) of NACE Rev. 2.

Austria.

Reference year: 2018.

Reference month: October 2018.

Not applicable.

-

The SES 2018 was conducted as a combination of primary and secondary statistics.

 With the exception of the variable "Citizenship", optional variables were usually not included in the survey.

Since 2002 every 4 years. 

In Austria, the Structure of Earnings Survey is conducted on the basis of a two-stage random sampling approach of enterprises (first stage) and employees (second stage).

The population used for sampling comprises 46 155 enterprises with at least 10 employees in NACE Rev. 2 Sections B-N and P-S and the approximately 2.7 million employees of these enterprises. 11 352 enterprises and around 212 000 employees were selected for sampling.

As a first step, a stratified random sample of enterprises was drawn from the business register. The variables economic activity (NACE Rev 2), location at level NUTS 1 and size of the enterprise (size categories: 10-19, 20-49, 50-99, 100-249, 250-499, 500-999, 1 000+) were used as stratification criteria.

In a second step, a simple random sample of employees was drawn within the selected enterprises. Depending on the size of the enterprise, the corresponding number of employees per enterprise was selected. In order not to over-burden enterprises, the maximum number of selected employees per enterprise was limited to 80.

 
Drawing the sample at the level of the local unit was not possible, because at the moment of sampling employees could be unambiguously allocated only at enterprise level. Enterprises with local units in different NACE divisions / NUTS 2 regions were, pursuant to the national implementing regulation[1]), required to allocate the employees selected to the local unit. This affected around 25% of the enterprises.

To ensure the quality of statistical results, the already tried and tested web-questionnaire (eQuest-Web) was used for the primary survey. The web-based questionnaire incorporated explanatory notes and plausibility tests on the individual survey parameters. Moreover, integrated classifications allowed the automatic recognition of NACE codes and occupations according to ISCO-08. A hotline was also available to provide information to enterprises.

The submitted data were subjected to several layers of plausibility testing at both micro and macro levels for the ex-post identification of any measuring errors. The first step was for the survey data to be examined and corrected at micro level. In a second step the survey data were linked to the secondary data and subjected to another plausibility test. After weighting the data, aggregates were subject to a third plausibility test (macro plausibility).

[Not requested]

-

Commitment to quality

http://www.statistik.at/web_en/about_us/responsibilities_and_principles/commitment_to_quality/index.html

[Not requested]

-

The results of the SES are used at national level mainly by various federal ministries, employer and employee organisations, universities and other research institutes, the media, enterprises and private individuals.

Information about the profile of users or the level of satisfaction with the data provided is not available for Austria.

[Not requested]

In Austria, responding to the SES is a legal requirement. The unit-response rate was 98% (see point 6.3.3.1. Unit non-response - rate).

[Not requested]

Coefficients of variation (see point 6.2.1. Sampling error - indicators).

The sample of enterprises is based on the business register. Therefore coverage errors are directly connected to the quality of the register data. In order to achieve a high level of completeness and relevance, Statistics Austria conducts ongoing technical comparisons between the business register and external sources of administrative data (register of companies, tax and social security data etc.). Moreover, information from economic surveys makes a substantial contribution to keeping the business register up to date.

Over-coverage was found with regard to 0.7% of the employees in the sample. These employees were no longer active in the company at the time of the survey and thus did not receive remuneration.

Statistical units in section O (public administration and defence; compulsory social security) were not part of the population, which meant that local units whose higher-level unit was allocated to section O in the business register were not recorded. According to the business register, this led to under-coverage of 6.9% of employees.

According to the employee estimation model in the business register[1] the under-coverage relates to the service sector mainly, in particular to education (section P -44%) and to parts of public health and social services (section Q -24%). In the education sector, it is mainly employees in public schools and childcare facilities who are not covered[2] and, in health and social work, the same applies to some employees in public homes and hospitals. Smaller economic sectors were also affected. In the area of "arts, entertainments and recreation" (section R -17%), employees of public libraries, archives, of museums and swimming baths are missing. In industry, municipal employees in section E "water supply and waste disposal" (-11%) were not recorded (see also point 8.1. Comparability -  geographical).

This can lead to either an over-estimation or under-estimation of earnings in these sectors.

[1] Employees are allocated to a local unit on the basis of an allocation key. This allocation key is based on an estimation model according to which the employees of the enterprise are allocated to the local units.

[2] Section P "Education" includes private and faith-based childcare establishments and schools, all universities and colleges, driving schools and the adult education sector.

See point 6.3.1. Coverage error.

[Not requested]

To prevent measuring errors, the already tried and tested web-based questionnaire was used for the primary survey. The submitted data were subjected to several layers of plausibility testing at both micro and macro levels for the ex-post identification of any measuring errors (see point 3.4. Data validation).

The secondary data were, in general, of very high quality. The business register was continually improved in order to make it more comprehensive and up-to-date (see point 6.3.1. Coverage error). The social security data came from monthly notifications to the social security institutions, meaning that employment relationships could be recorded precisely to the day. Tests were carried out upstream in the Statistics Austria database system to ensure the good quality of these data. For wage tax, data from the Austrian wage tax statistics were used which had already been verified by the tax statistics department of Statistics Austria. In this way, optimum data quality could be ensured in the calculation of gross annual earnings and annual bonuses and allowances.

The data from the education register were based on data from the 2001 census, which were constantly updated and supplemented according to the information sent by schools, universities and other educational establishments. The ongoing qualification of employees from other countries could be taken into account only if the training was completed in Austria, if the qualification was officially recognised, if the Public Employment Service Austria (AMS) provided the information or if an academic degree was entered in the Central Register of Residents (ZMR) (see point 6.3.3.2. Item non-response).

Enterprises are legally obligated to provide information [1]. In addition, the following measures were taken to limit non-response: firstly, reminder letters to enterprises that did not reply by the response deadline; then dunning letters sent to those enterprises that had not replied to the reminder letter; thirdly, reminder telephone calls and, fourthly, reports to the authorities for suspected failure to comply with information obligations (see also point 7.2. Punctuality).

[1] Regulation of the Federal Minister for Economic Affairs and Labour on the Structure of Earnings Survey in industry and parts of the service sector (Verdienststrukturstatistik-Verordnung), Federal Law Gazette (BGBl.) II No. 66/2007, as amended by BGBl. II No. 99/2011.

The overall unit response rate was 98.1%. The unit non-response rate of 1.9% (199 enterprises) can be broken down further into companies that are no longer active (0.7%) and refusals (1.2%).

See also point 6.3.4.1. Impuation - rate.

Due to the complete changeover to a web-based questionnaire that incorporates plausibility tests on the individual characteristics, it was possible to reduce the item non-response of the survey data. On the other hand, administrative data showed a slight increase of missing values. 

See point 6.3.3.2. Item non-response – rate

Survey data

Missing information about the variable 'Length of service with the enterprise' (0.1%) was calculated on the basis of information from social security data on the date of joining or leaving the enterprise.

Information on the variable 'Type of employment contract' (0.1%) was also completed with the aid of social security data.

For the variable 'Annual days of holiday leave'(0.7%), the statutory holiday entitlement was imputed, taking into account age and length of service.

Administrative data

The variables 'Gross annual earnings' and 'Annual bonuses and allowances' were imputed (4.1%) using regression analysis. The basis of the calculation was the information on gross monthly earnings from the survey in combination with information on the date of joining or leaving the enterprise.

Missing information on the 'Highest completed level of education' was imputed based on a multinomial regression, using the predictors sex, gross hourly earnings, age classes, occupation and citizenship (Austrian, EU-15 and other). Information on education was missing for 0.7% of employees with Austrian citizenship and 44.4% of employees with other citizenship. This is because the education register receives ongoing notifications only from educational institutions in Austria. The ongoing qualification of employees from other countries can be taken into account only if the training was completed in Austria, the qualification is officially recognised, the Public Employment Service Austria (AMS) provides the information or if an academic degree is entered in the Central Register of Residents (ZMR) (see point 6.3.2. Measurement error). The value of the variable highest completed level of education is therefore significantly restricted for non-Austrians. This means that the earnings of employees who are not Austrian, broken down by level of education, could be either over- or under-recorded.

To correct for unit non-response the net sample was adjusted to the population given by the selection framework. The weighting of the data, like the sampling, took place in several steps. The first step was to reproduce the sample design, by giving each enterprise the base weight of the sampled unit Wk, representing the reciprocal selection probability in the stratum h (Wk=Nh/nh). Using the total number of employees per enterprise as a basis, a sample weight was also calculated for each employee in the sample. Using the base weight Wk of the enterprise k, the number of employees Bk of this enterprise and the number of employees bk of the enterprise k in the sample, the base weight Wbhk was calculated for each employee in the sample unit.

W_bhk= W_k*B_k/b_k

Based on the allocation of employees to local units by enterprises, a breakdown by economic activity and region at the level of the local unit could be conducted on a European legal basis. For the weighting of the aggregated local units it was necessary, in a next step, to calibrate the weights in such a way that the sum of the weights of each NUTS 1 region cross-referenced with NACE sections matched the corresponding population.

Then, the second step of the weighting was to modify the basic weighting W_bhk, which initially was allocated to each aggregate, so that the sum of the weights of these units corresponded both with the relevant figures from the population in the cross-referencing of the NACE sections and of the NUTS-1 region and with the sum of all male and female employees in the population.[1]

[1] Weighting by gender has been carried out since 2006, as the Structure of Earnings Survey has been used as the source for calculating the gender pay gap throughout the EU.

[Not requested]

[Not requested]

[Not requested]

[Not requested]

As a consequence of the COVID-19 crisis Austria could not keep the deadline 30 June 2020 (t+18) for transmitting the SES 2018 microdata to Eurostat. The final data were delivered to Eurostat on 23 August 2020.

[Not requested]

[Not requested]

[Not requested]

The comparability of the results between Member States is restricted because of under-coverage in NACE Rev. 2 sections P, Q, R and E (see point 6.3.1. Coverage error).

Statistical units in section O were not part of the population, which meant that local units whose superordinate enterprise was allocated to section O in the business register were not recorded. The proportion of employees in those economic sectors (except R) was therefore much lower in Austria than the EU average.

The exclusion of section O from the survey population also affects the results for certain professional groups such as teaching professionals (ISCO 23), health professionals (ISCO 22), health associate professionals (ISCO 32) and personal care workers (ISCO 53).

Proportion of employees by economic activity in %

S: Eurostat. Not including Greece. Enquiry of 15 December 2020.

[Not requested]

The changes to the definitions between the 2002, 2006, 2010, 2014 and 2018 surveys are mainly the result of amendments to legal acts and classifications (NACE, ISCO, ISCED). Pursuant to Article 3 of Regulation (EC) No 530/1999, the inclusion of Sections M – O of NACE Rev. 1 was optional for the SES 2002. Furthermore, a derogation from Article 6 has been in force for Austria in 2002, whereby the statistical unit could relate to the enterprise rather than to the local unit.

Coverage                      

• 2002 Sections C-K of NACE Rev. 1
• 2006 Sections C-K and M-O of NACE Rev. 1.1
• since 2010 B-N and P-S of NACE Rev. 2

Statistical units              

• 2002 enterprises
• since 2006 enterprises/local units

Weighting                      

• 2002 enterprises/employees
• since 2006 local units/employees by sex

Classifications                  

• 2002 NACE Rev. 1, 2006 NACE Rev. 1.1, since 2010 NACE Rev. 2
• 2002 / 2006 ISCO-88, since 2010 ISCO-08
• 2002 / 2010 ISCED 97, since 2014 ISCED 11

[Not requested]

[Not requested]

[Not requested]

The characteristics 'Wages and salaries' per employee in the National Accounts and 'Gross annual earnings' in the SES are comparable only with certain caveats. Such comparisons lead to a discrepancy of -5%.

The main reasons for this discrepancy are:

• Statistical units in National Accounts are companies (or ‘establishments’), which are units between the enterprise level and the level of local units.
• The coverage of National Accounts includes enterprises with fewer than 10 employees.
•  'Wages and salaries' in National Accounts include estimates for tips or gratuities, in particular in section I (Accommodation and food service activities) and the divisions 49.32 (Taxi operation) and 96.02 (Hairdressing and other beauty treatment) in NACE Rev. 2.
• The number of employees in National Accounts corresponds to the annual average whereas SES exclusively relates to employees in the reference month (October). Seasonal variations thus arise as a result of the choice of reference period.
• Differences occur also regarding the calculation of full-time units (FTU), e.g. hours paid / hours worked, annual average / full year.
• The National Accounts include units whose enterprises come under Section O. This plays a role in NACE sections P, Q, R and E, in particular.

[Not requested]

See point 9. Accessibility and clarity. 

See point 9. Accessibility and clarity. 

See point 9. Accessibility and clarity. 

See point 9. Accessibility and clarity. 

See point 9. Accessibility and clarity. 

See point 9. Accessibility and clarity. 

See point 9. Accessibility and clarity. 

[Not requested]

[Not requested]

[Not requested]

[Not requested]

[Not requested]

[Not requested]

[Not requested]

[Not requested]