Czech Statistical Office (CZSO)

64 Labour Market and Equal Opportunities Statistics Department

Czech Statistical Office

Na Padesatem 81

100 82 Praha 10

Czech Republic

The Czech Republic (CR) has been conducting its labour cost survey since 1994, regularly every year, according to the Eurostat’s methodological recommendations and regulations. However, the survey also takes into account national terminology, recording methods, variable reporting, and other economic and legal conditions in force in the CR, so that the results can be also used by domestic users.

The data are broken down by economic activity in accordance with NACE Rev. 2 - Statistical classification of economic activities in the European Community.

The survey cover all the economic activities A-S defined by NACE.

Labour costs mean the total expenditure borne by employers in order to employ staff. Labour costs include compensation of employees with wages and salaries in cash or in kind and employers' social contributions, vocational-training costs, other expenditures, taxes relating to employment regarded as labour costs, less any subsidies received. All the definitions are compliant with EU Regulation on the Labour Cost Survey no. 1726/1999.

Enterprise with one and more employees.

The survey covers whole national economy: legal persons (including administrative regions, municipalities and semi-budgetary (subsidized) organizations), incorporated natural persons and government departments.

The whole territory of the country.

The data refer to whole year of 2020.

Not applicable.

The Business Register (BR) kept by the CZSO is used as the sampling frame. On its birth, each unit is assigned a unique identification number (which is also used for other purposes than for statistics) and is recorded in the BR under this number. The BR includes all legal persons in the CR as well as natural persons‑entrepreneurs, both incorporated and unincorporated. It also contains, inter alia, information on principal activity, number of employees, legal form, and SNA sector, which is used to produce a survey population and a sample of units (in the sampled part of the population).

The BR is continuously updated with information from administrative sources of data on the birth of units and their selected attributes, on their legal deaths and on some other facts. Information chosen from statistical surveys are also loaded in the BR (according to pre-set principles and agreed procedures). With the expanding use of administrative data, we gradually manage to supplement and update information on activity of individual businesses.

Every year

Reporting units provide their data electonically via DANTEWEB (web page) or email.

The data is checked and, if necessary, verified at the unit.

Not provided.

Not applicable.

The quality is ensured in accordance with the Quality Policy, see attachment (CZSO Data Revision Policy) .

Not provided.

LCS results constitute an important source of information on development of labour costs in various national economy industries and in individual regions. There is a growing need stemming from the competitiveness of Czech firms to compare these findings internationally, too.

The information on labour costs are provided to international institutions (Eurostat, ILO, OECD) and, in the CR, to state and scientific institutions and, if requested, employers, unions and media.  No systematic information on external users has been completed.

In the Czech Statistical Office (CZSO), LCS results are used for publishing and as source data for analyses and for calculation of the Labour Cost Index. They are also in-house used by the CZSO National Accounts in the form of special customized tables.

User satisfaction has not been expressed.

As regards compliance with the EC legislative, the main imperfection is that data on local units are missing. The LCS have used enterprises for reporting units, because the connection of enterprises with local units is not clear.

Regional breakdown to NUTS1 is not applied, as the NUTS1 level corresponds to the whole territory of the Czech Republic.

Persons employed by temporary employment agencies are not collected separately. No other data or breakdown than those requested by Commission Regulation No 1726/1999 as amended were included in the resulting tables transmitted to Eurostat. It means that 100% of all mandatory variables requested by EC was available.

Not provided.

See below.

Not provided.

There were five layers of inputs to match:

• (f) Full-version LCS /complete list of variables as regards the EC Regulation No 1726/1999 as amended/ had a scope of all the enterprises with 250 employees or more and a small sample of enterprises 10-249.
• (d) Version of LCS with lightly shortened list of variables was for sample of enterprises with 1-249 employees.
• (b) Version of LCS with more shortened list of variables on number of employees and earnings was for additional sample of enterprises with 10-249 employees.
• (a) The most shortened version of LCS only on basic indicators on number of employees and earnings was for a sample of minor enterprises with 1-9 employees.

There are conjunctions in the versions inside various strata so in one stratum there can be more versions possible.

• The structural business statistics (SBS) was the fifth layer; only two variables (number of employees, wages) were employed for LCS purpose, but the scope of enterprises covered is widest.

This structure was chosen to reduce the sampling error for the most important variables and, on the other hand, not to increase the burden of respondents. Structures of labour costs were then taken from the small samples and enhanced to levels examined on the bigger scopes – bottom-up.

The tables show coefficients of variation and standard errors.

See tables attached.

The enterprise population for 2020 LCS has been defined as on state of the Business Register and the sampling part specified on 31^st December 2020. Therefore, the coverage error should be minimised.

As the units are chosen from the Business Register at the end of the reference year, the under-coverage is supposed to be practically none.

See chapter 3.1.

Not provided.

When data are captured, interactive and batch checks are run in an automated way, using pre-determined check links between individual indicators in the questionnaire. The checks are of binding nature (the remedy shall be made in contact with the reporting unit) or informative nature. When an error or an anomaly is revealed, the person responsible for processing contacts the reporting units and corrects data (or confirms anomalies) according to the communication he/she received from the unit. It is requested that the entire questionnaire should be filled in; however, the respondent is allowed to produce an expert guess for some items. In the framework of the corrections, main problems the unit encounters in filling in the questionnaire are kept an eye on, so that the instructions for filling in the questionnaire the following year can be updated.

See below.

For unit non-response rate please see table under point 6.3, where there are response rates (non-response rate is 100% - response rate).

Item non-response rate has not been considered because all items must have been filled in (or it is empty completely).

These LCS data are processed and grossed up to be finally related to all units in the BR population of units classified to the principal activity and size group concerned. Preliminary tables are printed and the processing results are checked.

The data are supplemented and corrected again, if necessary, having been consulted with the reported unit again. After all these verifications and approvals, the results of the survey are considered to be semi-definitive and are to be published. After SBS had been gathered, the processing was made again using new data on number of employees and wages.

No systematic records of the frequency of these mistakes revealed in the process of data capture are kept. However, the mistakes are taken account of when explanatory notes on the individual indicators on the questionnaire are prepared for the following year.

The weights were recalibrated in line with non-response.

For units that did not respond and for which attempts to establish whether or not they were economically active at all were not successful (few cases), the imputation used averages the calculation of which also included inactive units (zero values), so that the degree of uncertainty concerning their economic activity would be taken account of (i.e. it was assumed that the probability of uncertainty corresponded to the share of provably revealed inactive units in the sample in a given imputation group.

Item non-response is not imputed. The questionnaire is checked so that it could be considered as filled in and the unit is contacted (if possible) in the event of incompleteness errors. Should a quantitative indicator remains unfilled in the questionnaire and the questionnaire still meets  prescribed logic checks, the value of such an indicator is considered and processed as a zero value.

Weight linear regression was used for grossing up. The number of enterprises, payroll basis of assessment or number of employees according to BR was used as auxiliary variables. The weights were inverse values of probabilities of including the unit in the sample (Horwitz-Thompson estimator).

See tables attached.

No models were used in LCS 2020 in the CR, because:

―  the tax year did not have to be adjusted to the calendar year, as the survey includes costs charged in the reference period to be paid;

―  data for reporting units with fewer than 10 employees measured on the questionnaire and grossed up to the basic population in the same way as data for the other reporting units;

―  no administrative sources were directly used in the survey.

Not applicable.

Not provided.

Not provided.

Not provided.

The whole process of running the LCS itself for a specific (reference) year takes one calendar year, consequently the data are matched with SBS in order to get final results.

Not provided.

Not provided.

A detailed time schedule of the individual stages of design and programme preparation, verification process and routine processing is given in the technical project to the LCS year T (year 2020). The main phases of the processing were made in accordance with the following deadlines:

― generation of basic population and sample of reporting units 31. 12. 2020

― mailing the questionnaires to reporting units 5. 3. 2021

― delivery of completed questionnaires from reporting units to the CZSO 31. 3. 2021

― ensuring maximal return of completed questionnaires, capture and checks of the questionnaire (carried out by the CZSO nationwide data processing departments) 16. 6. 2021

― loading net data into the CZSO source database and grossing up the data; printing preliminary output tables 30. 6. 2021

― checking and comparing the results of processing with results of other surveys on wages carried out by the CZSO and with LCS results for previous periods; supplementary corrections of data after consultation with the CZSO nationwide data processing division and/or reporting concerned, approval of data: 20. 8. 2021, 21. 9. 2021, 20. 10. 2021.

― publishing national LCS results in Czech and English 16. 11. 2021

― SBS data employed for LCS – 8. 4. 2022

Not provided.

― Definition of statistical units

Unlike the Eurostat methodology, the sample unit is an enterprise and not local unit. Seen spatially, this fact is not significant because regarding the Czech Republic, the whole country corresponds to the NUTS1 level. The enterprises are chosen from the Business Register: those with 250+ employees are taken all and those employing 1 to 249 employees are sampled by a stratified random sampling method. The LCS covers legal persons and natural persons, employing at least one person. Since 2008, the survey also includes the households sector (S14, unincorporated natural persons – entrepreneurs).

The strata are established with reference to economic activity and size of enterprise. Depending on the activity, the strata are broken down by CZ-NACE divisions. As for the size of enterprise (number of employees recorded in the Business Register), the following strata are used: 1-5, 6-9, 10-19, 20-99, 100-249 employees. Census-kind survey for enterprises with 250+ employees was used.

The sample sizes in individual strata differ both absolutely and relatively due to large differences in the number of units in the strata and differences in variability.

― Classifications

As requested NACE rev. 2 were used for the results to be send to Eurostat. Sampling method was also based on the NACE rev.2.

The national Classification of Economic Activities (CZ-NACE), which is fully compatible with the Statistical Classification of Economic Activities (NACE Rev. 2), is used for the national LCS (CZ-NACE 01 to 96). As requested, 2020 results for reporting units with 1+ and 10+ employees in selected economic activities (CZ-NACE 05-96) were transmitted to Eurostat, broken down to correspond to output tables A and B (see Commission Regulation No. 1726/1999 as amended).

― Variables

All of the variables correspond to the requirements of Commission Regulation (EC) No 1726/1999 as amended.

Not provided.

From its very beginning in 1994 the Czech LCS respected the definitions of variables given in relevant regulations of the EU. The bookkeeping records of most reporting units make it possible to quantify detailed items of wage costs, too, which is why there is no need to use estimates for these data. What is not measured in the survey only is data on persons employed by temporary employment agencies. The LCS 2000 was the first to measure data on apprentices. However, the results suggest that this phenomenon is rather insignificant. For this reason, results on apprentices are not published.

[Not requested]

Coherence with Labour costs index (LCI)

The increase in hourly total labour costs between 2016 and 2020 measured by LCS was 40 %, whereas LCI shows 37 %. The coherence between LCI and LCS growths seems relatively poor.

To search for the coherence with SBS is ineffective as the results of both task are finally made upon the same database. See above.

Not provided.

Coherence with National Accounts was not produced as there are large problems with comparability of results because of different use of  methodological concepts. The LCS results are used, among other things, as a data source for national accounts. The methodological concepts of employees, wages, wages in kind, and social security contributions are identical in both the LCS and national accounts. Information on wages and salaries is taken from corporate reporting (statistical or bookkeeping).The LCS data are a source for the in-kind component of wages and imputed social contributions to employees for all sectors and for the number of employees in sectors S13 and S15. The volume of wages and salaries derived from statistical or bookkeeping underlying data was raised by an estimate of wages and salaries intentionally not reported, connected with illicit employment or payments for work outside bookkeeping. The estimate was supplemented with LCS data on part of wages in kind (company cars used for personal needs, discounts on the company’s products sold employees at reduced prices or provided free, and housing contributions). The CZSO National Accounts Department make their own estimates of meals and accommodation provided free of charge or at reduced prices to employees of hotels, restaurants, canteens and in agriculture. Employers’ imputed social contributions (compensation for the time of sickness and other social contributions paid by the employer) are also estimated on LCS data, but corrected in some sectors using corporate reporting data.

Not provided.

Not provided.

1) Electronic publication Labour Costs year (Sample Survey) – code e3112-XX (in Czech and in English version), 110027-XX (since 2014).

Tabulated overviews of hourly, monthly and annual labour costs broken down by CZ-NACE, size of reporting units and region.

2) Electronic publication  Labour Costs – final data (2020) – publication code 110032-22 (in Czech and in English version)

This publication contains revised data of total labor costs. The data are in the same classification as the annual publication in point 1).

Some LCS annual time series data happened to be included also in the:

4) Statistical Yearbook of the Czech Republic

5) Labour Statistics: Time Series of Basic Indicators.

They have been also released on the CZSO web sites for free. No back information to reporting units – participants in the survey – are provided.

Internet:

See the following CZSO web sites for introductory information and selected tables from the publications above:

https://www.czso.cz/  (in Czech)

https://www.czso.cz/csu/czso/home  (in English).

Designed for users are also methodological descriptions of indicators, which contain the name, definition and code of indicator so that the uniqueness of the indicator will be ensured. The descriptions deal with indicators, which are part of the CZSO statistical surveys and used in the CZSO publication systems.

It is allowed to use only the indicators for designing statistical surveys and preparing publications that are kept in the meta-information system SMS. They are available to external users at

http://apl.czso.cz/iSMS/ukazvyb.jsp

Their descriptions are updated to reflect survey-related changes or changes in the contents of statistical outputs.

Not provided.

Not available.

The electronic format is based on Microsoft Office and PDF.

The survey has been caried out according to regulations EC 530/1999 and 1737/2005.

Not provided.

Not provided.

Not provided.

Not provided.

Not provided.

Not provided.

No comments.