Reference metadata describe statistical concepts and methodologies used for the collection and generation of data. They provide information on data quality and, since they are strongly content-oriented, assist users in interpreting the data. Reference metadata, unlike structural metadata, can be decoupled from the data.
Trade, Transport, Services, Tourism, and Environmental Statistics Department (52)
1.3. Contact name
Confidential because of GDPR
1.4. Contact person function
Confidential because of GDPR
1.5. Contact mail address
Czech Statistical Office Na padesátém 81 100 82 Praha 10 Czech Republic
1.6. Contact email address
Confidential because of GDPR
1.7. Contact phone number
Confidential because of GDPR
1.8. Contact fax number
Confidential because of GDPR
2.1. Metadata last certified
13 May 2025
2.2. Metadata last posted
13 May 2025
2.3. Metadata last update
13 May 2025
3.1. Data description
Retail trade turnover, excluding VAT.
Wholesale trade turnover, excluding VAT.
Turnover is published in the form of an index (not absolute figures) as an indicator of the economic cycle.
3.2. Classification system
NACE Rev. 2.
3.3. Coverage - sector
Activities covered: NACE Rev. 2 Division 45, 46 and 47. No size class is excluded from the survey.
3.4. Statistical concepts and definitions
Retail and whlesale trade turnover as defined in EC 1503/2006. Turnover excluding VAT - comprises invoices issued by the observation unit during the reference period. Turnover refers to flows during the calendar month.
3.5. Statistical unit
Enterprise. KAU from 1/2021.
3.6. Statistical population
All market enterprises with main activity in NACE Rev. 2 Section G. All size classes are included. The frame is the business register, which is regularily updated with demographic changes.
An overview on the structure of the population below represents the number of active units:
NACE 45 - 38 392 units.
NACE 46 - 67 356 units.
NACE 47 - 117 760 units.
3.7. Reference area
Czech Republic
3.8. Coverage - Time
Year 2000 - present, except NACE 46 which is only year 2015 -present.
3.9. Base period
For processing the year 2023, we used the base year and reference year 2015.
From March 2024 onwards (along with the publication of data for January 2024), the Czech Republic and Eurostat have published indices with base year 2021. All time series were recalculated (rescaled) to base year 2021 according to the release dates published in the release calendar. The base period for fixed-base indices changed from the average of the year 2015 to the average of the year 2021. New weighting schemes are used that are derived from results of the structural business statistics for the year 2021. In compliance with the Eurostat manual on short-term statistics, new weights were used for recalculation of data back to the year 2020, which entailed a revision of data for 2020–2023. Data for previous years were connected with updated results by means of conversion bridges by the annual overlap method.
We do not publish different base and reference periods nationally.
Index (%).
Calendar month.
6.1. Institutional Mandate - legal acts and other agreements
Act No. 89/1995 Coll. on the State Statistical Service of June 15, 1995, amended by Act No. 220/2000 Coll. and Act No. 411/2000 Coll.
The European Parliament and Council adopted the European Business Statistics (EBS) Regulation (EU) 2019/2152 on 27 November 2019, followed by the Commission Implementing Regulation 2020/1197 laying down technical specifications and arrangements pursuant to the mentioned EBS Regulation (General Implementing Act).
Data sharing with international organizations: Eurostat, OECD.
7.1. Confidentiality - policy
The principle of protection of individual data is assured by the law on protection of personal data and the Act No. 89/1995 Coll. as amended, which implies an active approach to confidentiality. The detailed rules are specified in the internal CZSO Directive No. 1/2022, which content is seen as confidential.
7.2. Confidentiality - data treatment
Data are only published down to an activity and size level at which they are not confidential. Data allowing direct or indirect identification of individual enterprise can only be published under written permission of all related subjects. The parameters for primary confidential data determination are confidential.
8.1. Release calendar
The Catalogue of Products containing Calendar of News Releases for the year is published at the end of the preceding year. The data are part of Retail Trade News Release and Retail Trade Time Series.
8.2. Release calendar access
Release calendar is available at the CZSO website.
Data are released, according to the Release Calendar, at 9 a.m. on the CZSO web site; no external user has prior access to the data before general release. Data are available at the same time for government and for public use; there is no pre-release access to data.
Data are regularly transmitted to Eurostat, shortly after national News Releases.
Transmission to Eurostat: monthly.
National dissemination: monthly News Release and time series.
10.1. Dissemination format - News release
Monthly News Releases (Retail Trade in Czech and English) - commentary on observed development accompanied by tables and graphs. They are published on the website of the Czech Statistical Office.
Data are available in our DataStat > Sectors > Trade > Short-term statistics on trade.
Tables produced:
Monthly value and volume year–on-year indices (neither seasonally adjusted nor calendar adjusted data, calendar adjusted data, calendar and seasonally adjusted).
Monthly value and volume indices (data 2021 = 100) (neither seasonally adjusted nor calendar adjusted data, calendar adjusted data, calendar and seasonally adjusted).
Level of breakdown: 2-digit, selected 3-digit and selected 4-digit headings of CZ-NACE 47, 3-digit headings od CZ-NACE 45 and 46.
According to the national legislation the access to the microdata is allowed for scientific purposes only. Each request is decided on individual basis by the relevant expert group.
10.5. Dissemination format - other
Time series are sent to Eurostat. Special data formats are available to users upon request.
10.6. Documentation on methodology
An elementary description of methodology is given on the CZSO website.
The whole process of data production is driven by internal documentation (Technical projects with programming specification and metadata in the Statistical Metadata System) to which CZSO employees have access.
Audit reports including all the test and analysis carried out within the audit are stored and the public parts are available on request (see also quality management).
11.1. Quality assurance
The key components of the quality assurance are the following:
Strategic documents of the CZSO: Mission, Vision and Priorities of the CZSO, Quality commitment, Priority tasks defined for each year
Irregular methodological audits represents an important tool in quality control. Audits are usually carried out by expert group comprising of internal and external experts (NA experts, representatives of universities, ministries etc.).
Important aspects and procedures of quality management within statistical production process are documented in by internal Statistical metadata system, including so-called Technical projects, event. by instructions for surveyors that are available for every statistics produced.
The indicator is relevant to various group of users.
Concerning accuracy, some problems are caused by misclassification; response rates are reasonable, especially at the time of revisions.
Release deadlines allow to meet requirements of both EU legislation and other users' requirements. All News releases are published on time.
There are no major problems with comparability and coherence of the indicator.
12.1. Relevance - User Needs
External users: Eurostat, OECD, Ministry of Industry and Trade, Czech National Bank, professional associations, media, general public.
Internal users: Quarterly National Accounts Department.
The CZSO primarily tries to fulfill its commitments towards the European Commission, however, generally we tries to deal with the requirements of all users equally.
Unmet user needs: Users would like to have more detailed data, i.e. data on sales by 4-digit CZ-NACE activity or even more detailed, geographical breakdown, absolute values of an indicator. These requests are not satisfied due to survey methodology, capacity and financial limits.
12.2. Relevance - User Satisfaction
No users satisfaction surveys are carried out, however, we are constantly in contact with different groups of users, e.g. ministries, Czech National Bank, different associations (e.g. Association of small and medium sized Enterprises and Crafts of Czechia), internet users etc.
12.3. Completeness
Czechia provides Eurostat with all data required by the STS Regulation.
13.1. Accuracy - overall
The sampling error in the data is not systematically measured, but believed to be negligible. The non-sampling errors are small and are mainly due to misclassification and other problems among small units. The first estimates are generally not biased. The results are final after the annual revision (in March of the following year).
13.2. Sampling error
The sampling error is estimated to be relatively small. For CV´s see section 13.3.
13.3. Non-sampling error
General problems:
Overcoverage and undercoverage is caused by high degree of misclassification of small units, especially natural persons and legal persons without employees, which are not VAT payers.
Non-response - see below for details. Non-responding units are re-contacted several times by e-mail, phone or mail.
Short deadlines sometimes force the respondents to submit data from unfinished book-keeping or estimates, which then leads to later revisions.
Demographic changes (mergers, splits, births and deaths) in reporting units and in the population as a whole are reported late.
Size-weighted unit reponse rate in 202409 for NACE 47 is 86,3, for NACE 45 is 94,8.
Coefficient of variation in 202409 for NACE 47 is 0,1143, for NACE 45 is 0,3807.
14.1. Timeliness
Data are published approximately 38 calendar days after the end of the reference month. Data for the reference month are provisional. Final data for the whole calendar year are published in March of the following year.
All questionnaires for the whole year are sent to the reporting units at the beginning of the year (January). Units are required to reply by the 20th calendar day after the end of the reference month. The survey (for the year) is closed approximately three months after the end of the reference year.
14.2. Punctuality
Releases delivered on time: 100 %.
15.1. Comparability - geographical
Data are fully comparable.
15.2. Comparability - over time
Time series are comparable over time.
15.3. Coherence - cross domain
Data from the monthly survey can be compared with the results of the structural business survey (SBS). Differences are due to methodology, mainly because the monthly survey focuses on the y-o-y index.
Coherence is assisted by use of the Business Register and a database coordinating main activity of enterprises over all surveys.
15.4. Coherence - internal
There are no problems regrading internal coherence of the time series in question.
NACE 45 and 47:
Turnover is surveyed by the questionnaire SP 1-12 (common with services). The questionnaire is sent to a sample of aproximately 1,4 thousand retail businesses. Respondent burden for retail enterprises is estimated on 1,8 thousand hours per year.
NACE 46:
Administrative data – value added tax (VAT) tax returns are the data source. Based on administrative data, total sales from sale of goods, services, and goods for resale excluding VAT in the relevant month are modelled. Therefore, there is no burden on respondents.
The cost of CZSO for the indicator are estimated as 5,7 thousand hours per year.
17.1. Data revision - policy
The system of regular revisions is briefly described on the CZSO website. Revision calendar is published on the CZSO website. Revised data are published on the same dates as published data.
Our regular revisions are:
Monthly Revisions: With the release of data for the current month, the data for the previous month are revised (for example: when data for February of the year Y are released data for January of the year Y are revised).
Annual Revisions: Annual revision of monthly estimates for all reference periods of the last elapsed year (Y) is released along with the appropriate News Release for January of the current year. The revisions are based on further knowledge and more accurate data acquired during the year. Data released in this revision are considered to be finalised data.
The same revision policy (for all types of revision: routine revisions, major revisions, non-scheduled revisions) is applied to data released nationally and transmitted to Eurostat.
Irregular revisions (such as changing of the processing program or correction of errors) are announced at the time of release.
Methodological changes are announced at the time of the change, major changes in advance.
The system of regular revisions is briefly described on the CZSO website. Revision calendar is published on the CZSO website. Revised data are published on the same dates as published data.
In accordance with the CZSO Data Revision Policy, standard revisions are made with the aim to enhance quality of a previously published piece of data by inclusion in the processing the reports (questionnaires) that were delivered late, corrections of data that were misreported, and by inclusion of updated administrative data. Aggregates and components are revised at the same time.
Our regular revisions are:
Monthly Revisions: With the release of data for the current month, the data for the previous month are revised (for example: when data for February of the year Y are released data for January of the year Y are revised).
Annual Revisions: Annual revision of monthly estimates for all reference periods of the last elapsed year (Y) is released along with the appropriate News Release for January of the current year. The revisions are based on further knowledge and more accurate data acquired during the year. Data released in this revision are considered to be finalised data.
During the processing of the revision, we analyze the impact of the revision. Revisions are not expected to cause time series break.
The same revision policy (for all types of revision: routine revisions, major revisions, non-scheduled revisions) is applied to data released nationally and transmitted to Eurostat.
Irregular revisions (such as changing of the processing program or correction of errors) are announced at the time of release.
Methodological changes are announced at the time of the change, major changes in advance.
The revisions between preliminary and definitive data for section G do not exceed 1 percentage point for most months.
TOVT: MAR = 0,3145 MR= -0,0810
TOVV: MAR= 0,3150 MR= -0,0662
18.1. Source data
Type of source: combination of Statistical survey (SP 1-12) and model based on administrative data source (VAT)
Frame on which the source is based: Business Register (RES).
We sticked with Statistical survey (SP 1-12) for several activity (CZ NACE) for which model based on VAT didn‘t give good results: 45.3, 47.5, 47.7, 47.8
Stratified sample: stratification by main activity (CZ NACE) and number of employees, proportional-to-sales sampling.
Employee size classes: 0-19, 20-49, 50-99, 100 and more.
For other CZ-NACE activity we use combination of Statistical survey (SP 1-12) and model based on administrative data source (VAT).
For each CZ-NACE activity we have determined threshold: all units above this threshold are included in sample of Statistical survey (SP 1-12)
Remainder is modeled on base of VAT - only monthly payers are used.
Frequency of updating the sample: A new sample is drawn every year and updated every month (only exhaustive strata).
18.2. Frequency of data collection
Data for the current month of the current year are collected monthly. Data for all months of the previous year are collected once at the beginning of the survey of the reference year for reporting units that were not included in the previous year’s sample (or that did not reply for the whole previous year).
Data for NACE Rev. 2 Division 46 are completely based on administrative data source (VAT) and for most units are collected monthly, for some quarterly.
Collected data refers to the full reference period.
18.3. Data collection
Questionnaire SP 1-12 is sent to all selected reporting units in NACE division 45 and 47 and selected service activities (units that are selected for the first time in the sample or those that did not reply in the preceding year receive an additional annex – Entry Questionnaire). Data for NACE Rev. 2 Division 46 are completely based on administrative data source (VAT).
18.4. Data validation
The electronic form of questionnaire contains built-in logical controls that draw respondent's attention to possible errors. Moreover, after receiving, data collected via electronic form of the questionnaire are further validated by statisticians. Data obtained via the postal form of the questionnaire are entered to the database using specific software (DANTE) with built-in logical controls. When the DANTE software indicates a possible error, respective statistician entering data to the database contacts the respondent with the aim to correct/validate data. Further plausibility and consistency checks are performed by statisticians during data compilation.
18.5. Data compilation
Missing data for the previous year are estimated from administrative data, missing data for the current year are calculated by the average year-on-year development in the stratum to which the unit belongs.
Turnover is grossed-up to the population using imputation. Grossed-up results are used only to calculate indices.
The year-on-year index is calculated from grossed up results and then chained to the base year. The methodology is based on a Laspeyres index. Turnover of the base year is used as the weight for each strata.
The files are converted to SDMX using internally developed application.
18.6. Adjustment
The Tramo/Seats method implemented in software JDemetra+ version 2.2.2. is used for seasonal adjustment. The CZSO uses manual selected ARIMA models.
Model are regularly reviewed once a year with a revision of the year. Values of parameters are recalculated every month and outliers are identified in the newly added observation.
Each month, when publishing adjusted time series, data from January 2021 have been recalculated backwards. Final non-adjusted data are published following the processing of all months of the reference year, in March of the following year at the latest. Together with a revision for a year, an adjustment model is changed and therefore the whole time series is recalculated.
The CZSO uses Arima and multiplicative seasonal adjustment decomposition. The critical value for outlier detection is 3,5. The filter lenght is automatically chosen. There is no seasonal breaks in series.
Seasonal and working day adjustment is performed on the following series:
We use the indirect method for seasonal and working day adjustment. We calculate other series by adding the values of their components (eg G471 is formed by the sum of G4711 and G4719).
Residual seasonality is checked automatickly by JDemetra+.
Consistency among the different levels of breakdown is carried out by indirect method.
Turnover is published in the form of an index (not absolute figures) as an indicator of the economic cycle.
13 May 2025
Retail and whlesale trade turnover as defined in EC 1503/2006. Turnover excluding VAT - comprises invoices issued by the observation unit during the reference period. Turnover refers to flows during the calendar month.
Enterprise. KAU from 1/2021.
All market enterprises with main activity in NACE Rev. 2 Section G. All size classes are included. The frame is the business register, which is regularily updated with demographic changes.
An overview on the structure of the population below represents the number of active units:
NACE 45 - 38 392 units.
NACE 46 - 67 356 units.
NACE 47 - 117 760 units.
Czech Republic
Calendar month.
The sampling error in the data is not systematically measured, but believed to be negligible. The non-sampling errors are small and are mainly due to misclassification and other problems among small units. The first estimates are generally not biased. The results are final after the annual revision (in March of the following year).
Index (%).
Missing data for the previous year are estimated from administrative data, missing data for the current year are calculated by the average year-on-year development in the stratum to which the unit belongs.
Turnover is grossed-up to the population using imputation. Grossed-up results are used only to calculate indices.
The year-on-year index is calculated from grossed up results and then chained to the base year. The methodology is based on a Laspeyres index. Turnover of the base year is used as the weight for each strata.
The files are converted to SDMX using internally developed application.
Type of source: combination of Statistical survey (SP 1-12) and model based on administrative data source (VAT)
Frame on which the source is based: Business Register (RES).
We sticked with Statistical survey (SP 1-12) for several activity (CZ NACE) for which model based on VAT didn‘t give good results: 45.3, 47.5, 47.7, 47.8
Stratified sample: stratification by main activity (CZ NACE) and number of employees, proportional-to-sales sampling.
Employee size classes: 0-19, 20-49, 50-99, 100 and more.
For other CZ-NACE activity we use combination of Statistical survey (SP 1-12) and model based on administrative data source (VAT).
For each CZ-NACE activity we have determined threshold: all units above this threshold are included in sample of Statistical survey (SP 1-12)
Remainder is modeled on base of VAT - only monthly payers are used.
Frequency of updating the sample: A new sample is drawn every year and updated every month (only exhaustive strata).
Transmission to Eurostat: monthly.
National dissemination: monthly News Release and time series.
Data are published approximately 38 calendar days after the end of the reference month. Data for the reference month are provisional. Final data for the whole calendar year are published in March of the following year.
All questionnaires for the whole year are sent to the reporting units at the beginning of the year (January). Units are required to reply by the 20th calendar day after the end of the reference month. The survey (for the year) is closed approximately three months after the end of the reference year.
