Statistics Belgium as part of Federal Ministry for the Economy (FPS Economy, SMEs, Self-Employed and Energy) 

Thematic directorate economy

Boulevard Albert II, 16, 1000 Bruxelles

Structural business statistics (SBS) describes the structure, conduct and performance of economic activities, down to the most detailed activity level (several hundred economic sectors). SBS covers all activities of the business economy with the exception of agricultural activities and public administration. Main characteristics (variables) of the SBS data category:

• Business demographic variables (e.g. Number of active enterprises)

• "Output related" variables (e.g. Net turnover, Value added)

• "Input related" variables: labour input (e.g. Number of employees and self-employed persons, Hours worked by employees); goods and services input (e.g. Purchases of goods and services); capital input (e.g. Gross investments)

Business services statistics (BS) collection contains harmonised statistics on business services. From 2008 onwards BS become part of the regular mandatory annual data collection of SBS. The BS’s data requirement includes variable “Turnover” broken down by products and by type of residence of client. 

The annual regional statistics collection includes three characteristics due by NUTS-2 country region and detailed on NACE Rev 2 division level (2-digits).

Statistical Classification of Economic Activities in the European Community (NACE):  NACE Rev.2 is used from 2008 onwards. Key data were double reported in NACE Rev.1.1 and NACE Rev.2 only for 2008. From 2002 to 2007 NACE Rev. 1.1 was used and until 2001 NACE Rev.1

The regional breakdown of the EU Member States is based on the Nomenclature of Territorial Units for Statistics (NUTS). 

The product breakdown is based on the Classification of Products by Activity (CPA) as stated in the Regulation establishing CPA 2008 and its amending Commission Regulation (EU) No 1209/2014 (from reference year 2015 onwards).

Starting reference year 2021 onwards SBS cover the economic activities of market producers within the NACE Rev. 2 Sections B to N, P to R and Divisions S95 and S96.
Until 2007 the SBS coverage was limited to Sections C to K of NACE Rev.1.1 and from the reference year 2008 to 2020 data was available for Sections B to N and Division S95 of NACE Rev.2.

From 2008 reference year the data collection Business services covers NACE Rev 2 codes: J62, N78, J582, J631, M731, M691, M692, M702, M712, M732, M7111, and M7112.
From 2013, as the first reference year, to 2020 information is published on NACE codes K6411, K6419 and K65 and its breakdown.

SBS constitutes an important and integrated part of the new European Business Statistics Regulation N° 2152/2019

Data requirements, simplifications and technical definitions are defined in Commission Implementing Regulation (EU) 2020/1197

Until reference year 2017, the enterprise statistical unit was the legal unit. Since 2018, statistical unit is in line with the Regulation (ECC) No 696/1993 and may consist out of one or more legal units.

- NACE sections B to N, P to R and Divisions S95 and S96 - Enterprises active in Belgium and foreign enterprises with a sign of activity in Belgium (employment, VAT turnover, annual accounts, ...) are included in the sample and report only for activities in Belgium.

- All size classes are covered

- The frame for SBS statistics is the business register, all data is provided on the enterprise level.

The reference area is the whole territory of Belgium.

1996-2021

Data are comparable between 1996 and 2007 and between 2008 and 2017. Until 2007, NACE Rev.1 is used. From 2008 onwards, results are computed according to NACE Rev.2. Until reference year 2017, the enterprise statistical unit was the legal unit. Since 2018, statistical unit is in line with the Regulation (ECC) No 696/1993 and may consist out of one or more legal units.

Data for sectors K, P to R and S96 are collected and published from reference year 2021.

Not applicable.

• Number of enterprises and number of local units are expressed in units.

• Monetary data are expressed in millions of €.

• Employment variables are expressed in units.

• Per head values are expressed in thousands of € per head. 

Ratios are expressed in percentages.

2021

In most cases, all data refer to calendar years. In cases, in which the source data used for compiling the data of the variable are only available for the fiscal year, and those data cannot be recalculated to cover the calendar year, the calendar year data may be approximated by data on the fiscal year for those statistical units.

Starting with reference year 2021 two new regulations currently form the legal basis of SBS:

• Regulation (EU) 2019/2152 repealing 10 legal acts in the field of business statistics (EBS Regulation), and
• Commission Implementing Regulation (EU) 2020/1197 laying down technical specifications and arrangements (general implementing act) pursuant to Regulation (EU) 2019/2152

Year 1995 was the first year for the implementation of the Council Regulation No 58/97 (SBS Regulation).

The Council Regulation No 58/97 has been amended three times: by Council Regulation No 410/98, Commission Regulation No 1614/2002 and European Parliament and Council Regulation No 2056/2002. As a new amendment of the basic Regulation it was decided to recast the Regulation No 58/97 in order to obtain a new "clean" legal text. In 2008 the European Parliament and Council adopted Regulation No 295/2008 and the provisions of this Regulation were applicable from the reference year 2008 to reference year 2020. Regulation No 295/2008 was amended by Commission Regulation (EU) No 446/2014.

Not applicable.

Regulation (EC) No 223/2009 of the European Parliament and of the Council, of 11 March 2009, on the transmission of data subject to statistical confidentiality to the Statistical Office of the European Communities.

Belgian Law on public statistics of 22 March 2006.

 The anonymisation method has not changed from previous years : 

Primary confidentiality calculated with SAS.

- Minimum frequency rule: if the number of enterprises in the cell is smaller than 3 --> Confidential

- Dominance rule: if one enterprise represents more than 85% of the weighted turnover (variable 250101) in the cell --> Confidential

Secondary confidentiality calculated with Tau-argus (linked tables approach when needed, modular approach for non-linked tables) 

The structural business survey is listed in the global publication calendar of Statistics Belgium.

https://statbel.fgov.be/en/calendar

Aggregated results which are not confidential are published on the website. Full aggregated results (including confidential cells) are transmitted to some public authorities (--> Institute of National Accounts, Belgian National Bank) and a limited set of registered main users on basis of confidentiality agreements.

Annual.

No press release concerning the results of SBS data is issued.

Pdf publications in French and Dutch -> October 2023

Cover_FR_19_Structuur_2020 (fgov.be)

Statistical yearbook -> end 2023

https://statbel.fgov.be/sites/default/files/files/documents/F_Kerncijfers_web_v2.pdf

SBS results are published on an annual basis in the Belgian Statistical DataWareHouse (BeStat) and can be accessed from https://statbel.fgov.be/fr/themes/entreprises/statistiques-structurelles-sur-les-entreprises

Online database: https://statbel.fgov.be/fr/themes/entreprises/statistiques-structurelles-sur-les-entreprises#figures

Microdata of the structural business survey can be obtained through a confidentiality agreement with Statistics Belgium provided that the applicant fulfils the provision of the Belgian Law on public statistics of 22 March 2006.

Data are sent to Eurostat, either to be used in European aggregates and to be released also as national data.

The data are available on the Eurostat website: https://ec.europa.eu/eurostat/web/structural-business-statistics

Not available publicly.

Not available publicly.

Belgium has been providing Eurostat with regular (annual) quality reports covering most of the categories of the ESS Standard for Quality Reports. Eurostat prepares a summary quality reports which is discussed in a yearly meeting with Member States.

According to the information available from the quality reports, the data providers have applied the recommendations available in the EU Regulations.

As we do every year, we carry out manual checks on values that have exceeded certain rates of change compared with previous years. These rates vary depending on the variables. By way of example, for turnover, the suitable standard interval is [82%;122%]. The other variables checked as a priority are the number of enterprises, the number of employees and persons employed, and the value added. 

More information is available in the Report from the Commission to the European Parliament and the Council on the implementation of the SBS Regulation.

Institute of National Accounts (http://inr-icn.fgov.be/nl)

National Bank of Belgium (http://www.nbb.be)

• survey values from the structural business survey
• yearly
• input for national and regional accounts (I.N.R.)

Federal Planning Bureau (http://www.plan.be)

• survey values from the structural business survey
• each 5 years
• for input-output tables

Price observatory (https://economie.fgov.be/fr/propos-du-spf/organisation/observatoires/observatoire-des-prix)

• survey values from the structural business survey
• yearly

National Bank of Belgium and Free University of Brussels (DULBEA - Department of Applied Economics - https://dulbea.ulb.be)

• ratio’s from SBS variables
• each 2 years
• for research project concerning wages structure in Belgium
• with confidentiality agreement

A general satisfaction survey is organised by Statistics Belgium (see https://statbel.fgov.be/fr/propos-de-statbel/qualite/enquete-de-satisfaction). There is no specific survey for the structural business survey.

Data is complete.

The overall accuracy is sufficient. 

As in previous years, there are some reasons inherent to the survey which may lead to certain biases or accuracy problems : 

• sampling and modelling errors can occur because of the rotating model used to draw the sample (see 13.3 and 16). For small businesses that are not surveyed every year, the evolution of the various variables is estimated on the basis of the evolution of VAT data. These estimates have limits in terms of accuracy, since VAT data may be not perfectly correlated with the other variables. 
• some errors on the NACE codes in the business register can also lead to misclassification in the sampling procedure : if a company selected in the sample for a NACE class turns out to be misclassified, the number of companies actually surveyed for the NACE class in question inevitably falls, leading to a reduction in representativeness. 
• for provisional data, annual accounts, VAT data and data of the previous years are used to compute the results, which can lead to some bias. However, provisional data are not disseminated by Statistics Belgium.

Coefficients of variation are calculated and transferred in different files. The coefficients of variation are calculated using the POULPE macro in SAS.  The aspects of the survey taken into account are the sample design (stratified simple random sampling), non-response correction and calibration.

The sampling error can be considered small, in 93% of the cases, the coefficient of variations are smaller than 10%.

The following methods are applied to take into account unit non-response:

• imputation: ratio imputation or hot deck imputation
• reweighting

Furthermore, the following measures are taken to minimize non-response:

• The survey is compulsory (Belgian Royal Decree of 18 July 2008)
• Three reminder letters are sent
• Telephone calls and visit are made to large enterprises

In order to calculate the non-response we took only the active and eligible enterprises into account (out-of-scope enterprises have been excluded). We weighted the unit non response rate with the number of persons employed. The overall weighted non-response rate is 5.00 %. Compared to the unweighted non-response rate of 16.36% this means that larger enterprises (in terms of employment) report better. In the overall contest, we evaluate this non-response rate as low.

For reference year 2021 the smaller entities from the trade and services sectors have been estimated using data on the same units from other sources. The use of past survey data and recent administrative data from several sources can introduce some bias but this is minimized through plausibility controls and checks.
Moreover, the estimation method concerns smaller enterprises. Therefore we assume that the overall effect on the accuracy of the estimates is limited.

The survey unit non-response may be not random. The potential bias is minimized by using imputations (if possible) for the non-responding enterprises. Moreover, the survey unit non-response is low, especially for the larger units. So we believe there is only a small potential bias resulting from survey non-response.

Errors in the business register can have an effect on sampling and on sampling weights. The exact impact on SBS results is unknown. Errors in addresses of statistical units in the business register also causes a coherence problem between national and regional results. The sum of the regional results is not identical to the national value, but the difference is small in most cases.

1% of the units initially included in the sample are deemed to be out of scope. These units are identified on basis of administrative data or after contact during the survey.

The data collection ended in April 2023. The post-collection phase ended in June 2023 and the results will be published in October 2023.

The data of 2021 were delivered few days after the transmission legal deadline.

The same statistical concepts are applied across entire national territory.

Data are comparable between 1996 and 2007 and between 2008 and 2017. Until 2007, NACE Rev.1 is used. From 2008 onwards, results are computed according to NACE Rev.2. Until reference year 2017, the enterprise statistical unit was the legal unit. Since 2018, statistical unit is in line with the Regulation (ECC) No 696/1993 and may consist out of one or more legal units.

We compared the characteristics Number of active enterprises (210101) and Number of employees and self-employed persons (220101) of Annex IX with the same characteristics (210101 and 220101) of Annex I-IV. Before 2021 some rather small inconsistencies due to differences in method (sample selection and grossing up versus full coverage using administrative records, activity criteria, delineation of non-market enterprises, etc.) existed.

By definition, the two statistics SBS and BD cannot have the same results, the first being the result of a survey, the other the result of administrative sources. For the same reasons, SBS and the national accounts may also diverge, since the national accounts are based in part on administrative sources. Since 2021, SBS and BD are now required to have the same figures for the number of businesses and employment, so adjustments have been made to BD to match the SBS figures exactly. Between the survey and the administrative sources, it is the survey that prevails. 

There may also be discrepancies between SBS and other surveys: the reason for this lies in NACE coding errors. As mentioned in 13.1, companies selected from the SBS sample as belonging to a NACE sector may report another main activity. These "assignment errors" imply adjustments in the results and in their weight in the universe: statistical calibration, specific to each sample. Discrepancies with another survey (e.g. LFS) are therefore inevitable. 

Comparison of variable Net Turnover (250101) with turnover statistics based on VAT declarations. Some inconsistencies due to difference in definitions and in methods.

Regional results do not add up to national results because some local units have no address in our business register and can therefore not be allocated to a NUTS-group.

The cost for Statistics Belgium can be estimated to 10000 man-hours for annex 1-4 and 8.

The time needed to fill in the questionnaire is estimated to 100 minutes (on average) by enterprise.

Statistics Belgium aim at reducing the burden on respondents by using administrative data (annual financial statement, VAT data, statistical business register, social Security database) to prefill the questionnaires that enterprises have to fill. A rotation system is also used for smaller enterprises in order to not select the same enterprises every year.

Revision if major errors in the data are detected. See also https://statbel.fgov.be/en/about-statbel/quality/revision-policy for the general revision policy applied by Statistics Belgium.

In recent years we did not revise the data.

Since we prefill the annual account data in the survey forms, we only send out the survey forms when annual account data are largely available (T + 9). Since only a fraction of the survey forms are returned at T + 10 we rely mostly on administrative sources (annual account data, VAT data) for producing preliminary data. For enterprises with no administrative data available, we use values from the preceding years or ratio imputation based on prior survey data and fresh administrative data.

At t+10 administrative data are already of good quality (especially social security data and VAT data; annual accounts data are almost complete).

For employment variable Number of employees and self-employed persons (220101) administrative data from National Social Security Database is our main source for both definitive and preliminary data. (At T + 10 we have almost no detailed information on the number of unpaid workers, but we use a proxy for the variable at this stage).

a) Survey

The sample is divided into two parts with a different methodology

• Surveyed enterprises --> PART 1
• Model based estimations for non surveyed enterprises --> PART 2

For both parts, the sample is stratified according to employment size class and turnover size class.

The sampling rates are the following: the total sample (parts 1 and 2) include 6% of the enterprises in the population, 76% of employment and 82% of turnover. Enterprises with more than 50 employees are exhaustively selected.

A minimum value of annual turnover is used as threshold and is set to 5,600 euro (according to VAT records).

The effective sample sizes for part 1 and 2 are respectively equal to 35000 and 15000 enterprises.

b) Administrative source

The following administrative sources are used for surveyed enterprises:

• Annual Financial Statement of enterprises (full or abbreviated model): balance sheet, profit and loss account, social balance sheet
• Statistical business register
• Social Security database
• VAT Administration Records: VAT flows supplier-customer

For non surveyed enterprises for which model based estimations are used, the following sources are used:

• Annual Financial Statement of enterprises (full or abbreviated model): balance sheet, profit and loss account, social balance sheet
• Statistical business register
• Social Security database
• VAT Administration Records: yearly data on turnover, purchases, investments, VAT flows supplier-customer and exportations

The administrative data described above are used for the following purposes for surveyed enterprises:

• data source, basic data for some characteristics (prefill in the websurvey)
• data source for imputation in case of non-response.

For non surveyed enterprises, the administrative data are used for the same purposes. In addition, these data are also used for imputation or for strata not covered by the survey.

We access micro data for the entire population. The frequency to which the used administrative data sources are updated is considered to be good and subject to several revisions with (increasing) degree of completeness.

c) Frame

The variables used for identifying principal and secondary activities are:

• Turnover from the SBS survey
• Employment (NACE from social security records)
• Value added (NACE from VAT records)

The top-down method is used to identify principal activities. The unit's principal activity is reviewed annually. The business register is updated weekly.

Annual data collection.

The SBS sample selection is carried out in August N+1. Letters are sent to the enterprises begin October. It is possible to answer via internet or on paper. The questionnaires are required to be returned to Statbel before the end of the year. The survey is mandatory. If businesses who have received questionnaires have not responded by the deadline, up to three reminder letters can be sent. Repeated telephone contacts are used to improve response rates. The number of units surveyed is about 35.000. Statbel supplements the collected data with administrative data from other administrations.

The following procedures are used to check the data:

• completeness checks (data integrity rules)
• validity checks (internal consistency)
• plausibility checks
• Controls of values from the SBS survey:

At data entry in the websurvey: data type (numerical or character field), compulsory fields, business checks (checking the logical consistency between survey items).
After data entry: controls of microdata: identifying potential problems (errors, discrepancies, outliers, item non-response, miscoding, …). Applying automatic edits and manual inspection and correction of the data.
Controls of the imputations: ratio checking (negative ratios or excessively high ratios), consistency checks, comparison with source data,..

Controls of the employment variables: comparison of social security data with data from the social balance sheet, internal consistency checks on social security data.

Macro controls of aggregated data: year-to-year variation, ratio checking, consistency checks.

The following methods have been applied to take into account non-response:

• imputation based on administrative sources (if possible)
• data from previous reference period
• hot deck imputation
• if imputation is not possible, a corrector factor is used in the weighting procedure.

The following methods have been applied for grossing-up the figures covered by the SBS Regulation in order to cover the entire population of enterprises:

The estimation based on a stratified sample. The initial sample weights are set for the different strata: number of enterprises in population/number of enterprises in the sample. The non responding enterprises are assigned to different status categories depending on the reason and components of the non-response (e.g. non contact of the selected units, refusal to participate, not eligible, cessation of activities during the reference year, cessation of activities after the reference year, …). The original sampling weights are adapted (with corrector factor) using these status categories.

There is no correction to bring the data in accordance with the reference period if the reference period differs from the calendar year.

No further comments.