Business statistics

Name: Business statistics
Topic: Establishments, enterprises and accounts

Division for Structural Business Statistics

The figures are published at the national level. For final figures, some variables at county and regional level are also published.

The statistics are published annually. Preliminary figures are published approximately 10 months after the end of the statistical year, while final figures are published approximately 16 months after the end of the statistical year.

The statistics are reported to Eurostat.

Collected and revised data are stored securely by Statistics Norway in compliance with applicable legislation on data processing.

Statistics Norway can grant access to the source data (de-identified or anonymised microdata) on which the statistics are based, for researchers and public authorities for the purposes of preparing statistical results and analyses. Access can be granted upon application and subject to conditions. Refer to the details about this at Access to data from Statistics Norway.

The business statistics provide detailed information on the activity in industries based on accounting figures and questionnaires.

The statistics adhere to Regulation (EU) 2019/2152 of 27 November 2019 on European business statistics (EBS).

Data on mining and quarrying (excluding industries 06 and 09.1) and manufacturing were first published in 1876.

Data on wholesale and retail trade; repair of motor vehicles and motorcycles" were first published in 1956.

Data on construction were first published in 1966.

Data on administrative and support service activities and real estate activities were first published in 1979.

Data on sewerage and remediation activities were first published in 1995, and water supply and materials recovery were included in 2008.

Data on other service activities were first published in 1995.

Data on parts of the information and communication, transportation and storage sectors, as well as accommodation and food service activities, were published in 1998, and final figures at both enterprise and KAU level were first produced in 1999.

From 2019, all of the above-mentioned statistics have been published under the title "Business statistics".

Data on education, human health and social work activitities and arts, entertainment and recreation, are published from year 2021 and onwards.

Data on activities of head offices, are published from year 2022 and onwards.

The business statistics are produced to meet the requirements of the EU’s regulation on structural statistics, based on the need for data in the national accounts. In addition, ministries and various trade associations request figures in this area. The business sector and other users also pay to receive specially adapted business statistics.

No external users have access to statistics before they are released at 8 a.m. on ssb.no after at least three months’ advance notice in the release calendar. This is one of the most important principles in Statistics Norway for ensuring the equal treatment of users.

The data collected for the business statistics includes information on employment. In these statistics, both full-time and part-time employees are counted as employed, and the figure represents an average for the year. Employment data is also found in the National Accounts and in other employment statistics.

While the business statistics can provide employment figures at a detailed industry level, and are consistent with other KAU-related variables, the National Accounts provide quarterly figures with a short production time. The Labour Market Survey and the register-based employment statistics also provide figures for personal identifiers such as age, sex, education and working hours. The register-based employment statistics also give figures down to municipality level.

In comparisons with other employment statistics, users should be aware that definitions and methods differ slightly between the various statistics, and that the figures will therefore vary. The business statistics apply the definitions stipulated in the EU’s regulation on European Business Statistics (EBS).

In addition to the business statistics, quarterly turnover statistics are also published.

The statistics are developed, produced and disseminated pursuant to Act no. 32 of 21 June 2019 relating to official statistics and Statistics Norway (the Statistics Act).

Regulation (EU) 2019/2152 of 27 November 2019 on European Business Statistics (EBS). In addition, Norway has, through the EEA agreement, undertaken to comply wit the European Parliament and Council Regulation (EC) No 1893/2006 and Commission Delegated Regulation (EU) 2023/137 establishing the statistical classification of economic activities NACE Revision 2.1.

All enterprises in the main industry classifications B-K, M-O and Q-T should normally be included in the statistics if they are registered with active business in Norway during the statistical year.

The EU’s regulation on structural statistics primarily requires statistics at the enterprise level. Statistics are compiled at the KAU level for the National Accounts and other Norwegian users. It is worth noting that the KAU-based figures show different values to the enterprise-based figures. This is because enterprises are placed in the industry subclassification that accounts for most of their activity, and can therefore be placed in a different industry to some of the activity in the enterprise.

From the 2022 statistical year onwards, sole proprietorships reporting operating income in their tax returns have been included in the population, regardless of the size of their operating income.

Enterprises in central government and social security administration, as well as county authorities and municipalities, i.e. units with sector code 6100 or 6500, are not included in the statistics. The organizational forms «Association/club/organization» (FLI) and «Foundations» (STI) are generally not considered market-oriented activities. However, some exceptions have been made:

• Top-tier football clubs in industry 93.120 are included. These are predominantly organised as FLIs. Clubs are considered market-oriented if they have played in the Eliteserien at least once during the last four years, including the statistical year.
• For the areas of education and health and social work activities, entities with the legal forms FLI and STI are also considered market-oriented.

Data sources

Where available, income statements, VAT data (VAT Register) and balance sheets are obtained from the tax authorities for the entire population, in addition to annual accounts from the Register of Company Accounts at Brønnøysund, and employment figures from ‘a-ordningen’. A questionnaire (often called a supplementary form) is also sent to the enterprises in the sample.

Sampling

The population consists of all active enterprises in the industry classifications during the statistical year. All multi-activity enterprises in the population are normally also included in the sample. In addition, the population is divided into levels by industry and number of employees. Based on these levels, a representative sample of single-activity enterprises is drawn for the sample. In practice, this may entail a full census count for some levels.

Data collection

The enterprises in the sample receive forms for digital reporting via Altinn before the summer. Enterprises that do not respond are subject to a compulsory fine. The forms provide detailed information on the enterprises’ finances, which together with the income statement, balance sheet and employment figures from ‘a-ordningen’ form a complete basis for the variables for which statistics are produced. Enterprises are asked to give a financial breakdown of their KAUs.

For enterprises that are not part of the sample, the turnover is retrieved from the income statement if available. If no income statement is available, the turnover is obtained from the VAT Register. Employment figures are still retrieved from ‘a-ordningen’, but are sometimes estimated based on turnover.

Editing and estimations

Editing is defined here as checking, examining and amending data.

Data from forms and income statements is checked and edited. The annual accounts are an important source for this work. Other sources include the previous year’s figures, the Central Register of Establishments and Enterprises and information from the enterprises themselves.

Turnover and employment are considered full-count variables and are used to estimate the remaining variables for the enterprises outside the sample.

From the 2024 reference year onwards, an automated, machine-learning-based method is used to control and assess survey data, including the allocation of enterprises. The methodology for estimation and imputation combines three main approaches, two of which are based on machine learning.

Model-based prediction is used to estimate missing values using machine learning models (CatBoost). The models first produce an estimate based on the current-year sample and then improve this estimate by incorporating patterns in industry structure and historical data.

Donor-based imputation uses machine learning (KNN-based method) to estimate missing values by identifying enterprises with similar characteristics and calculating values based on a distance-weighted average of the distributions of these enterprises.

In addition, a rate-based method is applied, where enterprises with unsuitable or extreme values are excluded before average rates within industries are calculated and used for selected variables.

Following the machine learning steps, the results are validated using rules and reconciliation checks to ensure internal consistency and coherence of the data.

Prior to 2024, estimation methods were as follows:

• From 2010-2023, the method for estimation was based on finding units outside the sample with a corresponding industry code that is similar to a unit in the sample, for the two variables full-count variables. These units were then allocated values for the other variables based on the given value in the sample. The ‘nearest neighbours’ method was used to select which units are similar to each other.
• The method for estimating missing data for non-sampled units was, up to and including 2010, based on rate estimation within predefined strata.

Not relevant

Employees of Statistics Norway have a duty of confidentiality.

Statistics Norway does not publish figures if there is a risk of the respondent’s contribution being identified. This means that, as a general rule, figures are not published if fewer than three units form the basis of a cell in a table or if the contribution of one or two respondents constitutes a very large part of the cell total.

Statistics Norway can make exceptions to the general rule if deemed necessary to meet the requirements of the EEA agreement, if the respondent is a public authority, if the respondent has consented to this, or when the information disclosed is openly accessible to the public.

More information can be found on Statistics Norway’s website under Methods in official statistics, in the ‘Confidentiality’ section.

To ensure confidentiality, the ‘suppression’ method is used in these statistics.

For comparisons with previously published figures, users should be aware that older data may have been altered in the revision of the previous year’s statistics. In addition, updates to industry codes in the Central Register of Establishments and Enterprises may result in changes for some of the industries. Such changes will not be reflected in the figures for previous years.

Employment figures were previously retrieved from the State Register of Employers and Employees, but since 2015, they have been obtained through ‘a-ordningen’. The employment figures are not directly comparable with previous years. More information on the differences between the two ways of obtaining employment figures can be found on this page for register-based employment, and in this article describing details of the correlation between old and new statistics.

The Standard Industrial Classification was revised in the statistical years 2025 (transition from SN2007 to SN2025) and 2008 (transition from SN2002 to SN2007). This affects comparability with previous years.

From 2018 to 2020 industrial classifications 06, 49.5 and 35 are not included. From 2021 division 06 is again included, in addition to division 35, group 49.5 and sections P, Q, R. Industrial classifications 68.201, 70.100 and 81.109 are not published.

Internal transactions between establishments in the same enterprise are not included in industrial classification C, manufacturing, from 2018.

From year 2022 the definition of “Enterprise” is updated. Legal units without autonomy in decision-making are included in the figures and consolidated with legal units that are in possession of autonomy in decision-making within the business group. Internal transactions between these units are excluded.

From year 2023, the method for calculating the following variables has been changed: "Turnover from principal activity", "Turnover from industrial activities excluding construction", "Turnover from service activities", "Turnover from wholesale and retail trade". Until 2022, the data for these variables were based on responses from the survey questionnaires. From 2023 onwards, the variables are calculated based on the industry classification codes of the establishments:
- Turnover from principal activity: NACE Rev. 2.0 Section 01, 02, 03.
- Turnover from industrial activities excluding construction: NACE Rev. 2.0 Section 05-33.
- Turnover from service activities: NACE Rev. 2.0 Section 58-63, 68-82.
- Turnover from wholesale and retail trade: NACE Rev. 2.0 Section 45, 46, 47.

From the 2024 reference year onwards, the production process for this statistic has been improved. The most important change concerns the calculation of enterprise figures, where a transition has been made to the use of automated allocation based on machine learning models. In addition, improvements have been made to the imputation of variables that are not census variables. This is described in more detail under “editing and calculations”. In some cases, these changes may lead to differences compared with previous reference years and may therefore affect comparability with years prior to 2024.

Statistical survey results can contain measurement and processing errors.

Measurement errors occur as a result of the respondent giving an incorrect answer either because they did not understand the question or they inadvertently gave the wrong information. A common example is numbers that are quoted in whole NOK instead of NOK 1000. Using familiar terms and providing easy access to guidance reduces this type of error.

Processing errors include coding errors and errors that occur during the transferring of information from the questionnaire or during editing.

Form-based surveys normally entail some level of non-response. Non-response by units can be due to conditions such as closure, bankruptcy, merger, demerger, holiday leave, submission errors etc. Partial non-response can be due to an oversight, missing data or other factors.

Non-response by critical units is followed up in the manual post-check. Other enterprises that do not return their forms are treated in the same way as enterprises that are not included in the sample. The use of compulsory fines reduces the non-response rate considerably, but not necessarily the uncertainty. For example, the respondents may choose to estimate figures just to avoid the fine.

About 95 per cent of the forms sent are normally returned. This means that the unit non-response rate is relatively low. However, the rate is somewhat higher in relation to answers to individual questions. This particularly applies to the breakdown of financial figures at the Local-KAU level.

Sampling errors are defined as the uncertainty that arises from producing the figures on the basis of a sample of units as opposed to the entire population. Sampling errors thus measure the expected deviation between the sample and what the result would have been if the entire population was included in the survey.

The variables included in the preliminary figures are based on a full census count. Sampling errors are therefore not relevant in this regard.

Several administrative registers are central to the work involved in producing statistics. The Register of Establishments and Enterprises is used both to define the population and to obtain identifiers and data. The Central Coordinating Register for Legal Entities at Brønnøysund, the VAT Register and ‘a-ordningen’ are used to obtain data on the units. This provides the basis for two types of registry errors that can affect the uncertainty of the statistics.

The most common errors are due to delays in registration. Such delays may be due to late reporting to the registers or the fact that changes are normally registered some time after they occur. The consequence is that the registers are not fully up-to-date at all times, which can, for example, lead to a bias in the sample or that the statistics are based on outdated information.

Identifiers such as industry code etc. in the Central Register of Establishments and Enterprises are also used when the sample is drawn. The quality of these identifiers will be crucial to the breakdown of the population into appropriate levels, and will therefore affect the quality of the basis from which the sample is drawn.

A revision is a planned change to figures that have already been published, for example when releasing final figures as a follow-up to published preliminary figures. See also Statistics Norway’s principles for revisions.

The Structural Business Statistics is an annual statistics, for which both preliminary and final figures are published.

Preliminary figures are published approximately 10 months after the end of the statistical year. The only variables included in these figures are number of enterprises, turnover and employment. The statistical unit is ‘enterprise’.

Final figures are normally published 16 months after the end of the reference year. The figures replace the preliminary figures for the statistical year, and also include far more variables. Figures for enterprises as well as LKAUs are included.

In accordance with Regulation 2019/2152, an annual quality report is submitted to Eurostat for these statistics. The report provides information on the quality of the sample in the statistics and includes details on sample coverage and size, response rates, and the coefficient of variation for certain variables.