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.
Activities covered: The survey covers NACE Rev. 2.1 Sections B to E.
3.4. Statistical concepts and definitions
The production index shows the evolution of gross value added at basic prices. Depending on the activity, enterprises report gross production, turnover, physical quantities of products, energy consumption, consumption of raw materials or hours worked.
No changes in data collection are foreseen, other than the use of VAT information to calibrate the results of the sample survey of turnover.
3.5. Statistical unit
Reporting unit Enterprise (legal unit). Observation unit(s) The enterprise and parts of enterprises (presumed to be KAUs).
3.6. Statistical population
The statistical population comprises the observation units (enterprises) that are economically active in the sectors B to E (NACE Rev. 2.1) in the period under review.
For 2023, there are ± 84.000 enterprises in the population.
3.7. Reference area
Geographical area covered: The geographical area covered by the survey is the Netherlands - no regions are excluded.
3.8. Coverage - Time
Date of first use as a source: 2000.
3.9. Base period
Base (reference) year: The base year at present is 2021. Once every five years the index is revised.
Indexes. Also percentage change published nationally.
Month.
6.1. Institutional Mandate - legal acts and other agreements
Legal basis: Statistics Netherlands is fully independent in terms of its statistical operations with respect to methodology and publications. Independence was granted by the Royal Act of 1899 and reconfirmed by law in 1996 and 2003. The CBS Law of 2003 is available in English and Dutch. The public can find it at the CBS website
Regulation (EU) 2019/2152 of the European Parliament and of the Council on European business statistics, repealing 10 legal acts in the field of business statistics (EBS-Regulation)
Commission Implementing Regulation 2020/1197 laying down technical specifications and arrangements pursuant to Regulation (EU) 2019/2152 of the European Parliament and of the Council on European business statistics repealing 10 legal acts in the field of business statistic (General Implementing Act)
Commission Implementing Decision 2021/1003 granting derogations to certain Member States with respect to the transmission of statistics pursuant to Regulation (EU) 2019/2152 (Derogations)
Statistiscs netherlands makes all the non-confidential data available on its website. Selected data in special formats are transmitted monthly to Eurostat, the IMF (International Monetary Fund) and the OECD (Organisation for Economic Cooperation and Development).
7.1. Confidentiality - policy
Statistics Netherlands is fully independent in terms of its statistical operations with respect to methodology and publications. Independence was granted by the Royal Act of 1899 and reconfirmed by law in 1996 and 2003. Data provision was made compulsory by law in 1936. Confidentiality is guaranteed. Individual data are never published without consent.
Treatment of confidential data: All activities which consist of fewer than four enterprises or in which one enterprise has a market share of 70% or more are confidential.
8.1. Release calendar
A rolling 5 months-ahead release calendar is published in the schedule by date of Statistics Netherlands. When approximate release dates are given, the precise dates are announced on the last working day of the week prior to the release.
The data are released simultaneously to all interested parties at 6:30 CET. One specific recipient, Eurostat, receives the data via Edamis. At the same time the data is also sent to subscribers by e-mail. The untitled press release is subsequently posted on the Statistics Netherlands Internet website.
Data are sent to Eurostat for EU aggregates and as well released as national data.
10.6. Documentation on methodology
Dissemination of documentation on methodology and sources used in preparing statistics: A methodological description is available, at Omzet en productie | CBS (Dutch only).
Description of standard tables produced: 44 branches are covered and published according to NACE Rev. 2.1 and ISIC. The other indices from the 194 compiled are not confidential, but are not reliable enough. They are used to calculate aggregated indices. No regional breakdown exists at the moment.
10.7. Quality management - documentation
Not available.
11.1. Quality assurance
There are different policies and procedures guaranteeing quality. Statistical improvements are reviewed by an independent department. There is also a list of methods that have been accepted as best practice. New statistics should use these methods or only deviate with good reasons. Every month the results of the IPI statistics are validated by a independent statistician.
11.2. Quality management - assessment
The general quality of the Industrial production index (IPI) is considered good. All quality aspects have been covered.
12.1. Relevance - User Needs
BCS provide statistical information necessary to monitor the competitiveness and performance of the business community in the EU. The BCS are used by different users (European Commission and ECB, national governments and central banks, economic analysts in private companies and financial institutions) and serve different purposes.
12.2. Relevance - User Satisfaction
User satisfaction is measured via download statistics of the online databank and with ad hoc user consultations. There is no further systematic collection of user opinion. Feedback is occasionally given in meetings, email or telephone contacts etc.
12.3. Completeness
Eurostat requirements are fulfilled.
13.1. Accuracy - overall
Production and average daily production: the first provisional results are published by CBS within five to seven weeks after the end of the reporting month. Until 70 days after the end of a quarter, the provisional figures may change as a result of subsequent response. The figures normally won't be adjusted until, no later than six months after the end of the year, the last observed response is processed. The effect of missing non-response is typically minor. The industrial production index is based on an integral dataset, no grossing up is carried out. Selection of companies is based on the national business register, the source for NACE-classifications is the Chamber of Commerce. NACE-classifications of companies are adjusted in the business register if activities don’t match.
13.2. Sampling error
The primary input for the production-index is turnover. The units belonging to the sections Mining and quarrying, Manufacturing and Production & distribution of electricity, gas and water are integrally observed, therefore sample margins do not apply.
13.3. Non-sampling error
The production index uses a panel to monitor the turnover development, which updates every month. The effect of non-response is generally small. Missing values are imputed automatically, the effect of the imputation method on the quality of the data is considered to be only minor as is confirmed by manual analysis and validation of the data. No models are used.
In addition to turnover, the production index also uses other sources such as volumes. Fewer companies provide data on volumes than on turnover. As a result flaws in volume figures have potentially more impact than flaws in turnover figures. Missing responses are followed up by reminders in first stage and then by telephone contacts.
QPI - size-weighted unit response rate of first estimates The Production Index Industry is unique in the sense that is has a wide range of different types of input, which makes calculating an exact size-weighted unit response rate (SWURR) from the microdata very time-consuming. Although (industrial) turnover is the most used one, a large part consists of data of working hours (maritime industry), volume indicators such as kilos (harvests), m³ (gas extraction), absolute amounts (number of cars produced, amount of chickens slaughtered) and more, just to name a few. Additionally some of the inputs we use are delivered to use as indices or growth rates, of which we do not have the exact microdata to begin with. As such we decided to calculate this QPI using only our turnover data. This is the most commonly used input in the PII and also the potential weighting variable that is known for the largest group of the population.
The monthly average SWURR of our first estimates over 2023 was 89,9% for the whole industry (B through E, excluding D353). It should be noted however that all periods are analyzed multiple additonal times in subsequent periods, with new responses coming in over time. After our annual revision the final weighted respons rate generally hovers around 98%.
14.1. Timeliness
First release: within 40 days after the reference month. Monthly corrections are made due to late reporting of enterprises. Three months after the end of the month under review the index is considered to be of good quality; subsequent changes are generally minor. The production index is final after one year, when December is final, with the first publication of April in the following year.
14.2. Punctuality
All releases are published according to the release calendar.
For 2023, all the deadlines were respected.
15.1. Comparability - geographical
Whole national territory is covered.
15.2. Comparability - over time
The time series 2005-2023 should be comparable over time, since no major changes in methodology or sources have been implemented in that period. Some smaller adjustments have been made to the business register in 2006 and 2009.
15.3. Coherence - cross domain
All production statistics are integrated into the system of National Accounts. Nominal production values are deflated by price indices. Quarterly information is used to improve monthly indicators.
15.4. Coherence - internal
The main aggregates (Both NACE and MIG) are always consistent with the subaggregates.
Higher NACE aggregates are calculated by weighing the lowest level NACE aggregates.
Chain linking is applied. The volume weights are updated every 5 years.
The sample survey is almost 5250 each month. The total burden for respondents is estimated to be about 16.000 manhours annually.
The cost for the statistical office (data collection, editing, processing etc) is estimated to be about 8.000 manhours annually.
These costs cover the STS and the national requirements.
17.1. Data revision - policy
The Statistical office release policy makes a distinction between adjustments, corrections and revisions.
The published results can be changed over time due to several reasons:
- Adjustments are made due to extra information from businesses: the data of the last four reference months are provisional and can be/are adjusted once a month if/when new information is available.
- Corrections for large, apparent mistakes (for example serious editing or processing errors): this type of change is applied very rarely.
- Revisions are due to the change of methodology or base year.
New figures are published once a month for BCS purposes. Figures of the three preceding periods can be adjusted due to extra information available.
The National and Eurostat revision policy is the same.
17.2. Data revision - practice
The data published are provisional. Final data are published after the first publication of April next year, when December is final.
Possible adjustments to first releases are due to better response.
In case of changes in methodology or major changes in the data, the public will be informed in advance by means of a press announcement (not translated on the Internet Site) one or two days in advance of the event. The last major change in methodology was in 2018.
QPIs: As requested the MR (Mean Revision) and MAR (Mean Absolute Revision) have been calculated for industry PI using (monthly) year-on-year growth rates for unadjusted and calendar adjusted data over the 2020-2022 period (base year 2015), comparing the data from the first made delivery of every month to the latest one. For the calendar adjusted series the MR was equal to -0,1 and the MAR equal to 0,6.
Due to the change in base year to 2021 = 100, the updating of the weights and changes in methodology it is not possible to calculate and provide the MR and MAR over 2023.
18.1. Source data
The source for the turnover information that is used to calculate production is a statistical survey. For some Classes of industry the information necessary to calculate production is collected more on a ad-hoc basis from individual enterprises or institutions. In particular in the food, basic metal, machine and car industry.
Regarding industrial production data source about 60 percent of the indicator in terms of weight relate to deflated turnover, 16 percent to measurement of physical output, 11 percent to net bookings, 9 percent to hours worked, 2 percent to calculated value added at constant prices, and the rest to various other methods.
Frame on which the source is based: the General Business Register (Algemeen Bedrijfsregister - ABR).
Sample or census: Census Survey with cut-off threshold.
Criteria for stratification: for mining and manufacturing industries a threshold of 10 employees is applied, for Manufacturing one of 20 persons employed. The selected enterprises represent around 90% of gross production. Units with fewer than 20 employees (not queried) represent around 10%.
Threshold values and percentages: about 5250 units are surveyed every month.
Frequency of updating the sample: Monthly.
18.2. Frequency of data collection
Periodicity of data collection 99% of the data is collected mainly monthly. The indices are calculated monthly and quarterly.
18.3. Data collection
Most of the data collection is done by electronic means, both e-mail and direct connection, after receiving a log-in code by mail. -
Questionnaires used in the survey are monthly questionnaire on sales and order positions and some additional branch-specific information. Quarterly questionnaire on stocks. Specific information is received from companies and from boards of producers.
18.4. Data validation
Data validation is performed as a final step before releasing/disseminating the data. The validation is performed by an independent statistician.
The data file structure is vailidated bij Eurostat itself when uploading the files through eDAMIS. The results of EBS are presented in excel-datatables and graphs. The validity of the results is done by examing the tables and graphs.
18.5. Data compilation
Estimates for non-response: non-response for turnover is treated using various imputation techniques. The most common estimation is done by using growth rates of the known average of a particular industry multiplied with the value of the previous month.
Estimates for grossing-up to population levels: the weights used for aggregations are based on the value added at basic prices of the most recent available year and includes the value added of the entire population.
Type of index: a simple Laspeyres index is used. Statistics Netherlands calculates the Production index as a chain index.
Method of weighting and chaining: the indices are weighted with the gross value added at basic prices according to the most recently available National Accounts.
18.6. Adjustment
As far as turnover is used for the calculation of the index, Producer price indices are used for deflation.
Working day and seasonal adjustment calculations: since 2000 working day adjusted figures (X12-Arima) have been published at the 2-digit level as well as seasonally adjusted figures (X12-Arima) for regroupings of the 2-digit level. Since 2024 calendar- and seasonally adjusted for many of the 3- & 4-digit series are supplied as well.
Regarding the provision of the MR and MAR: in the 2019 Metadata Work Programme we were requested to calculate these variables for ''industry''. As far as we know ''industry'' encompasses the whole of sections B, C, D and E (SBIs 06 through 39). However, even though we calculate and publish this statistic ourselves, we do not provide it to Eurostat. The largest aggregate we provide to Eurostat is activity ''BTOE36'', which excludes SBIs 37-39. As such we have used this variable to calculate the revision data, and not ''industry'' as a whole.
The production index shows the evolution of gross value added at basic prices. Depending on the activity, enterprises report gross production, turnover, physical quantities of products, energy consumption, consumption of raw materials or hours worked.
No changes in data collection are foreseen, other than the use of VAT information to calibrate the results of the sample survey of turnover.
Reporting unit Enterprise (legal unit). Observation unit(s) The enterprise and parts of enterprises (presumed to be KAUs).
The statistical population comprises the observation units (enterprises) that are economically active in the sectors B to E (NACE Rev. 2.1) in the period under review.
For 2023, there are ± 84.000 enterprises in the population.
Geographical area covered: The geographical area covered by the survey is the Netherlands - no regions are excluded.
Month.
Production and average daily production: the first provisional results are published by CBS within five to seven weeks after the end of the reporting month. Until 70 days after the end of a quarter, the provisional figures may change as a result of subsequent response. The figures normally won't be adjusted until, no later than six months after the end of the year, the last observed response is processed. The effect of missing non-response is typically minor. The industrial production index is based on an integral dataset, no grossing up is carried out. Selection of companies is based on the national business register, the source for NACE-classifications is the Chamber of Commerce. NACE-classifications of companies are adjusted in the business register if activities don’t match.
Indexes. Also percentage change published nationally.
Estimates for non-response: non-response for turnover is treated using various imputation techniques. The most common estimation is done by using growth rates of the known average of a particular industry multiplied with the value of the previous month.
Estimates for grossing-up to population levels: the weights used for aggregations are based on the value added at basic prices of the most recent available year and includes the value added of the entire population.
Type of index: a simple Laspeyres index is used. Statistics Netherlands calculates the Production index as a chain index.
Method of weighting and chaining: the indices are weighted with the gross value added at basic prices according to the most recently available National Accounts.
The source for the turnover information that is used to calculate production is a statistical survey. For some Classes of industry the information necessary to calculate production is collected more on a ad-hoc basis from individual enterprises or institutions. In particular in the food, basic metal, machine and car industry.
Regarding industrial production data source about 60 percent of the indicator in terms of weight relate to deflated turnover, 16 percent to measurement of physical output, 11 percent to net bookings, 9 percent to hours worked, 2 percent to calculated value added at constant prices, and the rest to various other methods.
Frame on which the source is based: the General Business Register (Algemeen Bedrijfsregister - ABR).
Sample or census: Census Survey with cut-off threshold.
Criteria for stratification: for mining and manufacturing industries a threshold of 10 employees is applied, for Manufacturing one of 20 persons employed. The selected enterprises represent around 90% of gross production. Units with fewer than 20 employees (not queried) represent around 10%.
Threshold values and percentages: about 5250 units are surveyed every month.
Frequency of updating the sample: Monthly.
Monthly.
First release: within 40 days after the reference month. Monthly corrections are made due to late reporting of enterprises. Three months after the end of the month under review the index is considered to be of good quality; subsequent changes are generally minor. The production index is final after one year, when December is final, with the first publication of April in the following year.
Whole national territory is covered.
The time series 2005-2023 should be comparable over time, since no major changes in methodology or sources have been implemented in that period. Some smaller adjustments have been made to the business register in 2006 and 2009.
