I.N.S.E.E. (Institut National de la Statistique et des Etudes Economiques),The National Statistical Institute of France.

Direction des Statistiques d'Entreprises (DSE), Business Statistics Directorate

Département des Statistiques de Court Terme (DSCT), Department of Short-Term Statistics

Division Élaboration des statistiques de production industrielle (ESPRI), Elaboration of industrial production statistics.

I.N.S.E.E, Timbre E330, Thomas DENOYELLE

88 Avenue Verdier,

92120 Montrouge

France

The PRODCOM survey is conducted by I.N.S.E.E.. This is an annual survey concerning the value and quantity of industrial products produced by enterprises in France and sold during the reference year.
It is carried out in compliance with EU Parliament and Counci Regulation of 27 November 2019 on European business statistics (No.2019/2152).

The survey used to provide PRODCOM data to Eurostat is not specific to Prodcom. It is called EAP ("Enquête annuelle de production" or, in English, Yearly Production Survey).
No actual release is made at national level regarding the PRODCOM classification, except for the Food, Agriculture and Sawmill sectors mentioned below. France releases data at PRODFRA 10-digit level and, until 2020, CPF45 level (close to the CPA).

There are two exceptions in the EAP survey where data are collected separately:

1. The first one is the Food, Agriculture and Sawmill sectors, covering NACE 10, 11, 12 and 16.10. These NACEs are treated by the Ministry of Agriculture, Agrifood industry and Forestry (around 13.000 enterprises are surveyed) and more specifically by their Statistics and Prospective Analysis Service.
2. The second exception is the steel domain, covering Prodcom 2410T, which is collected by A3M (Alliance for Minerals and Metals).

Other than that, I.N.S.E.E. deals directly with the Prodcom data through EAP. The EAP sample is approximately 33,000 enterprises, covering industrial activities and products (exceptions mentioned above are excluded).

PRODCOM data provide information on PRODuction of the European COMmunity and are compiled from data received from manufacturers. There were 4,043 products covered in 2022.

Available data include: value and volume of product sales, total production, number of enterprises, share of production under sub-contracting (EAP).

Classification of Products:
The PRODCOM list is divided into “Classes” corresponding to the classes of NACE rev 2. Within the individual classes, the headings of the list are set out in ascending order of their PRODCOM codes.
PRODCOM uses an eight-digit product code of the form XX.XX.XX.YY.

The first four digits of the code correspond to the 4-digit classes of NACE Rev. 2.
The first six digits are the Community Classification of Products by Activity (CPA) codes.
The last two digits provide a more detailed breakdown of the CPA classes into PRODCOM product headings.

There is a direct link between the PRODCOM classification and the EU foreign trade Combined Nomenclature (CN). Each PRODCOM 8-digit heading covers one or more CN 8-digit headings

At national level
The French system has its own nomenclature used to collect data for Prodcom and for the EAP (Enquête annuelle de production - Yearly Production Survey). What we use are 10-digit headings called PRODFRA, the first eight digits of the code corresponding to the 8-digit heading of Prodcom (except a few specific cases).

Under Prodcom, the NACE Rev.2 classification is covered as follows: sections B, C and E, including divisions 07 – 39, excepting 09 and 19.2.

The EAP survey and Food, Agriculture and Sawmill sectors survey covers all industrial enterprises, which are wholly, primarily or significantly engaged in industrial production and industrial services.

Under EAP, the NACE Rev.2 classification is covered as follows: sections B, C, D and E, including divisions 05 – 39, excepting 10, 11, 12 and 1610.

The survey is exhaustive, either for enterprises with 20 or more employees, or if their turnover exceeds a fixed threshold for all NACE Rev.2 sub-sectors, so that at least 85% of the sector turnover is covered. For the others, they are surveyed using a stratified random sample.

Enterprises classified in a sector other than industrial are surveyed in addition (the ones with secondary industrial sector having a turnover at least equal to the turnovers of enterprises in the exhaustive part (see 18.1)).

The purpose of the statistics is to report, for each product in the Prodcom List, how much has been produced in the reporting country during the reference period. The NACE codes on which Prodcom codes are based merely serve to identify the enterprises that should be surveyed in order to determine the amount of production of the product.
The products: the Prodcom List defines the products to be included in the survey. There are approximately 4,000 headings representing manufactured products and some industrial services.
The value and the volume of production: the value is expressed in Euros by Eurostat. The volume is expressed in a unit specified for each product.
The volume unit: the measurement unit, such as kilograms, square metres etc. used to indicate the volume of goods produced.
Production type: either Sold Production on Own Account (the value and volume of the product sold by the enterprise on its own account), Sold Production under Sub-contracted Operations (the value and volume of the product sold by the sub-contractor) or Actual Production (the volume of all production of the product, including both the proportion that is sold and the proportion that is retained by the enterprise for adding to stocks, using in further processing etc.).

Generally the reporting unit is the Legal Unit as a proxy for the enterprise, and so is the observation unit.

The survey population is made up of all industrial enterprises in NACE 05-39, as well as non industrial enterprises with one or more important secondary branches.

Our data cover France (Monaco and extra-regio NUTS 1 are excluded) and the French overseas departments (French Guiana, Guadeloupe, Martinique, Réunion and Mayotte). No regional breakdown.

The observed period is one calendar year.

Data from 2008 to 2013 are available at https://www.insee.fr/fr/statistiques/2504810.

Data for EAP 2014 are available at https://www.insee.fr/fr/statistiques/2503132.

Data for EAP 2015 are available at https://www.insee.fr/fr/statistiques/2503588.

Data for EAP 2016 are available at https://www.insee.fr/fr/statistiques/2905664.

Data for EAP 2017 are available at https://www.insee.fr/fr/statistiques/3596366.

Data for EAP 2018 are available at https://www.insee.fr/fr/statistiques/4200788.

Data for EAP 2019 are available at https://www.insee.fr/fr/statistiques/4640140.

Data for EAP 2020 are available at https://www.insee.fr/fr/statistiques/5423589.

Data for EAP 2021 are available at https://www.insee.fr/fr/statistiques/6676569.

For the Food, Agriculture and Sawmill sectors, the data can be found at http://www.agreste.agriculture.gouv.fr/enquetes/entreprises-agroalimentaires/statistiques-prodcom-et-enquete-de.

From 2021 on, data released by I.N.S.E.E are at Prodfra level. Data released by the Ministry of Agriculture, Agrifood industry and Forestry are released at Prodcom level.

Not applicable.

For volumes: units of measure are specific to each Prodcom code
For values: data are collected, stored and transmitted to Eurostat in THOUSANDS OF EUROS (k€), and released in EUROS (€) in the Eurostat database. VAT is excluded.

The data are from the calendar year 2022.

This survey is carried out in compliance with following regulations:

At EU level:

1. Regulation (EU) 2019/2152 of the European Parliament and of the Council of 27 November 2019 on European business statistics, repealing 10 legal acts in the field of business statistics (Text with EEA relevance), to be found at https://eur-lex.europa.eu/legal-content/en/TXT/?uri=CELEX%3A32019R2152

At national level:

The main legal acts concerning Insee and public statistics can be found at https://www.insee.fr/fr/information/1892119:

1. The role of Insee: Articles 32 and 33 of the "Loi n° 46-854 du 27 avril 1946"
2. The organisation of Insee: "Arrêté NOR : FCPO1621015A du 26 juillet 2016"
3. The list of surveys of enterprises, farms and public services: "Arrêté du 21 octobre 2020 portant approbation du programme d'enquêtes statistiques d’initiative nationale et régionale des services publics pour 2021 (enquêtes auprès des entreprises et des exploitations agricoles)", can be found at https://www.cnis.fr/arretes-au-journal-officiel-du-programme-denquetes-2022.

Not applicable at Member States level.

Regulation (EC) No 223/2009 on European statistics (recital 24 and Article 20(4)) of 11 March 2009 (OJ L 87, p. 164), stipulates the need to establish common principles and guidelines ensuring the confidentiality of data used for the production of European statistics and the access to those confidential data with due account for technical developments and the requirements of users in a democratic society.

The Statistical law of France stipulates that the following data are confidential:

1. individual statistical data;
2. statistical data which can be matched in a way that enables the identification of a specific statistical unit;
3. statistical information which aggregates data about less than three statistical units or about a population in which the relative share of the value of a surveyed parameter of a single unit exceeds 85 per cent of the total value of such parameter for all units in the population.

France's full policy on data confidentiality is available at:

https://www.insee.fr/fr/information/2387517

https://www.legifrance.gouv.fr/affichTexte.do;jsessionid=4BE8CF0C938023A97E8FBF5039E787C7.tpdjo02v_3?cidTexte=JORFTEXT000000888573&dateTexte=20130726

France is applying the principle of the active confidentiality.

The policy of French Statistics on confidentiality of Prodcom data is that a specific code is flagged as confidential only if at least one of the following two criteria are met:

• the largest reporting company has a share over 85 pct. of total sales of the commodity.

• there are less than 3 companies reporting sales of the commodity.

Additionnaly, a code can be flagged as confidenial in order to protect the national dissemination.

For 2022, 15,602 figures (value on own account, value on subcontracted operations, sold quantity on own account, sold quantity on subcontracted operations or actual producted quantity) have to be reported according to the Prodcom list. For France, in the June of 2023 dataset, 35% of them are confidential: about 22% because of the 3 companies criterion, 6% because of the 85 pct. criterion and 7% to protect national dissemination.

At PRODFRA level, there is only a definitive release in November-December N+1 or beginning N+2 (N is the reference year); no temporary release.

For the Food, Agriculture and Sawmill sectors: a temporary set of data is released in June N+1, and the definitive one is released in October N+1. A few revisions can be made in June N+2 and October N+2.

There is no release calendar available for the public.

In line with the Community legal framework and the European Statistics Code of Practice Eurostat disseminates European statistics on Eurostat's website respecting professional independence and in an objective, professional and transparent manner in which all users are treated equitably. The detailed arrangements are governed by the Eurostat protocol on impartial access to Eurostat data for users.

Prodfra/CPF45 data are published on I.N.S.E.E.'s website at: https://www.insee.fr (see Enquête annuelle de production)

PRODCOM data, as part of the statistical information collected and disseminated by the NSI, contain annual data obtained from the annual survey that comprise all industrial and non-industrial enterprises with their annual income from the industrial activity.

Not available.

Publications using EAP data can be found here: https://www.insee.fr/fr/statistiques?debut=0&theme=53&categorie=2&collection=115.

For the Food, Agriculture and Sawmill sectors, data are available at: https://agreste.agriculture.gouv.fr/agreste-web/disaron/!searchurl/cb4f332f-6400-4b9f-9790-2e5444f2663d!745fcc0d-4e89-4a99-b151-835d2a9be2df/search/.

The EAP databases are available online at: https://www.insee.fr/fr/statistiques?debut=0&theme=53&categorie=3.

For the Food, Agriculture and Sawmill sectors, data are available online at:

- for Food and Agriculture: https://agreste.agriculture.gouv.fr/agreste-web/disaron/!searchurl/cb4f332f-6400-4b9f-9790-2e5444f2663d!cda8b080-3e9e-4368-b41d-7a29c1da0be6/search/ ;

- for Sawmill: https://agreste.agriculture.gouv.fr/agreste-web/disaron/!searchurl/ee544c16-0429-48bc-b614-150cec98259e!cda8b080-3e9e-4368-b41d-7a29c1da0be6/search/ .

Not available.

Prodcom data are regularly transmitted to Eurostat.

Data are also distributed to the national and international users on the 4 and 10 digit-codes levels.

The main documentation is an internal desk manual describing the statistical production process.
For users, the main documentation is the Quality Declaration https://www.insee.fr/fr/metadonnees/source/s1193#documentation.

The main documentation is an internal desk manual describing the statistical production process.
For users, the main documentation is the Quality Declaration https://www.insee.fr/fr/metadonnees/source/s1193#documentation.

The main instrument for quality monitoring and assurance for French Statistics are the Quality Declarations, available at https://www.insee.fr/fr/metadonnees/source/s1193#documentation.

Prodcom is published with a high level of detail, its accuracy therefore depends on data being checked in several ways.

The main quality assurance activity for Prodcom is the data validation, see 18.4. concept.

The Information System provides a number of arithmetic and logical controls, both mandatory and optional between the items in each form.

There are a number of arithmetic and logical controls inside the PRODCOM form.

The System calculates both the totals and the unit price.

The PRODCOM questionnaire can not be completed successfully before all errors (detected by the checking program) have been corrected.

The supplied PRODCOM data are crosschecked against PRODCOM data of the previous year.

If there are interlinked errors, these are displayed after the validation of the whole set of reports.

The 3 main uses of PRODCOM data are:

• the knowledge of the detailed industrial activity (for Prodcom, trade associations, ...)
• the knowledge of the different activities of enterprises, so that the Principal Activity (APE) is known (for SBS, ...)
• to provide a yearly framing of the industry for STS (indices of production)

The main users of PRODCOM data can be presented in different groups, namely:
     a) EXTERNAL/ INTERNATIONAL (OUTSIDE NSI), like:
          - Institutions at European level: Eurostat
     b) EXTERNAL/ NATIONAL (OUTSIDE NSI), like:
          - Institutions at national level - governmental: Ministry of Agriculture, Agrifood industry and Forestry, other governmental bodies
          - Institutions at national level - non-governmental: National Bank, Trade associations
          - Media (the main channel of statistics to the general public), national and international
          - Researchers and students at international or national level
          - Companies, national and international

          - Professional organisations
     c) INTERNAL (INSIDE NSI), like:
          - National accounts
          - Environmental statistics
          - SBS
          - STS

No specific user satisfaction survey has been conducted for these statistics.

Since the PRODCOM coverage is included in the PRODFRA coverage, all the data requested by PRODCOM are available and transmitted to Eurostat.

Sample frame is taken from the I.N.S.E.E Business Registers (BRs).
Since all the industrial units are obtained from the BRs, all companies with their principal activity (APE) in the relevant NACE groups are in the sampling frame. Some others units with important industrial secondary activity are also in the sample thanks to data from SBS.

These enterprises are put into strates. For more details on the sampling method, see 20.1.

For more information on our estimations, see 13.3

For 2022, the response rate to the EAP survey was of 74% of all enterprises surveyed (June transmission). For the Food, Agriculture and Sawmill sectors, the response rate was 89%.

Sampling errors do not occur in the survey. The sample design applied is a cut-off sample, which mean that it is a full coverage survey with a cut-off threshold. There is no grossing up for units below the threshold. It is therefore not relevant/possible to calculate sampling error.

Reported data are first checked via our validation system, see 18.4.

In case of unit non-response (except for Food, Agriculture and Sawmill sectors), the value of the sold production is estimated with the value of the previous year, multiplied with the evolution of the unit's turnover from the VAT source. The same is applied to the sold quantity and total quantity. If the unit never responded before, there is no estimation.

Most cases of partial non-response are due to the fact that the enterprise could not estimate their sold quantity. If the quantity was given the year before, to estimate the missing one, we take the given value of the sold production, and we divide it by the unit price of the previous year. If no quantity was responded before, we use the median price of the product of the current year, of the respondent units.

The treatment of non response for the Food, Agriculture and Sawmill sectors is slightly different. For unit non-response, mass imputation is used: the N-1/ N evolution of respondants for a given product-variable is applied to the N-1 data. However, when there are not enough respondants for the product-variable, or when N-1 value is too important, N-1 value is imputed to N. In most cases, partial non-response is due to the enterprise not being able to estimate sold quantities. If the quantity was given the previous year, we divide the sold value of year N by the unit price of N-1 (of the enterprise/product) corrected by the N-1/N evolution of the average unit price for a given product. If no quantity was responded before, we use the average unit price of the product of the current year (of the other respondent units).

In some other cases, data from other surveys are available and used for imputation. 

For EAP:

The unit non-response rate, defined as the ratio of the number of units with no or not usable information to the total number of in-scope units, is at 56.6% (unweighted) in 2022. If we weigh this ratio with total turnover the unit non-response rate sits at 4.7%.

There is no item non-response on the sold value variable because a questionnaire cannot be valid until its turnover is allocated in its different products.
We define the item non-response rate for a variable as the ratio of the number of partial non-response values (so the unit responded but did not provide an information on this particular variable) over the number of values from surveyed units.
We weighted this ratio with the associated sold value since aggregating quantities with heterogeneous measurement units may not be relevant.

The weighted item non-response rate for the sold quantity variable is 9.4%.
The weighted item non-response rate for the total quantity variable is 4.0%.

We define the unweighted imputation rate for a variable as the ratio of the number of imputed values to the number of values requested for the variable.
We weighted this ratio with the associated sold value since aggregating quantities with heterogeneous measurement units may not be relevant.

The weighted imputation rate for the sold value variable is 8.0%.
The weighted imputation rate for the sold quantity variable is 12.4%.
The weighted imputation rate for the total quantity variable is 13.1%.

PRODCOM data are transmitted to Eurostat just before the 30th of June and in October.

PRODCOM data are not released by I.N.S.E.E..

At PRODFRA level: there is only a definitive release in November-December N+1 or beginning N+2 (N being the reference year); no temporary release.

For the Food, Agriculture and Sawmill sectors: a temporary set of data is released in June N+1, and the definitive one is released in October N+1. A few revisions can be made in June N+2 and October N+2.

The statistics are usually published without delay in relation to the scheduled date.

Prodcom data were transmitted succesfully to Eurostat on June 29th, 2023. A revision will ne sent in October 2023.

Not applicable at MS level.

An extensive revision of NACE  and CPA in 2007 has resulted in changes to PRODCOM estimates for the 2008 survey onwards.

Every 5^th year, there is a revision of the HS (Harmonized System) which will result also in more significant changes to CN and Prodcom. The latest major revision was in 2021.

Revision in NACE will also affect Prodcom. Prodcom data from 1995-2007 based on NACE Rev.1.1 have been backcasted to NACE Rev.2 by Eurostat.

Prodcom 2015 was the last year using CPA Rev.2 as reference before moving on to CPA Rev 2.1 for Prodcom 2016.

EAP (and Food, Agriculture and Sawmill sectors) data are articulated with data from customs, thanks to the link between Prodcom and Combined Nomenclature. It is also coherent with the price indices and industrial production indices from STS.

EAP is one of the sources which supplies SBS for the French industrial enterprises. Consistency checks are done through a division called ESANE (Elaboration des statistiques annuelles d'entreprises, Elaboration of yearly enterprises statistics).

Coherence ensured.

At the end of the EAP questionnaire, it is asked the needed duration (in minutes) to complete the questionnaire.

For 2021 the median duration was 30 minutes and the average duration was 1 hour and 15 minutes.

For Prodcom, data transmitted at the end of June are preliminary, and revisions are sent in October. The Prodcom 2022 data will be considered definitive by I.N.S.E.E. in October 2024 (temporary in June 2023, semi-definitive in October 2023). Usually, the only changes between semi-definitive and definitive data are headings of the food and sawmill industry (Prodcom 10, 11, 12, and 16.10).

Last year, about 41% of 2021 Prodcom headings were revised for the October of 2022 transmission. Sum value of total sold production has increaseded by 0.6% between June and October.

A data revision will be sent in October 2023 for the Prodcom 2022 data. Another revision will be sent in October 2024.

For the EAP, the current sample design consists of three classes of enterprises: EAP1, EAP2, EABIS. Each enterprise is composed of one or several legals units. For EAP1 and EAP2, a cluster sampling design is used and all the legal units of the sampled enterprises are selected.

• EAP1: This class is created by first creating a group of enterprises called "pure exhaustive". It contains all the enterprises with 20 or more employees, OR, a turnover of 5.000.000€ or more. It also contains the enterprises which are in a APE (principal activity) class with less than 10 enterprises. This "pure exhaustive" group is then completed to form EAP1. It is completed in a way that 85% of each APE class turnover will be covered. That is to say, for every APE class, enterprises are ordered by descending turnover. Then, we complete each APE class with the first enterprises until we cover 85% of the class. The last enterprise's turnover in EAP1 is called the cut-off threshold. In 2022 there were about 25,500 legal units in EAP1.
• EAP2: The other enterprises of the sample frame are put in classes depending on their APE, number of employees, and turnover. Then, they are surveyed using a stratified random sample. In 2022 there were about 6,500 legal units in EAP2.
• EABIS: This part of the sample is for the legal units with a non-industrial APE, but with an important industrial secondary activity. What we define as important is above the cut-off threshold of a given APE class. In 2022, there were about 1,000 legal units in EABIS.

Since this is an annual survey, data are collected from January until September.

EAP data collection is organized through the Regional Statistical Office in Normandy (in Caen). Data control is carried out at the regional level and at the national level.
I.N.S.E.E. uses paper and online questionnaires. Enterprises could answer on this website: https://entreprises.stat-publique.fr/ . They have to log in with their Siren number (National business register) and the password they received by mail.

The questionnaires:

• are customized for the enterprise, based on the products produced the previous year.
• contain the entire list of products within the corresponding NACE Rev.2 class, if needed.

A vast majority of respondants (98%) uses the online questionnaire.

For the Food, Agriculture and Sawmill sectors, data collection is proceeded the same way: Enterprises have to log in on this website: http://mesdemarches.agriculture.gouv.fr/demarches/entreprise-agroalimentaire-et/repondre-aux-enquetes-statistiques-73/ and choose the right survey.

Once the questionnaires are filled and sent to I.N.S.E.E they go through our system of controls and adjustments.

In a 1st phase, every questionnaire/ product/ variable triplet is given a score between 0 and 1 (1 being the best) based on the controls it undergoes. For example we check if the reported turnover is similar to the one we can find in other statistics like VAT source, and there are many other checks. If the global score is above a certain threshold, the questionnaire is validated. If not, the questionnaire is "fixed" as much as possible by using data from previous years and other sources.

Then it goes through a 2nd phase of controls and again, if the score is above a certain threshold, it is validated. If not, the agent responsible for the respective enterprise is warned and asked to contact the enterprise for a proper fill-in of the questionnaire.

Amongst the controls, some are essential and are blocking the questionnaire and others are simply warnings for the agent to make sure it is coherent.

Every Unit/Product couple is estimated at the beginning of the campaign (January) following the principles explained in 13.3.

Then, when a questionnaire is validated, all the data from the enterprise are replaced by the reported values.

Finally in June (and in October), the results are aggregated at Prodcom level and centralised in a table which is then sent to Eurostat.

Since this is an annual survey there is usually no seasonal or any other adjustment.

No comment.