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.
[FR1] National Institute of Statistics and Economic Studies (INSEE) (France)
1.2. Contact organisation unit
Direction des statistiques d'entreprises - Division Enquêtes thématiques et études transversales - Timbre E430
1.3. Contact name
Confidential because of GDPR
1.4. Contact person function
Confidential because of GDPR
1.5. Contact mail address
88 avenue Verdier
CS70058
92541 Montrouge Cedex
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
30 November 2025
2.2. Metadata last posted
30 November 2025
2.3. Metadata last update
30 November 2025
3.1. Data description
The data from the Global Value Chains Survey come from a survey conducted by the French National Institute of Statistics among enterprises with at least 50 employees or self-employed persons, operating in mainly market-oriented, non-agricultural sectors (i.e. sections B to N of the NACE Rev. 2 classification). This survey, conducted every three years, aims to improve knowledge of firms’ internationalization strategies.
The survey examines the fragmentation of production processes implemented by enterprises, leading them to outsource activities or relocate activities previously carried out abroad. It also covers cross-border purchases and supplies of goods, materials and services.
The GVC Survey is a European-initiative survey, mandatory for the first time in 2023.
The GVC dataset contains ten variables defined in Implementing Regulation (EU) 2022/918. These cover:
(1) employment by business function,
(2-5) enterprises purchasing or supplying goods and services abroad,
(6) the number of enterprises carrying out international sourcing,
(7-8) jobs created and lost resulting from international sourcing,
(9) motivations and barriers for sourcing, and
(10) events affecting GVC arrangements.
For variables (2)-(5), the dataset includes only enterprises that report at least 100,000 euros in goods or services purchased from or supplied abroad in the final year of the reference period.
All data are provided as absolute values and presented by NACE activity, business function, geographical partner area, size class, motivation and barrier types, and other breakdowns required by the regulation.
The data refer to 2023 for variables 1-5, and to the period 2021–2023 for the other variables.
Business functions are defined in the GVC implementing act, distinguishing core and support functions as well as specific functional categories such as production of goods, ICT services, marketing, engineering, and R&D.
Geographical classifications follow the breakdowns prescribed in Regulation (EU) 2022/918, including EU Member States, Extra-EU, and specific partner regions (e.g. China, India, USA and Canada, the Maghreb).
Classifications for job skills follow the high-skilled / not-high-skilled distinction set out in the GVC Compilers’ Manual.
3.3. Coverage - sector
GVC data cover enterprises with 50 or more employees and self-employed persons at the end of 2023 that belong to the NACE Rev. 2 sections B to N.
These activities are:
Section B – “Mining and quarrying”
Section C – “Manufacturing”
Section D – “Electricity, gas, steam and air conditioning supply”,
Section E – “Water supply, sewerage, waste management and remediation activities”
Section F – “Construction”
Section G – “Wholesale and retail trade; repair of motor vehicles and motorcycles”
Section H – “Transportation and storage”
Section I – “Accommodation and food service activities”
Section J – “Information and communication”
Section K – "Financial and insurance activities"
Section L – “Real estate activities”
Section M – “Professional, scientific and technical activities”
Section N – "Administrative and support service activities".
This does not include public administration or (largely) non-market services such as education and health.
3.4. Statistical concepts and definitions
All concepts and definitions follow Implementing Regulation (EU) 2022/918 and the GVC Compilers’ Manual. International sourcing refers to the partial or full movement of business functions abroad within or outside the enterprise group, during 2021-2023. Variables (2)-(5) apply only to enterprises exceeding the 100,000 euros threshold for goods/services purchased or supplied abroad. Business functions are divided into core and support categories and specific functional areas (e.g. ICT, management and administration). Motivations and barriers follow the fixed lists provided in the regulation, and importance factors use the standard four-level scale (very important, moderately important, not important, not applicable/do not know).
The structure of the questionnaire is divided into several topics, among wich are the following ones :
Business function
Domestic sourcing
Global value chain arrangements
International sourcing
Backsourcing
Motivation and barriers for international sourcing and backsourcing
Impacts of recent events on economic globalisation.
3.5. Statistical unit
The statistical unit used for the GVC statistics is the enterprise, as defined in Regulation (EU) 2019/2152 on European business statistics and the Business Register Regulation.
The enterprise is the smallest combination of legal units that is an organisational unit producing goods or services, which benefits from a certain degree of autonomy in decision-making, especially for the allocation of its current resources. An enterprise carries out one or more activities at one or more locations. An enterprise may be a sole legal unit.
3.6. Statistical population
The target population consists of all enterprises in NACE Rev. 2 sections B to N, with at least 50 employees and self-employed persons at the end of 2023. For variables (2)-(5), only enterprises reporting at least 100,000 euros of goods or services purchased from or supplied abroad are included, in line with the Annex of Implementing Regulation (EU) 2022/918.
3.7. Reference area
France (all territory, including overseas regions)
3.8. Coverage - Time
2023 is the reference year for global value chains arrangements and business functions.
For the questions on international sourcing, backsourcing, domestic sourcing and the effects of recent events, the reference period runs from 2021 to the end of 2023.
3.9. Base period
Not applicable.
Absolute values. This unit applies uniformly to all variables and all breakdowns required by the regulation.
According to the variables, it will represent enterprises, jobs or employees and self-employed persons.
The reference period for the GVC statistics is 2021-2023, in accordance with Implementing Regulation (EU) 2022/918.
2023 is the reference year for global value chains arrangements and business functions (variables 1 to 5).
For the questions on international sourcing, backsourcing, domestic sourcing and the effects of recent events, the reference period runs from 2021 to the end of 2023 (variables 6 to 10).
6.1. Institutional Mandate - legal acts and other agreements
GVC statistics are produced under the legal framework of European business statistics. The main legal bases are Regulation (EU) 2019/2152 on European business statistics and Implementing Regulation (EU) 2022/918, which defines the variables, population, breakdowns and transmission deadlines for the 2021-2023 reference period. The topic Global Value Chains is listed in Annex I of Regulation (EU) 2019/2152.
The survey is mandatory at national level, it has received visa No. 2024X066EC from the Minister of the Economy, Finance and Industrial and Digital Sovereignty, valid for the year 2024. Non-respondent units can be fined by the Litigation committee for compulsory statistical surveys after the data collection period.
6.2. Institutional Mandate - data sharing
Data are transmitted to Eurostat in accordance with Regulation (EC) 223/2009 and Articles 17 and 18 of Regulation (EU) 2019/2152, which govern data and metadata transmission.
Data are shared on demand within the national statistical system.
Researchers can access the data through the CASD (secure data access center), subject to approval by the Statistical Confidentiality Committee.
Insee makes available some non-confidential and sufficiently reliable data on its website.
7.1. Confidentiality - policy
Regulation (EC) No 223/2009 on European statistics (recital 24 and Article 20(4)) of 11 March 2009, 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.
Access to identifiable information is limited to authorised staff, and confidential data cannot be used for administrative, fiscal or legal purposes.
7.2. Confidentiality - data treatment
Data are transmitted via eDamis (encrypted) and delivered to a secure environment where they are treated. Flags are added for confidentiality in case results must not be disclosed.
At national level, the tool Tau-Argus is used to identify values to flag as confidential:
No cell of the table may concern less than three units.
No cell of the table may contain data for which a company represents more than 85% of the total.
No cell of the table can be calculated from other cells.
None of those cells can be deducted from others.
8.1. Release calendar
Some indicators calculated from microdata and a publication analysing results are planned to be released on Q1 2026
8.2. Release calendar access
Not applicable.
8.3. Release policy - user access
Data tables of indicators and a thematic analysis are publicly and freely available on our website, Insee.fr.
Information on publication releases is provided through Insee's social media account. Users can also subscribe to the newsletter on Insee.fr.
Triennial.
10.1. Dissemination format - News release
On the day of its release, the publication based on the data analysis is highlighted on the Insee website.
10.2. Dissemination format - Publications
At the end of each survey, a publication is produced highlighting the main results obtained. The publication includes tables, charts, analyses of the results, as well as methodological information.
Metadata are also published with the aggregate indicators.
10.7. Quality management - documentation
Information on quality is also published with the aggregate indicators.
11.1. Quality assurance
The European business statistics compliers’ manual for global value chains provides guidelines and standards for the implementation of the survey. Regarding the methodological aspects of the survey, national experts are consulted (sample selection, weighting adjustments).
During data collection, several checks are implemented: verifying that questionnaires are fully completed, and ensuring the consistency of the information provided for example. If any doubt arises, survey officers contact the entreprises to confirm their statements.
During post-collection processing, validation checks are performed by computing aggregates and breakdowns and comparing them with results from previous years. Units with a strong contribution to the aggregates are also checked carefully.
11.2. Quality management - assessment
Model questionnaire is translated to produce the national questionnaire. Methodological Manual guidelines are also used.
Units identified as important with regard to the survey topic are handled with particular care and are specifically followed up in cases of non-response. However, it is not possible to obtain responses from all units, even though the survey is mandatory. Post-collection processing procedures are therefore implemented.
12.1. Relevance - User Needs
Once available, the model questionnaire is translated into French and discussed with national stakeholders (from administrations and civil society) and users (e.g. researchers). The discussion includes fine-tuning the translations, and participants are also asked about their needs regarding the survey topic. Some French-initiative questions can be added, considering the overall response burden/length of the questionnaire.
12.2. Relevance - User Satisfaction
During the annual meeting with national stakeholders and users, participants provide feedback on their use of the survey.
Questions and feedback can also be received from users of the data published on the Insee website, as well as from researchers who access the microdata through the CASD (secure data access center).
No formal user satisfaction survey has been carried out specifically for the GVC domain.
12.3. Completeness
All mandatory questions were included in the French questionnaire. Regarding the optional questions, 65% were included.
For more details, see Annex '"GVC French questionnaire" below.
Overall accuracy is assessed using checks on coverage, internal coherence and consistency with related business statistics.
Some uncertainty may remain for the qualitative variables that require a subjective response, particularly regarding the impact of global events. In addition, international sourcing and backsourcing are relatively rare phenomena, so errors related to sampling and post-collection processing (such as weighting of non-respondents) may persist.
13.2. Sampling error
French data are obtained from a two-part design sample: an exhaustive component, including enterprises with 250 or more employees and foreign groups with subsidiaries in France, and a stratified component based on the NACE classification, the size, the export turnover and the type of entreprise. There are 4 types of enterprises: multinational groups headquartered abroad, multinational groups headquartered in France, entirely French groups, independant enterprises.
In total, 15,038 enterprises were sampled out of a sampling frame of 24,911 enterprises.
In post-collection processing, marginal calibration is applied. The results are therefore representative according to these criteria.
13.3. Non-sampling error
The response rate for the French GVC 2023 survey is 67%. Units were followed up several times to obtain their responses, with large units receiving particular attention. They received up to four reminder letters, and some were also contacted by email or phone. Nevertheless, it was not possible to obtain responses from all units. Post-collection processing procedures had to be implemented. Moreover, the phenomena being measured are quite rare, especially when breaking down the data to a detailed level such as by NACE sector. Coefficients of variation have been calculated, and when they indicated that data quality was insufficient, this was reported at the time of data delivery.
14.1. Timeliness
Data were transmitted to Eurostat within the legal deadlines: data were sent on september 24th of 2025, i.e. 6 days before the deadline.
14.2. Punctuality
All internal milestones in the French GVC production process were met in line with the planned schedule. Data collection took place between September and early January, as planned.
15.1. Comparability - geographical
The French questionnaire is a translation of the model questionnaire, and the methodological instructions provided by Eurostat (coverage, unit surveyed, etc.) were followed. The results are therefore comparable with those of other countries, subject to minor translation differences and to the differentiated handling of non-response.
15.2. Comparability - over time
This survey is the fourth conducted in France, after 2012, 2016, and 2020. In 2016, the survey was limited to small and medium-sized enterprises.
The questionnaire and methodology have changed little between GVC 2020 and GVC 2023, so the results are comparable.
15.3. Coherence - cross domain
Data from the GVC survey on sourcing (international, backsourcing and domestic sourcing) are unique, and no other statistical source can provide them.
As for the number of employees, it can be compared with that from the SBS; it is of the same order of magnitude, and differences may arise from declarations made by different respondents. The profiling of enterprises is also variable, which may explain some differences.
Trade data cannot be directly compared with ITGS and ITSS due to differences in the thresholds applied.
15.4. Coherence - internal
Internal coherence is supported by validation checks during data collection and also during the analyse of the results (logical rules are applied to verify relationships between variables).
Checks were specified prior to data collection, and the software used by the survey officers highlights inconsistencies so that the officers can verify them, sometimes by following up with the enterprises.
The overall response burden for the GVC data collection is considered manageable. The survey is limited to enterprises with 50 or more persons employed and is conducted only every three years.
There was 2 versions of the questionnaire:
one for enterprises that belong to a group, that included questions specific to groups. The questionnaire was sent to the legal unit head of the group or to the one identified as best able to respond.
the second version (shortened questionnaire) was for independant enterprises.
The median completion time for the questionnaire was estimated at around 30 minutes.
17.1. Data revision - policy
Should a considerable mistake be noticed after the timely data transmission, a corrected table would be produced and sent.
17.2. Data revision - practice
Should a considerable mistake be noticed after the timely data transmission, a corrected table would be produced and sent. The dataset provided to researcher through the CASD (secure access platform) would also be updated.
If an error is detected in the data or in the analysis published in the website of the French institute (Insee), it will be corrected, even years after, and a note will explain the nature of the error and the date of correction. These cases are very rare.
18.1. Source data
The primary data source is a dedicated survey of enterprises in NACE sections B to N with 50 or more employees or self-employed persons. The sampling frame for the GVC survey was constructed from national statistical business register reference database as of May 2024, containing both legal units (UL) and profiled enterprises (EP).
The filters applied to the reference database are as follows:
Independent legal units or profiled enterprises (legal units within the scope of profiled enterprises are not directly included in the sampling frame);
Employment size of 50 or more employees or self-employed persons;
NACE sections from B to N;
Activity status = 1 (active);
Market and operating units;
Units created before January 1st 2023;
Units located in France, including overseas regions.
The final sampling frame contains 24,911 units.
18.2. Frequency of data collection
The GVC data are collected every three years in accordance with Implementing Regulation (EU) 2022/918.
18.3. Data collection
Data collection took place from September 24 to January 25, i. e. around 4 months, to allow enterprises sufficient time to complete the questionnaire.
67% of enterprises sampled responded to the survey, almost all through a secure online platform. Less than 1% responded by a paper questionnaire, which was sent upon request.
During the data collection phase, a team of survey officers reviewed the responses provided by enterprises and contacted them whenever inconsistencies were detected or when key questions were left unanswered (for example, the number of jobs lost when enterprises said they outsourced).
Survey officers targeted the largest enterprises and those reporting globalisation activities.
18.4. Data validation
During the data collection phase, a team of survey officers reviewed the responses provided by enterprises and contacted them whenever inconsistencies were detected or when key questions were left unanswered (for example, the number of jobs lost when enterprises said they outsourced). Survey officers targeted the largest enterprises and those reporting globalisation activities.
Immediately after data collection, a data cleaning process was implemented to check the internal consistency of the questionnaires (for example, compliance with filter questions, breakdown of employees).
During post-collection processing, checks were carried out to verify the orders of magnitude, notably by comparing certain aggregates with those from the previous GVC survey. The dispersion of the weights was also carefully examined.
During the preparation of the data submission to Eurostat, numerous checks were carried out on the aggregates using logical rules provided by Eurostat concerning for example size or geographic areas.
18.5. Data compilation
After data collection, the first step was to check if the enterprises were still in the scope of the survey. Among the 15,038 enterprises sampled, only 24 were put out of scope, in terms of size, activity, or ceased activity.
At the time of sampling, certain units were identified, based on their importance within their activity sector with regard to the survey topic. These units (108) are non-substitutable. In the case of complete non-response, they are imputed. Total non-response for other units is handled through reweighting within homogeneous response groups (HRGs), which consist of units with similar characteristics (size, region of location, turnover, export turnover, number of legal units for profiled enterprises, NACE activity, year of creation, type of enterprise such as independent, French group,foreign group). Sample calibration methods were also used. They consist of adjusting the weights so that estimates of totals for quantitative variables match the true totals known from other sources, while not deviating too much from the weights derived from total non-response.
Some responding units may sometimes fail to answer all the questions relevant to them. For these variables, and for all variables of non-substitutable units, a method for correcting partial non-response is implemented. Correction of the filter variables is performed first. Two kinds of imputations are implemented: use of deterministic rules, wich are applied when the answer to a question can be inferred from another, for example a total for which each component of the sum is known, and the donor imputation method (hot-deck or nearest neighbor). This method allows, in particular, the correction of several closely related variables using the same donor (as in questions with multiple items, for example). Non-substituable units cannot be donors.
During the partial non-response correction, the consistency checks implemented during the data cleaning conducted immediately after collection are also carried out.
18.6. Adjustment
No adjustments are applied.
No additional comments were identified at this stage.
The data from the Global Value Chains Survey come from a survey conducted by the French National Institute of Statistics among enterprises with at least 50 employees or self-employed persons, operating in mainly market-oriented, non-agricultural sectors (i.e. sections B to N of the NACE Rev. 2 classification). This survey, conducted every three years, aims to improve knowledge of firms’ internationalization strategies.
The survey examines the fragmentation of production processes implemented by enterprises, leading them to outsource activities or relocate activities previously carried out abroad. It also covers cross-border purchases and supplies of goods, materials and services.
The GVC Survey is a European-initiative survey, mandatory for the first time in 2023.
The GVC dataset contains ten variables defined in Implementing Regulation (EU) 2022/918. These cover:
(1) employment by business function,
(2-5) enterprises purchasing or supplying goods and services abroad,
(6) the number of enterprises carrying out international sourcing,
(7-8) jobs created and lost resulting from international sourcing,
(9) motivations and barriers for sourcing, and
(10) events affecting GVC arrangements.
For variables (2)-(5), the dataset includes only enterprises that report at least 100,000 euros in goods or services purchased from or supplied abroad in the final year of the reference period.
All data are provided as absolute values and presented by NACE activity, business function, geographical partner area, size class, motivation and barrier types, and other breakdowns required by the regulation.
The data refer to 2023 for variables 1-5, and to the period 2021–2023 for the other variables.
30 November 2025
All concepts and definitions follow Implementing Regulation (EU) 2022/918 and the GVC Compilers’ Manual. International sourcing refers to the partial or full movement of business functions abroad within or outside the enterprise group, during 2021-2023. Variables (2)-(5) apply only to enterprises exceeding the 100,000 euros threshold for goods/services purchased or supplied abroad. Business functions are divided into core and support categories and specific functional areas (e.g. ICT, management and administration). Motivations and barriers follow the fixed lists provided in the regulation, and importance factors use the standard four-level scale (very important, moderately important, not important, not applicable/do not know).
The structure of the questionnaire is divided into several topics, among wich are the following ones :
Business function
Domestic sourcing
Global value chain arrangements
International sourcing
Backsourcing
Motivation and barriers for international sourcing and backsourcing
Impacts of recent events on economic globalisation.
The statistical unit used for the GVC statistics is the enterprise, as defined in Regulation (EU) 2019/2152 on European business statistics and the Business Register Regulation.
The enterprise is the smallest combination of legal units that is an organisational unit producing goods or services, which benefits from a certain degree of autonomy in decision-making, especially for the allocation of its current resources. An enterprise carries out one or more activities at one or more locations. An enterprise may be a sole legal unit.
The target population consists of all enterprises in NACE Rev. 2 sections B to N, with at least 50 employees and self-employed persons at the end of 2023. For variables (2)-(5), only enterprises reporting at least 100,000 euros of goods or services purchased from or supplied abroad are included, in line with the Annex of Implementing Regulation (EU) 2022/918.
France (all territory, including overseas regions)
The reference period for the GVC statistics is 2021-2023, in accordance with Implementing Regulation (EU) 2022/918.
2023 is the reference year for global value chains arrangements and business functions (variables 1 to 5).
For the questions on international sourcing, backsourcing, domestic sourcing and the effects of recent events, the reference period runs from 2021 to the end of 2023 (variables 6 to 10).
Overall accuracy is assessed using checks on coverage, internal coherence and consistency with related business statistics.
Some uncertainty may remain for the qualitative variables that require a subjective response, particularly regarding the impact of global events. In addition, international sourcing and backsourcing are relatively rare phenomena, so errors related to sampling and post-collection processing (such as weighting of non-respondents) may persist.
Absolute values. This unit applies uniformly to all variables and all breakdowns required by the regulation.
According to the variables, it will represent enterprises, jobs or employees and self-employed persons.
After data collection, the first step was to check if the enterprises were still in the scope of the survey. Among the 15,038 enterprises sampled, only 24 were put out of scope, in terms of size, activity, or ceased activity.
At the time of sampling, certain units were identified, based on their importance within their activity sector with regard to the survey topic. These units (108) are non-substitutable. In the case of complete non-response, they are imputed. Total non-response for other units is handled through reweighting within homogeneous response groups (HRGs), which consist of units with similar characteristics (size, region of location, turnover, export turnover, number of legal units for profiled enterprises, NACE activity, year of creation, type of enterprise such as independent, French group,foreign group). Sample calibration methods were also used. They consist of adjusting the weights so that estimates of totals for quantitative variables match the true totals known from other sources, while not deviating too much from the weights derived from total non-response.
Some responding units may sometimes fail to answer all the questions relevant to them. For these variables, and for all variables of non-substitutable units, a method for correcting partial non-response is implemented. Correction of the filter variables is performed first. Two kinds of imputations are implemented: use of deterministic rules, wich are applied when the answer to a question can be inferred from another, for example a total for which each component of the sum is known, and the donor imputation method (hot-deck or nearest neighbor). This method allows, in particular, the correction of several closely related variables using the same donor (as in questions with multiple items, for example). Non-substituable units cannot be donors.
During the partial non-response correction, the consistency checks implemented during the data cleaning conducted immediately after collection are also carried out.
The primary data source is a dedicated survey of enterprises in NACE sections B to N with 50 or more employees or self-employed persons. The sampling frame for the GVC survey was constructed from national statistical business register reference database as of May 2024, containing both legal units (UL) and profiled enterprises (EP).
The filters applied to the reference database are as follows:
Independent legal units or profiled enterprises (legal units within the scope of profiled enterprises are not directly included in the sampling frame);
Employment size of 50 or more employees or self-employed persons;
NACE sections from B to N;
Activity status = 1 (active);
Market and operating units;
Units created before January 1st 2023;
Units located in France, including overseas regions.
The final sampling frame contains 24,911 units.
Triennial.
Data were transmitted to Eurostat within the legal deadlines: data were sent on september 24th of 2025, i.e. 6 days before the deadline.
The French questionnaire is a translation of the model questionnaire, and the methodological instructions provided by Eurostat (coverage, unit surveyed, etc.) were followed. The results are therefore comparable with those of other countries, subject to minor translation differences and to the differentiated handling of non-response.
This survey is the fourth conducted in France, after 2012, 2016, and 2020. In 2016, the survey was limited to small and medium-sized enterprises.
The questionnaire and methodology have changed little between GVC 2020 and GVC 2023, so the results are comparable.
