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.
Institut National de la Statistique et des Etudes Economiques (French NSI).
1.2. Contact organisation unit
Direction des statistiques d'entreprises \ Département des statistiques de court terme \ Division des indices de prix à la production.
(Directorate of Business Statistics\ Department of Short Term Statistics\ Division of Producer Price Indices).
1.3. Contact name
Confidential because of GDPR
1.4. Contact person function
Confidential because of GDPR
1.5. Contact mail address
INSEE
Division IPP - timbre E310
88 avenue Verdier
CS 70058
92541 Montrouge cedex
France
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
24 June 2025
2.2. Metadata last posted
24 June 2025
2.3. Metadata last update
24 June 2025
3.1. Data description
Name of indicator/source: Import prices of industrial products, collected by the OPISE (Observation of Prices in Industry and SErvices) survey.
3.2. Classification system
Products are classified according to the French Product Classification (Classification des Produits Française : CPF rev. 2.1), a French adaptation of the European Classification of Products by Activity (CPA rev. 2.1), but using a more detailed breakdown.
3.3. Coverage - sector
Activities covered: For import price indices of industrial products, the EBS (European business statistics) regulation requirements comprise sections B to D of the CPA revision 2 of 2008, with the exclusion of the following products:
B07.21
Uranium and thorium ores
B09
Mining support services
C18
Printing and recording services
C24.46
Processed nuclear fuel
C25.4
Weapons and ammunition
C30.1
Ships and boats
C30.3
Air and spacecraft and related machinery
C30.4
Military fighting vehicles
C33
Repair and installation services of machinery and equipment
Not all products are covered by the survey, even among those requested by the EBS regulation. Conversely, import prices for C30.1 are released, even though this is not requested by the EBS regulation.
When prices are not gathered for an industry, they are imputed, so that the calculation of the "total industry" import prices takes into account the entire scope.
Importance of classes covered: In 2024, the OPISE survey covered about 94.9% of imports in the European scope and 93.3% of total industry (sections B to D).
In principle, only series whose response rate to the OPISE survey exceed 50% are disseminated.
3.4. Statistical concepts and definitions
Import price indices of industrial products track the development of transaction prices (converted to euros, therefore including the exchange effects) of goods and services from foreign industrial (and trade) activities sold on the French market. The measurement covers as much as possible cost, insurance and freight (CIF) prices, which is to say that it is assessed at the French border exclusive of import taxes, in line with European regulation No 2020/1197 (nonetheless, the price is extracted from firms databases, therefore it is more an « invoice price » than a « CIF price »). All imports are taken into account, whether they are internal to groups or not. Field surveyors regularly contact firms in order to define the elementary products and the kinds of “markets” to be included in the basket. The selected transactions are representative of the price change of a family of products. In practice, transactions are products with the biggest turnover within each family. The aim of Insee is to obtain a price close to the real price of transaction.
Planned changes in information collected: None.
Accounting conventions: The reported price often corresponds to the average price of the product bought over one calendar month.
3.5. Statistical unit
Reporting unit: Firms. Sometimes, several departments of a single firm (due to their organization) or a head office of a group for its affiliates.
Observation unit(s): Family of transactions = fine items of products crossed with a typology of "markets" (countries, kinds of providers).
3.6. Statistical population
The statistical population is defined by the customs statistics. There were 184,000 French firms importing industrial products in 2023.
3.7. Reference area
The survey covers French economic territory, including Corsica and overseas departments (Guadeloupe, French Guyana, Martinique, Mayotte and Réunion). Conversely, overseas territories (New Caledonia, French Polynesia, Saint Pierre and Miquelon, Wallis and Futuna) are excluded.
3.8. Coverage - Time
Date of first use as a source: The first import index was published in June 2007. Import prices were then backcasted by imports Unit Value Indices.
3.9. Base period
It is necessary to distinguish two different levels of aggregation: aggregation of representative transactions to four-digit CPF rev 2.1 levels and aggregation of CPF rev 2.1 levels to a higher CPF rev 2.1 level. These two levels of aggregation are called respectively basic aggregations and higher aggregations.
For basic aggregations, weights correspond to turnover of the year preceding the contact by the field officer.
For higher aggregations, indices are chain-linked. Weights used for compiling year Y indices stem from year Y-2 national accounts when Y>=2012, and from Y-1 national accounts when 2006<=Y<=2011. Before 2006, weights correspond to 2005 national accounts. The reference of production price indices is 100 in 2021 (annual average), so indices are labeled "base 2021".
Indices (reference 100 in 2021) on prices expressed in euros.
The periodicity is monthly.
6.1. Institutional Mandate - legal acts and other agreements
Legal basis:
The producer price indices are ruled by the European Business Statistics (EBS) Regulation (EU) 2019/2152 on 27 November 2019, followed by the Commission Implementing Regulation 2020/1197.
A yearly visa from the Minister of economy is given, according to the 1951 law about statistics (2025 visa: 2025M041EC)
This visa also covers producer prices indices in industry, services producer price indices and buildings’ maintenance and improvement work price index (~ division 43).
Obligation on units to provide data: The survey is compulsory. Non-respondents are sometimes fined.
Planned changes in legal basis, obligation to respond and frame used: None.
6.2. Institutional Mandate - data sharing
The statistical service of French customs provides Insee with a detailed file by firm, kind of product and partner country.
The data is shared with Eurostat, following the EBS regulation.
These industrial import prices, combined with agricultural import prices, provide import prices by SITC sections, which OECD downloads on the Insee website.
7.1. Confidentiality - policy
Dissemination of terms and conditions under which official statistics are produced, including confidentiality of individual responses:
The import industrial price indices are regulated by the 1951 law about statistics. Every surveyed firm is provided with information on statistical confidentiality.
7.2. Confidentiality - data treatment
Confidential data are not published. That happens in two cases: either less than three units purchase products of the index or one single unit represents more than 85% of the purchases in the product family.
8.1. Release calendar
Advance dissemination of release calendar : Import industrial price indices are published around m+30 (last working day of the following month).
8.2. Release calendar access
A monthly calendar is produced for the four following months. It can be found on the INSEE website.
8.3. Release policy - user access
Indices disseminated on the Insee website and data transmitted to Eurostat are strictly identical.
Data are transmitted to Eurostat the day before the national release (under "embargo"). The transmission to Eurostat uses the SDMX format.
Government has no access to data under embargo.
Monthly
10.1. Dissemination format - News release
A 5-page "Informations Rapides" is published on the INSEE website.
It covers domestic PPI (Producer price index), non-domestic PPI and import prices.
10.2. Dissemination format - Publications
National paper publications: No longer available on paper.
National electronic dissemination: See above our comments on press release; the 5-page "Informations Rapides" is on the INSEE website: both in French and in English.
A "quality file" for year 2014 is available in French on the Insee website (Gestion qualite).
11.1. Quality assurance
The European statistics code of practice is implemented as far as possible. In particular, selective editing is implemented in accordance with the EDIMBUS (Editing and Imputation in Cross-sectional Business Surveys) manual.
11.2. Quality management - assessment
The main quality problems are linked to one-off products, misunderstandings between firms and Insee about some elementary price series and non-response.
The DQAF (Data Quality Assessment Framework) of French PPI (very similar to import prices for source and technique) can be found on the IMF framework.
12.1. Relevance - User Needs
Uses of import price indices :
the European Central Bank (ECB) analyzes flows inside and outside the euro area
National Accounts use them as deflators
moreover, businesses may use some of them for contract indexation
businesses or macroeconomists can use them to track movements of import prices and materials' costs in their field, and evaluate their competitiveness compared to their competitors
international organisations such as Eurostat can compare them among Member States
12.2. Relevance - User Satisfaction
Several complaints are expressed every year about some PPI series by firms that use them for contract indexation.
12.3. Completeness
The OPISE survey covers CPA4 industries amounting to 94.9% of 2023 imports according to the European scope.
As far as we know, all Eurostat requirements are fulfilled.
13.1. Accuracy - overall
The main sources of error are the following ones:
non probability-sampling error (a "cut-off" technique is used)
aggregation of indices based on too few representative products
misunderstandings concerning price questionnaires
non-response
errors in data processing
handling of quality changes in products
handling of new products entering and old one exiting from the market
measurement difficulties in unique/one-off products
uncertainty of imputation models
The revision of "total industry" import price indices is rarely higher than 0.2% a month in absolute terms. No bias on revisions has been observed.
13.2. Sampling error
There is no sampling error strictly speaking as France uses a cut-off technique to select firms.
In every covered CPF4 commodity-group, main purchasers are surveyed. This could be a source of bias.
13.3. Non-sampling error
Multiple representative products may be identified when field surveyors contactfirms. Customs database is used for sampling, which can avoid both undercoverage and limit overcoverage when an activity is "refreshed". As time goes by, undercoverage may grow: new importing firms are not surveyed, new products are bought, and they are seldom surveyed (firms are able to "change products"', but they don't do it frequently).
The list of representative products are discussed between firms and field surveyors: this helps improve the response rate and the quality of the survey. Then, firms fill prices every month, on an Internet (or paper) form. If necessary, field surveyors are likely to contact the data provider long after the contact.
The main reasons for non-response are: representative products imperfectly defined, no purchase during the period under review, difficulties to estimate prices because of an unsuitable information system, change of contact inside firms, momentary "oblivion"... Non-responses are estimated through close groups of products, so as to limit their impact on aggregates.
The weighted response rate is approximately 80%. Some actions to speed up or increase the rate of response have been implemented. Web-based data collection offers a quicker response time and is favoured. In 2024 approximately 98% of surveys were sent by Internet.
Data editing focus on price changes that have the greatest impact on aggregates. Consequently, other errors may not be seen.
No specific models are used.
14.1. Timeliness
Timeliness: Indices are released near the 30th of the month following the reference month. They are provisional for three months. All markets, fine details and aggregates are available at the same time.
Timetable of data collection: Questionnaires for month M prices are sent at the end of month M. The response is to be made before the 10th of M+1. For internet respondents, if they do not complete the questionnaire, they can receive up to two reminder e-mails. The first is around the 10 of M+1. If they still do not answer, the second is sent to them around the 20 of M+1. For paper respondents, only one follow-up letter is sent around the 15 of M+1. If necessary, a follow-up call can be made until the deadline.
14.2. Punctuality
French release dates are always respected, with one exception in January 2009 due to a strike of Insee's civil servants.
The provision of price series to Eurostat is generally punctual, but some technical problems have induced an exceptional delay at the end of October 2013 (data were finally sent on the beginning of November 2013, indeed).
15.1. Comparability - geographical
Overseas departments are included in French economic territory, whereas overseas territories are excluded.
15.2. Comparability - over time
Import price indices are aggregated and chain-linked since January 2005. Before, they have been backcasted with the help of import PPI in base and reference 2005, with a fixed base technique, relying on weights of year 2005.
15.3. Coherence - cross domain
Industrial import prices are used to compile quarterly and yearly national accounts.
For yearly national accounts, "implicit import prices" (that is to say the ratio imports in value / imports in volume) may differ from yearly import prices when uses and supply balance require to do so.
15.4. Coherence - internal
No discrepancies between aggregates and sub-aggregates have been observed.
The cost of both PPI and import prices for Insee is about 7.5 Full Time Equivalent of a category A statistician, 21 Full Time Equivalent of category B or C civil servants, plus 7 field surveyors (assimilated to category A).
The average burden is estimated at 15 minutes per month for a respondent.
17.1. Data revision - policy
The policy of revision is strictly identical for both national release and data transmitted to Eurostat.
The revisions are announced at the time of change (an ‘r’ near the index shows that it has been modified since the previous release).
Scheduled revisions :
Indices are released near the 30th of the month following the reference month (M+30).
Firms are still able to change their price for the reference month for three months, so that indices may be updated at M+30, M+60 and M+90 respectively.
They also may be revised when CPF4 weights are updated (usually in May, so that indices from January to March may be revised with the May release).
Non-scheduled revisions :
Afterwards, indices are only revised if a big mistake is discovered. Then, a specific note is published on website (or in French), commenting the reason why and the impact.
17.2. Data revision - practice
Our press release discloses the revision brought to the aggregates' price changes for the 3 previous months.
Generally, consecutive values are close, there are few revisions : less than 0.2 point for aggregates.
Revisions between September 2024 and February 2025
Mean Revision (MR)
Mean Absolute Revision (MAR)
Import prices of industrial goods
0.00
0.04
18.1. Source data
Type of source: OPISE statistical survey, monthly for industrial import price indices.
Frame on which the source is based: Customs statistics.
Sample or census: A two-stage sample is drawn: the first stage selects firms, the second one chooses representative goods among the imports of firms (during a contact between field surveyors and firms).
Criteria for stratification: The largest firms are selected until at least 30% and most often 50% of the group of goods total imports is covered. Units that do not fall within these criteria might be selected to obtain a better coverage of a particular product family, especially when previously selected units appear to be misclassified. The choice of the products is made by field surveyors when they contact the firm. In accordance with the contact person inside the firm, a set of representative products is selected.
Threshold values and percentages: Firms with amounts less than EUR 3 million are generally excluded from the scope.
Frequency of updating the sample: Every 5 years, for each covered CPA class (4-digit level).
18.2. Frequency of data collection
Monthly, sometimes quarterly (especially when the product is not repeatedly purchased every month).
18.3. Data collection
Questionnaires used in the survey: The questionnaire is specific to the firm, but is built with a unique form for the whole sample (including import prices and SPPI (Services producer price index)). The questionnaire is pre-filled with the information previously received from the field surveyors and the prices communicated by firms themselves for previous periods. Available both on paper and on INSEE’s website via a single entry point (available in French only) for business surveys.
Planned changes in national questionnaires: None.
Data collection media: Both postal and electronic data collection are available. Respondents are given access to the website (available in French only).
Planned changes in data collection methodology: Experimentations are currently led to replace the field surveyors by a self-administered survey in some industries.
18.4. Data validation
Selective editing is applied in line with EDIMBUS manual, and a "Canadian questionnaire" is implemented online in order to collect reasons for price changes bigger than 10% in absolute value.
Price development >10%
Influential contribution
Yes
No
Yes
(to expertise by the price manager)
Contributes to expertise score calculation
If validated, the firm price is kept. Otherwise, the price is imputed (the firm price is not kept).
The price manager:
- validates the price change for some answers (1 to 3, some of 6) to the "Canadian questionnaire" without introduction of a quality coefficient,
-does not need to validate for other answers (4, 5 and some of 6) when justified: the software introduces itself a quality coefficient in order to impute price indices with cautious hypotheses
Contributes to expertise score calculation
Is not imputed
No
(not to expertise)
Does not contribute to expertise score calculation
Is imputed if not validated and if no answer to the Canadian questionnaire
Does not contribute to expertise score calculation
Is not imputed
18.5. Data compilation
Estimates for quality change linked with new products:
Many techniques are used, depending on the context:
Monthly overlap whenever possible
Imputation of the average price change in the series (by default)
Prices kept constant (for tariffs preferably)
Imputation of the average price change for the same respondent
Selection of another product from the same respondent
Selection of a similar product from another respondent
Imputation of the price change of the main raw material
When a product disappears, the firm is contacted in order to supply data for a new one.
Estimates for grossing-up to population levels: None.
Type of index: Chain-linked Laspeyres indices in reference 100 in 2021 (annual average), from CPA 4-digit upwards.
Method of weighting and chaining: The weights are proportional to the imports on the relevant zone (euro or non-euro area). This information is obtained from national accounts at aggregated levels and from the customs statistics at CPA 4-digit. Below the CPA 4-digit level, the aggregation is done according to the information gathered by field surveyors during contacts.
18.6. Adjustment
WDA, SA and other calculations
Gross data only. Data are neither working days adjusted nor seasonally adjusted.
IMF DQAF (Data Quality Assessment Framework) on PPI in English can be accessed on this page.
Name of indicator/source: Import prices of industrial products, collected by the OPISE (Observation of Prices in Industry and SErvices) survey.
24 June 2025
Import price indices of industrial products track the development of transaction prices (converted to euros, therefore including the exchange effects) of goods and services from foreign industrial (and trade) activities sold on the French market. The measurement covers as much as possible cost, insurance and freight (CIF) prices, which is to say that it is assessed at the French border exclusive of import taxes, in line with European regulation No 2020/1197 (nonetheless, the price is extracted from firms databases, therefore it is more an « invoice price » than a « CIF price »). All imports are taken into account, whether they are internal to groups or not. Field surveyors regularly contact firms in order to define the elementary products and the kinds of “markets” to be included in the basket. The selected transactions are representative of the price change of a family of products. In practice, transactions are products with the biggest turnover within each family. The aim of Insee is to obtain a price close to the real price of transaction.
Planned changes in information collected: None.
Accounting conventions: The reported price often corresponds to the average price of the product bought over one calendar month.
Reporting unit: Firms. Sometimes, several departments of a single firm (due to their organization) or a head office of a group for its affiliates.
Observation unit(s): Family of transactions = fine items of products crossed with a typology of "markets" (countries, kinds of providers).
The statistical population is defined by the customs statistics. There were 184,000 French firms importing industrial products in 2023.
The survey covers French economic territory, including Corsica and overseas departments (Guadeloupe, French Guyana, Martinique, Mayotte and Réunion). Conversely, overseas territories (New Caledonia, French Polynesia, Saint Pierre and Miquelon, Wallis and Futuna) are excluded.
The periodicity is monthly.
The main sources of error are the following ones:
non probability-sampling error (a "cut-off" technique is used)
aggregation of indices based on too few representative products
misunderstandings concerning price questionnaires
non-response
errors in data processing
handling of quality changes in products
handling of new products entering and old one exiting from the market
measurement difficulties in unique/one-off products
uncertainty of imputation models
The revision of "total industry" import price indices is rarely higher than 0.2% a month in absolute terms. No bias on revisions has been observed.
Indices (reference 100 in 2021) on prices expressed in euros.
Estimates for quality change linked with new products:
Many techniques are used, depending on the context:
Monthly overlap whenever possible
Imputation of the average price change in the series (by default)
Prices kept constant (for tariffs preferably)
Imputation of the average price change for the same respondent
Selection of another product from the same respondent
Selection of a similar product from another respondent
Imputation of the price change of the main raw material
When a product disappears, the firm is contacted in order to supply data for a new one.
Estimates for grossing-up to population levels: None.
Type of index: Chain-linked Laspeyres indices in reference 100 in 2021 (annual average), from CPA 4-digit upwards.
Method of weighting and chaining: The weights are proportional to the imports on the relevant zone (euro or non-euro area). This information is obtained from national accounts at aggregated levels and from the customs statistics at CPA 4-digit. Below the CPA 4-digit level, the aggregation is done according to the information gathered by field surveyors during contacts.
Type of source: OPISE statistical survey, monthly for industrial import price indices.
Frame on which the source is based: Customs statistics.
Sample or census: A two-stage sample is drawn: the first stage selects firms, the second one chooses representative goods among the imports of firms (during a contact between field surveyors and firms).
Criteria for stratification: The largest firms are selected until at least 30% and most often 50% of the group of goods total imports is covered. Units that do not fall within these criteria might be selected to obtain a better coverage of a particular product family, especially when previously selected units appear to be misclassified. The choice of the products is made by field surveyors when they contact the firm. In accordance with the contact person inside the firm, a set of representative products is selected.
Threshold values and percentages: Firms with amounts less than EUR 3 million are generally excluded from the scope.
Frequency of updating the sample: Every 5 years, for each covered CPA class (4-digit level).
Monthly
Timeliness: Indices are released near the 30th of the month following the reference month. They are provisional for three months. All markets, fine details and aggregates are available at the same time.
Timetable of data collection: Questionnaires for month M prices are sent at the end of month M. The response is to be made before the 10th of M+1. For internet respondents, if they do not complete the questionnaire, they can receive up to two reminder e-mails. The first is around the 10 of M+1. If they still do not answer, the second is sent to them around the 20 of M+1. For paper respondents, only one follow-up letter is sent around the 15 of M+1. If necessary, a follow-up call can be made until the deadline.
Overseas departments are included in French economic territory, whereas overseas territories are excluded.
Import price indices are aggregated and chain-linked since January 2005. Before, they have been backcasted with the help of import PPI in base and reference 2005, with a fixed base technique, relying on weights of year 2005.
