1.1. Contact organisation
Ministry of Agriculture, Food Sovereignty (Ministère de l'Agriculture, de la souveraineté Alimentaire)
Department of Statistics and Foresight Analysis (Service de la Statistique et de la Prospective)
1.2. Contact organisation unit
Department of agricultural, forestry and agrifood statistics (Sous-direction des statistiques agricoles, forestières et agroalimentaires: SDSAFA)
Unit on agricultural structures, aquaculture and forestry statistics (Bureau des statistiques sur les structures agricoles, l'aquaculture et la forêt: BSSAF)
1.3. Contact name
Confidential because of GDPR
1.4. Contact person function
Confidential because of GDPR
1.5. Contact mail address
Organisation:
MASA-SG-SSP
3 rue Barbey de Jouy
75349 PARIS 07 SP (France)
Project Unit:
SDSAFA-BSSAF
Complexe agricole d'Auzeville
BP 32688
31326 CASTANET TOLOSAN (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
8 July 2025
2.2. Metadata last posted
1 August 2025
2.3. Metadata last update
8 July 2025
3.1. Data description
The data describe the structure of agricultural holdings providing the general characteristics of farms and farmers and information on their land, livestock and labour force. They also describe production methods, rural development measures and agro-environmental aspects that look at the impact of agriculture on the environment.
The data are used by public, researchers, farmers and policy-makers to better understand the state of the farming sector and the impact of agriculture on the environment. The data follow up the changes in the agricultural sector and provide a basis for decision-making in the Common Agricultural Policy (CAP) and other European Union policies.
The statistical unit is the agricultural holding (farm). The aggregated results are disseminated through statistical tables. The data are presented at different geographical levels and over periods.
The data collections are organised in line with Regulation (EU) 2018/1091 and have a new structure, consisting of a core data set and several modules. The regulation covers the data collections in 2019/2020 (the agricultural census), 2023 and 2026. The data are as comparable and coherent as possible with the other European countries.
3.2. Classification system
Data are arranged in tables using many classifications. Please find below information on most classifications.
The classifications of variables are available in Annex III of Regulation (EU) 2018/1091 and in Commission Implementing Regulation (EU) 2021/2286.
The farm typology means a uniform classification of the holdings based on their type of farming and their economic size. Both are determined on the basis of the standard gross margin (SGM) (until 2007) or standard output (SO) (from 2010 onward) which is calculated for each crop and animal. The farm type is determined by the relative contribution of the different productions to the total standard gross margin or the standard output of the holding.
The territorial classification uses the NUTS classification to break down the regional data. The regional data is available at NUTS level 2.
3.3. Coverage - sector
The statistics cover agricultural holdings undertaking agricultural activities as listed in item 3.5 below and meeting the minimum coverage requirements (thresholds) as listed in item 3.6 below.
3.4. Statistical concepts and definitions
The list of core variables is set in Annex III of Regulation (EU) 2018/1091.
The descriptions of the core variables as well as the lists and descriptions of the variables for the modules collected in 2023 are set in Commission Implementing Regulation (EU) 2021/2286.
The following groups of variables are collected in 2023:
- for core: location of the holding, legal personality of the holding, manager, type of tenure of the utilised agricultural area, variables of land, organic farming, irrigation on cultivated outdoor area, variables of livestock, organic production methods applied to animal production;
- for the module "Labour force and other gainful activities": farm management, family labour force, non-family labour force, other gainful activities directly and not directly related to the agricultural holding;
- for the module "Rural development": support received by agricultural holdings through various rural development measures;
- for the module “Irrigation”: availability of irrigation, irrigation methods, sources of irrigation water, technical parameters of the irrigation equipment, crops irrigated during a 12 months period;
- for the module “Soil management practices”: tillage methods, soil cover on arable land, crop rotation on arable land, ecological focus area;
- for the module “Machinery and equipment”: internet facilities, basic machinery, use of precision farming, machinery for livestock management, storage for agricultural products, equipment used for production of renewable energy on agricultural holdings;
- for the module “Orchards”: apples area, pears area, peaches area, nectarines area, apricots area, small citrus fruit area, lemons area, olives area, grapes for table use area, each one by age of plantation and density of trees.
3.5. Statistical unit
See sub-category below.
3.5.1. Definition of agricultural holding
The agricultural holding is a single unit, both technically and economically, that has a single management and that undertakes economic activities in agriculture in accordance with Regulation (EC) No 1893/2006 belonging to groups:
- A.01.1: Growing of non-perennial crops
- A.01.2: Growing of perennial crops
- A.01.3: Plant propagation
- A.01.4: Animal production
- A.01.5: Mixed farming or
- The “maintenance of agricultural land in good agricultural and environmental condition” of group A.01.6 within the economic territory of the Union, either as its primary or secondary activity.
Regarding activities of class A.01.49, only the activities “Raising and breeding of semi-domesticated or other live animals” (with the exception of raising of insects) and “Bee-keeping and production of honey and beeswax” are included.
3.6. Statistical population
See sub-categories below.
3.6.1. Population covered by the core data sent to Eurostat (main frame and if applicable frame extension)
The thresholds of agricultural holdings are available in the annex.
Annexes:
3.6.1. Thresholds of agricultural holdings
3.6.1.1. Raised thresholds compared to Regulation (EU) 2018/1091
No3.6.1.2. Lowered and/or additional thresholds compared to Regulation (EU) 2018/1091
No3.6.2. Population covered by the data sent to Eurostat for the modules “Labour force and other gainful activities”, “Rural development” and “Machinery and equipment”
The same population of agricultural holdings defined in item 3.6.1.
3.6.3. Population covered by the data sent to Eurostat for the module “Animal housing and manure management”
Restricted from publication
3.6.4. Population covered by the data sent to Eurostat for the module “Irrigation”
The subset of agricultural holdings defined in item 3.6.2 with outside irrigated area and/or areas under glass or high accessible coverage.
Please note that, if the irrigated area is only based on kitchen garden, we did not fulfil the irrigation module (too many technical questions, not accessible for these farmers). Moreover, concerning rice, this crop is not systematically irrigated in the overseas territories; so, we did not ask about the irrigation module for these particular cases.
3.6.5. Population covered by the data sent to Eurostat for the module “Soil management practices”
The subset of agricultural holdings defined in item 3.6.2 with outside arable land.
3.6.6. Population covered by the data sent to Eurostat for the module “Orchard”
The subset of agricultural holdings defined in item 3.6.2, with any of the individual orchard variables that meet the threshold specified in Article 7(5) of Regulation (EU) 2018/1091.
3.6.7. Population covered by the data sent to Eurostat for the module “Vineyard”
Restricted from publication
3.7. Reference area
See sub-categories below.
3.7.1. Geographical area covered
The entire territory of the country.
3.7.2. Inclusion of special territories
Overseas territories are included:
- French Guiana
- Guadeloupe
- Martinique
- Réunion
- Mayotte
3.7.3. Criteria used to establish the geographical location of the holding
The main building for productionThe most important parcel by physical size
3.7.4. Additional information reference area
For farms without buildings: the most important parcel by physical size:
- from IACS, if the farm is included in IACS;
- with a collected physical address, if the farm is not included in IACS.
3.8. Coverage - Time
Farm structure statistics in our country cover the period from 1955 onwards. Older time series are described in the previous quality reports (national methodological reports).
3.9. Base period
The 2023 data are processed (by Eurostat) with 2020 standard output coefficients (calculated as a 5-year average of the period 2018-2022). For more information, you can consult the definition of the standard output.
Two kinds of units are generally used:
- the units of measurement for the variables (area in hectares, livestock in heads or LSU (livestock units), labour force in persons or AWU (annual working units), standard output in Euro, places for animal housing etc.) and
- the number of agricultural holdings having these characteristics.
See sub-categories below.
5.1. Reference period for land variables
The use of land refers to the reference year 2023:
- IACS 2023 is used for IFS 2023, which registers the crops present on the 15 of May 2023; nevertheless, pre-recorded values can be modified during the data collection period
- For farms not in IACS, we register the harvested crops in 2023, referring to the crop year from 01 November 2022 to 31 October 2023
- In the case of successive crops from the same piece of land, the land use refers to a crop that is harvested during 2023, or, if many, the one with the highest commercial value
5.2. Reference period for variables on irrigation and soil management practices
Reference crop year from 01 November 2022 to 31 October 2023.
5.3. Reference day for variables on livestock and animal housing
The reference day is 01 November 2023 for livestock variables. In case of sanitary vacuum on this date (poultry and piglets), the most recent livestock present on the farm is considered (usual advice for each survey).
The animal housing variables are not applicable for 2023.
5.4. Reference period for variables on manure management
The manure management variables are not applicable for 2023.
5.5. Reference period for variables on labour force
From 01 November 2022 to 31 October 2023.
5.6. Reference period for variables on rural development measures
The three-year period ending on 31 December 2023.
5.7. Reference day for all other variables
The reference day is, by default, 01 November within the reference year 2023.
6.1. Institutional Mandate - legal acts and other agreements
See sub-categories below.
6.1.1. National legal acts and other agreements
Legal actOther formal agreement
6.1.2. Name of national legal acts and other agreements
- Decree (of 13 July 2023) concerning the 2023 agricultural survey: Order of 13 July 2023 supplementing the order of 24 October 2022, approving the program of national or regional statistical surveys of public services for 2023 (surveys of businesses and farms)
- Approval of the National Council of Statistical Information, for the 2023 agricultural IFS survey, as a mandatory survey of general interest.
6.1.3. Link to national legal acts and other agreements
- Decree (of 13 July 2023) concerning the 2023 agricultural survey
- Approval of the National Council of Statistical Information
6.1.4. Year of entry into force of national legal acts and other agreements
Both legal acts mentioned in items 6.1.2 and 6.1.3 entered into force in 2023.
6.1.5. Legal obligations for respondents
Yes6.2. Institutional Mandate - data sharing
- There is a general agreement with the French National Payment Agency (ASP) for IACS data access.
- There is a general agreement with the Ministry of Agriculture, Food Sovereignty / General Direction for Food (DGAL) for bovine register data access.
- There is a specific agreement since 2021 with the National Research Institute for Agriculture, Food and Environment (INRAE-ODR) for processing rural development module data.
All these agreements are private, and thus not publicly available.
7.1. Confidentiality - policy
- In accordance with the Statistical Confidentiality Act (No 51-711 of 7 June 1951), the data collected are professional, and the interviewers and statisticians are bound by professional secrecy.
- Personal data are protected by the French Data Protection Act (No 78-17 of 6 January 1978) and the General Data Protection Regulation (Regulation (EU) 2016/679 of 27 April 2016).
7.2. Confidentiality - data treatment
See sub-categories below.
7.2.1. Aggregated data
See sub-categories below.
7.2.1.1. Rules used to identify confidential cells
Threshold rule (The number of contributors is less than a pre-specified threshold)Dominance rule (The n largest contributions make up for more than k% of the cell total)
Secondary confidentiality rules
7.2.1.2. Methods to protect data in confidential cells
Cell suppression (Completely suppress the value of some cells)7.2.1.3. Description of rules and methods
According to the threshold rule, the minimal number of contributors in a table cell should be 3, in order to publish the extrapolated total of that table cell.
According to the dominance rule, the contribution of the biggest contributor should be less than 85% of the extrapolated total of a table cell, in order to publish the extrapolated total of that table cell.
In the tables, data from fewer than three units and data where a single unit represents more than 85% of the information are replaced by "s" (secret).
For secondary confidentiality, we check all tables in order to prevent the recalculation of confidential values ("s" values) for both linked and isolated tables. If need be, one or more other cells are also marked with "s" to avoid recalculation.
7.2.2. Microdata
See sub-categories below.
7.2.2.1. Use of EU methodology for microdata dissemination
No7.2.2.2. Methods of perturbation
None7.2.2.3. Description of methodology
IFS 2023 national microdata files can be accessed for research purposes.
Access to microdata files (secure use files) for the needs of researchers is granted under an authorisation procedure, which is governed by law and requires a series of steps to be taken by the Statistical Confidentiality Committee.
The Statistical Confidentiality Committee ensures compliance with the rules of statistical confidentiality and gives its opinion on requests to communicate individual data collected during statistical surveys or transmitted to the official statistical system for the purpose of compiling statistics.
The remit of the Statistical Confidentiality Committee is set out in Law no. 51-711 of 7 June 1951 on legal obligation, coordination and confidentiality in the field of statistics.
In order to access confidential data under the Committee on Statistical Confidentiality, you must submit a request through the online request management portal.
The producer service (SSP) will examine and assess your application in the light of the conditions laid down in the opinion of the Committee on Statistical Confidentiality. Under the terms of Article 17 of Decree No. 2009-318 of 20 March 2009 on the National Council for Statistical Information, the Statistical Confidentiality Committee and the Committee for the Official Statistics Label, “the Statistical Confidentiality Committee shall deliver its opinion taking into account the nature and relevance of the work for which the application is made, the status of the person or body submitting the application and the guarantees it offers. It checks that the volume of information requested is not excessive in relation to the work justifying its communication and that this does not lead to excessive prejudice to the interests that the Act [No. 51-711 of 7 June 1951 as amended on the obligation, coordination and secrecy of statistics] was intended to protect.”
The use of each source must therefore be justified in relation to the objectives of the project. If you do not meet these requirements, the producer services may refuse their agreement to the examination of your file by the Committee. On the other hand, this agreement does not presage the final opinion that the Committee will express.
If the application has received the agreement of all the producer departments concerned, the secretariat will register it for submission to the Committee on Statistical Confidentiality for its opinion.
If the Committee’s opinion is favourable, you will then be able to access the requested data after the formalities for signature by the competent authorities and, for sources disseminated via the CASD (Centre for Secure Access to Data), after contractualisation between the CASD and your institution and a registration session (registration).
8.1. Release calendar
There is no online release calendar for IFS 2023.
8.2. Release calendar access
Not applicable.
8.3. Release policy - user access
The Statistical and Foresight Department of the Ministry of Agriculture, Food Sovereignty does not have a general release policy. In order to inform users, the department sends a newsletter by email. To receive it, you need to be registered; requests can be made at this website.
No specific policy is put into force concerning IFS.
For IFS 2023, neither a specific calendar nor information has been given to respondents or data users.
8.3.1. Use of quality rating system
Yes, another quality rating system8.3.1.1. Description of the quality rating system
The data is published with its 95% confidence interval, whatever the relative precision. This allows users to have access to the data while being aware of the accuracy.
In addition, if a definite change can be observed in relation to the 2020 census data, the data is marked in green, otherwise in red (irrespective of the accuracy of the data).
Data dissemination for farm statistics done after each survey, every 3 or 4 years.
10.1. Dissemination format - News release
See sub-categories below.
10.1.1. Publication of news releases
No10.1.2. Link to news releases
Not applicable.
10.2. Dissemination format - Publications
See sub-categories below.
10.2.1. Production of paper publications
No10.2.2. Production of on-line publications
No10.2.3. Title, publisher, year and link
Not applicable.
10.3. Dissemination format - online database
See sub-categories below.
10.3.1. Data tables - consultations
Not applicable.
10.3.2. Accessibility of online database
No10.3.3. Link to online database
Not applicable.
10.4. Dissemination format - microdata access
See sub-category below.
10.4.1. Accessibility of microdata
Yes10.5. Dissemination format - other
Regional statistical services have their own websites and may publish specific studies. There are no actual examples available.
10.5.1. Metadata - consultations
Not requested.
10.6. Documentation on methodology
See sub-categories below.
10.6.1. Metadata completeness - rate
Not requested.
10.6.2. Availability of national reference metadata
Yes10.6.3. Title, publisher, year and link to national reference metadata
- Methodological document explaining standard outputs coefficients and permanent pasture definitions, published by the Statistical and Foresight Department of the Ministry of Agriculture, Food Sovereignty.
- Methodological document for researchers (with microdata): not yet available (expected for September 2025)
10.6.4. Availability of national handbook on methodology
Yes10.6.5. Title, publisher, year and link to handbook
The IFS 2023 national handbook is not online, but it is available to researchers with microdata access.
10.6.6. Availability of national methodological papers
No10.6.7. Title, publisher, year and link to methodological papers
Not applicable.
10.7. Quality management - documentation
Not available.
11.1. Quality assurance
See sub-categories below.
11.1.1. Quality management system
Yes11.1.2. Quality assurance and assessment procedures
Designated quality manager, quality unit and/or senior level committee11.1.3. Description of the quality management system and procedures
The Ministry of Agriculture's Statistics and Forecasting Service (SSP) has developed a quality reference framework (Guide du projet statistique) based on the Generic Statistical Business Process Model (GSBPM). A 2022-2027 Quality Plan has been drawn up. It includes around fifty actions classified under 4 headings:
- I- Shared investment in administrative and private sources
- II- Documented, optimised and secure processes
- III- Strong focus on the public
- IV- Broad, modern and transparent dissemination and communication
The SSP's quality plan is part of the quality plan for French statistics led by the French national statistics institute (INSEE). The SSP's quality coordinator is the deputy head of the Statistics and Forecasting Service.
11.1.4. Improvements in quality procedures
Our quality plan is based on the 2022-2027 quality strategy for the official statistical service, which includes INSEE and the ministerial statistical services. This quality strategy was drawn up by the French SSP's quality monitoring committee, on which our quality consultant represents the SSP.
INSEE's Quality Unit monitors the action plan. Continuous improvement processes (feedback, user audits, etc.) are integrated into all operations, including IFS 2023.
For example, the project manager for the IFS 2023 survey has toured the regions to gather feedback from the Data Collection units of the regional agricultural statistics services.
Annexes:
11.1.4. Improvements in quality procedures - 2022-2027 quality strategy
11.2. Quality management - assessment
A report was published in August 2020 on the quality of statistical data published by the Statistical and Foresight Department of the Ministry of Agriculture, Food Sovereignty: "Qualité des données statistiques produites par le SSP".
12.1. Relevance - User Needs
A consultation with users of structural survey data was conducted in October 2022 in order to add national questions or items in the IFS survey. The consulted parties included:
- Researchers from the French National Institute for Agriculture, Food and Environment and universities, as well as all researchers with access to the 2020 census data via the safe data access center (CASD),
- Ministerial technical divisions using data, such as the Ministry of Agriculture and the Ministry of Environmental Policy,
- National agencies working in the agricultural field, including FranceAgriMer, Agence Bio, INAO, ASP, and INSEE,
- Professional organisations, such as farmers' unions (FNSEA, Confédération paysanne, Coordination Rurale, Modef),
- Technical agricultural institutes, including Arvalis, IDELE, and IFIP, and
- Regional statistics agencies.
Following the written consultation, which involved over 330 people via email, a final file detailing all requests and their outcomes (accepted, modified, or rejected) was distributed to everyone who responded.
12.1.1. Main groups of variables collected only for national purposes
For national purposes, one main group of variables has been added; it concerns the management of wastes produced on-farm (i.e. how they are treated, reused, burned, or evacuated by specialised firms). The French agency in charge of the waste management organisation asked for this in order to have a better knowledge of the efficacy of their actions and to determine if regional additional enforcement needs to be made.
For many groups of variables, some more detailed answer possibilities have been added (re-aggregated for Eurostat needs). For example, in the equipment module, we do not only ask for tractors, but also for sprayers, combine harvesters and many other machines. This request has been formulated by technical institutes and/or researchers working on specialised subjects to complete their data sources.
12.1.2. Unmet user needs
Some user needs have not been included in our survey because:
- they are not structural data which can vary each year (e.g. amount of nitrogen fertilisation),
- the data are covered by other national surveys like FADN and SAIO,
- the data are available in administrative sources,
- the data are too complex to collect.
Examples of requests not fitted: origin of the energy used on the farm (generally unknown), water consumption per crops (too complicate to collect), tank capacity of the sprayer (available in the database concerning sprayer inspection), year of perception of the support for young farmers (available in IACS files).
12.1.3. Plans for satisfying unmet user needs
For each survey, there is a committee to share user needs and design the questionnaire.
We will have to give better information to the users about the limits of sampled surveys: low accuracy at too detailed geographical levels, and too detailed questions/items that are not significant in relation to the number of records. To enhance user satisfaction, the context must be clearly defined.
12.2. Relevance - User Satisfaction
There are no procedures used to measure user satisfaction for IFS 2023.
12.2.1. User satisfaction survey
No12.2.2. Year of user satisfaction survey
Not applicable.
12.2.3. Satisfaction level
Not applicable12.3. Completeness
Information on not collected, not-significant and not-existent variables is available on Eurostat’s website, at the link: Additional data - Eurostat (europa.eu).
12.3.1. Data completeness - rate
Not applicable for Integrated Farm Statistics as the not collected variables, not-significant variables and not-existent variables are completed with 0.
13.1. Accuracy - overall
See categories below.
13.2. Sampling error
See sub-categories below.
13.2.1. Sampling error - indicators
Please find the relative standard errors on Eurostat’s website, at the link: CircaBC website.
13.2.2. Reasons for non-compliant precision requirements in relation to Regulation (EU) 2018/1091
Some variables have estimated RSE above applicable thresholds, with the RSE computation accounting for calibration. Margins used for calibration include the total number of statistical units in the sampling universe, totals from the bovine identification database, and surfaces from the IACS 2023 data. Consequently, variables correlated with total bovines and surfaces have a lower RSE when accounting for calibration, but not necessarily other variables.
Cases of non-compliance are almost all for livestock excluding bovines, and grapes for table use. In all these cases, this is due to the low quality of the sampling frame for these variables, and the fact the sampling design did not sufficiently account for this low quality. Notably, we do not have reliable administrative data sources on grapes for table use, as we have for grapes for wine. We do not have exhaustive administrative data on sheep, goats and pigs livestock, and in the case of pigs not for the detailed categories of the IFS.
For future IFS surveys, we will quantify the uncertainty surrounding those variables in our sampling frame, and design the sampling accordingly to have better RSE for these variables.
13.2.3. Reference on method of estimation
See in annex.
Annexes:
13.2.3. Methodology used to calculate relative standard errors
13.2.4. Impact of sampling error on data quality
Moderate13.3. Non-sampling error
See sub-categories below.
13.3.1. Coverage error
See sub-categories below.
13.3.1.1. Over-coverage - rate
The over-coverage rate is available on Eurostat’s website, at the link: CircaBC.
The over-coverage rate is unweighted.
The over-coverage rate is calculated as the share of ineligible holdings to the holdings designated for the core data collection. The ineligible holdings include those holdings with unknown eligibility status that are not imputed nor re-weighted for (therefore considered ineligible).
The over-coverage rate is calculated over the holdings in the main frame and if applicable frame extension, for which core data are sent to Eurostat.
13.3.1.1.1. Types of holdings included in the frame but not belonging to the population of the core (main frame and if applicable frame extension)
Below thresholds during the reference periodTemporarily out of production during the reference period
Ceased activities
Merged to another unit
Duplicate units
13.3.1.1.2. Actions to minimize the over-coverage error
Maintain of ineligible units in the records with assignment of 0 weights13.3.1.1.3. Additional information over-coverage error
Over-coverage occurs when units are included erroneously. To minimise the over-coverage error, the agricultural statistical service has been using a sampling frame of farms, which is updated with the results of thematic surveys and administrative data.
To mitigate coverage quality risks, before the IFS survey began, quality checks were performed on uncertain units in the sampling frame (those with undetermined status: ceased or active, agricultural or non-agricultural). Regional ministry staff were asked to verify and, if needed, correct the information in the sampling frame.
13.3.1.2. Common units - proportion
Not requested.
13.3.1.3. Under-coverage error
See sub-categories below.
13.3.1.3.1. Under-coverage rate
To indicate the degree of under-coverage, a numeric rate is not available.
However, units not present in the sampling frame could be included in two cases: non-active sampled units were asked for the identity of their successor, if applicable, and the successor was added to the sample if not already present. In Mayotte, the two-stage sampling allowed for the addition of new units in the second stage.
The number of newly added units was low: 1 247 units, mainly in the overseas departments, with 1 213 units.
These 1 247 holdings, belonging to the relevant population of the core but not included in the frame, are:
- very small farms: 1 011 units,
- small farms: 201 units,
- medium farms: 25 units,
- big farms: 10 units.
These farms are mainly (1 131 out of 1 247) farms with both crops and livestock production (545 units), cereals (377 units) and fruits (209 units).
13.3.1.3.2. Types of holdings belonging to the population of the core but not included in the frame (main frame and if applicable frame extension)
New birthsOther
13.3.1.3.3. Actions to minimise the under-coverage error
The frame used for the survey was built with the maximum of available administrative data sources:
- SIRENE (French business register): all active units with an agricultural activity code in the economic nomenclature,
- Common Agricultural Policy subsidies dataset: information on land and livestock (bovine, goat and sheep heads in subsidised farming) integrated,
- CVI (customs administration vineyard) dataset,
- BDNI (bovine identification national database) and other animal species databases: integration of information on bovine, goat and sheep cattle sizes, coupled with information on porcine cattle moves,
- Mutuelle Sociale Agricole (mandatory social security system for paid and independent agricultural workers): integration of annual information on contributors and paid workers,
- Fiscal information (benefits and micro-benefits): integration of the dataset,
- Other datasets for poorly covered activities with above administrative formalities, such as market gardening, horticulture, organic farming, horse farming and beekeeping.
13.3.1.3.4. Additional information under-coverage error
Due to the registration of all units with an economic activity in the business register, the obligation to fill in the IFS survey and the possibility of fines, the under-coverage is estimated to be very low. Regular checks are performed to ensure the frame's completeness; the annual coverage rate by activity was verified.
However, under-coverage exists for two main reasons: newly created units between the last updates and the field data collection and existing units that are absent from administrative data sources (notably small units and units in overseas territories).
13.3.1.4. Misclassification error
Yes13.3.1.4.1. Actions to minimise the misclassification error
Misclassification errors cannot totally be ruled out but are estimated to be minimal.
Holdings are initially selected from the administrative farm register (BALSA), which is constantly updated with the results of both thematic surveys and administrative data. This information can be incomplete, notably for non-bovine livestock.
Some holdings are misclassified into inappropriate strata in the sampling design. However, the region of the holdings is well-known. Determining the exact true value for farm types and standard outputs is complicated. This is primarily due to the use of classes, such as very small, small, medium, and large, for standard outputs.
For standard outputs, the misclassification rate is estimated at about 10%, and is lower for discontinuous classes (e.g., farms classified as big in the frame but actually very small or small). Regarding farm types, the misclassification rate is also estimated at about 10%.
13.3.1.5. Contact error
Yes13.3.1.5.1. Actions to minimise the contact error
Contact information is continuously updated. Information comes from administrative sources or directly from respondents in thematic surveys. Interviewers may collect additional information when needed.
13.3.1.6. Impact of coverage error on data quality
Low13.3.2. Measurement error
See sub-categories below.
13.3.2.1. List of variables mostly affected by measurement errors
Variables affected by measurements errors.
- Seeds and seedlings - outdoor (E0000T): depending on the crop type, seeds are included or not with the crop area or separated into the specific item "seeds and seedlings". This distinction is often confusing for interviewers and farmers alike.
- Irrigation controller (IR_CTRL): differences between automatic/precision/combined irrigation controllers are not very clear for farmers.
- Variable rate techniques (MAC_VRT): the inclusion and exclusion criteria are not clearly defined (sprayers, seeders, fertilisers, etc.).
13.3.2.2. Causes of measurement errors
Complexity of variablesRespondents’ inability to provide accurate answers
13.3.2.3. Actions to minimise the measurement error
Pre-testing questionnairePre-filled questions
Explanatory notes or handbooks for enumerators or respondents
Training of enumerators
13.3.2.4. Impact of measurement error on data quality
Low13.3.2.5. Additional information measurement error
To prevent measurement errors, the questionnaire was pre-tested, populated with IACS land variables, and surveyors received comprehensive online training prior to data collection.
Moreover, surveyors also had a handbook during data collection, and definitions were included in the electronic questionnaire survey.
13.3.3. Non response error
See sub-categories below.
13.3.3.1. Unit non-response - rate
See item 13.3.1.1.
The unit non-response rate is unweighted.
The unit non-response rate is calculated as the share of eligible non-respondent holdings to the eligible holdings. The eligible holdings include those holdings with unknown eligibility status which are imputed or re-weighted for (therefore considered eligible).
The unit non-response rate is calculated over the holdings in the main frame and if applicable frame extension, for which core data are sent to Eurostat.
13.3.3.1.1. Reasons for unit non-response
Refusal to participate13.3.3.1.2. Actions to minimise or address unit non-response
Follow-up interviewsReminders
Weighting
13.3.3.1.3. Unit non-response analysis
- Unit non-response analysis was conducted during data collection to estimate the size of non-response units (large farms or small farms); and their probability of being active. The goal was not to draw conclusions, but to target units most likely to be large and active for increased groundwork effort on data collection.
- Unit non-response analyses were conducted after data collection by comparing variables of respondents and non-respondents available in the sampling frame; and available in administrative sources. We detail these analyses and their results in the following bullet point.
- Unit non-response was corrected by re-weighting. We tried three different methods (Homogeneous response groups based on logit model, based on classification trees and based on sample design strata) and found very similar results on the variables of interest: eligibility status and land area. Consequently, we used the simplest and usual method of re-weighting by strata. In parallel, we tested models to predict ineligibility due to economic cessation based on sampling frame variables. Some variables were weakly predictive, notably farmer's age and lack of IACS declaration in 2023. The model predicted slightly more economic cessations among non-respondent, indicative of a potential non-response bias, but this observed bias was effectively corrected by the re-weighting procedure.
13.3.3.2. Item non-response - rate
Not available. The item non-response rate was low: generally 0 and unweighted rate <1% in the worst case.
13.3.3.2.1. Variables with the highest item non-response rate
Not available.
13.3.3.2.2. Reasons for item non-response
RefusalFarmers do not know the answer
13.3.3.2.3. Actions to minimise or address item non-response
Follow-up interviewsImputation
Other
13.3.3.3. Impact of non-response error on data quality
Low13.3.3.4. Additional information non-response error
Live checks for mandatory fields, implemented in data collection tools, are used to minimise item non-response.
For half of voluntary fields, live checks were also implemented in data collection tools.
13.3.4. Processing error
See sub-categories below.
13.3.4.1. Sources of processing errors
Data processing13.3.4.2. Imputation methods
Deductive imputationMean imputation
Random hot deck imputation
13.3.4.3. Actions to correct or minimise processing errors
To minimise processing errors, the information system is extensively tested and manual actions are minimised as much as possible. All corrections are made using R scripts (no manual adjustments), and before data is released, extensive checks and analyses are performed (e.g. preliminary data are sent to the regional services to perform a comparison with local information, or check the evolution's tendencies).
13.3.4.4. Tools and staff authorised to make corrections
Standard software tools like R (primarily) and Excel are used.
Only staff involved in the processing of the IFS 2023 is authorised to make corrections.
13.3.4.5. Impact of processing error on data quality
Low13.3.4.6. Additional information processing error
Not available.
13.3.5. Model assumption error
We studied the outliers. A model is used to compare outlier values (e.g. livestock numbers for cattle, sheep, goats, and poultry, or cultivated land areas) with administrative data. Additionally, ratios are calculated to detect inconsistencies between variables. Values are corrected when necessary.
14.1. Timeliness
See sub-categories below.
14.1.1. Time lag - first result
Not applicable, no first results.
14.1.2. Time lag - final result
Final results were published on 30 June 2025: 18 months after 31 December 2023.
14.2. Punctuality
See sub-categories below.
14.2.1. Punctuality - delivery and publication
See sub-categories below.
14.2.1.1. Punctuality - delivery
Not requested.
14.2.1.2. Punctuality - publication
Before data collection, the aim was to publish preliminary results in December 2024 and final data in February 2025.
Finally, no preliminary data have been published because the RSE for all variables was not available on time.
Final results were published on 30 June 2025, 124 days later.
15.1. Comparability - geographical
See sub-categories below.
15.1.1. Asymmetry for mirror flow statistics - coefficient
Not applicable, because there are no mirror flows in Integrated Farm Statistics.
15.1.2. Definition of agricultural holding
See sub-categories below.
15.1.2.1. Deviations from Regulation (EU) 2018/1091
No deviations.
15.1.2.2. Reasons for deviations
Not applicable.
15.1.3. Thresholds of agricultural holdings
See sub-categories below.
15.1.3.1. Proofs that the EU coverage requirements are met
Our national farm register contains all units having an economic activity on the agricultural domain. They are all included in the sampling design. Only those above national physical thresholds (lower than the European ones) have been collected.
For Eurostat, we apply the physical thresholds listed in Annex II of Regulation (EU) 2018/1091.
The IFS 2023 UAA and LSU correspond to IFS 2023 - IFS thresholds for the reference period.
| UAA | LSU | |
|---|---|---|
| IFS 2023 - IFS thresholds | 27 201 242 | 18 112 664 |
| IFS 2023 - national thresholds | 27 220 343 | 18 115 552 |
| IFS 2023 coverage: IFS thresholds vs national thresholds (%) | 99.93% | 99.98% |
15.1.3.2. Differences between the national thresholds and the thresholds used for the data sent to Eurostat
For the thresholds for the data sent to Eurostat, see item 3.6.1.
National thresholds: a holding is covered if it meets at least one of the following thresholds:
For land variables:
- UAA >= 1 ha
- specialised crops (tobacco, hops, aromatic or medicinal plants, seeds, fresh vegetables, flowers and ornamental plants, permanent crops, nurseries) >= 0.20 ha
- potatoes >= 0.50 ha
- cabbage for sauerkraut >= 0.20 ha
- asparagus >= 0.20 ha
- strawberries >= 0.15 ha
- fresh vegetables >= 0.50 ha
- flowers and ornamental plants >= 0.05 ha
- vineyards >= 0.10 ha
- vineyards for champagne >= 0.05 ha
- nurseries >= 0.05 ha
- crops under glass or high accessible cover >= 0.01 ha
For livestock:
- 1 male breeder
- 4 births of equidae animals
- 1 cow
- 2 bovine animals over 2 years old
- 1 breeding sow
- 1 fattening activity (for bovine animals or pigs or sheep or goats or rabbits)
- sheep: 6 breeding females
- goats: 6 breeding females
- rabbits: 10 breeding females
- 100 laying poultry
- incubation capacity: 1000 eggs
- 50 beehives
- 1 other animal production n.e.c.
Other thresholds regarding specific productions:
- 500 broiler poultry
- 50 fattened poultry
- 10 000 eggs
- 2 tons of endives
- 1 ton of mushroom
Thresholds are specific in overseas territories:
- Concerning labour force and OGA:
- we collect and publish national data on manager's family members for all legal forms (not only for sole holder holdings),
- we collect and publish national data on OGA for all legal forms, and we have a wider definition of OGA.
Concerning livestock:
- we use national LSU coefficients
- Concerning UAA: UAA disseminated at national level does not include common land areas.
15.1.3.3. Reasons for differences
National thresholds are different from Eurostat ones to keep comparability between time series, and to cover some specific productions in some regions.
15.1.4. Definitions and classifications of variables
See sub-categories below.
15.1.4.1. Deviations from Regulation (EU) 2018/1091 and EU handbook
Both the data disseminated at national level and the data transmitted to Eurostat present the following deviations from Regulation (EU) 2018/1091 and EU handbook:
- The irrigation data is not collected for kitchen gardens and, in the overseas territories, for rice (not systematically irrigated).
- The data on soil management practices is not collected for arable land under glass or under high protective cover.
15.1.4.1.1. The number of working hours and days in a year corresponding to a full-time job
The information is available on Eurostat’s website, at the link: CircaBC.
The number of working hours and days in a year for a full-time job correspond to one annual working unit (AWU) in the country. One annual work unit corresponds to the work performed by one person who is occupied on an agricultural holding on a full-time basis. Annual working units are used to calculate the farm work on the agricultural holdings.
15.1.4.1.2. Point chosen in the Annual work unit (AWU) percentage band to calculate the AWU of holders, managers, family and non-family regular workers
See item 15.1.4.1.1.
15.1.4.1.3. AWU for workers of certain age groups
See item 15.1.4.1.1.
15.1.4.1.4. Livestock coefficients
FR also uses national LSU coefficients for the national dissemination, available in the annex.
Annexes:
15.1.4.1.4. Livestock coefficients
15.1.4.1.5. Livestock included in “Other livestock n.e.c.”
Other livestock n.e.c. included in the file:
- Bees (A6710R)
- Deer (A6210)
- Fur animals (A6010)
- Male rabbits
- Rabbits for fattening
- Equidae for human consumption
- Equidae used as means of production
- Breeding mares in riding stables for race horses which also have breeding activities
- Camels and camelids
Other livestock n.e.c. NOT included in the file:
- Live swine wild species (A3200)
- Hybrid pigs
- European mouflon (Ovis orientalis musimon Pallas) if raised for the production of meat
15.1.4.2. Reasons for deviations
National disseminated results differ from Eurostat ones: see item 15.1.3.2. The reasons relate to keeping comparability in the national time series and to respond to national user needs.
The data transmitted to Eurostat present the deviations from Regulation (EU) 2018/1091 and EU handbook (mentioned in items 15.1.4.1 and 15.1.4.1.5), because:
- The irrigation data are too technical to be accessible for farmers who only irrigate their kitchen garden.
- In the overseas territories, contrary to what is stated in the handbook, not all rice crops are irrigated.
- The data on soil management practices questions are not relevant for arable land under glass or under high protective cover, especially for hydroponic crops: no tillage, no need for practices to avoid nitrogen losses.
- For “Other livestock n.e.c.” (A0030) (as presented in item 15.1.4.1.5), hybrid pigs, and mouflon (for meat production) are not present in France. It is not allowed to raise wild swine species. They can be hunted from nature, but cannot be bred in captivity.
15.1.5. Reference periods/days
See sub-categories below.
15.1.5.1. Deviations from Regulation (EU) 2018/1091
No deviations.
15.1.5.2. Reasons for deviations
Not applicable.
15.1.6. Common land
The concept of common land exists15.1.6.1. Collection of common land data
Yes15.1.6.2. Reasons if common land exists and data are not collected
Not applicable.
15.1.6.3. Methods to record data on common land
Common land is included in the land of agricultural holdings based on a statistical model.15.1.6.4. Source of collected data on common land
Administrative sources15.1.6.5. Description of methods to record data on common land
For IFS 2023, we used administrative data coming from the CAP to allocate common land areas. Total common land area, users, type of cattle and duration of occupation are collected in IACS.
The individual occupation is calculated by dividing the total area of the common land by the number of cattle present and its duration on the land during the year.
15.1.6.6. Possible problems in relation to the collection of data on common land and proposals for future data collections
Not applicable.
15.1.7. National standards and rules for certification of organic products
See sub-categories below.
15.1.7.1. Deviations from Council Regulation (EC) No 834/2007
No
15.1.7.2. Reasons for deviations
Not applicable.
15.1.8. Differences in methods across regions within the country
For national needs, specific questions adapted to the local context were included for overseas territories.
15.2. Comparability - over time
See sub-categories below.
15.2.1. Length of comparable time series
2 years
15.2.2. Definition of agricultural holding
See sub-categories below.
15.2.2.1. Changes since the last data transmission to Eurostat
There have been no changes15.2.2.2. Description of changes
There are no changes as both 2020 and 2023 are data collection years covered by the same Regulation (EU) 2018/1091.
15.2.3. Thresholds of agricultural holdings
See sub-categories below.
15.2.3.1. Changes in the thresholds of holdings for which core data are sent to Eurostat since the last data transmission
There have been no changes15.2.3.2. Description of changes
Not applicable.
15.2.4. Geographical coverage
See sub-categories below.
15.2.4.1. Change in the geographical coverage since the last data transmission to Eurostat
There have been no changes15.2.4.2. Description of changes
Not applicable.
15.2.5. Definitions and classifications of variables
See sub-categories below.
15.2.5.1. Changes since the last data transmission to Eurostat
There have been no changes15.2.5.2. Description of changes
There are no changes as both 2020 and 2023 are data collection years covered by the same Regulation (EU) 2018/1091.
15.2.6. Reference periods/days
See sub-categories below.
15.2.6.1. Changes since the last data transmission to Eurostat
There have been no changes15.2.6.2. Description of changes
Not applicable.
15.2.7. Common land
See sub-categories below.
15.2.7.1. Changes in the methods to record common land since the last data transmission to Eurostat
There have been no changes15.2.7.2. Description of changes
Not applicable.
15.2.8. Explanations for major trends of main variables compared to the last data transmission to Eurostat
Evolution of the number of holdings by legal status:
There has been a remarkable decrease in the holdings by legal status FARM_HLD and FARM_NFAM; after analysing the reasons it can be pointed out that the main one is the changes in the CAP policies.
From 2023, farmers older than 67 and receiving subsidies as retired cannot be eligible for CAP subsidies. Most part of them have quit their agricultural activity and transmitted the land and/or livestock to other farms.
The decrease is clear, and the IFS 2023 shows consistency with the IACS situation.
As consequence of the fact that many of the sole holders have quit their activity, there has been an impact in the holdings breakdown by SO_EURO in 2023, compared to 2020. They were generally managing very small farms with low or very low economical level. The IACS study confirms a decrease of share of holdings with very low agricultural standard output.
As per 2020 also in 2023, more than half of French holdings had no livestock (and the number of holdings without livestock seems increasing).
Evolution of main aggregates for quantitative variables:
| Variable code 2023 | Extrapolated value 2020 | Extrapolated value 2023 | Relative diff. 2023 vs 2020 | Comments |
|---|---|---|---|---|
| C1120T | 253 583.29 | 209 241.00 | -17.49 | IFS value confirmed. Not significant difference between IACS and IFS. |
| C1200T | 118 293.09 | 87 447.20 | -26.08 | IFS value confirmed. Not significant difference between IACS and IFS. |
| C1400T | 111 186.47 | 72 315.02 | -34.96 | IFS value confirmed. Not significant difference between IACS and IFS. |
| C1500T | 1 736 898.46 | 1 354 539.93 | -22.01 | IFS value confirmed. Not significant difference between IACS and IFS. |
| C2000T | 15 012.41 | 10 543.73 | -29.77 | IFS value confirmed. Not significant difference between IACS and IFS. |
| P1000T | 283 752.45 | 229 955.48 | -18.96 | Difference between IFS and IACS less than 2%. Values confirmed. |
| I3000T | 1 638.74 | 560.49 | -65.80 | Low value; difference cannot be considered as significant. |
| I1110T | 1 112 526.64 | 1 361 513.01 | +22.38 | IFS value confirmed. Not significant difference between IACS and IFS. |
| I1190T | 9 616.05 | 16 443.47 | +71.00 | IFS value confirmed. Not significant difference between IACS and IFS. |
| I2900T | 308.10 | 21.88 | -92.90 | Low value; difference cannot be considered as significant. |
| V0000_S0000TK | 53 226.56 | 67 060.73 | +25.99 | IACS value is the sum of all vegetables (in rotation with horticulture crop or not). We can consider that the ratio between the two is equal; the V0000_S0000TK - IFS 2023 value seems correct. |
| N0000T | 3 820.69 | 3 205.43 | -16.10 | Low value; difference cannot be considered as significant. |
| G9100T + G9900T | 68 627.68 | 103 249.05 | +50.45 | IFS value confirmed. Not significant difference between IACS and IFS. |
| K0000T | 1 598.40 | 1 241.37 | -22.34 | Low value; difference cannot be considered as significant. |
| J3000TE | 98 096.59 | 60 220.56 | -38.61 | IACS changes: less areas can be declared under this variable. Part of them are now integrated into permanent grassland. |
| W1200T | 6 242.71 | 7 514.95 | +20.38 | Low value; difference cannot be considered as significant. IACS does not cover properly such crop. |
| PECRS | 1 042.08 | 1 290.89 | +23.88 | Low value; difference cannot be considered as significant. |
| SRCAA | 165 538.77 | 103 927.31 | -37.22 | Not well covered by IACS. |
| U1000 | 103.01 | 68.15 | -33.84 | Low value; difference cannot be considered as significant. |
| I6000T | 7 674.50 | 14 626.79 | +90.59 | IFS value confirmed. Not significant difference between IACS and IFS. |
| A3110 | 4 943 607.00 | 4 191 383.22 | -15.22 | IFS value confirmed. "Livestock survey Nov. 2023" gives 4 331 654 units for A3110. Difference = -3.3%. |
| MOGA_NFAM_RH | 8 698.13 | 6 965.80 | -19.92 | Explained in the paragraph above. |
| SOGA_NFAM_RH | 10 643.21 | 4 409.34 | -58.57 | Explained in the paragraph above. |
15.2.9. Maintain of statistical identifiers over time
Yes15.3. Coherence - cross domain
See sub-categories below.
15.3.1. Coherence - sub annual and annual statistics
Not applicable to Integrated Farm Statistics, because there are no sub annual data collections in agriculture.
15.3.2. Coherence - National Accounts
Not applicable, because Integrated Farm Statistics have no relevance for national accounts.
15.3.3. Coherence at micro level with data collections in other domains in agriculture
See sub-categories below.
15.3.3.1. Analysis of coherence at micro level
No15.3.3.2. Results of analysis at micro level
We do not make any comparisons with other agricultural microdata sources.
The IFS questionnaires are prefilled with external information: CAP for crops, BDNI for bovine livestock, CVI for wine producers.
So, the information from these other sources is linked from the beginning, and we do not need to conduct further comparisons.
For other variables, no microdata are available.
15.3.4. Coherence at macro level with data collections in other domains in agriculture
See sub-categories below.
15.3.4.1. Analysis of coherence at macro level
Yes15.3.4.2. Results of analysis at macro level
Coherence cross-domain: IFS vs CROP PRODUCTION (main area in 1000 ha) in relative terms
Most of the values from the IFS 2023 are not significantly different from IFS 2020; differences with Eurobase values may be due to different definitions of variables (e.g. permanent grassland, fallow land). It seems that permanent grassland used for horses' breeders or centres is included in Eurobase but excluded in IFS (they do not reach the threshold due to a zero LSU coefficient).
In France, globally, there are more than 1 million horses, using more than for the FRD11 region, permanent grassland represents more than 90 000 hectares. Moreover, as said in the handbook, buffer zones are counted in “other arable land” in the IFS but as permanent grassland in Eurobase data.
Coherence cross-domain: IFS vs ANIMAL PRODUCTION (1000 heads) in relative terms
There are discrepancies between the two data sets but, the values of 48 of the 73 variables are with the confidence interval of the IFS 2023 value. For 25 of them it is not the case. France encountered some difficulties in sampling the livestock survey in 2023 in animal production statistics, namely for sheep and goats. However, following the validation of IFS 2023, the 2023 livestock survey data were corrected in order to ensure data consistency between IFS and the livestock survey. Despite this correction, the IFS values remain the most accurate.
Coherence cross-domain: IFS vs ORGANIC ANIMAL PRODUCTION (heads) in relative terms
For organic sheep and goats, France does not have any specific survey except 2020 census and IFS 2023. The values given for animal production statistics are coming from estimations made by the French Agricultural Organic Agency which provides values only for breeding females, but not for a specific date (average for the whole year). Therefore, the values coming from IFS are more reliable.
15.4. Coherence - internal
The data are internally consistent. This is ensured by the application of a wide range of validation rules.
See sub-categories below.
16.1. Coordination of data collections in agricultural statistics
The livestock survey took place in November 2023. To avoid double interrogation of farms included in the two surveys (IFS and Livestock), we decided to start the IFS data collection for these farms after the end of the livestock survey, in December 2023. The collected livestock information has been used to prefill the IFS questionnaires.
16.2. Efficiency gains since the last data transmission to Eurostat
Further trainingOther
16.2.1. Additional information efficiency gains
Compared to IFS 2020, we had:
- reinforced the e-training for interviewers,
- improved outlier detection using maps and consistency between variables.
16.3. Average duration of farm interview (in minutes)
See sub-categories below.
16.3.1. Core
Around 20 minutes.
16.3.2. Module ‘Labour force and other gainful activities‘
Around 15 minutes.
16.3.3. Module ‘Rural development’
Not relevant, use of administrative data only.
16.3.4. Module ‘Animal housing and manure management’
Restricted from publication
16.3.5. Module ‘Irrigation’
Around 5 minutes.
16.3.6. Module ‘Soil management practices’
Around 5 minutes.
16.3.7. Module ‘Machinery and equipment’
Around 10 minutes.
16.3.8. Module ‘Orchard’
Around 5 minutes (high variability depending on the number of concerned species)
16.3.9. Module ‘Vineyard’
Restricted from publication
17.1. Data revision - policy
No specific data revision policy.
Nevertheless, an unplanned revision may be required to correct an error found in the data after its release. In such cases, users are informed through a note sent by email.
17.2. Data revision - practice
No revisions to report.
17.2.1. Data revision - average size
Not requested.
18.1. Source data
See sub-categories below.
18.1.1. Sampling design & Procedure frame
See sub-categories below.
18.1.1.1. Type of frame
List frame18.1.1.2. Name of frame
BALSA - Database for agricultural statistics, our agricultural register (Base de Sondage pour la Statistique Agricole)
18.1.1.3. Update frequency
Monthly18.1.2. Core data collection on the main frame
See sub-categories below.
18.1.2.1. Coverage of agricultural holdings
Sample18.1.2.2. Sampling design
The sampling frame is built from the 2020 agricultural census, updated with administrative sources, notably to detect new units.
The sampling design is a stratified one-stage systematic random sampling. For the French overseas department of Mayotte, due to the inadequacy of the sampling frame, the sampling design is a one-stage cluster sampling (a sample of villages, with exhaustive survey of units in sampled villages, completed with a full coverage of the biggest units in the sampling frame).
Strata are built on three criteria: technical orientation of units, size measured with an estimation of the standard gross production (also used as the OSU_S1_CORE variable for systematic sampling within strata), and location. These variables are available for units in the 2020 census and can be imputed for new units known from administrative sources. The specific strata limits are computed using an optimisation algorithm (using INSEE Palourde R function and Koubi-Mathern method). The overall sample size is computed with the INSEE Palourde R function designed to meet several overlapping accuracy constraints. Sample allocation in strata (including full coverage strata) is computed using a Neymann formula based on the OSU_S1_CORE, with a minimum allocation constraint determined by Palourde. Full coverage strata cover about half the sample.
All modules use the same sample. There is no specific sampling design.
More precisely:
Systematic sampling sorted on standard output (OSU_S1_CORE). The stratification is based on:
- NUTS 2 and NUTS 3
- Standard output:
- Very small farms: Standard output <= 25 000 €
- Small farms: 25 000 € < Standard output <= 50 000 €
- Medium farms: 50 000 € < Standard output <= 100 000 €
- Big farms: Standard output > 100 000 €
- Farm type:
- Cereals, oilseed
- Other field crops
- Vegetables and mushrooms
- Horticulture
- Vineyards
- Fruits and other permanent crops
- Milk cattle
- Meat cattle
- Cattle combination (milk and meat)
- Sheep and goats
- Other herbivores
- Pigs
- Poultry
- Various granivores combined
- Various crops and livestock combined
- Non-classified farms
In order to obtain enough farms for the orchard module, the most detailed farm type (64 classes) has been used:
- Specialised fruit farms (except citrus, tropical fruit and nuts)
- Specialised citrus farms
- Specialised nut farms
- Specialised tropical fruit farms
- Specialised fruit, citrus, tropical fruit and nut farms: mixed production
- Specialised olive farms
- Farms with various combinations of permanent crops
For the construction of strata, variables (NUTS 2, farm type, standard output and NUTS 3) are crossed (in this order). Each stratum had to contain a minimum of 50 units. When it is not the case, strata are defined with a neighbour, by grouping in the following order: NUTS 3, then if necessary standard output, then farm type. Furthermore, the units for which only farm type is known on the one hand and those for which farm type and standard output are unknown on the other hand, are defined in “special” NUTS 2 strata.
In total, there are 1 986 strata (without full coverage strata).
The tool Palourde (Production d'allocations localement optimisées utilisables pour des résultats sur des domaines d'études or "Production of locally optimised allocations usable for results on fields of study" - Methodology Department of INSEE) was used. This tool optimises the allocation of the overall stratified simple random sample size with multiple objectives of published data. First, the algorithm calculates the minimum number of units to select from each stratum in order to respect the RSE on the variables of interest with some published data domains (solution by J. Bethel of the problem of optimal multivariate distribution of the sample). A minimum number has been determined and an algorithm (proposed by M. Koubi and S. Mathern) finalise the optimisation on a key variable of interest in the survey, while ensuring that the minimum number of units is correctly selected in each stratum. The chosen key variable of interest is standard outputs because it is overall correlated with the different variables for which the European regulation requires RSE. A minimum of six units is necessary in each stratum and the minimum sampling rate per stratum is 1/50. The resulting sample for France, excluding geographic exhaustiveness, must ultimately include 55 334 units to meet local constraints. The allocations per stratum are calculated to optimise the RSE on “standard output” variable with the constraint of the total allocation of 70 000 surveys.
18.1.2.2.1. Name of sampling design
Stratified one-stage random samplingStratified one-stage cluster sampling
18.1.2.2.2. Stratification criteria
Unit sizeUnit location
Unit specialization
18.1.2.2.3. Use of systematic sampling
Yes18.1.2.2.4. Full coverage strata
There are full coverage strata for:
- units with standard output above 500 000 € (or 250 000 € for vegetables and mushrooms, horticulture, and poultry type farms)
- units with 10 permanent salaried employees
- units with the same business register identifier (SIRENE)
18.1.2.2.5. Method of determination of the overall sample size
The sample size has been defined in order to meet several requirements:
- the cost of the survey: maximum 70 000 units in total,
- RSE by NUTS 2 under 5% for land and livestock variables in Annex V of Regulation (EU) 2018/1091.
Existing variables in the sampling frame (most of them known from the 2020 census) were used as proximate values for the survey variables to predict accuracy.
IFS accuracy requirements were analysed and used to establish 188 specific accuracy constraints. Usual sample size algorithms are not designed to answer for multiple, overlapping accuracy constraints. A simple method would be to compute minimal sample size for each constraint separately and select the maximum sample size for each domain, region or stratum, but this can lead to oversized, non-optimal sampling. To reduce the statistical burden, we used INSEE R function "Palourde" that optimises sample size for multiple accuracy constraints at once.
18.1.2.2.6. Method of allocation of the overall sample size
Neymann allocation18.1.3. Core data collection on the frame extension
See sub-categories below.
18.1.3.1. Coverage of agricultural holdings
Not applicable18.1.3.2. Sampling design
Not applicable.
18.1.3.2.1. Name of sampling design
Not applicable18.1.3.2.2. Stratification criteria
Not applicable18.1.3.2.3. Use of systematic sampling
Not applicable18.1.3.2.4. Full coverage strata
Not applicable.
18.1.3.2.5. Method of determination of the overall sample size
Not applicable.
18.1.3.2.6. Method of allocation of the overall sample size
Not applicable18.1.4. Module “Labour force and other gainful activities”
See sub-categories below.
18.1.4.1. Coverage of agricultural holdings
Sample18.1.4.2. Sampling design
See item 18.1.2.2.
18.1.4.2.1. Name of sampling design
Stratified one-stage random samplingStratified one-stage cluster sampling
18.1.4.2.2. Stratification criteria
Unit sizeUnit location
Unit specialization
18.1.4.2.3. Use of systematic sampling
Yes18.1.4.2.4. Full coverage strata
See item 18.1.2.2.4.
18.1.4.2.5. Method of determination of the overall sample size
See item 18.1.2.2.5.
18.1.4.2.6. Method of allocation of the overall sample size
Neymann allocation18.1.4.2.7. If sampled from the core sample, the sampling and calibration strategy
Not applicable18.1.5. Module “Rural development”
See sub-categories below.
18.1.5.1. Coverage of agricultural holdings
Sample18.1.5.2. Sampling design
See item 18.1.2.2.
18.1.5.2.1. Name of sampling design
Stratified one-stage random samplingStratified one-stage cluster sampling
18.1.5.2.2. Stratification criteria
Unit sizeUnit location
Unit specialization
18.1.5.2.3. Use of systematic sampling
Yes18.1.5.2.4. Full coverage strata
See item 18.1.2.2.4.
18.1.5.2.5. Method of determination of the overall sample size
See item 18.1.2.2.5.
18.1.5.2.6. Method of allocation of the overall sample size
Neymann allocation18.1.5.2.7. If sampled from the core sample, the sampling strategy and calibration strategy
Not applicable18.1.6. Module “Animal housing and manure management module”
Restricted from publication
18.1.6.1. Coverage of agricultural holdings
Restricted from publication
18.1.6.2. Sampling design
Restricted from publication
18.1.6.2.1. Name of sampling design
Restricted from publication
18.1.6.2.2. Stratification criteria
Restricted from publication
18.1.6.2.3. Use of systematic sampling
Restricted from publication
18.1.6.2.4. Full coverage strata
Restricted from publication
18.1.6.2.5. Method of determination of the overall sample size
Restricted from publication
18.1.6.2.6. Method of allocation of the overall sample size
Restricted from publication
18.1.6.2.7. If sampled from the core sample, the sampling strategy and calibration strategy
Restricted from publication
18.1.7. Module ‘Irrigation’
See sub-categories below.
18.1.7.1. Coverage of agricultural holdings
Sample18.1.7.2. Sampling design
See item 18.1.2.2.
18.1.7.2.1. Name of sampling design
Stratified one-stage random samplingStratified one-stage cluster sampling
18.1.7.2.2. Stratification criteria
Unit sizeUnit location
Unit specialization
18.1.7.2.3. Use of systematic sampling
Yes18.1.7.2.4. Full coverage strata
See item 18.1.2.2.4.
18.1.7.2.5. Method of determination of the overall sample size
See item 18.1.2.2.5.
18.1.7.2.6. Method of allocation of the overall sample size
Neymann allocation18.1.7.2.7. If sampled from the core sample, the sampling strategy and calibration strategy
Not applicable18.1.8. Module ‘Soil management practices’
See sub-categories below.
18.1.8.1. Coverage of agricultural holdings
Sample18.1.8.2. Sampling design
See item 18.1.2.2.
18.1.8.2.1. Name of sampling design
Stratified one-stage random samplingStratified one-stage cluster sampling
18.1.8.2.2. Stratification criteria
Unit sizeUnit location
Unit specialization
18.1.8.2.3. Use of systematic sampling
Yes18.1.8.2.4. Full coverage strata
See item 18.1.2.2.4.
18.1.8.2.5. Method of determination of the overall sample size
See item 18.1.2.2.5.
18.1.8.2.6. Method of allocation of the overall sample size
Neymann allocation18.1.8.2.7. If sampled from the core sample, the sampling strategy and calibration strategy
Not applicable18.1.9. Module ‘Machinery and equipment’
See sub-categories below.
18.1.9.1. Coverage of agricultural holdings
Sample18.1.9.2. Sampling design
See item 18.1.2.2.
18.1.9.2.1. Name of sampling design
Stratified one-stage random samplingStratified one-stage cluster sampling
18.1.9.2.2. Stratification criteria
Unit sizeUnit location
Unit specialization
18.1.9.2.3. Use of systematic sampling
Yes18.1.9.2.4. Full coverage strata
See item 18.1.2.2.4.
18.1.9.2.5. Method of determination of the overall sample size
See item 18.1.2.2.5.
18.1.9.2.6. Method of allocation of the overall sample size
Neymann allocation18.1.9.2.7. If sampled from the core sample, the sampling strategy and calibration strategy
Not applicable18.1.10. Module ‘Orchard’
See sub-categories below.
18.1.10.1. Coverage of agricultural holdings
Sample18.1.10.2. Sampling design
See item 18.1.2.2.
18.1.10.2.1. Name of sampling design
Stratified one-stage random samplingStratified one-stage cluster sampling
18.1.10.2.2. Stratification criteria
Unit sizeUnit location
Unit specialization
18.1.10.2.3. Use of systematic sampling
Yes18.1.10.2.4. Full coverage strata
See item 18.1.2.2.4.
18.1.10.2.5. Method of determination of the overall sample size
See item 18.1.2.2.5.
18.1.10.2.6. Method of allocation of the overall sample size
Neymann allocation18.1.10.2.7. If sampled from the core sample, the sampling strategy and calibration strategy
Not applicable18.1.11. Module ‘Vineyard’
Restricted from publication
18.1.11.1. Coverage of agricultural holdings
Restricted from publication
18.1.11.2. Sampling design
Restricted from publication
18.1.11.2.1. Name of sampling design
Restricted from publication
18.1.11.2.2. Stratification criteria
Restricted from publication
18.1.11.2.3. Use of systematic sampling
Restricted from publication
18.1.11.2.4. Full coverage strata
Restricted from publication
18.1.11.2.5. Method of determination of the overall sample size
Restricted from publication
18.1.11.2.6. Method of allocation of the overall sample size
Restricted from publication
18.1.11.2.7. If sampled from the core sample, the sampling strategy and calibration strategy
Restricted from publication
18.1.12. Software tool used for sample selection
R software was used (Palourde function - INSEE)
18.1.13. Administrative sources
See sub-categories below.
18.1.13.1. Administrative sources used and the purposes of using them
The information is available on Eurostat’s website, at the link: Additional data - Eurostat (europa.eu).
18.1.13.2. Description and quality of the administrative sources
See the Excel file in the annex.
Annexes:
18.1.13.2. Description and quality of administrative sources
18.1.13.3. Difficulties using additional administrative sources not currently used
Problems related to data quality of the sourceThe final validated data in the source would not be in time to meet statistical deadlines or would relate to a period which does not coincide with the reference period
Other
18.1.14. Innovative approaches
The information on the innovative approaches and the quality methods applied is available on Eurostat’s website, at the link: Additional data - Eurostat (europa.eu).
18.2. Frequency of data collection
The agricultural census is conducted every 10 years. The decennial agricultural census is complemented by sample or census-based data collections organised every 3-4 years in-between.
18.3. Data collection
See sub-categories below.
18.3.1. Methods of data collection
Face-to-face, electronic versionTelephone, electronic version
18.3.2. Data entry method, if paper questionnaires
Not applicable18.3.3. Questionnaire
Please find the questionnaire in annex.
Common land is assessed based on its use by farms. No specific questionnaire.
Annexes:
18.3.3. Questionnaire in French
18.3.3. Questionnaire in English
18.4. Data validation
See sub-categories below.
18.4.1. Type of validation checks
Data format checksCompleteness checks
Range checks
Relational checks
Comparisons with previous rounds of the data collection
18.4.2. Staff involved in data validation
InterviewersSupervisors
Staff from local departments
Staff from central department
18.4.3. Tools used for data validation
- Validation checks within data collection tools,
- R-programs validation, and
- XLSX files with aggregated results, comparing results between 2020 and 2023, for aggregated variables, at the national and NUTS 2 level.
18.5. Data compilation
Total non-response is corrected by reweighting within each stratum. Then post-calibration uses margins from the sampling universe (total number of statistical units in the sampling universe), from the bovine identification database, and from the IACS 2023 data within each NUTS 3 region. To ensure that calibration variables are identical to margins variables, we match respondent units with IACS declarations and retrieve the corresponding IACS variables. IACS declaration variables are then used as calibration variables for each respondent unit, as well as margin variables for the whole population.
18.5.1. Imputation - rate
There was no unit imputation, only item imputation (unit non-response is corrected with reweighting)
The item non-response is very low, with imputation for only a few variables and only a few values (less than 100 at most) for each variable. We did not compute a specific rate.
18.5.2. Methods used to derive the extrapolation factor
Design weightNon-response adjustment
Calibration
18.6. Adjustment
Covered under Data compilation.
18.6.1. Seasonal adjustment
Not applicable to Integrated Farm Statistics, because it collects structural data on agriculture.
See sub-categories below.
19.1. List of abbreviations
ASP – Agence de Services et de Paiement / National Payment Agency
AWU – Annual Working Unit
BDNI – Base de données nationale d’identification / Bovine identification national database
BSSAF – Bureau des statistiques sur les structures agricoles, l'aquaculture et la forêt / Unit on agricultural structures, aquaculture and forestry statistics
CAP – Common Agricultural Policy
CASD – Centre d'Accès Sécurisé aux Données / Centre for Secure Access to Data
CORE – General, crops and livestock variables of Annex III of Regulation (EU) 2018/1091
CVI – Casier viticole informatisé / Customs administration vineyard
DGAL – Direction générale de l'alimentation / General Direction for Food
EU – European Union
FADN – Farm Accountancy Data Network
FNSEA – Fédération des Syndicats d'Exploitants Agricoles / Federation of Farmers' Unions
GSBPM – Global Statistical Business Process Model
IACS – Integrated Administration and Control System
IDELE – Institut de l'Élevage / French Livestock Institute
IFIP – Institut du porc / French Pork and Pig Institute
IFS – Integrated Farm Statistics
INAO – Institut National des Appellations d'Origine / National Institute of Designations of Origin
INRAE-ODR – National Research Institute for Agriculture, Food and Environment for processing rural development module data
INSEE – Institut national de la statistique et des études économiques / National Institute of Statistics and Economic Studies
LSU – Livestock unit
MASA – Ministère de l'Agriculture et de la Souveraineté Alimentaire / Ministry of Agriculture and Food Sovereignty
NUTS – Nomenclature of territorial units for statistics
OGA – Other gainful activities
RSE – Relative standard error
SAIO – Statistics on agricultural inputs and outputs
SDSAFA – Sous-direction des statistiques agricoles, forestières et agroalimentaires / Department of agricultural, forestry and agrifood statistics
SG – Secrétariat Général / General Secretariat of the Ministry of Agriculture and Food Sovereignty
SGM – Standard Gross Margin
SIRENE – Système d'Identification du Répertoire des Entreprises / Business register
SO – Standard output
SSP – Service de la Statistique et de la Prospective / Department of Statistics and Foresight Analysis
UAA – Utilised agricultural area
19.2. Additional comments
IFS 2023 encountered three important difficulties:
- Changes in IACS policy: The new scheme for receiving subsidies from CAP now excludes the smallest farms and holders older than 67 years; this modification could not be taken into account before the survey, and resulted in a large number of ineligible units (> 4 500 units).
- Weather conditions: We had to face three major floods in the North of France during the survey, and a violent storm in the West; farmers were not able to answer the survey.
- Social disappointment: During the winter, we had to face farmers' anger. They did not want to respond to administrative requests anymore.
The two last points generated more than 3 000 non-responses from farmers.
Although these difficulties probably had a negative impact on the quality of the results, their overall effect remained low.
The data describe the structure of agricultural holdings providing the general characteristics of farms and farmers and information on their land, livestock and labour force. They also describe production methods, rural development measures and agro-environmental aspects that look at the impact of agriculture on the environment.
The data are used by public, researchers, farmers and policy-makers to better understand the state of the farming sector and the impact of agriculture on the environment. The data follow up the changes in the agricultural sector and provide a basis for decision-making in the Common Agricultural Policy (CAP) and other European Union policies.
The statistical unit is the agricultural holding (farm). The aggregated results are disseminated through statistical tables. The data are presented at different geographical levels and over periods.
The data collections are organised in line with Regulation (EU) 2018/1091 and have a new structure, consisting of a core data set and several modules. The regulation covers the data collections in 2019/2020 (the agricultural census), 2023 and 2026. The data are as comparable and coherent as possible with the other European countries.
8 July 2025
The list of core variables is set in Annex III of Regulation (EU) 2018/1091.
The descriptions of the core variables as well as the lists and descriptions of the variables for the modules collected in 2023 are set in Commission Implementing Regulation (EU) 2021/2286.
The following groups of variables are collected in 2023:
- for core: location of the holding, legal personality of the holding, manager, type of tenure of the utilised agricultural area, variables of land, organic farming, irrigation on cultivated outdoor area, variables of livestock, organic production methods applied to animal production;
- for the module "Labour force and other gainful activities": farm management, family labour force, non-family labour force, other gainful activities directly and not directly related to the agricultural holding;
- for the module "Rural development": support received by agricultural holdings through various rural development measures;
- for the module “Irrigation”: availability of irrigation, irrigation methods, sources of irrigation water, technical parameters of the irrigation equipment, crops irrigated during a 12 months period;
- for the module “Soil management practices”: tillage methods, soil cover on arable land, crop rotation on arable land, ecological focus area;
- for the module “Machinery and equipment”: internet facilities, basic machinery, use of precision farming, machinery for livestock management, storage for agricultural products, equipment used for production of renewable energy on agricultural holdings;
- for the module “Orchards”: apples area, pears area, peaches area, nectarines area, apricots area, small citrus fruit area, lemons area, olives area, grapes for table use area, each one by age of plantation and density of trees.
See sub-category below.
See sub-categories below.
See sub-categories below.
See sub-categories below.
See categories below.
Two kinds of units are generally used:
- the units of measurement for the variables (area in hectares, livestock in heads or LSU (livestock units), labour force in persons or AWU (annual working units), standard output in Euro, places for animal housing etc.) and
- the number of agricultural holdings having these characteristics.
Total non-response is corrected by reweighting within each stratum. Then post-calibration uses margins from the sampling universe (total number of statistical units in the sampling universe), from the bovine identification database, and from the IACS 2023 data within each NUTS 3 region. To ensure that calibration variables are identical to margins variables, we match respondent units with IACS declarations and retrieve the corresponding IACS variables. IACS declaration variables are then used as calibration variables for each respondent unit, as well as margin variables for the whole population.
See sub-categories below.
Data dissemination for farm statistics done after each survey, every 3 or 4 years.
See sub-categories below.
See sub-categories below.
See sub-categories below.
