Farm structure survey - administrative sources

Last updated: July 2024

Highlights

The use of administrative data in the production of agricultural statistics has increased considerably over time.

The number of variables collected directly from farmers in the EU decreased by 12% between 2020 and 2023.

Administrative data sources are used principally to replace variables, to impute unit/item non-response and for pre-filling.

Administrative sources are generally defined as data sources containing information that is not primarily collected for statistical purposes. Eurostat has encouraged countries to make appropriate re-use of data sources and to use fully administrative data as a way of reducing the burden on farmers. The use of administrative data in the production of agricultural statistics has increased considerably in recent years and has been incorporated into the legislative framework.

This article provides an overview of the variety of administrative sources used by countries in the Integrated Farm Statistics (IFS) 2023 in comparison to IFS 2020, and the purpose of their use. To show the change of use of administrative sources, the article compares IFS 2023 data, as provided by countries, with that from the IFS 2020 and from the Farm Structure Surveys (FSS) of 2013 and 2016.

Full article

Administrative sources in legislation

The use of administrative sources for agricultural statistics was included in legislation for the first time in Regulation (EC) No 1166/2008 of the European Parliament and of the Council , which set the framework for the implementation of three farm structure surveys (2009/2010, 2013 and 2016) and a survey on agricultural production methods (SAPM) (2009/2010).

Regulation (EU) 2018/1091 further strengthened the use of administrative data sources, other sources, methods or innovative approaches in addition to statistical surveys. According to Article 4 of this regulation, if the information from the administrative source is of at least equal quality to information obtained from statistical surveys, Member States may use information from:

The Integrated Administration and Control System (IACS) provided for in Council Regulation (EC) No 1782/2003 of the European Parliament and of the Council,
The system for the identification and registration of bovine animals provided for in Regulation (EC) No 1760/2000,
The system for the identification and registration of ovine and caprine animals established by Council Regulation (EC) No 21/2004,
The vineyard register implemented in accordance with Article 145 of Regulation (EU) No 1308/2013 of the European Parliament and of the Council,
The organic farming registers set up pursuant to Council Regulation (EC) No 834/2007,
Other administrative sources associated with specific rural development measures, cadastre or other administrative sources (which have to be specified).
Innovative approaches and methods like:

- Modelling

- Remote sensing

- Imputation

- Estimation

- Other innovative approach (to be specified)

Member States deciding to use ‘other sources, methods or innovative approaches’ need to inform the European Commission (Eurostat) during the year preceding the reference year and provide details concerning the quality of the data obtained from that source, method or innovative approach and the data collection methods used.

National authorities responsible for fulfilling the requirements of Regulation (EU) 2018/1091 have the right to access and use data, promptly and free of charge, including individual data on agricultural holdings and personal data on their holders contained in administrative files compiled on their national territory pursuant to Article 17a of Regulation (EC) No 223/2009.

Use of administrative sources by country under IFS 2023

Increasing use of administrative data

The IFS 2023 was conducted by EU countries, as well as three EFTA countries (Iceland, Norway and Switzerland). Eurostat encouraged the countries to make appropriate re-use of data sources and to use administrative data, as a way of reducing the burden on farmers.

The use of administrative sources represented about 40 % of all variables collected in the subsets of core (CORE), labour force (LAFO) and rural development (RDEV) by EU countries. The remaining 60 % came exclusively from the information collected from the farms by means of a survey under IFS 2023. This represents an increased use of administrative data (see Figure 1): the equivalent shares in 2020 and 2016 were 32 % and 29 % respectively.

Figure 1: Share of administrative data used for the variables in the CORE, LAFO and RDEV (%, 2020 and 2023)
Source: Eurostat calculations

For a growing number of countries, administrative data provided the majority of subset data

For eleven EU countries, for Norway and for Switzerland, administrative sources provided a majority of the IFS 2023 data for the variables in the subsets of core (CORE), labour force (LAFO) and rural development (RDEV). The highest share of variables covered in this way was in Slovenia (87 %), followed by Ireland (79 %), Spain (76 %) and Estonia (67 %).

This contrasted starkly with those countries that used little or no administrative data. There were four EU countries for which administrative sources were used for less than 1 % of the variables in the core, labour force and rural development modules; these were Portugal, Greece, Romania and Cyprus.

The sharpest increase in the uptake of administrative data was in The Netherlands (a 28 percentage point rise to 38 % in 2023), followed by Estonia and Italy (up 19 percentage points), Denmark and Ireland (up 18 percentage points) and Bulgaria (up 17 percentage points).

Nevertheless, there were some countries that bucked the upward trend. Less administrative data was used in 2023 in Sweden (a fall of 6 percentage points), Austria (down 4 percentage points) and Belgium (down 1 percentage point).

Overview of the purpose of administrative sources used in the IFS 2023 vs 2020

In 2023, administrative data sources were used for several purposes by countries, as follows:

To replace directly the values of the characteristic (denoted by Code 51)
To prefill the values (Code 52)
To impute for unit / item non-responses (Code 53)
To validate the data (Code 54)
For other reasons that should be specified in the comments (Code 59)
For model diagnosis (Code 71)
For external validation (Code 72)
For sensitivity analysis (Code 73)
To check a sub-sample (Code 74)
For other quality assessments that should be specified in comments (Code 79)

Code 99 is used when none of these options is relevant because the data was collected via a questionnaire.

EU countries used administrative sources for the core, labour force and rural development variables for a combined total of 3 331 purposes^[1]^[2] in 2023 (excluding code “99”), representing an increased use of 6 % compared with 2020.

A majority (54 %) of the uses of administrative data in 2023 were to replace directly the values of the characteristics - see Figure 2 – as was also the case in 2020 (56 %). Just over another one third of uses were for imputing unit / item non-responses (14 %), prefilling (14 %) and validation (7 %) combined. For these three cases, relative shares increased with respect to their use in 2020, the increase of 4 percentage points for validation being the highest. It should be noted that there was a sharp decline in the recording of use for other quality assessments, although this might be due to some re-allocation. Furthermore, there were no incidences in 2020 nor 2023 of administrative data being used for sensitivity analysis.

Figure 2: F2 Incidences of using administrative data for the variables in the CORE, LAFO and RDEV, by type of purpose (% of all incidences, EU, 2020 and 2023.
Source: Eurostat calculations

Overview of the administrative sources used in the IFS 2023 vs 2020

Countries were also required to indicate the type of administrative source used, as follows:

Not relevant - no administrative source or innovative approach (denoted by Code 00)
IACS - Integrated Administration and Control System (Code 01)
Bovine register (Code 02)
Ovine register (Code 03)
Caprine register (Code 04)
Vineyard register (Code 05)
Organic farming register (Code 06)
Genetically modified crops register (Code 07)
Rural development measures (Code 08)
Cadastre (Code 09)
Other administrative sources that should be specified in comments (Code 19)
Modelling (Code 21)
Remote sensing (Code 22)
Imputation (Code 23)
Estimation (Code 24)
Other innovative approaches that should be specified in comments (Code 29)

EU countries registered 3 311 entries regarding the type of administrative data source used for the core, labour force and rural development variables in the IFS 2023.

Of the various administrative sources used by EU countries, the most frequently referenced (43 % of all entries) was the Integrated Administration and Control System (IACS), followed by ‘other’ administrative sources (17 % of all entries) and the organic farming register (16 % of all cases) - see Figure 3. There was a higher incidence of use of the IACS and ‘other’ administrative sources in 2023 than there had been in 2020, whilst that of the organic farming register held steady.

Figure 3: F3 Frequency of use of types of administrative source for the variables in the CORE, LAFO, and RDEV variables (% of all incidences, EU, 2020 and 2023)
Source: Eurostat calculations

There were no incidences of remote sensing having been used as a source in either 2020 or 2023. Incidences of using imputation and modelling sources were lower in 2023 than in 2020.

Overview of non-significant and non-existing data

For core structural data, Article 5(3) of Regulation (EU) 2018/1091 states that “when a variable […] has a low or zero prevalence in a Member State, the variable may be excluded from the data collection subject to the Member State concerned providing information duly justifying its exclusion to the Commission (Eurostat) in the calendar year preceding the reference year”. A similar statement for the module variables is included in Article 7(9) of the same regulation. Analyzing developments is only possible for variables that are common to 2020 and 2023. For this reason, only the variables of the CORE, LAFO and RDEV are reviewed.

The prevalence of non-significant and non-existing data in the variables of the CORE, LAFO and RDEV was very slightly lower in 2023 than for 2020 for the EU and for a small majority of countries. Their use declined slightly from 13.4 % of variables for the EU in 2020 to 12.9 % in 2023 (see Figure 4).

Nevertheless, their use was well above the EU average and exceeded 20 % in Malta (32.7 %), Croatia (29.1 %), Sweden (22.3 %), Latvia (21.6 %) and Finland (21.2 %).

Figure 4: Share of non-significant and non-existing data in the variables of the CORE, LAFO and RDEV (%, 2020 and 2023)
Source: Eurostat calculations

Most of the reduction in the use of non-significant and non-existing data came from the variables in RDEV; there was a reduction from 21.2 % in 2020 to 17.2 % of variables in 2023. The change of share of non-significant and non-existing variables for CORE and the LAFO modules remained low (0.3 % decrease and 0.2 % increase respectively).

Figure 5: Share of non-significant and non-existing data in the variables of the CORE, LAFO and RDEV (%, 2020 and 2023)
Source: Eurostat calculations

Conclusions

In the past, administrative data was mostly used for building the sampling frame and calibrating the results (on the basis of the auxiliary variables), as well as for analysing and validating the survey results at macro level. More recently, there has been an increasing use of administrative data as a direct data source, thanks to developments in IT systems and closer cooperation between the different government authorities. This has resulted in lower expenditure on data collection and a significant reduction in the response burden of farmers.

Although the advantages of using administrative data are clear, there are also some disadvantages, misunderstandings and shortcomings. Administrative data are not necessarily error-free. Authorities also use collection processes for administrative sources, which may produce errors which are difficult to identify. Furthermore there may be specific quality issues such as over-coverage, under-coverage, misclassification, multiple listings, missing data, processing errors and coherence with other data sources.

An important tool that can help users monitor comparability and consistency across years and definitions is the analysis of the IFS quality reports, where countries provide methodological evidence of the potential breaks in time series, definition differences, and reference periods.

The analysis of the evolution of the NSNE variables in recent years shows that their use seems to have declined slightly for the common modules of the CORE, LAFO and more remarkably for RDEV.

Source data for tables and graphs

Farm structure survey - administrative sources: tables

Direct access to

Notes

↑ Sum obtained by adding all the purposes attributed with codes “51” to “79”.
↑ Countries could attribute different purposes to administrative data sources for the same variable.

[1] Sum obtained by adding all the purposes attributed with codes “51” to “79”.

[2] Countries could attribute different purposes to administrative data sources for the same variable.

[1]

[2]