Responsibility rests with Liaison Agencies of the Member States
The responsibility for FADN data collection rests with the Liaison Agencies, often together with agricultural research institutes. These either employ their own staff to visit the sample farms and to collect the data, or they contract this work out to accountants, universities, farmers' cooperatives or other organisations.
Confidentiality is maintained
Council Regulation No 1217/2009 of 30 November 2009 - the basic legal instrument establishing FADN, clearly states that data supplied to the Commission is treated in the strictest confidence.
Time period for data collection
The Community FADN Farm Return covers a period of twelve months. Member States have accounting years starting on different dates.
The Farm Return
The Farm Return is specified in Commission Regulation (EEC) No 2237/77 of 23 September 1977 and subsequent amendments until the year 2008 accounting included, then in Regulation (EC) 868/2008 from the financial year 2009. These same regulations contain detailed instructions on how the Farm Return is to be completed and provides definitions of the terms used.
The source of data at farm level
The required data are extracted from the appropriate inventory, cash book, ledger or journal kept by the farmer or field officer. In some Member States, the Liaison Agencies have drawn up special entry books to be completed periodically by the farmers.
Organisational structure for data collection
National committees comprise representatives of both the Liaison Agency and the bodies responsible for completing the Farm Return. They give guidance on the methodological aspects of the survey, such as the application of the European Union typology of farms, the selection of farms from the field of observation and the interpretation of terms. In some Member States, equivalent committees also exist at regional level.
DATA QUALITY CHECKS
The need for quality control
Decisions regarding agricultural policy in the European Union must be based on sound and accurate analyses. This means that FADN data themselves must be as sound and accurate as possible. The Liaison Agencies and the Commission take great care to ensure that any errors in FADN data are identified and corrected.
This chapter describes the procedures for ensuring a high quality of accounting data. The procedures used by the Liaison Agencies are outlined before the data is transmitted to the Commission. This is followed by a description of the procedures implemented by the Commission. The final section discusses the timetable for these procedures.
Quality control procedures followed by Liaison Agencies
Liaison Agencies invariably use one or more regional or national farm returns rather than the Community FADN Farm Return. This situation has arisen because many Member States were already running farm business/management surveys before the creation of a European network and thus had designed their own farm returns. Over time, the original farm returns may have been improved and adapted to suit the changing needs of users. They provide data firstly for the Member States' own purposes, and secondly, for FADN purposes. The objectives of the Member States may be different from those of the Commission. The Community FADN Farm Return is more limited in its coverage of farming activities than many of the regional and national farm returns.
Liaison Agencies use their own control procedures to maintain a level of data quality that may be higher or lower than the standard required by the Commission. When the data are considered error-free, the Liaison Agencies convert their national data to the Community FADN Farm Return as specified in Commission Regulation (EEC) No 2237/77 of 23 September 1977 and subsequent amendments until the year 2008 accounting included, then in Regulation (EC) 868/2008 from the financial year 2009. These procedures are summarised in a specific annex. Having done this, Liaison Agencies transmit the data to the Commission and submit it for inclusion with the quality procedures implemented by the Commission.
Quality control procedures implemented by the Commission
Quality control procedures at the level of the individual farm
The procedures implemented by the Commission for ensuring the quality of data at farm level are shown in annex. The first procedure is that of classification: all farms are classified according to European Union typology.
As detailed in the Methodology Chapter, to enable farm classification, Standard Output (SO) coefficients must be applied to each of the farm's enterprises. Farms must be classified by size and type before data controls can be performed. During this stage, farms may slip from the cell in which they were originally classified for selection purposes to another cell. Such a situation may arise because, for instance, the size of an enterprise may have changed between the time the farm was sampled and the end of the accounting year.
The second procedure is the running of the coherence tests. These consist of several hundred tests that try to detect and identify possible errors, inconsistent data and improbable values. Several levels of error are defined according to severity.
Errors can be of different types:
due to operator's mistake
due to mistakes in conversion programmes
Errors due to missing data
due to mistakes by the office completing the Farm Return
Type (4) errors - unlikely values - are identified by likelihood tests which compare the value of a variable, as given in the Community FADN Farm Return to the expected value. For instance, Liaison Agencies invariably have a good knowledge of crop yields in each region; a minimum and a maximum yield can thus be specified. The yield for each farm is compared to these expected limits. Farm Returns with values that fall outside the limits are then identified and examined.
The data quality control system operated by the Commission allows a flexible reporting of anomalies. It enables accountants and I.T. services in Liaison Agencies to locate:
errors that have arisen during the conversion of the national Farm Return to the Community FADN Farm Return,
errors due to misinterpretation of headings in the Community FADN Farm Return,
errors and omissions made during data collection.
It enables corrections to be made to individual Farm Returns and also identifies those returns that have a large number of errors and which may thus be rejected from the sample.
These tests help to create sub-samples that are appropriate for special analyses. They identify outliers, i.e. farms for which the value of one or more variables is significantly different from the mean value of the category to which the farm belongs. This may point to an error in data collection or data coding that was not found by the standard tests. Some farms may be outliers for logical reasons. For instance, in some regions pig farms generally operate on a small area of land because the pigs are permanently housed. It may be, however, that within a category of such farms there is one with an unusually large area of land because the pigs are "free-range" (allowed to forage in the fields and housed only at night). Such a farm will be identified as an outlier by the appropriate homogeneity test. When it is examined, the farm's special character will be revealed.
Quality control procedures at the aggregate level
The above section outlines the procedures applied by the Commission to individual farm data. When these have been completed, control procedures at the aggregate level are initiated. First the data are weighted and aggregated at the level of region, Member State, size class and type of farming. Continuity tests are then run. These compare the computed mean values of the standard set of variables to the expected mean values, i.e. the mean values that would be expected on the basis of previous trends.
For example, if the average land area of farms has been steadily increasing over the last three years, then it would be reasonable, all things being equal, to expect farm area to continue to grow at a similar rate. Thus for the current accounting year, mean farm area can be predicted (X ha) and compared to the observed value (Y ha).
The continuity test then computes the percentage deviation between X and Y. If this exceeds a predetermined threshold, then the computer programme generates a message giving both the percentage deviation and the absolute deviation. The data is then examined to see if there is a logical explanation for the apparent abnormality. If not, the data is corrected at the level of the individual farm.
The correction procedure is iterative during the quality procedures described above. Farm Returns that need to be corrected are processed by the Liaison Agencies and may have to be referred back to the regional level or to the original accounting office. However, some Farm Returns may be replaced because, in certain cases, national samples are larger than those required by the Commission for FADN purposes. Rather than correcting a Farm Return, it may be easier for the national Liaison Agency to replace it with another from the Member State's own sample.
Timetable for quality control and calculation of standard results
Once the data are accepted as error-free, the Commission can produce the Standard Results, perform other analyses and make forecasts (see Diffusion Chapter).
The completion of all the procedures needed to produce the Standard Results should take no more than 15 months from the end of an accounting year. In practice, some participants of the FADN have not always been capable of respecting the timetable set out in the legislation. Consequently, there happened delays in the publication of aggregates for the European Union as a whole. Constant efforts are made to follow the timeliness of FADN operation.
The following table shows the planning for the data quality checks for the accounting year N.
First quarter N+2
Second quarter N+2
By 31 July N+2
By end of September
delivery deadline for most of the Member States
Commission does basic data checks and exchanges comments, if any, with Member
results are published following the finalisation of the basic data checks
Commission has sent detailed feedback to all Member States
Commission has received explanations from all Member States and the data are
Commission presents Standard Results to the FADN Committee and data are made
available to users
Commission sends feedback on full data checks to Member States
States should reply within one month with explanations or corrections of the
data - more than one round of data verifications may be needed
soon as the data are final, facts-sheets with main results at Member State
level are published
required for data validation process: 10 months