Establishment and update of the statistical register
Using information from the administrative farm register (AFR), held by RVO*), and census information of several years Statistics Netherlands has built up a statistical farm register (SFR). Since 2015 information from the Business Register (BR), held by the Chamber of Commerce (Kamer van Koophandel), is used to update the AFR. From 2016 onwards only holdings with a registration in the BR compliant with Regulation 1166/2008 are included in the SFR. Registration in the Business Register is compulsory for every company and almost every legal entity in the Netherlands.

With the information from the BR and the AFR the SFR forms a virtually complete register of agricultural holdings in the Netherlands, thus providing a perfect basis for stratification and efficient sampling of subsequent agricultural statistics. The SFR contains name, address and several other characteristics of the holding and a unique registration number.

*) RVO: 'Rijksdienst voor Ondernemend Nederland' (‘Netherlands Enterprise Agency’) is part of the Ministry of Economic Affairs and works at the instigation of ministries and the European Union. RVO is a.o. responsible for the ‘combined data collection’ for the agricultural census, CAP subsidies, and enforcement of the manure law.

The main legal base for national policy information is the Agricultural Act. Furthermore many questions in the national census are based on the Manure law. Since the combined data collection also provides information for CAP subsidies the legal framework also includes the CAP regulations. Finally, since information from the agricultural census is also used for EU statistics (a.o. crops and livestock statistics), these regulations can also be considered part of the legal framework.

Each year the announcement to hold an agricultural census is published in the Official Government Gazette.

For Statistics Netherlands the agricultural census is part of the working program. The working program is established by the Central Commission for Statistics, an independent body in which many sectors of the society are represented. The status of the Central Commission for Statistics is established in the Statistics Netherlands Act.

Respondents receive a letter by mail, in which they are summoned to participate in the census. In case of lack of clarity or incompleteness of the forms the correspondents are contacted by checking personnel during the processing of the survey forms.

⁽¹⁾ according to Regulation 1166/2008: a single unit, both technically and economically, which has a single management and which undertakes agricultural activities listed in Annex I of the Regulation 1166/2008 within the economic territory of the European Union, either as its primary or secondary activity.

The economic size of the holding is determined by an extended size calculation. This uses more detailed SO-coefficients than the FSS (SO-coefficients for all extended crops and livestock characteristics). Extensive analysis, based on the agricultural census of 2009, was carried out to determine the level of the SO-threshold. A threshold of 3000 euros SO proved to be fully compliant with the physical thresholds of the FSS regulation.

Note: Strictly taken holdings with less than 0.7 ha hops (corresponding with 3000 euro SO) will be left out of the survey. However in practice it will rarely occur that holdings only have hops, so in practice (virtually) all holdings with hops will be part of the census. Furthermore hops is very small in NL (only 2 holdings, with 4 ha in total); until 2015 it was not even distinguished as separate crop in the agricultural census.

Reference date 1 April 2016:
• general characteristics;
• livestock;
• other gainful activities of the holding.

Reference date 15 May 2016:
• land.

Reference period 1 year preceding the reference date  (April 2015 – March 2016):
• labour force;
• irrigation.

Reference period 3 years preceding the reference date (2014, 2015, 2016):
• support for rural development.

In the following table, the working process phases are placed in time.

1. Determination of the content and format of the questionnaire
A number of organisations active in the agricultural sector in the Netherlands is involved in this process. The data need is inventoried involving a group of experts. This group of experts forms a permanent working group on the census characteristics. The working group meets every year in March or April to discuss the characteristics of the census of the next year. After approval of the list of characteristics, drafts of the questionnaire and explanatory notes are drawn up and validated by the working group. A large number of questions and characteristics are constant over the years. A small number of questions expires, changes or is newly introduced (if properly motivated). In 2016 the additional questions required for the Farm Structure Survey 2016 were added to the list of characteristics.

2. Development phase
Statistics Netherlands and RVO, in cooperation with the expert group, prepare the questionnaire by formulating the questions and the explanatory notes. Since 2006 it is possible to fill in the questionnaire using a web application. The number of respondents using the internet has since rapidly increased, and in 2016 the use of the web application became compulsory. RVO is responsible for building, testing and maintaining the web application. In the web application definitions and explanatory notes are available through information buttons. Controls are, as far as possible, integrated in the web application.

3. Training phase (RVO)
RVO is responsible for the training of the employees. Training is particularly aimed at temporary personnel for the helpdesk and the controlling process. Partly, this training is on forehand and partly it is on the job.

4. Data-collection phase(RVO)
Each holder receives a summon to participate in the census before the starting date of the survey (1 April). Farmers have to fill in the form before 15 May. Farmers that do not respond receive several postal reminders. After 15 May farmers may be fined or may receive less financial aid. The data-collection phase ended 1 September. The electronic forms are stored in the RVO database.

5. Pre-processing phase (RVO)
Many checks are already implemented in the web application. After the respondent submits his form, additional checks are performed by RVO. If necessary the respondent is contacted to check the given answers or to correct errors. During this phase data is regularly sent to Statistics Netherlands for further checking and analysis. By the end of January 2017 the pre-processing phase ended; at that time 95% of the forms were received and processed.

6. Processing and analysing phase
At Statistics Netherlands a more advanced check is performed on for instance internal contradictions and differences with the trends and results of previous years. Calculations are made for composed indicators. The data are rearranged, selected and transformed for dissemination purposes. Since the agricultural census is exhaustive the results are used as a sample base for other statistics such as yield assessments, labour and productivity statistics and environmental statistics.

7. Dissemination Phase
The main publication is the Statline database which is available on the internet (www.cbs.nl). In this database internet-users may select their own indicators and information topics. First results from the 2016 survey were published in September 2016 and the final results in April 2017. Several times a year short publications on specific subjects are presented in the form of newspaper reports or internet-magazine articles.

8. FSS delivery phase
The first phases focus primarily on the national census characteristics. After finalization of the national census characteristics attention shifted to the FSS characteristics. After additional analyses and checks against the FSS validation rules the dataset was delivered to Eurostat.

9. FSS/SAPM validation phase
After the delivery of the FSS dataset the validation process started. In this phase the validation tables were evaluated and the NMR was drawn up.

The information of the exhaustive agricultural census provides the basis for sample enumeration of more specific agricultural research, like crop yield predictions. Methods and designs used for those statistics are not discussed in this paper.

From 2016 onwards all holdings in the AFR with NACE 01.1 to 01.5 (with the exception of most of 1.49^*), and holdings in 01.6 that keep land in good agricultural and environmental conditions are selected for the census/FSS. During processing NACE is checked against the actual BR registration.

*) Holdings in 1.49 include raising and breeding of rabbits and fur animals. Raising and breeding of fur animals is not eligible for FSS; the other eligible activities for FSS under 1.49 are non-significant.

2. the Integrated Administration and Control System (IACS)
IACS is held by RVO. IACS is the control system for direct aid and agri-environmental schemes of the CAP, based on Council Regulation (EC) No 73/2009. IACS contains the same identification number as the AFR. The IACS register is continuously updated.  

3. the I&R (animal Identification and Registration) register on bovine animals
The I&R register is held by RVO. The register is based on Regulation (EC) No 1760/2000, which establishes a system for the identification and registration of bovine animals. The I&R register is continuously updated.

We did not participate in the conceptual design and subsequent related revisions of the administrative sources. However in the preparation and design of the questionnaire there is close cooperation with (governmental) holders of administrative sources in order to establish whether these meet the requirements (of FSS and agricultural statistics) and can be used (a.o. timeliness). Policy is that only data that cannot be derived from administrative sources can be asked directly.

The electronic form provides a number of online controls and validations, diminishing the need for additional telephone interviewing. Also the respondent receives only relevant questions (for example: questions about livestock only if there are animals on the holding), thus reducing the administrative burden.

During the processing and analysis phase farmers can be contacted by telephone in order to receive missing data or corrections of errors.

For efficiency reasons the questionnaire of the agricultural census nowadays also serves as an application form for the single payment scheme and animal, crop and arable land subsidies in the framework of the European Common Agriculture Policy (CAP), as well as for the enforcement of the minerals regulation (manure). The combined questionnaire and application form is extended with geographical information on the location of the crops (parcel registration).

This policy also means that registers and/or administrative data have to be used whenever available and suitable (under the condition of quality and continuity).

The mailing list for the combined data collection is based on the AFR. Much effort is put into taking care that the information in the AFR is up‑to‑date (both  in coverage and actuality). Since 2015 information from the BR is used to update the AFR; also other registers such as municipal register are used. This reduces response burden by preventing that respondents may be approached wrongly or more than once.

During data collection non-respondents receive several reminders, and risk being fined, and since the questionnaire also serves as application form for subsidies there is a strong financial incentive to timely fill in the form (after a certain time there will be cuts in payment).

In case of a census 1=3+4+5

Only for sample survey, in which case 2=3+4+5

these comprise 2 650 holdings below the threshold of 3000 euro SO, and 3 611 holdings with ceased activities

This item is a subset of 3.

4>4.1+4.2

this consists of new or changed non-responding holdings, for which there is no information for imputation available (which means that their activities are unknown, and it cannot be determined if they fall below the threshold of 3000 euro SO).

5=5.1+5.2

5.1>=5.1.1+5.1.2

6=5.2+5.1.2+4.2

Controls at data-entry
All questionnaires were filled in online. The online questionnaire already contains a lot of controls and validations, documented in the ‘functional design’ of the application.

Data validation is performed in two successive stages.

Data control at RVO
After data collection the data goes through an extensive error control phase. In this phase (additional) checks are made on missing values, valid values, unlikely values, range checks, checks of correlation in the data, checks of totals and so on. When necessary additional information is collected from the farmers by phone. Data that is checked and accepted by RVO is forwarded to Statistics Netherlands.

Control at Statistics Netherlands
Data processing by Statistics Netherlands involves further checks for hard and soft errors, and enrichment with additional information, such as total SO and typology. At Statistics Netherlands non-response imputations are performed (see section 6.3.4 Processing error). Further analyses are made at several levels of aggregation and comprise comparison with previous results and agricultural data from other sources.

The number of characteristics in the census is quite high, and more detailed than needed for FSS. To maintain continuity in the data most of the characteristics are repeated annually, or in a regular scheme. Of course developments in agriculture are reflected in changes in the list of characteristics. The topics that appear every year in the census are:
• land: arable crops, horticultural crops, grasslands, rough grazings, fallow land, natural territory,
• livestock: cattle, pigs, poultry, horses, rabbits, fur animals,
• labour force: holder, family and non-family, regular and non-regular.

For characteristics that do not change very fast, annual data are not necessary. For instance the area of irrigated land, the presence of a successor to the present holder is included in a list with a lower frequency. Major additions of this kind occur in the years when the Farm Structure Survey is held. Additional FSS characteristics are combined with these non-yearly topics if possible.

When preparing the census, there is an almost constant pressure to introduce new items in the questionnaire, thus extending it bit by bit and constituting a threat for the project as such. Meanwhile there is increasing pressure to reduce the administrative burden. New proposals can therefore only be taken into account if provided with proper motivation and legal base. Inclusion of new characteristics has to be accompanied by deleting less relevant characteristics. An impact analyses is part of decision process.

Several FSS characteristics are integrated to meet other EU-requirements and national information needs. Other EU-requirements comprise a.o. crops and livestock statistics; the national information needs are mainly for enforcement of the manure law, provision of subsidies and emission calculations. 

Categories of extended characteristics:
• livestock, mainly for enforcement of the manure law and emission calculations;
• arable crops, mainly for enforcement of the manure law and subsidies;
• horticultural crops, because of the specific economic importance of this agricultural sector in the Netherlands.

Although bias risks associated with non-response are low because of the very high response rate (about 96%), a separate analysis was performed, comparing trends and distribution between the imputed non-response units and the response units.
The analysis showed no difference in trends and distribution between the imputed non-response units and the response units. A final check comes from the evaluation of the census results.

There were only a few new characteristics compared to previous year, so almost all characteristics could directly be imputed from previous year. The new characteristics, that constitute a more detailed description of a characteristic from previous year, were calculated using analysis results from corresponding response units. The few other new characteristics were considered not to be present in the non-response unit.　

Note: The information imputed from previous year may also be based on imputation (because the unit did not respond in that year either). However, the use of imputed data is maximized to five years (units not responding for 5 years or more are assumed to have ceased activities and were removed from the population).

Completeness of online forms is largely enforced by the application. When (possible) errors are detected during the data-control phase at RVO or before finalization of the control phase at Statistics Netherlands, the farm holders are contacted to correct these errors. If errors are detected later on, an impact analysis is made in order to decide whether or not a correction should be made. The way these corrections take place (call-back, imputation, expert estimate) depends on the type and severity of the error.

Data collection ended 1 September 2016. The first provisional results were published already by the end of July 2016 (only main crops at country level). The first provisional results of all the regular census subjects at country level, were published by the end of September 2016, and the first complete set of preliminary results (all regular census subjects at country and regional level) were published in November 2016.

Final results of the 2016 census were published April 2017, 4 months after the last day of the reference year.

Publication of regular census subjects takes place according to a pre-determined release policy. The 2016 census was published according to the release plan.
Publication of FSS specific topics is on an ad hoc basis (on request or in special publications). Until now there have been no publications on FSS2016 specific topics.

The number of hours used to calculate Annual Work Unit is 2000.
Note: actual calculation of AWU is based on the average working hours per week, where 40 average working hours per week is considered 1 AWU, which corresponds to about 2000 hours per year.

Explanation:
According to Regulation 1166/2008 the scope of the FSS is determined by the agricultural activities listed in Annex I (i.e. NACE 01.1 – 01.6). Up until 2016 only information from the AFR and the combined data collection could be used to determine if a holding was part of the scope or not. Information in the AFR did not include NACE as such, but was determined from a temporary assessment upon registration of the holdings at RVO, and verified using information from the combined data collection. This however meant that some ineligible holdings (e.g. riding stables or children's farms) could not be uniquely identified. From 2016 onwards additional information from the BR is used, meaning NACE is available and can be used to determine eligibility. As a result, ineligible holdings, that were included in the scope before FSS 2016, could now be distinguished and excluded. These holdings mainly consist of very small farms without a registration in the BR (hobby farmers), riding stables, and organisations in nature preservation, resulting in a decrease of mainly sheep, equidae and other land. The new delineation has virtually no effect on arable land, permanent crops and most of the livestock (except for a large effect on sheep and equidae, and limited effect on goats and cattle).

Note:
The sheep on holdings not in NACE 01.1 – 01.6 are kept mostly for nature preservation (and some hobby farming) and not for meat or milk. 
As 'hobby farmers' are considered: natural persons engaged in agricultural activities, not registered in the BR (which is obligatory when engaged in economic activities), and holdings in the BR (and NACE 01.1 - 01.6), but below the thrreshold of 3000 euro SO.

FSS statistics for the Netherlands are coherent with other statistics within the agricultural domain (crops, livestock, ..).
Because of differences in definition of the units this may not be the case over domains.

Note:
We have not made a numeric analysis of the IACS register. This is because we do not use IACS as a separate register and link the information to FSS, but data is collected for FSS, subsidies (IACS), enforcement of manure law and emission registration combined. This means we do not have the IACS register as such, but we use the collected information directly (information that ultimately will also be taken up into the IACS register). From the data we receive from RVO we cannot directly identify the holdings that are part of IACS.

Dutch tables (under 'Landbouwtelling en biologische landbouw'):
- Landbouw; nationaal (crops, livestock and land use at national level), approx. 5000 hits
- Landbouw; regio (crops, livestock and land use by region), approx. 1000 hits
- Landbouw;gemeente (crops, livestock and land use, by municipality), approx. 5000 hits
- Landbouw;bedrijfstype, nationaal (crops, livestock and land use by farmtype at national level), approx. 1500 hits
- Landbouw;hoofdbedrijfstype, regio (crops, livestock and land use, by general farm type and region), approx. 3500 hits
- Landbouw; arbeidskrachten, regio (labour force by region), approx. 500 hits
- Landbouw; klassenindeling,regio (crops, livestock and land use, by size classes and region), approx. 500 hits
- Landbouw; economische omvang (information by economic size classes and farm type), approx. 1000 hits
- Landbouw; economische omvang, regio (information by economic size classes, general farm type and region), approx. 500 hits
- Landbouw; verbreding (other gainful activities by main farm type and region), approx. 500 hits
- Landbouw; biologisch, fase omschakeling (organic farming by stage of conversion), approx. 500 hits
- Landbouw; biologisch, nationaal (organic farming at national level), approx. 1500 hits
- Landbouw; biologisch, regio (organic farming by region), approx. 500 hits

English tables (under agricultural census):
- Agriculture; crops, livestock and land use by general farm type, region, approx. 1500 hits
- Agriculture; labour force by region, approx. 100 hits 

Statistics Netherlands also has several micro-data services for research, see:
https://www.cbs.nl/en-gb/our-services/customised-services-microdata

services are:

• Customized research, conducted by Statistics Netherlands
https://www.cbs.nl/en-gb/our-services/customised-services-microdata/customised-research

Statistics Netherlands can carry out research based on the research questions of the applicant. If necessary they can add and link researchers own datasets to statistical data sources available at Statistics Netherlands (including the agricultural census). The results of the study are published on the website of Statistics Netherlands and are thus accessible to everyone. Moreover, most publications appear in print as a research report. This work is carried out for a fixed rate per hour.

• Own research using data from Statistics Netherlands
https://www.cbs.nl/en-gb/our-services/customised-services-microdata/microdata-conducting-your-own-research

Under certain conditions, Statistics Netherlands makes available micro-data for statistical research. To be authorized to use this data the researcher must work at an organisation authorized by the Central Commission for Statistics (CCS). Research can be conducted on site at our offices in The Hague and Heerlen, or from the researchers' own workplace using a secure internet connection (remote access).

All datasets remain on the dedicated server at Statistics Netherlands. Before statistical results are released, all data is checked for the risk of disclosure. Statistical results (e.g. statistical tables) can be relocated so they can be used outside the secure environment.
Statistics Netherlands requires that all statistical results be published and made available to other interested persons and organisations. Statistics Netherlands publishes an overview of publications based on research based on micro-data from Statistics Netherlands.

• Advice on statistical research, see:
https://www.cbs.nl/en-gb/our-services/customised-services-microdata/advice-about-statistical-research

Statistics Netherlands can give advice about the set-up and implementation of statistical surveys and methods.

• Working group "Structure of Agricultural Holdings": Handbook on  FSS definitions 2016 (FSS WG/2015/1/11 revision 1)
• Working group "Structure of Agricultural Holdings": Eurofarm Manual for Data Suppliers Farm Structure Survey 2016 (Doc. FSS2016/DSM rev. 9)
• FO eGDI2016: functional design internet application, RVO

Besides regular publications via  the Statline database, the website or press releases, requests for information can be addressed at our Infoservice. In the case custom-made tables are prepared confidentiality rules are applied whenever needed.

For custom-made tables a distinction is made in information on 'sensitive' and 'non-sensitive' characteristics. Sensitive characteristics are characteristics that could lead to disclosure of individual personal data (age of the farmer, education level, ..) or of economic data on the individual farm (SO). For this type of tables the confidentiality regime is very strict and provisions are made that each cell contains enough cases so that no individual farms or farmers can be identified, otherwise the cell will be kept confidential (and special attention is given to secondary disclosure). For non-sensitive characteristics (e.g. number of animals or crop areas) a less strict regime is used (often these characteristics can be easily observed, or are directly available from farmers own website). In these cases it is accepted that information on individual farms may be derived. In all cases privacy protection has the highest priority, therefore all requests for custom-tables are scrutinized.