- all agricultural holdings with more than 2 hectares of arable land,

- holdings with a large number of livestock but with less than 2 hectares of arable land, and

- holdings with horticultural production.

Since its establishment the farm register was used as a sample frame for both farm structure surveys and other agricultural statistical surveys.

During the 1990s, the farm structure surveys were subject to few methodological and technical changes. The substitution of some censuses with sample surveys, together with the processing of statistics in PC-environment, led to a reduction of the costs for producing agricultural statistics. However, the substitution of some censuses with sample surveys inevitably led to lack of agricultural statistics on municipality level for the years the substitution occurred.

The Swedish accession to the European Union in 1995 created the need for adapting national agricultural statistics to the EU legislation. Until 2001, Sweden conducted farm structure surveys annually, switching every year between EU and national legislation. The main difference between these two consisted on the number of the characteristics surveyed. The national farm structure surveys met primarily national statistical requirements and therefore were far less extensive than the ones based on EU legislation. In 2001, national farm structure surveys were abandoned, thus embracing the surveys based on EU legislation as the sole Farm Structure Survey. Before 2001, the data collection for the EU Farm Structure Surveys was made through postal questionnaires to the farmers. Since 2001, however, the data collection methods included also web questionnaires, administrative registers and telephone interviews with the farmers.

Through the Official Statistics Ordinance (SFS 2001:100), promulgated in March 2001, the Swedish Board of Agriculture became the sole authority responsible for organizing and producing all agricultural statistics in Sweden. Until March 2001, this was the responsibility of Statistics Sweden.

1) the land register designation, area, land use and ownership and lease situation with respect to the property or part of the property where the business is carried on;

2) the occupations of the persons employed in the business and the year of engagement;

3) the keeping of livestock.

The owner of a property where another person is carrying on a business referred to in the first paragraph shall provide the name and address of that person and information concerning the area used in the business.”

In the provision SJVFS 2016:7, which took effect on April 17, 2016, the obligations of the farmers were further clarified. Through this provision, all agricultural holdings meeting the criteria set for the target population, were obligated to provide the requested information to the Swedish Board of Agriculture within the deadline set for the survey, June 23, 2016. The information provided should reflect the situation on the agricultural holding on the survey’s reference date, June 2, 2016. Holdings failing to provide the data requested could be subject to a penalty in accordance with the above mentioned act.

IACS - Integrated Administration and Control System

• holdings operated by natural persons,
• holdings operated by legal persons (estate of deceased person, Limited Liability Company, state, municipality, church, other).

• more than 2.0 hectares of arable land – outdoor, Fruits, berries and nuts (excluding citrus fruits, grapes and strawberries) – outdoor, Nurseries – outdoor, Short rotation coppice areas,
• at least 5.0 hectares of UAA (A_3_1),
• at least 200 m² area under glass (B_1_7_2+B_1_8_2+B_4_7),
• 2500 m² outdoor horticultural cultivation (B_1_7_1_2+B_1_8_1+B_4_1+B_4_2+B_4_3+B_4_4+B_4_5),
• at least 10 cattle (C_2) or 10 sows (C_4_2) or 50 pigs (C_4) or 20 sheep (C_3_1) or 1000 poultry (including chickens) (C_5).

-          one survey manager with overall responsibility,

-          one project manager with responsibility for writing the national legislation and recruiting and leading the FSS team,

-          four statisticians responsible for designing the population and the sample frames in accordance with EU precision requirements; for setting up criteria and procedures for data controls; developing methods for handling missing or incorrect data items, estimation procedures and other methodological issues,

-          five persons working with the design of the questionnaires and the information booklets,

-          two persons working with IT-applications, scanning definitions, web applications and archive functions,

-          two persons responsible for establishing the final register and for transmission of data tables to Eurostat,

-          14 persons working with the data collection: registering the questionnaires sent in by the respondents, doing basic checks on the data collected, processing of information and collecting the complementary information by telephone calls to farmers.

As the FSS 2016 was designed in a similar way as the previous farm structure surveys (from 1993-2013), there was no need for conducting a pilot survey.

The promotion of the census was done through the website of the Swedish Board of Agriculture. Two ads were also posted in two of the largest agricultural newspapers in the country.

Calendar overview

All agricultural holdings received a questionnaire. Questions regarding the main characteristics (on the holding, total areas, livestock excluding cattle, some important characteristics to locate the holding in registers and information on when the total holding is sold, rented or in some ways been closed) have been included in all questionnaires. A sample survey has been sent to approximately one third of the population where also questions about irrigation, other gainful activities, labour force, tillage methods, and manure application techniques have been included. 

For some characteristics, administrative registers were used to obtain data. Five administrative data sources were used during FSS 2016: IACS, the Bovine Register, the Organic Farming Register and the Register of Support for Rural Development. During the imputation process, the slaughter register, the Swedish education register, and the taxation register were also used. 

We combine IACS with the cattle register, and the last known value for other animal and horticulture production.

• counties (NUTS 3),
• area of agricultural land,
• number of animals of different kinds,
• specialised holdings with turkeys, hens, chickens, gilts,
• new holdings according to pig-, poultry-, sheep register and the horticultural division of the federation of Swedish farmers (GRO) plus holdings from the “Register survey”,
• holdings that do not longer exist in IACS or other registers.

The principles for building up strata were almost the same as for earlier farm structure surveys. However this time we also made a poststratification due to the fact that 634 new holdings were discovered in IACS after the frame was finalised. The total number of strata in the end in the file was 232 due to this issue.

• in the strata with holdings of large size in some respect. Total survey of these strata was important for acquiring high statistical precision and thus producing statistics with high quality.
• in all the strata containing new holdings from different animal registers. This was done because of the lack of information for these holdings which made it difficult to use correct pre-stratification.
• for holdings with more than 100 ha cereals + green fodder, more than 800 pigs, more than 100 sheep;
• for holdings with a lot of labour force, holdings with a lot of irrigation, holdings with large areas of fruit trees, holdings with a lot of turkeys, holdings with a lot of hens, holdings with a lot of chickens, holdings with a lot of boars and new holdings without animals and pasture of more than 10 ha.

The Bovine Register, in Sweden called the Central Cattle Register (CDB), was set up in accordance with Council Regulation (EC) No 1760/2000 of 17 July 2000. The register’s main objective is to make it possible to trace contagious animal diseases. The Commission approved the CDB in Sweden for use in pursuing this main objective from 1 November 1999. The register has also been approved to be used for other objectives and since 2000 it has been used, and served as the basis, for various EU cattle premiums (special beef premium, slaughter premium, suckler cow premium, extensification aid, etc.).

The CDB is made up of the following two main parts:
- One part contains information on the identity of individual animals, both living and dead, such as sex, date of birth, breed and age.
- Another part contains information on events relating to individual animals, such as birth, slaughter, purchase and sale as well as transfer between different holdings, etc.

Farmers with bovine animals must report each event for each individual animal, such as birth, transfer, sale for slaughter, etc., to CDB within seven days. A calf should be tagged within 20 days of birth and then reported within seven days to CDB. Late reporting will incur additional charges for the farmer. The reference date for the transmission of the information from CDB to the farm register was June 2, which is the same date as the reference date for the FSS 2016.

The Organic Farming Register is based on the Council Regulation (EEC) No 2092/91. In 2016 there were three control bodies on the market working with controls of agricultural holdings on which organic farming methods are applied. The control bodies are authorised by the Swedish Board for Accreditation and Conformity Assessment. The data concerning the characteristics on organic farming in FSS 2016 are based on existing information on individual holdings in registers at the control bodies.

The register of Support for Rural Development is based on the Swedish Board of Agriculture and contains information about all holdings seeking support for different kinds of rural development.

There is also a number of registers used to build the sampling frame and to give information for having the best possible sampling design. These registers are livestock registers for pigs, sheep and poultry combined with registers on slaughtering of animals. 

During the imputation process, the slaughter register, the Swedish education register, and the taxation register were also used. 

The Swedish Board of Agriculture security classifies all documents, databases etc. according to their content of information. For all information, whether it is in a database or in an Excel sheet, there is a person that is designated as the owner of the information. Risk analyses are made each year. The statistical unit is not included in the conceptual design of the administrative sources. However, due to the total response burden put on farmers we have the possibilities to make our opinion on possible deletion of data from administrative sources.

CDB: Metadata for CDB is described in a separate database table. Here, the property is described for all values stored, type label, description, field length, type of field, if included in the xml, where it comes from device, where to save, etc. The information on metadata is saved since the beginning of CDB.

Organic Register: Metadata for the organic register is described in a electronic way.

Register of Support for Rural Development: Metadata for the register of support for Rural development is described in a separate database table. Here, the property is described for all values stored, type label, description, field length, type of field, if included in the xml, where it comes from device, where to save, etc. The metadata information is saved since the beginning of register.

The main linkage between the holdings in the statistical farm register and the reporting units in CDB is the personal number for natural persons and organisational number for legal enterprises. These numbers are unique for each entity. In the farm register there is for every holding registered one personal number for each farmer on the holding. For units that mismatch according to this method, the linkage characteristics production location is used.
There can be some minor problems with linking some of the cattle to the right holding. This is because we may not have details of all of the persons in each holding. In that case, we may have to start a new holding.

Organic Register: The registers at the control bodies cover all holdings applying organic farming methods or are in the conversion period. The control bodies transmit the following information to the Board of Agriculture every year:

• the personal/organisational number of the holder or holding,
• the client number of the holding at the control body,
• what kind of crops are grown and the corresponding areas in combination with if the areas were under conversion or converted,
• organic production methods applied in the animal production.

Information on addresses and telephone numbers of the holder/holding are also transmitted by the control bodies.

In the FSS questionnaires, holders were asked to fill in their client number at the control body. By using the personal/organisational number of the holding combined with the information on client number at the control body and in the FSS questionnaires, a link between the Organic Farming Register and the statistical farm register could be established. In cases of mismatch, addresses and telephone numbers were also used to link holdings in the different registers. In the end, there were only a few mismatches left.

Register of Support for Rural Development: The definition of the reporting unit is the same as for the holdings applying for the single farm payment.
Hence, the identification of the units is also the same, i.e. the personal number for natural persons and organisational number for legal holdings. Should there be a mismatch, the client number at the Swedish Board of Agriculture is used.

1. sorting of incoming paper questionnaires (SF1, SF2, SJ1, SJ2, L),
2. registration of the questionnaires,
3. optical scanning (hereafter Scanning),
4. verifying,
5. transfer of data from the IT-system to the database.

When the questionnaires arrived at the Swedish Board of Agriculture, they were sorted by the FSS team in three different groups - one per kind of questionnaire (1). Then by scanning the barcode, each questionnaire was registered on the IT-system (2).

Written reminders by post and by email
During the data collection procedure, three reminders were sent out to the holdings which did not provide information within the deadline for the survey (June 23, 2016). An e-mail reminder was sent to holdings we had an email-address to, on June 21, 2016. The first written reminder was sent out on June 29, the second one on July 25, while the third one was sent out in August 20, 2016.

Telephone interviews/Follow-up interviews
Telephone interviews were used to complete partly answered questionnaires in order to reduce the non-response. The highest partial non-response rate was on the section of labour force and rural development of the FSS questionnaires.
Telephone interviews and follow-up interviews were planned in such a way as to give priority to important large holdings. About 5 % of the holdings were telephone interviewed by the FSS staff. Telephone interviews were conducted between August 10- October 15, 2016.

In case of a census 1=3+4+5

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

This item is a subset of 3.

4>4.1+4.2

5=5.1+5.2

5.1>=5.1.1+5.1.2

6=5.2+5.1.2+4.2

Questionnaire 1 - This was the smallest questionnaire where only data on areas and animals was collected. This questionnaire was received by all holdings that were not included in the FSS-sample.

Questionnaire 2 - In this questionnaire, all variables that were to be collected in the FSS were included, except data that was collected from administrative sources. The population that received this questionnaire were natural persons that were not included in a sample survey on manure and tillage methods conducted by Statistics Sweden in the same period.

Questionnaire 3 - The same questionnaire as questionnaire 2 except the question on manure and tillage methods that was deleted from this questionnaire because these holdings were included in the sample survey on the same topics conducted by Statistics Sweden. The holdings that received this questionnaire were natural persons.

Questionnaire 4 - In this questionnaire, all variables that were to be collected in the FSS were included, except data that was collected from administrative sources. The population that received this questionnaire were legal persons that were not included in a sample survey on manure and tillage methods conducted by Statistics Sweden in the same period.

Questionnaire 5 - The same questionnaire as questionnaire 2 except the question on manure and tillage methods that were deleted from this questionnaire because these holdings were included in the sample survey on the same topics conducted by Statistics Sweden. The holdings that received this questionnaire were legal persons.

The differences between questionnaires 2 and 4 and between questionnaires 3 and 5 are the questions on labour force where the questions were put in different ways. In questionnaires 3 and 5, questions on manure and tillage methods were not included.

• completeness checks (database),
• summation and arithmetic checks (IT-system),
• checks of extreme values (outliers) (IT-system),
• logic and relational checks,
• checks of unusual combinations (database).

Data, which did not meet one or more of the checking criteria, were checked manually. In many cases, the identified errors could be corrected using data from the questionnaires. In some cases, the respondents were contacted by telephone and in those cases also such information was asked for, which had not been completed in the questionnaires. The instructions, which were in force for the total survey process, generally permitted the data to be corrected or completed by the staff directly on PC.

The software used in this part of the process was SAS, and Excel.

Data validation has been done at central collection centre.

a) information about the holding: holder’s name, address, personal or organisational number, telephone number, e-mail address, client number in the administrative register for single farm payment, client number in the register of organic farming (at the control body), and production location number for bovine animals,

b) areas of different types of land,

c) areas of different crops,

d) set-aside areas under the EU aid programme,

e) number of livestock of different kinds,

f) organic farming: utilised agricultural area (converted and under conversion) and organic production methods in animal production,

g) farm labour force,

h) other gainful activity,

i)  irrigation,

j)  rural development,

k) soil cover,

l)  manure application techniques.

The statistical characteristics b)-l) were collected in order to produce the customary annual national statistics of farm structure as well as to fulfil the EU requirements on agricultural statistics.

The farm register was updated when respondents reported changes on holdings according to the instructions in the questionnaires. When the reported data were inadequate, further investigations were necessary. In general, the remaining degree of under-coverage was considered to be so small that no correction for this was needed.

To minimise the risk for duplicates, all holdings with more than 10 hectares of arable land and not applying for single farm payment were removed. Holdings above 10 hectares are therefore assumed to apply for subsidies. Of course this assumption is not correct in every single case, but it is assumed to be the best estimate of the total number of holdings. About 600 holdings were eliminated this way.

Due to the low non-response we did not do any further analysis in the bias from the non-respondents.

The weights were recalculated for the unit non-response on the sample characteristics.

The partial non-response has been considered to be the most serious type of non-sampling errors.

To avoid bias from partial non-response, much work has been done to create new imputation methods minimizing the risk for bias from imputation.

In the section on other gainful activities, there was a non-respnonse on 27 % of the respondents.

In the section on manure application techniques, there was a non-response on 55 % of the respondents.

For the section General characteristics, general information about the holding and the holder was obtained from different registers.  These registers include: pig-, sheep- and fowl registers; registers on memberships for different horticultural associations; Integrated Administration and Control System (IACS) for crop areas and CDB.

In cases where values were missing for a whole section of the questionnaire, such as the farm labour sector, the whole section was imputed with values from a similar farm which had answered this part of the questionnaire. A stratification based on the Swedish typology was used to categorise the holdings for these imputations. For the livestock section, the principle of last known value applied. For the irrigation section on the other hand, the last known values were used together with logical checks against the utilised agricultural area.

Single missing values were only imputed in few cases, and mainly within the labour force section. If for example the number of working hours was missing for one or more persons on the holding, then standard working hours on the holding were indirectly used for imputation. Further, with help of standard working hours on the holding the total expected number of hours worked on the holding was estimated. However, since the total standard working hours differs from the hours filled in by the holdings in the questionnaire, the quota between these two values on national level for 2010 was used as correction factor. These ´imputation hours´ were then distributed among the persons with missing value for the number of working hours.

The instructions, which were in force for the total survey process, generally permitted the data to be corrected or completed by the staff directly on PC.

• Livestock: 4 656;
• Agricultural holdings and holders: 1 300;
• Use of agricultural land: 3 872;
• Labour force: 288;

For January-March 2017, there was the following number of consultations:

• Livestock: 1 110;
• Agricultural holdings and holders: 443;
• Use of agricultural land: 1 099;
• Labour force: 79.
  
  

Code of Practice by the Statistics division at the Swedish Board of Agriculture.

For each publication made, we have an advanced documentation, describing in-depth all the steps required until the final publication.

We also made a special survey on drainage and irrigation at the same time as the FSS.

We also have a continuing improvement to collect every person that works in the holding. This is important, because we use that information to connect the holding to different registers that we have.

We are also planning to develop a completely new IT-system for data collection to be used during the Farm Structure Census 2020. The IT-system will be finished and tested for smaller surveys during 2019.