Holdings that reach specific minimum requirements in the size of utilised agricultural area (UAA) or for livestock or the crop areas for specialised crops are obligated to respond to the FSS. Due to structural changes in agriculture, the coverage thresholds have been adjusted in various ways over time, most recently prior to the 2010 Census of Agriculture. The survey program of the FSS, which provides an overview of the structure of agricultural holdings in Germany, was also adjusted over time to the respective prevailing data requirements for agricultural policy. Effects on chronological comparability of the data are mentioned in the respective methodology reports.

• Federal Statistics Law (Bundesstatistikgesetz - BStatG) of 22 January 1987 (Federal Law Gazette I p. 462, 565) as amended
• Law on Agricultural Statistics (Agrarstatistikgesetz - AgrStatG) of 17 December 2009 (Federal Law Gazette I p. 3886) as amended
• Act on Equal Status for Set-Aside and Agriculturally Used Areas (Gesetz zur Gleichstellung stillgelegter und landwirtschaftlich genutzter areas) of 10 July 1995 (Federal Law Gazette I p. 910) as amended

AGRA2010 =  processing and data editing program for agricultural statistics

AgrStatG  =  Law on Agricultural Statistics

approx.  = approximately

BMEL =  Bundesministerium für Ernährung und Landwirtschaft (Federal Ministry for Food and Agriculture )

BStatG  =  Bundesstatistikgesetz (Federal Statistics Law)

cf.  =  confer (compare)

EAFRD  =  European Agricultural Fund for Rural Development

EC =  European Community

e.g. =  exempli gratia (for example)

ESU = European Size Units

etc.  =  et cetera (and so forth)

EU =  European Union

EUROFARM  =  EC information system for farm structure statistics

Eurostat  = Statistical Office of the European Union

FADN  = Farms Accountancy Data Network

FSS  = Farm Structure Survey

GbR = Gesellschaft bürgerlichen Rechts (civil law company)

GENESIS =  Gemeinsames neues statistisches Informationssystem (Common new statistical information system)

ha  =  hectares

HIT =  Herkunftssicherungs- und Informationssystem für Tiere (central database on animal identification and registration (Bovine Register))

IACS =  Integrated Administration and Control System

IDEV  = Internet-Datenerhebung im Verbund (internet-based data collection of the statistical offices of the Federation and the Länder)

i.e.  =  id est (that is)

NE = not existing

No  = number

NS  =  not significant

NUTS  =  Nomenclature des unités territoriales statistiques (Nomenclature of Territorial Units for Statistics)

OJ  =  Official Journal

p. =  page

SAPM  = Survey of Agricultural Production Methods

StLÄ  =  statistical offices of the Länder

UAA  =  utilised agricultural area

• 5 ha of utilised agricultural area (A_3_1)

• or 10 bovine animals (C_2)

• or 50 pigs (C_4)

• or 10 breeding sows (C_4_2)

• or 20 sheep (C_3_1)

• or 20 goats (C_3_2)

• or 1 000 places for poultry (C_5)

• or 0.5 ha of hops (B_1_6_2)

• or 0.5 ha of tobacco(B_1_6_1)

• or 1.0 ha of permanent outdoor crops (B_4_1 + B_4_2 + B_4_3 + B_4_4 + B_4_5 + B_4_6)

• or 0.5 ha each of area under vines (B_4_4) , tree nurseries (B_4_5) or fruit trees (B_4_1)

• or 0.5 ha of outdoor vegetables or strawberries (B_1_7_1)

• or 0.3 ha of outdoor flowers or ornamental plants (B_1_8_1)

• or 0.1 ha of crops under glass or other accessible protective cover (B_1_7_2 + B_1_8_2 + B_4_7)

• or 0.1 ha of mushrooms (B_6_1).

Stricter thresholds exceeding the requirements of Regulation (EC) No 1166/2008 were set for some special crops (vineyards, nurseries or fruit-growing areas, flowers and ornamental plants outdoors, crops under high accessible protective cover including under glass and production areas for mushrooms). The additional limits for these crops prevent the survey coverage for these crops dropping too low (cf. BR document No 694/08¹). Furthermore, the previous threshold "1,000 heads of poultry" has been replaced by the more stable threshold "1,000 places for poultry".

The threshold for cotton production area in Annex II of Regulation (EC) 1166/2008 was not relevant for Germany, because cotton cannot be grown profitably for climatic reasons and therefore is not applied as a threshold.

The specified coverage thresholds for Germany therefore fulfil the conditions of Annex II and Article 3 of Regulation (EC) 1166/2008. Article 3 indicates that member states shall fix the thresholds at a level that excludes only the smallest agricultural holdings. This issue was already described in detail in the National Methodological Report for FSS 2010 ².

Geographic coverage:

As in the 2013 FSS, the sampling procedure for the 2016 FSS included the entire national area with one exception: alpine pasture cooperatives in Bavaria (common land units, cf. Section 8.1. Comparability - geographical - item 8.1), thus these units or the alpine pastures and meadows managed by them in the Bavarian Alps, were not included in the FSS 2016. 

1 Bundesrat document No 694/08, p. 59

2 http://ec.europa.eu/eurostat/web/agriculture/national-methodology-reports

The most important work steps of the 2016 Farm Structure Survey as well as the bodies responsible for each of the steps are illustrated below in Table 2.

Table 2: Survey organisation and calendar

[1] Ratio of the size of a sample (n) to the population size (N)

• If a NUTS1 region has associated NUTS2 regions with at least 10 000 holdings each, the sample size of this NUTS1 region is allocated to its NUTS2 regions by the method mentioned above.
• If a NUTS1 region consists of at least two NUTS2 regions with fewer than 10 000 holdings each, the NUTS2 regions are treated as one region. The sample size of the NUTS1 region is allocated to this combined region and its NUTS2 regions with at least 10 000 holdings each by the method mentioned above.
• If a NUTS1 region has several NUTS2 regions with at least 10 000 holdings and only one NUTS2 region with less than 10 000 holdings then the last NUTS2 region is treated like the other NUTS2 regions (only one case: DE12). The sample size of the NUTS1 region is allocated to its NUTS2 region by the method mentioned above.

In a second step the sample size of a NUTS2 region is allocated to the region-specific strata by a combination of the optimal allocation (Neyman) with the Standard Output as target variable and an allocation proportional to the square root of the number of agricultural holdings. The resulting sample size of each stratum is the weighted mean of the two methods, where the optimal allocation is weighted three times stronger than the proportional allocation.

Furthermore, data from agricultural liability insurance, from organic farming registers (organic farming yes/no), from animal disease fund as well as the Register of Establishments Keeping Laying Hens were used to update the population in the Farm Register.

An overview of administrative sources is contained in the following Table:

Table: Administrative sources – Name, legal base, time reference and updating

[1] Council Regulation (EC) No 1782/2003 of 29 September 2003 establishing common rules for direct support schemes under the common agricultural policy and establishing certain support schemes for farmers (OJ L 270, 21.10.2003, p. 1–69).

[2] Regulation (EC) No 1760/2000 of the European Parliament and of the Council of 17 July 2000 establishing a system for the identification and registration of bovine animals and regarding the labelling of beef and beef products (OJ L 204, 11.8.2000, p. 1–10).

[3] Council Regulation (EC) No 834/2007 of 28 June 2007 on organic production and labelling of organic products and repealing Regulation (EEC) No 2092/91 (OJ L 189, 20.7.2007, p. 1).

[4] Commission Directive 2002/4/EC of 30 January 2002 on the registration of establishments keeping laying hens, covered by Council Directive 1999/74/EC

Another problem with the use of IACS data in some Länder is that each variable item might not contain the entire area, but only the funded part of the area. 

Rural development measures:
In such cases, the measures paid were cited and not the approved measures.

In one Land the survey was not conducted directly by the statistical offices of the Länder, but in survey offices set up in municipalities. Survey offices are organisational units separated from regular administrative performance that carry out statistical tasks. The organisational separation ensures data protection. The work of a survey office primarily includes the distribution of questionnaires, controlling completeness and the complete counts and, if needed, supplementation of missing data by means of follow-ups.

Prior to the survey, the respondents were informed by the statistical offices via articles in agricultural journals, information events and an internet site about the upcoming survey. In addition, the farmers' associations have participated in the public relations work.

In order to increase response willingness, multiple reminder calls were made in some Länder. In most cases, however, a number of reminder and dunning letters were sent before the last resort was taken of citations, fine or penalty notices. Overall, the number of refusals could thus be minimised to 700 (or approx. 0.9% of all respondents).

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

Number of eligible non-responding holdings

5.1>=5.1.1+5.1.2

Of which 642 refused to respond (holdings that were surveyed but refused to respond) with the adapted weight and 58 new holdings (which were neither reweighted nor were they imputed because there was no information about these units). 

(refusals from the stratum for new holdings do not alter the weight) 

6=5.2+5.1.2+4.2

After the transmission of the edited Länder datasets to the Federal Statistical Office, these individual datasets were combined in one EUROFARM data record (data record for the data transmission to EUROSTAT). The extensive plausibility checks of the EUROFARM data record covered most of the checks prescribed in Annex VI of the Data Supplier Manual.

• Coordinates of the holding location
  The location of the holding is surveyed for the Farm Structure Survey using the official building coordinates in the form of geographical coordinates or Gauss-Krüger coordinates from administrative sources of the Land survey offices.
• Rented areas and rents
  The leased UAA and the relevant annual rents were broken down by the types of use arable land, permanent grassland and other UAA. In addition, first leases in the last two were listed as a separate item on the questionnaire. The rented UAA within an entire rented farm was identified separately.
• Catch crops
  Cultivation of summer catch crop in 2015 and winter catch crops from winter 2015 until May 2016.
• Market gardening module
  Special module for horticultural holdings with questions on high accessible protective covers/green houses (types (such as greenhouse constantly with lower or high temperature), kind of coverage and used energy sources/amounts) and on sources of revenue. Moreover horticultural training of the manager is surveyed.
• Keeping places
  This section encompasses the number of keeping places on the farm used in the last 12 months according to the respective types of poultry.
• Types and amount of manure applications
  This section requested additional variables on the amount and application techniques for farm manure.
• Annual net income in sole holder holdings
  Comparison of the holdings net income with the external net income of the holder/spouse
• Determination of profit and turnover taxation
  As in the previous Census 2010, the type of determination of profit and of turnover taxation was indicated here.

In addition, individual characteristics (coordinates of the holding location, legal personality, land use and crop production, catch crops, market gardening module, livestock and keeping places, determination of profit and turnover taxation and organic farming) were collected in a complete count in order to obtain regional results.

Please access the information in the file at the link: (link available as soon as possible)

In addition to sample-based errors, there are also non-sample-based, systematic errors. Non-sample-based errors can result from shortcomings in the survey technique, in the selection of the population of holdings or in the processing technique. Non-responses can, in turn, lead to systematic errors when the surveyed units provide no or incomplete information.

h = stratum
N_h = number of holdings in stratum h
n_h = number of sample holdings in stratum h
x_hi = variable value i (i=1,2,…..,n_h) in stratum h
= mean value of the variables in stratum h

For reasons of clarity, the standard errors (in %) were not published as error calculation results. Instead, a corresponding alphabetic character is placed after the representative resultant values, which stands for the respective error class of the simple relative standard error.

Allocation of the error classes of the simple relative standard errors:
A:                           to under                                ± 2 percent
B:            ± 2          to under                                ± 5 percent
C:            ± 5          to under                                ± 10 percent
D:            ± 10        to under                                ± 15 percent
E:            ± 15 percent and more

Results that fell under error class E were replaced in the national publication tables by the character “/” because the estimation error is then too great and the numerical value thus not reliable enough. In these cases, the sample size is too small for the assertion made. This can occur among variables that are too infrequent. The error marking is intended to enable users to sufficiently estimate the reliability for their purposes.

Several variable groups, such as, land use, livestock, organic farming, were surveyed in total in the 2016 FSS for national purpose, therefore in the national publication there were no RSEs shown for those variable groups. In national publications RSEs were only calculated for sample surveyed characteristics.

Attached RSEs were calculated on the basis of the sample survey, which are delivered to Eurostat. There are no cases where precision requirements are applicable and estimated RSE's are above the thresholds.

The population for the 2016 Farm Structure Survey was defined with great care. The Farm register, which serves as the basis for determining the population, is regularly managed and updated by the statistical offices of the Länder. Primarily various administrative sources as well as information from past surveys are used to update the register (cf. Section 3.1. Source data - item 4).

Moreover the questionnaire contains a question whether the holding reaches the coverage thresholds. Holdings which do not fulfill the thresholds are marked during data processing and excluded from further data processing. In general false non-responses only have an effect on the sampling error and can deteriorate the relative standard error.

These "false" non-responses did not change the extrapolation factor. How we dealt with non-responses and their extrapolation is explained in more detail in section 3.5. Data compilation - item 2.

Contact data are not always changed entirely in the questionnaire. Obvious incomplete or erroneous information (e.g. post code) or survey documents that cannot be delivered by post are corrected using public registries (telephone books, Internet), in part also using administrative sources, through enquiries with the register of residents, municipalities, trade or regulatory agencies as well as through queries among respondents.

5 022 holdings of the frame did not fulfill the survey thresholds and were excluded from further data processing. The proportion of out-of scope units in the gross sample was 6.0 percent.

All measurement errors were corrected – if recognised as such, for example through distinct deviations from previous year or experienced values – during data editing. Moreover a pretest was conducted with voluntary farmers to improve the questionnaire. In the context of the pretest, the performance and the usability (understanding / user-friendliness) of the online-questionnaire were tested.

While the extrapolation factor was adapted when possible for sample holdings that were “true” non-responses (in this case refusals), the “false”non-responses did not change the extrapolation factor. How we dealt with non-responses and their extrapolation is explained in more detail in section 3.5. Data compilation - item 2.

A non-response analysis was not conducted.

A total of 700 of 78 064 agricultural holdings refused to respond to the 2016 FSS, equaling a non-response rate of approx. 0.9%.[1] 

[1] This includes only “true” non-responses (cf. Section 6.3.3. Non response error - item 1.). 

Not available. 

The AGRA2010 processing program was the chief instrument for completeness and plausibility checks. This program stores 330 obligatory error tests, 207 facultative error tests and 15 automated corrections. In the following, we explain the differentiation of these error messages.

• Obligatory errors must be adjusted in all cases (e.g. missing age for an individual) and are obvious, unacceptable erroneous information or inconsistencies in correlations of data.
• Facultative errors occur when information or correlations of information are possible, but either are improbable or rare, taking into consideration the operating and economic circumstances in agriculture, or originate from chronologically different individual surveys and therefore need not necessarily match (e.g. maximum controls). In such cases, we checked whether and, if so, in what way correction of the relevant information is necessary through individual and targeted follow-ups with the holding or, from case to case, drawing on other information.
• Automatically adjusted errors are errors that can be corrected without a doubt and unequivocally based on the available information without follow-up interviews or data matching (e.g. by inserting missing total values).

The extensive plausibility checks cover the majority of the rules described in Annex 7 of the data supplier manual[1]. Due to a different approach in data editing, however, we cannot guarantee one-to-one implementation of the data editing rules. The transferred Eurofarm data material must consequently be verified for plausibility once again by Eurostat according to the rules in Annex VI of the data supplier manual. If the data editing program showed errors following the checks, these erroneous data had to be processed or corrected. Following these checks and the associated corrections no more missing or erroneous data should exist in the data material of the Farm Structure Survey.

[1] Eurofarm manual for data suppliers – Farm Structure Survey 2016, rev. 7, 20 December 2016

The imputation rates of item non-response are not available.

Errors in publications can also be the reason for data revisions. The treatment of errors in publications is prescribed at the Federal Statistical Office in a special guideline (Richtlinie zum Umgang mit Veröffentlichungsfehlern). Should publication errors occur they are allocated to error categories – depending on the severity of the error – and treated depending on the error category. Corrected data are then identified in the national publications by a special signature. The statistical offices of the Länder have comparable guidelines for handling publication errors or use a comparable procedure for revisions.

An additional data revision has been necessary due to errors in a national online publication on manure management (Series 2.2.2: Farm manure). The incorrect data were as quickly as possible corrected. The corrected data were flagged with a special signature. Additionally, errors, their reasons and treatments were described in the updated version of the publication.

The period between the beginning of the survey and publication of the first results from the entire feature set of the 2016 FSS (key figures in the January 2017 press release) was approx. 11 months.

However, according to the Law on Agricultural Statistics a holding must exhibit specific minimum requirements in the size of utilised agricultural area (UAA) or for livestock or the crop areas for specialised crops, i.e. it must reach specific coverage thresholds. In Germany, the threshold for poultry is 1.000 places (see 2.6-2).As a result, holdings that currently have no poultry (and no other animals or areas) but more than 1.000 places for poultry are included in the survey. 

The catalogue of variables for the 2013 FSS was created largely according to the definitions in the corresponding Eurostat handbook (Handbook on implementing the FSS and SAPM definitions – 12 October 2015).

Differences between national and EU definitions:

• Farm work: Under the national definition, work in other gainful activities is part of the time spent for farm work on the holding if the work is directly connected to the holding. This definition was used before 2010 (introduction of a separate question on hours worked in
  other gainful activities) and a change would cause a methodological break in the labour force time series.
• Other gainful activities: The variable “Processing of farm products” was worded in the questionnaire as “Processing and direct sales of farm products” because in Germany the processing of products is usually combined with subsequent direct sales of the products.
• Support schemes: The information on the receipt of rural development payments and on the ecological focus area were taken from administrative sources. The responsible administrative offices were therefore asked to transmit the authorised support schemes. However, the approved measures could only be identified for co-funded payments (i.e. when EU funds were used). Non-co-funded payments (i.e. if only federal or state funds were used) were usually recorded as sums by the agricultural administrations. In such cases, the measures paid were cited and not the approved measures. This can result in under-coverage.

  In some Länder there are combined rural development support programs, consisting, for example, of payments for agricultural areas in the scope of NATURA 2000 and payments related to the Water Framework Directive. In these cases, participation in both schemes was cited for the respective holding. Any resulting over-coverage is considered less problematic than under-coverage of measures.

• Ecological focus areas: Due to the Länder specific nature of the IACS support, the ecological focus areas also include areas outside the arable land in some Länder. This variable therefore can include areas of permanent crops and permanent grassland.

In 2015 only Bavaria reported the existence of common land units, namely 218 common land units with approximately 18,000 hectares of UAA. These are 0.23 percent of the holdings and 0.57 percent of the UAA of Bavaria. Most of these units are in the NUTS region DE27; but even here these units account only for 1.03 percent of the holdings in this NUTS region (157 holdings with 13,219 ha) and have only 2.61 percent of the whole UAA in this NUTS region. In the other regions of Bavaria, the proportion of holdings and UAA tends to zero or is even zero. The common land units from Bavaria are not included in the dataset.

In the other Länder, the share of the common land is even lower and if at all included in the data of the agricultural holdings under 'tenant farming' or 'share farming or other modes' and cannot be identified.

For the FSS 2016 survey, common land is a new characteristic and so we set it up as 'non-significant, and not collected and delivered to Eurostat under its own heading' but possibly under other codes.

[1] Council Regulation (EC) No 834/2007 of 28 June 2007 on organic production and labelling of organic products

The numbers of employees in agriculture in the Farm Structure Survey and the national employment accounts are not comparable as they are based on different methodologies. While the national employment accounts record all gainfully employed persons only once with their main employment, FSS covers all employees, regardless whether it is their main or secondary occupation. If the main and secondary employments are not in the same economic classification, deviations occur in the economic sector-related portrayal of results. This affects the many family workers having side-line jobs on farms in particular.

In addition, the results of the Farm Structure Survey differ from the information in the FADN in their allocation of agricultural holdings to socio-economic farm types: full and part-time farms. While in the FSS this classification is made exclusively for agricultural sole holder holdings, but not for group holdings and legal persons, the FADN includes group holdings, but do not typify legal persons either. Furthermore, the FSS and the FADN use different bases for classifying full and part-time farms. In the FSS questionnaire, all owners of agricultural sole holder holdings provide information on whether the income from the agricultural holding or from non-farm sources was higher. If the farm income was higher, the sole holder holding is considered a full-time farm; if the income from non-farm sources was higher, the holding is classified as a part-time farm. By contrast, for the FADN, enterprises with the legal forms of sole holder holdings and group holdings with 16 and more European Size Units (ESU) and at least one full-time worker are classified as full-time farms. All enterprises with the legal forms of sole proprietorships and partnerships with 8 to less than 16 ESU or less than one full-time worker are considered small and secondary occupation holdings. The test farm network (FADN) comprises only holdings from 8 ESU, whereby 1 ESU corresponds to a total standard coverage of 1,200 euros.

Furthermore, the Farm Structure Survey provides information that is partly comparable with variables from self-contained individual surveys (e.g. the survey on livestock for bovines, pigs and sheep). There are crucial differences in the respective survey methods with regard to the population, the thresholds and the reference dates. In the FSS, bovine livestock is reported according to the associated holdings while in the survey of bovine livestock the observed units are different and cattle is reported in total regardless to any thresholds. This can lead to differences with regard to the number of holdings or farms as well as the livestock between the surveys. In general, the Farm Structure Survey provides a structural overview of the agricultural variables while individual specialised statistics provide specific and more detailed characteristic information. More information is contained in the relevant quality reports at http://www.destatis.de/.

• Series 2.1.1: Holdings with forest areas
• Series 2.1.2: Land use of the holdings (Structure)
• Series 2.1.3: Livestock of the holdings
• Series 2.1.4: Business focuses and standard output
• Series 2.1.5: Legal personality and socio-economic holding types
• Series 2.1.6: Ownership and tenancy
• Series 2.1.7: Agricultural holdings with other gainful activities
• Series 2.1.8: Labour force
• Series 2.2.1: Holdings with organic farming
• Series 2.2.2: Farm manure
• Series 2.2.3: Holdings with viticulture
• Series 2.2.4: Horticulture in agricultural holdings
• Series 3.1.2: Land use of the holdings (UAAs)

The Fachserie contains tables as well as an explanatory section.

Not available. 

The transmission of information collected to the responsible supreme federal or Länder authorities in the form of tables with statistical results is allowed according to Article 98 (1) of the Law on Agricultural Statistics in conjunction with Article 16 (4) of the Federal Statistics Law. This applies even if individual table cells only identify a single case. Under Article 16 (6) of the Federal Statistics Law, it is possible to provide individual data (microdata) to universities or other institutions tasked with independent scientific research for scientific projects if individual data can only be assigned to the respondents or parties concerned with a disproportionately large investment of time, cost and labour. Research institutions may, in addition, apply to the research data centres of the federal and Länder statistical offices where they have various ways of obtaining access to data, such as the on-site use (via safe centres or remote execution) or off-site use (scientific use files, public use files).

The research data centres are obliged to ensure statistical confidentiality when offering access to microdata for scientific use. They may not release results that allow conclusions on individual cases.

For this reason, the research data centres check every result that is produced within an on-site use of microdata for statistical confidentiality. This happens according to set rules. If  results allow conclusions on individual cases, these results will to be blocked. [1].

[1] Cf. http://www.forschungsdatenzentrum.de/en

A quality report with explanations about the metadata and quality characteristics will be publicly available at www.destatis.de > Startseite > Publikationen > Qualitätsberichte > Land- & Forstwirtschaft, Fischerei > Strukturerhebungen. 

Reporting burden for the holdings was also reduced by using as many administrative data sources as possible. However, this procedure not always reduced the cost of data handling in the statistical offices too.

Primary confidentiality is applied for total (T) and representative (R) tables. Results of the representative survey were rounded published (thousands with one or two position after decimal point). The “natural confidentiality effect” of rounding was taken into account by means of a relevant adjustment of the p-percent rule. Results of total survey were published in complete length.

To prevent the disclosure of the primarily suppressed table elements by forming sums or differences in the tables, so-called secondary suppressions were carried out (secondary confidentiality) in addition to the primary suppressions. The secondary confidentiality was carried out manually by the Federal and Land statistical offices. The suppressed table elements based on primary or secondary confidentiality are marked with a dot in the publication tables.

The FSS results were tabulated and made confidential using the AMT and GHMAN Java applications developed especially for this purpose. While AMT supports the entire process of results tabulation including primary confidentiality, GHMAN enables manual entry of secondary confidentiality in the table cells.

1 Cf. Gießing, Sarah (1999): “Methoden zur Sicherung der statistischen Geheimhaltung”; Volume 31 of the series Forum der Bundesstatistik, published by the Federal Statistical Office, pp. 6-26.

Acceptance of the questionnaire and the questions it contains could be improved by shortening the survey length and making the questions more understandable. The relevance of the questions needs to be directly recognisable to the respondents. 

Not available.