Danmark

Since then the survey has taken place every year as a questionnaire-based survey where the farmer has received a questionnaire in a letter with an obligation to complete it. Nowadays, however, the contact with Statistics Denmark is digital.

Total censuses held in: 1977-83, 1985, 1987, 1989, 1999 and 2010.

Sample surveys held in: 1984, 1986, 1988, 1990-98, 2000-09 and 2011-17.

Generally the samples have been quite big, 20-50 percent of all farms.

The surveys have always had a threshold so that small holdings are excluded:

1977-82: FSS included farms with at least 0.5 hectares or at least a production with a value corresponding to 0.5 hectares with barley.

1983-1994: FSS included farms with at least 5.0 hectares or at least standard gross margin of 3 000 euros at 1985 prices.

1995-2009: FSS included farms with at least 5.0 hectares or at least a standard gross margin of 4 000 euros at 1990 or 1995 prices.

From 2010 Eurostat introduced a harmonization of the thresholds in FSS.

The agricultural activities include:

• All agricultural crops, also when only land maintained in good agricultural and environmental conditions.
• Equidae (horses)
• Bovine animals (cattle)
• Sheep and goats
• Pigs
• Poultry
• Rabbits, breeding females
• Bees
• Livestock not mentioned elsewhere (fur animals, excluding rabbits) 

The closest we could come to “the total number of farms in Denmark” would the number of applicants crop subsidies, which was 40.087 in 2016 according to IACS.      

1. An agricultural area of at least 5 hectares (A_3_1$ha)
2. A standard output of at least 7.500 euros
3. Fruits, berries and nursery area of at least 0,5 hectares (B_4_1$ha+B_4_5$ha)
4. Vegetables and strawberries of at least 0,5 hectares (B_1_7$ha)
5. Greenhouse and mushrooms of at least 1 000 m²  (0.1ha) (B_1_7_2$ha +B_1_8_2$ha +B_4_7$ha)
6. At least 10 cattle (C_2$heads)
7. At least 50 pigs (C_4$heads)
8. At least 10 sows (C_4_2$heads)
9. At least 20 sheep (C_3_1$heads)
10. At least 20 goats (C_3_2$heads)
11. At least 1 000 poultries (C_5$heads)
12. At least 40 fur animals

The source is most often the most recent FSS for each farm (FSS 2010-15) but could also be information from IACS or the livestock register.

According to the final national Danish publication, Denmark has 35.669 farms where 622 of them have fur animals as their only agricultural activity. They will be removed in the Eurostat publication, which thereby have 35.047 Danish farms. 

There are 20 farms in the sample sent to Eurostat with no livestock and where the agricultural land consists of only fallow land and permanent grassland no longer used for production. This is not a very common case but perfectly possible. There are additionally 3 farms fulfilling the same conditions except that they do have fur animals other than rabbits.
These farms have applied for single payment in the 2016 season. In order to obtain this support they have an obligation to keep the land in good agricultural and environmental condition.

19 farms sent to Eurostat have no agricultural land and no livestock at all. It would normally mean closed down farms which should not be included in the sample sent to Eurostat. However, for these farms something special applies: They are farms with poultries but they happen to have an empty animal house at the reference day.
So instead of reporting the number of chickens (or whatever) on the reference day they have instead reported a production of poultries for slaughtering in the recent year thereby indicating that they are still active farms.
We have chosen to include such farms in our own national publication and also send them to Eurostat. It should be noticed that these farms have complete and valid labour force information. 

Some of these “small farms” could, however, be big by virtue of their livestock (e.g. at least 20 sheep) or green house area. But this information on livestock and green house area is not included in IACS.

It is not possible to make a similar study for livestock using the livestock register for two reasons:

1) The exercise mentioned above for IACS defines as small farms the farms with both an area smaller than 5.0 hectares and an area with fruits, berries and vegetables smaller than 0.5 hectares. Here we suppose that these farms do not have livestock to any considerable content. This is normally true since many animals normally mean a big agricultural area. However, this rule does not work the other way round. It is quite normally to have some animals - but very few – in combination with a big area. So a farm with, say 5 sheep, cannot be classified as a small farm since it is extremely likely that it has more than 5.0 hectares of agricultural land.

It would in principle be possible to merge IACS and the livestock register whereby small farms could be identified as all farms below both the area and livestock thresholds stipulated in the regulation. Such an exercise is not as easy as it might sound and would involve certain match problems. Statistics Denmark has not yet found resources for such a study.  

2) Even though the livestock register has information on number of animals this information is much more uncertain than the similar area information in IACS. For all other livestock than cattle the number of animals is a mix of stock information and production figures and the figures are also often up to one year old.  

But it could be mentioned that Statistics Denmark in 2006 carried through a project with grants called “Danish TAPAS project on small units”. The study showed that Danish farms below the threshold have about 0.6 percent of the total standard gross margin of the whole Danish agriculture. The work can be found on CircaBC, see page 10. It makes it extremely unlikely that farms below the Danish threshold should possess more than 2.0 percent of the livestock units.

However, the preferred solution is to derive the municipality code from the property code belonging to the farm. It is because the business register information code could refer to the place where the farmer lives rather than where the farm activity takes place.

• Crops are collected from IACS where farmers should apply for crop subsidies no later than April 21 2016. The information on crops concerns the season 2015/2016 from the point of time where winter crops are sown – normally September or October 2015 - till the crops are harvested – normally August 2016. Evidently the time for sowing and the time for harvesting might differ depending on weather conditions. For green house crops and farms not applying for subsidies the survey day applies.
• Cattle are collected from the livestock register dated June 1 2016, pretty close to May 13.
• Labour force characteristics, including other gainful activities concern a period of one year prior to the survey date rather than the survey day itself.
• Rural development characteristics concern the calendar year 2014-2016. According to art 8.(d) of Regulation 1166/2008 this reference period should be 3 years.
• Education in the recent year concerns necessarily a period of one year prior to the survey day.
• Irrigated area in the recent year concerns necessarily a period of one year prior to the survey day.

• The questionnaire: The work on designing the Danish questionnaire began in September 2015. The online version of the questionnaire was prepared by our division of data collection - based on the paper version of the questionnaire - and was ready in the beginning of May 2016.
• Sample selection: The sample was selected by our methodological unit in April 2016.
• Sending the questionnaires: May 6 2016.
• Survey day: May 13 2016.
• Reminders:

  First reminder: Beginning of June 2016, about 7.500 farms
  Second reminder: End of June 2016, about 4.500 farms
  Third reminder: Middle of July 2016, about 3.000 farms
  Fourth reminder: From beginning of August 2016, about 1.000 farms
  About 50 farmers were reported to the police regarding FSS 2016. However, a non-respondent is not reported systematically to the police, but only after the second violation within a period of one year.  So still a certain non-response had to be tolerated.

• Validation of questionnaires: This work took place from May 2016.
• Integration with administrative data: January 2017.
• Calculation of extrapolation factors: May 2017.
• Publication: May 22 2017, results were published in a small online newsletter with detailed results.
• Data to Eurostat: July 2017, with certain later corrections.

Information on crops, cattle, organic farming characteristics and rural development characteristics are collected from administrative register but connected to farms in the sample. It means that in the end the FSS survey register will contain complete information, also on crops, cattle etc. exactly as if all questions had been on the questionnaire in the traditional way.  

Since the total FSS census in 2010 - where the number of farms was 42.099 – 10.473 have stopped their farm activities or have fallen below the threshold. 5.535 farms have since then been added to the population.

The Danish population must be defined as a list frame consisting of all farms on the Danish territory above the threshold. These farms are the survey units. 

The Danish population of farms has since 2010 been a part of the general business register. Before 2010 Statistics Denmark had separate registers for agricultural and forestry units.

Any business unit could, irrespective of the NACE code, be active in agriculture or forestry or both. For this reason a variable in the Danish business register “Active in agriculture” has been created. It can assume four different values:

0= Not active in agriculture or forestry
1=Active in agriculture, but not forestry
2=Active in forestry, but not in agriculture
3= Active in both agriculture and forestry

All units with value 1 or 3 are furthermore marked as small or big where big means above the thresholds mentioned in concept 2.6. Statistical population - item 2., big enough to be included in an agricultural survey.

The register is continuously updated with information from IACS and the livestock register where these routines are established:

- Farms in IACS and the livestock register should be updated with the active value 1 or 3 in the business register where this is not already the case.

- Big farms in IACS – at least 5.0 hectares of agricultural land or at least 0.5 hectares with fruit, berries or vegetables – should be updated as big farms in the business register if they have status as small farms.

Routines which could delete units from the list of active farms are not developed successfully yet. As it is now farms are mainly deleted in connection with agricultural surveys if a farmer when receiving the questionnaire tells that he has stopped all farm activities.  

Small farms below the threshold are still in the business register but marked as small so that they cannot be selected to a survey until information from either IACS or the livestock register indicates that they should be updated as big farms.

1) Region

Denmark has five different NUTS 2 regions and these regions are a part of the FSS sample stratification:

2) Typology

The standard output and the typology of the farms are calculated based on information from the most recent FSS, 2010-15 . For farms in the population with no survey information two rules are applied to describe the farms:

A) If the farm is in IACS and not in the livestock register the standard output and the typology are calculated based on crop information from IACS. The farm is assumed not to have any livestock.

B) If the farm is:

1) not in IACS but in the livestock register or

2) in both registers or

3) in none of the registers (could for instance be newly established horticultural farms)  it belongs to typology group 9000, which of course is not any real typology but merely means “typology and standard output unknown”.
All farms from this group are selected to the sample (unless they have as status as have a status in the business register as bankruptecy or they do not have a valid business number) and no size groups are created.

3) Size of standard output. Different size groups are applied for different typology groups.

• Agricultural area
• Number of cattle
• Number of pigs
• Number of fur animals
• Standard output

All IACS, the livestock register, the organic farm register and the rural development register live up to the statistical requirements so it has not been necessary for Statistics Denmark to influence the content. If any such need arises it is difficult to ascertain if it would be possible to change the content of an administrative register in order to meet statistical needs. That would depend on circumstances. An administrative body has no obligation to adopt its registers to meet statistical needs. We have, however, one positive experience going back to 1995 where the Ministry of agriculture agreed to divide beets into sugar beets and fodder beets.

IACS reflects the crop year 2016, which means winter crops sown in the autumn 2015 and spring crops sown in the spring 2016. As such IACS is assumed to meet the FSS standard and thus no adjustment procedure is necessary. 

The farmers report their land use in April for the given crop year. If changes are made the farmers are obliged to report the changes. The crops might not have a survey day as such but does rather concern the season. In 2016 the farmers had a deadline of April 21 2016. It means that IACS 2016 had information on winter crops sown in the autumn 2015 and spring crops recently sown.

• Bull-calves and steer-calves,- under 1/2 year, ca_1
• Bull-calves and steer-calves, 1/2-1 year, ca_2
• Bulls and bullocks, 1-2 years, ca_3
• Bulls and bullocks, 2 years and over, ca_4
• Heifer-calves, under 1/2 year, ca_5
• Heifer-calves, 1/2-1 year, ca_6
• Heifers, 1-2 years, in calf, ca_7
• Heifers, 1-2 years, not in calf, ca_8
• Heifers, 2 years and over, in calf, ca_9
• Heifers, 2 years and over, not in calf, ca_10
• Dairy cows, ca_11
• Cows kept for suckling, ca_12

The FSS categories required by Eurostat can easily be created from these 12 categories:

The livestock register is updated weekly or continuously by the cattle farmer regarding acquisition (either new born calves or purchased animals) and disposal of animals (either died of natural causes or sold). This update is also with information on suppliers and buyers of cattle. Furthermore the deliverance of cattle for slaughtering is verified by the slaughter houses and the animals died of natural causes by the carcase disposal plants.

Due to this tight system of control and validation unreported events must be assumed to be negligible, maybe even non-existent.

Approximately 5 percent of the Danish farms in the FSS do not apply for single payment. Generally these farms are:

• Horticultural farms with green house area but no crops on free land.
• Livestock farms with no agricultural land.
• Small farms which have not find it worthwhile to apply for a small area. But these small farms are normally not included in the Danish FSS.

IACS and the organic farm are delivered as SAS-data sets and  the rural development register and the livestock register as csv-files.

The choice of file type and data format is up to the data provider.

False matches are eliminated by using the following procedures, and they apply to both IACS and the livestock register:
a) Two or more farms have indicated on the FSS questionnaire the same register identification number.
b) A farm has indicated a non-existing identification number, probably a simple writing mistake.
c) A farmer has not indicated any number.
d) A farm has indicated an existing identification number, but neither the business number nor the personal civil registration code are the same in respectively the statistical register and IACS/livestock register. The farmer has most likely made a simple writing mistake and has accidentally chosen an existing number of a farm not included in the survey.

All these mistakes must be eliminated before the match can take place.

However, farmers who have difficulties with the web questionnaire may request a paper questionnaire or simply give the information on phone. 

• Internet: 12 446
• By post or telephone interviews: 1 524

These 13 970 answers include 13 087  farms sent to Eurostat, as well as farms below the threshold and closed down farms. Of the 13 087 farms sent to Eurostat, 398 are farms with only fur animals other than rabbits as the only agricultural activity.  

Questionnaires not completed online are all registered manually. Since the FSS in 2014 we do not use scanning as a tool for registering questionnaires.

Three reminders are sent to the farmers before they are contacted by phone, and only the biggest of the farms are contacted in this way.

It means that it is impossible to completely avoid non-response.

Farmers can find information on our web site regarding the survey. They can also find a phone number and an e-mail address if they wish to get into contact with Statistics Denmark.

Farmers having troubles with the online solution may request a paper questionnaire.

Farmers who wish to complete the questionnaire by phone are never denied this option but they must contact Statistics Denmark themselves. The reason is that the online solution is seen as the preferred solution. Other solutions should be reduced as much as possible.

In case of a census 1=3+4+5

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

(including holdings with fur animals other than rabbits) 

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

12.689

(excluding holdings with only fur animals other than rabbits)

• Warning checks to detect possible mistakes, for instance more than 30 horses. It is not necessarily a mistake but most likely it is.
• Logical mistakes. It could be no working time indicated by the farmer or that the irrigated area is bigger than the irrigable area.

The questionnaires are registered and also edited in an Oracle database. There are about 150 validation rules where some are meant as warning rules and some are meant as “serious” mistakes meaning the concerned problems have to be solved before the questionnaire can be accepted. In addition to these rules some ad-hoc controls might be carried through.

The tools used in the validation process are Oracle, SAS and Excel.

All the validation work is made at Statistics Denmark. The final survey register is stored at the PC network of Statistics Denmark. It is also the case for the special version of the survey register which Eurostat receives.

1) IACS: The target values concern farms applying for subsidies only.

Target values for extrapolation of the Danish FSS, farms based on IACS

* But with at least 0.5 hectares with fruits, berries and vegetables.

2) Pig surveys:

The quarterly pig surveys are specially designed to cover farms with pigs and it means that the sample error for pigs is lower than in FSS. The extrapolation has thus had number of pigs in the pig surveys in April and July as a weighted average of these two surveys as the target value:

Target value for extrapolation of the Danish FSS, based on the pig surveys

The target value for sows exclude young sows which the farmer have planned to use for breeding but which have not yet been inseminated. 

The April survey has been given a higher weight since it is a bit closer to FSS in May and also because the farmers in FSS who took part in the pig survey in April did not have to report pigs also to FSS.  

3) Fur animals from The Danish Fur animal Farmers Association:

All farmers with fur animals are assumed to be members of the association, which collects number of minks and other fur animals from the members. The association sends every year a register with individual farm information to Statistics Denmark.

Target values for extrapolation of the Danish FSS, based on information from the fur Farmers Association 

It should be noticed that the calibration method used may lead to extrapolation factors smaller than one.

• Crops: The Danish survey collects information on winter crops and spring crops for wheat, barley and rape.
• Pigs: There are 10 categories of pigs where the regulation demands 3 categories only.
• Cattle: There are 12 categories of cattle where the regulation only requires 7.
• Poultry: There 8 categories on the Danish questionnaire where the regulation requires only 3.
• Fur animals: The Danish questionnaire has from 2010 onwards had a question on fur animals where the regulation only requires a yes/no question on “other animals".
• Other animals: For sheep, goats, horses, rabbits and bee hives the Danish questionnaire is identical with the regulation requirements.
• The survey in 2016 had questions on production and areas with vegetables and strawberries.

The users are in particular EU, the ministries, farmer’s organisations, but also students and interested people in general. EU uses the statistics as a tool in the planning of the common agricultural policy.

The farm structure survey is discussed at meetings in user board on agricultural statistics. The members may put forward suggestions to new questions on the questionnaire as well new statistical tables for publication. This led for instance to that we in FSS 2017 had some simple questions on farms with use of precision technology.

Together with normal routines of data validation the quality of the FSS statistics could be assumed to be rather good.

352 farms were small meaning that their crops and livestock are lower than the thresholds mentioned in concept 2.6 Statistical population - item 2. They have completed the questionnaire on equal foot with all other farms but they are not a part of publication nor are they sent to Eurostat. Likewise no extrapolation is calculated for these survey units. Of these small farms:

• 88 have cattle
• 12 have pigs
• 51 have poultries
• 43 have sheep
• 16 have goats
• 76 have horses
• 147 have no livestock but only a small agricultural area

But when completing the farm structure survey the figure for the number of farms was 35.669 (including the holdings with only fur animals other than rabbits).

Thus the over coverage could be estimated to 37.161-35.669=1.492 farms or 4.0 percent.

The most difficult questions might concern the labour force section of the questionnaire, more specific work time of the farmer and his wife. It is of course not so difficult for those working full time on big farms. Neither is it so difficult for owners of very small farms working only a few hours per week to indicate "less than 1/4 of full time". The problem concerns farmers working less than full time but on the other hand more than just a few hours per week and maybe with varying work time over the season. For them it can be difficult to choose the correct of the three "in between working times".

In 2011 all in all 7 552 or a bit more than 50 percent did not respond to at least one SAPM question but completed the questionnaire otherwise. The missing information was imputed by means of imputation. 

No imputation work was done like this in the Danish FSS 2016. It might happen that a farmer does not complete the questionnaire in all details. Most often this is rather harmless:

• A farmer does not answer the section on irrigation but he lives in a region where almost no farmer irrigates so obviously the answer is simply “no irrigation”. 
• A farmer does not answer the section on greenhouses. But since he does not use to have greenhouse crops and since most farmers do not have greenhouse crops the answer is obviously no. 
• Also incomplete labour force information is often rather easy to correct by means of this manual imputation. For instance it is quite obvious that the holder of a big farm works full time and cannot have a work outside the farm. Likewise a young holder of a small farm most likely has the lowest work time and a work outside the farm.  

Some cases are more serious:

• A farmer known to have pigs does not indicate number of pigs. 
• A farmer known to have poultries forgets to answer the section on poultries. 
• A farmer known to have greenhouses forgets to indicate his greenhouse crops. 

When it comes to pigs we can often take the figures from the closest pig survey. But otherwise it is necessary to call the farmer.

It is not possible to identify any specific reason for item non-response other than normal carelessness. 

A re-weighting method is used in the Danish FSS. See item 3.5-2 for more information.

When it comes to pigs we can often take the figures from the closest pig survey. But otherwise it is necessary to call the farmer.

However, some qualitative assessments could be made:

1) About 90 pct. of all farmers complete the questionnaire online and here chances of a mistake caused by free completion are good since the farmers have to answer all questions. For instance he has to answer yes/no to whether he has sheep or not, and if he answers yes he has to indicate at least one sheep. The same rule applies to the other categories of livestock. Likewise he has to report his work time on the farm and also his wife’s work time (or answer that he has no wife or that his wife does not take part in the farm work). It means all in all that the completion of the questionnaire at least is consistent from a logical point of view but of course there is no guarantee that farmers could not by mistake write for instance 54 sheep instead of 45 sheep.               

The remaining 10 of the farms complete the questionnaire in other ways:

2) Some might call by phone asking for help to complete the questionnaire and we then almost always fill in the questionnaire immediately by interviewing the farmer. Also here, chances of a mistake caused by free completion are good since the interviewer knows the questionnaire and can secure that no questions are forgotten.

3) The remaining farms have asked for a paper version of the questionnaire to fill in the traditional way. Here risks of forgetting answering some questions evidently exist.

We have no certain information on the distribution of farms between group 2 or 3). A qualified guess is fifty fifty.  

Since the questionnaire did not contain many new items compared to the questionnaire in previous years we did not face any major problems, like for instance a huge number of farmers who misunderstood one or more questions. So only well-known measures were taken:

- Recontact to approximately 3 percent of the farmers by means of telephone calls.

- Manual imputation of about 10 percent of the questionnaires. (Qualified guess)

- The non-response was about 7 percent compared to 5 percent in 2013. 

Manual imputation is made for approximately 10% of the farms. 

Statistics Denmark has no special revision policy and practice differs from statistics to statistics.  

The Danish questionnaire on work time has these five categories:

So hours per year play no role when calculating AWU but it could be assumed that a person working exactly on full time has 7.4*225=1 665 hours per year.

It means that Denmarks finds itself in situation  “There have been some changes but not enough to warrant the designation of a break in series”.

Otherwise the users can find figures from 1982-2016 at a comparable level.

In the Eurostat publication farms with fur animals other than rabbits as the only agricultural activity are not included.

If categories in fact are not confused the FSS compared with IACS/the livestock register must be expected to be very close to each other but not necessarily identical for these reasons:

1) FSS is a sample survey and any result is subject to sample errors, also characteristics collected from an administrative register.

2) The thresholds used in FSS do not apply to IACS. It works in the direction of making IACS/livestock register results a little bit bigger than FSS results since IACS/livestock register include some small farms not covered by FSS. In 2016 IACS had 40.087 applicants for single payments. FSS had 35.669 farms published nationally and including farms with only fur animals other then rabbits. 

3) It is perfectly possible for a farmer not to apply for subsidies but still be covered by FSS. However, for farms growing cereals and other normal agricultural crops it is extremely rare not to apply for subsidies.       

The closest we come at a check is when comparing FSS with a recent pig survey:

Both surveys are sample surveys and are thus subject to normal sample errors. The difference is biggest for piglets but still no more than 2.1 percent.  

Also the Danish FSS is comparable with the FADN statistics as far as the farm concept is the same. BUT FADN has a higher threshold, namely 10.0 hectares of agricultural land or a standard output of 15 000 euros and excludes about 6.500 farms from FSS.

The principle is that as soon as a survey is ready for publication Statistics Denmark will publish a small newsletter of two pages with a few main results, and with focus on a particularly interesting development. The Danish FSS 2016 was published May 22 2017. At the same day detailed figures were also published on the Internet with for instance regional figures. The newsletter focused on farmer by type of ownership.

Figures from FSS 2016 were also published in Statistical Yearbook 2017 (published April 2017; the English version is only published online: https://www.dst.dk/en/Statistik/Publikationer/VisPub?cid=22257) and Statistical Ten years Review 2017 (published August 2017).

Statistical Yearbook 2017 (published April 2017)

Statistical Ten years Review 2017 (published August 2017)

Statistical Yearbook 2017: https://www.dst.dk/en/Statistik/Publikationer/VisPub?cid=22257

Here are a few examples of the statistical tables on FSS:

- [Farms](http://www.statistikbanken.dk/bdf11)
- [Labour](http://www.statistikbanken.dk/bdf307)
- [Crops](http://www.statistikbanken.dk/afg07)
- [Livestock](http://www.statistikbanken.dk/hdyr07)

[Agricultural and horticultural survey](http://www.dst.dk/en/Statistik/dokumentation/documentationofstatistics/agricultural-and-horticultural-survey) 

In practice it means that only a few colleagues at Statistics Denmark have the right to access the FSS survey registers.

Such a practice is in line with principle 5 in European Statistics Code of Practice on statistical confidentiality.

When designing statistical tables the aim is to secure that no table cells contain very few farms. There are no exact rules regarding "very few farms".