The first general agricultural census 2002 (GAC 2002) was carried out over the period 2 December 2002 – 31 January 2003.

The GAC 2002 data were processed at national level, development region level, county level and locality level and they were transmitted to Eurostat in the required format for the Eurofarm database containing 4 484 893 agricultural holdings.

The farm structure surveys 2005 and 2007 (FSS 2005 and FSS 2007) were conducted in accordance with the EU requirements based on the Council Regulation (EEC) No 571/ 1988 on the organisation of Community surveys on the structure of agricultural holdings with its subsequent amendments and the Commission Decision No 115/2000 regarding the surveys on the structure of agricultural holdings in 2005 and 2007 with its subsequent amendments.

The farm structure survey 2005 (FSS 2005) was a sample survey in accordance with the EU requirements and the national ones on the basis of a sample representative at national/development region/county level (NUTS 3) of approximately 8 % from the population registered in the Farm Register (FR) according to the results of GAC 2002. Out the total of 4 484 893 agricultural holdings a sample of 361 169 holdings was drawn with a margin of error of less than 5 %.

FSS 2007 was a sample survey based on a sample representative at national/development region/county level (NUTS 3) of approximately 8 % from the FR-registered population, updated with the information of FSS 2005. Thus, out of the total 4 480 664 agricultural holdings a sample of 354 742 agricultural holdings was allocated, with a margin of error under 5 %. The survey sample consisted of 336 299 agricultural holdings without legal personality and 18 443 agricultural holdings with legal personality, the latter being exhaustively surveyed.

The data collection for the general agricultural census 2010 (GAC 2010) and the survey on agricultural production methods 2010 (SAPM 2010) took place during the period 2 December 2010 – 31 January 2011. The preparations started in 2008 by setting up the legal frame and they ended in 2012 with the data being transmitted to Eurostat.

A single questionnaire was used for the data collection for GAC 2010 and SAPM 2010 including all the characteristics applicable to Romania mentioned in Regulation (EC) No 1166/2008 of the European Parliament and of the Council on farm structure surveys and the survey on agricultural production methods.

The farm structure survey 2013 (FSS 2013) was a sample survey, based on a sample representative at national/macro-region/development region/county level (NUTS3) of approximately 8,9 % from the FR-registered population, updated based on the information of GAC 2010. Thus, out of the total 3 859 043 holdings, a sample of 345 421 holdings was drawn, with a margin of error of less than 5 %. The survey sample consisted of 313 315 agricultural holdings without legal personality and 32 106 agricultural holdings with legal personality, the latter being exhaustively surveyed.

- Government Decision No. 957/2005 on the organisation and operation of the National Institute of Statistics with its subsequent amendments and additions;

- Government Decision No. 348/2016 on the approval of the National Annual Statistical Programme 2016;

- Government Decision No. 431/2017 on the approval of the National Annual Statistical Programme 2017;

- Law of the National Archives No. 16/1996, with its subsequent amendments and additions.

- Order of NIS President No. 1192/2015 on the setting up of for FSS 2016 team, changed and completed by the President’s Orders No. 1388/2016, No. 145/2017 and No. 502/2017

- Order of NIS President No. 446/2016 regarding the approval of statistical tools for FSS 2016, following the approval of Methodological Advisory Committee

The first two legal acts represent the general legal base for conducting the statistical surveys in Romania, as well as the provisions relative to the National Institute of Statistics, the producer of the Romanian official statistics, and the fundamental principles of official statistics.

The National Annual Statistical Programmes for 2016 and 2017 respectively were approved through the Government Decisions No. 348/2016 and No. 431/2017. They include the FSS 2016 sheets with the related activities for the two concerned years.

The FSS 2016 sheets contain information about:

- objective,

- survey type,

- coverage,

- number of observed units,

- main variables surveyed,

- data-providing units,

- collaborating institutions,

- responsibilities of the implied institutions,

- primary data collection method,

- data collection support,

- moment/periods of reference,

- data registration/collection period,

- main indicators resulted,

- data processing profile,

- data dissemination ways,

- data dissemination deadlines,

Thus, in the case of the agricultural holdings without legal personality the data registration was ensured by the interviewers, who benefitted from the Interviewer’s Handbook (definitions and methodological detail explanations of every variable researched, with concrete examples of completion of the questionnaire and possible cases that may be encountered on the field, for each questionnaire chapter).

The data on the agricultural holdings without legal personality were collected with the help of about 3000 interviewers recruited from agricultural/economic/IT and other domain experts with at least an average level of education. At county level, 42 co-ordinators were hired for a pre-established period (one co-ordinator for each county) and 126 IT operators.

The interviewers identified themselves with a card signed by NIS President proving his/her official identity for the position assigned.

- In performing his/her duties connected to FSS 2016, the interviewers were given the protection guaranteed by the law to the people implied in the exertion of the state authority.

The interviewers were given a fee set by a NIS President’s order, according to the FSS 2016 budget.

The interviewers assignments were the following:

- acceptance by signature of his recruitment as interviewer,

- participation in the training sessions organised by the county statistical offices,

- receiving against signature the interviewer’s folder (list of holdings to be interviewed, questionnaires, handbook, etc.) and checking the related contents,

- studying the instructions for filling-in the FSS 2016 questionnaires and observing their provisions,

- preserving the confidentiality of the data and information contained in the questionnaire (this obligation was included in the interviewer’s work contract),

- mandatory carriage of the card during the whole data collection period to prove the interviewer’s official quality,

- interviewing the declarants,

- informing the co-ordinators on the data collection stage on a permanent basis,

FSS – Farm Structure Survey

CAM – Methodological Advisory Committee

SIRUTA – Nomenclature of Territorial Administrative Units

DTS – County Statistical Offices

DSAM – Department of Agricultural and Environment Statistics

DGITIS – General Department for IT and Statistical Infrastructure

DAISAG – Department for Purchases, Investment and General Administration Services

ACS - Agricultural Crop Survey

LAPS - Livestock and Animal Production Survey

The agricultural holding as a statistical observation unit means a single unit, both technically and economically which has a single management and which undertakes agricultural activities by using agricultural areas and/or animal husbandry or activities meant to maintain the agricultural land in good agricultural and environmental conditions either as a main activity or as a secondary one.

The concerned agricultural activities are the following:

- cultivation of non-permanent crops,

- cultivation of permanent crops,

- crop cultivation,

- mushroom cultivation,

- animal breeding,

- crop cultivation combined with animal breeding,

- agricultural land maintained in good agricultural and environmental conditions.

The following categories of units are considered agricultural holdings only if they carry out agricultural activities as well:

- stables for racing/riding/galloping horses (i.e. the land used for riding horses training),

- fairs, slaughter houses (without animal breeding),

- hunting, forestry and logging,

The households that did not exceed a minimum threshold were not agricultural holdings and, consequently, they were not registered: they had as utilised agricultural area only the kitchen garden (an area equal or smaller than 15 ares) and grew only few poultry (less than 10) for own consumption.

The holding location was made by a strict observance of the elements hierarchically shown below according to:

- the most important parcel,

- location where most of the holding agricultural activities take place,

As regards the national characteristics the reference period was the 2015-2016 crop year.

The reference periods were as follows:

1. Crop year 2016 (1 October 2015 - 30 September 2016) for:

• Land use
• Soil and manure management practices
• Information on the irrigation
• Organic farming – crop sector
• People having worked in agriculture
• Other gainful activities

2. The last 3 years (2014, 2015 and 2016) for:

• Rural development measures

3. The moment of reference was 31 December 2016 for:

• Livestock numbers
• Organic farming – animal sector

The FSS 2016 main activities are presented in the General FSS 2016 organisation and carried out programme:

• At central level: General Department of Economic Statistics, Department of Agricultural and Environment Statistics, General Department for IT and Statistical Infrastructure, Department for Budget and Accountancy, Department of Human Resources, Department of European Affairs and International Cooperation;
• At county level: all 42 statistical county offices.

The main tasks of NIS were:

• General Department for IT and Statistical Infrastructure development of IT application for data processing at county and central level,
• The following activities were carried out within the Department of Agricultural and Environment Statistics: drafting of legal acts, design of questionnaire, Interviewer’s Handbook, the methodological guide, establishing the processing requirements at county and central level,  data checking and validation, using control tables, IT application use, data integrity analysis, non-response treatment, data grossing up, final tables design and achievement, publication preparation and making and Eurofarm file.
  • The recruitment and training of the interviewers, monitoring their activity throughout the survey and reception and analysis of questionnaire filling was the responsibility of the county offices.
  • Co-ordinators were trained by the county offices staff and in their turn they trained the local interviewers, at county level.
  • The data collection was done by direct interview of the holder or any other adult member of the holding in case of the agricultural holdings without legal personality or by self-registration by the holding manager or any competent person for the agricultural holdings with legal personality. 3 000 interviewers were hired for FSS 2016, each of them having to fill in about 101 questionnaires on average. As a rule, the interviewers had an agricultural or economic training and most of them took part in the previous structural surveys.
  • The completed questionnaires were received at county offices level, the data entry being ensured by additionally-hired staff. Also the county offices provided the first data validation. The county offices sent the data files to the headquarters where the other survey stages took place until obtaining the final results.
  • The data processing was performed as follows:

- at county level: the data entry made by additionally-hired staff, data validation, error solving, comparison with other sources, data integrity control, control tables, sending the files with correct data to the central level;

- at central level: data files reception from the county offices, data validation, error solving, control tables, non-response treatment, data grossing up, estimations of the main characteristics at county level and sending them for validation to the county offices, final estimations of all the surveyed characteristics, making the final tables, the publications and the Eurofarm file in order to send it to Eurostat in the specific format.

During all these phases a permanent contact with the statistical county offices was assured and each of the problems occurred was solved.

- county (NUTS 3);

- UAA size class (8 classes): 0-0,1; 0,1-0,5; 0,5-1; 1-5; 5-10; 10-50; 50-100; >100;

The final FSS 2016 sample was of about 303 000 agricultural holdings without legal personality and all the agricultural holdings with legal personality (~ 28 000). The total sample size was about 331 000 agricultural holdings.

- lack of an unique identifier between statistical Farm Register and other administrative agricultural registers;

- different definitions and methodologies for the observation units, variables (indicators);

• The duty to prove the official position as interviewer by showing the personal card when first visiting the agricultural holdings without legal personality;
• Interviewing a competent person within the agricultural holdings without legal personality, preferably the holder or any other adult member of the holding in full working capacity;
• Avoiding taking the interview in front of people who do not belong to the respective holding by explaining the fact that the information is confidential and can only be used for statistical purposes;
• Handing over a basic unfilled questionnaire to the interviewee so that he/she may follow the questions more easily. After the data are filled in, the basic unfilled questionnaire must be returned;
• Getting precise and sincere replies, the questions were clearly and politely formulated;
• In case the questions have several answering variants, the interviewee was shown a complete list of those so he/she may choose the correct variant;
• The interviewee must not be interrupted before finishing to answer even that he/she hesitates (the hesitation may be due to the fact that the respondent tries to remember various aspects related to the information requested);
• Interviewer’s return to the holdings that could not be contacted;
• Re-contacting the respondents (in the case of agricultural holdings without legal personality)

• Weekly monitoring of the data collection stage and coordinators intervention in case of delays behind or other malfunctions.

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

• at interviewer’s and coordinator’s level:

          - observance of work methodology;

          - reliability of questionnaire-registered data (correlations among indicators, comparisons with other sources, agreement with the sample, etc.);

          - observance of questionnaire filling rules.

• at county office level: automatic control made by the IT application.

The IT application was designed according to the processing requirements for the validation rules and the correlations established to enter the questionnaires into the database. The use of this application determines the input and logic validation of the questionnaire-registered data and if necessary the drawing up of a checklist. The validation rules included in the IT application contained checkings for every questionnaire chapter and among chapters. These lists were analysed by authorised staff and corrections were made where appropiate.

• at central level:

- Automatic control made by the IT application after the validation rules: the data centralised at NIS head office underwent an automatic control following the same procedures as at county office level. The errors spotted were also solved at county office level and the correct files were retransferred to NIS.

- Automatic control of data integrity, completeness and other aspects originally not included in the processing requirements. Automatic procedures were developed to analyse the agreement between database data and the ones in the sample taking into account the non-response and also procedures for analysing the data completeness against the completeness code and other analysing procedures for correlations and limits. Following such analyses automatic corrections were applied to the database.

- Data control in comparison with other sources. To this purpose, a specialist team was formed at NIS level in order to achieve the check of the centralised expanded data versus other sources: GAC 2010, FSS 2013, the crop survey and livestock survey.

- The Validation Rules included in the Data Supplier Manual  were used to validate the Eurofarm file.

Tools used for data validation:

The data were validated by means of several tools:

- manual control of the questionnaire-registered data;

- automatic control made by the IT application through the validation rules implemented;

- control by control tables;

- accounting records of the work on the holding (Y/N);

• The reference moment or period was specified on every chapter heading;
• The questionnaire included the arithmetical checks between rows;
• If the queries had to be ticked off, mention was made on the questionnaire if it was a single or multiple answering variant.

Throughout the interview, the interviewer had several obligations that contributed to the reduction of measurement errors:

• The obligation to decline his/her official quality as interviewer by showing the personal card when first visiting an agricultural holding without personality;
• Interviewing a competent person from the agricultural holding without personality, preferably its holder or any other adult member in full working capacity;
• Avoiding the interview in front of people that do not belong to the concerned holding by explaining the information is confidential and to be used only for statistical purposes;
• Handing over an unfilled questionnaire to the interviewee in order for him/her to be able to follow the questions more easily;
• To get precise and sincere replies, the questions were formulated clearly and politely;
• If the questions had several answering variants, the interviewee was presented a full list of them so he/she may choose the correct one;
• Taking down the replies as they were provided by the interviewee;
• Coming back to certain questions where the answer did not meet the arithmetical checks or if they did not correlate;
• Requesting the interviewee’s signature on the completed basic questionnaire to certify data quality.

There was not made an analysis of non-responses.

Once the questionnaires were collected a first data pre-validation was performed.

The filled-in questionnaires were re-checked by the co-ordinators and by county offices staff. In case of missing data, the unit was contacted in order to obtain these data. If this procedure did not function, the data were adjusted or imputed.

Data correction was made as follows:

• on-the-spot, if errors were included in the logical checks;

• tables at county level, containing data from FSS 2016 and FSS 2013
• tables at county level, containing data from FSS 2016, LAPS 2016  and ACS 2016

The persons authorized to make corrections in FSS 2016 data were:

• at territorial level: coordinators and statisticians involved in FSS 2016
• at central level:
  - statisticians involved in methodological activity of FSS 2016
  - IT personnel for automatic corrections

Corrections were made based on IT Application using Visual FoxPro.

The final results of FSS 2016 were sent to Eurostat according to the provisions of Regulation (EC) No. 1166/2008 of the European Parliament and of the Council accompanied by the National Methodological Report to the deadlines set in the grant agreement and the Annual National Statistical Programme 2017, at 31^st of December.

There occurred no differences between the national definitions and the EU ones as regards the characteristics or their classification.

The number of hours used for a full time-employee to calculate the Annual Work Unit was of 1 800 hours.

The range of days in the validation rules, used in order to establish the percentage bands was:

0 < X < 56 → „24”
56 ≤ X < 112 → „49”
112 ≤ X < 169 → „74”
169 ≤ X < 225 → „99”
X = 225 → „100”.

Taking into account this aspect, a special code for the legal status, “Town halls”, was created into the FSS 2016 questionnaire, under Chapter I, Section 2.2., code “8”, in order that the common land units are easier to be identified.

The common land area is also marked out at the point 6.1 `Type of tenure`.

However, the common land units may have other additional land, recorded under arable land crops, permanent crops, unutilised agricultural land, wooded land, but this area was not considered as common land.

In conclusion, all the units with code “8” – “Town halls”, having pastures and meadows area were considered in “Eurofarm” file as “common land units”, and the common land area was represented by their total area with pastures and meadows, marked out also at point 6.1. `Type of tenure`

The table hereunder presents the common land statistics by land type:

The table hereunder presents the situation containing the number of the common land units, split by each category of the common land. It is necessary to mention that are agricultural holdings having more than one category of pastures and meadows.

Thus, special holdings were set up for the registration of common land at town hall level which registered the whole area of pastures and meadows within a locality jointly used by various agricultural holdings.

 There are no significant differences concerning the area and the number of holdings with common land (thsd. ha)

a) Comparing the final results of FSS 2016 with the data obtained from the Agricultural Crop Survey (ACS) 2016, as follows:

When analysing the FSS 2016 and ACS 2016 data, one can notice very close results, the differences between the two statistical surveys being fairly small within the permissible 10 % limit for all UAA categories.

b) After comparing the results of FSS 2016 and those of the Livestock and Animal Production Survey 2016 (LAPS 2016), the following findings were obtained:

The data related to the main animal species resulted from FSS 2016, compared to those obtained from the LAPS 2016 show insignificant differences for bovines, sheep, goats, poultry and bigger ones for pigs.

Mention must be made that the livestock numbers for FSS 2016 are registered on 31 December and for LAPS 2016 on 1 December. The 12 % difference for pigs could be explained by the fact that some of the pigs were slaughtered during Christmas time when, traditionally, every family slaughters a pig.

The relatively small difference, of less than 10 %, between the FSS 2016 data and those obtained from the other annual agricultural surveys both in the crop and animal sector is another factor certifying the quality of the concerned results.

FSS statistics are not reconcilable with those obtained through other data sources or statistical domains, for some reasons, such as: precision requirements, different target populations or reference periods.

► At national level through:

•  Press release on 15 December 2017;

• Publication in 2 volumes (book + CD) on 29 December 2017:
• Volume 1: "Farm Structure Survey 2016 – General data at national level“;      
• Volume 2: "Farm Structure Survey 2016 – Data by macro-region, development region and county“.

The publications are bilingual (Romanian/English) and contain tables with results at national level and by macro-region/development region/county accompanied by methodological specifications and metadata.

Version in English: www.insse.ro

- Data collection questionnaire,

- Interviewer’s handbook,

Keeping the data confidentiality by NIS permanent staff is mandatory according to Law No. 226/2009.