/ EUROPEAN COMMISSION
EUROSTAT
Directorate E: Sectoral and regional statistics
Unit E-1: Agriculture and fisheries

Luxembourg, 31 July 2013

ESTAT/E1/ME/JS/AV/ga

DOC. CPSA/AE/118
Available in EN only
* Please note that paper copies of the documents will not be provided at the meeting*

Working Group
"Agriculture and Environment
– Pesticides Statistics"

Bech building – Room Quetelet
Kirchberg, Luxembourg

26 and 27 September, 9:30 A.M.

Chaired by: Mr. Marcel Ernens

4.2. Confidentiality and data dissemination*

* Document available on Circabc:

1.Introduction

Regulation 1185/2009[1] provides very strict instructions on how pesticide statistics can be disseminated and stresses that confidentiality must be safeguarded at all times.

This document is intended to identify the issues related to confidentiality and to describe how data will be disseminated by Eurostat. It is not intended to guide the Member States on their way of treating confidentiality.

2.Background

When Regulation 1185/2009 was prepared and discussed in the different institutions, much emphasis was put on safeguarding the confidentiality of the data.

  • All data, including the confidential ones, shall be transmitted to Eurostat (Art.3(2)).
  • Eurostat shall aggregate the data before publication (Art.3(4)). This means that Eurostat cannot publish any pesticide data on active substance level, all must first be aggregated to at least chemical class level (second level). In this work the confidentiality at Member State level must be taken into account.
  • The confidential data shall be used by national authorities and by the Commission, including Eurostat, for statistical purposes only (Art.3(4)). This could potentially have a negative impact on the calculation of pesticide risk indicators.
  • Member States are obliged to publish the data on the sales and agricultural use of pesticides on their websites, but taking due account of statistical confidentiality (Annex I, Section 5, paragraph 2 and Annex II, Section 5, paragraph 5).

3.Considerations

  • The sustainable use of pesticides directive (2009/128/EC[2]) states that pesticide risk indicators shall be established and calculated. The directive stipulates that the Member States shall calculate national risk indicators and the Commission at Community level by using statistical data collected in Regulation 1185/2009, and that the results shall be used to assess progress in the field. These risk indicators must be published.
  • The issue of confidentiality has been treated very differently depending on the Member State (see table 1). The treatment is obviously different depending on the source of the data (see table 2). If the data are collected from the importers or other central players, there is a bigger risk that the data are confidential than if the data are collected from the retail sellers.
  • It is in the interest of the public to have access to as much information on pesticides as possible, due to the potential risk to environment and human health. Therefore care should be taken not to assign confidentiality flags automatically, but only when there is a real risk of disclosure. Many of the most important pesticides, such as sulphur, glyphosate, 2,4-D, MCPA, MCPB, etc., have been on the market a long time and are now produced, imported and marketed in most countries by several companies so there is on larger markets hardly a risk for disclosure. The national pesticide authorities have to keep registers of all authorised companies on the market for a specific pesticide substance, so the information on this aspect is easy to retrieve through good collaboration.
  • It is likely that most data on the agricultural use of pesticides will be collected through farm surveys, or similar. This means that confidentiality will be a less of an issue for these data, as the risk of disclosure is minimal. Eurostat is of the opinion that the companies selling pesticides cannot be considered in this context, even if they are the sole actors on the market for a specific substance. It is only if they are the actual data providers that they can be taken into consideration.
  • The methodological unit in Eurostat has in collaboration with the Expert Group on Statistical Disclosure Control (EGSDC) drafted a model confidentiality charter. This can be adapted to each statistical domain, depending on the situation. See the Annex for a version adapted to pesticide statistics that could be adopted as the basis for confidentiality treatment in this specific domain.
  • The idea of the charter is that it will take account of the confidentiality rules applied by the countries in their work, so that the same rules are applied also in Eurostat. It is therefore important to identify these rules.
  • Based on the information received in the quality reports, in the suggested charter the assumption is made that Member States use the threshold rule. This means that the dominance rule (where a cell is considered confidential because a company dominates the cell in a manner allowing this company to be identified due to this) is not used. Obviously, if this assumption is false, the charter would need to be rewritten, but then there would be a need to change the data transmission format to allow for more than one code to be used as confidentiality flag. In order to allow a better understanding of this issue there are examples in the charter to describe the different alternatives.

4.Dissemination

It is in the interest of Eurostat to provide as much information as possible to the public, but at present too many confidential data are creating problems in disclosing data. As stated earlier, Regulation 1185/2009 does not allow Eurostat to publish data on active substance level, only aggregated data. Eurostat wishes to resolve the matters in such a manner that would allow publishing as many EU totals as possible on the lowest aggregated level. This in mind, Eurostat considers that it is in the public interest to ensure that as much information is disseminated as possible.

Regulation 1185/2009 states that Member States must publish the pesticide data on their websites. In order to give easier access to this information, Eurostat would like to publish the links to these websites on the Eurostat website. Member States are therefore urged to send these links to Eurostat.

5.Questions to the meeting

The meeting participants are asked to inform Eurostat on the following issues:

  • Do you agree to update the data on pesticide sales after reviewing the actual confidentiality of the data provided?
  • Do you agree to the confidentiality charter for pesticide statistics?
  • Can you provide more information on the national treatment of confidentiality as required by the charter?
  • Can you send links to the websites where you have published the pesticide sales data on your websites as required by the Regulation?

Table 1. Pesticide sales data delivery and confidentiality by country

Data delivered for / Data confidential for
Major groups / Categories of products / Chemical classes / Active substances / Major groups / Categories of products / Chemical classes / Active substances
Total possible number / 6 / 32 / 116 / 518
AT / 6 / 27 / 94 / 266 / 0 / 26 / 92 / 266
BE / 6 / 28 / 93 / 245 / 0 / 0 / 30 / 517
BG / 1 / 93 / 0
CY / 6 / 25 / 70 / 161 / 0 / 0 / 0 / 0
CZ / 6 / 28 / 85 / 224 / 0 / 0 / 0 / 0
DE / 6 / 29 / 87 / 262 / 0 / 0 / 64 / 260
DK / 149 / 0
EE / 5 / 19 / 40 / 76 / 0 / 3 / 6 / 12
EL / 6 / 25 / 78 / 209 / 1 / 2 / 27 / 144
ES / 5 / 29 / 91 / 301 / 0 / 0 / 0 / 170
FI / 0 / 0 / 0 / 138 / 0 / 0 / 0 / 124
FR / 6 / 30 / 27 / 518 / 0 / 0 / 0 / 132
HU / 1 / 26 / 88 / 252 / 0 / 2 / 6 / 17
IE / 5 / 20 / 2 / 1 / 1 / 6 / 72 / 166
IT / 0 / 1 / 0 / 273 / 0 / 1 / 0 / 176
LT / 5 / 22 / 52 / 113 / 1 / 6 / 27 / 85
LV / 5 / 23 / 57 / 128 / 1 / 22 / 54 / 114
MT / 5 / 23 / 48 / 92 / 0 / 0 / 9 / 24
NL / 6 / 32 / 110 / 224 / 0 / 15 / 72 / 216
PL / 6 / 31 / 88 / 237 / 0 / 0 / 0 / 518
PT / 4 / 23 / 4 / 195 / 0 / 0 / 0 / 0
RO / 6 / 25 / 80 / 209 / 0 / 0 / 0 / 0
SE / 153 / 0
SI / 6 / 27 / 77 / 203 / 0 / 0 / 0 / 0
SK / 5 / 25 / 76 / 193 / 0 / 5 / 39 / 169
UK / 6 / 27 / 79 / 192 / 3 / 21 / 75 / 192

Table 2. Data source and status consideration (as understood by Eurostat) by country

AT / Admin data from authorisation and approval holders as well as distributors to Austrian Agency for Health and Food Safety / Admin
BG / Administrative data according to the Law for Plant Protection. The entities allowed to distribute PPP on the market get licenced and announce the active substance quantities on annual basis. Ministry of Agriculture and Food / Admin
CY / Statistical survey to the suppliers of pesticides / Statistics
CZ / The State Phytosanitary Administration is responsible for data collection and processing administrative data / Admin
DE / Federal Office of Consumer Protection and Food Safety (BVL) conducts a total survey every year, covering alle holders of authorisations or parallel import permits for plant protection products / Statistics
DK / Danish Ministry of the Environment Environmental Protection Agency, the reporting of sales is obligatory to all approval holders / Admin
EE / Statistics Estonia conducts a (total survey). The surveyed population of importers and suppliers of plant protection products / Statistics
ES / Ministry of Agriculture, Food and Environment sent a statistical questionnaire to all the authorized holders / Statistics
FR / Ministère de l'agriculture, de l'agroalimentaire et de la forêt from a compulsory reporting system / Admin
HU / Research Institue of Agricultural Economics and National Food Chain Safety Office collect data that are obligatory and full-scale / Admin
IE / Department of Agriculture Food and the Marine, Pesticide Registration and Control Division, administrative data is sourced from all the product authorisation holders and marketing companies each year / Admin
IT / ISTAT - Italian National Institute of Statistics, a statistical survey carried out as a census / Statistics
LT / Statistics Lithuania, market survey as a full census, where data by active substances are collected from companies and enterprises, which are producers authorization owners. / Statistics
LV / State Plant Protection Service of Latvia, The data source is full survey of all respondents - PPP distributors which import or distribute PPP have to submit information for the previous calendar year. / Admin
MT / National Statistics Office contacted all importers for the data / Statistics
NL / National Statistics Office receives administrative data of sales by active substance from the Plant Protection Office (NVWA) and classifies them according to the classification of Annex III. / Admin
PL / Central Statistical Office carries out a full, compulsory survey on sales by producers and sales by importers using an electronic or paper questionnaire by of entities holding a PPP marketing and use authorisation. / Statistics
PT / Statistics Portugal (INE) uses data obtained from an administrative source from the National Authority (DGAV) / Admin
RO / National Institute of Statistics carries out a statistical census survey covering the certified pesticide selling units and including the economic operators and the certified natural persons registered for PPP merchandising on the Romanian market. The data collection took place in March 2012, using field operators based on face-to-face interview questionnaires / Statistics
SE / Swedish Chemicals Agency (KemI) uses data from authorisation holders that are obliged to supply data on the exact amount of product placed on the Swedish market / Admin
SI / Statistical Office of the Republic of Slovenia uses data on pesticides products sold in wholesale on the domestic market that are yearly reported to the Administration of the Republic of Slovenia for safety food, veterinary medicine and plant protection / Admin
SK / Statistical Office of the Slovak Republic uses data from Central Controlling and Testing Institute in Agriculture (CCTIA) which gathers reports on pesticides placed on market that authorization holders/holders of a parallel trade permit for pesticides trade are obliged by law to report data / Admin
UK / Data are collected by the British Crop Production Council (BCPC) on behalf of the Office for National Statistics. A census of businesses was carried out using the list of active substances contained within Annex III of EC Regulation 1185/2009 / Statistics

Confidentiality Charter
for the treatment of confidentiality
at EU level

Pesticide statistics

1. Introduction

2. Glossary

3. Identification of sensitive cells at the national level (primary confidentiality)

4. Assumptions

5. Rules for publication / protection of EU aggregates

6. Legal references

7. Annex A

Page 1 / 18

1.Introduction

One of Eurostat’s main duties is to disseminate data aggregated at European Union level. The Commission needs these statistics in order to develop and monitor EU's policies. They also meet an increasing demand from users outside the Commission: national administrations and private businesses. In order to make data useful for users,they should be available to the maximum extent possible while guaranteeing that no data on individual respondents is disclosed.

The main purposes of this charter are:

  • to lay down practical rules in order to guarantee that the confidentiality pattern of data disseminated by Eurostat is consistent with that of the data disseminated at the national level;
  • to lay down the confidentiality rules to be applied to the EU aggregates taking into account the practices and laws in the area of confidentiality in the Member States.
  • to ensure a proper protection of confidential data and to encourages Member States to report them and
  • to facilitate the systematic treatment of confidentialityand thus, to reduce significantly the need to consult Member Stateseach time a change occurs either into the confidentiality pattern or in the size of an EU aggregate.

The Glossary of terms and references to legal acts complete the charter.

2.Glossary

EU totalquantitative measure at EU level of a particular variable

Confidential cellthe cell of a table which has to be protected due to the risk of statistical disclosure

Confidential clusterthe group of countries contributing to an EU aggregate or a EU total whose data is confidential for a particular variable

Confidential statistical datathe data which are subject to confidentiality clauses; data used by the national statistical institutes and Eurostat are considered confidential when they allow statistical units to be identified, either directly or indirectly, thereby disclosing individual information (Regulation (EC) 223/2009 Article 3)

Contributora statistical unit (e.g. business/enterprise) taking part in an aggregate (e.g. confidential cluster/national total/EU total/EU aggregate)

European/EU aggregate quantitative measure at EU level of a particular variable for a characteristic or a set of characteristics of a population

National totalquantitative measure at national level of a particular variable

Primary confidentialityit concerns tabular data, whose dissemination would allow attribute disclosure. The two main reasons for data to be primary confidential are too few units contributing to a cell or the dominance of one or two contributors in a cell. The limits of what constitutes "too few" or "dominance" vary between statistical domains and across Member States.

Secondary confidentialityapplication of SDC methods to some safe cells in order to prevent disclosure of primary confidential cells through recalculation (based on additive feature of tabular data and/or hierarchical structure of statistical data). It could act solely onthose EU aggregates incorporating confidential data but safe for publication. The EU aggregates not incorporating confidential data can neither be suppressed nor rounded/perturbed as they are disseminated by NSIs.

Statistical confidentiality it means the protection of data related to single statistical units which are obtained directly for statistical purposes or indirectly from administrative or other sources. It implies the prevention of non-statistical utilisation of the data obtained and unlawful disclosure.

Suppressionit means that the values of confidential cells are not shown in the table, but replaced by a symbol such as "x".

3.Identification of sensitive cells at the national level (primary confidentiality)

Member States may apply different rules to identify confidential data. Also, Member States have complete freedom in setting up their systems of rules to protect their national statistical data.

The most common sensitivity rules are:

  • Threshold rule: data (either a cell or a marginal total)in a table is confidential if the number of contributors is less than a specified number (threshold). In otherwords, a minimum number of contributors in a cell is needed to publish the data inthat cell.
  • Dominance rule (or (n,k) rule): according to this rule,a cell isbe regarded as confidential if the n largest units contribute more than k% to the cell total.
  • P-percent rule: it relates to the largest 2 contributors of a cell within tabular data. From general knowledge, the second largest contributor can estimate the value of the first contributor within p%.

To make the rules protective, all qualifiers (i.e. p, k and n) are decided by each Member State and are confidential for users. In addition, a Member State might use different qualifiers across domains or among variables in a domain.

The p-percent rule and dominance rule make sense only if it is assumed that intruders are able to identify the largest contributors to the cell. This depends on the statistical domain, the breakdown and the variable. The p-percent rule is the most effective concentration rule and its use is recommended.

4.Assumptions

The confidentiality charter is developed around the following principles:

  • Same national confidentiality rules and qualifiers apply to compileEU aggregates.

In order to ensure a consistent protection of confidential data, Eurostat applies national rules and qualifiers when treating EU aggregates. For instance, if a national confidential total dominates an EU aggregate, the qualifiers against which the disclosure risk is assessed, are the national qualifiers of the concerned Member State.

–dominance rule: the dominance criterion refers to n concerned individual statistical units and is evaluated for each variable separately. If negative values occur, the dominance will be evaluated on the basis of the absolute value of the figures. The dominance threshold (percentage of the confidential EU subtotal that should not be exceeded) will be agreed with the individual countries. For countries for which the dominance rule does not apply at national level, there is no dominance threshold applied for EU aggregate either.

–p-percent rule: for the sake of simplicity and clarity and because the p-percent rule is used by very few Member States, Eurostat suggests to replace it by using the (2,k) dominance rule. According to the ESSnet-handbook SDC (see page 121), if we define k as 1/(1+p), then the (2,k) rule is more protective than the p-percent rule. Eurostat would prefer thus to overprotect some EU aggregates and to keep the procedure for treatment of confidentiality simple and easy to implement. However, this recommendation holds solely in the current situation when very few countries (i.e. 2 out of 27) use the p%-percent rule.

  • Secondary confidentiality treatment: priority to the dissemination of EU totals over EU aggregates.

Suppression is by far the most commonly method used to protect confidentiality in tabular data, but it can not be applied on national totals, because, unless they are declared confidential, they may already have been published by Member States. Suppression can apply solely on those EU aggregates which include confidential cluster. The suppression of EU aggregates shall maximize the number of disseminated EU totals.

  • Structured and standardized confidentiality reporting to Eurostat

The burden on Member States should be kept at a reasonable level. Therefore, the confidentiality reporting shall be integrated into the future standard on data and metadata transmission. The currently used system of flags is the following: