52008DC0160

Report from the Commission to the Council and to the European Parliament on the implementation of Regulation (EC) No 1177/2003 of the European Parliament and of the Council of 16 June 2003 /* COM/2008/0160 final */


[pic] | COMMISSION OF THE EUROPEAN COMMUNITIES |

Brussels, 31.3.2008

COM(2008) 160 final

REPORT FROM THE COMMISSION TO THE COUNCIL AND TO THE EUROPEAN PARLIAMENT

on the implementation of Regulation (EC) No 1177/2003 of the European Parliament and of the Council of 16 June 2003

REPORT FROM THE COMMISSION TO THE COUNCIL AND TO THE EUROPEAN PARLIAMENT

on the implementation of Regulation (EC) No 1177/2003 of the European Parliament and of the Council of 16 June 2003

1. INTRODUCTION

Under the terms of Article 16 of Regulation N° 1177/2003 of the European Parliament and of the Council of 16 June 2003 on the EU-SILC (Community Statistics on Income and Living Conditions) project, 'No later than 31 December 2007, the Commission will submit a report to the European Parliament and the Council on the work done under this Regulation'.

This report documents the progress made by Member States, Candidate and EFTA countries, together with the Commission, with respect to the provisions of the abovementioned EP and Council Regulation and the associated Commission regulations issued during the period 2003-2007.

2. MAIN POINTS

2.1. SILC: a source of reference data for income and poverty analysis at EU level

Since the adoption on 16 June 2003 of Regulation N° 1177/2003 on the EU-SILC (Community Statistics on Income and Living Conditions) project, significant progress has been achieved.

EU-SILC was launched gradually between 2003 and 2005 in all EU Member States and has become the source of data for the analysis of income distribution and social inclusion at EU level. In 2005, cross-sectional data for all 25 Member States plus Norway and Iceland became available. These data have been used for producing common indicators for the social protection and social cohesion process (overarching, social inclusion, pensions and health and long-term care) included in the Joint Social Inclusion and Social Protection (JSISP) report, which is a component of the Open Method of Coordination (OMC).

EU-SILC was launched in 2003 on the basis of a gentlemen’s agreement in six Member States (Belgium, Denmark, Greece, Ireland, Luxembourg and Austria) as well as in Norway.

In 2004, under Regulation N° 1177/2003 of the EP and Council, EU-SILC was implemented in twelve EU-15 countries (Germany, Netherlands and the United Kingdom delayed the launch for one year) as well as in Estonia, Iceland and Norway. In 2005, EU-SILC was operating in all EU-25 countries, plus Iceland and Norway. Bulgaria, Turkey and Romania launched EU-SILC in 2006, and Switzerland followed suit in 2007. EU-SILC is likely to be launched at a later date in the Former Yugoslav Republic of Macedonia and in Croatia.

From the very beginning, SILC has become an annual operation in all countries.

2.2. Legal framework

The legal framework of the project has also been developed. Apart from the EP and Council regulation, five implementing Commission regulations - on 'Sampling and tracing rules', on 'Definitions', on the 'List of primary (annual) target variables', on 'Fieldwork aspects and imputation procedures' and on 'Quality reports' - were approved by the Statistical Programme Committee (SPC) and published in the Official Journal (OJ) of the European Union on 17 November 2003 in the case of the first four and on 9 January 2004 for the last one. In addition, a Commission Regulation has been published annually containing the list of secondary target variables, i.e. modules introduced in EU-SILC, with the possibility of a topic being repeated every four years or at longer intervals. Commission regulations on the inter-generational transmission of poverty (2005 module, N° 16/2004 of 6 January 2004), on social participation (2006 module, N° 13/2005 of 6 January 2005), on housing (2007 module, N° 315/2006 of 22 February 2006) and on over-indebtedness and financial exclusion (2008 module, N° 215/2007 of 28 February 2007) were published in the OJ.

2.3. EU funding

The annual EU funding of the project has been ensured through grant agreements concluded with NSIs. For the first four years of SILC data collection, all Member States plus Norway and Iceland received EU funding for carrying out the project (in Estonia, the four-year EU funding started not in the first but in the second year of data collection). For the two new Member States that joined in 2007 (Bulgaria and Romania), EU funding started for the 2006 operation.

In total, EU funding amounted to around €6.5m for the 2004 data collection and around €11m for the 2005-2007 data collections respectively.

2.4. Collection and dissemination of SILC data

Overall the collection of SILC data has been a success. Eurostat, together with the network of the National Statistical Institutes (NSIs), has been able to collect, check and issue these data within reasonable periods of time. However, there is a significant gap of 2 years between the reference year of the collected data and the latest available economic data.

For the cross-sectional component of EU-SILC, two waves of data (2004 and 2005) have already been sent to Eurostat. The third wave exercise (year 2006) is currently in progress. The SILC 2004 operation concerned 15 countries (twelve of the 15 old Member States plus Estonia, Norway and Iceland), while 27 countries (all 25 EU Member States plus Norway and Iceland) participated in the 2005 round. All countries concerned were able to send Eurostat the 2004 data on time, either by the end of October or by the end of November depending on the country. For the 2005 operation, two Member States (Slovenia and the United Kingdom) encountered a delay of a few months in finalising the cross-sectional data.

These data were then disseminated by Eurostat both in the form of aggregated tables on the Eurostat website and in the form of anonymised micro-data files that the research community can access under strict conditions. Malta, however, opposed the release of micro-data. The anonymisation of individual records is obtained by removing or recoding variables that could be used indirectly to identify individuals or households. The cross-sectional anonymised micro-data files were issued in June 2006 and June 2007 for the SILC 2004 and 2005 information respectively. Of the fifteen countries that launched SILC in 2004 (12 of the 15 old Member States plus Estonia, Norway and Iceland), 14 were part of the 2004 issue (with the exception of Iceland). The 2005 issue covered 26 of the 27 countries concerned (all 25 Member States, except Malta plus Norway and Iceland).

The longitudinal component of EU-SILC information gathering on 4-year individual trajectories will not be fully operational until 2010. So far, only the longitudinal information corresponding to the 2004/2005 SILC data has been released. This release was limited to 13 countries, as the other countries either only started SILC in 2005 or were experiencing problems in the validation of the SILC longitudinal component.

There were also delays in the issue of longitudinal data, as the longitudinal UDB for 2004/2005 was not available until October 2007.

Once the project has reached a mature stage (from 2010), there will be two types of release each year:

- A cross-sectional anonymised User Data Base (UDB) with the most recent wave of EU-SILC;

- A longitudinal anonymised UDB containing individual trajectories of a maximum of 4 years.

2.5. Content of EU-SILC

EU-SILC is a multi-dimensional instrument focusing on income, but at the same time covering housing, labour, health, demography and education, thus making it possible to study the multidimensional approach of social exclusion.

It comprises primary (annual) and secondary (module) target variables.

The primary target variables are either household information or individual information (for persons aged 16 and more) and are grouped into areas.

- At household level, five areas are covered (1) basic/core data, (2) income, (3) housing, (4) social exclusion, (5) labour information;

- The personal level is grouped into five areas (1) basic/demographic data, (2) income, (3) education, (4) labour information and (5) health). Childcare information is also collected for children under 12 years of age.

The secondary target variables (modules) are introduced annually from 2005 in the cross-sectional component of the instrument only. The first EU-SILC modules relate to the following:

2005: Inter-generational transmission of poverty

2006: Social participation

2007: Housing conditions

2008: Over-indebtedness/Financial exclusion

2009: Material deprivation

2.6. Collection unit

EU-SILC covers only people living in private households . Persons living in collective households and in institutions are excluded from the target population. The impact of excluding old people living in institutions, people with disabilities and other vulnerable groups, including the homeless, from the samples may differ from one country to another. Some vulnerable groups living in private households may also be under-represented because they are not easy to reach. However, such groups require specific data collection methods, which are not compatible with the EU-SILC framework.

In EU-SILC two types of collection of household and individual variables have been allowed. In most countries (the non-register countries), all members of selected households aged 16 or over were given a personal questionnaire which they were asked to fill in. In the register countries (DK, FI, IS, NL, NO, SE, SI), only a selected household respondent receives a personal questionnaire, and household and income variables are collected either through a register or from the selected respondent.

2.7. Sampling issues

On the sampling side, all countries collect information on representative probability samples of the target population. Germany is an exception, where an existing quota sample was used at the launching of EU-SILC; however, this non-probabilistic sample will gradually be phased out up to 2008 in accordance with Regulation N° 1177/2003.

In order to be able to produce the cross-sectional and longitudinal information each year, with the longitudinal component covering at least four years, Eurostat has recommended using a rotational panel. For this purpose the sample is divided into four sub-samples, called "rotation groups", each of which is representative of the whole population. The idea is then to rotate out one group each subsequent year and follow up the people in the other groups. Most of the 27 countries have adopted this four-year rotational design recommended by Eurostat. In Norway and France, a rotational design over a longer period (8 and 9 years respectively) is used.

Some countries have also used alternative survey structures. Luxembourg has chosen to use a pure long-term panel, which is supplemented every year with a new sample in order to ensure cross-sectional representativity. Sweden has two separate operations: a cross-sectional survey and a pure long-term panel. Lastly, Finland uses a modified rotational schema (in which the basic two-year rotational panel is supplemented by the follow-up of two sub-samples for two additional years, to be able to have a four-year trajectory).

Administrative data are used by countries where part of the SILC information is available. In the so called register countries" (DK, FI, IS, NL, NO, SE, SI), most income components and some demographic information are obtained from administrative registers. Other personal variables are obtained mainly by telephone interview. In all other countries except Ireland and Germany, the full information is obtained by means of a survey among households and face-to-face interviews with household members. In Ireland, for the households which explicitly gave their agreement (around 25% of the households), the information is obtained from administrative registers. Germany uses a mail survey in which individuals self-administer the questionnaires.

2.8. Sample sizes

As can be seen from the following table, for the SILC 2005 operation almost all countries have met the minimum effective sample size for the cross-sectional component as defined in Regulation N° 1177/2003. The only exceptions are Czech Republic, Netherlands and Portugal.

It is to be noted that a specific derogation was given to the Czech Republic to have a smaller sample size for the first year of data collection.

Achieved effective household sample size (based on achieved sample)[1] | Minimum effective household sample size (as defined in regulation) |

Belgium | 4939 | 4750 |

Cyprus | 3533 | 3250 |

Czech Republic | 3687 | 4750 |

Denmark | 5957 | 4250 |

Estonia | 3970 | 3500 |

Germany | - | 8250 |

Greece | 4283 | 4750 |

Spain | 9088 | 6500 |

France | 8787 | 7250 |

Hungary | 5328 | 4750 |

Iceland | 2958 | 2250 |

Ireland | 4680 | 3750 |

Italy | 15625 | 7250 |

Latvia | 3284 | 3750 |

Lithuania | 4353 | 4000 |

Luxembourg | - | 3250 |

Malta | 3459 | 3000 |

Netherlands | 3931 | 5000 |

Norway | 5991 | 3750 |

Austria | 4476 | 4500 |

Poland | 13183 | 6000 |

Portugal | 2817 | 4500 |

Finland | 8020 | 4000 |

Slovakia | 5147 | 4250 |

Slovenia | 7892 | 3750 |

Sweden | 6133 | 4500 |

United Kingdom | - | 7500 |

For the longitudinal component, the situation can only be evaluated at this stage for the 15 countries that carried out SILC in 2004 and for which 2004 and 2005 longitudinal data are available, namely twelve EU-15 countries (Germany, Netherlands and the United Kingdom delayed the launch for one year) plus Estonia, Iceland and Norway. The table shows that Belgium and Ireland do not meet the minimum size requirement for the first two years of the longitudinal component.

Achieved effective household sample size (based on achieved sample)(1) | Minimum effective household sample size (as defined in regulation) |

Belgium | 2851 | 3500 |

Estonia | 3091 | 2750 |

Denmark | 3250 |

Greece | 3456 | 3500 |

Spain | 6394 | 5000 |

France | 6757 | 5500 |

Iceland | 1908 | 1700 |

Ireland | 2118 | 2750 |

Italy | 11145 | 5500 |

Luxembourg | 2500 |

Norway | 3656 | 2750 |

Austria | 3089 | 3250 |

Portugal | 3250 |

Finland | 3312 | 3000 |

Sweden | 4866 | 3500 |

2.9. Reports and studies

In accordance with Regulation N° 1177/2003, all 15 countries that launched SILC in 2004 submitted an intermediate and a final quality report respectively before the end of December 2005 and 2006. On that basis, Eurostat produced, in June 2006, a single EU quality report combining the intermediate and final national quality reports; this was because, in the first year of data collection, in the absence of longitudinal information there is very little difference between the two types of reports.

For the SILC 2005 operation, Eurostat received intermediate quality reports by the end of December 2006 from all 27 countries concerned, and it produced a synthesis which is now in the finalisation phase (there has been a slight delay).

The process in relation to the final quality reports for 2005 is now ongoing.

The intermediate report includes sections on aspects ranging from accuracy (sample design, sampling and non-sampling errors, method of data collection, interview duration), to comparability (basic concepts and definitions, income components) and coherence.

The final report includes parts relating to relevance, accuracy (sample design, sampling and non-sampling errors, method of data collection, imputation procedure, imputed rent, company cars), comparability (basic concepts and definitions, income components, tracing rules), timeliness and punctuality, accessibility and clarity, and coherence.

In addition, the Commission has launched several actions to evaluate the quality and, in particular, the comparability of the SILC information:

- Methodological studies have been launched covering key issues for comparability at both EU and national levels. This has been achieved through the launching of a call for tenders which is intended to cover issues such as the impact of the definition of household, comparability of administrative and survey data, the method of data collection, as well as the different treatment of negative income, and sampling issues. In addition, grants were offered to National Statistical Institutes to carry out methodological studies on the impact on comparability of the use of national sources (for example, studies on the use of registers as against survey data; impact of using different methods for estimating imputed rent; analysis of coherence with National Accounts or other survey information; study of attrition);

- A Methodology Task Force has been set up, which has discussed issues such as imputation techniques, the treatment of lump sums, imputed rent, and the status of private pension plans within income;

- From 6 to 8 November 2006 Eurostat organised, together with Statistics Finland, a Conference on the SILC project which was held in Helsinki and at which all dimensions of the project were reviewed and an action plan was drafted;

- Eurostat is systematically computing standard errors for the income-based indicators during the first year for which these data are available.

2.10. Publications

Apart from releasing micro-data to researchers and aggregated tables on the Eurostat website, Eurostat produced, in September 2005, a small publication (Statistics in Focus) containing the cross-sectional indicators commonly used as part of the Open Method of Coordination (OMC) on social inclusion. This publication, entitled 'Income, poverty and social exclusion in the EU25', uses the information from SILC 2004. A new issue based on SILC 2005 data is currently being prepared by Eurostat.

In addition, EU-SILC is now listed as the key source for the social cohesion chapters of more general publications, like the Joint Social Inclusion and Social Protection (JSISP) report, the 'Annual Progress Reports', the report on ' Measuring progress towards a more sustainable Europe - 2007 monitoring report of the EU sustainable development strategy (and previous SDS monitoring reports)', 'The Eurostat Yearbook', the 'Social Situation report', the Pocketbook 'Key figures on Europe', and the Panorama on men and women in the European Union.

However, dissemination of SILC information still needs to be improved. For example, two kinds of analysis (and corresponding publications) need to be prepared: a statistical publication (monograph including statistical analysis) plus a series of short statistical analyses in the same format as the Eurostat 'Statistics in Focus' publications.

3. CONCLUSION

Since the adoption of Regulation N° 1177/2003, SILC has become the source of reference data for statistics on income distribution, poverty and social exclusion analysis at EU level, and these data have been used in the context of the 'structural indicators' and OMC exercises. The legal framework has been developed and an adequate EU funding has been guaranteed for the first four years of data collection in each country. The collection and dissemination in the form of tables and micro-data to researchers of the SILC cross-sectional data is under control although there is still margin for progress in terms of timeliness. The longitudinal component of EU-SILC is in the initial stages and will not be fully operational until 2010. The only longitudinal information currently available is limited to 13 countries. A significant amount has been invested in the evaluation of SILC data quality in the form of quality reports and methodological studies. A methodological Task Force, as well as a SILC Conference held in Helsinki in November 2006, also studied a number of quality issues in depth. Lastly, even though several publications were produced, further work is to be done on the dissemination of the SILC information.

[1] Achieved effective sample sizes are obtained by dividing the achieved sample size by the so called design effect. For this table, design effect has been estimated by Eurostat for the first wave of EU-SILC (2004 or 2005). This information is not yet available for DE and UK. Separate calculations are available for LU but are not strictly comparable. It can be considered that deviations of less than 10-15% from regulation targets are in the margin of tolerance implied by the degree of accuracy in the estimation of the design effect by Eurostat. Figures in bold are below this tolerance band.