The International Conference and a Methodological Workshop on EU-Silcaimed At

/ EUROPEAN COMMISSION
EUROSTAT
Directorate F: Social Statistics and Information Society /
Doc. Eurostat/F/07/SDG/01/03/EN
Point 2:Feedback from the helsinki Conference "Comparative EU Statistics on Income and Living Conditions: Issues and Challenges"
Strategic Development Group
Brussels, 2 February 2007-01-18
54 rue Joseph ii
Building J-54
Room 2/121
/ EUROPEAN COMMISSION
EUROSTAT
Directorate F: Social Statistics and Information Society
Unit F-3: Living conditions and social protection statistics /

Feedback from the Conference “Comparative EU Statistics on Income and Living Conditions:Issues and Challenges” (Helsinki, 6-8 November 2006)

1.Introduction

Eurostat and Statistic Finland organised an international conference and a methodological workshop on EU-SILC the Community Statistics on Income and Living Conditions in Helsinki on 6-8 November 2006. At the end of 2006, the collected data and derived indicators will cover for the first time all 25 European Union (EU) countries plus Norway and Iceland. EU-SILC, which replaces the EU-14 European Community Household Panel (ECHP), is becoming the EU reference source for income, poverty and social exclusion. On the one hand, EU-SILC raises some new issues regarding the EU common indicators already in use - especially with regard to the income concept(s) to be used for calculating the income-based indicators (as it follows closely -though not strictly- the recommendations of the Canberra Group and therefore provides detailed information on income components, in gross form). On the other hand, EU-SILC should allow to (better) address some policy areas that have not been (satisfactorily) covered to date.

The international conference and a methodological workshop on EU-SILCaimed at:

(a)Reviewing the EU-SILC instrument in all its dimensions: content, quality, methodology, implementation

(b)Identifying domains for improvements, taking into account the needs of “institutional” and academic users.

(c)Promoting the exchange of best practices in the various areas concerned.

(d)Improving the monitoring of the social situation in the EU, meaning that indicators, comparability among countries and comparability over time were of key importance throughout the Conference and the workshop.

The Conference and Methodological Workshop can therefore be seen as a concrete follow-up to the plea in favour of further methodological studies and visibility around the EU-SILC instrument, which was made in the independent Report presented at the Luxembourg Presidency Conference on “Taking forward the EU Social Inclusion Process” (Luxembourg, 13-14 June 2005)[1].

To ensure the high standard of the event, the scientific content of the conference was put under the responsibility of a scientific committee composed of Tony Atkinson (Oxford University, UK), Michel Glaude (EUROSTAT, European Commission), Markus Jäntti (Åbo Akademi University, Turku, Finland) and Eric Marlier (CEPS/INSTEAD Research Institute, Luxembourg).

The event brought together data producers (Eurostat and EU-SILC National Data Collection Units), (potential) institutional users (European Commission and Member States including members of the Indicators Sub-Group of the EU Social Protection Committee as well as OECD, WHO…), academic experts and the research community in general.

The event was split into an international conference and a methodological workshop

The international conference consisted of 5 different sessions:

• The introductory session aimed to review the EU-SILC instrument from an EU and national perspective: review of the EU-SILC framework, the preliminary outcomes and the challenges for the implementation.
• The second session aimed to review conceptually the EU-SILC instrument: discussing pros and cons of the EU-SILC approach in various dimensions of income measurement in the framework of policy assessment (Open Method of Coordination…) and academic research. In particular the focus was put first on the comparison of EU-SILC concepts with Canberra recommendations defining a framework for developing internationally comparable income statistics. Second, the focus was put on specific issues of comparability over time when structural changes are recorded and over heterogeneous regions.
• The third session aimed to focus on some specific aspects of income measurement and some complex income components: namely the extreme income and their impact on poverty and inequality indicators, the measurement of property income, the distributional impact of non cash income components such as imputed rent and company cars, the measurement of self-employment statistics, and the net gross conversion issues.
• The fourth sessions aimed to explore the non-income dimensions of EU-SILC and assessing their ability to fulfil the EU and country needs in terms of other EU indicators for social inclusion. Specific recommendations were expected to come from this session. Specific attention was put on material deprivation indicators, on heath variables, on gender issues (employment and income) and socio professional status in its relation with poverty (social classes). Children perspective could not be addressed as planed because of late withdrawal of one contributor.
• The fifth session aimed to review variouscomparability issues in particular in defining a framework to carry out studies on comparability for EU-SILC and borrowing from previous studies on ECHP to assess comparability of EU-SILC.
• The last session aimed to address the relevance of EU-SILC to be used for national and comparative EU monitoring of some key aspects of social protection and social inclusion. It collected main recommendations from the different sessions and identified the needs for monitoring and modelling purposes of institutional users represented by DG EMPL, OECD, ISG and academic users represented by two renowned experts and benchmarked these needs against the data currently available in EU-SILC. A concrete outcome of this session was expected to be a set of recommendations and suggestions for improving the EU-SILC instrument. A specific attention was put on the use of EU-SILC for micro simulation policy monitoring.

Key experts in the their respective fields, from both academic and official statistics worlds, were asked to prepare contributions supported by empirical results on the basis of the EU-SILC Users’ Data Base (UDB), so as to ensure the papers to be robust on both theoretical and empirical grounds. Speakers were expected to come up with concrete recommendations in their conclusions and quite often did so. Discussants were invited to discuss and “tie” papers together. The UDB was made available for free to all contributors who needed it for their interventions. All relevant background material on the project was also made available to the researchers and specific support was provided in case of specific enquiries.

The methodological workshop aimed at reflecting upon and exchanging good practices for the different aspects of the EU-SILC process. Its different phases from the design to the publication including data collection, editing imputation, estimation, quality reporting (accuracy measurement and coherence analysis) were reviewed systematically.

2.Users’ recommendations and needs: a review

The conference was a unique opportunity to collect opinions, recommendations and needs from (potential) key users about the EU-SILC instrument.

The main recommendations for the future of EU-SILC that were expressed by speakers, discussants, panel participants and/or the audience can be structured and summarised as follows:

According to the participants, EU-SILC project achievements are substantial. However there is a need to carry out evaluations and to communicate results of these in order to increase the confidence in the EU-SILC instrument to be used as the basis for cross-country comparisons and for the monitoring and evaluation of anti-poverty and social exclusion policies. The confidence is critical at national level where EU-SILC should get 100% acceptance and recognition.

The short term is to be devoted to consolidation (evaluation, streamlining and reporting). Changes in the content should occur parsimoniously and extension of the list of variables cannot be envisaged (i.e. if there is a need for a few new variables, these should be “compensated” by dropping some of the existing ones).

Coverage and content of EU-SILC

• EU-SILC is a most impressive statistical resource, but it should not be seen as a “miracle source” covering all aspects of income and living conditions. In order to guarantee data quality, choices need to be made as to which (micro-data) information is to be collected through EU-SILC and which should rather be collected through other instruments of the European Statistical System. Basically, EU-SILC primary aim is to collect data on income and living conditions with the necessary auxiliary information required to derive relevant classification variables for analysis.
• The relevance of current content of EU-SILC is unquestionable: policy making should be data driven and the redefinition of Lisbon strategy verging towards employment and growth makes it only more important to have data on social cohesion to see the redistribution impact of the measures upstream. The income and employment dimensions appeared to be of great value forinternational studies.
• The EU-SILC framework proved to follow closely the Canberra framework and is thus suited for international comparisons. Some improvements could however be recorded in the detailed definition of specific components like the definition of property income.

• Some important income components to be available from 2007 need to receive much attention in the next years. This includes in particular self-consumption and imputed rent.

• On the complex issue of imputed rent, there was agreement among conference participants on:
• the importance of collecting/ producing the required information; and thus also
• the urgent need to agree a clear set of common guidelines for calculating it in the context of EU-SILC and to carry out in-depth analyses of the robustness and comparability of the information.

Opinions were however split on the exact “status” to give to imputed rent once robust and reasonably comparable information would become available, even though there was a clear preference for not automatically considering it as part of the total household income. It should be up to individual data users to decide whether or not to include imputed rent in the total household income depending on the purpose of their analysis and indicators.

• The instrument could be adjusted to better meet emerging issues, in particular child poverty (add child specific items, more detailed demographic information) and the sharing of resources between household members. Deprivation items specific to children could be added. For child care, the children access dimension could be developed. The need for access to health care and mental health which has been negatively prioritised in favour of unmet needs for health care variables could be reassessed.
• Regarding the need related to annual modules, the themes of the modules should be more linked to primary variables. The modules are not appropriate to address new issues. As much as possible, the module process should not hamper the development of secondary indicators at national level
• EU-SILC excludes certain groups of the EU population. It is important to keep this limitation in mind when carrying out analysis on EU-SILC data, and to emphasise it when presenting EU-SILC results (see below: Presentation).

Income-based indicators and treatment of extreme values

• The sensitivity of some income indicators to extreme values can be quite important. Parametric-tail imputation seems to offer a good trade-off between robustness and information withholding. “Winsoring” (i.e. top and bottom coding) is appealing for its simplicity. Trimming is less recommendable.

Comparability of EU-SILC

• A key issue, that will require constant methodological monitoring, will always remain the assessment of the impact of using different sources and esp. register information versus interview information. The use of administrative data records is considered as a positive factor (following the example of Nordic countries) because it can resolve item and unit non-response problems and lead to more accurate data. However,administrative data are not exempt of problems because the administrative framework can lead to non comparability even between register countries.
• In this context, it is important to develop systematic cross-comparisons/ validations of EU-SILC results with other data sources: registers, National Accounts, ESSPROS data, other household surveys...
• There is a clear need for an increased level of standardisation of some aspects of the EU-SILC project (e.g. self-employed -including farmers- income; capital income; imputation; weighting…). While it is important that EU-SILC remain a flexible instrument in order to anchor it in each national statistical system, it is equally important that this flexibility be accompanied with clear and detailed guidelines to be strictly followed by countries so as to ensure a satisfactory level of comparability.
• More thoughts should be given about the links that can be made between statistics based on the old (European Community Household Panel – ECHP) and those based on the new (SILC – Community Statistics on Income and Living Conditions) instrument. Policy monitoring requires long time series which EU-SILC will not be able to offer before a few more years of data collection. Furthermore any significant forthcoming change in the methodology would need backward reconstruction of series.

• Non-monetary deprivation should receive closer attention in EU-SILC. The list of items aimed at covering it and the international comparability of these items need to be analysed.
• The quality of the translation of the questions/ items in the different EU languages has to be more carefully looked after (e.g. deprivation items, but also health questions…).

Timeliness

• Timeliness will remain a key requirement for institutional users such as the Commission and national ministries despite the fact that EU-SILC has met the original target. The timeliness needs to be further improved to better cope with the deadlines imposed by the Joint report on social protection and social inclusion to be available for the Spring European Council. However, the methodological workshop has demonstrated that EU-SILC process is complex because of the necessity to deal with different year processes at the same time. The level of standardisation achieved by countries at their 2nd or 3rd wave is quite high. Significant timeliness improvements will probably rely more on innovative solutions than streamlining.

Accuracy

• The reporting on the accuracy and reliability of the EU-SILC indicators used in the context of the OMC should be systematic. Standard errors should be made available and visible. On this issue, the methodological workshop showed that despite very encouraging results obtained by Eurostat for the first wave in order to benchmark national calculations, it is not yet clear which method could be used for routine production when we will have to cope with the complexity of combining several waves. The variance estimation in presence of imputation will remain a majorchallenge as well as how to communicate with end users on this issue.

Analysis

• Regional variation in prices is potentially important and needs to be considered as part of a sensitivity analysis. This refers to the adjustment of incomes by region for comparison with a given national standard; it does not refer to the use of separate standards that vary by region.

• It is worth exploring the potentiality of EU-SILC with respect to individual-based as opposed to household-based analysis; the same logic applies to the importance of analysing within-household distribution of income. This is one way in which to investigate the gender dimension (at present not sufficiently highlighted).
• The EU-wide approach (e.g. EU-wide at-risk-of-poverty line) should receive more attention provided that the micro comparability of data is satisfactory.

Accessibility and presentation of EU-SILC results

• The offer of statistics computed from EU-SILC should be diversified and should not concentrate on the only deprivation indicators. In general, statistically unreliable results should not be published for the sake of political credibility. The communication to the public should be enhanced at EU-level by the publication of a pocketbook oriented to the public.
• EU-SILC results (and in particular indicators) should be accompanied by some basic, descriptive statistics on the population not covered by EU-SILC. It is particularly true of the population living in institutions (e.g., when studying poverty among the elderly). It is also true of the homeless.
• Basic, descriptive statistics should also be provided on those groups that are under-represented in EU-SILC (e.g. certain groups of migrants, some ethnic minorities…) and ways of improving their coverage should be explored.
• We need to modernise the way in which we present the evidence, in order to allow more incisive analysis and increase the visibility of the findings. Some useful concrete tracks were proposed in some of the conference papers.
• Regarding the accessibility of EU-SILC micro data for research, participants clearly indicated that a reduction of price and the release of non critical constraints on micro data access will always have a positive impact of the degree of use of the data and the development of new indicators. The access to confidential micro-data, the remote on-line access as well as access for non EU bodies should be fostered.
• The value added of EU-SILC anonymised UDB can be increased if techniques like micro-simulations, micro data linking for imputation or with other institutional databases are permitted. In particular, taxation simulation requires information about incomes (and assets) available at the individual level and with more details then currently available in EU-SILC. The same holds for social benefits (and income sources in general) which should be made available at the most disaggregate level possible. Furthermore, housing costs variables should be provided in the detail in which national questionnaires collect the information. Separate income adjustment factors for non-response, for missing individual(s) and for missing values for responding individuals should be supplied.
• Finally, the importance of the contact with the scientific community and the users in general was stressed. There was a call for regular research conferences to be organised. For example, the next one could take place once EU-SILC longitudinal data are available. Ideally, yearly meetings (similar to the Helsinki one) including policy makers not only at EU but also national level would be useful to maintain EU-SILC in adequacy with emerging needs. The setting upof a users’ groupfor analytical research and for instance a working paper series would be very valuable. There is also a need for training interested users on how to use the EU-SILC dataset. The EuroPanel Users Network (EPUNet)[2] ) concept developed in the context of the ECHP will provide a good basis to start from.

Longitudinal dimension of EU-SILC