EUROSTAT
Directorate F: Social statistics /
Doc. Eurostat/F/12/DSS/01/2.1 EN
Item2.1
Modernisation of Social Statistics
meeting of the european directors of social statistics
luxembourg, 21 March 2012
BECHbuilding
MODERNISATION OF SOCIAL STATISTICS
1Executive summary
The Wiesbaden memorandum adopted by the ESSC in September 2011 is providing a clear mandate and guidance for the modernisation of social statistics. It is the purpose of this paper to provide an overview of the state of play as regards the key areas for development identified, i.e. the availability of reliable and up-to-date sampling frames, the re-use of existing data and in particular administrative data, and the streamlining of core social surveys.
Progress on the modernisation of social statistics can best be achieved in close partnership between MemberStates and Eurostat. Therefore communication structures are proposed that aim at encouraging the sharing of good practices as regards the modernisation of social statistics.
The Directors of Social Statistics are invited:
-to comment on the strands of actions proposed and to inform about their priorities as regards actions to be tackled at EU level
-to inform about related actions carried out or planned in their own countries
-to inform about which of the areas would be of priority interest to be addressed in the planned workshop
2Proposed amendments to Regulation (EC) 223/2009
The modernisation activities in Eurostat are embedded in a wider process of implementing the vision on the production method of EU statistics. In this context, the Proposal for a Regulation of the European Parliament and of the Council amending Regulation (EC) 223/2009 on European statistics contains elements of particular relevance for social statistics.
The proposal for the new version of Article 5 (1) aims at strengthening the coordinating role of NSIs within the national system of developing, producing and disseminating European statistics. Given the large share of social statistics under the responsibility of other actors (e.g. ministries of the interior, ministries of health, ministries of education), this article would support a streamlining of statistical activities shared between various actors.[1]
Equally important is the proposal of a new article aiming at facilitating the usability of administrative records for statistical purposes. Firstly, this article foresees that NSIs shall have the right to timely and freely access and use all administrative records and to integrate these administrative records with statistics, to the extent necessary for the development, production and disseminationof European statistics. Secondly, this article foresees that the NSIs shall be consulted on and involved in the initial design, subsequent development and discontinuation of administrative records built up and maintained by other bodies in order to facilitate the further use of these records for statistical purposes. They shall have the right to coordinate standardisation activities concerning administrative records relevant for statistical data production.[2] If adopted, this article would greatly enhance the usability of administrative records for statistical purposes, particularly in those Member States in which the usage is current severely limited by law.
3Activities targeting the key areas for development
By adopting the Wiesbaden memorandum on a new conceptual design for household and social statistics on 28.9.2011 (see annex), the ESSC defined the key areas for development as being reliable and up-to-date sampling frames, the re-use of existing data and in particular administrative data, and the streamlining of core social surveys.
The following description aims at presenting all ongoing and currently planned actions aimed at implementing these key areas and highlighting the links between the various activities.
3.1Streamlining of core social surveys
3.1.1. General approach
The first proposal aiming at a rationalisation of social surveys ("the pillar approach") was presented to the DSS in March 2011. While the aim of rationalising the current situation was welcome, several Member States considered the proposal as incompatible with national plans and with the need to take into account national demands. As a result of the discussions, an Expert Group was created with the mandate to address the issue of survey integration.
The first meeting of the Expert Group, which took place in June 2011 issued two recommendations:
-In order to overcome the drawbacks linked to the size of the building blocks envisaged in the pillar approach, the Expert Group recommended in the first stage to modularise the social surveys, i.e. breaking them down into smaller building blocks[3].
- In order to evaluate the realistic options for integrating these building blocks in the second stage, the Expert Group recommended investigating the feasibility of a number of issues such as sub-sampling as well as practicalities such as case management.[4]
These two recommendations were endorsed by the DSS in September 2011.
Work as regards the modularisation is currently under way. It aims at splitting existing surveys into smaller building blocks that do not only focus on a specific topic (i.e. the traditional meaning of modularisation) but at the same time are coherent from a survey methodology point of view. The Expert Group had recommended evaluating frequency and accuracy requirements; additional aspects likely to be considered include statistical unit, reference period and target population. The Expert Group will be consulted before the DSS meeting on this matter via written consultation. Once proposals for the modularisation for a first set of surveys will have been established, the Expert Group will be consulted in a meeting that is currently scheduled to take place in the week 11 – 15 June 2012.
The ultimate goal of modularisation is the creation of building blocks that can be combined in a more optimised way than is currently the case both as regards the survey characteristics and as regards relevant linkages at micro-level. Current social surveys have been developed taking into account the user needs as regards specific topics. The methodology of each survey can be considered as a "one size fits all" for all the topics dealt with. By taking into account all core social surveys, more efficient designs could become possible, e.g. by limiting the panel component to those variables for which this component is essential. At the same time, efforts will be devoted to creating designs that allow more relevant analytical linkages between topics.
As recognised by the Expert Group, an alternative combination of the building blocks raises a number of methodological and feasibility issues. While work on the first stage (modularisation) is not yet sufficiently advanced to clearly identify these issues, it is likely that sampling design and calibration will be relevant issues. Therefore, methodological work has started to address these issues. Practicalities such as the suitability of existing case management systems to such surveys architecture will be addressed once work on the redesign is more advanced.
The goal is to find an architecture for the European social surveys which is able to combine a common design at European level with flexibility of implementation at national level. The work will be considered a success, if it improves the current situation as regards the main objectives of the streamlining process[5] and if the benefits achieved outweigh the costs incurred.
3.1.2. Review of SILC and LFS
The previously decided processes of reviewing SILC and LFS interact closely with and form part of the streamlining process. Firstly, the reviews address design issues such as precision requirements or the relevance of the longitudinal component. Secondly, the reviews address user expectations as regards the survey contents and potential changes will need to be anticipated in the course of the streamlining process.
The exercise of reviewing SILC was launched in 2011 taking stock of the expectations that had been expressed so far by users and producers.
Concerning the design of SILC, these expectations relate in particular to the timeliness of the estimates, the representativeness of the national samples (in particular in the context of improved regional coverage), the simplification and harmonisation of sampling designs, the longitudinal component, as well as the role and periodicity of annual ad-hoc modules (versus the implementation of regular rolling modules).
Concerning the content of SILC, these expectations concern the central role SILC should play in measuring quality of life[6], the need for additional information on wealth and consumption in order to better address issues related to poverty and social exclusion, the willingness to improve the estimation at household level of social transfers in kind, the interest in incorporating some variables of former modules as well as the need for a critical evaluation of the necessity of all primary variables currently collected.
This list of expectations shows contradictions and competing priorities. At the occasion of the first task force meeting, priority has been given to five themes that are now being addressed in five sub-groups on the following issues:
-Contents of SILC, with the aim of modularising SILC
-Modes of data collection, including use of administrative data and mixed mode data collection
- Improvement of the timelines, with the development of scenarios based on flash estimates, improved production chain, model based estimations
-Improvement of the methodology, in particular modernising quality reports, precision requirements and regionalisation
- Longitudinal component, with the production of several scenarios
Similarly, a comprehensive review of the LFS was launched in 2011 based on experience from ten years of operation on the basis of Council Regulation 577/98. The issues to be addressed can be grouped under two broad themes: (i) maintaining and increasing the relevance of the survey, (ii) modernisation.
Amongst the items addressed are the review of variables, harmonising the measurement of the labour status, making the panel structure compulsory, clarifying the precision requirements, improving the measurement of main status, improving timeliness and providing for multi-mode design.
3.2Multimode data collection
The use of mixed modes in social surveys is current practice in many Member States. In that context, most have already made the move from PAPI to computer-assisted modes (CAPI, CATI). Recently, some national statistical institutes have started considering web-based data collection (CAWI) for surveys among persons and households.
This provides the unique opportunity to coordinate and harmonise the introduction of CAWI across the ESS from the onset. Against this backdrop and with a view to the modernisation of social statistics the ESSC decided last year to launch a dedicated ESSnet on data collection for social surveys using multiple modes in 2012. This ESSnet should (i) review available tools for web-based data collection and work towards standardised solutions that can be shared, and (ii) establish a knowledge database on the methodological and organisational aspects of mixed-mode design.
Work to analyse challenges and pitfalls will focus on the LFS. However, the aim is to deliver results that can be used for all social surveys.
The call for proposals will be launched end of March/beginning of April.
3.3 Statistical data matching
One of the criteria of success for the streamlining of social surveys is an increased analytical potential of the data collected through an integrated system. Supplementary to a traditional integration in the form of sub-samples or follow-up surveys, Eurostat is currently also investigating the potential of statistical data matching to increase the analytical potential of available datasets.
A pilot project has started in 2011, adopting an ex-post approach, i.e. applying statistical data matching to existing micro data sets. Current areas of work comprise case studies on producing regional estimates for poverty and health indicators based on SILC, EHIS and LFS, on producing joint information on quality of life indicators based on SILC, EQLS (European Quality of life Survey) and ESS (European Social Survey) and on producing integrated statistics on income, consumption and wealth based on SILC, HBS and HFCS (Household Finance and Consumption Survey of the ECB).[7]
In the course of the work, the need for a better de facto harmonisation of current surveys became apparent, including the need for an improved coherence of concepts and core social variables. Moreover, inherent limitations of statistical matching techniques could be tackled through a better integration of surveys and an improved potential for modelling. Therefore, the focus has been enlarged to encompass also an ex-ante perspective and to define the conditions that would need to be met in order to ensure that statistical data matching will produce good results.
3.4Harmonisation of variables
A key aspect for the streamlining of social surveys and for the related activity of statistical data matching is the de facto harmonisation of variables aimed at measuring the same or similar concepts across surveys. This aspect has already been addressed in the past in the work on the 'core social variables'.
It is planned to build upon this work and to develop it further along three strands. First or all, the de facto harmonisation of the 'core social variables' needs to be improved in order to ensure that harmonisation is not impacted by the application of specific filters or diverging methodological guidelines. Secondly, it might be worth investigating the benefits of a harmonisation of more detailed breakdowns of the 'core social variables'. Thirdly, the modularisation of social surveys is likely to reveal further sets of variables that exist in several surveys and would benefit from harmonisation.
The ambition could evolve from aiming at having a limited set of compulsory core variables to a more extended set of harmonised variables that can be used in surveys as appropriate.
3.5 Re-use of existing data for statistical purposes
A second key area for development identified in the Wiesbaden memorandum concerns the re-use of existing data and in particular administrative data. An increased use of administrative data for statistical purposes clearly has the potential of reducing burden on respondents as well as reducing cost for statistical agencies, whenever it can serve as a substitute to data collection through statistical surveys. As mentioned in the beginning, the proposed amendment of Regulation (EC) 223/2009 carries the potential of improving considerably the usability for statistical purposes of administrative data.
Beyond the legislative barriers to the use of administrative data for statistical purposes, the use of such data is often limited by lack of quality. Accuracy and reliability might be impacted in particular when the data elements of interest are not the core elements of interest to the owner of the data source or when the coverage is incomplete. Timeliness and punctuality might be impacted when the timing of the updates is inconsistent with the requirements from a statistical point of view. Coherence and comparability might be impacted when the definitions and concepts used are not in line with those required for statistical purposes.
In order to improve the de facto (and not only de jure) usability of administrative data for statistical purposes, a project will be started that aims at developing a commonly accepted quality framework to guide the choice of data from administrative data sources and at reviewing statistical techniques that carry the potential of successfully transforming raw data from administrative data sources to high-quality statistics.
Another strand of re-using existing data concerns the possibilities of re-using the digital footprint for statistical purposes, in particular the developments of using the internet as a data source for official statistics. While most applications relate to price statistics, information society statistics or mobility statistics, some applications are being developed in the area of social statistics, in particular job vacancy statistics. Eurostat is closely monitoring these developments.
3.6 Development of reliable and up-to-date sampling frames
A third key area for development identified in the Wiesbaden memorandum concerns the reliability of sampling frames for social statistics. Given the size of the task, Eurostat is currently preparing an ex-ante evaluation for a project aiming at increasing the availability of reliable and up-to-date sampling frames for European social statistics.
This project is linked to the future of the census and Eurostat will share its initial views with Member States in this year's Seminar of the Conference of European Statisticians devoted to the topic "Challenges for future population and housing censuses" to be held on 6 June 2012.
4Governance and communication structure as regards Member States
The DSS is considered as the formal group for steering the modernisation of social statistics and its role has been officially recognised in the Wiesbaden memorandum.[8] Given the multitude of producers of official statistics, it is important that the members of the DSS ensure good communication and coordination with all national producers.
It is proposed to put in place specific Expert Groups with time-limited mandates in order to develop specific strands of the modernisation. Currently such a group already exists for the area of streamlining social surveys with representatives of Germany, Estonia, Ireland, Spain, France, Italy, the Netherlands, Austria, Finland, Sweden, United Kingdom and Norway. Other groups might be created according as appropriate.
Subject area working groups will be regularly informed about the state of play on modernisation of social statistics. Specific agenda items will be introduced where appropriate, e.g. an item on streamlining surveys in those groups dealing with surveys.
Bilateral meetings with teams in NSIs that are either responsible for the modernisation of social statistics in general or for specific areas of modernisation would prove helpful in sharing experiences and establishing a network.
In order to increase the network effect while keeping burden on participants limited, a CIRCABC interest group has been created and discussion forums will be opened. Additionally, a dedicated two-day workshop on modernisation of social statistics is planned to take place in autumn for the sharing of experiences between the Member States, Eurostat and other international bodies such as the ECB, the OECD and UNECE.
Annex
Wiesbaden memorandum
New conceptual design for household and social statistics
The DGINS
CONSIDERING
1. The EU treaty (art 3): well being, free movement, full employment, social progress, equality between women and men, combating social exclusion and discrimination being fundamental objectives