/ EUROPEAN COMMISSION
STATISTICAL OFFICE OF THE EUROPEAN COMMUNITIES
Directorate E: Social and regional statistics and geographical information system /

DOC. E0/HARM/23/2000

Working Group

Harmonisation of Social Statistics

2-3 May 2000

Luxembourg, BECH Building

Point 3 of the agenda

A framework for harmonisation of social statistics, draft

April 2000

A framework for harmonisation:

Key social indicators, core variables and a framework for the joint use of administrative sources, register and survey data.

Pieter Everaers,

Directorate E, Social and regional Statistics

Eurostat

Luxembourg

Table of content

  1. Introduction
  2. Background

-Harmonisation of social statistics

-Integration

-Social indicators

  1. A set of key social indicators
  2. Core variables

- A short description of the provisional list of core variables

  1. An integrated European data base

References

Annex

The action plan on harmonisation of a set of core variables

1.Introduction

The IVth session of the Mondorf seminar on the future of Social Statistics (March 1998) resulted in the approval of a plan for the harmonisation of a set of core variables on persons and households, as a step to a more elaborate integration of surveys, administrative data and registers. (ref.1). The Action Plan for the harmonisation of a set of core variables (ref.2) is mainly based on the achievements of this seminar.

The objectives of for the harmonisationAction Plan as formulated in 1998 are

1)the construction of a set of harmonised core variables,

2)to integrate data from surveys, administrative sources and registers.

The Action Plan (see Annex 1) describes the steps to be taken to reach these objectives. It proposes the selection of core variables to be related to a set of Eurostat Key Social Indicators. With the relation to the Key Indicators the harmonisation work is also related to the developments in European Social Policy as the indicators are considered to be the main monitoring instruments for the convergence in the social field of the EU Member States.

The main products of the harmonisation Action Plan are the following.

  1. A set of Eurostat key social indicators and a description of how they have to be calculated;
  2. A set of harmonised core variables on persons and households, the recom-mended concepts, definitions and measurement rules;
  3. A meta information system on sources for core variables for social statistics in Europe;
  4. An overview on well developed and applicable linking methodology for combining data from different sources,
  5. Building blocks for the integration of data from surveys, administrative sources and register information.

The aim of this paper is twofold. It aims to give an overview of the background of the Eurostat harmonisation project as formulated in the Action Plan. Secondly it aims to link the harmonisation work to developments in related fields such as the combined use of administrative data and registers and surveys, the developments in input harmonised surveys such as the European Community Household Panel (ECHP) and the target structure of core variables in existing harmonised surveys.

In section 2 the background of the harmonisation program is described. In section 3 the selection of the Key indicators is given (main product A). The relation between the key indicators and the core variables is included in Annex 2. The work done on the selection of a set of core variables, key indicators and the description and publication of the list of variables and key indicators are the main steps of the Action Plan achieved in 1999 and 2000. The list of core variables (main product B), and descriptions of their content are given in section 4. In section 5 some remarks are made on the future structure of a set of linked statistical sources for social statistics. This section refers to – recently (1999) started – work on the integration of data from different sources and the linking methodology (Main product D and E). Pilot studies in this field (e.g. related to the future of the ECHP) are underway. With respect to Main product C, Meta information systems, Eurostat is planning to launch (during 2000) a call for tender for the development of a documentation system as a helpful tool for the harmonisation work.

2.Background

Harmonisation of Social Statistics

The Maastricht Treaty and later the Amsterdam Treaty clearly put the social policy on the agenda of the European Commission. Since the mid nineties, via several special summits, the social domains have grown in importance. For example, the Social Action Plan (ref. 3) and the Employment Guidelines (ref. 4) describe a need for comparable statistical information. This information allows (in great detail) to monitor the developments (convergence or divergence) in European social matters.

In the early nineties, Eurostat, like several NSI’s, has set up projects to reach more comparability of data in the field of social economic developments. Examples are the regulation of the Labour Force Survey (LFS) questionnaire and the harmonisation of the Household Budget Survey (HBS). Since 1994 input harmonisation is implemented in the European Community Household Survey (ECHP) (see also Grais, ref. 5 for an overview of harmonisation efforts).

The advantages of sources like the ECHP are obvious. However, considering user demands, these sources have disadvantages. With respect to detailed (regional) information on specific domains they are considered inferior to many Member States' specific data sources. The European harmonised surveys enable EU wide com-parative analysis on specific domains. However, a combined analysis of background characteristics and varying theme specific variables is -because of the limited scope of the harmonised surveys and the sample size and sample design - hardly possible.

This issue of differences between 'Eurostat's harmonised sources' and the 'best national sources' has been discussed in large detail in the second session of the Mondorf seminar (1996). The decreases in budgets as well as the increase in the use of administrative data and registers led several Member States to prefer alternative methods to the launching of a new large European survey. These methods, to reach the level of comparable results, are based on the concept of integration. Since 1996 the integration of data from different sources has been high on the agenda of Eurostat.

Integration

Integration (of social statistics) is defined as the combination of data from different sources of social statistics in order to obtain information that is superior to the information provided by the source data as such.

The concept of integration applies to the combination of data derived from different sources (output harmonisation, using macro data) as well as the combination of sources (input harmonisation, integration of surveys). The first application uses statistics for the same variable, derived from different sources, as linking element of other statistics from these sources. The second application concentrates on the combination of questionnaires or questionnaires and administrative/register information of individual persons or households.

The first step in integration is harmonisation of concepts and definitions. Harmonisation work starts with selecting the main sources (for specific variables) and comparing the variables (definitions, concepts, and classifications) as well as the quality of the outcomes. By co-ordinating the concepts, procedures etc (e.g. via selecting the best practices) the practical steps in the harmonisation work start.

Social indicators

Based on a Commission demand dating from 1993, Eurostat launched in 1995-1996 a project to develop a set of social indicators covering all domains of social statistics. The method to derive the indicators was discussed at the second Siena group and second session of the Mondorf seminar (1996); alternatives ranged from a general social survey to a top down approach using existing information (ref. 6). This discussion influenced the start of the current harmonisation project by the proposal to organise the harmonisation work, with the aim to develop a consistent set of indicators, in a top down approach. A set of social indicators should be based on a set of variables already available in Member States sources (administrative sources, registers or surveys) or in the already harmonised surveys. (ref. 7) The discussion focused not only on data itself but, as an important source for quality and comparability, also on the differences in concepts and definitions used in national versus international statistics and in the consequences of differences in measurement methods.

The Eurostat project on social indicators was related to the need for European social reporting, following the social policy intentions as conceived in the Maastricht treaty. The direct policy relevance is a strong point of the indicator approach. As policy agendas can change frequently the demand for specific indicators to monitor developments also changes. This causes problems with respect to continuity of the indicators and the flexibility of the basic data gathering. This was seen as an argument in favour of using existing statistical information of statistical systems already composed for other questions.

The indicators published in the Pocketbook on Living Conditions 1998 (ref. 8) and in the 1999 issue were an implicit combination of informative and problem oriented indicators: the formulated goals of the social indicators project point in the direction of a general description of the quality of life in the Member States of the EU. The result of the project was considered a first stage on the longer road towards indicator development so that the extended list of indicators would lead to the selection of a more limited set of isolated, small spectrum macro indicators.

3.Key social indicators

There is a clear tendency towards a more explicit formulation of policy aims by the Commission departments as well as other policy institutes. The Social Action Programme 1998-2000 and the Employment Guidelines are the best examples of this tendency. Well-selected indicators are seen as useful tools for informing policy makers on the main processes in their field of responsibility. The recently developed set of Eurostat Key Social Indicators, is limited to the fields with a large policy interest. However, whenever a field becomes of interest new key indicators should be developed and added to the existing set.

A total of 15 indicators has been identified to monitor the most important social policy concerns. The selected Key Indicators are described in detail in the Report on the Social Situation in Europe 2000 (ref. 11).

4. Core variables on persons and households

The key indicators cover several domains of social statistics. They also differ with respect to the complexity and - as a consequence - the source on which they are based. The key indicators, their numerator and de-numerator, can be entangled (parsed) to the level of the original statistical sources or (when available) to the results of integrated systems. The procedure allows the core variables to be delineated. In Annex 2, the relation between the key indicators and core variables is given. As the key indicators have to be comparable between Member States, the core variables also have to be available at a sufficiently high level to allow the (harmonised) calculation of the key indicators which are comparable between Member States.

Below the list of core variables is given. The variables mentioned as core variables are directly related to the key indicators. The related variables were proposed earlier to be part of the core variables. They are not part of the key indicators presently. For the variables a recommendation is made on the concept and definition and their classification is given in brackets.

Core variableRelated variables

Place of usual residence (Census) Region (NUTS)

Locality (Census)

Sex (Census)

Age (Census)

Private household (Eurostat recom.)Legal marital status (Census)

De facto marital status (Census)

Housing unit (Census)

Household relationship (***)

Educational attainment (ISCED 97)

Labour status (ILO)

Economic activity (NACE)Occupation (ISCO 88 COM)

Hours worked (Census)

Disposable income (Eurostat)Status in employment (UN)

Main activity status (***)

Health status (***)

Current education/training (***)

Country of birth (Census)

Country of citizenship (Census)

Tenure status of household (Census)

Number of rooms (Census)

Socio economic status (***)

5. An integrated European data base

As indicated at the end of section 3, the result of the social indicator project was marked as a point for further work. Especially the international comparable dataset needed had to be elaborated. It was argued that the multidimensional nature of policy systems requires statistical information systems by means of which the interrelationships between policy objectives can be mapped out (ref. 9). In the theme description of the Siena Group Meeting Maastricht 2000 (ref. 10) the principles of the integration of data from different sources and aim and characteristics of statistical information systems are discussed in detail. It is clear that at national level the approach of integration into a statistical system is more accessible than on the European level.

Currently many sources are available for statistical information on social issues. Arguments for more harmonisation can be based on the fact that these sources differ with respect to content but also with respect to sample, questions, measurement errors and non-measurement errors. As a consequence, results on main characteristics will differ between these sources. The number of sources in use for producing European comparable analysis is limited; however, differences in outcomes from these sources will confuse the users. The aim of the harmonisation project is to avoid this confusion and to provide authoritative statistical information.

Another result of the large amount of different sources is the fragmentation. Specific issues are measured via one survey but not included in others. An analysis of the relation between this specific issue and variables available in other sources is not possible. One of the main starting points for the discussion on harmonisation was the efficient use of existing sources.

The availability of harmonised core variables in European data sources for social statistics allows the development of an integrated European database. This integration is built on the comparability of variables and classifications of survey data and administrative data and registers. There are numerous advantages of such an integrative approach. Conflicting information can be avoided, validity of separate sources can be tested, new expensive surveys can be avoided, and relations between different domains can be exploited.

The ECHP (input harmonised), the LFS (target harmonised) as the HBS (long history in harmonisation) can be characterised as the best sources for showing differences in the social and economic aspects of the Member States, as they are controlling many methodological issues of the surveys. Member States’ best national sources make use of the very specific characteristics of a country, they are not built for international comparisons, but to give the best possible (level of detail, reliability, precision) statistical information on (MS specific) issues.

The use of data from harmonised Eurostat sources (LFS, ECHP, HBS, and TUS) and from best national sources is a combination of maximum comparability and maximum level of detail and precision. This combination is an effective method to develop an integrative system with a high level of comparability between MS and precision within MS. A challenging part of the harmonisation project is the construction of this European integrated data set based on harmonised sources and best national sources. The integration process is mainly based on categorical matching. The categories are close to the individual level. The result of this integration process is not a micro data set, but a (or a combination of several) giant multi dimensional table, the dimensions being the core variables. By bringing in thresholds for the minimum number of cases per cell, this structure avoids the well-known confidentiality problems and the problem of ecological fallacy, e.g. the misuse of spurious correlation at the micro level.

The core variables should meet some criteria to function as a linking pin. To assure a high quality of the data, non-response on these variables should be low or missing, the questions based on the variables should be easy to answer, especially for measuring facts.

In this structure the harmonised Eurostat surveys play an important role in the harmonisation program. By using or changing to the use of the harmonised variables they promote the use of the harmonised variables and classifications. They also allow large scale pilots and research and development work.

The ECHP as well as the LFS are yearly surveys, the LFS is in many MS based on continuous surveys. These surveys do hardly change on a year to year basis in methodology, sample size, etc. They are very well suited to develop time series. This stability is, next to the comparability, an important advantage. The methodologies of an integration system will not change much when the reference survey (the linking pin) does not change. This implies that fundamental research on development only has to be done once. Based on this characteristic a high grade of consistency in time is also guaranteed.

An important element in the proposed structure of an integrated set of social data, based on the harmonised set of core variables, is the availability of at least one input harmonised (yearly) survey in which all the harmonised variables are included. To enable a maximum level of calibration over time and over Member States this has to be a stable instrument, organised independently of domains.

References

  1. Conclusions of the fourth Mondorf seminar on the Future of Social Statistics: social statistics and quality. Eurostat, Luxembourg 1998.
  2. Everaers, P.C.J. Action Plan on the Harmonisation of a set of core variables on persons and households. Eurostat, Luxembourg, October 1998.
  3. Social Action Plan 1998-2000. European Commission, Brussels, April 1998.
  4. Employment Guidelines. European Commission, Brussels, 1998.
  5. Grais, B. Statistical Harmonisation and Quality; Paper presented at Mondorf IV seminar, Luxembourg, March 1998.
  6. New Challenges for Social Statistics: guidelines and strategies (report of the second Mondorf seminar). Eurostat, Luxembourg, March 1995.
  7. Report of the third Mondorf seminar on the future of Social Statistics: Use of administrative registers and dissemination strategies. Eurostat, Luxembourg, October 1996
  8. Living Conditions in Europe. Pocketbook on Social Indicators. Eurostat, Luxembourg, 1998.
  9. Social Indicators and social reporting. Results of the inventory exercise in the EU and EFTA countries, Eurostat, Luxembourg, September, 1996.
  10. Accounting in social statistics and indicators for social development. Theme

description, Siena Group meeting Maastricht 2000, Statistics Netherlands, 1999.