Comparable Indicators of Inequality Across Countries
it possible to examine how their circumstances changed and, in particular, to measure the persistence of low in-
come and poverty. This comes at a cost: attrition - losing respondents from one survey to the next - is a general
problem with longitudinal surveys, and was substantial in some countries over the life of the ECHP. The impact on
the composition of the samples generally seems to have been limited, but it remains a consideration in longitudinal
analysis.13 Data from the ECHP have been made available to researchers in the form of a USER Data Base or UDB,
as described in Eurostat (2001c, d).
The ECHP was discontinued in 2001 (or 2002 in a few countries), and was replaced by the European Statistics
on Income and Living Conditions (EU-SILC) which is currently the EU reference source for indicators of poverty
and social inclusion. EU-SILC is coordinated by Eurostat and covers all 27 Member States of the EU (plus some
others such as Iceland, Norway and Turkey). It was launched in 2003 in six Member States, and by 2005 was oper-
ating in all of the then EU-25 countries, subsequently extending to Bulgaria and Romania when they joined the EU
in 2007. EU-SILC takes a very different approach to the ECHP: rather than a common design and questionnaire, it
is based on a common framework for the production of specified statistics on income and living conditions. Mem-
ber States have considerable autonomy in the areas of sampling design, questionnaire editing and data compilation,
but the “target” variables required are tightly defined in the framework Regulation, which also sets out common
guidelines and procedures, common concepts (such as what constitutes a household and how income components
are defined) and classifications aimed at maximising comparability of the information produced. Both household
surveys and administrative registers can be used to produce the data, provided they are “linkable” at the micro-
level; while there is a longitudinal element, it generally involves following an individual only for four years, and
the cross-sectional element can be separate from the longitudinal one.14
Considerable effort is being invested by Eurostat and national statistics offices into assessing the quality of
the data produced by EU-SILC, with each country being obliged to provide a detailed annual quality assessment
report, and Eurostat producing an overall assessment of quality each year focused on accuracy, timeliness and
comparability, and covering areas such as sample design, sampling and non-sampling errors, mode of data collec-
tion and imputation procedures (see for example Eurostat, 2010).15 The switch from the ECHP to EU-SILC as the
main source of data and indicators on income and living conditions in the EU was aimed at improving the robust-
ness, comparability and timeliness of the data, but is a potentially significant source of discontinuity between the
two time series, and also needs to be taken into account in cross-sectional analysis.16
13 See Eurostat (2002), Watson (2003), Behr, Bellgardt and Rendtel (2005).
14 For general descriptions of EU-SILC and discussion of its key features see Clemenceau and Museux (2007), Marlier et al (2007), Wolff,
Montaigne and Gonzalez (2010).
15 For valuable discussions of accuracy in EU-SILC see also Wolff, Montaigne and Gonzalez (2010) and Verma and Betti (2010).
16 On which see Eurostat (2005), Marlier et al (2007), Atkinson, Marlier, Montaigne and Reinstadler (2010).
Page • 23