Brian Nolan, Ive Marx and Wiemer Salverda
This arises from two sources: some deliberate changes in the way income is defined and measured, and the
shift from an “input-harmonised” to an “output-harmonised” data collection framework. Changes in the income
concept in EU-SILC compared with the ECHP include the following:
• EU-SILC obtains estimated values for imputed rent and the private use of a company car from 2007;
• Regular inter-households transfers to other households are deducted from the income of the donor
household in EU-SILC, whereas in the ECHP amounts received from this source were included but
payments were not deducted.
• Lump-sum tax adjustments are deducted from/ included in household income in EU-SILC but were
not in the ECHP.
• Negative incomes for the self-employed were put at zero in the ECHP but can be reported in EU-SILC.
The more fundamental change was from the harmonized questionnaire employed across the countries par-
ticipating in the ECHP to the quite different data-gathering framework being employed with EU-SILC, as well
as the shift from a wholly longitudinal ECHP design subject to serious attrition over time to a much more limited
longitudinal element in EU-SILC. In the latter, a set of key target variables is specified, and individual countries
have scope to decide how best to obtain that information from one or more surveys and/or administrative sources.
The potential impact on measured levels of inequality across countries, and the implications for those measures
over time linking those produced from ECHP and EU-SILC, has to be taken seriously in employing these key data
sources for comparative analysis.
While LIS, OECD and ECHP/EU-SILC provide for a variety of indicators of income inequality and poverty,
they do not provide a continuous annual time-series over a substantial number of years for a wide range of coun-
tries. Investigations requiring such data generally draw on the inequality dataset originally compiled by Deininger
and Squire (1996) and subsequently built on and adapted by others. This is confined to the Gini coefficient, for
which Deininger and Squire brought together observations from a wide range of sources for developed and devel-
oping countries in a dataset subsequently used extensively in cross-national studies. Unfortunately, as Deininger
and Squire themselves pointed out, the observations are rarely comparable across countries or even over time
within a single country because many are based on different income definitions (e.g., gross or net) and different
reference units (e.g., households or persons). Building on Deiniger and Squire, the World Income Inequality Data-
base (WIID) created by the World Institute for Development Economics Research of the United Nations Univer-
sity (UNU-WIDER 2008) incorporates data from additional sources and provides the most comprehensive set of
Page • 24