Page 108 - Invited Paper Session (IPS) - Volume 1

IPS102 Sigita G. et al.
observations. If there are less than 3 donor observations for each recipient, the
backward selection of matching variables is reiterated reducing the number of
variables. The whole matching process is replicated 100 times. The last step
then consists of re-calibrating the EU-SILC weights in the matched data set
adjusting them to a number of consumption margins. This step is also
repeated 100 times.
To join wealth data from HFCS to the matched income-consumption data
set, data are stratified according to three matching variables: Household type,
food consumption quintile and tenure status. Within each of these strata, the
two data sets are ranked according to gross income data. Then, for each
receiving observation the closest donor observation of the cumulative
distribution function is selected. Again, the process is repeated 100 times.

3. Results
3.1. Micro-Macro links for income and consumption
Figure 1 shows the data gaps for household disposable income (HGDI) for
the EU Member States and EFTA countries. In 2015, the HGDI average data
gap for the EU-28 between EU-SILC and National Accounts was 27%. In 2015,
for Austria the data gap for HGDI was 20%, corresponding to 80% coverage
rate for EU-SILC and national accounts data. Figure 1 shows the contribution
of each income component to the HGDI data gap. The largest contributors to
the data gaps on average were: operating surplus (9 percentage points),
property income (8 pp) and self-employed income (8 pp). Operating surplus
is not part of HGDI definition in EU-SILC. Data suggest that the wealthiest part
of population is not very well covered in EU-SILC. In addition, both property
and self-employed income have medium/low conceptual consistency in the
both data sources. It should be noted that even if there was no aggregated
HGDI data gap, there could be data gaps for income components that offset
each other. Denmark and Norway provide such examples.

97 | I S I W S C 2 0 1 9