Estimation of the joint distribution and linkage micro

Income, Consumption and Wealth in the EU
Estimation of the joint distribution and linkage micro-macro
Pierre Lamarche & Sigita Grundiza
Eurostat, European Commission
Abstract
The European Commission has stressed the need to bring social indicators on a par with macroeconomic indicators within the macroeconomic governance. Income, Consumption and Wealth (ICW) are three
key dimensions that determine the economic well-being of people or
material inequalities. ICW situation for the individual describes the
level and realization of economic opportunities. The data availability
of the joint the distributions of IWC and their dynamics are needed to
fill the data gaps for policy making; moreover, strong discrepancies between National Accounts aggregates and survey results underlines the
need for a better reconciliation of the ICW statistics.
Introduction
The Sen-Stiglitz-Fitoussi report [1] issued in 2009 has underlined the necessity of focusing on the household perspective,
while also considering income and consumption jointly with
wealth. Another important points of this report consists of recalling the importance of distribution and in particular considering the introduction of distributional elements in National Accounts.
• On the long run, increase of the integration of data collections,
in particular for surveys.
Data and methods
Joint distribution
The different available sources on ICW harmonized at the EU
level are exhibited in Table 1. The information on the joint distribution remain quite limited in the different surveys, due to
burden response. Harmonization efforts have also to be carried
out in order to achieve survey comparability: in particular, the
concepts used for collecting income in HBS and HFCS are not
always the same and do not always make a cross-country analysis possible.
Surveys Income Consumption Wealth
EU-SILC
HBS
HFCS
X
yes
yes
no
X
partial
partial
no
X
Table 1: European surveys collecting ICW data
Figure 1: Aggregate saving rate for sector S14-S15 compared with the proportion of households declaring having difficulties to make ends meet, 2013
If aggregates are able to shed light on well-being, they do not
tell the whole story. Classical poverty measures [2] already provide distributional information, focusing on income; the interplay between the three dimensions should help to better understand households’ vulnerability. From this viewpoint, an approach similar to the Household Account, involving in a dynamic way income, consumption, saving and ultimately wealth
should be considered.
It is also possible to describe the pattern of saving behaviors
according to different characteristics of the households (age –
linked with the life-cycle theory, as shown on Figure 7, income,
household type...).
In spring 2016, EU-Surveys were conducted among representatives of EU-SILC and HBS in order to gauge how far (or close)
we are from a full-integrated survey. The survey integration turn
out to remain fairly limited; very few countries conduct HBS and
EU-SILC on the same sample. Additionally, the pieces of information regarding the two other dimensions of ICW they collect
are usually limited.
Another solution consists of a modular approach; complementary modules could be considered to be included in the respective questionnaires of the different surveys. Such an approach will be tested in the coming year on EU-SILC with the
Over-Indebtedness, Consumption and Wealth module. Finally, a
”second-best” solution – statistical matching [3]– has been implemented on existing data.
Linkage between micro- and macro-data
The linkage between micro- and macro-statistics requires to first
map the concepts coming from the surveys and the ones used in
National Accounts: this work has already been done for the different dimensions of ICW [4]. We then analyze the coverage
rate of the mapped concepts, as in Figure 3.
Figure 2: National aggregate saving rates for sector S14-S15 compared with
saving rates as in surveys, combined EU-SILC and HBS, 2010
Figure 5: Median and aggregate saving rates in DE according to the age of
the reference person, 2010
The assessment of the quality of the entire process relies also
on ”natural” experiments, when income and consumption are
collected in the same survey, as it is the case for some of the
European countries. In this case, the matching is performed as if
consumption was not observed in SILC and the obtained result
is compared with the benchmark (as for instance it is done with
the EVS sample in DE, figure 6).
Figure 6: Distribution of saving rates in DE according to HBS and SILC,
2010
The dissemination of the obtained results comes necessarily
along with estimates of the uncertainty: the data are multiplyimputed and the Conditional Independance Assumption is relaxed using the Fréchet bounds as an estimation of the range of
plausible values.
Figure 7: Lorenz curve for income, consumption, wealth and savings in AT,
2010
Figure 3: Coverage rates for the different components of income, 2013
However, this approach poses the question of the comparability of estimates coming from the National Accounts and the ones
obtained with surveys. As shown on Figure 2, there is little if any
correlation between savings rates coming from the two sources.
Main Objectives
The work undertaken at Eurostat follows two separate leads that
are ultimately meant to converge into one single set of indicators:
• Estimation of the joint distribution of income, consumption
and wealth, performed at the micro-level.
• Linkage between micro- and macro-statistics. This linkage
entails not necessarily making all figures equal; careful attention has to be brought to conceptual and methodological
differences that exist beween micro- and macro-statistics.
The results show good coverages for wages and social benefits,
whereas it performs poorly for income from self-employment
and property income: breaking down these components of income according to the micro-data turns out to be more demanding in terms of assumption, since the reasons for such low coverage rate may be related to the sampling design as well as underreporting phenomenons.
Results
The purpose of the estimation of joint distribution is, inter alia,
to complement the already existing indicators on poverty; the
fused data makes it possible to compute statistics on the interplay between income and consumption in the EU-28 (Figure 4).
Eventually this project is aimed at producing distributional indicators at the meso-level, i.e. that are consistent with the National Accounts and that embed information from the microlevel. The last DGINS conference in September 2016 insisted
on the several following aspects:
Conclusions
• Statistical matching makes it possible to link the three ICW
dimensions at the cost of strong assumptions. Publishing the
results constitutes challenges as the data are experimental and
cannot be reduced to point estimates. It also underlines the
need for more survey integration.
• The effort for the micro-macro linkage has to be pursued, as
the consistency between the different data sources is essential.
References
[1] Stiglitz J, Sen A, Fitoussi JP, et al. The measurement of
economic performance and social progress revisited. Reflections and overview. Commission on the Measurement of
Economic Performance and Social Progress, Paris; 2009.
[2] Atkinson AB, Marlier E, Montaigne F, Reinstadler A. 5. In:
Atkinson AB, Marlier E, editors. Income poverty and income
inequality. Eurostat, European Commission; 2010. p. 101.
• Development of a harmonized ICW statistical framework, in
particular with regards to a multi-source approach.
[3] D’Orazio M, Di Zio M, Scanu M. Statistical matching: Theory and practice. John Wiley & Sons; 2006.
• Closer cooperation between survey statisticians and national
accountants, between statistical offices and academic world
and between the international organizations (Eurostat, OECD,
ECB).
[4] Mattonetti M. European household income by groups of
households. Eurostat Methodologies and Working Papers,
Publications Office of the European Union, Luxembourg;
2013.
Figure 4: Median and aggregate saving rates in the EU-28, 2010