Wie Eltern ihr Erwerbsleben arrangieren und das

Download Report

Transcript Wie Eltern ihr Erwerbsleben arrangieren und das

EU-SILC from a Research Perspective
Heike Wirth & Christof Wolf
Topics covered
•
•
•
•
2
Strengths of EU-SILC
Flexible implementation of EU-SILC
Selected issues regarding data comparability
Opportunities for longitudinal analysis with EU-SILC
Strengths of EU-SILC
•
•
•
•
•
•
3
Coverage of countries
Coverage of topics
(Private) Household survey
Cross-sectional and longitudinal data
Good and improving data documentation
Access for research purposes free of charge
(but more demanding under new regulation)
Flexibility
4
Flexible implementation of EU-SILC
• EU-SILC is based on a common framework
•

guidelines: concepts, definitions, classifications, procedures
Target variables, i.e. ex ante harmonization
• Within this framework high flexibility regarding data
generation
•
•
Accommodates the national conditions and needs (+)
Potential to limit cross-national comparability (–)
• While the input side might be diverse, the output side is
harmonized (standardized microdata set)

5
i.e. problems of data comparability are not directly visible
Flexible implementation of EU-SILC
Some potential sources of non-comparability
Different sampling strategies
(2) Different survey designs
(3) Different modes of data collection
(4) Different field work periods and procedures
(5) Different national questionnaires
(6) Different reference periods
(7) Different nonresponse rates
(8) Different attrition rates
…
(1)
6
Comparability
7
Why is comparability so important?
EU-SILC is the central data source for social reporting in Europe
• Social indicators based on EU-SILC are used
•
•
•
8
to assess countries’ places in relation to each other
to learn from others’ best practices
to evaluate policy measures
Selected issues regarding data comparability
1. Different survey designs and response rates
2. Different modes of data collection
3. Ex-ante output harmonization: Wording of questions
9
Comparability 1: Survey design and response rates
• Survey design
•
Rotational panel: variations across countries in the number of
rotations.
Most countries 4, but 8 in NO, 9 in FR and full panel in LU
(in the future possibly 6 or more waves)
• Response rates and attrition vary throughout Europe
10
SILC response rates 2007 (only new rotational group)
100
80
60
40
20
BE
HU
LV
ES
UK
PL
CZ
EE
AT
DE
SI
BG
FI
EL
IT
IS
MT
NL
CY
FR
PT
RO
SE
SK
DK
NO
0
11
Source: Eurostat: Proposal for revising the design of EU-SILC longitudinal component. Item 4; 5th Meeting of the Task-Force on the revision of the EU-SILC legal basis.
EU-SILC retention rates
(a) households, (b) individuals re-interviewed the following year, in %
100
95
90
85
80
75
UK
AT
SI
BE
NL
LV
DK
BG
IE
LU
SP
IT
HU
NO
EL
LT
FR
SE
PT
EE
PL
IC
FI
CZ
CY
SK
RO
70
% of eligible households in which at least one member was interviewed the next year
% of eligible individuals in the sample who were interviewed the next year
12
Source: Iacovou et al (2012) from EU-SILC longitudinal files, release 2008-4, unweighted
Comparability 2: Different modes of data collection
Sources of EU-SILC data could be:
•
•
•
survey(s)
register(s)
combination of survey(s) & register(s)
 Data could come from one source or two sources
 Issue of concern: Substantial findings of EU-SILC such as
indicators used in social reporting may differ due to the
diversity in the data collection across countries
13
Different modes of data collection (2010)
Information/
Interview
completed from
14
Survey
Register
73.1%
3.4%
Both: Survey &
Register
Full Record
Imputation
22.8%
0.7%
Survey countries
Register countries
BE, BG, CY, CZ,
DE, GR, ES, EE,
FR, HU, IT,LT, LU,
MT; AT, PL, PT,
RO, SK, UK
DK, FI, SE, NL, IS, LV, SI, IE
Use of registers for different domains
15
Different modes of data collection
• Measures in surveys and registers may be based on different
concepts, e.g.
•
Earnings information in registers


•
tax-based (non taxed earnings?)
different points in time when income and tax are collected (self-employed,
temporary workers)
Employment, Unemployment


evidence that information on unemployment in survey and registers differ
in a significant way at the individual level
survey: e.g. memory errors regarding employment situation in the past
• Consistency problems if combining information from different
sources?
16
Different modes of data collection
• Mixed modes of data collection in surveys
•
Personal interview (respondent)




•
CATI (Computer Assisted Telefon Interview)
CAPI (Computer Assisted Personal Interview)
PAPI (Paper and Pencil Personal Interview)
self-administered (respondent completes the questionnaire him/herself)
Proxy-interview (respondent has someone else answer the questions
for him/her)
• Type of interview might affect the response and thus reduce the
comparability between countries and
for countries with sequential mixed mode between waves
17
Different modes of data collection (2010)
18
SURVEY COUNTRIES
REGISTER COUNTRIES
Interview
Register & Interviews
73.1%
22.8%
Face to Face
Face to Face
• CAPI 29.1%
• PAPI 42.3%
• CAPI 0.4%
• PAPI 11.7%
CATI
CATI
3.9%
65.6%
Selfadministered
5.5%
Selfadministered
0.4%
Proxy
Interview
Proxy
Interview
18.3%
22.0%
Proxy interview by country – ‘register countries’
% of proxy interviews
Country
Iceland
proxy interview
0.0
Nederland
1.2
Sverige
2.1
Ireland
23.7
Latvia
23.7
Norway
23.8
Slovenia
24.6
Suomi
Danmark
As a rule only 1 person in hh is
interviewed, who answers also
for all other hh members
42.7
48.6
Datasource: UDB_c10R_ver 2010-2 from 01-08-12, own computation
19
Proxy interviews by country – ‘survey countries’
% of proxy interviews
Slovak Republic
4.3
Czech Republic
Ellada
8.0
Hungary
Belgique
8.6
Luxembourg
United Kingdom
10.7
Bulgaria
Oesterreich
13.7
Portugal
Romania
15.3
Espana
Lithuania
15.6
Cyprus
Deutschland
18.8
Estonia
Italia
19.0
France
Poland
19.2
Malta
Datasource: UDB_c10R_ver 2010-2 from 01-08-12, own computation
20
19.3
19.8
20.1
20.3
20.5
21.9
23.0
24.3
27.6
28.9
Flexibility in modes of data collection
• Quality of proxy interviews might depend on the reason of
the proxy interview
1. a respondent is not accessible or willing to give an interview
•
proxy interview are cofounded with other characteristics like
age or sex
2. producers take proxy interviews as a mean to lower costs
•
21
data producers might make efforts for a random selection of
proxy respondents
Comparability 3 – Different questionnaires
• SILC is ex ante harmonized, i.e.
variables which are delivered by the NSIs to Eurostat are
defined in regulations & guidelines
(=> standard EU-SILC definition)
• But there is no common SILC questionnaire
•
•
22
questionnaire design varies (e.g. order of questions)
wording of questions varies (e.g. ‘How often do you usually ..’
or 'How often during a usual year do you …?)
Research example
 Research example Gash (2011): Methodological issues in
comparative research. European Workshop to Introduce the
EU-SILC and EU-LFS Manchester
Research interest
• How does unemployment affect social engagement?
• EU-SILC Module (2006) on Social Participation
•
•
•
•
23
Frequency of contacts/getting together with friends &relatives
Ability to ask relatives, friends, neighbours for help
Participation in formal and informal activities
Participation in cultural/sport events
Research example
Main findings:
• Broad agreement in the questionnaire wording across
countries, but
1.
2.
3.
Some countries provide examples of social participation
others not
Some countries mention reference periods others not
Some countries prompt that respondents should exclude
people they live with others not
 Might have an effect on the reported frequencies of
contacts
24
Source: Gash, Vanessa (2011): METHODOLOGICAL ISSUES in COMPARATIVE RESEARCH
25
Comparability 3 - Output harmonization
• When Eurostat knows about problems arising from different
wording or other deviations in the questoinnaire it reports
this
• Most national questionnaires are available
• Check documentation!!!
26
Opportunities for Longitudinal
Analysis with EU-SILC
27
Opportunities for Longitudinal Analysis
Main topics studied with SILC (Eiffe & Till 2013)
• Income studies
• Poverty studies
• Labour market studies
• Limitations arise because SILC is a short-term panel,
i.e. a maximum of 3 transitions
28
Income studies
Income distribution
•
What are the consequences of income gains and losses on
income inequality and poverty levels?
•
How do regional economic and labour market structures as
well as national institutions contribute to changes of income
level and income distribution?
Income dynamics
•
How much does income mobility vary across European
countries?
29
See: Franz F. Eiffe and Matthias Till. 2013. The Longitudinal Component of EU‐SILC Still Underused? NetSILC2:
Working Paper 1/2013.
Income studies
Impact of socio-economic events on income
• Impact of having a disabled person in a household
• Changes in women’s contribution in Italian families
• Effect of partnership breakdown on individual income
30
See: Franz F. Eiffe and Matthias Till. 2013. The Longitudinal Component of EU‐SILC Still Underused? NetSILC2:
Working Paper 1/2013.
Poverty studies
• How long do individuals or households remain in poor living
conditions?
• How often do Europeans experience poverty over their life
span (or at least over four years)?
• What are the profiles of households who remain in poverty
for longer periods?
• What are the events/determinants that trigger poverty
transitions?
31
See: Franz F. Eiffe and Matthias Till. 2013. The Longitudinal Component of EU‐SILC Still Underused? NetSILC2:
Working Paper 1/2013.
Labour market studies
• What patterns of occupational mobility can be observed in
Europe?
• How difficult is it to leave unemployment?
• How do labour market dynamics differ across countries?
• Can difference between countries be explained by different
institutions, e.g. welfare state arrangements?
See: Franz F. Eiffe and Matthias Till. 2013. The Longitudinal Component of EU‐SILC Still Underused? NetSILC2:
Working Paper 1/2013.
32
Possible problems of SILC longitudinal
• Different attrition rates could be a problem
• If there is a correlation between attrition and income or
others variables this would be a problem
However, income bias related to attrition seems to be low
33
Household participation in SILC by
Income Quintiles in previous year
% Non-Response / % Response
2.0
1.5
1.0
0.5
0.0
1
2
3
4
5
Income Quintiles
34
Source: Eurostat: Proposal for revising the design of EU-SILC longitudinal component. Item 4; 5th Meeting of the Task-Force on the revision of the EU-SILC legal basis.
Thanks for listeing!
35