Cross-national differences in determinants of multiple
deprivation in Europe
Francesco FigariEPUNet Conference
May, 8th 2006
Research questions
• Which is the longitudinal relationship between income (and other determinants) and deprivation in the European countries?
• Which are the reasons for the deprivation differentials across Europe?
Multiple deprivation• Multidimensional approach → outcome elements
(Townsend 1979)
• Concept and measurement → indirectly and directly (Ringen 1987, 1988)
• Deprivation → Exclusion from minimum living standards (Nolan and Whelan 1996)
• Social exclusion → Relationship between current income and living conditions indicators
Motivations
• Policy → at the EU level necessity of quantitative indicators to monitor the Lisbon Strategy
• Conceptual → “poverty” ≠ “deprivation”
• Empirical → mismatch “poverty” – “deprivation”→ different determinants across countries
Empirical literature review
EU level• Nolan and Whelan (1996, …)
– Methodological measurement aspects– Identification of different dimensions of deprivation– Relationship between income poverty and deprivation– Determinants of “consistent poverty”
UK • Berthoud, Bryan and Bardasi (2004)
- Longitudinal relationship between income (and other determinants) and deprivation
Data
• ECHP
• 1994 – 2001
• 11 countries(excluded: Germany, UK, Luxembourg and Sweden)
Items and Dimensions• 24 non-monetary indicators (Eurostat 2002)
• Factor analysis (Whelan et al. 2001) 5 dimensions + Overall
– Basic 7 items: housing, social activities, diet, clothes
– Secondary 6 items: durables
– Housing facilities 3 items: services
– Housing deterioration 3 items: structural parts
– Environment 5 items: noise, pollution, crime, space, light
• Cronbach’s for overall deprivation
0.76 0.74 0.65 0.82
IT NL PT
0.70 0.70 0.65 0.79 0.73 0.69 0.71
ES FI FR IEAT BE DK EL
Average number of items lacking in the Overall Deprivation
0
1
2
3
4
5
6
7
819
94
1995
1996
1997
1998
1999
2000
DK
NL
B
F
Irl
I
EL
E
P
A
Fin
Deprivation indexPrevalence weighting procedure within each country and each wave
- Each item (j)
- Each household (h)
Normalisation
DW= 0.99+0.83 = 1.82
I p IW
tv 1 99.35% 0.99dishwasher 1 82.57% 0.83
jthjt
Whjt pII
J
j
Whjtht ID
1
100*
11
J
jjt
J
j
Whjtht pID ]100,0[htD
Deprivation indexCountry-specific and time varying weights
to compensate for variations in deprivation due to the trend of possession over time and social and cultural differences across countries
At household level and attributed to the individuals to follow them across waves
Just an indicator and not a direct measure of deprivation: weak set of assumptions questionable choice of the indicators formulation of questions in terms of non affordability or unwillingness focus on some specific areas of consumption
The minimum value is not a censored point it cannot be considered as the direct realization of the true and latent deprivation value
Overall deprivation index - 2000
05
1015
20
DK IE NL AT BE FI FR ES IT EL PT
Average overall deprivation score Poverty rate
AT BE DK EL ES FI FR IE IT NL PT
99th percentile 34.90 36.19 32.45 53.75 42.60 36.29 40.86 36.34 42.99 31.72 63.41
Average 6.12 6.49 5.07 17.2 10.13 7.05 8.05 5.13 10.64 5.62 18.23
% with zero 35.16 36.39 47.54 1.84 21.53 32.86 29.97 46.81 15.30 38.20 7.00
Coeff. of Variation 1.253 1.287 1.424 0.738 0.994 1.192 1.168 1.559 0.937 1.314 0.806
Correlation between Overall Deprivation index and equivalent income
AT BE DK EL ES FI FR IE IT NL PT
-0.20 -0.22 -0.19 -0.41 -0.32 -0.26 -0.33 -0.28 -0.36 -0.24 -0.40
Overlap Income - deprivation
010
2030
4050
%
AT DK NL FI BE IE ES FR EL IT PT
Poor and deprived in 2000
Persistently poor and persistently deprived in 1998-2000
Persistently poor in 1998-2000 and deprived in 2000
6
810
12
510
15
810
12
1015
20
68
1012
1015
20
1820
2224
26
1012
1416
18
2022
2426
28
67
89
10
810
1214
5000 10000 15000 4000 6000 80001000012000 5000 10000 15000 4000 6000 80001000012000
4000 6000 8000 1000012000 2000 4000 6000 8000 10000 2000 4000 6000 8000 2000 4000 6000 8000
2000 4000 6000 5000 10000 15000 4000 6000 80001000012000
Denmark Netherlands Belgium France
Ireland Italy Greece Spain
Portugal Austria Finland
Ove
rall
Dep
rivat
ion
Inde
x
Equivalent income PPP
Low income - deprivation
Econometric specification
itititit vD βx• Overall Deprivation Score
• Income (deflated at 2000 prices and exressed in PPS) Social transfers Education level Employment status (employed, unemployed, inactive) Health status Housing tenure (owner, mortgage, tenant, free) Number of adults, number of children, family type,
proportion of elderly• Time-specific effect: dummy variable for each year• Individual-specific unobserved effect
Random and Fixed effects models
itD
itx
tiv
Decomposition of deprivation gap
ccccD β̂xˆ Average predicted deprivation score
ConstantReturnssticsCharacteriGAP
)ˆˆ()ˆˆ(ˆ)( ABABBAABAB DD ββxβxx
Averaging approach:
contribution of each variable
Returns
111
sticsCharacteri
111 )ˆˆ(ˆ)( ABBAAB ββxβxx
Results: Fixed EffectsHausman specification tests suggest a preference for the FE specifications
General robustness of the results across countries according to the expectations
Income negatively associated
the impact of the first lag is stronger than of the current income
the second lag are still statistically significant in most of the countries
Employment status moving into and out of the labour market is as important as being in or out of it
Education level Difficulties to capture the impact of the achievement of a new education level.
Housing tenure penalty of moving in rented houses
Decomposition of deprivation gap
-50
510
15A
bsol
ute
valu
es
NL AT FI IE BE FR IT ES EL PT
Characteristics Returns Constant
Decomposition of deprivation gap
-5 0 5 -5 0 5
PT
EL
ES
IT
FR
BE
IE
FI
AT
NL
PT
EL
ES
IT
FR
BE
IE
FI
AT
NL
Characteristics Returns
Current income 1st lagged income 2nd lagged income
Income
Income
Decomposition of deprivation gap
Employment status of household head
-.6 -.4 -.2 0 .2 .4 -.6 -.4 -.2 0 .2 .4
PT
EL
ES
IT
FR
BE
IE
FI
AT
NL
PT
EL
ES
IT
FR
BE
IE
FI
AT
NL
Characteristics Returns
workingref unemployref inactref
Employment status HH head
Decomposition of deprivation gap
Housing tenure
-1 -.5 0 .5 1 -1 -.5 0 .5 1
PT
EL
ES
IT
FR
BE
IE
FI
AT
NL
PT
EL
ES
IT
FR
BE
IE
FI
AT
NL
Characteristics Returns
Own house Outstanding mortgage Rented Free-rent
House tenure status
Decomposition of deprivation gap
Family structure
-1 0 1 2 3 -1 0 1 2 3
PT
EL
ES
IT
FR
BE
IE
FI
AT
NL
PT
EL
ES
IT
FR
BE
IE
FI
AT
NL
Characteristics Returns
Family composition Lone parent
Family composition and lone parent
Decomposition of deprivation gap
Constant
0 5 10 15 20
PT
EL
ES
IT
FR
BE
IE
FI
AT
NL
Returns
Constant
Cross-country conclusionsHigh deprivation countries
Contribution to a reduction of the gap:- increase in income- home ownership - job market participation of household members
Contribution to an increase of the gap:- family composition- (fixed country effect)
Policy implications
• More comprehensive policies in addition to income policies
• Primary attention to long term poverty status
• Definition of the eligibility criteria of the beneficiaries of public policies
Further developments
• Dynamic analysis• short and long term effects of the socio-economic
determinants• persistence of deprivation over time • impact of the determinants given the initial
deprivation conditions.
Average number of items lacking in the Basic Deprivation
0
0.5
1
1.5
2
2.5
3
3.5
4
4.519
94
1995
1996
1997
1998
1999
2000
DK
NL
B
F
Irl
I
EL
E
P
A
Fin
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
1994
1995
1996
1997
1998
1999
2000
DK
NL
B
F
Irl
I
EL
E
P
A
Fin
Average number of items lacking in the Secondary Deprivation
• BASIC DIMENSIONCan the household afford…
...keeping its home adequately warm?
...paying for a week's annual holiday away from home?
… replacing any worn-out furniture?
… buying new, rather than second-hand, clothes?
… eating meat, chicken or fish every second day?
… having friends or family for a drink or meal at least once month?
… paying scheduled rent/mortgage and utility bills of the house?
• SECONDARY DIMENSIONAffordability of…
… car
… tv
… video recorder
… micro wave
… dishwasher
… telephone
• HOUSING FACILITIES DIMENSIONDoes the dwelling have…
… bath or shower?
… indoor flushing toilet?
… hot running water?
• HOUSING DETERIORATION DIMENSIONDoes the accommodation have…
… leaky roof?
… damp walls, floors, foundations…?
… rot in window frames or floors?
• ENVIRONMENT DIMENSIONDoes the accommodation have…
… noise from neighbours?
… shortage of space?
Is there any pollution, grime, or other environmental problem…?
Is the accommodation too dark / not enough light?
Is there crime or vandalism in the area?
Variables•Dependent variable Overall Deprivation Score
• HH head and HH members– Income (deflated at 2000 prices and in PPS): current and lagged– Social transfers– Education level (less secondary school, secondary school or +)– Employment status (employed, unemployed, inactive)– Health status – Housing tenure (owner, mortgage, tenant, free) – Number of adults, number of children, family type (single, couple with kids, couple without kids, lone parent), proportion of elderly – Dummy variable for each year
Econometric model
• Random Effects
• Fixed Effects
itititit vD βx
)(β)xxx()( iitiitiit DDD
)}(){(β)xx()()( iitiiitiit vDD 11
Results: Fixed EffectsHausman specification tests suggest a preference for the FE specifications
Income negatively associated
the impact of the first lag is stronger than of the current income
the second lag are still statistically significant in most of the countries
Employment status moving into and out of the labour market is as important as being in or out of it
becoming inactive has a significant and positive impact in DK, FR, EL, NL, PT and FI.
if the proportion of person employed increases in the household the impact is statistically significant that in DK, FR, EL, ES, PT and FI.
Education level negatively associated (RE) but difficulties to capture the impact of the achievement of a new education level.
Results: Fixed EffectsHealth status negatively associated
an improvement in the health status is important in terms of deprivation reduction.
Housing tenure penalty of moving in rented houses rather than in an own house
presence of outstanding mortgage has a negative effect on deprivation score in BE, DK, NL, PT.
the coefficients associated to a free-rented house are always positive (no AT, IE, NL) reflecting the generally poor nature of these houses.
Family structure positive effect of becoming lone parent in BE, DK, FR, ES, NL
positively associated with number of adults and children