Monitoring child well-being in the EU: measuring cumulative deprivation
Keetie RoelenGeranda Notten
ISCI Conference, 27 July 2011
The breadth of poverty
The breadth of poverty
The breadth of poverty
Academic and policy relevance
fits widespread and increased attention for child poverty; seeks to address questions around construction of measures, overlap
between indicators and measures and concurrent implications for policy
We know about child well-being in the EU at:• Macro-level: Bradshaw et al., 2006; IRC RC 7; OECD, 2009• Micro-level: TARKI, 2011
BUT we know little about: • overlap and breadth of child poverty• composite measures of poverty at micro-level
This study
• Overlap and breadth of child poverty in the EU– Across domains– Across countries– Underlying factors
• Options for constructing a multidimensional measure of cumulative child well-being in the EU
Data
• EU-SILC 2007, cross-sectional data
• Germany, France, Netherlands and UK
Domains and indicators
Housing conditions - Dwelling has leaking roof, damp walls/floors/foundation, rot in window
frames or floor- Dwelling is not comfortably warm during winter time- Dwelling is overcrowded
Neighborhood conditions- Pollution, grime or other environmental problems- Crime violence or vandalism in the area
Access to basic services- Accessibility of primary health care services- Accessibility of compulsory school
Domains and indicators - continued
Financial means - Household has payment arrears on mortgage/ rent, utility bills, loan
payments- Household can’t afford meal w. meat, chicken, fish (vegetarian equivalent)
every 2nd day- Household can’t afford paying for one week annual holiday away from
home- Household can’t afford a computer for financial reasons- Household can’t afford a car for financial reasons- Ability to make ends meet (very difficult)
Monetary poverty- 60% of median income
Domain deprivation
Overlap patterns
Overlap patterns
Monetary vs. multidimensional poverty
A, B or AB (as %
of total population)
A - deprived but not income
poor (as % of A+B+AB)
B- income poor but not
deprived (as % of A+B+AB)
AB - deprived and income
poor (as % of A+B+AB)
odds
Neighborhood problemsDE 35.1 [33.2,37.0] 60.4 24.8 14.8 1.84*FR 35.8 [33.4,38.2] 56.1 29.1 14.8 1.61*
NL 37.6 [35.2,40.0] 63.2 29.1 7.7 0.71
UK 49.3 [47.1,51.6] 53.4 28.7 17.8 1.18Difficult access to basic servicesDE 32.7 [30.9,34.6] 32.5 62.1 5.3 1.43*
FR 27.0 [25.0,29.1] 7.1 89.9 3 1.1
NL 26.2 [23.9,28.6] 24.9 70.3 4.9 1.1
UK 31.1 [28.9,33.3] 3 95.4 1.7 1.75*
Monetary vs. multidimensional poverty
A, B or AB (as %
of total population)
A - deprived but not income
poor (as % of A+B+AB)
B- income poor but not
deprived (as % of A+B+AB)
AB - deprived and income
poor (as % of A+B+AB)
odds
Housing problemsDE 32.9 [31.0,34.8] 57.8 21.9 20.4 3.30*FR 35.9 [33.7,38.1] 56.1 19.2 24.7 4.03*
NL 32.4 [30.0,34.9] 57.1 25 17.9 2.59*
UK 40.4 [38.1,42.6] 43.2 29.3 27.5 3.23*Financial strainDE 41.4 [39.4,43.3] 66.3 9.4 24.2 5.48*
FR 45.0 [42.8,47.1] 65.1 8 26.9 6.30*
NL 28.5 [26.1,31.0] 51.2 24.2 24.6 4.92*
UK 47.3 [45.1,49.6] 51.5 12.9 35.7 5.99*
NeighborhoodProblems
Difficult accessto basic services
Financialstrain
Overlap (%) Odds Overlap (%) Odds Overlap (%) OddsHousing problemsDE 9.5 2.02* 7.1 1.45* 15.9 4.00*
[8.4,10.8] [1.66,2.46] [6.2,8.3] [1.18,1.79] [14.5,17.4] [3.29,4.85]FR 10.5 2.14* 4.7 1.33 19.4 4.53*
[9.0,12.2] [1.74,2.64] [3.7,5.8] [1.00,1.77] [17.6,21.3] [3.69,5.56]NL 8.5 1.71* 3.9 1.17 9.1 3.02*
[7.2,10.0] [1.33,2.19] [2.9,5.2] [0.82,1.66] [7.4,11.1] [2.29,4.00]UK 11.4 1.34* 5.3 2.17* 18.2 3.70*
[10.0,13.0] [1.09,1.66] [4.0,6.8] [1.56,3.01] [16.2,20.4] [2.99,4.57]Neighborhood problemsDE 6.5 1.17 12.4 1.72*
[5.5,7.7] [0.93,1.46] [11.1,13.8] [1.43,2.07]FR 3.8 1.14 13.3 1.83*
[3.1,4.6] [0.87,1.51] [11.7,15.0] [1.51,2.22]NL 4.4 1.24 7.9 1.83*
[3.6,5.3] [0.94,1.64] [6.6,9.4] [1.41,2.38]UK 5.7 1.80* 16.5 1.44*
[4.6,7.0] [1.32,2.46] [14.8,18.4] [1.19,1.74]Difficult access to basic servicesDE 10.7 1.67*
[9.5,12.0] [1.38,2.03]FR 6.3 1.24
[5.2,7.5] [0.96,1.61]NL 4.4 1.78*
[3.4,5.9] [1.26,2.52]UK 7.0 2.17*
[5.6,8.6] [1.61,2.92]Source: own calculations with EU-SILC, wave 2007. * means significant at a 1% level.
Factors influencing domain deprivation
• Single-parenthood:Significantly increases probability to being financially strained, income poor
and experiencing housing problems
• No or low work intensity in household: Significantly increases probability to being financially strained, income poor
and experiencing housing problems
• Living in rented dwellings:Significantly increases probability to be financially strained and experiencing
housing problems and, to a lesser extent, being environmentally deprived, and income poor
• Low educational attainment parents:Significantly increases probability to being financially strained and income poor
In conclusion
A diverse picture
• Limited overlap with considerable size and group differences between indicators of monetary and multidimensional poverty;
• Considerable differences across countries;
• Indicators of monetary poverty and multidimensional poverty can not serve as a proxy for one another;
• Higher levels of overlap are not necessarily an indication of higher odds for experiencing cumulative deprivation
What are appropriate measures of cumulative deprivation?
• Why?- more deprivations are worse than one- one headline statistic is practical
• EU policy context- search for child specific indicators- many single indicators, one composite index
• Criteria- sensitive to changes in breadth of deprivation- intuitive interpretation
Aggregation option I
Simple headcount vs. adjusted headcount
• Simple headcount =
or the proportion of poor in the population
• Adjusted headcount = (x1 deprivation) (x3 deprivations)
or the proportion of deprivations in the population
Aggregation option II
Absolute vs. relative poverty line
Absolute: poverty line=2, headcount=4
Relative: poverty line=median, headcount=6
3
2
2
1
1
0
0
0
0
4
Headcount, absolute versus relative
Adjusted headcount, absolute versus relative
Adjusted headcount, absolute in UK
In conclusion
• Adjusted headcount (CDI) with cumulative deprivation threshold of 1 works best
• Can be complemented with headcount with higher cumulative deprivation threshold
• Need other method to determine relative cumulative deprivation threshold