The distributional impact of in kind public benefits in five European countries
Alari Paulus Institute for Social and Economic Research, University of Essex
Holly SutherlandInstitute for Social and Economic Research, University of Essex
andPanos Tsakloglou
Athens University of Economics and Business and IZA
Paper presented at the conference “European Measures of Income, Poverty, and Social Exclusion: Recent Developments and Lessons for U.S. Poverty Measurement”
U.S. Census Bureau, University of Maryland School of Public Policy and Association for Public Policy Analysis and Management, Washington DC, Wednesday, November 4, 2009
Motivation
• Usual practice in empirical distributional studies, as well as EU inequality and poverty statistics: distributions of disposable monetary income
• But individuals derive utility from the consumption of non-purchased as well as purchased commodities (privately or publicly provided)
• Implications for inter-temporal, cross-country and some inter-personal comparisons because of changes/differences in
– extent of public provision
– division into cash and non-cash (both public and private)
• Here we focus on the results of an empirical study estimating the value and incidence of publicly-provided health care and education and subsidised social housing in 5 EU countries
• We also consider the welfare interpretation of measures of cash + non-cash incomes
Background
• Building on previous cross-national comparisons by multi-national teams (e.g. Smeeding et al. 1993 RIW)
• AIM-AP project (Accurate Income Measurement for the Assessment of Public Policies) FP6 2006-2009
• Non-cash incomes: one part of a 3-part 11-partner project
– (Non take-up of benefits and tax evasion)
– Indirect taxes
• Private as well as public non-cash: here focus on public components
• 7 countries: here focus on 5 only: BE, DE, EL, IT, UK
• Underlying aim: to improve and broaden scope of EUROMOD analysis: here focus on basic results and use EUROMOD simulated cash incomes as starting point
• National “state of the art” methods/analysis + comparable “best possible” methods/analysis: here focus on the latter
• To find out more: www.iser.essex.ac.uk/research/euromod/aim-ap-project
Methods (1)Microdata
Country Dataset Date of
collection
Reference time period for incomes
Tax-benefit system
BE Belgium EU-SILC 2004 2003 2003
DE Germany German Socio-Economic Panel 2002 2001 2001
EL Greece Household Budget Survey 2004/ 5 2004 2004
IT Italy Italian version of EU-SILC 2004 2003 2003
UK UK Family Resources Survey 2003/ 4 2003/ 4 2003
Methods (2) Estimating values for in kind rent subsidies
Subsidy = Market rent – rent paid
• Rent paid: taken from data (before any HB)
• Market rent estimated as for imputed value of owner occupation: opportunity cost (rental equivalence) approach
• For private market tenants: gross rents modelled as dependent variable (using characteristics of dwelling, area and/or occupants); coefficients used to impute gross market rents for otherwise similar rented dwellings
• Social tenants: 20% of the households in UK; 7% DE; 5% BE; concentrated in lower income groups
Methods (3)Estimating values for education and health care
• Public education – OECD Education at a glance per student spending by level of
education – Primary, secondary, tertiary, disregarding other stages (pre-primary,
post-secondary non-tertiary, etc), matched into microdata for students in non-private education
– Particular issues related to tertiary: part/full time; fees; R&D expenditure; where do students live and how are they treated in surveys?
– In all countries: more beneficiaries in lower income groups (varies by level of education)
• Public health care– Social expenditures on health care services per capita (SOCX -
OECD); by age see Marical et al. (2006); matched into microdata by age
– Insurance value approach– In all countries: much higher expenditures at older ages
7
Methods (4) General remarks
• Static incidence analysis under the assumption of no externalities
• Partial short term analysis (taxes and social insurance contributions given)
• Benefit shared by all household members (“un-spent” household income)
• No inefficiencies in the production of public services
• Modified OECD equivalence scale
Non-cash income components as % of household cash disposable income: all households
0
5
10
15
20
25
30
BE DE EL IT UK
% o
f d
isp
osa
ble
in
com
e
Rent Subsidy Education HealthSource: EUROMOD
Non-cash income components as % of household cash disposable income: bottom quintile
0
10
20
30
40
50
60
70
BE DE EL IT UK
% o
f d
isp
osa
ble
in
com
e
Rent Subsidy Education HealthSource: EUROMOD
Percentage change in inequality indexes after adding non-cash benefits
-70%
-60%
-50%
-40%
-30%
-20%
-10%
0%
Belgium Germany Greece Italy UK
Gini Atkinson (0.5) Atkinson (1.5)
Percentage with less than 60% median income (cash vs cash + non cash)
0
5
10
15
20
25
30
all children elderly all children elderly
cash disposable income cash disposable income + non-cashtransfers
BE DE EL IT UK
Preliminary conclusions and a qualification
• The methodology adopted in the project is in line with the existing literature
• In all countries the inclusion of the three non-cash benefits appears to lead to a substantial decline in measured levels of inequality and poverty
• Broadly similar effects in each country; no re-ranking (except for sub-groups)
• Rent subsidies: no doubt that the measured changes have a straightforward welfare interpretation
• Education and health care: not necessarily so, due to the use of “conditional” equivalence scales
• The “alternative scenario” (implicitly, paying for the services or insuring against their costs out of disposable income) changes the institutional framework: the corresponding needs (for education and health care) should be taken into account
• Also, public goods theory: provision in equal quantities; not in proportion to the individuals’ incomes (scale invariance of inequality indices)
Accounting for health care and education needs (1)
• Assuming that y is household income, k is the amount of extra needs of the household members for health and education, e the OECD scale, and e’ the new scale,
the following should be valid for the household to remain at the same welfare level:
y/e = (y+k)/e’ and e’ should be equal to
e’ = e(y+k)/y• But how to estimate k?
Should it be a function of demographics alone or demographics and income?
Accounting for health care and education needs (2)
• An empirical approach which simply assumes needs=expenditure would take us back (almost) to square 1
• Except – private education, dropouts, (non-compulsory education), (private
health insurance)
– variation in spending across countries: take EU averages (or minima or maxima) to represent the “in principle” cost of meeting needs
• per capita amounts, adjusted by GDP relative to EU mean
• Crude calculations, sensitivity tested, to illustrate the implications of the approach
• To be less crude within the same framework requires better, more detailed, data
Change in Gini coefficient with addition of 3 non-cash benefits without equivalence scale adjustment (Baseline) and with adjustment (Scenarios 1 & 2)
E-scale adjustment: k is EU mean heath care per capita + education per student spending (GDP-adjusted)
Scenario 1: Compulsory education only in needs
Scenario 2: All education levels in needs
Belgium Germany Greece Italy UK
Baseline -22.8 -21.3 -16.5 -20.3 -21.0
Scenario 1 (-0.3) -4.0 1.5 -2.2 -2.0
Scenario 2 0.2 -2.4 3.1 -0.7 -1.1
Final conclusions
• It is important for comparability to measure the incidence of non-cash incomes across the cash income distribution
• Importance of this type of analysis in a life-cycle framework
• Housing subsidies (and private imputed rent) should be included in income, to improve comparability
• A welfare interpretation of adding to cash income the value of public health care and education requires that needs for health and education are accounted for in the equivalence scale
• Our crude empirical experiment suggests that the net effect of adding non-cash benefits would then be small
• The effect could well be larger for comparisons with non-EU countries
Proportional changes in inequality indices(Imputed Rent + Education + Health)
-60
-50
-40
-30
-20
-10
0
BE DE EL IR IT NL UK
Gini
Atkinson0.5
Atkinson1.5
Re-ranking effects
BE DE EL IR IT NL UK
M A M A M A M A M A M A M A
Gini 2 2 3 4 6 7 4 3 5 6 1 1 7 5
Atkinson0.5 2 2 3 4 5 5 4 3 6 7 1 1 7 6
Atkinson1.5 2 2 3 4 4 5 5 3 7 7 1 1 6 6
FGT0 2 2 3 4 6 6 7 5 5 7 1 1 4 3
FGT1 2 2 3 4 6 6 5 5 7 7 1 1 4 3
FGT2 2 2 4 5 6 4 3 3 7 7 1 1 5 6
M: Monetary
A: Augmented
Importance of consumption of own production
Greece From own
farm production
From own non-farm
production
From other households
From employer
ALL
Gini -1.6 -0.2 -1.4 0.1 -3.2
Atkinson0.5 -3.3 -0.5 -3.0 0.1 -6.6
Atkinson1.5 -3.8 -0.7 -3.9 0.2 -7.9
FGT0 -2.8 -0.8 -2.7 0.0 -6.1
FGT1 -6.8 -1.1 -4.4 1.0 -13.1
FGT2 -9.4 -2.1 -7.2 1.1 -18.9
Proportional change in % Germany
flat.4 flat.8 pred.wage
Gini -13.9 -22.8 -18.3
Atkinson 0.5 -25.5 -39.6 -34.0
Atkinson 1.5 -31.5 -44.2 -38.9
FGT0 -18.7 -27.7 -19.9
FGT1 -33.2 -47.0 -41.5
FGT2 -44.4 -60.3 -56.5
Calculation of k (needs)
For each household with n members (i=1,…,n) with
different characteristics (such as age) the needs for
education and health care are assumed to be:
n
i HNi
HEUiHNi
ENi
EEUiENi S
Sk
S
Skk
1
where kENi and kHNi are, respectively, national spending
for public education and public health care for persons
with characteristics i, SENi and SHNi are national spending
figures for public education and public health care
expressed as a share of the country’s GDP per capita (i.e.
they are equal to kENi/GDPpcN and kEHi/GDPpcN,
respectively) and SEEUi and SHEUi are the corresponding
(unweighted) averages for EU15