Turkey:
Expanding Opportunities for
the Next Generation
Francisco Ferreira and Jesko Hentschel
April 19, 2010
Motivation Inequality of Opportunity
Child Development ECD Reflections
Outline
1.
Motivation
2. Inequality of Opportunity
3. Child Development Today
4. Early Childhood Development
5. Reflections
2
Motivation Inequality of Opportunity
Child Development ECD Reflections
Turkey’s income inequality remains high by European
standards, and “middling” by international standards
Gini Index for Income per capita
70.0
60.0
45.0
50.0
40.0
30.0
20.0
10.0
0.0
Source: United Nations University – WIDER Data (2003)
3
Motivation Inequality of Opportunity
Child Development ECD Reflections
Still, Turks appear to be unusually inequalityaverse.
The share of the population that strongly agrees that
"the gap between rich and poor should be reduced in the country"
80
70
66
60
50
% 40
35 36
38
41 41
45
43 43 44
47
49
51 51 52 52
54 55 55
58 58
69
61
30
28 29 29
30
20
23
20
10
0
Source data: Life in Transition Survey, 2006
4
Motivation
Inequality of Opportunity Child Development ECD Reflections
What is the main reason why there are some
What is the Main Reason why there are
people
needin in
Turkey
some
peoplein
in need
Turkey
today?today?
Inevitable
part of life,
3%
Unluckiness,
8%
Laziness,
25%
Injustice,
65%
Source data: Life in Transition Survey, 2006
Source: Life in Transition Survey (2006)
5
Motivation Inequality of Opportunity
Child Development ECD Reflections
• Does that perception of unjust inequality reflect unequal
opportunities, more than differences in effort or merit?
1. Can we quantify the extent of inequality of opportunity in
Turkey?
2. How does the process of intergenerational transmission of
life chances affect Turkish children today?
3. What are the implications for social policy going forward?
• There is an established literature on inequality of opportunity in
theories of justice:
Dworkin (1981): What is Equality? Part 1: Equality of Welfare; Part 2: Equality of
Resources”, Philos. Public Affairs, 10, pp.185-246; 283-345.
Arneson (1989): “Equality of Opportunity for Welfare”, Philosophical Studies, 56,
pp.77-93.
Cohen (1989): “On the Currency of Egalitarian Justice”, Ethics, 99, pp.906-944.
Roemer (1998): Equality of Opportunity, (Harvard University Press)
Fleurbaey (2009): Fairness, Responsibility and Welfare (OUP)
6
Motivation Inequality of Opportunity Child Development ECD Reflections
•
More recently, there have also been a number of efforts to
measure inequality of opportunities empirically:
•
•
•
•
•
•
van de Gaer, Schokkaert and Martinez (Economica, 2001)
Roemer et. al. (J. Pub. E., 2003)
Checchi and Peragine (2005)
Bourguignon, Ferreira and Menéndez (RIW, 2007)
Lefranc, Pistolesi and Trannoy (RIW, 2008)
Lefranc, Pistolesi and Trannoy (J. Pub. E., 2009)
• And the concept has been used to inform the policy debate in
development circles:
•
•
•
•
World Development Report 2006: Equity and Development
Barros, Ferreira, Molinas and Saavedra (2009)
ADB: (Zhuang, 2010) Poverty, Inequality and Inclusive Growth in Asia
UNDP: LAC Human Development Report 2010
7
Motivation Inequality of Opportunity Child Development ECD Reflections
• But what is inequality of opportunity, exactly?
• Most of this applied literature has drawn on John Roemer’s
definition of equality of opportunities:
• Advantages (such as income, education, well-being) are determined by
individual efforts, as well as pre-determined circumstances.
• Circumstances should be treated as morally irrelevant.
• Thus:
“Leveling the playing field means guaranteeing that those who apply equal
degrees of effort end up with equal achievement, regardless of their
circumstances. The centile of the effort distribution of one’s type
provides a meaningful intertype comparison of the degree of effort
expended in the sense that the level of effort does not”
(Roemer, 1998, p.12)
8
Motivation Inequality of Opportunity Child Development ECD Reflections
•
This definition has been formalized as follows:
•
Consider a finite population of agents indexed by i  1,...N
•
Each i is characterized by
•
C is a vector of J elements, each of which is discrete and can take a
finite number of values xj.
•
Define a partition   T1 , T2 ,...,TK 
•
Roemer’s definition implies:
yi , Ci , ei 
y k    y l  ,   0,1; Tk , Tl  
•
Invert the quantile function to obtain the “strong criterion” of
inequality of opportunity in Bourguignon, Ferreira and Walton (2007)
and Lefranc, Pistolesi and Trannoy (2008).
F k  y   F l  y , l , k Tk  , Tl  
Motivation Inequality of Opportunity Child Development ECD Reflections
•
Empirically, the estimation of Fk(y) is very data-intensive. There is a trade off
between implementing a full-distribution identification criterion, and recognizing
the role of the full set of observable circumstances.
•
An alternative approach (CP, 2005; BFM, 2007; FG, 2008) is to consider a weaker
empirical identification criterion for E. Op.:
 k  y    l  y , l , k Tk  , Tl  
•
Proposed as an empirical identification criterion, given the above trade-off and a
desire to incorporate a richer typology in the empirical analysis.
•
Measuring inequality of opportunity would now seem to require assessing
differences in the smoothed distribution  ik , rather than on the marginal
distribution y =  y1 ,..., y N  .
• Two obvious candidate measures:
 a  I  ik 
 
I  ik
r 
I y
Motivation Inequality of Opportunity Child Development ECD Reflections
• Computing such a measure requires data on two types of
variables :
– Advantage: an outcome that everyone can be reasonably presumed to
value: e.g. income, earnings, educational achievement
– Circumstances: pre-determined and beyond the control of the
individual: e.g. family background, race or ethnicity, birthplace
• High quality data sets that contain both sets of variables
are rare in developing countries, and were not available in
Turkey
– In particular, most household or labor surveys have limited information
on family background for today’s adults (i.e. the incomes, education
levels, or occupations of their parents)
11
Motivation Inequality of Opportunity Child Development ECD Reflections
The Data Configuration in Turkey
HBS
TDHS
• Surveyed in 2003
• 8,500 households
• Surveyed in 2003-04.
• 10,836 household
• 8,075 ever-married women
• Detailed information on
consumption
• but no information on
parental background
• No information on incomes
or consumption...
• but good information on
family background
• common set of household
“asset” variables
• common set of household
“asset” variables
Motivation Inequality of Opportunity Child Development ECD Reflections
The Data Configuration in Turkey
• Proposed methods to circumvent data configuration:
nk
 yk   2

n
ˆrN    k
Var  y 
.05
Then compute:
0




Density
 x pi  x p
yi   a p 
 s
p
p

.1
.15
• Method 1: compute a Filmer-Pritchett (2001) “wealth index” for
each household, as the first principal component of a vector of
assets (x) that includes:
– durable goods
Distribution of the asset index (with weights)
– housing characteristics
– access to amenities.
-10
-5
0
Asset index
5
10
Table 1: The Household wealth index
Principal components and summary statistics for asset indicators
Variable
Mean
Std. Dev. Scoring factor
(/sd):
full set of
variables
Has gas or electric oven
Has microwave oven
Has dishwasher
Has blender/mixer
Has DVD/VCD player
Has washing machine
Has video camera
Has iron
Has satellite antenna
Has vacuum cleaner
Has air conditioner
Has television
Has video
Has cable TV
Has camera
Has CD player
Has cellular phone
Has computer
Has internet
Has private car
Has motorcycle
Has bicycle
Works own or family's agricultural land
Source of water for drinking (ordered variable)
Piped water inside dwelling
Type of toilet (ordered variable)
Toilet inside dwelling
Type of floor material in dwelling (ordered variable)
Dwelling is owned by a household member
Dwelling is rented
Dwelling is a lodging
No rent paid for dwelling
Other type of dwelling
Number of members per sleeping room
0.712
0.072
0.221
0.392
0.317
0.783
0.035
0.851
0.143
0.756
0.047
0.947
0.073
0.062
0.339
0.182
0.671
0.116
0.063
0.258
0.045
0.193
0.137
0.501
0.742
0.675
0.782
0.041
0.620
0.248
0.014
0.116
0.002
2.412
0.453
0.259
0.415
0.488
0.465
0.412
0.184
0.356
0.350
0.429
0.212
0.223
0.259
0.240
0.473
0.386
0.470
0.320
0.242
0.437
0.208
0.394
0.344
0.861
0.437
1.946
0.413
0.520
0.485
0.432
0.118
0.321
0.040
1.223
Number of members per room
Observations
1.325
10,836
0.872
0.234
0.138
0.257
0.269
0.218
0.243
0.140
0.221
0.106
0.263
0.140
0.128
0.153
0.164
0.249
0.205
0.223
0.222
0.196
0.195
-0.009
0.116
-0.136
0.105
Scoring factor
(/sd):
common set of
variables
0.191
0.331
0.258
0.265
0.208
0.205
0.138
0.212
0.240
0.252
0.316
0.295
0.251
-0.026
-0.182
0.244
0.224
0.266
0.219
-0.043
0.062
0.039
-0.031
-0.017
-0.133
-0.047
0.043
-0.043
-0.165
Notes: mean and standard deviation of the ownership, access to amenities and dwelling characteristics (full
and reduced) set of variables, and scoring factors for the first principal components, divided by the standard
deviation.
Motivation Inequality of Opportunity Child Development ECD Reflections
The Data Configuration in Turkey
• Proposed methods to circumvent data configuration:
• Method 2: Combine information from the two surveys to impute
estimated consumption into the TDHS.
– Estima ln c  X  w   on the ancillary sample (HBS), for common set
of X, w.
~
ˆ
– Impute ci  exp  ' xi  ˆ' wi   i into the main sample (TDHS), using linear
prediction from the HBS, and a bootstrapped, heteroscedasticity-consistent
estimate of the residuals.


.8
Log annual expenditure observed in the HBS and predicted in the DHS
.6
Then compute:
 
0
.2
.4
E0 ik
r 
E0  y 
4
6
Predicted in the TDHS
8
Observed in the HBS
10
Motivation Inequality of Opportunity Child Development ECD Reflections
Circumstances and Types
Table 4: Partition of the population by circumstances
Characteristics
Pop. Share
Percent
Type of area
Rural area
Urban area
57.0
43.0
Birth region
East
Central
West
22.3
45.7
32.0
Mother's education
no education or unknown
primary education
Secondary education
higher education
68.8
28.3
2.5
0.3
Father's education
no education or unknown
primary education
Secondary education
higher education
43.0
48.5
6.8
1.7
Mother tongue
Other language
Turkish
14.9
85.1
Number of siblings
less than 3
21.7
4 to 5
48.8
6 to 8
25.0
9 or more
4.5
Sample of ever-married women aged 30-49
Source: TDHS 2003
4x4x4x3x2x2=768
Motivation Inequality of Opportunity Child Development ECD Reflections
Table 6: Measures of Inequality of Opportunity for ever-married Turkish women
Economic outcomes measure
Inequality measure
Parametric decomposition
Overall inequality
Total share of inequality of opportunity
Non parametric
Parametric
Partial shares associated with circumstances
Type of area
Birth region
Mother's education
Father's education
Mother tongue
Number of siblings
Asset index
(main)
Asset index
(subsidiary)
Imputed
consumption
Variance
Imputed
consumption
(no residuals)
Variance
Variance
Imputed
consumption
(no residuals)
MLD
Log
Imputed
consumption
6.01
[0.13]
4.14
[0.12]
26985
[1895]
57512
0.255
[0.007]
0.362
0.358
[0.012]
0.311
[0.012]
0.357
[0.014]
0.302
[0.014]
0.407
[0.023]
0.334
[0.021]
0.264
0.426
[0.015]
0.374
[0.016]
0.323
0.116
[0.009]
0.004
[0.004]
0.058
[0.008]
0.074
[0.009]
0.030
[0.005]
0.029
[0.006]
0.114
[0.009]
0.003
[0.004]
0.064
[0.009]
0.074
[0.011]
0.022
[0.005]
0.025
[0.005]
0.073
[0.007]
0.016
[0.005]
0.113
[0.018]
0.085
[0.015]
0.016
[0.004]
0.031
[0.005]
0.042
0.154
[0.011]
0.039
[0.009]
0.136
[0.016]
0.128
[0.015]
0.073
[0.008]
0.070
[0.010]
0.107
Variance
Observations
5229
Bootstrap standard errors in brackets. Bootstrap S.E.s not reported for imputed consumption with
residuals, given double boostrapping.
Source: TDHS 2003 with consumption imputed from HBS 2003. Sample: ever-married women,
aged 30-49
0.195
0.008
0.069
0.048
0.010
0.018
MLD
Log
0.262
0.025
0.097
0.090
0.051
0.049
Motivation Inequality of Opportunity Child Development ECD Reflections
Inequality of Opportunity for Wealth
Decomposition of wealth inequality in Turkey
100%
Efforts, luck and
circumstances not
measured
90%
80%
70%
60%
68%
50%
40%
30%
20%
10%
32%
Circumstances: Lowerbound estimation for
inequality in wealth asset
index attributable to
opportunities
Partial shares of opportunities contributing to
inequality
Birth region
1%
Birth Area
(Urban/Rural)
39%
Number of
siblings
9%
Language
10%
Mother's
education
18%
Father's
education
23%
0%
1
18
Motivation Inequality of Opportunity Child Development ECD Reflections
Opportunity profiles
• The non-parametric decompositions used to calculate
lower-bound estimates of a cardinal measure of
inequality of opportunity can also yield various
orderings of types:
– Complete orderings can be generated by choosing a particular
moment or percentile of the type-specific distributions, and
ordering them accordingly
– Two examples:
*  T1 , T2 ,..., TK  |  1   2  ...   K
– “Opportunity Profile”:
– “Opportunity-deprivation profile”:
   T1 , T2 ,..., T j ,..., TJ |     ...   ;    , k  J ; and
*
1
2
J
J
k
J 1
N
j 1
J
j
 N  N j
j 1
Motivation Inequality of Opportunity Child Development ECD Reflections
Opportunity profiles
Motivation Inequality of Opportunity Child Development ECD Reflections
Inequality of Opportunity for Educational Achievement
• Using PISA results (2006), between a quarter and a third of
overall educational achievement variation can be traced to
underlying inequality of opportunities.
• That range depends on alternative treatments for selection
into the PISA sample.
• While gender emerges as a key correlate of enrollment, as we
will report later on, it is not a significant correlate of
achievement.
• Family background variables are the key circumstances
“explaining” achievement.
• Books, durables and cultural possessions more important
than parental occupation.
22
Motivation Inequality of Opportunity Child Development ECD Reflections
Inequality of Opportunity for Educational Achievement
23
Motivation Inequality of Opportunity
Child Development ECD Reflections
The intergenerational transmission of opportunity
continues to shape children’s life chances today.
Intergenerational Opportunity Groups
Birth place of
Mother
Education of
maternal
grandparents
Low Intergenerational
Opportunity Group
Rural East
Less than primary
completed
High Intergenerational
Opportunity Group
Urban
West/Center
At least primary
completed
24
Motivation Inequality of Opportunity
Child Development ECD Reflections
Child nutritional outcomes differ considerably across the two
groups of children.
0
.1
.2
.3
.4
Height-for-Age Measure
Height for Age Measures
LINOG and HINOG
(indicates for
stunting)
-6
-4
-2
0
Standard deviations from mean
Lowest opportunity group of mothers
2
4
Highest opportunity group of mothers
Note: Stunting is often defined as height for age less than two SDs below the25mean.
Motivation Inequality of Opportunity
Child Development ECD Reflections
Relative Risk of Child Poverty, Stunting and Iodine efficiency
by Intergenerational
Opportunity
Relative Risk of Child
Poverty, StuntingGroup,
and Iodine2004
Deficiency,
by intergenerational opportunity group, 2004
2.0
1.7
1.5
1.5
1.0
0.8
0.5
0.0
-0.5
-0.7
-1.0
-0.6
-0.8
Child poverty (below age 5)
Stunting
Iodine deficiency
Note: Risk relative to average incidence for all children aged 5 and younger (poverty rate is 36.2%,
stunting rate is 12% and iodine deficiency rate is 50%).
Source: Staff calculations based on Turkey Demographic and Health Survey (2003/2004)
26
Motivation Inequality of Opportunity
Child Development ECD Reflections
Cognitive Development Inputs at Household Level for 36-47
month old children in Turkey
Cognitive Development Inputs at Household Level for
36-47 month old children in Turkey
86.8
90
77.1
80
65.3
Score
70
60
76.6
63.5
57.4
50.9
50
35.3
40
30
20
12.8
10
Mean for low SES
Mean for middle SES
Mean for high SES
Responsivity score (of mother towards the child)
Language stimulation score (of mother for the child)
Learning materials score (avaliable to child at home)
Source: Data from The Study of Early Childhood Developmental Ecologies in Turkey,
2008
27
Motivation Inequality of Opportunity
Child Development ECD Reflections
Cognitive Development Scores of 36-47 month old children
in Turkey
Cognitive Development Scores of
36-47 month old children in Turkey
Score
1.5
1.3
1.0
Tifaldi test- receptive
language score
0.5
Corsi test- short term
memory score
0.6
0.1
0.0
Mean for low SES
-0.5
-1.0
Mean for middle SES
Mean for high SES
-0.4
-0.8
-0.6
Source: Data from the Study of Early Childhood Developmental Ecologies in
Turkey, 2008, 36-47 month old children and their families.
28
Motivation Inequality of Opportunity
Child Development ECD Reflections
Enrollment Age Profiles by Opportunity Group and Gender,
(2004)
7
8
9
10
11
age
12
13
.6
14
15
.2
.5
.4
.6
.7
enrollment rate
.8
.8
.9
1
Enrollment-age profiles within low opportunity group
1
Enrollment-age profiles by opportunity group
7
8
9
10
11
age
low opportunity: grandparents did not go to school and mother born in a rural area of the East region
high opportunity: grandparents did go to school and mother born in an urban area of another region
Girls
12
13
14
Boys
29
15
Motivation Inequality of Opportunity
Child Development ECD Reflections
Under-5 Mortality Rate (2006, per 1,000 live births)
U5 mortality rate per 1000 live births
40
35
Indonesia
Mexico
30
Turkey
25
20
15
Brazil
Argentina
Russia
10
Chile
Poland
5
Korea
0
0
10,000
20,000
US
UK
Ireland
Spain
Germany
FranceJapan
Italy
Finland
30,000
40,000
50,000
GDP per capita (constant prices)
30
Motivation Inequality of Opportunity
Child Development ECD Reflections
Early Childhood Development
• Development of child begins pre-natally, learning
begins directly after birth
• All basic brain development takes place in early years
– on which later growth and learning ability depends
– Stunting (low height for age) in early ages reduces
intelligence
– Iodine deficiency leads to mental retardation
– Insufficient cognitive stimulation reduces learning
abilities
31
Motivation Inequality of Opportunity
Child Development ECD Reflections
Coverage of Various ECD programs in Turkey by Age
Category
Intervention
Coverage of Age Group
Pregnancy Monitoring, MOH (2004)
~ 80%
Immunizations - DPT, MoH (2004)
~ 90%
Growth and Psycho-Social Monitoring - Family Doctors, MoH (2008)
< 10 %
Private day-care centers, nurseries and community driven models (2008)
Preschool (Ages 4-5-6)
Parent training, MONE (2008)
Toddler and posttoddler (18-36
months)
Infancy (0-18
months)
Prenatal
Child Development Phase
SHÇEK Community Centers (2008)
~3%
100 %
coverage line
<1%
Parent training, MONE (2008)
~ 3%
Growth and Psycho-Social Monitoring - Family Doctors, MoH (2008)
< 10 %
< 2%
Home-based MOCEP - for 6 year group only, MONE (2008)
~ 6%
Private preschools (2008)
~ 1%
Public preschools & kindergartens - for 5-6 year group, MONE (2008)
~ 30 %
0%
20%
40%
60%
80%
100%
32
Motivation Inequality of Opportunity
Child Development ECD Reflections
Pre-primary Gross Enrollment for 3-6 year olds (%)
2006
120
France
Germany
Mexico
Preprimary Enrollment (Gross)
100
Korea
Italy
Sweden
Russia
80
Japan
UK
60
Finland
Poland
Chile
40
US
Indonesia
Kazakhstan
20
Turkey 2006
0
0
10,000
20,000
30,000
40,000
50,000
GDP per capita (constant prices)
33
Motivation Inequality of Opportunity
Child Development ECD Reflections
Pre-primary
Enrollment
3-6 year olds (%)
Graph
Z: PreschoolsGross
and Daycare
Access, byfor
socioeconomic
status and rural/urban 2006
location [2008]
90
Mothers in sample (%)
76.6
76.2
80
73.9
68.2
70
62.7
60
54.0
50
40.7
34.7
40
26.2
30
20
18.8
10
0
Low SES
Middle SES
High SES
Socio-economic status
There is pre/nursery school in neighborhood
Rural
Urban
Location
There is elementary school in neighborhood
Note: Socio-economic status is defined as an index that combines income, household assets and the
educational level of adults in the household.
Source: The Study of Early Childhood Developmental Ecologies in Turkey (2008), Koc University.
34
Motivation
Inequality of Opportunity Child Development ECD Reflections
Graph ZZZ: Pre-School Enrolment [2008] and Human
Pre-School Enrolment
and (2004),
Human by
Development
Development
Province Index, by
Province
% of children in preschool (%)
60
Sakarya, 54.1
50
Amasya, 48.7
40
Trabzon, 40.2
Erzurum, 37.6
Burdur, 36.8
30
Siirt, 28.7
20
Istanbul, 14.6
Van, 11.7
Ağrı, 8.6
10
0
0
20
40
High HDI
Ranking of Province in HDI index
60
80
Low HDI
35
Motivation Inequality of Opportunity
Child Development ECD Reflections
• Spending-incidence by age group:
• Centrally financed expenditures that are not contributionfinanced
• Age-beneficiary relationship derived for each individual
budgetary program
• Health: age-specific beneficiary weights from existing
utilization/cost study. Also applied to non-contributory
treasury financing of SSI deficit;
• Education: each functional item with specific breakdown
(e.g., part of primary expenditure going to pre-school age)
• SP: program by program according to nature
• Simulations: how does intra-household distribution of cash
transfers (especially pensions) alter distributional picture?
36
Motivation Inequality of Opportunity
Child Development ECD Reflections
Per Capita Social Expenditures by Age Group in Turkey
Per Capita Social
Expenditures by Age
Group in Turkey
(2008) Security
(Central
Government,
Excluding
Social
(Central Government, Excluding Social Security Contributions)
Contributions, 2008
2.5
Education
TL in 2008
Social Protection
2.0
Health
1.5
1.0
0.5
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
Age Group (0-90)
37
Motivation Inequality of Opportunity
Child Development ECD Reflections
Benefit Simulations of ECD policies: Pre-School Education
• Impact evaluation of one year pre-school programs: one-year
additional net schooling impact (Ertem et al, 2006, Kaytaz 2005)
• Microsimulation approach a la Bourguignon, Ferreira and Leite
(2008)
• Increases in education attainment based on individual/family
characteristics and given mean expansion at national level
• Fertility choices
• Occupational choices (unemployment, formal, informal,
independent)
• Earnings changes (formal, informal, independent) – constant
returns assumption
38
Motivation Inequality of Opportunity
Child Development ECD Reflections
Benefit Simulations of ECD policies: Pre-School Education
benefits!
• Expanding education of the 20-39 year olds today by one year
(equal to measured impact of pre-school education)
• Two channels:
increased education on earnings
occupational distribution
• Impacts:
8 percent increase of mean earnings of all
individuals 20-39 years of age
14 percent increase in female labor force
participation rate
5 percent reduction of poverty
39
Motivation Inequality of Opportunity
Child Development ECD Reflections
Aggregate Income and Poverty Reduction Effects of
Aggregate Female
Income and
Poverty
Reduction
Effects of
Increased
Labor
Force
Participation
Increased Female Labor Force Participation
10.0
6.9
3.4
5.0
% points
0.0
-5.0
-10.0
-8.0
-15.0
-15.0
-20.0
Poverty Reduction
Full-time
Income Growth
Part-time
40
Motivation Inequality of Opportunity
Child Development ECD Reflections
A. inequality of opportunity shown for determining
education and wealth outcomes. Region less
important than area;
B. inequity travels across generations: children’s life
chances are significantly impacted by their
grandparents socio-economic profile
C. child development diverges strongly across
opportunity groups and poverty at very early ages,
especially for girls. Traditional solidarity system
does not compensate for disadvantage
41
Motivation Inequality of Opportunity
Child Development ECD Reflections
D. Pro-equity policies would aim to weaken the link
between circumstances and outcomes in early ages –
early childhood development policies
E. Today, social expenditures are not reaching children
below school age and coverage of disadvantaged
children is low
F. Quo vadis, social policy?
42