earnings, education and competences: can we reverse inequality · 2016. 10. 14. · 4 but...
TRANSCRIPT
1
Earnings, education and competences:
can we reverse inequality ?
Daniele Checchi
(University of Milan and LIS Luxemburg)
2
Educational policies are often invoked as good instruments for reducing income inequality. Do we possess strong empirical evidence ? We know that some reforms (for example increase in compulsory education) increase schooling, with heterogeneous impact among genders and proxies for abilities. However unobservable ability and/or sorting of individuals makes it difficult to obtain reliable measure of the causal impact of educational policies. Educational policies are difficult to measure, since they capture an institutional change, which can be more qualitative than quantitative.
3
from Brunello-Fort-Weber EJ 2009
4
But educational reforms can work in different point of the ability distribution. Reforms extending pre-primary schooling and/or expanding the access education (via raise in leaving age for compulsory education or in tracking age, removing barriers to university admissions) and/or increasing teacher qualifications exhibit positive correlation with average years of education in the population and negative one with inequality and intergenerational persistence. Let us label these reforms as inclusive.
Inclusive policies
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
popu
latio
n fr
eque
ncy
5
Reforms increasing school autonomy and accountability as well as university autonomy are also positively correlated with mean educational attainment, but also with inequality and persistence. Similar properties are also associated to reforms related to financial support to university students. Let’s identify these reforms as selective.
Selective policies
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
popu
latio
n fr
eque
ncy
6
Human capital embodies both quantity (formal schooling, certification) and quality (competences) dimensions: raising one does not necessarily implies raising the other. The two are correlated but which is exogenous ?
-50
5-5
05
-50
5-5
05
-4 -2 0 2 4 -4 -2 0 2 4
-4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4
Austria Belgium Czech Republic Denmark Estonia
Finland France Germany Ireland Italy
Netherlands Norway Poland Slovak Republic Spain
Sweden England Northern Ireland
nu
mera
cy (
sta
nda
rdis
ed
)
years of schooling (standardised)
7
LM participation / employability
Where should the policies attack educational inequalities ?
family schooling labour market
ability
demographics (gender, ethnicity)
family background
(parental education, books
at home)
youth competences
(PIRLS, TIMMS, PISA)
educational attainments
(degrees, years of education)
adult competences (PIACC)
labour earnings / incomes
8
Eric A. Hanushek, Guido Schwerdt, Simon Wiederhold and Ludger Woessmann. 2013. Returns to Skills around the World: Evidence from PIAAC. IZA DP No. 7850
9
Unfortunately we observe competences when adult, ignoring what may have occurred when people were young. We would need more longitudinal datasets where we observe test scores when young, schooling experience, labour market transitions and competences when old. Recall of past events does not solve the problem, since people tend to make their lives coherent when recalling.
10
Inequality in competences, years of schooling, gross labour earnings (from dependent employment and from total employment)
ATBE
CZ DE
DK
EL
ES
FI
FR
GB
HU
IE
IT
LVNLNO
PTSE
SI
SK
.14
.16
.18
.2.2
2
gin
i in
dex
ma
th te
st s
core
.08 .1 .12 .14 .16gini index years of education from attainment
AT
BE
CZ
DE
DK
EL
ES
FI
FRGB
HU
IE
IT
LV
NL
NO
PT
SE
SI
SK
.25
.3.3
5.4
gin
i in
dex
dep
.em
plo
yee
s g
ross
ea
rnin
gs
.14 .16 .18 .2 .22gini index math test score
AT
BE
CZ
DE
DK
EL
ES
FI
FRGB
HU
IE
IT
LV
NL
NO
PT
SE
SI
SK
.25
.3.3
5.4
gin
i in
dex
dep
.em
plo
yee
s g
ross
ea
rnin
gs
.08 .1 .12 .14 .16gini index years of education from attainment
ATBE
CZ
DE
DK
EL
ES FIFR
GBHU
IE
IT
LV
NL
NO
PT
SE
SI
SK
.3.3
5.4
.45
.5
gin
i in
dex
tota
l no
nne
ga
tive
gro
ss la
bou
r e
arni
ng
s
.25 .3 .35 .4gini index dep.employees gross earnings
11
Income inequality impact of educational reforms (reduced form)
-.15 -.1 -.05 0 .05 .1
of one standard deviation increase in reform variables
Impact on Gini index on dependent employment earnings
public preprimary compulsory begin age
compulsory end age tracking age
standardised test school accountability
teacher autonomy university access
12
We may choose a more modest goal: ensuring equality of opportunities to every citizen, irrespective of the final outcome. How to measure inequality of opportunities ? Following Roemer, one may think that all differences attributable to circumstances are to be considered unfair. Data from the 2005 and 2011 waves of the European Survey on Income
and Living Conditions (EUSILC) specific modules data for attributes of each respondent's parents during her childhood period when aged 14-16.
13
We consider individual characteristics as circumstances independent from individual responsibilities: parental education (low, intermediate, secondary, college) gender native (country of birth being the same of the country of residence) age (six cohorts from 30 to 60)
We account for the existence of unemployed imputing to all individuals with zero income their expected income (namely the conditional income corrected by the probability of self-selection into employment). Measuring inequality of opportunity and comparing with total inequality shows that the two concepts are not coincident.
14
Inequality of opportunity – EUSILC 2005 and 2011
.05
.1.1
5.2
.25
Gin
i in
dex -
dis
posab
le in
co
me
s im
puting
exp.incom
es
Den
mar
k
Slo
veni
a
Swed
en
Finland
Lith
uania
Por
tuga
l
Nor
way
Cze
ch R
epub
lic
Cro
atia
Latvia
Bul
garia
Est
onia
Franc
e
Bel
gium
Hun
gary
Pol
and
Spa
in
Italy
Irela
nd
Rom
ania
Gre
at B
ritai
n
Aus
tria
Net
herla
nds
Gre
ece
Ger
man
y
Luxe
mbo
urg
Switz
erla
nd
2005 2011 95% CI
Inequality of opportunity
15
Total and opportunity inequality – EUSILC 2005 and 2011
AT
BE
CZ
DE
DK
EE
EL
ES
FI
FR
HU
IE
IT
LT
LU
LV
NL
NO
PLPT
SESI
UK
.05
.1.1
5.2
.25
Ex-
ant
e o
pp
ortu
nity
ineq
ua
lity
- G
ini d
isp
osa
ble
inco
mes
.2 .25 .3 .35 .4Total inequality - Gini disposable incomes imputing exp.incomes
2005
AT
BE BG
CH
CZ
DE
DK
EE
EL
ES
FI
FRHR
HU IE
IT
LT
LU
LV
NL
NO
PL
PT
RO
SESI
UK
.05
.1.1
5.2
.25
Ex-
ant
e o
pp
ortu
nity
ineq
ua
lity
- gi
ni d
isp
osa
ble
inco
me
s
.2 .25 .3 .35 .4Total inequality - Gini disposable incomes imputing exp.incomes
2011
Total inequality and opportunity inequality - disposable incomes
16
AT
BE
CZ
DE
DK
EE
EL
ES
FI
FRHU
IE
IT
LT
LU
LV
NL
NO
PLPT
SESI
UK
.05
.1.1
5.2
.25
Ex-
ante
op
port
un
ity in
equ
alit
y -
Gin
i dis
po
sab
le in
com
es
.2 .25 .3 .35 .4Total inequality - Gini disposable incomes imputing exp.incomes
Changes in total and opportunity inequality - 2005-11
17
Educational expenditure and inequality of opportunity
AT
AT
BE
BEBG
CH
CZCZ
DE
DE
DK
DK
EE
EE
EL
EL ES
ES
FIFI
FR
FR
HU
HU
IE
IE
IT
IT
LT
LT
LU
LU
LV
LV
NL
NL
NO
NO
PLPLPT
PT
RO
SESE
SI
SI
UK
UK
51
01
52
02
5
ex-
an
te o
pp
ortu
nity
ine
qua
lity
imp
utin
g e
xp.in
com
es
3 4 5 6 7 8Public expenditure in education as % of GDP
AT
AT
BE
BE BG
CH
CZCZ
DE
DE
DK
DK
EE
EE
ES
ES
FIFI
FR
FR
HU
HU
IE
IE
IT
IT
LT
LT
LV
LV
NL
NL
NO
NO
PLPLPT
PT
RO
SESE
SI
SI
UK
UK
51
01
52
02
5
ex-
an
te o
pp
ortu
nity
ine
qua
lity
imp
utin
g e
xp.in
com
es
0 5 10 15 20Expenditure in pre-primary education as % of education expenditure
18
Summing up: educational attainment should incorporate schooling and achievements both dimensions are endogenous, being correlated with parental background and unobservable abilities educational reforms affect the distribution of both schooling and competences not clear whether one dimension dominates the other one would need to ascertain how competences are formed, and whether
they are primitive measures (i.e. prior to schooling experience) longitudinal surveys and/or administrative data can answer this question