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