Earnings, education and competences: can we reverse inequality? Daniele Checchi (University of Milan and LIS Luxemburg) 1
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. 2
from Brunello-Fort-Weber EJ 29 3
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. population frequency.14.12.1.8.6.4.2 Inclusive policies 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 17 18 19 2 21 4
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..14 Selective policies population frequency.12.1.8.6.4.2 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 17 18 19 2 21 5
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? Belgium Czech Republic Denmark Estonia Finland France Germany Ireland Italy Netherlands Norway Poland Slovak Republic Spain -5 5-5 5-5 5 Austria -4 England 2 4-4 -2 2 4 Northern Ireland -5 5 Sweden -2-4 -2 2 4-4 -2 2 4-4 -2 2 4 years of schooling (standardised) 6
Where should the policies attack educational inequalities? ability demographics (gender, ethnicity) family background (parental education, books youth competences (PIRLS, TIMMS, PISA) educational attainments (degrees, years of education) adult competences (PIACC) LM participation / employability labour earnings / incomes family schooling labour market 7
Eric A. Hanushek, Guido Schwerdt, Simon Wiederhold and Ludger Woessmann. 213. Returns to Skills around the World: Evidence from PIAAC. IZA DP No. 785 8
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. 9
Inequality in competences, years of schooling, gross labour earnings (from dependent employment and from total employment) gini index total nonnegative gross labour earnings.25.35.14.16.18.22.3.4.2 gini index dep.employees gross earnings SI SK CZ HU AT IE LV NO IT EL GB FR PT ES.8.1.12.14.16 gini index years of education from attainment SI CZ SK HU AT IE LV NO IT.8.1.12.14.16 gini index years of education from attainment EL GB FR ES PT BE BE.35.45.25.35.3.4.5.3.4 AT CZ FR BE SK HU SI IE LV NO.14.16.18.2.22 gini index math test score IE BE SK EL CZ ES SI NO AT ES GB IT.25.3.35.4 gini index dep.employees gross earnings PT PT FR IT HU EL GB LV 1
Income inequality impact of educational reforms (reduced form) Impact on Gini index on dependent employment earnings of one standard deviation increase in reform variables -.15 -.1 -.5.5.1 public preprimary compulsory end age standardised test teacher autonomy compulsory begin age tracking age school accountability university access 11
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 25 and 211 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. 12
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 3 to 6) 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. 13
Inequality of opportunity EUSILC 25 and 211.25 Inequality of opportunity.15.2.5.1 Denmark Slovenia Sweden Finland Lithuania Portugal Norway Czech Republic Croatia Latvia Bulgaria Estonia 25 211 95% CI France Belgium Hungary Poland Spain Italy Ireland Romania Great Britain Austria Netherlands Greece Germany Luxembourg Switzerland 14
Total and opportunity inequality EUSILC 25 and 211 Total inequality and opportunity inequality - disposable incomes 25 211 Ex-ante opportunity inequality - gini disposable incomes.15.2.1 AT IT BE FR NO CZ SI EE LV LT LU EL ESIE UK HU PT PL.5.25.15.2 BE CZ NO SI AT CH LU EL UK RO IT PL BG HU IE FR EE HR ES PT LT LV.5.25.1.2.25.3.35.4 Total inequality - Gini disposable incomes imputing exp.incomes.2.25.3.35.4 Total inequality - Gini disposable incomes imputing exp.incomes 15
.25 Changes in total and opportunity inequality - 25-11 LU AT ES IE EL.15 BE NO CZ IT FR EE LV LT UK HU PT PL SI.5.2.1.2.25.3.35.4 Total inequality - Gini disposable incomes imputing exp.incomes 16
Educational expenditure and inequality of opportunity ex-ante opportunity inequality imputing exp.incomes 1 15 2 25 EL LU LU EL RO ES IE CH UK AT IT UKAT HU PL IT BE PLPT EEFR BG HU LV IE CZ EE FR CZ ES LTLV LT SI PT SI BE NO NO 1 15 2 25 IE IE CH UK 5 AT ES UKATIT RO PL HU PT BE PL IT EE FR BE LV NO HU EE FR CZ CZ ES LT NO PT SI SI LT LV BG 5 3 4 5 6 7 8 Public expenditure in education as % of GDP 5 1 15 2 Expenditure in pre-primary education as % of education expenditure 17
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 18