Educational Policies and Income Inequality

Saturday, 4 July 2015: 8:30 AM-10:00 AM
TW2.1.04 (Tower Two)
Herman van de Werfhorst, University of Amsterdam, Amsterdam, Netherlands
Daniele Checchi, University of Milan, N/A, Italy
Thanks to the abundant availability of comparative data on student achievement, we have come to know a lot about the association between characteristics of educational systems and student learning. Student achievement tests like the Programme of International Student Assessment (PISA), the Trends in International Mathematics and Science Study (TIMSS) and the Programme for International Reading and Literacy Study (PIRLS) have helped us to understand whether educational policies are related to the distribution of student performance, and inequality in performance by social and ethnic background (Hanushek and Wössmann 2011; Kerckhoff 1995; Van de Werfhorst and Mijs 2010; Brunello and Checchi 2007; Marks 2005; Cobb-Clark et al. 2012). Educational policies that received attention include, among others, tracking, accountability regulations, and length of compulsory schooling.

            However, there are three shortcomings in the present literature. First, most of the studies have looked at educational policies in a static way, whereas policies in fact change over time within countries. Given the cross-sectional nature of the comparative datasets the cross-sectional focus is understandable, but to assess whether policies affect outcomes (in a causal meaning) one would also want to exploit within-country temporal variations. Second, the literature has mostly concentrated on student achievement at a relatively young age (mostly students aged between 8 and 15). Other outcomes in educational distributions, most notably the attained level of education, has received far less attention from a policy perspective. In other words, most attention has been paid to the ‘quality’ of education rather than its ‘quantity’. This is unfortunate, because we do not know whether policies that are related to the quality of education are also related to the quantity of schooling in societies. Thirdly, studies focusing on educational policy have mostly addressed educational outcomes. It is heavily understudied whether educational policies have, through their effects on educational distributions, also repercussions on income and earnings distributions in societies. In our view it is crucial to understand whether policies affect educational and income distributions, because the effectiveness of educational systems may be assessed not only in terms of the skills and qualifications that are produced, but also in terms of the stratification in society that is generated through educational policies. After all, educational policy is often geared towards preparing youth for adult life, so to understand whether this preparation is successful one needs to study distributions in earnings and income in addition to distributions in educational outcomes. 

            In this paper we contribute to understanding the associations between educational policies, educational distributions and income distributions by combining four sorts of data. First, as a starting point we have exploited comparative mathematics achievement data from various years: the First International Mathematics Study of 1964 (FIMS64), the Second International Mathematics Study of 1980-1982 (SIMS80), and the Trends in International Mathematics and Science Study of 1995 (TIMSS95). This collection provides us with mathematics achievement data, as indicator of the ‘quality’ of education, for a multitude of cohorts and countries, split out by gender. Second, for the cohorts and countries for which data were available in these student achievement tests, we have collected data on educational policies, including policies on compulsory education, school and/or teachers’ autonomy, and tracking age (see Braga et al. 2013). Third, using data from Eurostat, we have collected data on educational attainments measured by the (median) years of schooling required to attain a specific degree, following the ISCED classification. This is again done for each combination of cohort, country and gender separately. Fourth, for each combination of country, cohort and gender we have calculated income inequalities using the European Community Household Panel (ECHP) and the European Union Statistics on Income and Living Conditions (EUSILC), the latter being official European Union data on income statistics.

These four data sources enable us to study whether educational policies are related to the distribution of quality and quantity of education, and whether policies and educational distributions (of both quality and quantity) are related to income inequality. By controlling for country-specific and time fixed effects, and by separating age and cohorts effects, we believe that our assessment of policy effects is stronger than in most other studies. Our results indicate that inequality in education (measured both at quality and quantity levels) affect earnings inequality. In addition, since educational inequality respond to educational reforms, we are able to identify educational policies (like later entry into compulsory education or introduction of standardised tests) capable to reduce income inequalities thirty years later.