7. Inequality in educational achievement

Bas van Leeuwen
International Institute of Social History
Jieli Li
International Institute of Social History

As noted recently by the OECD Secretary-General Angel Gurría, “the dream of ‘quality education for all’ is still far off”.1 Indeed, even today, education is far from being universally available, with educational inequality having wide implications for both individuals and societies. First, educational differences translate into inequalities in a range of other well-being outcomes, such as earnings, longevity, morbidity and subjective well-being.2 Second, higher educational inequality may also reduce benefits for society as a whole, for example, a lower demand for public goods such as health care, threats to the rule of law, a lower investment in R&D and the adoption of new technologies, or higher crime rates (Card and DiNardo, 2002[1]). A third rationale for reducing educational inequality is equity and equality of opportunity. Access to education must be fair, i.e. not dependent on personal characteristics, and inclusive, i.e. with the same rules applying to all (Field, Kuczera and Pont, 2007[2]).3

Educational inequality thus has significant effects on both individuals and societies. This explains the wide scholarly interest in the topic. Many studies such as Meyer (2011[3]) have shown educational inequality in a country to be strongly related to the historical development of its education system.4 This is one reason why this chapter documents trends in educational inequality from the 19th century to the present. We do this, following the literature, by using relative and absolute measures of inequalities in the completion of formal education. After describing concepts, sources and data quality, we present the main features identified from the historical evidence. Overall, our evidence shows a strong decline in the Gini coefficient of educational attainment, which is caused mostly by a reduction in the share of people without any formal education. Measures of absolute inequality show an increase in the first half of the 20th century and a decline afterwards. We conclude by briefly discussing the link between educational inequality and GDP per capita, and by highlighting some priorities for future historical research.

Education, defined as the facilitation of learning, has varied strongly across countries and over time. In order to enhance international comparability, an International Standard Classification of Education (ISCED) was created in 1976. Even though this prompted some standardisation of educational statistics, it has not been fully implemented by all countries. Even though most countries are currently using the 2011 ISCED classification when reporting their educational statistics, the basic definitions remain the same as those used in previous versions, with formal primary education starting roughly from age 5-7, with a duration of around 6 years, and offering basic reading, writing and arithmetic; secondary education, covering lower and higher secondary schools, also commonly lasting 6 years in total; and tertiary education usually lasting around 3 or 4 years (UNESCO Institute for Statistics, 2012[4]). As a continuous measure is needed for the purpose of computing educational inequalities, some scholars have mapped these layers into mean years of schooling, e.g. Morrisson and Murtin (2009[5]) and Lee and Lee (2016[6]).

A commonly used measure of inequality is the Gini coefficient, which has the property of being independent of mean and population size and it is easy to interpret, ranging between 0, in the case of perfect equality, and 1 (or 100 if reported as an index) in the case of perfect inequality. For those reasons, this measure is widely used in the literature on educational inequality, e.g. Lopez, Thomas and Wang (1999[7]), Thomas, Wang and Fan (2003[8]); Földvári and Leeuwen (2011[9]).5 Other authors prefer alternative measures for two main reasons.

First, being a relative measure, the Gini coefficient does not capture the scale effect of higher educational attainment: if, for every individual, the number of years of education doubles, the Gini coefficient does not change even as the absolute difference in the number of years in education between any two individuals doubles. Whether this is important or not depends on how one views inequality. If one considers inequality as intrinsically relative (e.g. person A has 50% more years of education than person B), then the Gini coefficient is the preferred measure; conversely, if one considers that what matters are absolute differences between people (e.g. person A has 5 years more education than person B), then an absolute measure such as the standard deviation is preferable. As pointed out by Amiel and Cowell (1999[10]) even though a sizeable group of respondents to surveys considers absolute differences as important, most respondents consider inequality as a relative concept.

Second, the Gini coefficient declines when the average number of years of education increases. That is, changes in the value of the Gini coefficient for education are, to a large extent, driven by the declining share of persons without formal education (Castelló and Doménech, 2002[11]; Morrisson and Murtin, 2009[5]). On the contrary, absolute measures of inequality6 – which include the absolute Gini (the Gini index times mean years of education) (Jenkins and Jäntti, 2005[12]; Niño-Zarazúa, Roope and Tarp, 2017[13]; Bandyopadhyay, 2018[14]) and the standard deviation (Ram, 1990[15]; Meschi and Scervini, 2014[16]) – record the absolute dispersion (in years of education) of education among individuals in a society. With absolute measures, when more people complete basic education, inequality will initially rise but, when more than 50% of the population has achieved basic education, additional persons with education will cause inequality to decline. For those reasons, absolute measures of inequality, such as the standard deviation, have been used alongside the Gini index to provide a more complete picture of educational inequality, e.g. Ram (1990[15]).

Because of a lack of systematic historical censuses, educational inequality is often calculated based on datasets of educational attainment (the highest level of formal education completed by each person), converted into a continuous measure of “years of schooling”. There are a great number of such databases, e.g. Cohen and Soto (2007[17]), Barro and Lee (2013[18]), de la Fuente and Doménech (2015[19]), and Samir and Lutz (2017[20]), but these data refer mostly to the period after 1950, with two datasets providing projections for the future (Barro and Wha Lee, 2015[21]; Samir and Lutz, 2017[20]). Yet, there are fewer datasets covering historical periods, e.g. Morrisson and Murtin (2013[22]); Van Leeuwen and Li (2014[23]); Tamura et al (2019[24]); and Lee and Lee (2016[6]). Only two datasets have the advantage of covering a broad range of countries spanning back to 1870. The first is the one presented by van Leeuwen and Li (2014[23]), which is based on a combination of the dataset by Morrisson and Murtin (2009[5]) and by Földvári and van Leeuwen (2014[25]); the second is the one from Lee and Lee (2016[6]).

Choosing which dataset to use is complicated since all have their merits. An advantage of the Van Leeuwen and Li dataset is that it is annual, while the Lee-Lee data are quinquennial. Conversely, Lee and Lee (2016[6]) provides data for men and women separately. A further difference is that the Van Leeuwen and Li dataset covers the population aged 15 and older, while the Lee-Lee data covers the population aged 15-64. This obviously causes some differences in trends, but levels of both databases have very high correlations of close to 96%. For these reasons, both databases are often used in historical research.

Yet, there are also more critical observations about both datasets. For example, de la Fuente and Doménech (2015[19]) note that, for OECD countries after 1950, the reliability of the Barro-Lee database, on which the Lee-Lee data is partially based, ranks considerably below other available series. Conversely, Lee and Lee (2016[6]) argue against the use of literacy data in the Van Leeuwen and Li dataset to gauge educational attainment in all types of educational institutions (i.e. both public and private), preferring to limit their remit to public schools. Yet, since private schools make up a significant portion of schools both in Europe and in the non-Western world, using data limited to public schools implies a significant underestimate of attainment when going back in time, leading to higher estimates of inequality. For example, the share of the population over age 15 with some attainment (irrespective of the level) in 1870 in China is 24.8% in the Van Leeuwen and Li dataset and 0% in Lee and Lee. To assess the reliability of both estimates, one may look at Chinese enrolment ratios for 1870. These vary between close to 0% (for Yugan county in Jiangxi province) to over 40% (for Chongning county in Sichuan province), with an average of close to 20% (Yugan, 1947[26]; Cao, 2013[27]), thus providing evidence in favour of the higher estimate of educational attainment reported by Van Leeuwen and Li. Similar observations apply to other countries. These differences in mean years of schooling have significant effects on inequality estimates: whereas, using the Van Leeuwen and Li data, the global Gini coefficient is 79.0 in 1870 (Table 7.2), Lee and Lee provide a measure of 90.5. Because of these considerations, the dataset from Van Leeuwen and Li is used in this chapter to calculate educational inequality over the period 1870-2010.

Our assessment of the quality of the educational inequality estimates is based on the method used to collect the underlying data, sources and indicators; we rate this quality as high quality (1), medium quality (2), moderate quality (3) or low quality (4). An assessment of the quality of data on mean years of education was already reported in van Leeuwen and Li (2014[23]). The data used for this chapter, being a slightly improved version of those in van Leeuwen and Li (2014[23]), are quite comparable.

The quality measures are reported in Table 7.1. Starting with the data from 1950 to 2010, these are sourced from Van Leeuwen and Li, which for the modern period are based on direct education data and for earlier years on back-projection. Starting with the most recent period, comparison of these attainment-based data with the highest quality level individual-level data available from the International Social Survey Programme (ISSP) Research Group (2017[28]) results in a strong correlation. Since these ISSP data are high quality, but not formally produced by an official statistical agency, we assess data quality at level 2 for all world regions since 1980, as well as for Europe, the Western Offshoots, and Latin America and the Caribbean for the period between 1950 and 1980. Varying by country, most data between 1870 and 1950 are based on a backward extrapolation of censuses using various data such as population and enrolments, which deteriorate in quality the further one goes back in time; hence, these are assessed at level 3.

We start by reporting global trends in educational inequality. Table 7.2 shows that the Gini coefficient of years of schooling at the world level decreases over time. This decline mainly reflects a fall in the share of people without any formal education, from around 73% in 1870 to 43% in 1950 and 15% in 2010. The standard deviation points to increasing world inequality.

Yet, this global pattern obscures the trends for the various regions of the world. As shown in Table 7.3, the Gini coefficient of educational inequality declines in all regions, but more so for less developed regions. For the standard deviation, regions with lower education levels (all regions except Western Europe and the Western Offshoots) experienced rising inequality up to around 1990. Western Europe and the Offshoots, however, show broadly constant standard deviations.

It is important to note that, even though for both measures of inequality trends among all world regions are roughly similar, this does not apply to their levels. Whereas in terms of the Gini coefficient, the Middle East and North Africa, sub-Saharan Africa, and Latin America and the Caribbean display the highest levels of inequality, for the standard deviation prior to 1950 Western Europe and the Western Offshoots had the highest levels of inequality, arguably because of the scale effect of higher average levels of education.

One question is whether cross-country differences in educational inequality can be attributed either to structural differences in countries’ education systems or simply to lagging development in mean years of schooling. To test this, Figure 7.1 looks at whether some countries have a different Gini index when they reach a certain level of mean years of education later than other countries. We find no evidence of rising or declining inequality for the (mostly developed) countries that reached an average of 5 years of schooling before the 1950s, suggesting that, within this group of countries, differences in inequality are mainly due to a lagging educational development. Even though there is no strong effect, we do see some evidence of higher educational inequality among the (mostly less developing) countries that reached an average of 5 years of schooling after around 1950, thus suggesting that countries experiencing educational expansion only recently featured increasing structural inequalities in their education systems.

These patterns of less equal education systems in less developed countries have to be viewed in a historical perspective. Indeed, the highest literacy rates (implying a lower Gini index) in the 19th century occurred in Western Europe and the Western Offshoots. These high literacy rates reflected the existence of an education system that was linked to economic, religious and political developments of individual countries (Fuller, Edwards and Gorman, 1986[29]; Boli, 1989[30]). A very different pattern prevailed in former colonies, whose education system was often partly modelled on the coloniser’s conditions rather than locally prevailing conditions (Bolt and Bezemer, 2009[31]). The importance of this factor can be seen when examining why the Gini index in former French colonies is higher than that of former British colonies (Garnier and Schafer, 2006[32]). For example, Cogneau and Moradi (2014[33]) try to isolate this effect of the coloniser country and conclude that the lower educational Gini of former British colonies can be attributed to the more flexible British system based on missionary schools (White, 1996[34]), which connected better to local society than the highly regulated French system and hence resulted in higher literacy rates (Cogneau and Moradi, 2014[33]).

This historical persistency does not mean that the educational systems of countries are always inexorably tied to their past. Educational development depends, to a large extent, on government policy. As pointed out in the introduction, reducing educational inequality can deliver both private benefits (benefitting the individual who invested in education) and social returns (accruing to society as a whole), such as enhancing equity and opportunity, and the priority attached to each goal can change over time. After World War II, capitalist countries generally pursued an educational policy aimed at enhancing private returns to education by stimulating post-primary education. Other countries, such as socialist countries, often started mass literacy campaigns based on the expansion of basic education, which enhanced social returns and furthered equity (Stites and Semali, 1991[35]). In China, for example, Mao’s education drive between 1950 and 1970 resulted in a reduction of the share of persons with no education from 61% to 32%, causing a drop in the Gini coefficient of East Asia over this period. Likewise, Dupraz (2019[36]) studied the effect of the colonial legacy for Cameroon, which was divided into English and French parts, but reunited after independence. Using this as a natural experiment, he found that, even though people born in the English part of the country had, on average, one year more schooling than those born in Cameroon’s regions under French rule, this difference disappeared after World War II due to a policy of stimulating schools in the French part. It has to be stressed though that, rather than conforming to an ideal-type, policies are frequently mixed or changing over time, e.g. Samoff (1991[37]). For example, in the 1950s-1970s, various socialist countries moved away from their focus on social returns and equity and prioritised increasing private returns instead.

Large declines in the Gini index of educational inequality occurred in all world regions. Yet, as pointed out before, this does not apply to the standard deviation, as higher shares of persons with some education imply higher standard deviations up to the point where more than 50% have some level of education, after which the standard deviation declines again. These trends are often captured in the literature by the notion of an educational Kuznets curve (Ram, 1990[15]; Thomas, Wang and Fan, 2003[8]; Castelló and Doménech, 2002[11]). This notion implies that, when a country moves from zero to maximum education, it will first experience an increase in inequality and later a reduction (Ram, 1990[15]), creating an inverse U-shaped curve (see Figure 7.2). In 1870, the United States, with a standard deviation of 2.77 and an average of 5.6 years of schooling, was on the declining path of this curve, although inequality there was still higher than in most developing countries, which remained on the path of rising educational inequality. This inflection point for educational inequality had however significantly increased by 2010. As found by Ram (1990[15]) and Thomas, Wang and Fan (2003[8]), the standard deviation in years of schooling peaks at around 7 years of education, i.e. a level around 3.5 years higher than in 1870. In 1870, only Western Europe (with the exception of some southern European countries) was on the declining portion of the curve. Hence, increasing educational attainment in Western Europe has, since 1870, led to lower inequality when measured by the standard deviation.

Looking at other world regions, various countries from sub-Saharan Africa were in 1870 on the increasing part of the curve, and remain there even today. In terms of policy, an absolute measure of educational inequality such as the standard deviation implies that less developed countries face a trade-off between increasing the mean years of education and reducing inequality, while developed economies, whose mean years of schooling are well above 7, can, by investing further in education, reap both benefits at the same time.

No such trade-off exists when considering a relative measure of educational inequality such as the Gini index. Yet, this does not mean that countries that have passed the 7-year threshold do not face choices in the structure of their educational investments: investments in tertiary education are more likely to increase inequality, even expressed in the Gini index, than investments in basic education.

Since, as described in this chapter, measures of educational inequality may, at times, differ substantially from one another, it is unsurprising that the same differences exist in the correlation of these measures with per capita GDP. Looking at the Gini index, based on the models from Lopez, Thomas and Wang (1999[7]) and Földvári and van Leeuwen (2011[9]), we might expect a negative effect, i.e. the higher the per capita GDP, the lower the educational inequalities. Yet, while some studies found this effect to be significant (Lopez, Thomas and Wang, 1999[7]; Castelló and Doménech, 2002[11]), others, correcting for missing variables such as physical capital, found it to be small or insignificant (Földvári and van Leeuwen, 2011[9]; van Damme, 2014[38]). Since our correlations do not correct for missing factors, Figure 7.3 shows a negative and significant correlation between GDP per capita and the Gini index for educational inequalities (Panel A), a correlation that is broadly unchanged over time.

The result is different for the standard deviation (Panel B), which runs from positive (i.e. higher GDP per capita is associated with higher educational inequality) before 1950, to negative (i.e. higher GDP per capita is associated with lower educational inequality) in the second half of the 20th century (Figure 7.3).This may be attributed partly to a scale effect: European countries and their Offshoots had both higher levels of GDP per capita and higher educational inequality before around 1950, while the less developed countries had low levels of both per capita GDP and educational inequality, resulting in positive correlations. Yet, as shown in Figure 7.2, over time educational inequality in the less developed countries increased, whereas it declined for the developed countries, thus leading to declining, and eventually negative, correlations.

Inequality in educational attainment can be expressed in various ways, depending on one’s perspective. If relative differences are considered more important, measures such as the Gini index are preferred. However, if absolute differences in years of education are seen as relevant, absolute measures of inequality, such as the standard deviation, are preferred.

Overall, the Gini index of educational inequality at the world level declined significantly over the 19th century and even more so in the 20th and early 21st centuries. This reduction has been caused largely by the reduction of the share of people without any formal education. On the contrary, the standard deviation of educational inequality first increases over time when a growing share of the population participates in formal education, and then declines when this share exceeds 50% of the population. These patterns of rising and declining educational inequality, as measured by the standard deviation, are captured by the notion of the educational Kuznets curve. With the exception of some less developed countries that are still on the rising part of this curve, most countries reached the peak of the standard deviation at some point in the 20th century. These exceptions are noteworthy since, when countries are on the upward trajectory of educational inequality, they face a trade-off between increasing the mean years of schooling and reducing educational inequality.

These patterns in educational inequality are, however, tentative, and much more research is necessary. For example, in this chapter we focused on inequality in years of education, ignoring the large literature on the quality of education. After all, an increase of one year in tertiary education will have a different impact on educational inequality than the same increase in primary education. Likewise, there may be large differences in the quality of primary education between countries. At present, various studies use the OECD’s Programme for International Student Assessment (PISA) to capture the ability of 15-year-olds in e.g. reading and mathematics. Unfortunately, PISA scores are unavailable for historical periods. One way to overcome this would be to use proxies for educational quality, such as wages or pupil-teacher ratios, which may be available in historical periods. Finally, future research should also focus on personal characteristics, such as gender or place of residence, which may influence trends in educational inequality, e.g. Lee and Lee (2016[6]) and Ali, Benjamin and Mauthner (2004[39]).


[39] Ali, S., S. Benjamin and M. Mauthner (2004), The Politics of Gender and Education: Critical Perspectives, Palgrave Macmillan, London.

[10] Amiel, Y. and F. Cowell (1999), Thinking about Inequality: Personal Judgment and Income Distributions, Cambridge University Press, Cambridge.

[14] Bandyopadhyay, S. (2018), “The Absolute Gini is a more reliable measure of inequality for time dependent analyses (compared with the Relative Gini)”, Economics Letters, Vol. 162, pp. 135-139, https://doi.org/10.1016/j.econlet.2017.07.012.

[21] Barro, R. and J. Wha Lee (2015), Education Matters: Global Schooling Gains from the 19th to the 21st Century, Oxford University Press, Oxford.

[18] Barro, R. and J. Wha Lee (2013), “A new data set of educational attainment in the world, 1950-2010”, The Journal of Development Economics, Vol. 104, pp. 184-98, https://doi.org/10.1016/j.jdeveco.2012.10.001.

[30] Boli, J. (1989), New Citizens for a New Society. The Institutional Origins of Mass Schooling in Sweden, Pergamon Press, Oxford.

[31] Bolt, J. and D. Bezemer (2009), “Understanding long-run African growth: colonial institutions or colonial education?”, The Journal of Development Studies, Vol. 45/1, pp. 24-54, https://doi.org/10.1080/00220380802468603.

[41] Bukowski, P. (2019), “How history matters for student performance. Lessons from the partitions of Poland”, The Journal of Comparative Economics, Vol. 47/1, pp. 136-75, https://doi.org/10.1016/j.jce.2018.10.007.

[27] Cao, N. (2013), Population Sequel Collection In The Period of the Republic of China, National library Press, Beijing.

[42] Card, D. (2001), “Estimating the return to schooling: Progress on some persistent econometric problems”, Econometrica, Vol. 69/5, pp. 1127-60, https://www.jstor.org/stable/2692217?seq=1#metadata_info_tab_contents.

[1] Card, D. and J. DiNardo (2002), “Skill-biased technological change and rising wage inequality: Some problems and puzzles”, The Journal of Labor Economics, Vol. 20/4, pp. 733-783, https://doi.org/10.1086/342055.

[11] Castelló, A. and R. Doménech (2002), “Human capital inequality and economic growth: Some new evidence”, Economic Journal, Vol. 112/478, pp. 187-200, https://doi.org/10.1111/1468-0297.00024.

[33] Cogneau, D. and A. Moradi (2014), “Borders that divide: Education and religion in Ghana and Togo since colonial times”, The Journal of Economic History, Vol. 74/3, pp. 694-724, https://doi.org/10.1017/S0022050714000576.

[17] Cohen, D. and M. Soto (2007), “Growth and human capital: Good data, good results”, The Journal of Economic Growth, Vol. 12/1, pp. 51-76, https://doi.org/10.1007/s10887-007-9011-5.

[19] de la Fuente, A. and R. Doménech (2015), “Educational attainment in the OECD, 1960-2010. Updated series and a comparison with other sources”, Economics of Education Review, Vol. 48, pp. 56-74, https://doi.org/10.1016/j.econedurev.2015.05.004.

[36] Dupraz, Y. (2019), “French and British colonial legacies in education: Evidence from the partition of Cameroon”, The Journal of Economic History, Vol. 79/3, pp. 628-68, https://doi.org/10.1017/S0022050719000299.

[43] Eyre, L., T. Lovell and C. Smith (2004), “Gender equity policy and education: Reporting on/from Canada”, in The Politics of Gender and Education, London: Palgrave Macmillan.

[44] Feldmann, H. (2016), “The long shadows of Spanish and French colonial education”, Kyklos, Vol. 69/1, pp. 32-64, https://doi.org/10.1111/kykl.12102.

[2] Field, S., M. Kuczera and B. Pont (2007), “No More Failures: Ten Steps to Equity in Education”, https://www.oecd.org/education/school/45179151.pdf.

[25] Földvári, P. and B. van Leeuwen (2014), “Educational and income inequality in Europe, ca. 1870-2000”, Cliometrica, Vol. 8/3, pp. 271-300, https://doi.org/10.1007/s11698-013-0105-3.

[9] Földvári, P. and B. van Leeuwen (2011), “Should less inequality in education lead to a more equal income distribution?”, Education Economics, Vol. 19/5, pp. 537-54, https://doi.org/10.1080/09645292.2010.488472.

[40] Fourie, J. and C. Swanepoel (2015), “When selection trumps persistence. The lasting effect of missionary education in South Africa”, TSEG/ Low Countries Journal of Social and Economic History, Vol. 12/1, pp. 1-30.

[29] Fuller, B., J. Edwards and K. Gorman (1986), “When does education boost economic growth? School expansion and school quality in Mexico”, Sociology of Education, Vol. 59/3, pp. 167-81, https://doi.org/10.2307/2112341.

[32] Garnier, M. and M. Schafer (2006), “Educational model and expansion of enrollments in Sub-Saharan Africa”, Sociology of Education, Vol. 79/22, pp. 153-75.

[28] ISSP Research Group (2017), “International Social Survey Programme: Social Inequality IV - ISSP 2009.”, Cologne: GESIS Data Archive.

[12] Jenkins, S. and M. Jäntti (2005), “Methods for summarizing and comparing wealth distributions”, Social Research. ISER Working Paper Series.

[6] Lee Wha, J. and H. Lee (2016), “Human Capital in the Long Run”, The Journal of Development Economics, Vol. 122, pp. 147-169, https://doi.org/10.1016/j.jdeveco.2016.05.006.

[7] Lopez, R., V. Thomas and Y. Wang (1999), Addressing the Education Puzzle: The Distribution of Education and Economic Reform, The World Bank.

[16] Meschi, E. and F. Scervini (2014), “Expansion of schooling and educational inequality in Europe: The educational Kuznets Curve revisited”, Oxford Economic Papers, Vol. 66, pp. 660-80, https://doi.org/10.1093/oep/gpt036.

[3] Meyer, H. (2011), ““Path dependence in German and American public education: The persistence of institutional difference in a globalizing world””, in Mitchell, D., R. Crowson and D. Shipps (eds.), Shaping Education Policy: Power and Process, Routledge, New York, https://doi.org/10.4324/9780203837351.

[22] Morrisson, C. and F. Murtin (2013), “The Kuznets Curve of human capital inequality: 1870-2010”, The Journal of Economic Inequality, Vol. 11/3, pp. 283-301, https://doi.org/10.1007/s10888-012-9227-2.

[5] Morrisson, C. and F. Murtin (2009), “The century of education”, Journal of Human Capital, Vol. 3/1, pp. 1-42, https://doi.org/10.1086/600102.

[13] Niño-Zarazúa, M., L. Roope and F. Tarp (2017), “Global inequality: Relatively lower, absolutely higher”, Review of Income and Wealth, Vol. 63/4, pp. 661-84, https://doi.org/10.1111/roiw.12240.

[48] Psacharopoulos, G. (1979), “On the weak versus the strong version of the screening hypothesis”, Economics Letters, Vol. 4, pp. 181-85.

[15] Ram, R. (1990), “Educational expansion and schooling inequality: International evidence and some implications”, The Review of Economics and Statistics, Vol. 72/2, pp. 266-74, https://doi.org/10.2307/2109716.

[20] Samir, K. and W. Lutz (2017), “The human core of the shared socioeconomic pathways: Population scenarios by age, sex and level of education for all countries to 2100”, Global Environmental Change, Vol. 42, pp. 181-92, https://doi.org/10.1016/j.gloenvcha.2014.

[37] Samoff, J. (1991), “Socialist education?”, Comparative Education Review, Vol. 35/1, pp. 1-22, https://www.jstor.org/stable/1188013.

[35] Stites, R. and L. Semali (1991), “Adult literacy for social equality or economic growth? Changing agendas for mass literacy in China and Tanzania”, Comparative Education Review, Vol. 35/1, pp. 44-75, https://www.jstor.org/stable/1188015.

[45] Subramanian, S. and D. Jayaraj (2013), “The evolution of consumption and wealth inequality in India: A quantitative assessment”, Journal of Globalization and Development, Vol. 4/2, pp. 253-81, https://doi.org/10.1515/jgd-2013-0018.

[24] Tamura, R. et al. (2019), “Economic Growth in the Long Run”, The Journal of Development Economics, Vol. 137/March, pp. 1-35, https://doi.org/10.1016/j.jdeveco.2018.10.010.

[8] Thomas, V., Y. Wang and X. Fan (2003), “A new dataset on equality in education: Gini and Theil Indices of schooling for 140 countries, 1960-2000”, The Journal of Educational Planning and Administration, Vol. 17/1, pp. 5-33.

[46] Trent, W. (1984), “Equity considerations in higher education: Race and sex differences in degree attainment and major field from 1976 through 1981”, The American Journal of Education, Vol. 92/3, pp. 280-305, https://www.journals.uchicago.edu/doi/abs/10.1086/443750.

[4] UNESCO Institute for Statistics (2012), International standard classification of education: ISCED 2011, Montreal: UNESCO Institute for Statistics.

[47] van Bavel, J. (2012), “The reversal of gender inequality in education, union formation and fertility in Europe”, Vienna Yearbook of Population Research, pp. 127-54.

[38] van Damme, D. (2014), “How Closely is the Distribution of Skills Related to Countries’ Overall Level of Social Inequality and Economic Prosperity?”, OECD Education Working Papers, No. 105, OECD Publishing, Paris, https://dx.doi.org/10.1787/5jxvd5rk3tnx-en.

[23] van Leeuwen, B. and J. van Leeuwen-Li (2014), “Education since 1820”, in How Was Life?: Global Well-being since 1820, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264214262-9-en.

[34] White, B. (1996), “Talk about school: Education and the colonial project in French and British Africa (1860-1960)”, Comparative Education, Vol. 32/1, pp. 9-26, https://doi.org/10.1080/03050069628902.

[26] Yugan (1947), Fifth Township, Fourth District, Household Registration Books, Shangrao City: Yugan county, Shangrao City, Jiangxi Province.


← 1. http://www.oecd.org/education/launch-of-oecd-2015-education-at-a-glance.htm.

← 2. Some authors have claimed that private returns cannot be calculated, as education does not signal higher productivity but rather functions as a screening tool for innate abilities. Yet, as pointed out by Psacharopoulos (1979[48]) and Card (2001[42]), even though screening might occur at the hiring phase (when employers have little knowledge about one’s productivity), this effect ebbs away over time when employers become more capable of assessing the true productivity of individual workers.

← 3. Equity depends on a great many factors. The role of race (Trent, 1984[46]), gender (Eyre, Lovell and Smith, 2004[43]; van Bavel, 2012[47]) and income has been widely discussed in relation to educational differences.

← 4. Often these historical differences are proxied by border effects (Bukowski, 2019[41]), the experience of colonialism (Feldmann, 2016[44]) or religion (Fourie and Swanepoel, 2015[40]).

← 5. There are some studies that included, besides the Gini, also the Theil index. See e.g. Thomas, Wang and Fan (2003[8]) or Morrisson and Murtin (2013[22]).

← 6. There also exist intermediate measures between relative and absolute inequality, but we will ignore those here (Subramanian and Jayaraj, 2013[45]).

Metadata, Legal and Rights

This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area. Extracts from publications may be subject to additional disclaimers, which are set out in the complete version of the publication, available at the link provided.

© OECD 2021

The use of this work, whether digital or print, is governed by the Terms and Conditions to be found at https://www.oecd.org/termsandconditions.