Published by EH.Net (June 2016)

Robert J. Barro and Jong-Wha Lee, Education Matters: Global Schooling Gains from the 19th to the 21st Century. New York: Oxford University Press, 2015. xi + 289 pp. $35 (hardcover), ISBN: 978-0-19-937923-1.

Reviewed for EH.Net by Sun Go, School of Economics, Chung-Ang University.

Can we analyze the role of education in economic, political, and social development using cross-country panel data? Robert J. Barro (Paul M. Warburg Professor of Economics at Harvard University) and Jong-Wha Lee (Professor of Economics and the Director of the Asiatic Research Institute at Korea University) have long been studying this subject and have published numerous academic works on it for more than twenty years. Education Matters is a comprehensive volume of their contributions to the literature on human capital and development, compiled by creating and analyzing their own long-term panel data at the country level. The book is mainly composed of two parts. The first part (Chapters 2 and 3) explains how they created cross-country panel data on average school years, and projects the growth of educational attainment to 2040. The second part (Chapters 3, 4, and 5) presents their analysis of the panel data on the effect of educational attainment on growth, fertility, and political institutions. In Chapter 5, in particular, Barro and Lee create another set of educational attainment data considering the quality of schooling, and repeat their analysis using the new quality-adjusted data set.

A conspicuous contribution of Barro and Lee is the creation of the cross-county educational attainment data sets. They first collect 146 countries’ enrollment rates of school-age population at the elementary, secondary, and tertiary levels at five-year intervals from 1950 to 2010. Most of the enrollment data are collected from UNESCO statistics based on each country’s census reports. By applying school-entering ages, term lengths, and the dropout rates to the enrollment rates by year and country, they compute the average years of schooling of the young cohorts, such as the 15 to 19 or the 20 to 24 year-old population. The average years of schooling for the older cohorts are estimated by applying cohort and education-level specific mortality rates to the educational attainment of younger cohorts under the assumption of no adult education. Finally, the educational attainment of the working population of a country for a year are calculated by the average years of schooling of the five-year-interval birth cohorts weighted by their population sizes. In addition to the total educational attainment data, Barro and Lee also estimate country-level educational attainment by sex. This is the baseline data that they have created and updated since the 1990s.

The book introduces another historical panel of educational attainment at the country level from 1870 to 1945, which is constructed in a similar way. The starting point is again a collection of the enrollment rates at elementary, secondary, and tertiary levels of 89 countries from 1820 to 2010 at five-year intervals. They calculate the enrollment rates using various historical statistics of school enrollments and school-age populations. Historical enrollment statistics are compiled from diverse sources such as Databanks International, Mitchell’s International Historical Statistics, Benavot and Riddle (1988), Lindert (2004), U.S. Bureau of Education’s Annual/Biannual Reports, Barnard (1854), and Monroe (1911). School-age population statistics are collected from Mitchell’s International Historical Statistics, the United Nations’ Demographic Yearbooks, and the League of Nation’s Statistical Yearbooks. However, due to the limited availability of historical enrollment data, a portion of the enrollment rates are created by linear interpolation or estimation assuming a logistic trend. About 38 percent of the school enrollment rates of total population from 1870 to 1945 are either interpolated or estimated. The share of the artificial data is greater for the female population and the period before 1870. Using the historical enrollment rates, Barro and Lee estimate the historical data on educational attainment by sex in a similar way to their baseline data for 1950-2010.

The two data sets are freely downloadable from the authors’ webpage ( The baseline data on educational attainment from 1950 to 2010 have already been widely used by researchers in social sciences, and their unique historical panel is expected to attract the interests of scholars. The historical panel will be useful in capturing long-run trends and examining over-time correlation between the expansion of formal schooling and other variables in the long run. However, the Barro-Lee historical panel of educational attainment may not be the best for identifying moments of change in educational expansion or research that requires identifying the timing of variation in formal schooling, as it contains values structurally estimated only by the trend.

Barro and Lee also present various results from the cross-country panel analysis using their own data sets. Although they have created balanced panels of educational attainment, the data sets for further analysis become unbalanced panels because other dependent and control variables, such as GDP per capita, fertility, and the democracy index, are not available for all the countries and years. Their development accounting shows that about 6 to 20 percent of the cross-country variation in output per worker can be explained by educational attainment. The contribution of human capital to economic growth is estimated to be a bit higher in growth accounting. The authors also present results from the three-stage least squares regressions with country fixed effects, which use lagged explanatory variables as instruments to deal with a possible endogeneity problem between education and the outcome variables — the growth rate of GDP per worker, fertility, and the democracy index. The results are not exactly the same as in the existing literature. The effect of educational attainment on growth is weak and statistically insignificant. The effect on fertility differs by gender. Women’s schooling leads to lower fertility, while men’s schooling is positively associated with fertility. The effect on democracy is nonlinear. Controlling for the country fixed effects, the higher average years of schooling, particularly of women, raise the democracy index at a decreasing rate. The panel regressions using the baseline data of 1950-2010 and the historical panel of 1870-2010 return similar results.

In Chapter 6, Barro and Lee repeat the panel regression analysis using another cross-country panel of 70 countries containing the quality-adjusted human capital stock measures from 1960 to 2010 at five-year intervals. The quality-adjusted human capital stock is calculated by weighting the average schooling years of each five-year birth cohort by the relevant test scores representing the quality of education and associated labor market returns. Their collection of standardized test scores spans 134 countries from 1965 to 2010 at the elementary and secondary school levels, despite a significant portion of the observations being missing, especially for the earlier period. Barro and Lee again fill the missing observations with estimates by linear interpolation or regional trends. The real and artificial standardized test scores for each five-year interval then become cohort-specific aggregate measures of school quality at each level of schools. The quality-adjusted panel of educational attainment is constructed in a similar way to the previous data sets. Further, from the panel IV regressions, Barro and Lee find that quality-adjusted educational attainment has a positive effect on the growth rates of GDP per worker if average years of schooling are controlled.

Education Matters offers a bird’s eye view of the role of education in the long-run development in the global context. It clearly shows the pioneering endeavor of Robert Barro and Jong-Wha Lee for the construction and analysis of their unique cross-country panels of educational attainment data. Anyone interested in cross-country analysis on the effect of human capital on economic, social, and political outcomes will undeniably find this volume a practically helpful starting point. This book also contains good teaching resources for undergraduate courses, such as maps showing the expansion of formal schooling in the world or figures presenting correlations between the average years of schooling and other socioeconomic indicators. On the other hand, the book may not be perfect for studying what really happened in history, as descriptions of historical or institutional backgrounds are not sufficiently accompanied by the valuable work of data construction and analysis. The book also contains little discussion of the contributions by economic history research to the literature on the rise of formal schooling and its associated effects on various outcomes since the nineteenth century.


H. Barnard (1854), National Education in Europe: Being an Account of the Organization, Administration, Instruction and Statistics of Public Schools of Different Grades in the Principal States, New York: C.B. Norton.

A. Benavot and P. Riddle (1988), “The Expansion of Primary Education, 1870-1940: Trends and Issues,” Sociology of Education, 61(3): 191-210.

P. Lindert (2004), Growing Public, Cambridge, UK: Cambridge University Press.

P. Monroe (1911), A Cyclopedia of Education, New York: Macmillan.

Sun Go is an Associate Professor of Economics at Chung-Ang University. His research focuses on the development of public school finance in the nineteenth-century United States and twentieth-century Korea.

Copyright (c) 2016 by EH.Net. All rights reserved. This work may be copied for non-profit educational uses if proper credit is given to the author and the list. For other permission, please contact the EH.Net Administrator ( Published by EH.Net (June 2016). All EH.Net reviews are archived at