Daniel Cohen

Daniel Cohen

Associate Chair, Founding Member of the Paris School of Economics

Professor and Director of the Economics Department ENS

Campus Jourdan – 48 Boulevard Jourdan 75014 Paris

Phone +33(0)1 80 52 13 51

  • International Trade
  • Macroeconomic dynamics
  • Financial economics
  • Macroeconomics and asset accumulation
  • Monetary and Fiscal Policy

International Educational Attainment Database



The Cohen-Soto database (2007) compliled in the article "Growth and human capital: good data, good results", has been updated by Cohen-Leker in 2014.

It gives the educational attainments of the population of 95 countries, every ten years, from 1960 to 2020. The database displays the average years of education of the population above 15, above 25, and between 25 and 64 years old. The data are also disaggregated by 5 year age groups.

Download the Database

The database is available in excel or stata format. For any inquiries: laura.leker@psemail.eu



Three main sources are used: (i) the OECD database on education; (ii) national censuses or surveys published by UNESCO’s Statistical Yearbook and the Statistics of educational attainment and illiteracy; and (iii) censuses obtained directly from national statistical agencies’ web pages.

The parcimonial use of censuses

We rely on a limited number of censuses per country, usually drawn from a single source, to have consistent series over time. For countries for which there has been changes in the classification system of education since 1960, we use exclusively the census based on the latest classification and apply the backward extrapolation to calculate the earlier dates. This avoids measurement errors coming from the alterations of the classification systems.
For some countries, we exclusively use enrollments to build a measure of the educational attainments by age groups.

For further information on the number of censuses, their sources, and schooling duration per country, see here.

Backward and forward extrapolation of censuses

We assume that the average years of schooling of a cohort remains unchanged once it has reached the age of 25. Therefore, we extrapolate backward or forward the average years of education of age-groups above 25, using as a benchmark informations on younger or older age-groups in censuses conducted at a previous or a later date. This is possible provided that the age-groups used as benchmarks are also above 25. Missing cells remain therefore for some groups, mostly those aged 15 to 19, and 20 to 24, when extrapolating backward, and sometimes older ones when extrapolating forward, as well as for the oldest groups of the population when extrapolating backward, since they may be already dead at the date of the census used as a benchmark. We rely on enrollments to fill the missing cells.


For some countries, all the series are based on enrollments. For those for which we use censuses, we rely on enrollments only to fill the missing cells when we extrapolate backward or forward. We carefully take into account the drop out rates and the repetition rates given by UNESCO to compute net intake ratios and therefore educational attainments.

For a more detailed discussion about the methodology, see Cohen-Soto (2007)

Related academic articles

Daniel Cohen & Marcelo Soto, 2007. "Growth and human capital: good data, good results", Journal of Economic Growth, Springer, vol. 12(1), pages 51-76, March

Daniel Cohen & Laura Leker, 2014. "Health and Education: Another Look with the Proper Data", Manuscript.