Learning about Learning
More years of schooling and more education are not the same
As reviewed by Lant Pritchett
This book, a masterful summary of a decades-long research project, should clear up once and for all one of the most damaging confusions in all of economics, particularly development economics. That confusion is the widespread practice of treating “schooling” and “education” as synonymous.
Contrast the statement in the United Nation’s 1948 Declaration of Human Rights that “Everyone has the right to education” (Article 26 (1)) with the UN’s Millennium Development Goal Target 2.A, “Ensure that, by 2015, children everywhere, boys and girls alike, will be able to complete a full course of primary schooling.” The latter assumes that meeting a target for years of schooling will achieve the goal of education.
If there has been one element of conventional wisdom about what it would take for countries to develop (in all senses of the word)—and in particular to achieve high levels of economic productivity (measured crudely as GDP per worker)—it is that expanding education is key to development and economic growth.
Unfortunately, political leaders, policy makers, international donors, development economists, and the public at large made the easy and natural elision of “education” to “schooling” and set about to expand schooling, on the premise that expanding time spent in school will produce economic growth, prosperity, and development.
At the task of expanding schooling, the world has been phenomenally successful. In 1950, the average person in the developing world aged 15 to 64 had spent 2.1 years in school. By 2010 the average adult in a developing country had spent 7 years in school. In the 60 years from 1950 to 2010, schooling increased by more than twice as much as in all previous human history combined.
Moreover, this expansion was equal across categories of developing countries: schooling went up by about five years in democracies and autocracies, in corrupt countries and noncorrupt countries, and, remarkably, in developing countries that had rapid economic growth and in those that had slow economic growth.
Consider Ghana and Thailand. From 1960 to 2010, Ghana expanded schooling by 6.7 years to reach an average of 7.75 years in the adult population, which seemed an amazing success. But GDP per capita in constant PPP (i.e., GDP converted to international dollars based on purchasing power parity) increased by barely 10 percent, from 2,107 to 2,354, which was disappointing stagnation. Thailand increased schooling by only 3.4 years to reach about the same level as Ghana, 7.5 years—so it accomplished less schooling expansion—but GDP per capita increased eightfold from 986 to 8,628.
The expansion of schooling around the world has led to some striking disconnects between the level of schooling and levels of income and productivity. As of 2010, the average adult in Haiti had 5.2 years of schooling. That is the same number of years as the average adult in France had in 1970. But France in 1970 was a fully developed country and economy by any measure, with GDP per capita in PPP of 14,500. Haiti in 2010 was not developed in any sense of the word, with GDP per capita in PPP of only 1490.
By at least the mid-2000s, it was clear that the assertion that rapid expansion of schooling would inevitably lead to higher rates of economic growth was demonstrably false. The association between the percentage rate of growth of schooling years (or change in natural log years) of the labor force and economic growth conditional on the growth of measures of physical capital was negative. (As the book discusses, the literature on economic growth had taken to mostly avoiding the question of how the expansion of schooling affected growth by examining the association between economic growth and the lagged level of schooling, which had some formal rationale, but was decidedly odd. Even in that empirical approach, the lagged level of schooling was not a robust or empirically important correlate of growth.)
This background is important for understanding the contributions of this book. One possible reaction to a book showing that “education is important for the economic growth of countries” is to say “yes, but we always knew that,” but that is clearly not the case. As the book introduction relates, this book was initially meant to be completed in 2006. At that time, the development community (countries and donors) was hopelessly confused about the connection (or lack of connection) between schooling and education. What was being promoted and spent for and measured by countries and the development community was additional years of schooling. Yet it was obvious (and not just in the sophisticated econometric sense, but also in the more important “just look out the window” sense) that the massive ongoing expansion of schooling was not having the hoped-for impacts in many countries.
Hence the book makes three major contributions:
First, Hanushek and Woessmann exploit recently connected data from TIMMS and PISA and other international assessments to construct the most sophisticated and comprehensive measures yet seen of learning across countries and over time. While the researchers rely heavily on math and science performance, they don’t make the claim that these are the only important parts of education. But they do claim that these are key indicators that have been widely measured for many years and, hence, can be used to measure competence as distinct from “time served” in school.
Figure 2.3 showing “learning” across regions and countries and Figure 2.4 showing the weak association between schooling and learning (e.g., China has twice the learning of Venezuela with about the same number of years of schooling) are alone worth the price. The low levels of learning in developing countries (including middle-income countries in Latin America) are beyond words (though terms like “shocking,” “depressing,” and “dispiriting” come to mind).
Second, the authors show that, if one looks empirically at which variable better explains countries’ growth, education (proxied by their test scores) or schooling, the answer is all test scores. Not some mixture of test scores and schooling, but all test scores. Anyone saying anything about economic growth and education from now on needs to have Figures 3.1 and 3.2 in mind.
Third, the policy implication of this connection between test scores and economic growth is that improving learning has massive economic returns. But this should not lead anyone into a new false elision between “learning” and “more spending” or “more inputs.” Reforms that are going to have strong and sustained effects on learning are going to be systemic (affecting the incentives of agents at all levels) and contextualized—and hence will be hard to implement and sustain. Now that the connection between learning and economic progress is settled, the next generation of research has to focus on how to produce the learning gains where they are most needed.
Lant Pritchett is professor of the Practice of International Development at the Harvard Kennedy School.