Why do some countries grow rich and others languish in poverty? This is the question addressed in a very important new book, The Knowledge Capital of Nations, by Eric Hanushek and Ludger Woessmann, and in a paper by the same authors titled, “Universal Basic Skills: What Countries Stand to Gain,” published by the OECD with an introduction by Andreas Schleicher. The book and paper make a very carefully documented and well-reasoned case that the academic skills of a nation’s labor force, as measured by the skills they leave high school with, contribute far more than any other single factor to national economic growth in the long run.
The question of why some countries grow rich and others do not is not new. Economists and politicians have been looking for good answers to this question for centuries. But it has been a frustrating quest. As long ago as the 17th century, an English economist suggested that the skills of the people in the workforce would have to be taken into account in explaining economic growth. Adam Smith, a century later, opined that the expense of a worker’s education would be more than paid for by the profits to be made from the skills acquired by that worker. More recently, as mathematical analysis has become the beating heart of economics, leading economists have been trying to quantify the contribution made by the education and training of the labor force to the growth of national economies.
But this effort to quantify the contribution of education to national economic growth has had its problems. At first, economists added up all the contributions of things they could identify and measure that contributed to economic growth and then attributed most of the rest to education and technological innovation. That was not particularly satisfying. Then they used the average numbers of years of education among the workers in a nation as a measure of how well-educated that workforce was. When they correlated the number of years of education in the workforce over time with economic growth in the same economy, they got a significant effect, but they also got lots of anomalies.
Take Latin America and East Asia, for example. Back in 1960, average income in Latin America substantially exceeded that in East Asia. And the workforces in Latin America were much better educated, as measured by average years of education attainment. The income head start, combined with their great advantage in educational attainment, should have enabled the Latin Americans economies to have grown much more quickly than the Asian economies, leaving East Asia as a whole far behind. But, as we all now know, that is not what happened at all. The East Asian economies quickly outstripped Latin America with growth rates that have eclipsed Latin American growth rates for decades. How could that have happened if education is the principal driver of economic growth? Over time, questions of this sort have led analysts and policymakers to have less faith in investments in education as an essential driver of economic growth.
But, as it turns out, it is all in the measure that is being used. In this case, the measure of education and skill being used was average number of years of educational attainment. What Hanushek and Woessmann wanted to know is what would happen if, instead of using quantity of education as the measure of workforce quality, we used the quality of education instead. Back in 1960, there was no reliable measure of education quality that could be used on an international scale to compare school academic achievement. Years of attainment were all that the economists of the time had to go on. But, since then, such measures have been developed, most importantly TIMSS, PIRLS and PISA. The researchers used the simple average of all observed math and science scores between 1964 and 2003 for each country. The long time scale was necessary because it takes a long time for changes in school policy to affect what actually goes on in classrooms, and then for what goes on in classrooms to affect national economic output. The results are startling.
Let’s take the Latin America anomaly. It simply goes away. When economic growth rates for Latin America and East Asia are plotted against school test scores, they are almost perfectly correlated. When economic growth rates for the two regions are plotted against years of attainment in school, there is hardly any correlation at all. This holds true over a period of fifty years. It turns out that the early development economists were right in concluding that education has a very strong bearing on economic growth. The problem was that they were using the wrong measure of education.
What is really eye opening is not the finding that education makes a big difference in economic outcomes but how much of a difference it makes. The achievement-based models used by the authors accounted for three-quarters of the international variation in economic growth rates. Compare this with the old models, based on educational attainment, which accounted for only one-quarter of the economic growth in the countries studied. The authors calculate that a difference of one standard deviation in the cognitive skills of a country’s workforce is associated with approximately two percentage points of higher annual growth in GDP. A full standard deviation is a lot to expect under any circumstances, but one quarter of that (about 25 points on the PISA scale) has actually been accomplished by Germany, Mexico, Poland and Turkey recently and in Finland earlier. So let’s imagine what a gain of one half of one point in the annual growth rate of GDP (the amount produced by an average gain of 25 points on the PISA scale) would do for the United States. According to the Congressional Budget Office, the cumulative losses from the Great Recession from 2008 to 2012 were about $4 trillion. But an improvement of just 25 points on average for American students would bring in about $67 trillion over the next 50 years. How’s that for a return on your investment?
Andreas Schleicher, in his introduction to the paper by these authors, points out that the economic gains to be had by raising achievement are so great as to make it possible to pay the entire cost of a nation’s current education system from just the increases in economic growth that result from the increased achievement. As impressive as these effects are, the researchers found that the effects of education on economic growth rates have themselves been growing at an even faster rate in recent years, confirming the findings of others that what a person or a country earns is now more dependent on what they know and can do than ever before.
Projections of this sort are based on models. The researchers make assumptions, plug them and the data into their models and watch the results pour out the other end. What is really interesting about the work of these two authors is that they not only share, in detail, the design of their models, the nature of the assumptions they are making and the data they are plugging in to their models, but they adjust the models to reflect a whole range of conceptual models and assumptions. And, time after time, they show us that the general direction of the conclusions is not affected in any important way by these adjustments.
This is only one aspect of the care the authors have taken to anticipate their critics and refute them in advance. Another has to do with the question of causation. Correlation, as we all learn in college, is not cause. Perhaps these very strong relationships between school academic achievement and economic growth do not reflect the contribution that highly educated workers make to economic growth but rather reflect the fact that wealthy economies can afford to spend more on education than others not so wealthy. This possibility is convincingly dealt with in a clever analysis to which the authors devote a whole chapter. Again and again, the authors use their models, the conceptual frameworks that underlie them, the data they have gathered and their statistical skills to make their points in a very convincing way.
With one exception. By the end of the next-to-last chapter, the reader wonders how anyone can doubt the author’s claims that school achievement is a very powerful contributor to economic growth at a national scale. But that leaves the convinced policymaker with an important question. Just what is it, exactly, that I need to do to make sure that my country is doing what is necessary to raise the academic skills of our high school students so we can enter the ranks of the wealthiest countries?
At this point in the narrative, all the models, data and statistical finesse of the rest of the book cannot be used to answer this question. While I can easily agree that the best research concludes that the quality of a nation’s teaching force has a very strong bearing on the achievement of its students, I have a little more trouble with the use of relative teachers’ salaries as the measure of teacher quality. I have even more trouble with basing the claim that school choice is an essential component of high-quality schooling based on the claim that high proportions of students enrolled in Catholic schools produce superior national education results. The authors would have done much better, in my opinion, to have left this part of the argument to others. Indeed, on this point, I would refer you to Andreas Schleicher’s clear, well-argued and passionate introduction to the paper the authors did for the OECD, which he ends with a much more compelling summary of what the data says about what makes for effective education systems than the one supplied by the authors.
So far, I have stuck to the main line of the story—how higher school achievement leads to stronger economies. But there is a subtext related to equity that has many dimensions and is very important. First there is Andreas Schleicher’s invitation to the authors to write a paper focusing on what the economic outcomes would be for countries that succeed in bringing all their students up to the Level I proficiency on the PISA tests, the lowest level on the PISA scale. One quarter of United States students do not meet this standard. In many underdeveloped countries, that percentage is much higher, but in many developed countries, it is much lower. If the United States were to achieve this goal of universal basic skills, it would add $27 trillion dollars to the gross national product over the working life of the students. But the authors show that, in addition to adding an enormous amount to the wealth of the country as a whole, it would greatly reduce the growing gaps in income that are beginning to threaten the social fabric of the country. Schleicher points out that this is very different from and much more acceptable to many Americans than addressing income inequality by using tax policy to take earned income from one class of citizens and give it to another.
But better education is not an elixir. The models do not produce the predicted results when national economies cannot absorb better-educated workers. Sound fiscal and monetary policy still matter. Many countries whose economies depend largely on the sale of natural resources, especially oil, have failed to educate their citizens or prepare in other ways for the day when their natural resources run out or are no longer needed.
But these points do not in any way detract from the contribution that Hanushek and Woessmann have made with their analysis. It is stunning.