Higher Education and Economic Growth

Tom Wood

There are many reasons to believe in the value of liberal higher education. Some, but certainly not all, of the reasons are economic ones. These economic considerations are arguably not the most important ones for supporting higher education, but in hard economic times like the present, it becomes particularly important for political reasons if nothing else to make the case for higher education in terms of economics. 

If that case cannot be made, higher education runs the risk of being perceived as nothing more than a luxury in which only the rich who can afford it might indulge. To maintain continuing public support for higher education on a broad basis, particularly in hard economic times, requires evidence that higher education leads to greater financial returns and greater economic productivity and growth. Can this case be made, not just for the individual (for which the case is widely acknowledged to be compelling), but for the general society as well? Economically, are there social returns to higher education that justify continued support by states and by the federal government?
 
There is a considerable body of research in economics that addresses this question—just as there is on the subject of “credentialism” or the “sheepskin effect,” which I addressed in my previous essay. Like the previous one, this paper is an attempt to bring research in economics into the discussion about the value of higher education.
 
The econometrics of economic growth (a primer)
 
A non-economist and outsider like me is struck, first of all, by the sheer amount of data that is available on global economic growth. The CIA, the World Bank, the International Monetary Fund, the UN, and others have accumulated a very large mass of data on economic growth at both the regional and country-by-country level.
 
A map of the world’s real growth rate broken down by country is available here for 2007 (based on CIA World Fact Book estimates). One of the most easily discernible patterns is that developing countries have, in general, higher growth rates than developed, advanced economies. Although this pattern came as a surprise to me, the view that, all other things being equal, developing economies will catch up with more advanced, developed economies has been part of economic theorizing on growth almost from the outset. An early assumption in the economic growth literature was that there are diminishing returns to capital; consequently, developing economies should catch up with developed economies. This is called the theory of convergence or the catch-up effect in the economics literature, and it was built into one of the first models of economic growth—the exogenous growth theory of Solow-Swann.
 
Even within this general picture, however, there are further surprises. Take Angola and Turkmenistan. We sometimes read news items about such countries, but it is not typical of such articles (at least in my experience) to do so in the context of outstanding economic growth; instead, they are usually treated as deeply troubled countries with fairly dysfunctional economies. And in fact they are. Their high growth rates in 2007 are mostly explained by their oil and gas reserves, some of which were discovered fairly recently. Economists try to identify fundamental determinants of economic growth that lie beyond and behind such comparatively unpredictable factors.
 
Economists formalize their economic growth theories in models that predict growth over the long-term, focusing on factors or variables that are not haphazard, irregular, or fortuitous. Standard inputs (independent variables) in these equations include physical capital, investment, and labor. Such factors are distinguished from other factors that are recognized as sometimes important but are not treated as inputs in the models. Total factor productivity (TFP) is defined by economists as the variable that accounts for that part of the total output of an economy that is not accounted for by the inputs of the model. Weather, to the extent that it is variable and unpredictable, is considered a total-factor productivity variable.
 
Physical capital and labor (total labor hours for the economy), because they are more tangible, were the first to be included as inputs in econometric growth models. Today, economic growth theorists recognize that less tangible factors like human capital, technology, and technological efficiency play an extraordinarily important role in economic growth. These factors are now included in current economic growth models as input variables. Most of this paper will focus on one such “intangible” factor involved in economic growth—human capital—and especially on the dimension of education, including higher education, in the development of human capital.
 
Economists look beyond annual reports of economic growth in order to identify long-term determinants of growth. When longer time-periods are considered, the picture looks considerably different from annual report snapshots like the one above. Consider, for example, the tables, maps, and graphs on this page of the World Bank web site. Figure 4.1 gives the annual growth rates internationally of GDP, population, and GDP per capita for the period 1965-1999. Map 4.1 on the same page gives the GDP per capita growth rates for the time-period 1990-1999.
 
Data covering longer periods of time get us closer to what we think of intuitively, and that economists have come to recognize, as the systematic, underlying, fundamental determinants of economic growth. They show some significant differences from the yearly snapshots that deserve notice. Note that the CIA map considered above for 2007 gives total GDP, whereas the World Bank map gives per capita GDP. Population growth accounts for much of the difference between the two maps, but by no means all of it. Another salient difference is that the World Bank’s data and maps no longer show a clear picture of convergence. Poor, developing countries no longer head the charts of economic development. Angola and Turkmenistan, for example, are no longer leaders according to this data. They have fallen to less than 0-9% or less than 0% growth. But this is not an invariable pattern. Some developing countries do register impressive growth rates: e.g., Peru, China, India, Malaysia, and Indonesia.
 
Some advanced economies also have respectable rates of growth. Instead of being laggers, for example, the United States, the UK, and Australia register GDP per capita growth rates of 2-2.9 percent. This kind of data has led economists to revise the theory of convergence in a number of ways. Recent economic models attribute a fixed rate of return, rather than diminishing rates of return, to investment or capital. And while economists have not abandoned the theory of convergence completely, they think now in terms of conditional convergence. Advanced economies, it is now recognized, can grow faster than the earlier models recognized, and poorer, developing countries do not necessarily catch up.
 
Economists have found a number of different kinds of variables to be useful in predicting economic growth: physical capital, investments, labor (total labor hours worked), the technological level of the economy, the openness of the economy to trade and its adherence to free markets and market competition, and human capital variables, including the skills and education of its work force.
 
A recent paper—actually a keynote address to an OECD conference in 2000, by Robert J. Barro, can be taken as a representative example. Barro, who has been ranked as the second most cited economist in the world, is a Harvard economist and Senior Fellow at the Hoover Institution. “Education and Economic Growth,” his OECD conference address, is an update of Barro’s widely cited and influential 1995 NBER paper, “The Determinants of Growth.” Since the World Bank address is an update and is available for free online, I will use it for the purposes of the present discussion.
 
Below are some of the input variables that Barro includes in his modeling of economic growth:
 
  • The level of per capita GDP
  • The ratio of government consumption to GDP (a high ratio has a significantly negative effect on growth) 
  • The rule of law, including the security of property rights and a strong legal system. Barro has found that this indicator has the most explanatory power for economic growth. He uses as the indictor here the assessments of a number of consulting companies that advise clients on the attractiveness of countries as places of investment. 
  • International openness (i.e., openness to international trade). The indicator here is the ratio of exports and imports to GDP.
  • Inflation rate. This has, according to Barro, a marginally significant, negative effect on growth.
  • Fertility rate (negatively related to growth in output per person).
  • Investment ratio
  • Terms of trade (the ratio of export prices to import prices)
  • Education, measured by the average years of attainment and scores on internationally comparable examinations.
  • Health variables, such as life expectancy at birth and the infant mortality rate.
Computers have made the statistical analysis of large amounts of this kind of data feasible. The objective is to assign values or weights to the factors that are thought to be important or that have been found to be important to economic growth. Economists call this process estimating the coefficients (parameters) in the equations that model growth, or simply estimating the equations. The mathematics used is regression analysis. The regressions assign weights or values to the parameters, and tell the econometrician how important the various independent variables are as determined by the data. Another aim is to predict as much of the economic growth for a country or region as possible, given the data.
 
The study of economic growth is a subfield of econometrics. There is a good introductory discussion of econometrics on Wikipedia, from which the following equations and discussion have been adapted.
 
A simple linear equation using some of the variables cited above would have the following form:
 
Q = β0 + β1GDP + β2International openness + β3Education + ε
 
Regression analysis is used to estimate the unknown parameters β0, β1, β2 etc. (i.e., the coefficients) in the equation, using measures of GDP, international openness, mean years of education of the work force, etc. The symbol ε represents the measurement error. The aim of a model is to reduce this term to zero (i.e., to leave none of the correlation unexplained).
 
In the economic growth literature the models are actually more complicated than this, because the typical growth model uses something called panel data and panel analysis. A panel has the form
 
            Xit, i= 1,…,N t = 1,…,T,

This sounds quite complicated, and apparently it is. But one must keep in mind the simple idea behind what economists are doing here. One could try to do the same thing in one’s head, by eyeballing maps and figures of economic growth and stocking your mind with what seem to be relevant facts about countries, and then estimating. But you would probably go out of your mind in the process, and in any case any conclusions one might draw would not carry conviction, since someone else using the same informal, unsystematic procedure would almost certainly reach different conclusions.
 
= α + β’Xit + uit.
where i is the individual dimension (e.g., an individual country) and t is the time dimension. A general panel data regression model is written as yit
 
Still, what the econometricians who study growth are doing is not different in principle from the one just described. The main difference is that econometricians try to be precise with data, and build formal models that can be tested against the data. Models and data are set out in a formal way, and then run by computer.
 
Research on economic growth using years of education as a parameter
 
Jacob Mincer (1922-2006), the founder of labor economics, emphasized “educational achievement,” defined as years of education, as an important human capital variable. The earliest econometric work on growth predicted that school achievement (mean years of schooling of the labor force) was a significant factor in economic growth. Barro’s “The Determinants of Growth” confirmed this prediction:
 
The results show a significantly positive effect on growth from the years of schooling at the secondary and higher level [emphasis mine—TW] for males aged 25 and over …. On impact, an extra year of male upper–level schooling is therefore estimated to raise the growth rate by a substantial 1.2 percentage points per year. (p. 15)
 
[G]rowth is predicted by male schooling at the upper levels but not by male schooling at the primary level. However, primary schooling is indirectly growth enhancing because it is a prerequisite for training at the secondary and higher levels. (p. 16)
 
Another interesting finding of Barro’s in “The Determinants of Growth”—and one that generated a could deal of controversy—was that female education at various levels is not significantly related to subsequent growth. The data did indicate, though, that female schooling is important for other indirect indicators of economic development, such as fertility, infant mortality, and political freedom.
 
Cognitive skills and economic growth
 
One of the important differences between Barro’s 1995 NBER paper “The Determinants of Growth” and his 2000 address to the OECD conference (“Education and Economic Growth”) is that the latter uses data on cognitive skills in its analysis of education as a determinant of economic growth, whereas the earlier paper considered only educational achievement, measured by years of schooling.
 
The incorporation of cognitive skills in the analysis of the education component marked a significant advance in the analysis. As Barro says in his OECD paper:
 
Many researchers argue that the quality of schooling is more important than the quantity, measured, for example, by years of attainment. Barro and Lee (1998) discuss the available cross-country aggregate measures of the quality of education. Hanushek and Kimko (2000) find that scores on international examinations—indicators of the quality of schooling capital—matter more than years of attainment for subsequent economic growth. My findings turn out to accord with their results.
 
However, Barro’s papers to date lack an extensive discussion about cognitive skills and their role in economic growth. For such a discussion we must turn to the work of Eric A. Hanushek, one of the first researchers to emphasize the importance of cognitive skills in econometric analysis.
 
Eric A. Hanushek is a Senior Fellow at the Hoover Institution at Stanford and a member of the Koret Task Force on K-12 Education. Hanushek and a number of coworkers have summarized some of their most important findings on the importance of cognitive skills in “Education and Economic Growth“ (henceforth Hanushek et al. 2008). This paper appeared in the Hoover Institution publication Education Next. I shall be using this paper for much of the present discussion, because it is available for free online. More details are available in a paper called “The Role of Cognitive Skills in Economic Development” by Hanushek and Woessmann (henceforth H&W 2008). This more technical paper appeared in 2008 in the Journal of Economic Literature.
 
I give below some of Hanushek’s major findings about the role of cognitive skills in economic growth:
 
  • Models that include direct measures of cognitive skills account for about three times the variation in economic growth than models that include only years of schooling. The former models are also far more robust than the latter.
  • If one country’s test-score performance was 0.5 standard deviations higher than another country during the 1960s—a little less than the current difference in the scores between such top-performing countries as Finland and Hong Kong and the United States—the first country’s growth rate was, on average, one full percentage point higher annually over the following 40-year period than the second country’s growth rate.
  • A work force with high cognitive skills can raise economic growth by about two-thirds of a percentage point every year.
  • If the United States had raised its cognitive skills scores on PISA (one of the international cognitive skills exams) from its below average rating for OECD countries by 50 points (or 0.5 standard deviations) from 1989 to 2000, bringing it to the same level of leading countries like Hong Kong, Taiwan, and Finland, U.S. GDP would by 2015 be 4.5 percent greater than in the absence of any such gains. That 4.5 percent increment in GDP is equal to the total the U.S. currently spends on K–12 education.
  • There is a significant interaction effect between the cognitive skills factor in economic growth and the degree of openness of the country to competition and foreign trade. In closed economies, the effect of cognitive skills on economic growth is significantly positive, though modest, but in open economies like the U.S. the effect size is much larger. Thus, cognitive skills are much more important for open economies like ours than they are for closed economies.
Higher education and the measurement of cognitive skills
 
Although K-12 education in the U.S. falls significantly behind that of our major economic competitors (see below), American economic growth has been quite respectable, particularly for an advanced or developed economy. The fact that the American economy has outperformed what would have been expected on the basis of its performance in K-12 education has puzzled econometricians. One factor in explaining the discrepancy between econometric expectations and performance is that the United States scores well on many of the other measures that economists have found to be important predictors of productivity and healthy economic growth. It scores well, for example, on the rule of law (e.g., the security of property rights) and international openness and competition. Hanushek has cited these factors as part of the explanation of the discrepancy, but he has also mentioned another likely factor. Although American K-12 education is weak, American higher education is strong by almost any international measure:
 
[T]he analysis of growth rates across countries emphasizes quality of the primary and secondary schools of the United States. It does not include any measures of the quality of U.S. colleges. By most evaluations, U.S. colleges and universities rank at the very top in the world. A number of models of economic growth in fact emphasize the importance of scientists and engineers as a key ingredient to growth.
 
As Hanushek has pointed out (H&W 2008: 368, fn. 35), in the 2007 Shanghai rankings of the world’s research universities, the U.S. had seventeen of the top twenty universities and fifty-four of the top ninety-nine. In a 2007 professional ranking based on graduates who were CEOs at Global Fortune 500 countries, U.S. institutions had ten of the top twenty-two places and twenty-four of the top fifty-nine places.
 
Since cognitive skills at the more basic K-12 level are important for growth, it isn’t unreasonable to think, as Hanushek does, that the more advanced skills developed at the post-secondary and higher education levels are important, too. But is it so?
 
Econometricians have been able to predict the impact of K-12 on economic growth, using measures of both years of education (“educational achievement”) and cognitive skills as the parameters (input variables). Unfortunately, econometricians are not in the same position with regard to higher education. As I will show below, the only research that disaggregates higher education from the lower education sectors in the modeling of economic growth has used expenditures for the instrument—and expenditures have proven to be a weak indicator in the econometric analysis of K-12 education.
 
Hanushek and others can show that cognitive skills at the more basic level of K-12 are important for economic growth because good tests (PISA, TIMMS, IALS, ALL) are available for cognitive skills at this level. Similar, internationally calibrated tests for cognitive skills are needed at the level of higher education. Such comparisons should be available well within a decade for at least many of the thirty OECD countries, including the United States. An OECD project called the International Assessment of Higher Education Learning Outcomes (AHELO) is underway to do just that. With such tests in hand, economists will be able to run the same kind of regressions for higher education that have been used to predict the impact of cognitive skills at the K-12 level on economic growth.
 
A cautionary note: years of education is a significant measure and it still matters
 
Work by Hanushek and others has demonstrated that cognitive skills are a far stronger determinant of economic growth than other dimensions of education, like years of education. It is important to be clear, however, that this more recent work has not shown that years of education or other, earlier educational measures are irrelevant to economic growth.
 
A number of studies have found evidence that education contributes to economic growth, and they have managed to show this even in the absence of data on education’s contribution to cognitive skills. Two examples worth mentioning—both utilizing very different methodologies—are the 2001 paper by Krueger and Lindahl, and the 2002 paper by Bassanini and Scarpetta. (See also the 2007 paper, which is available for free online, by Arnold et al., “Solow or Lucas?.) As we have seen, Barro in his earlier work also found that secondary and higher levels of schooling are a significant factor in economic growth, even when years of education are used as the parameter.
 
In addition to their groundbreaking findings about the importance of cognitive skills in economic growth, Hanushek et al. also found that educational achievement (defined as years of schooling) is correlated with economic growth (Hanushek et al. 2008: 66):
 
[W]e looked just at the impact of average school attainment on the economic growth rate. … When we performed this analysis, we found, as other economists before us, that when the average number of years of schooling in a country was higher, the economy grew at a higher annual rate over subsequent decades. Specifically, we found that, across the 50 countries, each additional year of average schooling in a country increased the average 40-year growth rate in GDP by about 0.37 percentage points. That may not seem like much, but consider the fact that since World War II, the world economic growth rate has been around 2 to 3 percent of GDP annually. Lifting it by 0.37 percentage points is a boost to annual growth rates of more than 10 percent of what would otherwise have occurred, a significant amount.
 
Hanushek’s work has found that the educational achievement variable adds nothing to the predictive power of equations for economic growth beyond what is contributed by the cognitive skills variable. However, this finding does not mean that educational achievement (measured simply by years of schooling) does not contribute to economic growth. What the conclusion does support is the conclusion that more attention and study have to be given to what kinds of education and education policies support the development of cognitive skills.
 
Two studies of higher education and economic growth
 
The works I cited above by Krueger and Lindahl, Bassanini and Scarpetti, Barro, and Hanushek included data on post-secondary and tertiary education, but none of those studies focused specifically on the contribution to economic growth by higher levels of education.
 
Tertiary and higher education were, however, the special focus of two studies we will now consider. The studies provide evidence that higher education is an important factor in economic growth, although both studies suggest that this is true only given certain conditions. The studies found that investments in higher education have a significant payoff for countries with very advanced economies. Although the United States is regarded by economists as the world’s most technologically advanced economy—to use the economists’ jargon, it is the country that defines the “world’s technological frontier”—one of the studies found that the payoff for expenditures on higher education might vary widely even among the states in the U.S., depending on the level of their technological development.
 
One of the two papers, by Aghion et al. 2009, is available for free online as a Brookings Institution publication. Its earlier incarnations were a Harvard economics department mimeo dated 2005 and an NBER conference paper published in the same year. Since the update is available for free online, I will be using that version of the paper in the discussion here. The other paper I shall be discussing, Vandenbussche et al. 2006, is available only in the Journal of Economic Growth, which requires a one-time fee or subscription.
 
The Aghion et al. study is based on a comparison of three states in the U.S.: Alabama, West Virginia, and Massachusetts. The Vandenbussche et al. study is based on a comparison of OECD countries for which it was possible to differentiate between tertiary and higher education expenditures, on the one hand, and expenditures on primary and secondary education, on the other. Both studies were designed to measure the economic growth effects of lagged expenditures aimed at the tertiary and higher education sector. (The expenditures were lagged because it takes time for such expenditures to have any effect.) The fact that the studies found significant results using as the instrument expenditures on higher education (treated as external or exogenous shocks) is fairly striking, because Barro, Hanushek and others have found that the economic growth effects of expenditures on K-12 education, while not non-existent, are quite weak.
 
Aghion et al. (2005/2009): expenditures on higher education as a factor in economic growth in the U.S.
 
As I’ve just mentioned, Aghion et al. analyze data on three states in the U.S.: Alabama, Massachusetts, and West Virginia. To assist the reader, it will be helpful to look at some of the data about these three states that are of interest to econometricians. This will help to justify and motivate both the selection of these three states for comparative study and the general methodology that Aghion et al. adopt.
 
Most research on economic growth involves international or regional comparisons rather than within-state comparisons. Aghion et al. had the interesting idea of looking for economic growth comparisons within a country—in this case the United States. The U.S. is by a considerable margin the largest economy in the world, and it is quite heterogeneous from state to state on important economic measures.
 
We will look at three of these measures: (a) personal income; (b) the percentage of the population holding degrees in higher education (Bachelor’s and advanced), and (c) the mean cognitive skills of the respective populations, measured by tests at the K-12 level. Although Aghion et al. do not discuss the data I will be discussing here in their own paper, a brief review of the kind of data that is available will help to explain their finding that the impact of expenditures on higher education can vary considerably across states in the U.S.
 
(a) Data on personal income state-by-state is collected by the U.S. Census Bureau. According to the Bureau’s Fact Finder, the median per capita income in 1999 for the three states we will be considering were as follows:
 
            Alabama                      $18,189
            Massachusetts             $25,952
            W. Virginia                   $16,477
 
(b) The three states also vary considerably according to the percentage of the population holding Bachelor’s degrees. Here is the breakdown for the three states (as of 2007) according to the data available from the U.S. Census Fact Finder:
 
Alabama                      21.4%
Massachusetts             37.9%
W. Virginia                   17.3%
 
Here is the breakdown for the percent of people 25 years and over who have completed an advanced degree (as of 2007) from the same source:
 
Alabama                       8%
Massachusetts             16%
W. Virginia                   6.6%
 
(c) The states of the U.S also vary widely in students’ scores on important cognitive skills tests. The following data is taken from a fascinating report issued by the American Institutes for Research in 2007, entitled Chance Favors the Prepared Mind: Mathematics and Science Indicators for Comparing States and Nations. The report is available for free online. Its author was Gary W. Phillips, who in 1999-2002 was Acting Commissioner of the National Center for Education Statistics (NCES) within the U.S. Department of Education.
 
Phillips used the U.S. Department of Education’s NAEP scores to calibrate scores internationally with the Trends in International Mathematics and Science Study test (TIMMS). His report includes international comparisons of these calibrated science and math scores for each state, with two separate figures (for math and science) devoted to each of the 50 states.
 
Consider some of the details on the first state: Alabama. In the two-part Figure 1 of Phillips’ study (p. 25), NAEP scores for Alabama have been projected onto the constructed TIMMS scale for the purpose of international comparisons. It can be seen from this figure that 17 countries perform statistically better in mathematics than Alabama (these are indicated in the figure by the taller black bars to the left of Alabama):

1. Singapore
2. Hong Kong
3. Republic of Korea
4. Chinese Taipei
5. Japan
6. Belgium (Flemish)
7. Netherlands
8. Hungary
9. Estonia.
10. Slovak Republic
11. Australia
12. Russian Federation
13. Malaysia
14. United States (determined by the mean for all 50 states)
15. Latvia
16. Lithuania
17. Israel.
 
There are 10 countries whose mathematics performance is statistically similar to Alabama
(indicated in the same figure by the white bars surrounding Alabama):
 
1. England
2. Scotland
3. New Zealand
4. Sweden
5. Serbia
6. Slovenia
7. Romania
8. Armenia
9. Italy
10. Bulgaria.
 
And so on.
 
Alabama has a low ranking compared to most other states and to other countries in the study. The same is true of W. Virginia (Figure 51 on p. 75). But not all the states of the U.S. have low rankings. Massachusetts, for example, has a fairly respectable ranking (Figure 23, p.47 of the study). Only four countries, all of them Pacific Rim countries, do better than Massachusetts in math skills on Phillips’ calibrated international scale.
 
The reader is encouraged to review the findings for his or her own state of residence or of special interest, and to do the same for the Census Bureau’s Fact Finder data for personal income and degree holders. To this reader, anyway, the findings were something of an eye-opener. I confess that the results were better for the U.S. in general than I had feared. There has been so much bad-mouthing of American education, and of American K-12 education in particular, that I was relieved to find that we were in the competition at all. On the whole, however, the data are bad news for the U.S., because our most fearsome economic competitors are significantly outperforming us on these important measures.
 
Phillips puts it this way (p. 23):
 
The results in this report represent both good news and bad news. The good news is that most states are doing as well or better than most foreign countries. If you think of states and nations as in a race to prepare the future generation of workers, scholars and citizens to be competent and competitive in a technologically complex world, then the states are in the middle of the pack. The bad news is that even our best-performing states are significantly below the highest performing countries.
 
We are now in a position to return to the Aghion et al. paper and to explain the idea behind its methodology.
 
Although Aghion et al. do not discuss the foregoing kinds of data in their paper, and simply assume it as background and context, the data demonstrates clearly the heterogeneity of American states on the kinds of measures that are of interest to econometricians. States vary significantly in the level of the cognitive skills of their K-12 students; in their personal income statistics; and in the percentage of their populations that holds Bachelor’s and advanced degrees. Furthermore, there is a good deal of overlap on all of these measures. This justifies treating states in the U.S. as separate nations or economies, in the same way that countries have been treated by econometricians in international studies of economic growth.
 
Differences in important economic measures among the states suggested an econometric model to Aghion et al. that could predict the role of higher education in economic growth. They reasoned, as other economists had before them, that economies differ in the extent to which they imitate technology and innovate technology. Countries or states further from the technological frontier imitate technology developed in more advanced economies; countries and states closer to the frontier must innovate new technology.
 
This distinction was at least implicit in work in economic growth theory almost from the beginning. Economists began by assuming—and this has been borne out by subsequent evidence and research—that undeveloped economies will tend to grow faster than developed ones, because the former have the advantage of being able to simply adopt technologies that have been developed elsewhere. This is an easier task than innovating or developing new technologies, which is what advanced economies have to do to keep growing. In economics, this phenomenon has a name—convergence—but since many important exceptions to it have been found, economists now work with the concept of conditional convergence, according to which undeveloped economies will develop faster only if certain conditions are met.
 
Aghion et al. reasoned that in the U.S. public investments in higher education would have different effects depending on a state’s proximity to the technological frontier. To test this hypothesis, Aghion et al. selected three states: Alabama, Massachusetts, and West Virginia. These were selected for a couple of reasons. First, they vary quite widely in their proximity to the world technological frontier, with Alabama and West Virginia being at the low end, and Massachusetts at the upper end, on all the measures considered above. The second reason for selecting these three states is that Aghion et al. found significant increases in expenditures on higher education in these three states that could be treated as exogenous variables in their model. The three states have been represented either in the U.S. Senate or the House of Representatives by chairmen on the Appropriations committees. Because they occupied these powerful positions, these representatives were in a position to “satisfy incumbents” with, among other things, earmarked federal spending on higher education. Aghion et al. tested whether the increases in federal funding (regarded as “exogenous shocks”) had different effects in these three states. They also looked to see how the economic growth patterns of these states compared with the growth patterns of states similar to them, measured by their proximity or distance to the “technological frontier.”
 
Aghion et al. found that, while these infusions of federal subsidies might have boosted the economic growth of all three states, only Massachusetts was benefited to any significant extent. Within the 15 years covered by this study, Massachusetts had caught up with California and even started outpacing California in economic growth (pp. 27-28). On the other hand, there was no significant benefit to Alabama and West Virginia from their increased federal subsidies. Indeed, the authors question whether the monies that flowed to Alabama and W. Virginia might have been put to better use in developing the basic skills of their work forces through increased investments in K-12 education.
 
Aghion et al. (p. 25) put it this way:
 
The three cases we consider concern two far-from-the-frontier states, Alabama and West Virginia, and a close-to-the-frontier state, Massachusetts. In each case, a legislator’s membership on the Appropriations Committee led to an infusion of federal research funding over and above the amount allocated to states with similar geography and technology. We show that payback in this form generally led to increased numbers of degrees of a high-brow type. However, we find no evidence that the payback generated increased growth in the two far-from-frontier states, nor do we observe a prior increase in these states’ proximity to the technological frontier that might have justified the increase in funding (if we reason in terms of our model). In contrast, we find that Massachusetts did experience increased growth that coincides with its legislator using his position on the Appropriations Committee to generate substantial investments in research universities.
 
Vandenbussche et al. (2006): higher education as a factor in economic growth in the OECD
 
There are 30 countries in the OECD (Organization for Economic Cooperation and Development): Australia, Austria, Belgium, Canada, Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Japan, Luxembourg, Mexico, Netherlands, New Zealand, Norway, Poland, Portugal, Slovakia, South Korea, Spain, Sweden, Switzerland, Turkey, United Kingdom, and the United States. Vandenbussche et al. (2006) examined economic growth in 19 of the OECD countries for which all the needed data for the study was available: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Greece, Ireland, Italy, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom, and the United States.
 
Like Aghion et al. in the study just discussed, Vandenbussche et al. focused their analysis on lagged public education expenditures as the principal instrument. They also used data on educational attainment that distinguished between the percentage of the population that had attained tertiary levels of education and those with only primary/secondary levels of schooling. Two different data sets were used for this purpose: Barro and Lee 2001 and De la Fuente and Domenech 2006. As it turned out, both data sets yielded very similar results, so the findings were robust. Vandenbussche et al. were also able to use Unesco data on public expenditures on all levels of education between 1950 and 1990. Estimating the model involved using the log of the lagged expenditures per capita on primary/secondary and on tertiary levels of education, respectively.
 
Although the study found significant heterogeneous effects for such expenditures, it also found (p. 115) that most and maybe all OECD countries would benefit from having a larger fraction of skilled workers (defined as a larger fraction of workers with tertiary education). Their study found a very strong, positive, and significant interaction effect between the tertiary years of education and proximity to the technological frontier. All countries with a productivity above 74% to that frontier benefit by the presence of adults with tertiary education.
 
One interesting and important finding of the study is that in economies in the top 10% of the skill distribution (which for 2000 included four countries), proximity to the technological frontier has a positive effect on subsequent growth. I mentioned above the economic theory of conditional convergence, according to which advanced economies achieve significant levels of economic growth (given certain conditions) even though they have to compete in the global marketplace with less developed countries that can simply imitate established technologies—a much easier task than innovating new ones. The finding of the Vandenbussche et al. study underlines the importance of attaining high levels of skills in the population as one of the most important conditions for achieving continued high growth in advanced economies.
 
The range of economies that can benefit from increased expenditures on higher education may be somewhat narrower according to the Aghion et al. study than the one the Vandenbussche et al. study describes. This is an important question that needs to be resolved by further research. Both studies do show, however, that for the most advanced economies, increased expenditures on higher education are likely to promote economic growth. This is a remarkable finding, since both studies used expenditures on tertiary or higher education as the principal instrument. As I noted above, increased K-12 expenditures have proven to be a fairly weak predictor of economic growth.
 
A recent report, Higher aspirations: An agenda for reforming European universities (2008), has recommended that European countries increase their support of higher education in order to be more competitive with the United States. (One of the co-authors of Higher Aspirations was Philippe Aghion; another co-author was Caroline Hoxby, currently a professor of economics at Stanford, a Senior Fellow of the Hoover Institution, and a member—like Eric Hanushek—of the Koret K-12 Education Task Force.) Europe now spends 1.1% of its GDP on tertiary and higher education. The Higher Aspirations report recommended that European countries get closer to the 3.3% that the U.S. is spending on higher education in order to be competitive in the global marketplace.
 
Studies like those of Aghion et al. and Vandenbussche et al. have sharpened the findings of earlier work on economic growth, like Barro’s, that showed the importance of secondary and higher level education for economic growth. Such studies are particularly important because some analysts have claimed that, while basic education at the K-12 deserves public support, tertiary and higher education does not, or at the very least does not deserve more than it already has. (E.g., Wolf 2002.) The evidence from economics research indicates that this view has it exactly the wrong way around, at least for very advanced economies.
 
As Hanushek and others have pointed out, the question is complicated by the fact that K-12 feeds into higher education, so any weakness in K-12 will have an adverse impact on higher education. Still, the evidence shows that economic growth in advanced countries is highly dependent on the ability of the higher education sector to develop cognitive skills beyond the basic level, with primary and secondary school levels constituting an important indirect, “feeder” effect. For advanced economies especially, a strong K-12 sector is a necessary but hardly a sufficient condition for continued economic growth in the competitive global marketplace. The real payoff of improvements in K-12, at least for advanced economies, appears to come primarily at the level of tertiary and higher education.
 
What kinds of cognitive skills are important for economic growth?
 
Econometricians, who are increasingly convinced of the importance of the cognitive skills of the labor force as a determinant of economic growth, have speculated about the mechanism involved. In building their economic models and theories, some economists have theorized that higher education plays an important role in economic growth because modern economies demand technological innovation. Accordingly, in many current models it is primarily scientific and engineering expertise that drive economic growth. (Think Silicon Valley, the paradigmatic example of technologically driven economic growth.)
 
This emphasis on science and math skills is reflected in the international comparisons of test scores we have considered above. Hanushek et al. focus on these skills, as did G. W. Phillips in his comparisons between other countries and individual states of the United States. One of the most widely used and discussed international tests—Trends in International Mathematics and Science Study, or TIMSS—is devoted exclusively to the assessment of math and science skills.
 
Math and science skills are important cognitive skills to study, for at least two reasons. First, having a cadre of scientists and engineers that can innovate new technologies to drive economic growth is important in advanced economies. This technological innovation requires training in specific disciplines. Second, math and science skills are important because training in these fields has been found to be a good way to develop more general or generic cognitive skills.
 
While having a cadre of workers with specific skills in technological fields is undoubtedly an important factor in economies like ours, it is doubtful that this is the only mechanism driving growth in advanced economies. If it were the sole mechanism, and the cognitive skills of other graduates contributed little or nothing, then we should have to view science and math grads as doing all the heavy lifting when it comes to economic growth: graduates in all other disciplines would, in effect, just be dead weight. This is hard to reconcile with the relatively low percentage of students in STEM fields, even in economically advanced countries. Only 6.4 percent of U.S. graduates earned their degrees in engineering in 2004. Even in other OECD countries, the proportion of students graduating with degrees in engineering—while higher than in the U.S.—is fairly low: Canada (7.8 percent), France (12.4 percent), Italy (15.5 percent), Germany (16.5 percent), and Japan (20.2 percent). It is unlikely that these graduates are the only ones contributing to economic growth in these countries.
 
Another reason for thinking that math and science skills are not the only engines of economic growth is that econometricians have found that scores on literacy tests and scores on math and science tests are highly correlated (Hanushek and Woessmann 2008: 625, fn. 25):
 
It should be noted, however, that results from different tests, even when described as measuring different parts of the distribution of cognitive skills, tend to be highly correlated—particularly at the country level. Hanushek and Zhang (2008) correlate the literacy test results from IALS (identified as measuring basic skills) with the higher order math test results from the Third International Mathematics and Science Study (TIMSS) for comparable age groups in 1995 and find a simple correlation of 0.73. This consistency suggests that the available assessments of cognitive skills may capture a wider range of knowledge and skills than would be suggested by the descriptions of the individual tests.
 
A correlation of 0.73 is a very high correlation in the social and psychological sciences; accordingly, it can be expected that future research will show that economic growth is correlated with reading and literacy skills as well. The correlations that researchers have found between reading or literacy skills and math and science skills should caution us against supposing that only STEM fields—important as they undoubtedly are—make a contribution to economic growth in advanced economies.
 
The Collegiate Learning Assessment (CLA), which I discussed at some length in a previous paper, provides corroborative evidence of the high correlation that Hanushek and others have found between math and science skills and reading and literacy skills. A preliminary study of the first two years of a college cohort showed that gains on the cognitive skills measured by the CLA varied depending on the student’s major or field of concentration. Students concentrating in education, business, and social work made significantly fewer cognitive gains in their first two years of college study. Those who had emphasized math and sciences during the same period did significantly better, and those who had emphasized humanities and the social sciences did as well as students with concentrations in math and science.
 
I also discussed this issue at the K-12 level in my “Education and Intelligence—A Response to Charles Murray.” As I pointed out there, it is a mistake to view reading tests like NAEP’s as tests of reading or grammatical skills and nothing more. As the U.S. Department of Education itself insists, the NAEP reading tests involve tests of reasoning ability and not just grammatical skills. Doing well on the tests requires the student to use analytical and cognitive skills that transcend simply knowing English and the meaning of the sentences. Critical thinking, analytic reasoning, problem solving, and written communication skills are engaged by these tests.
 
The NAEP tests appear to have been the model for the Collegiate Learning Assessment (CLA), so it is particularly interesting that an OECD project called the International Assessment of Higher Education Learning Outcomes (AHELO), which I have mentioned above, is adapting the CLA in the construction of its own test of cognitive skills for cross-national studies and comparisons.
 
The AHELO feasibility study consists of four strands: a generic skills strand, a discipline-specific strand, a learning-in-context strand, and a value-added strand. The value-added strand will attempt to measure how much value has been added to the student’s cognitive skills by the college experience. This strand will focus on gains, not just on measuring college outputs. The discipline-specific strand will include a performance assessment component in order to measure how well students have been prepared for jobs in the career field they have chosen. Engineering and economics have been selected for this part of the feasibility study.
 
The most interesting strand of AHELO, at least for our present purposes, is the general skills strand. AHELO defines generic cognitive skills in the following way:
 
Building a bridge leaves no margin for error. Nor does criminal law, where an oversight could have a devastating effect on a person’s future. Engineers and lawyers have little in common, except that skilled analysis lies at the heart of both their professions. Analytic reasoning is defined under AHELO as a “generic” skill, applicable to any number of fields. Other generic skills include critical thinking, the ability to generate fresh ideas, and the practical application of theory.
 
What AHELO has called analytic reasoning probably comes pretty close to describing our intuitive notion of critical thinking, and both may in turn come close to what we mean by intelligence—or at least to the concept of developed intelligence.
 
Developing intelligence is one of the most important functions of education. Even the transmission of an intellectual or cultural heritage is only possible when students have attained at least a minimal level of critical or analytical intelligence. For example, readers of this essay are encouraged to reflect on the CLA’s sample performance assessment test item involving the proposed purchase by a corporation of a small private plane by its sales force. How meaningful could it possibly be to try to transmit the tradition of human learning to students who performed poorly on this performance assessment task? Analyzing such a problem is no different in kind from analyzing a Shakespearean sonnet, or critically analyzing the arguments in the Federalist Papers, or discussing the causes of the American Civil War or the Great Depression. The latter tasks require a lot more knowledge about history and other matters, but so far as analytical or critical reasoning is concerned, the main difference is that the latter tasks are a lot harder.
 
It is significant, therefore, that the preliminary results of the CLA show that college students do make significant gains on generic cognitive skills and reasoning ability. The findings show that higher education does much more than simply impart facts in academic disciplines. It actually in a broad, general, and fundamental way makes college students smarter.
 
It is a no-brainer that the U.S. needs to develop advanced cognitive skills in order to prepare students for managerial and professional jobs. Higher education is the education sector that has this responsibility. And while reasonable people should be open to alternative suggestions, no one since the advent of our modern technological society and economy has thought of a way to develop these skills systematically and in the way needed except through education that is directly or indirectly publicly funded.
 
The international standing of American higher education
 
Although they have been among the top scoring countries internationally on tests of cognitive skills like TIMMS, PISA, IALS etc. at the K-12 level, Pacific Rim countries also recognize the economic importance of the advanced skills that tertiary and higher education aim to develop. Two of these Pacific Rim countries—Korea and Japan—typically achieve proficiency levels on these exams—about 50 points higher than the U.S. average. Despite their success in attaining proficiency levels of achievement on cognitive skills tests at the K-12 level, Korea and Japan are participating in the OECD’s projected AHELO study of advanced cognitive skills. Korea is a participant in the generic skills strand, and Japan is a participant in the engineering strand.
 
As I’ve already mentioned, Hanushek has invoked the high quality of American higher education to explain the fact that U.S. economic growth rates have been higher than would be expected giving our middling scores on cognitive skills at the K-12 level. Hanushek’s high opinion of U.S. higher education is shared by other countries. In the Shanghai University, Times Higher Education, and other international rankings, no other country in the world approaches the U.S. in rankings of the world’s best colleges and universities. And while no one from economically competitive countries like Korea and Japan comes to the U.S. to enroll in our public schools at the K-12 level, students from these countries do flock to American colleges and universities.
 
I have ocular evidence of the high regard that high-performing East Asian countries have of American higher education in my home town of Berkeley. My path to the main library on campus (which I frequent) takes me past a new, imposing building called the Chang-Lin Tien Center for East Asian Studies. The Center has been named after a previous chancellor of the university (Chang-Lin Tien), and was largely built with funds that Tien raised from alumni of the university from Pacific Rim countries. In another quadrant of the campus—the engineering and sciences sector—are other imposing buildings that were constructed with funds generated in large part by Tien’s prodigiously successful fundraising efforts in East Asia among UC Berkeley alumni. Many of these alumni did their undergraduate or graduate work at U.C. and then returned to their home countries, where they helped build their nation’s strong economies and in the process, apparently, made a lot of money.
 
I often think of Chancellor Tien when I hear criticisms of American higher education. (Tien was also a fierce opponent of Prop. 209 and a staunch supporter of racial preferences in admissions at U.C., but that, as they say, is another story.) To judge from international rankings, and the large numbers of foreign students enrolled in American colleges and universities, it is safe to say that American superiority in higher education (unlike American K-12 education) is unquestioned internationally. Nor are these judgments confined to elite, selective research institutions like Berkeley. International students in this country seem to be everywhere. I have been struck by how many foreign students there are on community college campuses in California, and this impression can be backed up with data. (See the section “Trends in International Student Enrollment at Community Colleges” on p. 8 here.) It is true that community colleges enroll far fewer international students than doctoral/research universities in the U.S., but they also enroll far more of them than baccalaureate colleges and almost as many as Master’s granting institutions.
 
When critics complain about the quality of American higher education, listen to them because their criticisms are often valid and important—but keep in mind that it is important to maintain perspective as well. When colleges and universities are criticized, it is also relevant to ask—Good or bad relative to what? In the judgment of other countries and in the recognized international rankings, American colleges and universities across a broad range of institutions look pretty good.
 
A moving target?: economic growth and American higher education in the future
 
The research discussed in this paper is based on data that covers fairly long time intervals; most cover time-periods of a decade or more. (Some go back as far as the 1960s, though this is now unusual.) Even when the panel data does not go back beyond the 1990s, it is still legitimate to ask how valid the data is for some of our present concerns.
 
One of the principal arguments for expanding funding for higher education, or even for maintaining existing levels of funding, is that Americans need to be equipped with enough cognitive skills to adapt and survive in a rapidly changing environment. But the objects under study here—higher education, the American economy, and our international competitiveness—are themselves very much a part of the changing scene. How well can the economic research discussed in this essay throw light on what is happening now and what will happen in the future?
 
In particular, it is reasonable to ask how the current economic downturn might redraw the picture that has been painted here. One prominent aspect of the current economic crisis is high unemployment and underemployment. (In underemployment, workers either want to work full-time but can find only part-time work, or are forced into low-wage jobs even though they have high-wage skills.) The bad news at the beginning of the month was that the jobless rate in the U.S. had soared to 9.7 percent. 216,000 jobs were lost in August alone.
Almost 7 million jobs altogether have been lost during this recession, the most severe economic slowdown since the Great Depression of the 1930s.
 
Most economists are predicting that high unemployment and underemployment, and the concomitant loss of consumer confidence, will significantly impede any rapid economic recovery. Many of the 6.9 million jobs shed since the recession began in December 2007, analysts predict, will never come back. Some economists are predicting that economic growth in the U.S. will fall to as low as 2 percent next year. Even more worrisome, this is a global slowdown, and the severity and nature of the causes suggest that the effects might be long-term.
 
It is important to keep the notion of an under-performing economy in mind in order to properly evaluate a common argument made by those who are skeptical of the value of higher education. This criticism or skepticism is not new, but the current economic downturn seems to have given it renewed force recently. The skeptics have asked: If higher education is a significant contributor to economic growth and productivity, why are so many college graduates unemployed or underemployed?
 
Actually, there are many answers to the query that have nothing to do with the alleged worthlessness of higher education. There are millions of college graduates each year. (In 2003-04, American public and private colleges graduated 2,755,202 students.) The total number of college graduates is, of course, much larger. As a purely statistical matter, the number of individuals who are either unemployed or underemployed at any given time in such a large pool will also be large (for any number of reasons), even when the economy is good.
 
The problem becomes much worse, however—and a matter of concern for economists and policy makers—when the economy is in a recession or depression. What happens then is a cascading effect. Many college graduates do become unemployed or underemployed. However, even in a bad job market, college graduates are more employable than those without a college degree. Instead of becoming temporarily unemployed or underemployed, like many college graduates, workers without a college education often become permanently underemployed or unemployed, and therefore part of a severely depressed underclass.
 
Here is how a recent OECD publication, Education at a Glance 2009, puts it:
 
While enrolments for 15-19 year-olds have been steadily rising in most countries…, this still leaves an important minority who leave education without acquiring a baseline qualification. Across OECD countries, 42% of the 25-64 year-olds with less than an upper secondary qualification are not even employed…. Even those with higher levels of education are vulnerable if they become unemployed. Young people with lower qualifications who become unemployed are also more likely to spend a long time out of work: in most countries over half of low-qualified unemployed 25-34 year-olds are long-term unemployed. … In contrast, as noted before, those in work enjoy high wage premiums for completing tertiary education – over 50% in most countries.
 
The cascading effect occurs because employers favor college graduates over non-graduates. It has been argued that this preference on the part of employers is irrational, but this is a rather odd argument, coming as it does most often from those who think of themselves as defenders of free markets and the great wisdom of the entrepreneurial class. Furthermore, the claim does not withstand examination. There are perfectly good reasons why employers prefer college graduates. I presented some of them in a previous essay called “The Sheepskin Effect.”
 
The argument that goes from data about unemployment or underemployment to claims that a significant part of the U.S. workforce is overeducated also assumes significant levels of unemployment and underemployment as the norm. Unless we are indeed entering a protracted period of significantly slower growth, the critics’ argument lacks traction. Assuming that the economy is able to get back on its feet and start functioning again at something close to functional full employment, all the foregoing arguments showing that higher education contributes significantly to economic growth will remain in full force. Indeed, as the manufacturing sector continues to shrink (as most analysts and economists continue to believe it will), and the professional and managerial sector continues to grow (as economists and analysts also predict), the role of higher education in contributing to economic growth can only increase. As I pointed out in my previous discussion of the Collegiate Learning Assessment (CLA), the skills measured by the CLA are precisely those that are required by professional and even entry-level managerial jobs.
 
Unemployment and underemployment do raise significant issues about college costs. Unemployment and underemployment are exacerbated for college graduates if, as is increasingly the case, they are saddled with high debt when they graduate. Consequently, recent trends if they continue will change the cost-benefit analysis of acquiring a college degree. The trends could also result in differently weighted coefficients in the wage equations in labor economics and in economic growth models
 
There are, of course, many other criticisms to be made of American higher education. The National Association of Scholars has made many of them. Besides being much too expensive, higher education in the U.S. is infected with political correctness; spends far too much on functions unrelated to teaching and research (sports, lavish residence halls, costly student services that have nothing to do with learning, etc.); and fails to give nearly enough attention to teaching and student outcomes at the undergraduate level—to mention just a few selected problems.
 
At the same time, it is important to be aware of higher education’s strengths. If the U.S. fails to give higher education the continuing support that it needs and deserves, America’s commanding position in the world of higher education will be lost. This is not just a matter of prestige, like beating the Soviets to the moon. It would have a dire consequence for an advanced economy like ours—lower economic growth.
 
Many valid and important criticisms can and need to be aimed at American higher education, but as I’ve argued here at some length, the claim that it hasn’t contributed significantly to our productivity and economic growth doesn’t appear to be one of them.
 

BIBLIOGRAPHY
 
2009
 
Philippe Aghion, Leah Boustan, Caroline Hoxby, and Jerome Vandenbussche. “The Causal Impact of Education on Economic Growth: Evidence from the United States.” Brookings Papers on Economic Activity. Spring 2009 Conference Draft.
 
Ben Wildavsky. “How America’s mania for college rankings went global.” Washington Monthly September-October 2009.
 
2008
 
Philippe Aghion, Mathias Dewatripont, Caroline Hoxby, Andreu Mas-Colell and André Sapir. Higher aspirations: An agenda for reforming European universities. Bruegel blueprint, Volume V, July 2008.
 
Eric Hanushek, Dean Jamison, Eliot Jamison, and Ludger Woessmann. 2008. “Education and Economic Growth: It’s not just going to school, but learning something while there that matters.” Education Next, Hoover Institution, Spring 2008, pp. 62-70.
 
Eric A. Hanushek and Ludger Woessmann. 2008. "The Role of Cognitive Skills in Economic Development." Journal of Economic Literature (2008) 46(3): 607–68.
 
2007
 
Jens Arnold, Andrea Bassanini and Stefano Scarpetta. "Solow or Lucas?: Testing growth models using panel data from OECD countries." OECD Economics Department Working Papers No. 592, Dec 20 2007.
 
Daniel Cohen and Marcelo Soto. “Growth and human capital: good data, good results.” Journal of Economic Growth (2007) 12:51–76.
 
2007a Eric A. Hanushek and Ludger Woessmann. “The role of school improvement in economic development.”NBER Working Paper 12832.
 
2007b Eric A. Hanushek and Ludger Wößmann. 2007. “The Role of Education Quality in Economic Growth.” World Bank Policy Research Working Paper 4122, February 2007.
 
Gary W. Phillips. Chance Favors the Prepared Mind: Mathematics and Science Indicators for Comparing States and Nations. American Institutes for Research. November 14, 2007.
 
2006
 
A. De la Fuente and R. Domenech, R. “Human capital in growth regressions: How much difference does data quality make?” Journal of the European Economic Association (2006) 4(1), 1–36, 2006.
 
Jérôme Vandenbussche, Philippe Aghion, and Costas Meghir. “Growth, distance to frontier and composition of human capital.” Journal of Economic Growth (2006) 11: 97–127.
 
2005 Philippe Aghion, Leah Boustany, Caroline Hoxby, Jerome Vandenbussche. “Exploiting States’Mistakes to Identify the Causal Impact of Higher Education on Growth.” NBER Conference Paper.
 
2002
 
A. Bassanini and S. Scarpetta. “Does Human Capital Matter for Growth in OECD Countries? A Pooled Mean Group Approach”, Economics Letters (2002) 74(3), 399-405.
 
Eric A. Hanushek. “The Long Run Importance of School Quality.” Education Next, Hoover Institution.
 
Alison Wolf. Does education matter?: Myths about education and economic growth. Penguin Books (2002).
 
2001
 
Robert J. Barro. “Education and Economic Growth,” in J.F. Helliwell, ed., The Contribution of Human and Social Capital to Sustained Economic Growth and Well-Being, OECD, 2001. Keynote speech, OECD Symposium: The Contribution of Human and Social Capital to Sustained Economic Growth and Well-Being. Canada, Château Frontenac, Québec City, March 19-21, 2000. (Available free online.)
 
Barro, Robert J. and Jong-Wha Lee. “International Data on Educational Attainment: Updates and Implications,” Oxford Economic Papers, July 2001. (Available free online as Harvard Economics Department working paper.)
 
Alan B. Krueger and Mikael Lindahl. "Education for Growth: Why and for Whom?" Journal of Economic Literature 39(4), 1101-1136, December. (Also available online.)
 
J. W. Lee and R. J. Barro. “Schooling Quality in a Cross-Section of Countries,” Economica, November 2001. NBER Working Paper 6198, September 2001.
 
Lant Pritchett “Where Has All the Education Gone?” The World Bank Economic Review, Vol. 15, No. 3 (2001), pp. 367-391.
 
2000
 
Mark Bils and Peter J. Klenow. “Does Schooling Cause Growth?” The American Economic Review, Vol. 90, No. 5 (Dec., 2000), pp. 1160-1183.
 
Eric A. Hanushek and Dennis D. Kimko. “Schooling, Labor-Force Quality, and the Growth of Nations.” The American Economic Review, Vol. 90, No. 5 (Dec., 2000), pp. 1184-1208.
 
1999 David Card. “The causal effect of education on earnings.” Chapter 30. Handbook of Labor Economics, O. Ashenfelter and D. Card eds., Volume 3 1999). Elsevier Science.
 
1997 Jong-Wha Lee and Robert J. Barro. “Schooling Quality in a cross section of countries.” NBER Working Paper 6198.
 
1996 Robert J. Barro. Determinants of economic growth: a cross-country empirical study. Cambridge, MA: National Bureau of Economic Research. NBER Working Paper 5698. Prepared for the Lionel Robbins Lectures, delivered at the London School of Economics, February
20-22, 1996. (Not available free online, but see Google Books.)
 
1995 Doo Won Lee and Tong Hun Lee. “Human capital and economic growth: tests based on the international evaluation of educational achievement. Economics Letters 47 (1995) 219-225.
 
1994 Jess Benhabib and Mark M. Spiegel. “The role of human capital in economic development: evidence from aggregate cross-country data. Journal of Monetary Economics 34 (1994) 143-173.
 
1977 Zvi Griliches. “Estimating the Returns to Schooling: some Econometric Problems. Econometrica, Vol. 45, No. 1 (Jan., 1977), pp. 1-22.
 
 
Online and Miscellaneous Resources
(in no particular order)
 
Wikipedia: Econometrics
 
Wikipedia: Economic growth.
 
 
 
Wikipedia: Panel analysis

Wikipedia: Panel data
 
 
2009 Materials on the OECD’s International Assessment of Higher Education Learning Outcomes (AHELO):
 
 
 
 
 
 
 
Angus Maddison's Historical Dataseries (series for almost all countries on GDP, Population and GDP per capita from the year 0 up to 2003)
 
Angus Maddison: The World Economy: A Millennial Perspective (OECD Development Centre Studies)
 
Angus Maddison: The World Economy: A Millennial Perspective/ Historical Statistics (Development Centre Studies) (Paperback)
 
Angus Maddison. The World Economy. Published OECD Publishing. Volume 1: A Millennial Perspective and Volume 2: Historical Statistics. PDF format.
 
 
Paul M. Romer. “Beyond Classical and Keynesian Macroeconomic Policy.” From Policy Options, July-August 1994. [Paul Romer's plain-English explanation of endogenous growth theory.]
 
 
Andrew C. Porter and Adam Gamoran (2002). Methodological Advances in Cross-National Surveys of Educational Achievement. Board on International Comparative Studies in Education, National Research Council. National Academies Press. (See also Google Books.)
 
Lapointe, A.E., Askew, J.M., and Mead, N.A. (1992). Learning science. The International Assessment of Educational Progress (Rep. No. 22-CAEP-02). Princeton, NJ: Educational Testing Service.
 
Lapointe, A.E., Askew, J.M., and Mead, N.A. (1992). Learning mathematics. The International Assessment of Educational Progress (Rep. No. 22-CAEP-01). Princeton, NJ: Educational Testing Service.
 
 
 
  • Share

Most Commented

September 16, 2019

Slavery Did Not Make America Rich

'King Cotton' isn't King

September 18, 2019

Most Read

January 03, 2011

May 26, 2010