A review of The Race between Education and Technology, by Claudia Goldin and Lawrence F. Katz

The academic establishment has been effusive in its praise of Claudia Goldin and Lawrence Katz'snew book, The Race between Education and Technology. Larry Summers, Harvard guru and chief economic advisor to President Barack Obama, says "this is empirical economic scholarship at its finest." Princeton's Alan Krueger adds, "this book represents the best of what economics has to offer," sentiments echoed by the University of Chicago's Steven Levitt and University of Rochester's Stan Engerman. Let me be a bit of a skunk at the love fest by offering an alternative view.

Goldin and Katz, professors of economics at Harvard, argue that American higher education contributed, during the 19th and most of the 20th century, both to America's long-term economic growth and the achievement of its egalitarian ideals. In the past three decades, however, a dramatic slowdown in the growth in educational attainment has contributed to something of a reduction in American economic growth and, most importantly to the authors, a sharp rise in economic inequality.

More specifically, Goldin and Katz argue that education enhances productivity, and the recent slowdown in the growth of skills has reduced productivity and income growth. Because technological advances are continuing as fast as ever, and since these tend to favor higher skill attainment, the demand for educated Americans is rising rapidly, which, other things being equal, increases their relative wages. Until about 1975, this rise in demand was counteracted by an equally or even more impressive growth in the supply of highly educated workers, so relative wages did not accelerate, and wage-induced inequality did not rise-indeed, it even fell at times. The slow growth in educational attainment since 1970 or 1975, however, has slowed the supply increase, so relative wages of the educated are rising relatively rapidly, increasing income inequality, which "many commentators" believe "can contribute to social and political discord." What should we do? The authors recommend that we promote better educational performance at the K-12 level through smaller class sizes and higher teacher salaries. We should increase spending for early childhood education. And we should deal with rising college costs by "more generous college financial aid for low-income youth…." To deal with rising inequality, they also favor increasing income tax progressivity and judiciously using labor market "institutional interventions" like theminimum wage. In short, increase government spending and regulation, and make taxation more progressive—more or less the prescriptions adopted by the Obama Administration in its stimulus bill and proposed federal budget.

The authors are industrious and detailed in making their arguments, mining neglected historical documents such as the Iowa State Census of 1915. There is even a four-page discussion, complete with regression results, of the role that electricity played in promoting high school graduation in the early 20th century. The authors spent literally years doing the research on educational attainment and economic inequality that gives the book something of an authoritative aura.

Yet I think they make a number of implicit assumptions that are debatable and call into question the accuracy of the story they tell, not to mention the proposed remedies for the alleged problem. Let me outline six: educational attainment lowers inequality; lower inequality is needed and desirable; higher educational attainment promotes economic growth; more public spending will promote higher attainment;therefore, more public education spending will mean more economic growth;and finally, differences between human beings in cognitive skills and other personal attributes are relatively unimportant.

The notion that enhanced educational attainment will raise income equality rests on the assumption that an increase in the proportion of educated workers relative to less educated ones will lower wages of the former group relative to the latter. This is what probably happened around 1970 when college enrollments were soaring. It might well happen that way, but statistical work that Daniel Bennett and I have been conducting suggests that the relation between educational attainment and income equality is not so clear-cut. Some preliminary results even suggest the opposite: higher educational attainment is associated with more inequality. But even if Goldin and Katz are right, there are other reasons why conventional policy responses may have little effect. To cite one problem, my associates and I at the Center for College Affordability and Productivity have observed that the statistical relationship between state higher education appropriations and the proportion of adults with college degrees is not positive, even after allowing for various lags for graduation rates to catch up with the higher spending.

Goldin and Katz clearly believe greater income equality is good for America. They grudgingly admit that "some degree of economic inequality may [emphasis added] be desirable to spur incentives…." But there is no discussion of the equality-efficiency trade-off. Why is the income distribution of 1970 "good" and that of 2008 "bad"? How do we know? Inequality's alleged threat to social order is not documented, and some polling results suggest interpersonal variations in income are less controversial than the authors suggest, Obama's election notwithstanding. Moreover, the inequality in the distribution of lifetime income, or of consumption spending, is far less substantial than suggested by income figures for any single year.

* * *

It is true that areas with relatively high levels of educational attainment have typically grown more than areas with less educated populations. But the question is: will the resources devoted to increasing the proportion of well-educated people generate a return, in terms of incremental income, that justifies the expenditure? In reality, human beings differ, and those with the greatest motivation to acquire higher skills havelargely done so. Is it necessarily true that increasing the proportion of students attending college will promote higher incomes? Attending college is not costless, and those costs rise and the benefits decline as we reach further down into the pool of talent. The authors approvingly talk of surging educational attainment in Europe—but fail to note that economic growth rates there have fallen, averaging levels lower than in the U.S. in recent decades. Surging European educational attainment has been associated with falling, not rising, growth rates.

Goldin and Katz assume that higher spending on higher education will increase attainment rates. States spending more on higher education do indeed get higher college attendance, but only modestly so. Most such increases do not go to hold tuition charges down or increase financial aid to the poor, and therefore it isn't surprising that we do not find a positive relationship between state spending in higher education and the proportion of adults with college degrees. The call for more public spending on education implies it will enhance economic growth rates, and the authors even state that "the short-run fiscal burdens of increased spending on education are likely to be more than offset in the long run with increased tax revenues from a more productive workforce and lower public spending to combat social problems." Yet when we run numerous regressions using statewide data on the relationship between higher education spending and economic growth, we typically get a negative relationship: higher spending, lower growth. Money taken from the market-disciplined competitive sector to finance less efficient universities partially isolated from market forces by third-party payments seems to have a negative growth effect.

Goldin and Katz implicitly assume that education causes the difference in incomes and productivity between high school and college graduates. Underlying their argument is an unwritten assumption that all people are created equal in terms of cognitive endowments, motivation, discipline, etc. It is extremely difficult to correct econometrically for all the human variations in characteristics, and I believe at least some of the wage differentials that the authors associate with education relate to the fact that typically college graduates are brighter, harder working, and more dependable, than those who leave school after getting a high school diploma.Had they not even gone on to college, these young strivers would have earned more than the existing pool of high school graduates.

* * *

College is a screening device—a point ignored by the authors. Employers hire workers with college degrees because the diploma raises the probability that the graduate will be talented and diligent, because the colleges themselves have screened their entrants,admitting those with some prospects for success, and then not graduating close to half of them (another point ignored by Goldin and Katz), many for being deficient academically. While their book has a bibliography of well over 500 entries that covers 28 pages, there is not a single reference to The Bell Curve (1994), the well-known work of Richard Herrnstein and Charles Murray on the importance of cognitive endowments to educational attainment and vocational success.

Nowhere in this book do you learn that for every 100 students entering high school, only about 20 have four-year college degrees 10 years later. The problem often is not access but attrition. Nor do we learn that in 1970 only three out of every 100 mail carriers had college degrees, compared with 12 today. Do college-educated mail carriers promote economic growth? I doubt it. Maybe Richard Freeman's The Overeducated American (1976) is still relevant. Do we need more than 15,000 Americans with advanced degrees fixing people's hair? One study says that, five years after graduation, only 61% of college graduates are in jobs requiring a college degree.

In addition, the rising differential between high school and college earnings beginning in the late 1970s corresponds nicely with the beginning of enforcement of Duke v. Griggs Power (1971)a decision making it nearly impossible for employers to test prospective workers, forcing them to rely more on colleges to screen applicants for competency. (Goldin and Katz ignore, by the way, the stagnation in the high school/college earnings differential for females in the past 15 or 20 years.) The authors also accept soaring education costs as a given, not criticizing significantly the perverse effects of unionization and rapidly rising salaries for professors with light teaching loads. Perhaps higher education access would be better enhanced by making professors teach more, and reducing administrative and recreational expenses that contribute to the college cost explosion. Perhaps we should do cost-benefit analysis on more academic research. The authors' implicit faith that the ways of the academy are fixed and invulnerable to change, even as technology forces changes on everyone else, is one this writer cannot embrace.

In short, Goldin and Katz's errors of omission rival their errors of commission. The Race between Education and Technologyis an interesting book with some good data and historical analysis, but it fails to persuade because its authors seem blinded by a desire to increase resources for the business that feeds them.