The college wage premium—the difference between average earnings among those with a college (but no graduate) degree and those who do not attend college—has increased substantially in recent years while the premium for those who attend “some college” without actually earning a degree has not changed at all. This fact leads many observers to conclude that a college degree is the best way for young adults to attain the skills they need to earn more and thus reverse growing inequality. (See Figure 1.)

Figure 1


This view, however, has serious problems. The idea that not enough people are graduating from college implies that the much-reported rise in income inequality is thanks to a “shortage” of highly skilled college grads able to meet the labor market’s need. That idea has been conclusively debunked. The fact that the wage premium only kicks in when a college student receives a diploma, rather than gradually appearing in the cross section of people who go to college for one, two, or three years but don’t earn a degree, casts serious doubt on the idea that it’s the skills content of college that matters. Furthermore, the fact that buying an expensive degree correlates with high income certainly doesn’t imply that causation runs from buying the expensive degree to the high income.

So where in the academic literature did this notion of the four-year college degree as the solution to labor market inequality arise? The idea that the college wage premium reflects a rise in the labor market’s demand for skills stems in large part from a 1992 paper by Harvard University economist Lawrence Katz and University of Chicago economist Kevin Murphy, who argued that since there’s been both a rise in the college wage premium and a rise in the proportion of the population with college degrees, demand in the market for skilled labor has increased against a somewhat elastic but essentially unchanged supply curve.

Along similar lines, Katz and another Harvard professor Claudia Goldin, published a paper in 2007 that tracks the college premium over the long run and posits that its dynamics are explained by the race between education and technology. They argue that “skill-biased technological change” creates a demand for college degrees that takes time to be reflected in the skill composition of the workforce.

The story about the race between education and technology leaves questions unanswered. First, it cannot explain the significant differences between the income distributions across countries, especially at the very high end. Technological change and the distribution of individuals’ skills seem to be uniform across countries, at least in the developed economies, and yet their income distributions are very different. For instance, the distributions of harmonized standardized test results for high school math students in the United States and France are basically the same, yet the top ten percent of income earners accrue 25 percent of total labor market income in France and 35 percent in the United States- and the shares are even more skewed higher up the distribution.

Second, the argument that increasing inequality is caused by a shift in the demand for scarce skilled labor is only theoretical: it’s not at all clear where that technological change comes from. Every attempt to operationalize the theory of skill-biased technological change has run up against problematic data. Since the late 1990s, most of the increase in the college wage premium (which has not grown much during the last fifteen years) is due declining absolute wages for those with less education. That is the exact time period in which the “IT revolution” is supposed to have had a wide impact on experience of the middle class in the labor market. And it has had an impact—on the industrial mix of workers, but not on their wages. If there is a race between education and technology, currently the runners are tied: both supply and demand for skills have shifted such that wages are unaffected.

The potential harm in misattributing rising income inequality to a race between education and technology — a race that technology is winning — is that it could lead to perverse policy prescriptions. Trying to get more enrolled college students to undertake the cost of finishing their degrees might lead to yet further tuition hikes, especially if that route receives government subsidies, without significantly improving their outcomes in the labor market or reducing inequality overall.