False signals on the need for college graduates

Catherine Rampell of the New York Times recently looked at job growth by education and interprets the data as suggestive of the “polarization” of employment where employer demand is growing at the most and least-educated categories but falling for those in the middle. Rampell uses this analysis to argue that “college is worth it.” My analysis of the employment trends in recent years, elaborated below, runs directly counter to a common interpretation of the “polarization” hypothesis, which is that we have a growing unmet need for college graduates. I’m all for giving everyone the opportunity to get the most training and education they want. And, I’m sure that there are plenty of economic and non-economic (i.e., health, citizenry) ways we’d be better off with more college graduates. I’m just not persuaded that economic data are screaming out that we need to greatly accelerate the supply of college graduates.

I’m not much of a fan of the hypothesis that recent technological change is leading to a “polarization” of wages. I’m especially suspicious when analysts lump all college graduates together in their analysis, combining those with four-year college degrees (22 percent of employment) and those with advanced or professional degrees (11 percent of employment). This is because college graduates (those with bachelor’s degrees only) have not fared well in the labor market for at least 10 years—real wages are no higher than 10 years ago—while those with advanced degrees have seen their wages grow strongly. I have covered this ground in Education is Not the Cure for High Unemployment or for Income Inequality. In fact, those economists who argue that employment and wages are polarizing, such as Larry Katz of Harvard or David Autor of MIT, are pretty clear that employment outcomes for about half of college grads are part of the “middle” that’s faring poorly. That’s why it is misleading to use those analyses to argue that having more people go to college is the answer to growing wage inequality or middle-class wage stagnation; getting onto the better wage track requires either getting an advanced or professional degree (not just a college degree) or joining a clear subset of college graduates. That being the case, the arguments of those seeing “polarization” in the data lay out a very narrow track to good earnings and, in my view, further raise the issue of the need to make sure that those without college degrees, and many with college degrees, have good quality jobs.

More extensive elaboration of these issues will have to wait for another time to explore. Now, it is worth digging into Rampell’s analysis, especially since it reaches conclusions contrary to two blog posts (read here and here) where I presented data showing that the falling unemployment among college graduates (unfortunately, because of data availability, using all college graduates, bachelor’s or higher) over the last two years was primarily due to labor force shrinkage rather than strong employment growth. For context, it should be noted that in earlier work I documented that the unemployment rate doubled for every educational group during the period of rising unemployment, including those with a college degree or further education.

Rampell presents the absolute employment growth over the last 12 months (December over December), noting college graduate employment was up over a million, high school graduate employment down about 550,000 and employment up by roughly 125,000 for high school “dropouts.” I take this a step further in Table 1 and look at changes over the last two years (from when unemployment peaked) and improve the analysis by using quarterly data (less volatile) and calculating percent growth in employment (necessary since the education groups are different sizes, “dropouts” being less than one-fourth the size of college graduates).

Table 1: Employment growth by education, ages 25+

Less than high school High school Some college College or more
2009-10 -2.5% 1.0% 0.9% 0.5%
2010-11 1.9% -1.0% 0.3% 2.3%
2009-11 -0.6% -0.1% 1.2% 2.8%

Both years have a different pattern. Over the last four quarters there was weak employment growth for the middle two education groups and stronger growth at the top and bottom ends. The year before, however, saw weaker employment growth among college graduates (up 0.5 percent) than for those with high school (up 1.0 percent) or “some college” (up 0.9 percent). Looking over the last two years as a whole one finds employment growth better the higher the education level. This is not strong evidence of “polarization,” or even the bastardized version of the hypothesis (the one where all college graduates are presumed “winners”).

The best way to analyze these trends is to examine the employment rates (employment divided by population) of each group, as done in Table 2. This scales the changes to the size of the population involved. Overall, the employment rate has not changed over the last two years, rising just 0.1 percent. Unemployment fell by 1.2 percentage points, from 9.9 percent in 2009:4 to 8.7 percent in 2011:4, but that is entirely explained by a shrinkage of the labor force. Interestingly, the employment rate has declined for all but the “dropouts” and has declined the most for those with a college degree (or more) and those with “some college.” As observed at the start, these results run directly counter to a common interpretation of the “polarization” hypothesis, that we have a growing unmet need for college graduates and that one of our key challenges is to greatly accelerate the supply of college graduates.

Table 2: Changes in employment-to-population ratio, 2009-11


Quarter 4


2009 2010 2011 09-10 10-11 09-11
All 58.4 58.3 58.5 -0.1 0.2 0.1
Education, 25 years and older
Less than high school 39.2 39.3 40.6 0.1 1.3 1.4
High school 55.1 54.9 54.7 -0.2 -0.2 -0.4
Some college 64.4 64.2 63.5 -0.2 -0.7 -0.9
College degree or more 73.6 72.9 72.7 -0.7 -0.2 -0.9