When Is College Worth It?

Today's RealClearPolicy slate features a piece from Dylan Matthews of the Washington Post claiming that college is a good investment. It pushes back against arguments in the vein of critics like Charles Murray and Richard Vedder, who argue that too many people are going to college.

Several years ago, I wrote a few pieces for National Review outlining the Murray/Vedder view (here's one from the website), but I hadn't kept up with the debate, and Matthews's piece features a lot of newer research on the subject. What follows isn't necessarily a response, but more of an overview of the topic and a reconsideration of my views in light of the updated evidence. I have six main points to make.

1. It's not about college, it's about college-for-all.

Matthews spends a considerable amount of time explaining that attending college is a good idea for the average student. This is true -- but to the best of my knowledge, no college critic disputes it. For a lot of high-paying jobs, you simply need a college degree, no matter how smart you are; today's doctors, teachers, engineers, and so forth would not be nearly so well off if they'd stopped after high school.

What we argue is that not all students benefit from college the way that the average student does, and that efforts to draw even more students to college could do more harm than good, because such efforts are focused on students who today rationally decide that college isn't for them. We especially bristle at the notion that America should aspire to send all students to college.

2. How do you calculate a college wage premium?

As Matthews notes, it doesn't really prove anything that college grads tend to make more than people who stop after high school. Getting a diploma requires a lot of intrinsic qualities -- intelligence, diligence, etc. -- and those qualities would benefit today's college grads even if college didn't exist. When we're debating whether to send a certain group of students to school, we have to ask how those students in particular will fare with and without college, not how grads and non-grads fare in general.

To calculate the wage premium for a certain group of people, you need to factor in what schools they'll be attending, what majors they'll pursue, their odds of graduating, and what opportunities the job market will offer them. Then you need to factor in tuition, as well as the opportunity cost of missing several years in the work force to study. And if you're a policymaker rather than a parent, you need to look at the government money that's spent too.

Economists do their best to sort this out through intense statistical work, but it's a difficult problem, and even the best studies are to be taken with a grain of salt -- this type of work is just too imprecise to prove much, despite its complexity. For example, it apparently wasn't until 2008 that someone realized that college graduates are more likely to live in expensive urban areas. Once you correct for this, the overall college wage premium is reduced significantly. How many other things are these analyses missing?

That said, a study Matthews cites makes a plausible argument (with a lot of technical calculations to go with it): Though marginal students often get less of a return on investment than the average student, they can still see worthwhile income gains -- if you pick the right students. As one of the study's authors, James Heckman, told Megan McArdle: "If you’re not college ready, then the answer is no, it’s not worth it."

If Heckman is right, we'll need to choose very carefully, because about two-thirds of high-school grads head to some kind of college already, and about one-third of students aren't college-ready in even a single subject area according to the ACT.

3. So, what policies might add marginal students in a productive manner?

Naturally occurring experiments can give us a good way to evaluate policies, and Matthews cites a number of studies that take this approach. But while all of them point in the direction of "more education is good," it's hard to draw many conclusions more precise than that.

For example, in one study, researchers looked at students who spent slightly longer than others in high school before becoming old enough to drop out, thanks to the days their birthdays happened to fall on. These students seemed to benefit. But the policy implications for college, as opposed to high school, are far from obvious.

In another, researchers looked at the distance students lived from colleges, and found that living near a campus increased income. This might suggest efforts to bring college closer to people who might benefit from it. But as Matthews notes, areas with colleges might make people richer for reasons having nothing to do with education.

Still another study found benefits from the Job Corps program. But this program isn't college -- it's vocational education, which is what college-for-all critics often encourage in lieu of college.

Easily the most convincing study Matthews cites looked at Florida students who were slightly above and slightly below the GPA cutoff (usually 3.0) for Florida International University. Above that cutoff, many students are able to attend a less-selective state school instead of a community college. On a graph matching the students' GPAs to their future income, there's a sudden jump in income at the cutoff, suggesting that better access to college benefited these students, and that it might be a good idea to encourage students in this ability range to go to four-year schools. However, this is one study covering a very narrow slice of the college population in a single state. I'd like to see this method applied to other data.

4. By the way, what do basic descriptive data say about college students?

Even if some students who could benefit from college aren't going, we can't conclude that all the students attending now belong there. A few simple statistics undergird most critiques of college-for-all, and they're worth repeating here.

While 66 percent of high-school grads head to two- or four-year colleges in the fall, almost half fail to graduate in six years -- and six years is pretty much the norm for four-year-degree students. Up to 40 percent of undergrads need remedial courses because they are not ready for college-level work. According to one assessment, close to half of students don't measurably improve their skills in a number of areas over their years in college. Meanwhile, even before the recession, about 25 percent of recent grads were "mal-employed," meaning they were in a job that didn't actually require a degree.

Staunch defenders of the "college wage premium" often point out that college is correlated with moderately higher earnings even in these cases -- i.e., college dropouts and mal-employed grads make a bit more than people who stopped after high school. But it's awfully difficult to see how, say, a given waitress or bartender is more valuable with a college degree than without.

More likely, people who managed to get into college and/or graduate have other qualities that employers value, as outlined above in No. 2. At most, their college admission or diploma signaled those qualities during the hiring process, thus increasing income in a way that had nothing to do with anything they learned.

5. If college is sometimes just "signaling," what are the policy implications?

According to Matthews, arguments like the one I just made don't "really negate the finding that college causes higher earnings. [They just present] a slightly cynical explanation of how it is that college causes that earnings bump."

It's not mere cynicism, though. If many college degrees are just signaling underlying abilities, and the time spent earning them doesn't actually mean anything, we should be finding cheaper ways to measure and signal abilities while cutting subsidies for college.

6. Charts!

In researching this post I spent some time with the National Longitudinal Survey of Youth, which has interviewed a cross-section of Americans born between 1980 and 1984 annually since 1997.* I came up with charts that nicely illustrate both sides of this debate.

Here's one in which students are sorted according to their decile rank on the ASVAB, which is similar to an IQ test, as well as their income in 2010. As you can see, even within each grouping, students with better degrees made more. (When there were fewer than 20 respondents in a category -- e.g. folks with lowest-decile ASVAB scores and advanced degrees -- I left it off.)