Thursday, November 19, 2015

The Free College Problem

Allison Schrager is an economist who writes about retirement and how to hedge risk in more unconventional situations. But in this article, she addresses the question of free college and whether or not it addresses the bigger problem.

Her arguments echo several being brought up as free college emerges as a Democratic platform item.

The first, largest issue is college completion.

Poor students are far less likely to finish college than their rich counterparts. And that includes poor kids who are smart and get high scores on, well, anything. Here's a chart that lays it out:

Poor students apply to less selective schools, and their are fewer poor students who rank as high achievers (which is unsurprising since "high achiever" means "good standardized test score-getter" which we know doesn't correlate closely with poverty).

It's possible that tuition costs are part of what forces poor students out of school, and that free tuition might help. But there's also a strong case to be made that poor students take all of the problems of poverty to college with them. It's not just that it costs money to pay tuition to go to college; it costs money just to be there, to live in a lifestyle that is in many ways upper class. It's like tossing students over the wall into an exclusive swimming pool without ever checking to see if they can swim.

And here's another depressing factoid. We can talk about how hard it is for poor students to finish college, but data suggests that middle class students have a lousy completion rate as well.

The second issue is just how much value a college education provides.

Folks keep discussing college degrees as if there's a direct correlation between degree and lifetime earnings. The emphasis on Common Core college and career readiness is predicated on the notion that if everyone had a college degree, everyone would be making much more money. Some of this notion is based on the work of guys like Raj Chetty, whose research is hugely doubtworthy. And it fails the common sense test-- if everyone had a college degree, what would happen to minimum wage jobs? Would McDonald's start paying big bucks to burger flippers with BA's? Would those jobs just vanish somehow?

What research suggests repeatedly is that your eventual earning power is best predicted by that of your parents. I've seen various charts for these data, but here's one that Schrager uses

In other words, a man who comes from the lowest SES level who gets a Bachelors degree will still make about a third of the lifetime earnings of a rich-kid high school grad.

I've seen various numbers associated with this argument, but the basic point remains unchanged-- college does not remotely come close to magically erasing the effects of your SES of origin.

So does all this mean that free tuition is a bust of an idea? I don't think so-- a college degree is still worth having (though good welding certification is also an excellent career move). But to suggest that free college will cure societies ills, reverse social injustice, and revitalize America's stalling social mobility-- well, it's not going to do those things. It's foolish to expect it to, and even more foolish to institute free tuition and then declare, "Mission accomplished," and stop looking for better solutions for the underlying issues.


  1. "We can talk about how hard it is for poor students to finish college, but data suggests that middle class students have a lousy completion rate as well."

    So mostly rich kids finish college. How much of that is because they are more likely to have what it takes to finish college (whether academic or other skills), and how much is because rich parents can hint about the size of the donation they'd like to make if only Junior's grades were higher?

    Also, how much of the value of a college education comes from the connections made there? Which, of course, would naturally be greater for rich kids attending elite schools than for poor kids attending For Profit U? College is often just the junior version of the golf course.

  2. If anyone is interested in the attempts to estimate the returns to education and the issues that make this difficult can look here:,%20Econometrics-%20Facts%20and%20Data%20Issues/Econometric%20Issues%20in%20Higher%20Education.pdf

    My biggest problem with paying for education using general tax funds is that with any politically plausible tax structure and the social structure of post secondary education it involves a transfer of resources from the relatively poor to the relatively rich.


    One of the papers cited in the link above (Dale, S., Krueger, A., 2002. Estimating the payo4 to attending a more selective college: an application of
    selection on unobservables. Quarterly Journal of Economics 117, 1491–1528.) compares students accepted at highly selective universities but who ended up attending less selective universities (like the one where I teach) to those who attended the highly selective universities. They found that there was a significant gain for minority students and students from relatively poor households to attending the highly selective private university but there was no gain to majority students who did not come from poor households. The authors speculate that the connections made an elite college are not as important to relatively wealthy white students as they are to relatively poor African-Americans.

  3. I should add that there is an interesting contrast between this post and the previous post. If test scores simply measure SES status (as the last post suggested), shouldn't we see that low income students graduate at the same low rate no matter their score on mathematics tests and high income students graduate at the same high rate independent no matter their score on mathematics tests?

    Instead of seeing no relationship, we see that relatively poor students scoring in the highest quartile on standardized math exams graduate from college at over 8 times the rate of relatively poor students scoring in the lowest quartile on those exams and relatively wealthy students in the highest quartile on standardized math exams graduate at a little under 4 times the rate of relatively wealthy students in the lowest quartile.

    1. I don't understand the contradiction. Yes, poor kids who are good at tests are much more likely to graduate college than poor kids who aren't - no surprise. But well-off kids are much more likely to be good at tests overall.

    2. Madeleine,

      The post below this one concludes that "...this research does stand as one more data point regarding standardized tests and their ability to measure SES far better than they measure anything else."

      The post here points out that standardized tests appear to do a good job of measuring the traits that will enable any student, no matter their income level, to graduate from college.

    3. BSTs measure the ability to test well, a useful enough skill, though incidentally it isn't especially interesting to teachers (in fact, it's an embuggerance – it's much harder to assess learning when your students are deft test-takers).

      But the important point is that poor students are much less likely to be good at tests than better-off ones. And BSTs don’t shed any light on that.

      So - when you hear that a particular school gets great BST results, this means only that it's in a good neighborhood. And we can find this kind of thing out without imposing endless demands on students.

    4. Madeleine,

      If your exams do not help you asses learning by your students, perhaps you are using the wrong exams. If your exam results do not conform to your assessment of student learning, perhaps you should reexamine your non-exam based assessments.