Tuesday, November 1, 2011

College or not?

I was rather taken aback by this awful Brookings piece on the choice of whether or not to go to college.

My major concerns:

The discussion gives no hint that the labor market payoff to college may differ between the average student and the marginal student. Absent credit constraints and assuming full information, the marginal student is presently indifferent between the two choices. We are not in a full information, no credit constraint world but it is still quite likely that many if not most of those presently not attending college are making the correct choice (and some of those who are attending are making the incorrect one).

There is no mention of the fact that the earnings payoff to college depends a lot on what you study.

There is no mention of the fact that the earnings payoff to college depends on what college you attend.

While the piece mentions that finishing college is very important, it does not highlight the dismal completion rates of those who seek a four year degree but start at a two-year college or a mediocre four-year college. These student often end up with non-trivial debt but little earnings impact.

Unconditional earnings effects are presented with only a brief and dismissive mention of the importance of non-random selection into college. In fact, conditioning on "ability" as measured by tests scores reduces the earnings effect of college non-trivially.

More minor concerns:

No mention is made of the uncertainty implicit in estimates of earnings effects that rely on a synthetic cohort assumption. That is, it is assumed that old people today are a good proxy for what young people to day will earn if they do or do not get a college degree.

It is not correct to simply ignore the room and board costs of college. Yes, young people have to eat and live somewhere, but the private costs of both may be much lower at home.

In thinking about risk, it is important to remember that the nature of the riskiness associated with the college non-college choice is somewhat unusual. It is well known that earnings risk is lower for college graduates, in part because they are much likely to be unemployed. At the same time, credit constrained students can pile up a lot of debt in college and thus experience a long period of negative net worth that people who do not go to college are spared. So there is an asymmetry of sorts here.

Standard errors?

I recognize that the piece is trying to reach a popular audience, but I think that even a popular audience needs to hear about many of these issues.

In sum: ugh. And Michael Greenstone is a very smart guy. So I blame the other author, whom I do not know.

Via Brian McCall

No comments: