Bryan Caplan  


I, Christmas... Exchange and the Human Conditi...
Most researchers oversell their results.  After re-reading Dale and Krueger's latest piece on the selectivity premium, however, I suspect that they are greatly underselling their results.  They haven't just undermined the value of academic selectivity; they've confirmed the value of personal ambition.

Background: Labor economists routinely find that degrees from more selective colleges raise income more than degrees from less selective colleges.  D&K replicate this result using standard control variables: your SAT score, high school GPA, etc.  But then D&K add controls for (a) the average SAT of the schools you applied to, and (b) the number of applications you submitted.  After adding these controls, the selectivity premium actually vanishes.  In their latest paper, this is true if you measure selectivity by (a) your college's average SAT, (b) your college's net tuition, or (c) your college's Barron's Index.

D&K conclude that the selectivity premium is probably greatly exaggerated, at least on average.  But I draw a much bigger lesson: D&K have (a) proposed very plausible measures of what we intuitively call "ambition," and (b) shown that ambition so measured has a huge effect on income.  As far as I can tell, D&K's measures are much stronger than the other measures of "non-cognitive ability" (especially the NLSY's Locus of Control and Self-Esteem scores) that economists have been toying with over the last decade.

Here are D&K's log income results for their 1989 cohort.  The "Basic" results don't control for their measures of ambition; the "Self-Revelation" results do.


D&K focus on the fact that, after controlling for ambition, the premium for attending a school with an average SAT 100 points higher goes from +5.6% to -0.8%.  What's truly remarkable, though, is the size of the ambition premium.  Applying to schools with an average SAT 100 points higher has a +9.9% premium.  Applying to four additional schools has a +9.8% premium.  Notice, moreover, that the payoffs for SAT scores and high school GPA only moderately decline after controlling for ambition; Dale-Krueger's measures capture something fairly novel about a young adult's character.

This presumably doesn't mean, of course, that you can greatly increase your income by mailing out lots of Hail Mary applications.  Instead, it means that having the ambition to apply to lots of good schools greatly increases income.

To be fair, Dale and Krueger explicitly state that they're trying to control for ambition.  But they remain focused on connection between ambition and the selectivity premium.  Given the strength of their results, though, they should be pushing for a complete overhaul of the return to education literature.  What is the effect of a year of education on income... controlling for ambition?  What is the effect of education on unemployment... controlling for ambition?  What is the college major premium... controlling for ambition?  What happens to the sheepskin effect... controlling for ambition?  Inquiring minds want to know!

COMMENTS (10 to date)
Ignoto Fiorentino writes:

What reason is there to think that this reflects ambition as opposed to any other quality-related variable that is unobserved by the statistician but observed by the student and the college, and thus leads the student rationally to aim high and the college rationally to admit the student?

Ignoto Fiorentino writes:

E.g., quality of the student's high school or of the student's recommendation letters, neither of which Dale-Krueger control for?

Ambition could explain why a student applies to a selective college. It doesn't explain why they get in.

Charlie writes:

The study controls for the Average SAT or the college the student gets into. Thus, the study controls for where the student goes to school (and that turns out not to matter).

Kevin Driscoll writes:

@Charlie I don't think that's right. Controlling for Average SAT of the attended school doesn't mean that they remove this factor from the analysis. It just means that they track it carefully so that they are always comparing "apples to apples."

D&K found that among students who applied to a similar level of schools, there was basically no benefit to attending the more selective school. On the other hard, among students who attended similar schools those students who applied to many more selective programs showed a significant premium.

They had to control for average SAT of schools applied to so that they could make these comparisons. They just divided up their sample into a larger number of groups and found that one group (the 'ambitious' ones) saw basically all of the benefit of attending a selective school.

EclectEcon writes:

Why is there such a high premium for being Asian, ceteris paribus? Is that another proxy for ambition? ability?

MikeP writes:

Why is there such a high premium for being Asian, ceteris paribus? Is that another proxy for ambition? ability?

Being selected from a population of 3 billion as someone who has or whose parents have the ambition and/or ability to do high skilled study and work in the US?

Glen writes:

Why call this "ambition"?

More likely, SES. More selective schools cost more to get in, and depending on how much your parents make, this limits where you can go.

As for SAT, the affluent handle it the way they always have -- private tutoring. How many parents do you know that actually send in the applications themselves? Ah-huh. Or hire a private counselor to do it for them. Check.

Z Kromer writes:


It seems a bit of a leap to link D&K's controls to a general measure of ambition.

Ambition in the sphere of college application process, sure, but even here it is not obvious the student is driving the application behavior.

The controls measured-- the number of applications submitted and average SAT of schools applied to-- probably reveal as much about the applicant's parents, high school guidance counsellors, and zip code as they do about the student himself.

And these three factors really boil down to choices the parents have made. I would be willing to bet that controlling for parental achievements and geography (down to the zip code) would yeild results similar to D&K.

MG writes:

It is good that the study controls (implicitly, though the class cohorts) for "era of application". Nowadays, some of what may have been teased out - of ambition - may now be obscured by today's prevailing fine-tuned application strategies that may overstate or understate ambtion. For example, early action applicants who get in will bring into the data set only one app, understating latent ambition.

MingoV writes:

I think these types of analyses would work much better if they looked at the colleges (or schools) within the universities or technical institutes. A university may have a good reputation primarily because of its excellent college of liberal arts, but it may have a poor college of science. An ambitious science student would not apply there, an ambitious philosophy student would. The mean SAT scores might be vastly different between colleges at a university. I went to a technical institute, and the mean SAT scores of science and engineering students were far higher than those of fine arts, graphic arts, photography, and criminology students. Splitting out colleges and their mean SATs might give significantly different results.

Comments for this entry have been closed
Return to top