What does the Dale and Krueger education paper really say?

It was reported in the media as showing that, controlling for all the right variables, going to an elite college or university as an undergraduate doesn't really matter for your future prospects or income.  But Robin Hanson, with money on the line, investigated further.  After reading the relevant pieces closely, he reports [what follows is Robin, not me, but with the multiple indentations I haven't indented everything again]:

"In fact his original 1998 working-paper abstract said:

We find that students who attended colleges with higher average SAT scores do not earn more than other students who were accepted and rejected by comparable schools but attended a college with a lower average SAT score.  However the Barron's rating of school selectivity and the tuition charged by the school are significantly related to the students' subsequent earnings. 

Half Sigma screams from the rooftops:

Based on the straightforward regression results in column 1, men who attend the most competitive colleges [according to Barron's 1982 ratings] earn 23 percent more than men who attend very competitive colleges, other variables in the equation being equal.

23 percent is quite a bit of money, it’s almost like getting two college degrees instead of one!  They also discovered that there was a benefit to attending a more expensive school. The more expensive tuition resulted in a lifetime internal rate of return of 20% for men and 25% for women."


I feel much better about the $40,000 I spent on my Harvard education back in the late 1970s and early 1980s.

Those higher returns are probably industry-specific. For instance, what if you just focused on English majors or Philosophy majors? Conditional on SAT scores, is there still an Ivy premium? Dartmouth and Harvard grads make a lot of money b/c there's selection into high-earning occupations, like finance, associated with them.

Could it be that people that went to more expensive schools come from wealthier families? Isn't the greatest predictor of future income the income of your parents?

That 2006 Atlantic Monthly article did go on to say:
A study by Caroline Hoxby ... suggests that graduates of elite schools do earn more than those of comparable ability who attended other colleges. Hoxby studied male students who entered college in 1982, and ... projected that among students of similar aptitude, those who attended the most selective colleges would earn an average of $2.9 million during their careers; those who attended the next most selective colleges would earn $2.8 million; and those who attended all other colleges would average $2.5 million.

So...by those figures, the difference is about 16% for 'most selective' vs 'all other' and 3.5% for 'most selective' vs 'next most selective'. My translation of this into advice for prospective students is that if you are sure you are headed for a high-paying profession (and even more if what attracts you about it is the high pay), then by all means attend the most selective, expensive private institution that will accept you. BUT...if you know you're headed for a lower-paying profession (or aren't at all sure), go to the state school. Huge student loans and lower-paying work is a really unpleasant, life-stunting combination. And you also don't want to put yourself in a position, at age 20, where you feel you have to keep on pursuing a degree and a profession that no longer appeals just because it's the only way you're ever going to dig yourself out of your student loan debt.

True, but the coefficients are much attenuated. In general, however, trying to difference out the unobservable skills in the matching/self-revelation models really knocks down the effects of school characteristics, which is their point and one missed with the "23 percent is a lot of money" comments.

If I were to argue against their results, I would do so on the grounds that their comparison aleviates one source of bias but creates another: what kind of people get into Harvard but then decide to go to a lesser-ranked school? People who want to play basketball at UNC or football at Florida State? People who think they don't need a Harvard diploma to succeed (think of famous Harvard dropouts)? That's a completely different kind of selection bias.

Is there a possible confounding personality variable here? Among the set of students facing a choice between a very high prestige college and another college, it seems possible that those choosing the very high prestige college are, on average, more confident, or more focused on external signs of achievement, than those who choose a less prestigeous college that may seem less frightening or a better fit? And might such students be likely to earn more money independently of where they went to college? This is pure speculation, and I'm not sure how you would correct for it, if the self-selection exists at all.

How about if the student goes to the cheaper school and you give him the difference in cost in cash.

Like I like to say a high school drop out with a backhoe can make more that high school with a shovel.

The missing piece in this discussion is that these papers were
done back in 1998. Tuition in private schools have since soared
substantially compared to what they were and what state schools
have done. I suspect those rates of return on the more expensive
schools are not going to look quite so wonderful if measured today,
although the selectivity relations may well still hold.

Krueger also has studies saying the minimum wage has no effect on employment. Therefore, we cannot believe any of his research.

It's amazing how clueless some of you are about our labor markets. We have a career track based system. The career track you get onto determines your life's earning potential. The prestige of your academic credentials makes it more likely that you will get onto a better career track.

How can you people not understand this? Why do you think so many parents put so much effort into ensuring that their children get into the good colleges. They understand how the system works.

Adam's comments above are correct: the 23% estimate, as the paper makes painstakingly clear in many pages of discussion, cannot be interpreted as causal. It is disingenuous to quote it as if the authors are claiming otherwise.

The preferred estimates are in the "matched applicant" columns, which are attempts to control for the selectivity which biases the "straightforward" models in a robust manner. Table 7 shows that the Barron's index of college selectivity is not statistically significant in these models. The estimates in Table 8a suggest an elasticity for men of earnings to tuition of about 0.05 with a t-stat of about 2.1. For women the point estimate is an elasticity of 0.012 which is not statistically significant. Taken together, these are pretty sketchy estimates on which to base a claim that more expensive schools benefit the average student. However, the models with interactions suggest that more expensive schools do benefit both men and women from poor families.

As the paper points out on page 10, this method of controlling for selection bias is hardly bulletproof, as the problem of explaining why one student accepted at both Harvard and Podunk goes to Harvard whereas another goes to Podunk is not confronted. If it is the case that the student who went to Harvard is unobservably more able than the one who went to Podunk, the matched applicant models are also biased. The paper notes they may even be more biased than the plain OLS estimates.

Put me in the camp that thinks paying a lot to go to undergrad is mainly a waste of money.

Most of the big money jobs require a graduate degree and if you have good grades and test scores you will get in to good grad schools from pretty much anywhere.

Plus once you get somewhere (work or grad school) your talent is really what makes the difference.

The only caveat to me is if you want to go into certain jobs with a high snob factor (investment banking) right out of college it may make sense to go to the high prestige school.

The biggest waste to me seems to be to pay a lot to go to a second tier private school and major in a field that has weak job prospects. In that case all you do is graduate with a boat load of debt. Better to be debt free.

Living in a state with good public universities (North Carolina) I will be strongly encouraging my daughter to go to a public undergrad. It worked out great for my wife and I, two Podunk U grads (NC State, Mich State) who still managed to get into a position in our early 30's where we will be seeing a income tax icrease from Obama. And we are far from alone in our circle of friends.

Remember that there has been a significant shift in admissions and cost at the most elite college over the past 20 years that is probably not in that 1998 data. They used to be the home of the elite social classes, and now they are much more tilted towards nerdish academic performance with lavish scholarship money. I have heard that Penn State U. has the highest average family income of any PA school, including Penn and Swarthmore.

I'll agree with Adam and Chris Auld. It seems as if Robin Hanson is taking a more accurate and less biased piece of data and replacing it with a more biased and less accurate piece of data to make a point.

Comments for this post are closed