Very sadly, Princeton labor economist Alan Krueger died by suicide last week.  Krueger was much-loved in the profession. He produced an enormous amount of work on the minimum wage and for a couple of years served as Chair of President Obama’s Council of Economic Advisers.  He also thought a lot about education and famously declared in one paper (co-authored with Stacy Dale) that the effects of attending an extremely selective university (i.e. one of the top Ivy League Schools) on later earned income was “generally indistinguishable from zero” after taking various considerations into account. 

This finding has been written about in many places (such as the Atlantic, various Malcolm Gladwell pieces, etc) and it has achieved a kind of peculiar cult status: it’s one of those stats everyone knows, but literally nobody believes.  I mean, really, no one.  Tell an American their kid can go to Virginia or Harvard and the long-term outcomes will be exactly the same and my guess 100% of them will say “great, my kid’s going to Harvard!”  I’ve always found this to be fascinating and so I thought it would be worthwhile to go back over that famous 2002 paper (and a follow-up paper he wrote with Dale a decade later) to see what he actually said and whether there was a reason people might be skeptical.

Here is the original paper: Estimating the Payoff to Attending a More Selective College: An Application on Observable and Unobservables.  The problem it sets out to solve is, more or less as follows: “look, we know graduates of super-selective universities make more money than graduates from less selective universities, but what we really want to know is to what extent that is a function of the fact that they choose better students or that they actually add more value?”  This is not an easily answered question: to actually answer this, one would need to have access to long-run economic data of identically matched students (well, matched on academic ability, anyway) who attended schools of different levels of selectivity.  This is not easy to do since data sets which tell you about income (tax records, labour force surveys, etc.) don’t contain data on academic ability or alma mater, and data on academic backgrounds contains no data on lifetime income streams. 

The genius of this particular paper (as is true of many economics papers) is the way in which Card and Krueger used a unique data source to answer the question: namely, the College and Beyond Survey (CBS), which was a project of the Mellon Foundation in the mid-90s.  What this survey did was gather a sample of graduates from 34 colleges across the US, including 4 of the most selective in the country (Columbia, Yale, UPenn and Princeton) from the matriculating classes of 1951, 1976 and 1989.  From the participating institutions, they got data from their admissions file and their scholarly career; the Foundation then contracted the research firm Mathematica to contact the graduates and ask them questions about, among other things, a) their current (i.e. 1996) occupation and earnings and b) what other schools to which they had considered applying back when they went to school.  What Card and Krueger found is that the average level of school selectivity (that is, the SAT scores required to get into a school) ceased to have any effects on earnings or occupation once the students’ own SAT scores were taken into account.  They then verified that result using data from the National Longitudinal Survey of 1972 high school graduates which contains similar but not identical data – and got a similar though perhaps less powerful result (due to sample size).  The one exception to this finding was lower-income students, who did see higher incomes (Krueger hypothesized that it was because it allowed them to form professional connections they would not otherwise have found). 

So, one left-wing triumphalist interpretation of this piece at the time was “Ha!  Public education is just as good as private education!” and there was some truth to that.  But the results were actually trickier than that.  In effect, what it suggested was that when it comes to long-term outcomes of post-secondary, the extra money the Ivies spent doesn’t really change very much, which could undermine arguments for the utility of extra public fundingtoo.  And even the exceptions to the rule – lower-income students – were hypothesized to do better not because of extra investment but because of access to personal networks.  It made you wonder, in effect, what the purpose of any investment in higher education was for, if all long term outcomes could be predicted by SAT scores.

Now, there are a couple of potential criticisms of this research.  One is that the CBS is a pretty limited sample of colleges, most of which are reasonablyselective if not exceedingly so.  I think this is a valid critique of anyone trying to extrapolate the results to argue that money or selectivity never matters, but if the only point you want to make is that holding SATs constant, a top selective school makes no difference compared to a second-tier selective or flagship public, then this is irrelevant.

The bigger issue, perhaps, is that the result was derived from a sample of graduates who entered university in 1972 – the point in American history where the private Ivies were at their lowest ebb relative to the public, both in terms of income and selectivity.  It was also a time before the finance industry went berserk, sucking in an ever-greater percentage of American GDP, spending it on a tiny group of people who – as Lauren Rivera memorably showed in her brilliant but horrifying book Pedigree: How Elite Students Get Elite Jobs – are drawn overwhelmingly from these top privates.  It could simply be that what was true in the mid-1970s is no longer true in the mid-2010s.

We’ll never know.  In another paper from 2011, Krueger and Card linked the data from the 1989 CBS cohort to Social Security admin data, which allowed them to examine the first 15 years of that cohort’s income stream, and found results essentially identical to 1976.  This is stronger evidence for the enduring nature of the phenomena he described, but even 1989 was still early days in terms of the widening gap in private college affluence and the transmogrification of Wall Street.  There’s still a possibility that today’s parents are right to ignore Krueger and Card’s findings.

Even though the CBS ended with a 1989 cohort, I feel sure that somehow, Alan Krueger would have found a way to continue to measure this effect into later cohorts.  Maybe someone else will pick up the challenge; we can only hope it will be someone as committed to improving lives through smarter public policy as he was.