Babcock-Marks and Signaling
By Bryan Caplan
Babcock and Marks could reply, of course, that the return to college
would have been even greater if schools had maintained standards. But
the more natural inference is that studying is mostly wasteful signaling.
If one student cuts his effort by 40%, he gets low grades and looks
bad. But students in general can still cut their effort by 40% without
noticeably impairing their future productivity in the real world.
Philip Babcock objects that “Falling Study Times Don’t Imply College Is a Wasteful Signal”:
[I]t’s simply not the case that, as Bryan Caplan suggests,
“the natural inference” to be drawn from a rising college wage premium
in the presence of declining study time is that college is just
“wasteful signaling.” … [A]ny of these explanations
could be right even though study time is declining… This could simply mean, for example, that the effect of
skill-biased technical change is larger than previously thought, so
that even more modest increases in human capital now generate
significant wage gains.
To be clear, I never said that falling study times imply that college is signaling. Indeed, I specifically mentioned that a human capital theorist could simply insist that the return to education would have risen even more if student effort hadn’t declined. My claim, rather, is that the Babcock-Marks’ findings make the signaling model more probable. Here’s why:
Imagine that Babcock and Marks had found that study time declined by 100%. College students ceased studying altogether. Suppose further that the return to education still sharply increased during the same period. This would be exceedingly difficult for a human capital story to explain. After all, its central thesis is that students acquire job skills by studying!
While you could still posit a massive countervailing skill-biased technical change, it strains credulity. In contrast, it’s quite a bit easier for a signaling model to explain why the market might reward students who don’t study. Maybe they can show off their quality merely by gaining admission to college, then failing to get expelled for moral turpitude.
Of course, Babcock and Marks don’t find that studying disappeared entirely. But probabilistically, my point still holds. If studying sharply declines, and the labor market doesn’t care, this raises the probability that something other than straightforward human capital acquisition is at work.
Most of us believe that productivity-enhancing skills can be learned in
college (I certainly wouldn’t want to go to a doctor who hadn’t gone to
college.) We also believe there may be a signaling component to
education, as well. In our piece, we simply note that if learning is a
part of the university mission, or if studying leads to any meaningful
increase in human capital, then these declining study times should be
I completely agree that actual education is a mix of job skill acquisition and signaling.* The most precise way of stating my point is that, given Babcock and Marks’ findings, we should upwardly revise our estimate of signaling’s share of the mix. But Babcock’s wrong to conclude that “declining study times should be disturbing” either way.
It depends on the human capital/signaling mix. Even if you think that human capital acquisition has positive externalities, you have to weigh these against the negative externalities of signaling. And if you don’t think that human capital acquisition has positive externalities, then a voluntary fall in study time is reassuring, not disturbing. Students internalize the cost of lower human capital acquisition, and society reaps the benefit of lower negative externalities of wasteful displays of scholarship.
* However, for most ailments I would be happy to go to a doctor who went to a trade school rather than college – at least in a society where this wasn’t too weird.