William Dickens wrote:
>
> >My other side couldn't disagree more strongly. A significant percentage
> >of undergrads should really be in some kind of 2-year trade school
> >instead.
>
> Ahhh... But then you have to deal with the very empirical evidence that kicked this
>whole discussion off. If you think that people are making irrational (essentially
>random) decisions to go to school then the OLS estimates of the rate of return are
>right and the real returns to education are the best you can get anywhere. You may
>not think that they are getting much out of college, but the average return to each
>additional year of education is very high... the first order effect is something like
>at least a 6.5% increase in income for each year of school completed whether you get
>a degree or not.
----------------------------------------------------
OK, now for my long reply.
It is child's play to drastically shrink the number from 6.5-7.5% real
rate of return. The 7.5% number, you will recall, is the estimated NLSY
return controlling for AFQT scores. I'll use that for the sake of
argument.
1. This estimate treats the social and private cost of tuition as 0.
(A bizarre assumption to begin with if the conclusion is that there
should be more money for education!) Even a low estimate of tuition
costs at 50% of foregone earnings shrinks the return from 7.5% down to
5%. (7.5/[1+.5]) A more reasonable tuition estimate at 100% of
foregone earnings drops it to 3.75%. (7.5/[1+1])
2. The estimate doesn't factor in the analog of "default" costs.
People report how many years of schooling they *completed*. If you
don't work and pay tuition for a year, but fail out, those are resources
down the drain. Even a 1% annual default rate pulls down the estimate
from 3.75% to 2.7%. (1.0375*.99)
3. I am still convinced that OLS has to give an overestimate because
the marginal student gets less than the average student. Yes, in
theory, students with high discount rates might get more, as you pointed
out. But in practice, a high discount rate makes you a very poor
student - you sleep and party rather than attending class and doing hw.
In my experience, this effect dominates.
When I explain average vs. marginal returns to my students, I ask them
to compare the willingness to pay of the *average* Attack of the Clones
viewer vs. the marginal viewer. I'm sure the former is many times
higher. In theory, there could be a winner's curse - the people who
underestimate the most don't go. But I can't believe that matters much
in practice in movies or anything else.
The marginal return couldn't be more than 75% of the average return.
That brings us down to 1.8%.
4. Since you keep mentioning current disutility, I don't think you can
reject my next point: most people find school unpleasant, even relative
to work. Professors are a biased sample! If you add on a few thousand
dollars worth of disutility per year, that's maybe 25% of the cost of
foregone earnings, bringing our revised real return down to 1.5%.
I can't make this argument in good conscience, though, without granting
that more educated people probably enjoy their jobs more and have more
non-cash amenities, so the starting real return should have been
higher. Even if we put that starting number at 10% instead of 7.5%,
though, re-doing the numbers leaves us at 2.3%. Less than the real
return on T-bills, last time I checked. And riskier.
Private, Family, and Social Return
Note that that 2.3% rate of return is what could be called the "family"
rate of return. The *student's* rate of return is indeed higher if the
family pays. So maybe it is a no-brainer for most students, though it
shouldn't be for their parents.
What about the social return? Let's ignore externalities for the time
being, and just adjust for the fact that government pays part of tuition
expenses. If family tuition bills only pay for 2/3 of the full cost,
then the social return drops to 1.7%. If the family pays only half,
it's down to 1.3%.
Adjusting for Externalities
I readily grant that there are large externalities of education. But I
think they are NEGATIVE. Signalling isn't everything, but I am
convinced that a decent fraction of the private return is created by
making other people look bad because they endured fewer years of painful
schooling. Credentialism is all around us. Back when most professors
didn't have Ph.D.'s, was education worse? I doubt it.
Imagine half of the gross return (cash and non-cash) to education is
from signalling. Then that 1.3% social rate of return falls to .14%.
The positive externalities, in contrast, are pretty trivial. Less
crime? I can't believe you get much bang for your buck. My JLE piece
does find that more educated people "think more like economists," and
presumably better voters. But a lot of that effect is probably just
IQ. And universities also probably make people more liberal, leading
them to support wasteful programs such as ... education subsidies!
Naive efficiency analysis recommends Pigovian *taxes* on education,
though given how badly government has messed things up here it's
probably best for them to do nothing.
Obviously I'm taking a highly contrarian position here. But frankly, I
think labor economics exhibits a gross leftist bias on the topic of
education. They are doing everything they can to exaggerate the
benefits of education subsidies, and consign the signalling model to a
theory ghetto.
It's unclear how much labor economists' cheerleading has actually
influenced policy. But I do think it's clear that the real social
return to education has been pushed down to less than 1% by our current
system of subsidies. What a waste of time and money.
--
Prof. Bryan Caplan
Department of Economics George Mason University
http://www.bcaplan.com [EMAIL PROTECTED]
"He wrote a letter, but did not post it because he felt that no one
would have understood what he wanted to say, and besides it was not
necessary that anyone but himself should understand it."
Leo Tolstoy, *The Cossacks*
