Bryan Caplan  

Two Educational Heresies: Ability Bias vs. Signaling

Economists Overconfident?... Opting Out: Do Efficiency and ...

I believe that IQ matters quite a lot for earnings, and that a lot of education is mere signaling. But I've recently reached an epiphany about the relationship between these two heresies.

Note that the two heresies are distinct.

You can believe that IQ matters quite a lot for earnings, but still think that education teaches nothing but bona fide job market skills. If this is so, then comparing the earnings of college graduates to high school graduates overstates the private benefit of education. Why? College graduates were smarter to begin with, so they would have earned more money than the typical high school graduate even if they didn't go to college. Labor economists call this "ability bias."

Similarly, you can believe that a lot of education is mere signaling, without thinking that IQ by itself puts money in your pocket. Suppose that the world is rigidly credentialist, so that no one will even consider a person without a degree for anything beyond a low-skilled job. If this is so, then comparing the earnings of college graduates to high school graduates overstates the social benefit of education. Why? Because part of the effect of education is just to make yourself look better compared to other people without increasing production.*

However, signaling by itself does not imply that the private benefit of education is any less than it seems to be: If you want to cash in on your brains, you've got to suffer through school first.

Last, note that while these two positions are logically distinct, they are compatible. You can consistently think that a lot of education is mere hoop-jumping to distinguish yourself from other people, AND that even if a high-IQ person refused to jump academic hoops, he'd still end up earning more money than a low-IQ person who was unable to jump academic hoops. And that is exactly my view.

Why should you care if either or both heresies is true? Prudence alone tells you to worry about ability bias, because it leads brainy people - like you, valued reader! - to overestimate your payoff from education.

In contrast, from the standpoint of prudence, it makes no difference whether education is signaling or not. What counts is that employers value it; why they value it makes no difference. However, whether education is signaling or not makes a great deal of difference from an ethical viewpoint: The world as a whole is better off if people have more productive skills, but not if people jump through more academic hoops.

As a professor, it's admittedly contrary to my interest for people to understand either of my favorite educational heresies. If they understood ability bias, they'd be less likely to go to college; if they understood signaling, they'd be less likely to vote to subsidize college. But who listens to me, anyway?

* There is some social benefit of signaling, because it helps match abilities to jobs; but a lot of it is still socially wasteful because you could have matched people about as well if everyone got, say, half as much education.

Comments and Sharing

COMMENTS (9 to date)
aaron writes:

I might disagree that an unintelligent person is uncapable of jumping acedemic hoops.

conchis writes:

I'm not sure the heresies are as distinct as you make out:

(1) To believe that IQ matters for earnings, but that there is no signalling in education, you need to deny any correlation between IQ and getting more education: otherwise employers have every incentive to use it partly as a signalling device to identify those with higher IQ.

(2) To believe that a lot of education is signalling, without believeing that IQ improves productivity, you need to think that education is signalling something else that does improve productivity, otherwise no-one would use it as a signal. This is easier than believing (1) (you might think it signals motivation/ambition for example), but it's still not entirely unproblematic (the motivation/ambition argument is pretty circular - motivated/ambitious people will do whatever brings success).

Tom Schofield writes:

Income and IQ may correllate in the mass, but surely we have all encountered well-paid and expensively educated buffoons. On the other hand, highly-compensated,intelligent autodidacts seem pretty scarce - outside entrepreurial circles, anyway.

Based only on personal experience, I believe that your second proposition in undoubtedly true but that the first must be hedged with multiple caveats.

blink writes:

I agree that, to a large extent, education signals productivity. If so, we should see a higher return to education where performance is more costly to observe. This is intuitively appealing – athletes, actors, and entrepreneurs receive a low return to education, while the number of degrees earned determines a large part of teachers' pay at all levels.

These comparisons may not be fair, however, so I propose a test. In several industries, technological improvements allow employers to measure productivity more accurately and at a lower cost than in the past. To the extent that education is signaling, the return to education will decline as this technology is adopted. Perhaps data are already available for an informal confirmation.

edj writes:

Bryan I generally agree. It does seem there are a few exceptions like Tom mentions, but I think you are right on as usual.

You are awesome, btw.

eric writes:

I'm an econ phd, and as a professional nonacademic, I think that the axiom of equal innate ability is a big problem with labor economists, and other economists, because people are manifestly unequal in innate ability, not just in particular, but in groups. The Korean adoption data would be sufficient in less loaded arenas to convince people.

I find it ironic that economists pride themselves on being realists in the sense of emphasizing selfishness, and thus objective, though it seems rather dismal to others. But they are still very pollyannish in terms of assuming that, say, the poor are just as smart as the rich, only with different levels of social connections, education, and of course wealth. Many poor people are of low IQ, and this is relevant to their future poverty. It's a statistical relation, with exceptions, but it's still very relevant (the fact that some black people are rich, but on average they are less wealthy than whites is prima facia evidence of racism to many, but the fact that most poor people are of low IQ is considered irrelevant because some low IQ people are rich and vice versa--plus, it seems so unfair).

I remember the hatchet job in the JEL on The Bell Curve, sanctioned by the best and brightest of economic research, which basically argued that because y=f(x,y), you simply can't tease out the effect of x. Labor economists routinely tease out relationships in multicollinear worlds in other aspects with great care, and great success, but it seems to these same people that estimating IQ's effect is impossible (Heckman's other tact, that IQ is malleable, was abandoned when he found that IQ was pretty consistent over an individual's life). It really turned me off to economics as an academic discipline, because this was just a partisan smear campaign against a result they didn't like (they had the temerity to say that it was "irresponsible" to present such results to a broad audience without presenting it first to an academic journal--we want monopoly control of the axiom of equality!).

conchis writes:

Is it really that much of a heresy in Labor economics to think that IQ matters? The entire literature using IVs to try to measure the returns to education is premised on the idea of ability bias: that some people are inherently more able than others and that those people tend to get more schooling. OK, so it's also half-premised on measurement error, and it's also true that most of the results tend to suggest that the latter is a bigger problem, but I don't know any labor economists who are seriously convinced that it doesn't play a role. (And more recent work estimating dynamic programming models seems to reach the opposite conclusion.)

(All of which is consistent with the criticisms of the Bell Curve if you attribute it to the general academic tendency to require overwhelming proof before being willing to accept anything anybody else says.)

Niclas Berggren writes:

This post made me think of an article by Weede and Kämpf, the abstrct of which reads:

Average scores on IQ tests are more strongly related to economic growth than standard indicators of human capital formation are. This is demonstrated for the 1960s to 1990s with multiple regressions. By and large, institutional improvements or more economic freedom and average intelligence exert an approximately equally strong influence on growth rates.

See the Kyklos citation.

Bourg writes:

I'm an econ phd, and as a professional nonacademic, I think that the axiom of equal innate ability is a big problem with labor economists, and other economists, because people are manifestly unequal in innate ability, not just in particular, but in groups.

I don't which school you did your PhD in, but you could have used a little bit of exposure to labor economics. Why don't your try and do a little search using the following words "returns, schooling, training, education, ability"? The problem of differences in innate ability is the one problem every labor economist who deals in these issues is aware of. I mean the chief example given in omitted variable bias is a log wage equation with an ability measure.

Comments for this entry have been closed
Return to top