Bryan Caplan  

Signaling that I'm Not Signaling About Signaling

Rebranding Additive Shocks... Funding Dis-Education...
During our last few lunches, Robin has challenged me to clarify my position on educational signaling - and help my critics to do the same.  And more recently, he named "You have little interest in getting clear on what exactly is the position being argued" as a sign that "[Y]our opinions function more to signal loyalty and ability than to estimate truth."  Implication: If I clarify what my critics and I are arguing, I raise the probability that I'm a truth-seeker.  Or in other words, I can signal that I'm not signaling about signaling.

Here are the key points of contention as I see them:

1. On average, how much does learning course material causally increase students' marginal productivity?

My answer: It accounts for roughly 20% of the gross average return to education.

2. At the margin, how much does learning course material causally increase students' marginal productivity?

My answer: It accounts for roughly 20% of the gross marginal return to education.  The final years of course material are no more or less relevant in the real world than the early or middle years.  Even kindergartens spend a shockingly small fraction of their time teaching reading, writing, math, and other skills that their students will eventually use on the job.

3. On average, how much does schooling causally increase students' marginal productivity by shaping their character, a la Eliza Doolittle or the Marines?

My answer: Relative to a carefree life of play, this accounts for another 20% of the gross marginal return to education.  But relative to getting a job, the average character-shaping benefit of education is roughly zero.  For very young children with negative marginal productivity on the job, I'll admit the character-shaping effect is more positive.  But by the time they're teen-agers, the character-shaping effect is mildly negative relative to employment.  As I told Bill Dickens:
Work inculcates the worker ethos; school inculcates the student ethos.  The two are only moderately correlated.  The most obvious differences: Work offers much more tangible rewards for good performance, and much harsher punishments for bad performance, than almost any school.  School teaches students the wrong life lessons: Excellence doesn't lead to money or status, and disruptiveness won't get you fired.

Even worse, school often indirectly inculcates counter-productive character traits.  Students spend a lot of their energy trying to show their fellow students that they're defiant, cool, etc.
4. At the margin, how much does schooling causally increase students' marginal productivity by shaping their character?

My answer: Mildly negatively.  See above.

5. On average, how much does schooling increase students' wages by signaling their pre-existing marginal productivity?

My answer: Roughly 80% of the gross average return to education is signaling.

6. At the margin, how much does schooling increase students' wages by signaling their pre-existing marginal productivity?

My answer: Roughly 80% of the gross marginal return to education is signaling.

7. On average, what fraction of signaling has positive social value because it improves the match between workers and jobs, causally raising workers' marginal productivity?

My answer: Tough call, but my best guess is 20%.  IQ tests are great, but education also signals other important traits that are too subtle and/or fakeable to easily test.

8. At the margin, what fraction of signaling has positive social value because it improves the match between workers and jobs, causally raising workers' marginal productivity?

My answer: Another tough call, but I'll say 0-5%.  Adding an extra year to a nation's average education level does little to help employers better distinguish the top 5% of workers from the bottom 5%.

Anyone care to further refine the debate?

Comments and Sharing

COMMENTS (23 to date)
Peter writes:

A few other questions I'd like you to answer:

On average, what is the consumption value of socialization among a diverse and relatively high functioning peer group during a university education?

At the margin, what fraction of the accounting and opportunity cost of university education is regained in consumption of networking, leisure, and interesting coursework?

At the margin, how much is an individual student's marginal productivity improved by socialization during college?

At the margin, how much is an individual elite (Ivy League and echelon) university student's marginal productivity improved by socialization during college?

Other than education, how would one signal the following traits effectively (assuming one is at the 90th percentile for each): conscientiousness, ability to stick to a commitment for multiple years, ability to conform to instructions.

Greg writes:

I suppose it's a starting point, but this all seems too general to be very useful. I think we need a more granular debate about the value of education. For example, a biology degree for someone who goes on to be a biologist increases productivity substantially, while that same degree probably doesn't mean too much if that person goes into banking. So there's a matching issue to consider.

Taking a skeptical view of education has its merits, but perhaps you're over-stating the value of jobs as training as well. There are many, many jobs that also fail to show that excellence leads to money or status or that low effort and poor work will get you fired. You are likely over-valuing the opportunity cost of education.

andy writes:

When I am asked which percentage of the 2 master studies are useful to me (didn't finish the first, finished the second - completely different subject), I would say that 20% is exagerating. If you ask what percentage raises my marginal productivity for the employer (I don't mean signalling), it would be even less. And even the signalling is not _as_ important in computer programming as one would expect in different areas.

I am in Czech republic and there are very funny things going on. The government decided to raise the quality of state employess; in many areas a college diploma is necessary. Guess what - you can found a private college pretty easily. It's a big question if the volume of new colleges isn't sparked by state employees willing to retain/have a job.

Some schools had corruption problems recently; the funniest thing is that it centers mostly on the Law schools. The biggest university Law department lived through big scandals several years in a row because of bribery in entrance exams (a friend told me that this in a way didn't end). A different college (again, Law....) gave diplomas to people who didn't study AT ALL.

One almost sees the answer to some of your question: thanks to government regulation you have to have a law diploma for certain jobs. The 'knowledge' itself, however, is not that important. Therefore you would see huge incentives to 'buy' the diploma. If I take the technical universities, it's a little bit different. Even if you have a diploma, you still have to be good, otherwise you would lose the job quickly. And even if you don't have a diploma, you may still get a good job if you are good enough.

And if we come to PhD studies, the case is completely different; it's not study per-se, but more a employment requirement. I would say most of the PhD people are going to have some kind state-funded job.

A small question lurks in my head: what percentage of the signaling is not signalling, but actually a 'marginal wage productivity' improvement because of government regulation?

Miguel Madeira writes:

The most powerful argument against signalling theory - the big difference of incomes for different degrees.

If the function of education was largely signalling, with will be largely indeferent for the an employer if you have a degree of Medicine or Philosophy (both signalling that you are inteligent, applied, etc), then people with a degree in Medicine or Philosophy should have a similar income.

Miguel Madeira writes:

Another point:

There are two types of degrees. In one type (History, Economics, Sociology, etc.) you "learn things", in other type (Engineer, Medicine, Accounting, etc.) you "learn do do things".

The signal effect will be higher for the first type of degrees; my point - being Bryan Caplan an economist (a degree of first type), could be possibel that, because that, he is overstimating the value of siganlling in general?

Hyena writes:

I don't think this clarifies the issue in a useful way. It simply lays out in bullet points everything which has been discussed by you and in other's responses to you. I don't think there was ever a point where anyone was unclear as to what was being argued.

At this point, it's more important that you demonstrate your position isn't predicated on bias, starting with evidentiary biases like availability.

Nathan Smith writes:

There may be a need for improvements in empirical research design in order to test how well people are matched with jobs that use their skills. I haven't read everything by a long shot, but I've read some of the literature and I've never found a paper that explores this well.

However, I think the big weakness in your argument at this point is theoretical. It's just not clear how, in theory, so much wasteful expenditure on education could be a sustainable market outcome. That people are devoting a lot of costly resources to education is a fact. Why? If education is neither an efficient means neither of forming nor or revealing endowments of human capital, why do people do it? If individuals do it because the individual, as opposed to social, returns to education are high, why do employers make them high by paying so much more to (at least many) college graduates? If it's just that state subsidies distort education and the market would achieve efficient outcomes if they were terminated, that's easy to understand, but that doesn't seem to be Bryan's position.

Maybe mass college education *isn't* a sustainable market equilibrium. Maybe it's a bubble and we're about to see it pop. But unless Bryan is willing to go on record as saying that's the case, I think a theoretical paper explaining how inefficiently high investment in education can be an equilibrium would be in order. That theory could then inform empirical research designs that to test it.

RPLong writes:
ability to conform to instructions

Buried at the very end of Peter's comment is the real reason for mandatory education. Everyone knows it, and it is becoming less and less of an "elephant in the room." Conforming to instructions is exactly what it's all about.

Adam Ozimek writes:


I think some more empirical foundations for these arguments would be the right next step. The debate thus far has mostly just been logic and theory.

A lot of the debate seems informed by anecdote, and while you point out rightly that "P.S. If anyone naturally overestimates the on-the-job usefulness of education, it's academics like us. After all, by definition a big part of our job is to teach students the material we learned when we were in their shoes." keep in mind that if anyone naturally underestimates the usefulness of education it's autodidacts like yourself.

tperich writes:

I'd expand on Greg's point a bit. I think the return on signaling vs. skills gained from course material has high variability across professions and schools.

I received an MBA from a "top ranked" school known for it's intense nature and aggressive workload. I would not be successful at my post degree job without the knowledge I gained from my education. All that knowledge could have been learned on the job, but instead of two years, it would have taken 10 years in several different roles, with lots of mistakes.

However, I'd acknowledge the more than equal signaling effect of my MBA to help me get the job. In my particular case, you can only get my job if you go to a "top ranked" MBA program. So, the signaling impact is quite powerful.

Sam writes:

Suppose a child is sent to an excellent private school due to their family's wealth. This school naturally filters into the Ivy League, and they signal their way to the top of society. If return on education is only 20%, this student may not be as excellent as the stamp on their diploma suggests.

Law firms, banks, and all other high-paying places to work are interested in value. They shell out hundreds of thousands of dollars to have a few 20 year olds work 80 hour weeks. Why, if their bottom line is money, would they ever hire so many Ivy League graduates, if signaling alone accounts for 80% of the return on education?

Nathan Smith writes:

Btw, the obvious may as well be stated: Bryan is free from *self-interest bias* in the present case, since if he succeeded in drastically shrinking the education sector, he'd likely put himself out of a job. In that respect, he has an advantage over his critics among academics.

Bob Knaus writes:

Peter's last question is the most important one - what are the alternatives to education which send the same signals?

I have an answer. Start your own IT business, run it for 13 years, when it finally fails single-hand your sailboat from Miami to San Diego. Then get a job with a management consulting firm.

It worked for me. But for most, a few years of college would have been easier.

OneEyedMan writes:

School is often deliberately made more fun rather than to maximize skill formation in the short term. By keeping young adults entertained they stick with their educations longer, leading to greater skill formation. This makes it a problem to say how much of the time is really being spent on learning.

Floccina writes:

We must balance the benefits of schooling against the cost of additional years of education to the productivity of the nation.

I am also interested in what could be done to improve lives if school was focused more on education rather than signaling. People often say that we need more college graduates but people seldom say what skills and knowledge that they see as lacking and what is the easiest way to get that to the people who need it most.

I think that a separation between education and testing would be positive. Schools work at cross goals and IMO testing (signaling) too often wins out. Rather than delivering the most valuable information to the most students in the most easily learnable form, in order to prove our high standards, we choose rigor that leaves many behind and even in those that learn it fails to drill the simple fundamentals deep enough.

I think that lately much education policy and hand ringing is driven by national pride and a sort of nation against nation competition.

floccina writes:

I saw this PBS debate. The big focus seemed to be how do we get more of our people to get college degrees than other countries.

For what purpose? If all you want is more degrees you can just make college easier cheaper and more enjoyable. But that would destroy the value of the signal. Notice there is no discussion of what skills and knowledge would help people most and what is the best way to get it to the people who need it most.

I think the myth busters TV show is educational. Many games computer and others exercise thinking. History channel teaches history. PBS teaches cooking and gardening but none of these produces a signal and so the are not discussed when in debates like above. I consider this proof that schooling is primarily signaling.

Brian Clendien writes:


See wall streets journal top 25 colleges which fortune 500 firms recruit from. There is only one Ivey league and only about a 4th are expensive schools. The real question is if I am as smart and motivated as an Ive leauge graduate, what is the PV of my income verses their when one takes into consideration the cost of college. Say the discount rate is the current U.S. 10 year bond rate.

I think if one removes the family connections, their is no diffrences between a degree from Havard and a degree from a state school like Florida/Michican/Texas A&M.

Paul writes:

the real problem is the distortion of the institution with the massive subsidies and union control. maybe its signaling (I think Brian's correct) maybe its not. but we really won't be sure until the institution is freed from its statist chains... ; )

Jon writes:


You estimate that 20% of the gross average return to education is due to learning the course material, 20% is due to changed character traits, and 80% represents signaling. Did you mean to say 60% for the last number?

Also, you used 'gross marginal return' in your answer to question 3 where I think you meant 'gross average return'.

Philo writes:

I think you were quite successful in signaling your interest in *finding the truth*.

A minor comment: You write: "Even kindergartens spend a shockingly small fraction of their time teaching reading, writing, math, and other skills that their students will eventually use on the job." Well, how much "reading, writing, [and] math" can five-year-olds learn? On the other hand, the social skills (not quite "character-shaping") acquired in kindergarten may be very valuable on the job.

RPLong writes:

I think the "social skills" arguments are untenable. They always assume that in absence of schooling there would be no alternative means to acquire social skills. This is fairly obviously a false assertion.

That's saying nothing of the value of relinquishing our parental responsibility of socializing our children and handing it over to the state to do it for us...

Maximum utility writes:

Do you think Industrial arts/vocational education may increase marginal productivity more than the liberal studies education that you are discussing here?

stuhlmann writes:

I think college, and especially graduate school, was a wonderful preparation for the working world, especially the current version of life as work. Basically almost all other aspects of life were allowed to atrophy, so I could devote myself to my studies. And the whole time I had the feeling that no amount of studying was ever enough - there was always more needing to be done. Yes, when I finished a paper or project, it was finished (though maybe I should proof read it again), but in order to complete that paper/project I had to let work in other classes slide for a while. Plus I worked part time to support myself. It was such a relief to graduate and go to work. A 60 or 80 hour week was nothing in comparison.

Comments for this entry have been closed
Return to top