Bryan Caplan  

School Is to Submit: My Critique

Don't Let the IRS Ruin Your Da... David Beckworth has a new podc...
Robin Hanson and I have long agreed that education is mostly signaling.  But after reading The Case Against Education, my forthcoming book on educational signaling, Robin proposes a bold new theory of what's really going on.  At least on my reading, Robin largely abandons signaling in favor of an enriched version human capital theory.  Indeed, Robin's new story is very much in the spirit of Tyler Cowen's beasts-into-men version of the human capital model, no doubt explaining Tyler's enthusiasm for Robin's innovation.

As you'd expect, I disagree.  While Robin makes some keen observations, he's fundamentally wrong.  Point-by-point, with Robin in blockquotes.

In his upcoming book, The Case Against Education, my colleague Bryan Caplan argues that school today, especially at the upper levels, functions mostly to help students signal intelligence, conscientiousness, and conformity to modern workplace practices. He says we'd be better off if kids did this via early jobs, but sees us as having fallen into an unfortunate equilibrium wherein individuals who try that seem non-conformist. I agree with Bryan that, compared with the theory that older students mostly go to school to learn useful skills, signaling better explains the low usefulness of school subjects, low transfer to other tasks, low retention of what is taught, low interest in learning relative to credentials, big last-year-of-school gains, and student preferences for cancelled classes.

So far, so good.

My main problem with Caplan's story so far (he still has time to change his book) is the fact that centuries ago most young people did signal their abilities via jobs, and the school signaling system has slowly displaced that job signaling system. Pressures to conform to existing practices can't explain this displacement of a previous practice by a new practice. So why did signaling via school did win out over signaling via early jobs?

While I don't dwell on history, my book does answer the question, "Why does schooling pass the market test?"  My answer is: "Market test?!  Government showers almost a trillion dollars a year on the status quo, and you call that 'passing the market test'?!"  But why do governments so favor conventional education?  My answer: The politics of Social Desirability Bias.  "Making sure every child has the best possible education" sounds wonderful despite its absurdity.  When individuals spend their own money, of course, they at least ponder whether what sounds wonderful is really worth the cost.  For collective spending, in contrast, Social Desirability Bias reigns supreme.

Like early jobs, school can have people practice habits that will be useful in jobs, such as showing up on time, doing what you are told even when that is different from what you did before, figuring out ambiguous instructions, and accepting being frequently and publicly ranked relative to similar people.

One factual disagreement: Robin repeatedly claims that schools "publicly" rank students.  But at least in the modern U.S., schools strive to keep students' performance private to protect students' feelings.  Robin can insist that performance isn't "really private," but the fact remains that drawing attention to a student's failure is a serious offense.  Try emailing your class the list of everyone's grades, and you'll see what I mean.

Otherwise, though, I agree with Robin's description.  My book freely admits that school - especially K-12 - provides some socialization benefit.  But I still say: Compared to what?  School provides useful socialization relative to staying home playing videogames.  But not relative to actually doing a job.

But while early jobs threaten to trip the triggers than make most animals run from domination, schools try to frame a similar habit practice in more acceptable terms, as more like copying prestigious people.

There is quite a bit of research on the returns to vocational education and early job experience.  Both are at least as effective at raising future income as traditional education, at least after adjusting for students' pre-existing academic strength.  Robin has my references.


Schools work best when they set up an apparently similar process wherein students practice modern workplaces habits. Start with prestigious teachers, like the researchers who also teach at leading universities. Have students take several classes at at a time, so they have no single "boss" who personally benefits from their following his or her orders. Make class attendance optional, and let students pick their classes. Have teachers continually give students complex assignments with new ambiguous instructions, using the excuse of helping students to learn new things. Have lots of students per teacher, to lower costs, to create excuses for having students arrive and turn in assignments on time, and to create social proof that other students accept all of this. Frequently and publicly rank student performance, using the excuse of helping students to learn and decide which classes and jobs to take later. And continue the whole process well into adulthood, so that these habits become deeply ingrained.

This sounds vaguely plausible if you're talking about college.  But it's an outlandish summary of K-12 education.  And if you're trying to understand the historical evolution of  education, K-12 became widespread first, and remains the dominant form of education to this day.

Why "outlandish"?  For starters:

1. In elementary school, students do have a single boss - one teacher they spend almost all their time with.

2. Attendance isn't optional for K-12, and the students have almost no class choices until high school.

3. K-12 teachers give endless busywork: simple assignments with clear-cut instructions, not complex assignments with ambiguous instructions.

4. To repeat, performance rankings are not public.  While it's impossible to preserve total secrecy, schools sincerely try to protect weaker students' privacy to spare their feelings.

When students finally switch from school to work, most will find work to be similar enough to transition smoothly. This is especially true for desk professional jobs, and when bosses avoid giving direct explicit orders. Yes, workers now have one main boss, and can't as often pick new classes/jobs. But they won't be publicly ranked and corrected nearly as often as in school...

The opposite is true.  At least in the U.S., schools preach and practice an egalitarian ethos.  No student is entitled to boss other students around merely because he's a "better student."  That's why their group projects are hell on earth.  Employers are much more comfortable publicly ranking workers, and encourage the more-skilled to correct the less-skilled.


In sum, while students today may mostly use schools to signal smarts, drive, and conformity, we need something else to explain how school displaced early work in this signaling role. One plausible story is that schools habituate students in modern workplace habits while on the surface looking more like prestigious forager teachers than like the dominating bosses that all animals are primed to resist. But this hardly implies that everything today that calls itself a school is equally effective at producing this benefit.

Let me propose a variant on Robin's story.  Namely: While school is not and never was a good way to acclimate kids to the world of work, it does wrap itself in high-minded rhetoric or "prestige."  "Teaching every child to reach his full potential" sounds far nobler than "Training every child for his probable future."  As a result, making the political case for ample education funding is child's play.  Education's prestigious image in turn cements its focal status role, making academic achievement our society's central signal of conformity

But while the problem is vast, it is not deep.  Slash public funding for education, and standard estimates of demand elasticity say that kids will spend many fewer years in school - and many more years learning practical skills on the job.

P.S. Aside: In his build-up, Robin also remarks:
When firms and managers from rich places try to transplant rich practices to poor places, giving poor place workers exactly the same equipment, materials, procedures, etc., one of the main things that goes wrong is that poor place workers just refuse to do what they are told. They won't show up for work reliably on time, have many problematic superstitions, hate direct orders, won't accept tasks and roles that that deviate from their non-work relative status with co-workers, and won't accept being told to do tasks differently than they had done them before, especially when new ways seem harder.
According to the best empirical evidence, multinational corporations are well-managed around the world.  The main problem of development isn't that people in poor places won't individually submit to foreign direction, but that people in poor places won't collectively submit to foreign direction.  "Letting foreigners run our economy" sounds bad, but individuals are happy to swallow their pride for higher wages.  Voters and politicians in LDCs, in contrast, loathe to put a price on pride - and therefore hamstring multinationals in a hundred different destructive ways.

COMMENTS (15 to date)
J. Joseph Porter writes:

To fill out your explanation of the shift from apprenticeship-as-signaling to school-as-signaling:

Education reformers like Horace Mann ( believed that schooling was the way to instill civic virtues, *not* just to prepare future workers. That's why they were so passionate about expanding public education.

Hanson is right, then, to point out that the shift from apprenticeships to schooling is puzzling at first glance. But employers certainly weren't the ones who made the push for this shift! Instead, it was education reformers and politicians who brought it about.

Tom Davies writes:

Typo: I think LCD should be LDC.

Glen Whitman writes:

With regard to how education displaced early work experience, I read a long time ago (sorry, can't give you a source) that adult workers lobbied for compulsory education to reduce competition from younger workers. So that would be a public choice explanation for the death of the old equilibrium.

BJ Terry writes:

Your post doesn't contain a link to Mr. Hanson's post, unless I've missed it. It can be found here for other readers.

Jon Murphy writes:

I don't know if you watch Rick and Morty, but your title of this post reminded me of this scene from Season 1.

Glen Smith writes:

School,given signal theory is true and the signal performs itsfunction, tends to drive the cost of work away from the employer towards the employee.

Bryan Caplan writes:

Thanks for the corrections, readers!

Matt Skene writes:

It's not just that colleges don't publicly rank students, it's actually illegal to. Technically, it's illegal to tell a student "Good job" while handing back their test, or email a student with her own grade to a non-personal email address. FERPA rules are crazy restrictive.

Random writes:

I have done group projects where better student helped me. Was no problem. It was course requirement or anything. Groups were recommended by course but how groups worked inside was up to them. Ours worked just fine. We were all motivated though.

Joe writes:

I've gatta stick up for video games here.

The right kind of video game totally provides socialization, TF2, Overwatch, CS global offensive, Starcraft in team games, cooperation is the key.

Moreover, you get to be socialized to ppl of basically every income and of varied age and sex.

You can also make millions if you become very good.

Many video games are at least as good at socialization as sports/school

Brian writes:


I think it's easy to show that signaling does not account for most of the value of achool beyond ability bias.

Signaling is basically a reputational effect. It's not simply, for example, that college signals for intelligence, conscientiousness, and conformity, but that prestigious colleges signal those things even more than non-prestigious ones. So the signaling theory would predict that colleges with better reputations would lead to better outcomes for their graduates. And colleges with the best reputations, like the Ivys, would have the best outcomes, even after adjusting for ability bias.

Yet studies show this doesn't happen. Have you dealt with the results of Dale and Kruger in your book? How about the recent Economist calculation of economic value added (adjusted for ability)? Here's a comparison of US News reputation rankings versus the Economist rankings for the Ivy League schools.

School (rank)_____Reputation Score___Value-Added %-ile

Princeton (1)__________4.9/5.0__________39th

Harvard (2)____________5.0______________99th

Yale (3)_______________4.9______________0th

Columbia (4)___________4.9______________19th

Penn (9)_______________4.8______________98th

Dartmouth (12)_________4.8______________65th

Brown (14)_____________4.8______________11th

Cornell (15)___________4.8______________84th

As is easily seen, there is NO correlation between reputation and economic value added. Ivy League colleges span the entire range of value added relative to student ability. (And yes, Yale ranks almost dead last in economic value added.) This is impossible under a signaling model.

As a side note, schools of certain bloggers fair very well under the Economist's methodology, with Bentley at 99th percentile (#5) and GMU at 96th percentile (#40), despite having so-so reputational scores. It's odd that a GMU faculty member would be making this argument when his own college seems to refute it.

Miguel Madeira writes:

«One factual disagreement: Robin repeatedly claims that schools "publicly" rank students. But at least in the modern U.S., schools strive to keep students' performance private to protect students' feelings. »

I suspect that this is an US excentricity - at least in Portugal, the grades of each class are afixated in a paper in the wall; every student (and their relatives) in the school can know the grades of any other student

Brian Holtz writes:

Dale and Krueger reportedly found that

people who attended elite colleges do not make more money than do workers who were accepted to the same institutions but chose less selective ones instead

The signaling model doesn't predict that it's impossible to signal high ability without a certificate from a selective college. Nor does the signaling model predict that selectivity will correlate with value-add. Indeed, if value-add correlated with prestige, that would be strong evidence against the signaling model. I.e. if Harvard could consistently mold random/average college students into Harvard-level achievers, then the signaling model would be disproven.

P.S. I wonder if the Dale/Krueger study controlled for these students bragging about having rejected an acceptance at a selective college.

Peter D Jones writes:

"My book freely admits that school - especially K-12 - provides some socialization benefit.  But I still say: Compared to what?  School provides useful socialization relative to staying home playing videogames.  But not relative to actually doing a job"

What kind of job can an eleven year old, without full literacy or numeracy, do in a modern economy?

weareastrangemonkey writes:

Cowen's Second Law: There is a literature on everything.

In this case it belongs to a long Marxist tradition. I think this view on education goes back to Marx himself, but a more modern variant can be found in Bowles' and Gintis 1976 Schooling in Capitalist America. Bowles and Gintis are also noted in the work on "non-cognitive" skills by Heckman.

Or you can hear Chomsky give a very similar description to Hanson's back in 1989:

Comments for this entry have been closed
Return to top