Bryan Caplan  

The Roots of Signaling Denial

PRINT
Numeracy and Risk in Air Trave... What's the Correlation Between...
The signaling model of education fits first-hand experience.  It fits the psychology of learning.  It explains otherwise very puzzling facts like the sheepskin effect.  There are few theories in economics harder to doubt.  But many economists continue to do so.  Empirical labor economists seem to be the worst offenders.

What gives?  The charitable explanation, of course, is that the critics have fantastic counter-evidence.  I've read enough of the literature to say that this simply isn't so.  40% of the economists who criticize the signaling model conflate it with ability bias.  Another 40% of the critics attack the straw man view that education is nothing-but-signaling.  The remaining 20% dismiss the best evidence on implausible methodological grounds - or have standards of proof so high that no theory could possibly measure up.

If the charitable story is wrong, we have to turn to uncharitable stories.  What's the real reason why economists reject the signaling model of education?  The top candidates:

1. Status quo bias.  Critics, like everyone else, hate to change their minds.  They've disbelieved in signaling for ages, and they're not going to let anyone take their current opinion away from them.

2. Social pressure.  Economists feel strong social pressure to disbelieve signaling.  Even though many economists think signaling is empirically important, high-status economists don't.  And economists are very status conscious.

3. Provincialism.  Signaling theory comes from economics departments.  But most of the best empirics for signaling come from psychology, sociology, and education departments.  Economists don't read the research of these outsiders, and wouldn't take it seriously if they did.

4. Liberalism.  Arrow and Stiglitz - two of the seminal figures in signaling theory - have impeccable liberal credentials.  But economists still tend to see educational signaling as a free-market story.  After all, the signaling model suggests that massive government subsidies to the beloved education industry could actually be exacerbating the market failure of excessive education rather than correcting the market failure of insufficient education.

5. Education loving.  Professional economists were almost invariably good students - if not outright teacher's pets.  They enjoyed their educational experience - or at least enjoyed it a lot more than most students.  Education is a cherished pillar of their identity.  Economists treat the signaling model as an affront to the institution they love.

6. Panglossianism.  Many economists are Panglossian: Whatever exists, exists for a socially good reason.  If massively subsidizing education is a bad idea, why does every developed country on earth do so?  Even some market-leaning economists are strangely prone to this view.  Some actually treat the signaling model with the same hostility as any other market failure model - even though the signaling model strongly recommends lower government subsidies.

My personal judgment:

#1 isn't a big deal.  Economists embrace all sorts of weird views.

#2 is a big deal for practicing labor economists.  But most economists don't even realize that the consensus of labor economists largely dismisses signaling.

#3 is a big deal.  Economists are as guilty of the "Not Invented Here" syndrome as anyone.  Maybe more so due to our superiority complex vis-a-vis other social sciences.
 
#4 is tempting, but overrated.  If you actually talk to economists at the AEA, you'll learn that support for massive government education subsidies is bipartisan.  Mainstream Republican economists are more supportive of school choice, and less supportive of the "everyone should go to college" idea.  But only hardcore libertarian economists favor significant cuts in government support for education.

#5 is a big deal.  Regardless of their political views, economists were good students and love education.  Even I do, in some sense.  Converting such economists to signaling is like persuading them to renounce their religion.

#6 is a big deal for market-leaning economists, especially the Chicago-trained.  Liberal economists are never consistent Panglossians.  They constantly complain even when they're in charge!  But some market-leaning economists' fundamental belief isn't that markets are efficient, but that the world is efficient.  Donald Wittman's The Myth of Democratic Failure is a case in point.

Any explanations I'm missing?  For the sake of argument, take "signaling is false" off the table.



COMMENTS (30 to date)
ajb writes:

[Comment removed for rudeness. Email the webmaster@econlib.org to discuss your comment privileges. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

Clay writes:

7. Formal education is easily objectively quantifiable. The signaling model doesn't provide a better metric for ability and skill. Even Caplan demands people back their arguments with metrics. A flawed measure is preferred over mere opinions.

Philo writes:

#2: Why do high-status economists think signaling is unimportant?

Ben writes:

Clay writes "7. Formal education is easily objectively quantifiable. The signaling model doesn't provide a better metric for ability and skill. Even Caplan demands people back their arguments with metrics. A flawed measure is preferred over mere opinions."

This doesn't get to the lack of acceptance of the signaling model. In fact, if people used this argument, they would be admitting that the signaling model holds true, and their answer would essentially be "So, what should anyone do about it?" Bryan is saying that the signaling model is not even being accepted.

Also, I thought the signaling model does either implicitly suggest or explicitly provide alternatives, e.g., IQ testing. Why waste time and money on an "education" if an employer could identify the desired traits more easily (and be legally permitted to test IQ as part of hiring decisions)? I thought that the implications of the signaling model do make a case for alternatives.

Brian Shelley writes:

I presuppose latent religious elements, but here's my take.

#5. Converting such economists to signaling is like persuading them to renounce their religion.

Exactly.

Many secular people see education as the agent of redemption in society. A pseudo-Jesus. If, as the signaling model suggests, that much of this is a fraud it has enormous ramifications for the priestly class. It defrocks their sense of moral superiority by rendering their rites of passage through the education establishment to empty rituals. Their pseudo-Jesus is a fraud.

Roger Sweeny writes:

5a. Economists (at least the ones who publish articles on human capital, signalling, etc.) work in universities. Universities say they are about transmitting knowledge to the next generation. If the major function of universities is signalling, they--and their colleagues across the university--are taking money under false pretenses. These academics are living a lie. Nobody wants to believe that.

Moreover, most teachers want to believe that they are helping make the country more democratic, providing a way up the ladder of mobility, etc. If they are instead largely signalling already existing hierarchy, they may be part of the problem rather than part of the solution!

Bostonian writes:

PhD economists use what they learned in college and graduate school and may think others find their coursework similarly useful on the job.

If education is about signalling, so that having a B.S. from MIT signals a higher IQ than being a high school dropout, it suggests that races with a higher fraction of members with the aptitude to get a B.S. from MIT are smarter. Economists don't want to sound like Steve Sailer (who I admire).

Fazal Majid writes:

8) Rational self-interest: educational institutions are a major source of funding and employment for economists, and any decrease in government funding would adversely impact their income, even for those who don't work for education. This goes a long way to explain why support for the gravy train is bipartisan.

Glen Smith writes:

Business leaders have a strong desire to believe or at least promote the HC view (but that would make them frauds). Much of the welfare they receive from our government, the warm fuzzies they get and whining they are allowed is dependent on that belief. For instance, if HC is not true, they are not cost-sharing job training and if they don't believe it true, they're even worse than stupid, they are criminal. So, government, schools and business all have a rational self-interest to deny the signal model.

Finch writes:

While I wouldn't put it the same way, Bostonian has probably captured it.

People think "That way lies Steve Sailer."

I'd also give some credit to #5. If you are tenured faculty, education clearly worked for you.

Steve writes:

People deperately want to believe the human capital model because the alternative’s too depressing. The signaling model says people aren’t changed much by education, which means that we can’t use it to lift up underperforming individuals or groups.

The idea that not everyone can be a rocket scientist is a bitter pill to swallow, and cheery, optimistic people do what they can to avoid swallowing it.

Here’s Charles Murray in the tenth anniversary edition of The Bell Curve, commenting on the root of the anger his book produced:


“Watching the reaction to the book, theologian Michael Novak and economist Thomas Sowell have written in similar terms how the dominant intellectual stance toward public policy continues to invest everything in a few core beliefs about society as the cause of problems, government as the solution, and the manipulability of human beings in reaching the goal of equality. For persons who hold this view, Novak writes, The Bell Curve‘s message cannot be true, because much more is at stake than a particular set of arguments from psychological science. A this-worldly eschatological hope is at stake. The sin attributed to Hernstein and Murray is theological; they destroy hope.'"

Tom writes:

Here's a guess: Economist's own personal experience.

Some academic economists have worked in non-academic settings as economists (economic consultancy, econ departments of large firms, etc.), but almost all at least are aware that there are opportunities like that out there for them. If you're a high-end non-academic economist, you do need to know at least some good econometrics (although probably not much in the way of theory). So, from their own experience, economists probably have a higher prior that what they learned at university is useful on the job.

Steve writes:

On a somewhat unrelated note: when do you plan to release your book? We seem to be seeing more and more complaints about the current college model...might be better to strike while the iron's hot.

Glen S. McGhee writes:

Fazal's #8 is good -- the conflict of interest makes it impossible to embrace as morally corrupt a system that supports your life style.

And Steve is right too. James E. Block has written expensively on the development of the liberal self (i.e., autonomous self), which human capital theorists pander to. His latest book looks at education institutions.

Floccina writes:

#7 Normal human over optimism (maybe if you only asked depressed economists you would get a different result).

So it might be better to try to convince them that our current schools are 80% signalling and 20% education and so get them thinking optimistically about what can be done to get us to 80% education and 20% signaling.

One could be very optimistic about 80% signalling and 20% education, in that it leaves plenty of room for improvement. People do not want to believe that the bottom 40% of students are beyond improvement through education.

MingoV writes:
#5 is a big deal. ... economists were good students and love education.... Converting such economists to signaling is like persuading them to renounce their religion.
I agree that this is a big deal. I saw plenty of signaling while attending a technical institute in the 1970s. It was minimal but present in the College of Science and the College of Engineering. It was rampant in graphic arts (printing), photography, sociology, and criminal justice. It was least rampant in the College of Fine Arts, almost certainly because art portfolios were much more important than degrees.

Most students who were in the signaler group were unashamed of saying that they were in school for the degree, and they didn't care much about learning or grades (as long as they passed required courses). I was appalled at the waste (money and time), but the pure signalers were happy to party more than study.

John Fast writes:
#4 is tempting, but overrated. If you actually talk to economists at the AEA, you'll learn that support for massive government education subsidies is bipartisan. Mainstream Republican economists are more supportive of school choice, and less supportive of the "everyone should go to college" idea. But only hardcore libertarian economists favor significant cuts in government support for education.
If signaling theory is seen as supporting significant cuts in government support for education, then the fact that almost all economists favor "massive government education subsidies" means that almost all economists have a motive to deny the evidence for signaling theory.

Am I overlooking something here?

Tom West writes:

People deperately want to believe the human capital model because the alternative’s too depressing.

I have to second that. The super-simplified alternative to "education will save us" is "genetics are destiny", and let's face it, that's not a society many want to live in.

I would say that signaling is one, but not the only one, purpose of education. If it was the only purpose, then there would be no need for studying different disciplines.

Joel writes:

Here are two reasons why academics (not just economists) might be less inclined to believe in the signaling model:

(1) the academic hoop-jumping process virtually guarantees that all of your colleagues are smart and hard-working, which gives you no first-hand reason to doubt the non-signaling models. Whereas most non-academics have lots of experience working with people who are "highly educated" and yet still dumb and lazy. (One of the dumbest people I ever worked with was a UChicago MBA.)

(2) being an academic means that one of your job responsibilities is teaching students the things that you learned in school. In particular, this means that the things you learned in school are super-relevant to your job. Non-academics are much more likely to have the opposite experience.

Bill Kaplan writes:

It seems to me signaling theory works for less technical matters, but becomes attenuated when talking about highly technical matters. I presume that the professoriate least supportive of the theory are also the ones dealing with the least technical matters.

Glen S. McGhee writes:

Bill Kaplan, have you read Rivera?
w.asanet.org/journals/asr/dec12asrfeature.pdf

Eric Falkenstein writes:

I think you're too optimistic about the susceptibility of economists to change their beliefs. On what big issue has a dominant economic belief ever been changed, within 10-20 years, based on logic, theory and systematic data?

Models, theory, and data, are great ways to defend beliefs, but those big beliefs (ie, not some technical point) come first. Think about planting a seed for future debate, as opposed to changing minds. It's more realistic.

Tom West writes:

Alternatively, with respect to a question like this with strong social implications, the answer one gives may be a strong signal to others as to what sort of person you are.

Believing in educational signalling may well be a signal that economists would rather not broadcast, especially if you aren't that sort of person.

Brian writes:

Reasons 5 and 8 (& 5a) all play a role, the real reason is that, well..., signaling is not quite WRONG--it plays a role--but human capital is also a major contributor. You say you've looked at the evidence, but I've found very few of your arguments from the evidence persuasive. I'd say you prematurely dismiss counter-evidence because you're invested in the signaling model.

So what are some of the arguments against the signaling model?

--It assumes a massive and universal market failure. Why would employers wait for the college signal when the same information is largely available from high school and SAT-type tests? Why not just hire all the high-school seniors who get accepted to elite colleges? You get them for cheaper and you get more productive years out of them. Frankly, such a massive market failure is improbable.

--It fails to explain why companies that hire for technical backgrounds hire only those with specific technical degrees AND that those who hire for humanities-related jobs hire only those with humanities-related degrees.

--It fails to explain how the education enterprise in the younger grades is mostly about human capital (even YOU have admitted that it's important to read and do basic math) but mostly about signaling in high school and college. This position depends on believing that higher education becomes magically ineffective compared with lower grades. You might as well argue for the Easter Bunny.

--It fails to explain why education still provides a significant wage premium even after accounting for ability and degree effects.

Anon writes:

[Comment removed pending confirmation of email address. Email the webmaster@econlib.org to request restoring this comment. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

Miguel Madeira writes:

Bostonion:"If education is about signalling, so that having a B.S. from MIT signals a higher IQ than being a high school dropout"

Brian: "Why would employers wait for the college signal when the same information is largely available from high school and SAT-type tests?"

I think that you are forgetting that, in Caplanian version of signalling, what education signals is hard work and conformism, not intelligence.

Roger Sweeny writes:

In the Caplanian version of signalling, education signals hard work and conformism AND intelligence. There is a Holy Trinity.

Bill Kaplan writes:

Glen,

Thank you. When I'm done with my elite job for the week I will make sure to read it on my less-than-elite weekend.

Brian writes:

"I think that you are forgetting that, in Caplanian version of signalling, what education signals is hard work and conformism, not intelligence."

Miguel,

I am well aware that signaling includes hard work, conformism, and intelligence. But how does college signal any of those things better than high school? After all, any student who gets into an elite college is doing precisely that--using high school to signal those characteristics to the college. Why wouldn't employers short-circuit the college process by using college acceptance information to get an early jump on the best candidates? The employers would get well-signalled employees for less money (because they didn't go to college first) and potentially get more of their productive years. The employees would benefit by saving the cost of college and making a good income right out of high school.

One problem (of many) with the signaling model is that college is far too expensive to be used for signaling alone. There are much less expensive ways to get the same signal. Nor is this signal cost-free for employers. Besides the opportunity costs mentioned above, employers have to pay a college wage premium in part to cover the cost to the student. This is usually indirect through simple market forces, but sometimes becomes direct, such as when employers cover the cost of college for their employees. Employer-based tuition benefits are impossible to explain in the signaling model, since such companies would be placed at a huge disadvantage if the classes provide no productivity gain.

Comments for this entry have been closed
Return to top