Bryan Caplan  

David Autor Signals Wit and Insight

I Believe These Are Related... Tyler's Telling Question About...
I don't normally associate MIT with "funny," but David Autor's notes on signaling definitely qualify:
Testing signaling versus human capital models of education

Does it seem plausible that education serves (in whole or part) as a signal of ability rather than simply a means to enhance productivity?

• You obviously learn some valuable skills in school (engineering, computer science, signaling models).

• Many MIT students will be hired by consulting firms that have no use for any of these skills. Why do these consulting firms recruit at MIT, not at Hampshire College, which produces many students with no engineering or computer science skills (let alone, knowledge of signaling models)?

• Why did you choose MIT over your state university that probably costs one-third as much? Is this all due to educational quality, or is some of it credentialism?

Harder question: How do you go about empirically distinguishing the human capital from the signaling model?

1. Measure whether more educated people are more productive? (Would be true for either model.)

2. Measure people's productivity before and after they receive education- see if it improves. (Conceptually okay, very difficult to do.)

3. Test whether higher ability people go to school? (Could be true in either case- certainly true in the signaling case.)

4. Find people of identical ability and randomly assign some of them to go to college. Check if the college educated ones earn more? (Both models say they would.)

5. Find people of identical ability and randomly assign them a diploma. See if the ones with diplomas earn more. (A pure test of signaling.)
Funny - and insightful too.

COMMENTS (6 to date)
Ziqiang He writes:

Personally speaking, I agree that education is a significant reflection on one's productivity to a certain extent.
The reason why some firms hire MIT students may be not the students have competitive skills but they are able to get into MIT. If a student is qualified to get into MIT, do you doubt his/her quality? I would not anyway. He/she must be intelligent or diligent, because MIT is one of the top universities in the U.S.. As long as they get well trained, they will do well in many things.

ajb writes:

The real question is why there isn't a greater premium for MIT than the Ivies. MIT is harder to graduate from (on average probably, at minimum certainly) than HYP or Stanford. I would guess, minimum standards for admission are also a bit higher (given that everyone needs to pass some math and physics) so why doesn't the market reward that?

Finally, why should an MIT student who doesn't graduate after 3 years or who gets a sub B average get punished relative to A students at very weak, very grade inflated non-Ivy schools? [Consider that the former has virtually no chance of admission to med school nor can he get interviews for many jobs where the A student at the worst podunk U still has a shot.]

What signalling model predicts these facts?

RPLong writes:

I would actually like to see study #5 conducted.

Kevin writes:


5. Find people of identical ability and randomly assign them a diploma. See if the ones with diplomas earn more. (A pure test of signaling.)


I would actually like to see study #5 conducted.

As would I.

Actually, there are two simple ways to test this, the second being the much easier:

1) Recruit some undergraduates (class credit or money rewards) and randomly assign them a level of education (to keep it simple: High School Dropout, High School Graduate, Some Post-Secondary, Post-Secondary Graduate) and measure their "ability" (and personality and whatever else). Send them out to apply for any and all positions they can find and measure the number of call-backs/interviews (even the number of positions they actually get - or is that taking things too far?) and even the hourly wage and/or yearly salary they would earn for the jobs they got call-backs/interviews for.

In the analyses, look at the differences in callbacks/interviews and wages. Signaling theory says that the between-subjects effects of (fake) education should be doing the heavy lifting.

2) The very simple version. Make up some fake resumes that vary only on the level of education. Shop them around to any and all positions available and collect results as above. Same predictions.

Dan Carroll writes:

On #5: yes, funny, but no, not easily testable. The test is whether the diploma is a signal of completing an education or a signal of ability that is not relevant to the education received. If one assumes the human capital model, then employers will use the diploma as a signal of the underlying human capital acquired while in school. If one assumes the signaling model, then the diploma is a signal of the individuals innate ability. While employers often do some verification, it is not practical to verify that the individual actually 'earned' the degree.

I believe most employers have consciously bought into the human capital model, but unconsciously adopt the signaling model when hiring.

Kevin L writes:

The problem with #5 is that a diploma is not the only signal of having been educated. I doubt that someone, even of decent ability, with a fake diploma would pass the interview process - unless such a person was already a con artist. They'd be even less likely to make it through the workplace for a month without their educational veracity coming into question.

Since I've graduated from engineering school, I've noticed that most employers in my line of work don't just put education requirements, but also experience requirements on their job postings; so what about job experience signaling? Is an employer looking at job longevity, stability, promotions, and responsibility looking for signals of innate ability or human capital gained in the crucible of real-world experience?

Comments for this entry have been closed
Return to top