« "It's not the economy, stupid" | Main | Is religion for controlling men? »
Education isn't mainly about signalling
We find that employer learning about productivity occurs fairly quickly after labor market entry, implying that the signaling effects of schooling are small.
Here is much more. And here is more yet; this second paper estimates the speed of employer learning and uses that estimate to bound the value of the signal at no more than 28 percent of the value of education. I consider this devastating to the signaling hypothesis. How can ?? years of schooling be needed to signal your quality, if your employer often knows your quality within months?
In my view education is mainly about indoctrination to give you more productive habits. So yes it is learning, but not in the way they might have told you, and that is why it so often does not feel like learning.
Posted by Tyler Cowen on March 26, 2008 at 07:33 AM in Education | Permalink
Comments
Could it simply be that what we mistook for signalling was purely self selection, and that schooling has practically no benefit at all?
Posted by: josh at Mar 26, 2008 7:47:31 AM
It's signaling to potential mates, not to employers. A typical Ivy League liberal arts education (added to one's existing skills) is as beautiful and useful as a peacock's feathers.
Posted by: DK at Mar 26, 2008 7:52:59 AM
If we'd let the Boskops live, maybe we coulda asked them.
http://discovermagazine.com/2008/mar/21-the-extinct-human-species-that-was-smarter-than-us?utm_campaign=DISCOVER%20Magazine%20Human%20Origins%20Newsletter%203%2E24%2E2008&utm_content=tygerx9@gmail.com&utm_medium=Email&utm_source=VerticalResponse&utm_term=The%20Extinct%20Human%20Species%20That%20Was%20Smarter%20Than%20Us
Posted by: burger flipper at Mar 26, 2008 8:12:37 AM
Without having yet read the papers, I would point out that it also still may serve as an ex ante signal. While employers may be able to quickly discern quality once an individual is hired, if the costs of firing an employee is sufficiently high it may still be a valuable signal to employers.
Posted by: Tucker at Mar 26, 2008 8:26:06 AM
Education signals conscientiousness, not ability. Throwing out the resumes of non-degreed applicants allows employers to choose from a pool of people who meet some minimum level of conscientiousness. Then they can "learn" who is productive and who isn't. Also, saying signaling is "only" 28% of education's value is a bit like saying aerodynamic considerations are "only" 28% (or whatever) determinant of a car's fuel efficiency.
Posted by: Steve Miller at Mar 26, 2008 8:30:14 AM
The findings you report ring true to me.
But signaling is not only a matter of signaling some narrowly defined workplace productivity.
There is also a kind of signaling of cultural attitudes--and I don't mean good work habits. I mean social and political outlooks, esp. establishment attitudes. This matters to a heck of a lot of career paths, especially highly paid ones. Having been through college and especially grad school generally signals that one knows to bow establishment presuppositions, superstitions, and taboos. Part of this is learning how to bow, but I think part is also signaling.
Posted by: Daniel Klein at Mar 26, 2008 9:16:05 AM
>"In my view education is mainly about indoctrination to give you more productive habits."
Is there really no way for other way employers to test whether you have productive habits? Four years of school seems like a very expensive way to establish that you have good habits.
And so what is the purpose of a graduate degree? Are your habits so much better after seven years of school instead of four?
Posted by: Phil at Mar 26, 2008 9:16:24 AM
How do these studies account for education in the sciences, engineering, and mathematics, or do they even bother? I think it's pretty clear that 4 years of English or History education is overkill for creating productive managers for the local mall stores, but how do you suggest we train the people that design computers and cars and whatnot?
Posted by: billb at Mar 26, 2008 9:26:42 AM
I think Dr. Cowen misses that the cost of hiring an employee and firing them 3 months later far exceeds the salary of that employee. Selecting the individual takes a lot of time and money from the organization. Additionally, there may be severance pay, CORBA, and all sorts of other stuff to consider. Finally, there is also the possibility of opening oneself to lawsuits.
Oh, and there is training. Training a new employee takes more money and time away from existing workers.
Altogether, if education is a worthwhile signal, you should use it--even if you will be able to verify the employees usefulness in a few months, it will cost you a lot of time and effort to get to that point. Having the information now instead of 3 months from now is of great valuable.
Posted by: bobvis at Mar 26, 2008 9:42:24 AM
I consider this devastating to the signaling hypothesis. How can ?? years of schooling be needed to signal your quality, if your employer often knows your quality within months?
I'm with the other commenters who have questioned this conclusion. Let's say I have a retail business. I might be able to tell within 3 months whether a particular employee does a good job and shows potential. But I don't want to hire people at random and see if they work out; that wastes my time and energy. So suppose it turns out that if I hire college graduates (without regard to major), 50% of them turn out to be hard workers; but if I hire high school graduates, only 20% of them turn out to be hard workers. So I might look for a college degree as a signal that the person is a hard worker, at least compared to other potential hires. After the hiring, I'll find out for sure in short order, but I still need a screening device up front so that I don't waste even more time hiring the wrong people.
So that's all obviously made up, but what's implausible about the notion that education could often serve as a signaling device even while employers can assess quality on the job as well?
Posted by: Stuart Buck at Mar 26, 2008 10:18:30 AM
As rent-seekers, those who confer diplomas desire require payments that seem to them sufficient in return for the signal they permit the prospective graduate to display. That's why they require "??" years.
Think of it this way -- if you were given the opportunity to hire 2 recent high school graduates as summer employees -- one going to Harvard and one going to, say (I was going to say "GMU" but never mind) Northern Va Community College, you might favor the one going to Harvard. Signal given, and received.
Posted by: hwinva at Mar 26, 2008 10:34:01 AM
If your employer or mate privately learn about your real productivity, that doesn't obviously or quickly translate into higher wages or a better mate. If they see you are better than expected, that might embolden you to quit and look for a new deal, but high switching costs put a wedge between productivity and both wages and mating value.
Posted by: Robin Hanson at Mar 26, 2008 10:38:06 AM
I'm not sure why:
1) You feel your identity is "established" by your mid 20's - lots of people go back to school far later in life and I doubt it is because they are mistaken as to the value of schooling;
2) neither model appears to account for the joy of schooling, I don't think you can explain liberal arts or university without taking into account people enjoy it - sitting around discussing philosophy is enjoyable itself, and people spend an awful lot on enjoyable hobbies.
Posted by: yakov at Mar 26, 2008 11:20:46 AM
I'm a PhD student at Michigan doing some research on this subject myself, at the moment, except I'm looking at signaling and learning in the context of college quality (read, selectivity) rather than quantity (years). I'm finding stronger, though conditional, support for the signaling hypothesis at jobs early in the career, and I suspect for many of the reasons put forth in previous comments (high hiring/firing costs). The employer learning model is more interesting in that the effects of having attended a more selective school do not fade with time, even when I include AFQT a la Lange or Altonji and Pierret. While the evidence is mixed, I think reports that the JSH is dead appear to be greatly exaggerated.
Posted by: Brad Hershbein at Mar 26, 2008 11:53:06 AM
What self-respecting professional would take a job at a company that might fire her after three months?
Posted by: Kevin at Mar 26, 2008 11:57:42 AM
If you disbelieve the signalling model of education, I would love to see a post explaining where you think sheepskin effects come from.
Posted by: Jebs at Mar 26, 2008 12:37:09 PM
See the paper (link below) by ARCIDIACONO et al. It suggests higher ed. leads to direct revelation of ability.
There is a large return to human capital investment, so education can add significantly to human capital & still be (in some cases) a signal.
http://www.nber.org/~confer/2007/hiedf07/bayer.pdf
Posted by: Tim perri at Mar 26, 2008 12:54:18 PM
I think the point is that the entire labor market learns about your productivity after your entry into the labor market. IE, it's not just YOUR employer that figures out how productive you are, but everyone else, too. Hence, the signaling effect shrinks with time.
It certainly jives with the evidence that students who get accepted into Ivy League schools, but don't go, usually end up with the same wages that Ivy Leaguers receive eventually (though, right out of college, Ivy Leaguers earn more).
Posted by: Robert Olson at Mar 26, 2008 1:27:24 PM
A lot of good points made here. I would add:
1. It entirely depends on the job. Some jobs require skills which can be shown easily before hire and having that skill also demonstrates study skills (required to learn the skill), etc. Software development is an obvious example. What a college degree in those cases signals is: (1) Willingness to endure 4 years of, probably low-grade, education in the field - submissiveness, (2) Some higher probability of a rounder education than the self-taught.
2. I don't buy the argument that non-professional school doesn't teach you anything worthwhile for a job. Learning about medieval literature, as a pre-requisite for a bank manager job, or something. I think its only a signal, and I would certainly think it stupid if it is used as a requirement, rather than one of many potential signals, but I do think that the education helps to hone an ability to reason, to think in new ways and understand arguments made by others, to think abstractly, and so forth.
I know some claim to have debunked this, but not to my satisfaction. And, having been in school and out of school, on and off for many years, I believe my own experience and not the weak papers I have read on the subject. Also, somebody showed that the longer you stay in school the more abstractly you think, which supports my hypothesis.
Posted by: liberty at Mar 26, 2008 1:31:00 PM
billb: "How do these studies account for education in the sciences, engineering, and mathematics, or do they even bother?"
Really. How can anyone lump liberal arts education with education in sciences and then proclaim that "it is not learning in the sense they may have told you"?
Here's some starting salaries for various college majors:
chemical engineer ...$64K
electrical engineer .$57K
computer science ....$53K
accountant ..........$47K
financial analyst ...$47K
nurses (4 year) .....$44K
statistician ........$43K
Does anyone really believe that graduates with these degrees learned nothing more than "indoctrination in productive habits" in four years of college? Of course not. The knowledge they acquired is why graduates with these and other "hard" degress earn $15K to $25K more than journalism and history majors.
One should not become a surgeon without understanding anatomy. One should not become a mechanical engineer without understanding thermodynamics and fluid mechanics. One should not become an accountant without understanding Sarbanes-Oxley and derivatives accounting.
Why shouldn't employers expect the professionals they hire to be trained well in such concepts? If not in a guided program of study, where would such knowledge be acquired?
Posted by: John Dewey at Mar 26, 2008 4:50:19 PM
Caplan responds: http://econlog.econlib.org/archives/2008/03/tyler_on_signal.html
Posted by: TGGP at Mar 26, 2008 5:58:08 PM
Now I'm no economist but in machine learning this would be called a cascade. The basic idea is that you use some easily obtainable attribute to screen out obvious failures and this makes learning the true function from the remaining data easier.
The flaw in your argument, Tyler, is that you're assuming that it will take employer the same amount of time (~ 3 months) to ascertain an employee's quality if the employer does not use the education signal.
Posted by: Matthew Dutton at Mar 26, 2008 11:44:28 PM
If it is about quality why not just use potential employee's SAT scores? Is a university education really much better at predicting outcomes than the SAT alone? I'd imagine it evens out given variations between universities.
Posted by: yakov at Mar 27, 2008 12:06:01 AM
Besides the economic costs of firing an employee after just a short time, the hiring manager also experiences a loss of face. It's an open admission of a mistake in the hiring process. So there's a social cost to such an action.
Posted by: jult52 at Mar 27, 2008 7:58:31 AM
A very unique file hosting website, that allows you to upload/share and search textual files, such as MS word, PDF, text, MS Excel etc. very helpful resouce for students and researchers, thousands of available documents...
http://www.maifile.com
Posted by: aamir at Mar 27, 2008 1:01:40 PM
