signaling theory and credentialing theory in sociology
A loyal reader asks me to comment on a recent exchange between Econlog’s Bryan Caplan and economics professor and blogger Noah Smith. Specifically, Noah Smith attacks Bryan for his strong defense of the signaling model of education. The theory asserts that the main reason that education correlates with income is that is a signal of intelligence and work ethic, not learned skills. I.e., employers like college graduates because they are good workers, not because they have useful skills.
Smith calls signaling theory a “fad,” even though the main papers were written by Arrow and Spence decades ago!! He also offers arguments in favor of human capital theory, which deserve their own response and have been debated in the literature. For example, he offers the argument that education provides networks. On this blog, MIT’s Ezra Zuckerman has argued that the overall explanatory power of social networks is weak. UNC’s Ted Muow is also a bit skeptical about the value of networks in labor markets.
But I want to step back – what do sociologists think about human capital and signalling? Well, it’s safe to say that opposition to human capital is not a fad. A core theory in the sociology of education is Randall Collins’ credentialing theory. And it’s been around for decades. On this blog, we had a discussion of signaling and it was split – about half the readership (which is mainly soc and management PhD students and faculty) thought that the education/income correlation is due to signalling. Furthermore, sociologists such as Richard Arum and Josipa Roska have documented the lack of learning in college, which strongly supports signalling.
So it’s not a fad, Critiques of human capital are an important part of economics and sociology. The debate will continue.
50+ chapters of grad skool advice goodness: Grad Skool Rulz ($2!!!!)/From Black Power/Party in the Street!!
orgtheory.net 
Surely this depends on the type of college education? If X has a degree in computer science, I would argue that this is much more than a simple signal. In contrast, if one has a business degree and is about to become a salesperson or responsible for a few clients, it might be much more about signalling and the ability to learn. Again, if one has to get into banking and work the financial markets, one has learned techniques and an overall knowledge that will be indispensable, and expensive for the company to transfer – naturally depending on the size of the companies.
Of course – one might be in the “Fuck nuance” ASA section…
LikeLike
Carsten B
May 1, 2015 at 1:00 pm
@Carsteen B: I haven’t seen that ASA section on my annual membership renewal notices. Who’s the section chair?
LikeLike
Howard Aldrich
May 1, 2015 at 2:10 pm
The funny thing about this debate is that most sociologists of education who critique HCT have never actually read Schultz/Mincer/Becker. It is similar to those who ridicule functionalism, without having actually read Parsons/Davis & Moore/etc. In that sense, critiques of HCT within the soc of ed might not be a ‘fad’, but they do entail some degree of intellectual posturing, as in: ‘I hate HCT more than you, I am such a radical’.
LikeLike
RPM
May 1, 2015 at 2:14 pm
Why does it have to be either or? A signal derives its value from its rough correlation with underlying quality. This means, of course, that others can free ride on the signal. However, (from Spence) it is expected that a minimal level of competence is present to obtain the ability to possess the signal. It is also important to remember that the evidence is at the mean.
This means: if Tom attends a top school and ignores learning and just coasts he will be better off than Tom’ who acted similarly at a worse school. However, if Tom’ improves his skill, there is a point where he will outperform Tom.
Since a signal is such because there is a “bar” to obtain it, this implies that it has a value to those who obtain it. What this does not imply is that skill can be learned to overcome a lack of signal.
I have served as a reference (at a signal owning institution) and I always get the following question: “How would you categorize [student’s] ability to do “X”? This is evidence that firms buy into the signal, as I am sure my student was more likely to be chosen than her ability counterpart at another institution, however, without the human capital to perform they would be passed on.
In short, as search costs increase signals matter but when they decrease the emphasis shifts to skill/quality.
LikeLike
To Signal or Not to Signal
May 1, 2015 at 8:52 pm
Arum and Roksa also point out that there is a sizable group of students who get hired after a college education (signaling) in which they did not learn anything but then lose their jobs, while the grads who learned more are more likely to keep them…fwiw.
LikeLiked by 1 person
Mikaila
May 1, 2015 at 8:52 pm
@To Signal: Both signaling and human capital posit a correlation with quality, but the issue is why and how much can be assigned to skill learning and jumping through hoops. There are major policy implications for each answer. If it is mostly human capital, then spending on education is clearly good. If it is signaling, education might be wasteful.It matters.
@Mikaila: Great comment! A follow up question: is long term employment due to the actual skills learned, or are people who learn skills just better workers?
LikeLike
fabiorojas
May 1, 2015 at 8:57 pm
Well, then there is another question: does learning some skills make you better able to learn other skills in the future?
LikeLike
Mikaila
May 1, 2015 at 8:59 pm
On a slightly tangential note, here is some other evidence that hiring may be based more on credentialism, while compensation may be less so: https://www.insidehighered.com/quicktakes/2015/04/30/do-prestigious-law-degrees-really-matter
LikeLiked by 1 person
Mikaila
May 1, 2015 at 9:01 pm
@Mikalia: It’s tricky. Bryan appeals to the “transfer learning” lit in psychology that say unequivocally, no school does not make you a better learning. The way the test it is by seeing if by learning skill X, you can learn unrelated skill Y. Answer: no, no, no. But maybe people who learn in college sort into jobs where the learning is cumulative. An engineering student, might move into a science related field. So learning calculus would help. But transfer learning can’t work for most arts and science grads since most fields don’t draw on philosophy, history, etc.
LikeLike
fabiorojas
May 1, 2015 at 9:02 pm
@Carsten: Yes, it surely varies by degree type. Engingeering grads are way more likely to rely on human capital, while philosophy grads get the college premium exclusively through signalling. But most people aren’t engineers, they are in arts and sciences and professional tracks like business.
LikeLike
fabiorojas
May 1, 2015 at 9:05 pm
@Carsten: the relationship between the field of the degree and later occupation is, in the US, fairly modest, even among STEM majors. IIRC, only about half of engineering majors wind up in any STEM occupation, let alone in engineering occupations. (The Census Bureau has some interesting interactive graphs on this.) Point is merely that degrees may play a signalling function, too, even in the STEM fields.
LikeLike
krippendorf
May 1, 2015 at 11:29 pm
@Howard: Sorry, it is a session, not a section, according to Kieran’s Twitter feed
@Krippendorf. Of course it is also signalling, in all fields.
@Fabio: Arts teaches languages, and in Denmark / Europe a ton of positions strongly depend on being fluent in various languages. Again, learning will play a huge role, even in such a discipline. And even if it is mostly signalling – it might be 40% learning, and still very important for the future employers. I don’t know what data is already out there, but we need nuanced data from various disciplines within various institutional conditions, in order to answer these questions.
To lead the discussion in a related different direction: What I don’t understand is why everyone in Denmark needs 5 years of university education. I certainly see plenty of master students who don’t seem able to benefit enough (whatever that is) from the extra 2 years of education, on top of the bachelor. 3 years of learning is often adequate (whatever they learned) but right now one needs the extra 2 years of (mostly?) signalling.
LikeLike
Carsten B
May 2, 2015 at 8:48 am