Bryan Caplan  

Signaling Rules: Today Hollywood, Tomorrow the World

Alfred and Mary Paley Marshall... The Air Travel Revolution...
My dear friend and colleague Tyler Cowen thinks the signaling model of education is, roughly speaking, empirically irrelevant.  He's repeatedly urged me to stop barking up what he sees as a very wrong tree.  I was pleasantly surprised, then, to see that when he analyzes the role of nepotism in Hollywood, he takes my "diamonds in the rough" problem seriously:
Imagine a talent selection system with many different levels of filters and many, many applicants and also few winners.  The first level could be something as simple as "does anyone even look at your photo shoot or ask you for an audition?"  Let's also say that nepotism gets you past the first filter, or maybe a bit more, but not past the final filters.  They won't let you star in a movie just because you're Goldie Hawn's daughter (by that time most of her clout is gone).  Nonetheless relatives of famous actors, actresses, etc. still will end up considerably overrepresented on the screen.

There is also someone known who can vouch for you, albeit not always with perfect credibility: "Believe me, if you give my brother this role, he won't ruin the movie promo efforts with a cocaine addiction."  And so on.

You will be remembered more easily: imagine a director saying "hey for this bit part, why don't we get what's-his-name, you know the brother of [xxxx]."  It is then easier to work your way up.

All perceptive and plausible, Tyler.  But if you admit that true worker productivity is very costly to observe in Hollywood, if you see a long list of fallible filters in play, why are you so closed to the possibility that the same kinds of problems and the same kinds of imperfect solutions permeate the rest of the labor market?  Finding diamonds in the rough may be especially difficult in Hollywood, but don't good job openings throughout the economy routinely attract hundreds of resumes? 

The parallel is clear: If you apply for a good role in Hollywood, your head shot ends up in the trash unless you know a Hollywood insider; if you apply for a good job outside of Hollywood, your resume ends up in the trash unless you have a college degree.

COMMENTS (9 to date)
Tyler Cowen writes:

I don't get your characterization of my view. I've noted to you many times that I think signaling might be as much as one-third of the value of a college degree. And that for many MBAs it is probably much higher yet.

BC writes:

If the value of a college education was purely in its signalling value, then why do chemical companies tend to hire chemical engineering (ChemE) graduates and electronics companies electrical engineering (EE) graduates? Under the signalling model, if ChemE grads were considered smarter, harder working, and more conformist than EE grads, then electronics firms should prefer to hire them over EE grads. Conversely, if EE grads were considered smarter, harder working, and more conformist, then chemical companies should prefer to hire them. Yet, we see a universal preference in the market for neither ChemE nor EE grads. Chemical companies prefer ChemE grads, while electronics companies prefer EE grads, indicating that the content of their respective educations, not just signalling value, is important to their employers.

Mr. Econotarian writes:

Family of actors/directors/producers get through the same way the relatives of nobility do: there is some human desire to give power & popularity to a family line. This may be due to an innate assumption that much behavior is inherited.

Jesse writes:

I don't think anyone is arguing that there is no skills development value in specialized technical degrees. Do media firms strongly prefer media studies degrees?

Mark Little writes:

But Hollywood does not have barriers to entry through credentialing. (You don't need a degree to get an acting job.) The parallel doesn't hold.

The empirical problem for the economic signaling model is to tease out the signaling effects from the other motivations for education with which it is confounded. How could that possibly be done?

The demand for education is driven by a mixture of: human capital investment (in both the traditional skills/knowledge form and in Cowen's version); credentialing (the BS is now effectively a 'white-collar union card'); economic quality signaling (a la Caplan); and social status signaling.

These factors interact and reinforce one another. They are also highly heterogeneous both over students and over fields of study. In the absence of controlled randomized experiments (which would necessary have rather complex designs to be able to disentangle multiple higher-order interactions), how could the effect sizes be validly estimated?

(One thing seems sure: the sort of wage regressions that Bryan seems fond of will not do the job.)

Such experiments are impossible of course. (We cannot rerun US history in a world in which credentialing incentives are absent, etc.)

Steve Sailer writes:

Hollywood has more nepotism for blue collar and executive jobs than for the glamour jobs. The film crews I pass by every week on the streets of Los Angeles are overwhelmingly non-Hispanic white, whereas all other outdoor work crews are Hispanic dominated. That's because most of the jobs on a set are union jobs and to belong to the union you need to have a relative or friend.

Nobody ever seems to think of suing Hollywood for disparate impact discrimination because we all know Hollywood is so liberal and would never dream of discriminating against Latinos, so your lying eyes must be misleading you.

Krishnan writes:

Re: BC - good points... I do not have a problem accepting the idea that a "degree" provides a "signal" to potential employers - the problem is that this "signalling" theory can be used to explain almost anything - ... almost seems like a "theory of everything" - and when presented with data that does not directly support the idea of signaling - the response is something like this "Well, it is not all signaling - perhaps 20 or 40 or whatever percent" - and so on (So, when confronted with why for some jobs a Chemical Engineering graduate may be better, the response would be "yea, sure - we are not claiming that signalling explains everything" (and so on and so on ...) - Explains why "economics" and such remain very much subjective - almost anything can be explained - after say several thousands of words using hundreds of pages where you get lost after a few paragraphs

guthrie writes:

@Steve, clearly you've never seen 'Bowfinger' which, by all accounts, is a documentary... ;)

Floccina writes:

BC makes a very good point. I have wondered the same thing but on the other hand a know a few engineers that workout of the field that they were schooled in, and I do not know that many engineers. what percentage of engineers work in the field that they studied seems like an excellent empirical question. These days it seems many engineers work as computer programmers.

It could also be that taking chemical engineering signals that you are good at and interested in the type of things that chemical engineers do.

I cannot believe that Tyler thinks that 2/3 of college is something other than signalling. I wonder if he is for pushing most people to go to college. I would think that even 1/3 would be sufficient to cover the cost!

Comments for this entry have been closed
Return to top