Debating the Signaling Model of Education

At Econlog, GMU economist Bryan Caplan and Princeton economist Bill Dickens have been debating the signaling model of education. See this post for Bryan’s most recent contribution and links to earlier parts of the debate. Bryan argues that a large part of our education spending (perhaps as much as 80%) is socially wasteful “signaling.” It is a kind of arms race where students try to get more education than than their rivals in order to signal their conscientiousness, conformity, and intelligence to potential employers. Crucially, however, much of the information learned is actually not needed for their careers; the real objective is just to rack up better-looking credentials than the Joneses in order to look good to employers.

Both sides make many good points. Overall, I am not persuaded by Bryan’s argument, at least not yet. The crucial objection, raised by Dickens, is that if most education expenditures are primarily about signaling, it should be possible to find other, cheaper ways to signal these desirable traits to employers. Bryan in fact concedes that “intelligence is fairly easy to observe (even in a regime where IQ tests are only semi-legal).” For example, applicants can submit their standardized test scores even if employers don’t require them to do so. Intelligence can also be signaled by getting a high grade in one or a few difficult courses at the high school or college level. You don’t really need four years of college grades. So the debate really turns on the extent to which it’s possible to find easier and cheaper ways to signal conscientiousness and conformity. Here, Bryan argues that there is an adverse selection problem:

[C]onscientiousness and conformity are often hard to spot – especially when people have a strong incentive to fake them. Even worse, low educational attainment relative to IQ is a strong signal of low conscientiousness and conformity. So when employers interview a smart person with little education, they infer that the person is well below-average in other productive traits.

As Bryan sees it, a cheaper or quicker method of signaling (e.g. – a college that takes only one or two years to complete) will tend to attract noncomformists and slackers, the types of people whom most employers seek to avoid. As a result, they will shun graduates of such institutions. This key part of Bryan’s argument is not entirely persuasive. For one thing, the cheaper or quicker method will not attract a disproportionate number of slackers if it is hard to pass. Consider, for instance, a college that will give you a degree in only one year, but requires you to pass a series of extremely difficult courses that are very strictly graded. If higher education is primarily about signaling and the actual content of courses doesn’t matter very much, that type of program will attract hardworking, capable people eager to get into the work world faster and at lower cost. It should spread quickly. Indeed, employers might even start to look askance at the slackers who spend four years hanging out and socializing at conventional colleges.

A second relevant consideration is that conscientiousness and conformity is better signaled by good work at boring and unpleasant tasks than at relatively interesting ones. If you do the latter well, it could just be because you enjoyed them, not because you are dedicated and trustworthy. In most four year colleges, students have considerable choice as to which courses to take, and can usually avoid those they find boring or off-putting. By contrast, many blue collar and service jobs have extremely boring and unpleasant elements that are hard for workers to avoid. If your goal is to signal conscientiousness and conformity, a year of good performance at McDonald’s is probably a better signal than a year of academic success at most colleges. And unlike college, McDonald’s doesn’t charge tuition and pays you a salary (even if a small one).

When I was in high school, I did a lot of babysitting and lawn work. These jobs were generally boring and repetitive, and I often hated them. Yet, for the most part, I did fairly well. My effective performance of these tasks was a much better signal of conscientiousness and dedication than my work in various academic classes, especially the ones I took in college where I had a free hand in picking most of my courses. Indeed, what could be a better signal of conscientiousness and conformity than the fact that people were willing to entrust their children to me, sometimes for many hours at a time?

Yet few if any white collar employers cared about this part of my record. Had I tried to get a job based on a combination of my standardized test scores (signaling intelligence) and glowing recommendations from the people I did babysitting and lawn work for (signaling conscientiousness and conformity), I probably wouldn’t have done very well.

I suspect that my experience was not atypical. Perhaps most employers are simply too stupid or too tradition-minded to hire workers based on these credentials alone. But, as economic history shows, the first employer to recognize and correct a major inefficiency in hiring labor is likely to get a huge competitive advantage. Over time his rivals will have strong incentives to copy his innovations.

In sum, I think that Bryan overstates the extent to which signaling drives education expenditures. Like Dickens, I conjecture that successful completion of college courses often improves people’s qualifications even if the specific knowledge they learn has very limited market value in itself. For example, it could do so by improving the students’ reasoning ability, writing ability, or organizational skills. Bryan doesn’t deny this completely, but his argument can only work if such effects are very small relative to the impact of signaling. At the same time, I agree with him that the education system has numerous inefficiencies, many (though by no means all) of them caused by government subsidies and regulation. I’m just skeptical that the signaling arms race is nearly as big a part of the problem as he contends.