I postulate the wage return to signalling as going away within five years, in say a career of forty years
Some remarks:
1. Why does Bryan have to ask anyone’s opinion about the relative weight of signalling? Perhaps because this question may never be settled empirically. If it could be settled easily, then that $20 bill would have been picked up by now. If it cannot be settled easily, then perhaps the reason is that the signalling model is in many respects “observationally equivalent” to the human capital model.
2. Tyler is suggesting that the signalling proportion of the wage premium for a college degree ought to decay as employers learn from experience about workers, reducing the premium they are willing to pay for an old signal. On the other hand, human capital obtained in school might not decay to the same extent. However, I am confident that someone can come up with a signalling story that is consistent with persistence of a premium or come up with a human capital story that is consistent with decay.
Another take on #1 is that for many economists the question has already been easily settled in favor of human capital. I know that in my grad labor classes the large cohort variance of earnings over time was the most important stylized fact in justifying the importance of human capital/learning. Perhaps they should have been less sure of themselves in light of your #2 possibility but people seem very confident that this stylized fact requires there be a form of human capital in any story trying to explain lifecycle wages. Perhaps they are wrong about this and Bryan will convince us otherwise but that is my reading of the state-of-belief among labor economists.
Another way to say this may be that it is (relatively) easy to write down human capital or sorting models that hit the bulk of stylized facts about panel earnings data without relying on signalling mechanisms but very very hard to write down a signalling model to explain the same data without relying on a human capital mechanism.
Try:
“Macroeconomics Falsifications”
Although falsification is used as the test of “Science,” there is no reason it can not be used as a test of the validity of “thinking” that is “Non-Scientific” (though it may purport to be otherwise).
The first commenter has a very good point.
I can see degrees being ignored five years after you’ve gotten your first job. However, the comparison is between that person and the person who didn’t have a degree at all. Five years later, I’d expect the non-degree person is either just getting started or is still struggling to find someone to give them a chance.
Of course, it depends on the degree. I truly wish people would separate out STEM degrees in these discussions. There’s something deeply wrong about lumping together a Chem E. degree and a Philosophy degree.