Private sector unions have almost disappeared, but occupational licensing is all the rage.  Almost 30% of all workers need a license to do their jobs – and licensed workers earn roughly a 15% wage premium.

Occupational credentials are one common licensing requirement.  According to this working paper by Kleiner and Krueger, 43% of all licensed workers are required to have a college degree.  This is consistent with one of Arnold’s favorite alternatives to the signaling model: Employers value degrees in useless subjects because they’re legally required to do so.

Still, the question remains: How much of the return to education is a disguised licensing premium?  Kleiner and Krueger create a brand new data set that allows them to directly answer this question.  Even though they find a large licensing premium, they conclude that licensing only mildly inflates the return to education.  Their results:


The first two columns show standard return to education estimates.  Ignoring licensing, a year of schooling appears to boost your earnings by 10.9%.  Adjusting for licensing, a year of schooling appears to boost your earnings by 10.5% – .4 percentage-points less.  Columns 3 and 4 measure education with degrees instead of years and re-estimate.  Once again, the coefficient on education changes in the expected direction, but only a little bit.

I freely admit that licensing isn’t the only way that government can artificially inflate the demand for useless degrees.  Government pay scales are another plausible candidate.  But still, given the prevalence of licensing and the substantial licensing premium, I would have expected a larger effect.