Human capital, signaling, and the pattern of returns to education

This paper analyzes the implications of the pattern of returns to education for human capital and signaling models. The US Current Population Survey shows diploma effects and small returns to the interval just before college graduation. A human capital model explains this pattern only under special circumstances unlikely to be stable. A Spence-type signaling model can explain diploma effects, but not the low return prior to graduation unless there is substantial measurement error, I propose an extension of the signaling model where agents Lire initially unsure of their ability.