[econophile]

Riverman,
The signalling theory of education, one of the two major hypotheses labor economists use to explain the large returns to education, gives an explanation people might go to school even if they learn anything useful there. The idea is that employers can't tell high-ability and low-ability workers apart, at least not in a job interview. High-ability workers are more productive than low-ability workers (although the employer can't observe productivity until some time after they hire a new employee), so firms are willing to pay more for high-ability workers. But if the firm offers higher wages for high-ability workers, they need a way to tell the workers apart before hiring them.

College provides a way for workers to reveal their ability. To get a college degree, students must complete a series of annoying tasks. If these tasks cause sufficiently more hardship for low-ablity students than for high-ability students, then the low-ability students will not go to college despite the higher wages offered to college graduates. And high-ability workers will go to college because firms will only offer high-wages to college graduates (workers that have proven their high ability).