Why do university graduates generally earn more than non-graduates? The finding that returns to arts degrees for men are very low sheds light upon this.
Such degrees should be a signal of high ability; entry requirements for many such courses are as high as for sciences, and completion of such courses are signals of high communication skills at least.
And these courses also create human capital; languages and essay writing are useful skills.
That such degrees pay badly is, therefore, not obviously consistent with either the human capital or signalling theories of how education boosts earnings.
There is, however, a third theory that doesn't get the attention it deserves. It's the idea of incentive-enhancing preferences discussed here (short pdf) and here (long pdf).
The idea is that bosses want workers with the right attitude - ones they can trust. And arts' graduates can't be relied upon to have the right attitudes. They're interested in soft, fluffy, arty-farty things like culture and books rather than the hard bottom line of business. So they're not so employable.
This problem is worse for men than women because of cultural stereotypes. We expect women to like fluffy things like art, books and languages - it's part of their charm. So a woman with an arts degree isn't signalling such wrong attitudes as her male equivalent, and so is more employable (relative to other graduates) than him.