Marriage is an implicit contract that is very difficult (and costly) to renege on. But it is an institution that has been around for a very long time, so it must be satisfying some need from both parties. There must be some Economics involved, right?
Gilles Saint-Paul offers an explanation: it is all about the children. Just from biology, women know which children are theirs. But to convince men they are the legitimate fathers, women need to offer a marriage contract and thereby forgo potentially better opportunities for other men. Men are fine with this contract, despite the fact that they could have mated with other women, because they are assured of the legitimacy of their children, and that it is worthwhile investing in them. For this contract to be credible, women must be able to commit to the marriage and thus accept penalties in case of adultery. Men do not need to, though, because it works against their interest to recognize their children.
What does this entail for investment in human capital? Assume human capital is at least partly inherited. Clearly, men will provide more effort if they can recognize their children. But it matters also what kind of marriage matches happen. By committing to a man, a woman passes on future better opportunities. But, unlike some other contracts, she is unable to transfer utility if she marries a male of lesser quality. This means that the marginal utility of consumption is lower for a women of higher human capital and she will require a much larger transfer from the male for forgoing future opportunities. For men, the more human capital they have, the lower the marginal utility and the more they are willing to invest in their kids.
It turns there are two possible equilibria here: A so-called Victorian one, where there is positive assortative matching with women marrying men with the same human capital. But there can also be a so-called Sex and the City one, where men marry women with lower human capital, and there are high human capital ones that are not satisfied with the remaining low human capital men. The latter is more likely if the distribution of human capital becomes more unequal, as some women (dumb blondes?) can underbid more intelligent ones. In aggregate terms, this means also that there is less talent in the next generation.