There are a lot of good reasons to marry an educated partner. Among them is that his/her income is higher, and theory has consistently pointed to the fact that one's own returns to education should be higher. This is known as the supermodularity of the marriage return function: the second derivative of outcomes with respect to both education levels is positive. This implies that the return to education becomes even higher, as the couple's surplus increases even more due to the supermodularity. But how much?
Pierre-André Chiappori, Bernard Salanié and Yoram Weiss use an extract of the US Census to confirm that supermodularity is present is very significant manner. It varies by cohort, though, with in particular younger females seeing stronger returns than young males. This means that women have done better in three dimensions in recent decades: they closes a large part of the wage gap with men, they get a higher marital college premium and they marry better. All this compounds to remarkable progress for women, who also work more and get a larger share of the marital surplus.