The marriage-rate trajectories of the more- and less-educated began to diverge in the mid-1980s. Although college-educated men and women marry later than those with less education, they are now substantially more likely to be married between ages 30 and 50 than those without a college degree. For college-educated women, the proportion of mothers who are single has remained low; for high-school dropouts, high-school graduates, and women with some college, single-motherhood has increased dramatically.
McLanahan (2004) emphasizes how the disparate patterns in marriage and fertility across socioeconomic groups affect the inequality of parental resources available to children. Mothers with lower levels of education have their children at younger ages, are less likely to marry and more likely to divorce, and have lower levels of employment than highly-educated mothers.
...Becker's analysis assumes that the gains to marriage are generated by specialization and the division of labor, implying that those with higher wages will tend to marry those with lower wages. If both spouses are working, however, then most of the gains to marriage may not arise from specialization and a division of labor between home and market, but instead from the joint consumption of household public goods (Lam, 1988). In this case, marriage market equilibrium implies positive assortative mating.
The last paragraph is interesting food for thought. When Ozzie is not looking for someone to fix dinner and mop the floor, he marries a college graduate instead of Harriett. The upshot is that greater equality within the household leads to greater inequality between households.
The article appears in the most recent Journal of Economic Perspectives.