A brand new job market paper has some interesting data on assortative mating and its impact on income inequality. Drawing on data from the Panel Study of Income Dynamics (PSID), the paper
investigates the evolution of assortative mating based on permanent wage in the U.S. since the late 1960s; quantifies its impact on rising family-wage inequality; and, finally, tries to understand the factors behind this evolution. It documents a significant increase in assortative mating, as measured by couple’s permanent-wage correlation, between families formed in the late 1960s and in the late 1980s. I then show that changes in the degree of assortative mating accounts for a sizable amount of the increase in family wage inequality across these family cohorts. This finding shows focusing on the time trends in permanent-wage inequality is not enough to understand the mechanics of increasing family wage inequality. Note, this finding does not rule out a feedback mechanism. It might be that increasing family wage inequality incentivized individuals to care more about their spouse’s wages, causing a higher degree of marital sorting along permanent wage, which in turn mechanically increased family wage inequality (pg. 23).
What is “a sizable amount”? The data show that “the trend in assortative mating explains more than one-third of the increase in family wage inequality” (pg. 11; emphasis mine). He explains,
I use the Gini coefficient in the left panel, as the measure of inequality, whereas the variance of logarithm is used in the right panel. Solid lines show the actual evolution of family wage inequality with these measures. In the dashed lines, however, I construct the counterfactual family wage inequality by holding the empirical copula distribution fixed at its initial form while letting the permanent-wage distribution vary. The dashed lines thus show the counterfactual evolution of family wage inequality if no change occurred in assortative mating. The left panel shows the rise in the Gini coefficient would be 40% lower, and the right panel shows that the rise in the variance of (log) family wage would be around 35% lower under the counterfactual scenario. The increase in assortative mating thus accounts for a significant amount of the increase in family wage inequality (pg. 11).
