Everybody is talking about inequality and everybody seems to have their own favorite culprit, from globalization, to deregulation, to skill-biased technical change, and political capture. Yet a recent paper by Greenwood, Guner, Kocharkov and Cezar Santos (Greenwood, Jeremy, et al. “Marry Your Like: Assortative Mating and Income Inequality.” National Bureau of Economic Research, No. w19829, 2014) suggest an additional factor much closer to home: whom you marry.
The argument is simple: Go back to the 1950s when few women worked. As a result, back then household income inequality was largely determined by the income of the men of the house. However, as it was more likely that women married to poorer men worked, household income inequality was actually slightly lower than it would have otherwise been with male earnings only.
Now, fast-forward to the present day, when most women in the developed world have joined the labor force and contribute significantly to household income. All else equal, this should reduce household income inequality as women’s income contributes to smoothen the disparity from men’s income in household income inequality.
People tend to marry people with a similar profile
However, people don’t just marry randomly. They tend to marry people with a similar profile, both in terms of income and educational background. This result is on the one hand very intuitive to understand. Romantic serendipities aside, an individual is more likely to primarily be exposed and relate to another individual having a similar social and educational background and, ceteris paribus, more educated and wealthier parents will create more opportunities for their children*.
Compared with a world in which people married randomly, this turns out to be a large contributor to household income inequality. Greenwood et al. (2014), using data on hundreds of thousands of households from the United States Census Bureau in the period 1960 to 2005, find that a growing tendency to find a partner from a similar educational and professional background is exacerbating income trends . In a regression analysis they find that the additional impact of a husband’s education on his wife’s education relative to the baseline year (1960) is increasing, implying the degree of assortative mating has increased.
They confirm this result constructing contingency tables for years 1960 and 2005 finding both that the ratios of the sums of diagonal elements for both years is larger than one. This finding implies that the number of matches between husband and wife with the identical education level is larger than what would occur if matching were random.
Economists use the Gini-coefficient to measure income inequality. It is always between zero and one with zero being complete income equality and one being complete income inequality (one guy has the income of the whole society). Second, the ratios increase over time. If matching in 2005 between husbands and wives had been random, instead of the pattern observed in the data, the Gini coefficient would have fallen from the observed 0.43 to 0.34. Interestingly, 0.34 is exactly what household income inequality was in 2005. Hence, completely random marrying would have cancelled the rise in household income inequality since 1960.
The policy implications of this insight are not clear; the analysis does demonstrate the multifaceted causes of rising income inequality, and shows that they go way beyond the scope of just economics.
Authors: Prof. Morten Olsen and Ria Ivandic