[data] Throughout the 20^th century, the relationship between women’s human capital and men’s income was nonmonotonic: while college-educated women married richer spouses than high school-educated women, graduate-educated women married poorer spouses than college-educated women. This can be rationalized by a bi-dimensional matching framework where women’s human capital is negatively correlated with another valuable trait: fertility, or reproductive capital.

Such a model predicts non-monotonicity in income matching with a sufficiently high income distribution of men. A simulation of the model using US Census fertility and income data shows that it can also predict the recent transition to more assortative matching as desired family sizes have fallen.

…In this paper, I posit that underlying these apparent contradictions is the fact that human capital investments that yield greater income may also decrease another desirable marriage market trait, “reproductive capital.” Women’s fertility decreases with age, and human capital investments that increase income also delay marriage and childbearing and increase spacing between births. I first show that older age at marriage is linked to lower spousal income for women, aligning with experimental findings that men value women’s age through the channel of fertility (Low2023a). I then document for the first time that husband’s income has historically exhibited a nonmonotonic pattern in wife’s education: additional education up to a college degree was associated with increased spousal income, but education beyond college was associated with decreased spousal income. This pattern cannot be rationalized by a traditional unidimensional model but can be easily explained by a bi-dimensional model where income is negatively correlated with fertility.

I outline a transferable utility matching model between men characterized by income and women characterized by income and fertility. A latent human capital type impacts both income and fertility. This contributes to a growing literature showing that truly multidimensional models, as opposed to index frameworks, may be crucial in understanding matching patterns, since valuations of non-income traits likely vary with income (Coles & Francesconi2011, 2019; Dupuy & Galichon2014; Chiappori et al 2017; Lindenlaub & Postel-Vinay2023; Galichon et al 2019; Galichon & Salanié2022). I demonstrate that with a surplus function that is supermodular in both incomes and income and fertility, nonmonotonic matching on incomes can appear. The stable match will depend on the trade-off between human and reproductive capital in women’s type distribution relative to men’s income distribution. I provide a simple condition such that there always exists a man rich enough that he prefers a higher fertility but poorer woman to a richer and less fertile woman.

…If men indeed value fertility as a marriage market trait, it suggests that time-consuming human capital investments would be a double-edged sword for women: on the one hand, human capital carries higher earning, a presumably positive attribute likely to help attract a high-income spouse. On the other hand, income-increasing investments take time, decreasing what could be another valuable asset on the marriage market: reproductive capital.

While much empirical work categorizes all women with college degrees as “college plus”, the reproductive capital hypothesis suggests that women with college degrees and graduate degrees may have very different marriage market outcomes, since women with college degrees only could still marry quite young and have large families. Moreover, graduate degrees are correlated with the types of high-investment careers that may continue to interfere with time to have children: the tenure track, the partner track, surgical residencies, and climbing the corporate ladder.

…These facts suggest a second factor that is decreasing in education, even as income rises. Thus, I introduce the concept of reproductive capital, which depreciates with age.⁷ Table 1 shows just how substantially highly educated women’s fertility differed historically from those with college or lower degrees, using 1970 US Census data. Highly educated women had almost 1⁄2 fewer children on average and were only 2⁄3^rds as likely to have more than 4 children compared with college-educated women. By contrast, there is little difference in these family size metrics between those with college degrees and those with high school education or some college. In addition to marrying older than all other educational levels, highly educated women may also be more likely to make post-education career investments that delay childbearing and increase spacing between children.

See Also: