Education, Income, and 'Marriageable' Men

In the U.S., about 40% of kids are born to unmarried mothers. For decades, scholars and commentators have debated why this number rose so much in the latter half of the 20^th century—and why marriage has continued to thrive among those with college degrees while deteriorating among those with lower levels of education.

A new paper from Clara Chambers, Benjamin Goldman, and Joseph Winkelmann helps to unpack some of the dynamics here, diving into the complicated interplay among marriage, income, education, and fertility. It supports a version of the “marriageable men” hypothesis, which posits that marriage has declined among the less educated because many men at the bottom of the socioeconomic ladder are not seen as suitable marriage partners, especially in a world where women have greater economic opportunities and/or access to safety-net benefits.

The study is largely framed around a minor paradox. It’s commonly thought that women place a high value on “marrying up”—and for several decades now, women have been more likely than men to attain BAs. If women insist on marrying someone with equal or higher education to their own, it’s college-educated women who should have trouble finding men to marry.

Some educated women do experience struggles along these lines, of course. But statistically, marriage for college-educated women has been remarkably steady over time. Women born from 1940 all the way to 1980 have had about a 70% chance of being married in their 40s if they had four years of college. Those without college degrees had about the same chance if they were born in 1940, but in the 1980 cohort—the first of the Millennials—only about half of those without degrees are married in their 40s. 

The solution to this riddle is that many college-educated women are happy to marry non-college-educated men—around a fifth to a quarter have long done so—but they tend to skim off the most economically stable ones (in many cases likely still “marrying up” income-wise). Chambers et al. vividly illustrate this with trends in men’s earnings.

Between the 1930 and 1980 birth cohorts, the authors estimate, the average real earnings of non-college-educated men in their 40s dropped from roughly $56,000 to $50,000. I’d caution against overinterpreting that decline by itself,¹ but it serves as a baseline for an informative exercise. If you separate non-college-educated men into two groups based on whether they were married to a college-educated woman, the trends differ markedly. The minority of non-college-educated men grabbed up by women with BAs had higher and increasing wages between these cohorts ($61,000 to $68,000), while the rest of the men saw earnings drop from about $56,000 to about $46,000.

If you’re a non-college-educated woman in a dating pool of mostly non-college-educated men, that’s not good news—though future research might explore the dating-market dynamics here. I’m curious what it is that apparently gives college-educated women a leg up in the dating pool for less-educated men. Do men care about their partners’ education level more than stereotype would have us believe, do higher-earning men without college degrees run in social circles with more educated women, or what?

Next, the authors look at modern census data, rather than focusing on trends over time. The idea is essentially to look at variation in non-college-educated men’s employment while holding women’s college attendance constant: In other words, if women in two geographic areas² have the same rate of female college attendance, but men without college degrees have higher employment rates in one of them, what happens?

The answer is that non-college-educated women are much more likely to get married when non-college-educated men are doing better: “In areas with the lowest employment rates for non-college men (bottom 2.5% of the ... distribution), the marriage rate for non-college women is 44.5%, compared to 66.0% in areas with the highest employment rates for non-college men (top 2.5%).” The effect on college-educated women’s marriage rates is significantly smaller. The results are similar when the data are disaggregated by race, and when other variables such as incarceration are used in place of employment.

Fertility works a little differently. The authors write:

For college women, fertility declines almost one-for-one with marriage rates as [employment] for non-college men decreases. In contrast, for non-college women, the association between fertility and [employment] for non-college men is approximately 60% weaker than the association with marriage. In areas where men are less likely to work, non-college women continue to have children, but often outside of marriage.

This study contributes to a growing body of research about the role of economics in marriage’s collapse: Men’s economic performance, including their performance relative to women’s, seems to matter. And today, about a tenth of men, ages 25-54, are out of the labor force entirely.

Fundamentally, the story is that as many lower-skilled men have fallen behind and women’s own opportunities have grown, marriage has suffered. It’s not that women, even highly-educated women, categorically refuse to marry men without college degrees—but rather that many non-college-educated men are economically floundering or facing other serious problems. And since sex and childbearing don’t necessarily stop when marriage does, the result is many kids being born into unstable family arrangements.

Perhaps there is no turning back the clock. But restoring a healthy balance of work, income, marriage, and children among the non-college-educated should be among policymakers’ highest priorities.

Robert VerBruggen is an IFS research fellow and a fellow at the Manhattan Institute.

1.  The authors use an inflation adjustment, the CPI-U, that’s arguably too aggressive, and since Americans have gotten more education over time, “non-college” represents a shrinking and more negatively selected group over time. See Scott Winship for more on men’s wages over time, inflation, and education.

2. Specifically, “Public Use Microdata Areas,” which cover about 140,000 people on average and are the smallest geographical level for which individual-level data are available.