Fortune has been chronicling how money shapes marriage since long before this study — the magazine’s Julie Connelly coined the term “trophy wife” itself in a 1989 cover story, a phrase that’s since become an inescapable shorthand across pop culture for wealth-driven pairings. Now, new research suggests the more consequential story about marriage and money is happening in the opposite direction: not among the wealthy elite, but in the widening gap between non-college women and the shrinking pool of “marriageable” men.

A National Bureau of Economic Research working paper tracked marriage patterns across U.S. birth cohorts from 1930 to 1980 and found that college-educated women have kept their marriage rates stable even as their pool of similarly educated men shrank dramatically. The ratio of college men to college women fell from 1.8 to 0.8 over that span, and there are now 1.6 million more women than men enrolled in four-year colleges nationwide, according to the study by economists Clara Chambers of Yale, Benjamin Goldman of Cornell, and Joseph Winkelmann of Harvard.

“Four-year college rates for men stagnated essentially many decades ago, whereas for women, it’s been a steady increase,” Goldman told Fortune. “That’s really been the key factor that’s led to the changing ratio over time.”

The paper, “Bachelors Without Bachelor’s: Gender Gaps in Education and Declining Marriage Rates,” complicates the mutation of the “trophy wife” era into the “power couples” era: as wealthy Americans increasingly marry within their own income and education tier. This new research shows the opposite pattern playing out further down the income ladder, where college-educated women are increasingly pairing with men who skipped college altogether.

The shift also reflects the labor market of the era, Goldman said. Well-paying, often unionized manufacturing jobs once let men build a solid living without a four-year degree, while fields that have drawn more women, like nursing and teaching, increasingly required one.

College women adapted by marrying down. The share of marriages in which the wife has a four-year degree and the husband doesn’t has quadrupled, from 2.3% of all women in the 1930 birth cohort to 9.6% in the 1980 cohort. Crucially, they’re marrying the highest earners without college degrees, whose incomes rose from an average of $61,400 to $68,400. Non-college men who didn’t marry a college-educated woman saw their own average earnings move the opposite direction, falling roughly $10,000 to about $46,000 — underscoring how the gains from a shrinking pool of “marriageable” men have concentrated at the top rather than lifting the group as a whole.

“If you look historically at marriage patterns in the U.S., women who went to four-year colleges predominantly married men who also went to four-year colleges,” Goldman told Fortune. “But there’s simply not enough of those guys to go around today. You’d either expect to see more college-educated women not marrying, or they’d have to marry across education boundaries.”

Goldman cautioned that the paper doesn’t examine why individual couples pair off the way they do—only these market-level patterns. He also pushed back on reading a status gap into these marriages, pointing to a more typical example like someone who went to nursing school marrying a small-business owner, or a well-paid electrician or HVAC technician. 

That pattern matches what Institute for Family Studies researchers have called “skimming” — college-educated women aren’t rejecting non-college men wholesale, but they disproportionately end up with the most economically stable ones, leaving less-stable non-college men further behind in the marriage market.

“It’s not clear in those marriages that there’s kind of like a power imbalance just because of the education dimension of it,” he told Fortune.

That cuts against a narrative that’s spread dating commentary—essays touching on the difficulty of finding a partner as a highly educated woman have proliferated in outlets like The New Yorker and The Atlantic. But the study’s own data show college-educated women are, in fact, marrying at higher rates than any other group in the analysis.

That anxiety around scarce, high-status male partners isn’t new—it’s the flip side of the dynamic Fortune first named in 1989, just moving down the income ladder from the CEO’s second wife to the electrician’s wife.

“It’s not to say that that’s off base or not people’s experience or anything like that, but part of what motivated us here was this puzzle, which is when you look at the data, that is the demographic group in the U.S. that is still marrying at the highest rates,” Goldman added. 

As college women absorbed these men, the pool of “marriageable” ones available to non-college women—defined in the study as those earning above the national median and not already married to a college graduate—collapsed from 72.9% of non-college men in the 1930 cohort to 35.3% in the 1980 cohort, a drop of more than half.

The effect on actual marriage rates is stark: non-college women’s likelihood of being married fell from 78.7% to 52.4% across the cohorts studied, while college-educated women’s marriage rates held steady above 70% throughout the same period.

This finding data led the researchers to predict that Americans born in the mid-1990s are on track to become the first cohort where more than half of non-college women won’t be married at age 45.To build that forecast, the researchers looked at how many people in a birth cohort were already married or living with a partner in their early 30s, then applied that same relationship from past cohorts—where the age-45 outcome is already known—to project forward. 

The dynamic echoes a well-known 2014 paper, “Marry Your Like,” by economists Jeremy Greenwood, Nezih Guner, Georgi Kocharkov, and Cezar Santos, which found that rising sorting by education and income since 1960 has been a significant driver of the growth in U.S. household income inequality.

Goldman noted the underlying trend shows up even without the forecast: marriage and cohabitation rates for non-college women in their early 30s have kept falling in every cohort through the one born in the mid-1990s.

“I actually think this type of thing resonates more quickly with a normal person than it might with someone who went to a super fancy college,” Goldman said. He pointed to nationally representative survey data—like the University of Michigan’s Monitoring the Future survey of high school seniors—showing that most Americans still list having a family among their central life goals. 

“There’s work in social psychology and other fields that’s shown, at least correlationally, that being married is a key input to health, longevity, and avoiding loneliness in old age,” he said.

‘What affects one group naturally affects the other’

Goldman’s study dovetails a larger trend of the economic unraveling facing men without college degrees. About 16% of non-college-educated men now live with their parents, double the portion of their college-educated counterparts, as rents have risen 150% since 1960 while their wages have barely budged. 

Scott Winship of the American Enterprise Institute previously told Fortune the decline of marriage itself is feeding the problem: “I think marriage really is the sleeper issue here,” he said, noting that working-class men who once could have “tolerated higher housing costs” as married breadwinners are now single and thus more likely to work less and live at home. 

There’s also a growing gender imbalance in the workplace that is also hitting men harder. Women now hold more U.S. payroll jobs than men for just the third time in history, a change economists previously told Fortune looks structural rather than cyclical. Female-dominated sectors like healthcare have driven more than half of recent U.S. job growth, while male-heavy fields like manufacturing and tech have stagnated—compounding the earnings erosion the NBER researchers link directly to marriage rates.

“I would view our paper as kind of diagnostic at a very high level,” Goldman said. “It’s very clear, both in the data over time and across places, that the environments where you see declining marriage rates for working-class women are precisely the environments where men are doing worse economically.”

Other research on the question points in different directions, he noted. Economist David Autor and his coauthors found that manufacturing decline during the China Shock reduced marriage rates in the hardest-hit communities. Economists Melissa Kearney and Riley Wilson, studying localized fracking booms that raised men’s earnings, found increases in both marital and nonmarital births—but no corresponding increase in marriage rates.

Not all economists agree assortative mating itself has intensified. A 2016 NBER paper by Rania Gihleb and coauthors argued that once measured properly, there’s little compelling evidence that educational assortative mating rose over the past half-century — a reminder that Goldman’s findings describe a shift in who marries whom by credential, not necessarily proof that Americans have become choosier about matching status overall.

The pattern bears out geographically, though: the marriage gap between college-educated and non-college-educated women narrows in regions where non-college men face lower joblessness and incarceration rates, reinforcing Goldman’s point that men’s local economic standing and women’s marriage prospects move together.

Goldman was careful to note that his own paper doesn’t establish causation.

 “The work we do in the paper is not enough to attribute the decline in economic outcomes among men as having caused the changes in marriage patterns we observe,” he told Fortune. “It’s more noting that they seem to occur at the same time.”

He also resisted the idea that the findings amount to a tradeoff between helping men and helping women. 

“What affects one group naturally affects the other, because you need a pair,” he said.