I. What Is The Hypergamy Hypothesis?

“Female hypergamy” (from now on, just “hypergamy”) is a supposed tendency for women to seek husbands who are higher-status than themselves. Arguing about educational hypergamy (women seeking husbands who are more educated than themselves) is especially popular, because women are now (on average) more educated than men - if every woman wants a more-educated husband, most won’t get them, and there will be some kind of crisis.

Freddie de Boer’s Demographic Dating Market Doom Loop presents an argument that educational hypergamy is lowering marriage rates¹. He writes:

The fact that women are earning so many more degrees than men has social consequences . . . It will, I’m sure, shock you to learn that research done with dating app data (which has the advantage of being more honest than mere self-described preference) shows that women place dramatically more stock in a man’s education and income level than men do when searching for a woman partner [...]

As more hard-charging women flood a given dating market, while the number of eligible men drags behind because of increasing advantages for women in school and the workplace, fewer and fewer women are likely to find themselves with a partner they consider marriageable [...] Career women are faced with a growing structural disadvantage of insufficient suitable partners, which is exacerbated as they age because of men’s continuing preference for younger women.

On the other hand, Marginal Revolution recently highlighted a paper finding that “in England and Wales 1837-2022 . . . there was never within this era any period of significant hypergamous marriage by women.”

So who’s right? Is this going to be one of those annoying cases where they’re both right in different senses?

II. Complicated Definitional Issues

You didn’t think you would get away without this section, did you?

Hypergamy can be “absolute” or “relative”. Absolute hypergamy means that eg a husband is more educated than his wife. Relative hypergamy means that a husband’s educational rank among men is higher than his wife’s relative educational rank among women. That is, if the second-most educated woman (who has 25 years of education) marries the first-most educated man (who has 20), this would be relative hypergamy (she married a higher-ranked man) but absolute hypo-gamy (she married a man with less education than her).

If everyone got married, then by definition, on average marriages would in absolute terms be educationally hypo-gamous². After all, women have more education than men. The average wife would have [average woman’s education level], the average husband would have [average man’s education level], and since the average woman is more educated than the average man, on average women would be marrying less-educated men.

And if everyone got married, then by definition, on average marriages would in relative terms be educationally equal. After all, on average, the average man has 50th percentile educational rank among men; the average woman has 50th percentile educational rank among women, therefore the average husband and wife would share the same educational rank.

So hypergamy is only interesting and worth talking about in two cases:

First, if not everyone gets married, In this case, maybe the most educated 5% of women don’t get married, the least educated 5% of men don’t get married, and every man can pair up with a woman who’s less educated than he is. Since in fact not everyone gets married, this is easy to check, and most studies on hypergamy check it.

Second, even if everything averages out to average, the slope could be different at the top vs. bottom of the distribution. Suppose that women want more educated men, but men want something else out of women (let’s say beauty). The most educated man would pair off with the most beautiful woman, and so on to the least educated man pairing off with the ugliest woman. On average, husbands and wives would have the same education. But in the top 50% of the distribution, husbands would on average be more educated than their wives; at the bottom half, husbands would on average be less educated. Why? Assume that education and beauty aren’t especially correlated. Then we expect women at any given tier of beauty to be about 50th percentile for education. So the top half of the male educational distribution is married to (on average) 50th percentile women (who they outrank), and the bottom half of the male educational distribution is also married to (on average) 50th percentile women (who outrank them). So although the entire distribution is equal-gamous, the top is hypergamous and the bottom is hypo-gamous. Not all of the studies we’re going to look at check for this, but some do.

Since we already know women will (on average) be absolutely educationally hypogamous (because they’re on average more educated than men) and absolutely financially hypergamous (because they on average make less money than men)³ most of the rest of this post will be focusing on relative hypergamy.

III. People Like Hypergamy

DeBoer cites these dating site data (source) to show that men’s romantic success varies more with status than women’s:

I have some qualms about the data source⁴, but it doesn’t matter; everyone else agrees. Wikipedia:

Studies of mate selection in dozens of countries around the world have found men and women report prioritizing different traits when it comes to choosing a mate, with both groups favoring attractive partners in general, but men tending to prefer women who are young while women tend to prefer men who are rich, well-educated, and ambitious.

IV. But Educational Hypergamy Is Pretty Dead

There are good reasons to be skeptical of educational hypergamy in particular.

In the modern day, women are more likely to educationally “marry down” (ie marry a husband less educated than themselves) than to “marry up”⁵:

Women (on average) get more education than men, so you would expect to see some effect like this just from the demographics. Is the effect stronger or weaker than you would expect from demographics alone? If it were weaker, you could argue that there is some tendency toward educational hypergamy, but not enough to counteract women’s demographic advantage.

But it’s either stronger or the same. This American study and this French study both investigate this question by comparing the educational distribution of real relationships compared to the expected distribution you would get from throwing random men from the population at random women. The American study is confusingly phrased and might either be saying it’s stronger or just getting stronger over time. But the French study is very clear: women are marrying down in education more often than you would expect by chance. They write:

It appears that social norms, proxied here by the choices made by individuals beyond the constraints of partner availability, are merely the reflection of the most frequent couple configurations in a given cohort, observed in Figure 1. In other words, the growing proportion of couples where the women is more educated than her partner – an inevitable demographic consequence of the rise in women’s educational attainment – seems to have produced a change in the preferences of individuals, in the social norms that are internalized or imposed upon them by their social environment.

What about unmarried women? Is it possible that some women are eagerly accepting the new female-dominated reality, but the study is missing other women who refuse to marry down and so avoid matrimony entirely?

If this were true, you might expect educated women to be disproportionately single compared to educated men. But this isn’t true:

This is a new phenomenon; fifty years ago, there would have been more educated women staying single. But for now, educational hypergamy seems dead.

V. Class Hypergamy, Income Hypergamy

Education closely tracks social class. But maybe it’s not exactly what we’re looking for here. Women get more education than men, which means that in any given family, the daughter will (on average) get more education than the son. Does that make her “higher class”? What if we looked at class directly?

And what about income? If women marry hypergamously in search of men who can provide for them, wouldn’t we expect income to be the most direct measure of this ability? Priscilla Chan (medical school graduate) married Mark Zuckerberg (college dropout), but probably doesn’t feel like she got a raw deal or “married down”. Should we be looking at this too?

I was able to find five studies⁶ that investigated in this level of detail. Here are the results from women’s perspective (ie a down arrow means women tend to marry down):

Everyone who looked at education agreed that women tended to marry down, even in relative terms.

Everyone who looked at income agreed that women tended to marry up, even in relative terms.

Class was more complicated⁷. The Swedish study said in the abstract that it found women tended to marry down in terms of class, but I had trouble finding that effect in the data, and it looked pretty small if it existed. The Norwegian study said they tended to marry up, but with the same caveat. The two American and one English study were explicit about very close class matching, sometimes implausibly close.

Is this contradictory? On average, classier people earn more money. So if women are “marrying up” for money, surely they would, just by accident, also marry up a bit in class, right?

The Norwegian study found some evidence of this. They found that on average, husbands were 8 percentile points higher in income than wives (ie a 50th-percentile-of-income woman would marry a 58th-percentile-of-income man). But husband’s parents were only 0.75 percentile points richer than wives’ parents. They say that based on known parent-child correlations, husbands with 4 pp higher income should have 0.75 pp richer parents. But in fact it’s husbands with 8 pp higher income, having 0.75 pp richer parents. I think this is because the husbands are being selected for higher income, and so their parents regress to the mean, but I’m not sure. In any case, a pretty substantial income difference (8 pp) reduces to a barely-measurable class difference (0.75 pp).

Clark and Cummins don’t measure income, but they do find a very small (~0.5 pp) status advantage for husbands’ fathers; they round this off to zero, but maybe by the Norwegian logic it represents a substantial income difference. However, this difference disappears and even reverses by the last period in their study (1980 - 2021). Although they don’t directly measure income, there’s either no income hypergamy, or it’s happening in the total absence of class hypergamy. How could this be?

Here’s one model; suppose class decomposes into income + education. Everyone’s first goal is to marry the highest-class person they can, but women prefer husbands whose class is lopsided towards income, and men prefer wives whose class is lopsided towards education.

A summary of the Swedish paper mentions that:

Couples with similar education levels, and those with a more highly-educated male partner, actually had more equal incomes than would be predicted if couples were matched randomly. However, in couples where the woman had the education advantage, random matching predicted that more women would be the higher earner than actually were. As in Qian’s study, highly-educated women appear to have an especially strong preference for men who out-earn them.

If I understand right, this suggests that education and income funge against each other; if someone marries down educationally, they’re more insistent on marrying up financially, and vice versa. This provides some evidence that men and women are trying to match overall social status, but select for different aspects of the social status package.⁸

But here’s another model: women whose careers are going badly relative to their class expectations are more likely to marry (since the opportunity cost from abandoning their dead-end career to raise children is lower), and men whose careers are going well relative to their class are more likely to marry (they can delight their equal-class wives with more resources than they expect).

I think either of these patterns would both produce the observed effect of women marrying men of the same class but higher income even though income and class are correlated.

VI. Nobody Cares If You’re Pretty

If both men and women care about their partners’ looks and status, but men care proportionally more about looks, then women must care proportionally more about status. So one form of the hypergamy hypothesis goes: since men seem to care more about looks than women, women probably marry up.

Clark and Cummins, authors of the English study, reverse this logic. In the study above, they show that on average husbands and wives are of equal class (contrary to the predictions of class hypergamy). But what’s the variance? Do men and women marry down equally often, or equally rarely?

A different Clark and Cummins study says: rarely. They find a correlation of 0.8 between wives and husbands’ social classes in England, which is “unchanged 1837-2021”. For context, a correlation of 0.8 is about the correlation between the SAT score of the same person taking the SAT two different times.

They remark that this means people in England can’t care very much about their partners’ looks:

There is also ancillary evidence that physical attraction cannot have been a very significant factor in marriages in any period 1837-2021, based on the correlation observed in underlying social abilities. In terms of family status, matching in marriage was nearly symmetrical between men and women. The male preference for physical attraction would be predicted to reduce the correlation between marital partners in family status. But this was the same effect for women as compared to men.

As noted above, if men prize physical attributes in mates which are uncorrelated with family status then the correlation of status in marriage will decline. However there is ancillary evidence that the importance of physical attributes in forming matches must always have been modest. In a related paper we estimate the correlation of underlying social abilities for brides and grooms in marriage in England as constant at around 0.8 1837-2021 (Clark and Cummins, 2022). This very strong correlation implies that the importance for men in making a match of physical appearance must have been limited, or else physical appearance in women was strongly correlated with social abilities⁹

That is: if men cared about looks more than women, then they would trade off status for looks: faced with a choice between an average-looking woman of the same class, or a beautiful woman of a slightly-lower class, they would choose the beautiful-but-poor woman. But this would mean men would marry lower-class women more often than women married lower-class men, which would imply a less-than-perfect status correlation between husbands and wives. But the data show a pretty perfect status correlation between husbands and wives. Therefore, men can’t care about looks.

Might both men and women care equally about looks? But then both sexes would be frequently marrying people of different classes. And the marriages in Clark and Cummins’ study were generally so well status-matched that this seems pretty rare. So it seems more like neither gender cares about looks than that both do.

How do we square this with the fact that obviously men care a lot about whether women are attractive or not?

Maybe this isn’t as common-sensically wrong as it seems. I know many rich male Google programmers, but I have never seen any of them marry a stunning black girl from the ghetto. Why not? Wouldn’t the hypergamy hypothesis pronounce this a good deal for both of them? He gets a beautiful wife, she gets a rich husband? And it’s not just a race thing, I’ve also never seen them marry a beautiful hillbilly from West Virginia, or a beautiful farmer’s daughter from Modesto. I don’t even really see them marry a beautiful girl from the suburbs with a community college degree.

And when I think of unattractive women I know, I can’t get a good read on whether or not they’re less likely to be married. But the ones who are married are married to husbands of similar social class as themselves. Nor do their husbands seem defective in other ways - except maybe that they’re also no lookers (research does show that husbands’ and wives’ attractiveness correlate at about r = 0.4)

So maybe attractive rich people marry other attractive rich people, attractive poor people marry other attractive poor people, and so on, but there’s no tendency for ugly rich people to marry attractive poor people.

Is this by choice or social necessity? That is, when a rich man marries an average-looking rich woman, is that because he prefers her to a beautiful poor woman, or just because he doesn’t know any beautiful poor women well enough to ask them out? While it’s true that rich men might not know too many beautiful poor women, this itself seems to require explanation; if this was as good a deal as the hypergamists think, they would actively take steps to find them, or there would be social institutions to make such matches happen. Also, the rise of online dating makes it trivial to meet people outside your social class, but it seems to produce the same kind of class-matched couples as offline dating did. Also, rich people meet poor people all the time. Poor people are their secretaries, servants, waitresses, and Uber drivers. Sometimes they have casual sex with these people. They just don’t (usually) marry them. I think it’s choice.

I worry the economists (and PUAs, and social critics, and other people who talk about hypergamy) have screwed up by reducing “class” to “income”. True, men are usually the breadwinners and won’t need their wives to support them financially. But whether it’s nature or nurture, high-status successful women tend to raise high-status successful children; men know this, which incentivizes them to seek high-status successful wives regardless of their financial situation. Also, men have to live with their wives. They want someone who shares their norms and values. For upper-class men, that means upper-class women. So men have strong reasons not to “marry down” regardless of income. And even though women do care about men’s ability to provide, they’re thinking about these things too.

VII. Are Hypergamists Happy?

Do marriages go better when husbands are higher-status than wives?

This answer is the same as all the others: no for education, yes for income.

I’d heard that if husbands married up educationally, they were more likely to get divorced later. That was true in the 1970s, but not anymore. Esteve et al:

In societies in which wives have more education or earn more than their husbands, do relationships suffer? Evidence from the United States speaks against these fears. Prior to the 1980s when men clearly had more education than women and hypergamy was normative, men who married women with more education were more likely to divorce. However, as the situation reversed and wives now have more education than their husbands, the association between wives’ educational advantage and divorce has disappeared. Among marriages formed since the 1990s, wives with more education than their husbands are now no more likely to divorce than other couples (Schwartz and Han 2014). A similar trend is observed for couples in which women out-earn their husbands (Schwartz and Gonalons-Pons 2015). This evidence is consistent with the notion that, at least in the United States, couples have adapted to the changing realities of the marriage market and are no longer at increased risk of divorce when women have the educational or earnings advantage. A recent study for marriages formed in Belgium in the 1990s found that those where the husband has more education than the wife are more likely to dissolve than marriages in which the wife has the educational advantage. In line with the American findings, the same study also found that the latter type of marriage is relative more stable in regions and municipalities where they are more common (Theunis et al. 2015).

My interpretation is that there’s nothing inherent about educationally hypo-gamous marriages that makes divorce more likely. But when hypo-gamous marriages are weird, only weird people will marry hypo-gamously, and weird people are more likely to divorce. This is a constant pattern across sociology. There is a stigmatized thing. Statistics show the stigmatized thing has bad outcomes. Then the thing becomes unstigmatized, normal people start doing it, and then the outcomes are fine.

But wives earning more than their husbands is still weird, and still (on average) goes less well. Bertrand, Kamenica, and Pan discuss various negative consequences. I especially like this graph:

…which shows a “cliff” between the wife earning very slightly less than her husband, and very slightly more. Some of these marriages either ended in divorce, or never happened in the first place. The study claims that 23% of the decline in marriage comes from women being less able to find higher-earning men and unable to tolerate lower-earning ones, though I can’t begin to evaluate whether that’s true. Here is a funny article about how it’s hard to study this effect, because when wives earn more than their husbands, both partners lie about their incomes to deny this.

What about looks?

Most of the studies I found were from one team in Florida which puts a lot of effort into showing why everyone else who thinks differently is wrong. I don’t know this team and I don’t know whether to trust their results, but they find pretty conclusively that marriages where the wife is more attractive than the husband are happier (see also here). In these marriages, both the husband and the wife are nicer to each other than in the reverse scenario. Attractive people are no happier in their marriage than unattractive ones overall; it only matters that the wife is more attractive than her husband.

I conclude that the Taliban is throwing acid at the wrong gender. If we just disfigure all the men, everyone can be happy at the same time¹⁰!

VIII. What About The Gays?

So far this has been pretty heteronormative. I originally thought it wouldn’t make sense to talk about homosexual hypergamy, since there’s no way to mark which partner marries up vs. down. But this paper on “Gay Hypergamy And Rainbow Kinship” manages to be mildly interesting anyway.

It points out that gays (especially gay men) are more likely to have wildly hypergamous marriages - ie actually marry someone from a very different social class - compared to straights. I can’t find direct data on class, but here’s showing that 9% of straight relationships are inter-racial, compared to 21% of gays, and 17% of lesbians¹¹, which seems to back up the idea of less sorting.

I mentioned before that I never see a rich male Google programmer dating a stunning woman from the ghetto. But I have heard of gay relationships like this. Why? Commenters suggest that gays mostly meet their partners through “the gay community”, which takes a cross-section of society through a direction mostly uncorrelated with race and class. ¹²

And for a laugh, compare the way academic papers talk about hypergamy in straight couples (source):

Female hypergamy can only be understood in the context of a male-dominated society, being both a consequence of this norm and a vehicle for its perpetuation. In a model where women are generally in a position of inferiority, the reproduction of the social order requires that male dominance be respected within the family, the central institution for the interiorization of gender roles. Under this rationale, men choose wives of lower social status than themselves, and vice-versa. Such behaviours reflect the interiorization of dominant norms, the desire to avoid social disapproval – either explicit or implicit – or simply to forestall the negative feedback to which couples who violate these norms are exposed . Male superiority within the couple also ensures that, in most social situations, the roles assigned to man and woman are spontaneously assumed, thereby consolidating gender stereotypes in the face of realities that may challenge them.

Versus hypergamy in gay couples (source):

Gay male sexual cruising serves, I suggest, as an underappreciated cultural resource for the creative construction of those ‘families of choice’ and ‘invincible communities’ that scholars have identified as the distinctive character of non-heterosexual family and kinship formations. In particular, the unfettered pursuit of masculine sexuality facilitates opportunities for individual social mobility and for forging rainbow kinship ties that have not yet attracted much attention from scholars or activists.

IX. Conclusions

Educational hypergamy has gone into reverse. Now that women dominate education, they’re actively seeking less educated men, and vice versa. This seems to be because educational imbalances in favor of women have become normative; education is now a “proper” “feminine” trait.

In contrast, income hypergamy is still widespread, important, and causing problems for non-compliers. Is the norm weakening over time? It’s hard to tell.

Despite this, men and women display an equal and stunning degree of class homogamy. Men may use their class-based market value to purchase a little more education in a mate, and women to purchase a little more income, but both genders consider class first and foremost.

Looks don’t seem to figure into this at all. There’s not much trade of better looks for higher income. Instead, each quadrant in the (rich, poor) x (pretty, ugly) matrix pairs off with itself.

Gay people of both sexes sort on class less heavily than straight people.

Women’s rising share of education isn’t directly damaging the marriage market. Women’s rising share of income might be, with one study suggesting it’s responsible for 23% of the decline in US marriages. By analogy to education, it’s possible that if women ever earned more than men, society would switch to accepting this and marriages would happen as normal. But as long as this is still uncommon, the norm against it persists and women who earn too much have a hard time.

Finally, a practical question: to maximize your odds of getting a desirable spouse, should you make more money or less? For men this is easy: earn more. For women, it’s a harder question; earning more raises your status (which ought to get you a higher-status man), but also decreases your chances with men who make less than you. This study says that “income is not associated with the probability of marriage for women”, and it seems more likely to get you a better partner than a worse partner, so probably you should go ahead and get rich. But it’s possible that income is partly serving as a proxy for class, and on a causal level income has some totally different effect. So this one could still go either way.