How When Harry Met Sally Explains Inequality

A new study says that educated people marrying each other has increased inequality by 25 percent.

Wikimedia Commons

Inequality has exploded the past 30 years, because of the usual suspects: technology, Wall Street, Harry, and Sally.

Okay, it probably isn't fair to blame Billy Crystal and Meg Ryan for our widening income gap. But it is fair to say that When Harry Met Sally tells us something about why the rich have been getting so much richer than everyone else. That's high-earning college grads marrying each other—which a new paper estimates has increased inequality by 25 percent.

We used to live in a Mad Men world. Few men went to college, even fewer women did, and not many women, period, worked outside the home—not that they had many opportunities if they did. In 1960, 42.5 percent of married women hadn't graduated from high school. 39.6 percent had only done that. And just 37.7 percent of all women had or were looking for a job. The stereotype is that men married their secretaries, if their wives did work, and there's something to that—though highly-educated people did still pair up at high rates back then.

You can see that in the chart below, which breaks down 1960 marriages by each partner's education. It shows how much more or less likely a marriage was between, say, two college grads than what you would've expected from random chance. And it shows that for every combination of education level: people who didn't finish high school (HS-), high school grads (HS), people who attended some college (C-), college grads (C), and people with advanced degrees (C+). Anything more than 1 means there were disproportionately more of that type of marriage; anything less means the opposite.

(Note: I've highlighted all the homogenous pairs—HS- and HS-, C and C, etc.—with dotted lines).

All the data comes from a new paper that looks at how much people marrying someone like themselves—what economists charmingly call "assortative mating"—has contributed to inequality. The answer for 1960: not much.

Sure, people with at least a college degree were much more likely to marry each other than by chance. But this was a small group. It only made up 4.5 percent of all marriages. And it wasn't as if couples with two college degrees had two high incomes. Even highly-educated women didn't have many opportunities to earn a lot if they did work outside the home (which many didn't). In other words, most households still only had one big earner, relatively-speaking, regardless of the wife's education.

That made marriage a different institution back then. Men weren't looking for a soulmate, let alone someone to help pay the bills. They were looking for a good wife and mother. So they were willing to marry "down." Now, it doesn't look like that if you focus on how disproportionately likely college-educated men and women were to trade I-dos. But take another look, this time at the shaded areas. Men with a college degree were about as likely to marry women with only a high school degree as you'd expect by chance. And men with a college degree or more were twice as likely to marry a woman who hadn't finished college. This meant there were almost twice as many college-educated men married to women who weren't than were for the simple reason that many more women hadn't graduated from college than had.

I'll Have the Marriage My Classmate Is Having

That world is gone now. It's gone, because more people, especially women, go to college. Because women have the chance to work—and earn—more. Because couples need two incomes to afford the same lifestyle that one used to buy. Because people who graduate from college cluster in neighborhoods away from those who don't. And because people want more from marriage nowadays.

Among affluent marriages, it's a When Harry Met Sally world, where working men and working women of similar education and income levels bond and marry over similarities—movies, art, paprikash—rather than the differences that defined traditional marriage (he works/she cleans). This new arrangement promises a better world, but it has increased inequality. Here's how.

Five big trends have transformed the home and the workplace the past few decades. Those are more education, more women in the workforce, more financial insecurity, more marital expectations, and more assortative mating. Let's consider them one-by-one.

1) More education. In 1959, only  percent of 14.8 men and 7.6 percent of women between the ages of 25 and 29 had graduated from college. In 2012, it was 29.8 and 37.2 percent. And by 2011, women had surpassed men in advanced degrees too.

2) More women in the workforce. In 1959, 37.2 percent of women had a job or were actively looking for one. In 2012, it was 57.7 percent. That said, it's still hard for women to lean in when they have to plan their careers around when they have kids, when daycare is so damn expensive, and when husbands don't always do their share of the housework. That's why men still dominate the highest-paying jobs that require an inhuman number of hours as a show of your semi-feudal obligation to work. And even if women do break through, they still don't get paid as much as men do for the same work—though, as Hanna Rosin points out, the gap is more like 9 cents for every dollar men make, and not 23 cents.

3) More financial insecurity. You used to be able to afford a solidly middle-class lifestyle with just a high school diploma. That seems inconceivable today, when two college degrees aren't always enough, depending on how pricy your zip code is. But it's true. In 1960, a married couple of high school graduates made 103.1 percent of the average income. In 2005, that same couple would have made 82.7 percent of the average—and that's despite women working more. College-educated couples haven't fallen back, but outsized inflation in things like housing, healthcare, and college tuition have made staying ahead harder than getting ahead.

4) More marital expectations. Men don't just want a homemaker, and women don't just want a good provider. Men and women both want a good partner. That means someone to help with the mortgage, the homework, the housework, and watch House of Cards with. Someone ... like themselves.

5) More assortative mating. Take a look below at the same chart breaking down marriage by each partner's education, but updated for 2005. The first thing you'll notice is how much more likely high school dropouts are to marry each other than by chance. But look again. Every group is disproportionately more likely to marry someone like themselves than any other.

Now, it sure doesn't look like highly-educated people are more likely to marry each other today than before. But that's just because there's more of them. In 1960, there were so few people with college, let alone advanced, degrees that any marriages between them would be disproportionate. Today, higher ed is more democratized, but marriages aren't. The opposite, actually. You can see that in the shaded parts above—college-educated men are much less likely to marry women who aren't than they were before. By 2005, someone who had at least graduated from college was more likely to marry someone else who had than not: 24.2 of married couples both had college degrees, while 22.2 percent had a single partner with one.


It's a classic story. Boy meets girl in college. Girl dislikes boy, but then asks to be friends five years later. Boy says no, but changes his mind mind after they both go through bad breakups. Girl and boy start falling for each other. Boy freaks out after they finally act on their sexual tension. Girl doesn't want to see boy anymore. But then boy confesses his love, and they live happily ever after—with two nice incomes.

Marriage has changed, because women's opportunities have changed. Women graduate more, they work more, and they earn more than they used to. These are all good things. But marriage has also changed, because people want new things from it. Men don't want a homemaker, and women don't want a provider. Men and women both want a partner, someone who can help with their emotional and financial needs. So they wait until they've settled into their careers to tie the knot, and they try to find someone who's doing the same. This is also a good thing. 

But the consequence of all these good things is more inequality. College grads were already leaving everyone else behind as robots, globalization, and de-unionization have eaten away at less-skilled workers' wages. But now college grads are marrying each other at higher and higher rates. The researchers estimate that assortative mating has increased our Gini coefficient—where 0 is perfect equality and 1 is perfect inequality—from 0.35 to 0.43 compared to a world where we married like we did in 1960.

So what? It's not like we should worry about women working more, and people marrying who they love. And it's not like assortative mating is why the super-rich have turned today into a new Gilded Age. Inequality is a fractal: The top 0.01 percent are to the top 0.1 percent what the top 0.1 percent are to the top 1 percent what the top 1 percent are to top 10 percent what the top 10 percent are to the bottom 90 percent. Corporate lawyers and consultants getting hitched might explain the rise of the top 95 to 99 percent, but it doesn't explain the top 1 percent—let alone the dynastic incomes above that.

Inequality is a complicated, complicated thing. There are plenty of usual, and some unusual, suspects that have been steadily pulling our income gap further and further apart. It's hard to say which are causes and which are effects, or if that distinction even matters. But there is one thing we can say for sure.

Harry and Sally are a cute couple, but they should probably be paying higher marginal tax rates.