The Lemon Market: Dating

Start with who's on the apps. Among 18-to-29-year-olds using them, 73% are men (Pew, 2022). That's 2.7 men for every woman.

Over a whole year the ratio is gentler, about 1.5 to 1, because women try the apps and leave faster. But at any moment, the men far outnumber the women.

And most of the women on the apps aren't really there to date. About 38% of single women 22-35 used a dating app in the past year, but only about one in five of those actually went on dates from it: 7.8% of single women. The rest browse, swipe and chat, for the ego boost or in case someone great turns up.

Of the ones who are dating, 75% spend months of that year on at least one man who was only ever there for something casual. And about 18% to 41% of them pair off on a path to marriage. The apps look like a market of millions of eligible women; in a year it's about 1.2 million who actually date, each shown more men than she could ever see.

Women's likes follow a power law, and the median woman likes very few men: only 4.5% of the profiles she's shown. The median man likes 26%.

Put those together with the unbalanced gender ratio and the average woman gets about a hundred likes a day. If she has time for five conversations, she can answer the top 5%. Any woman with strict filters ("only men 6 feet or above with grad degrees") set them purely to cut down the flood. Her standards didn't get crazy for no reason; they reflect the overwhelming amount of attention women get on the apps.

This is what it looks like from the men's side. The median man is liked by about 5% of the women who see him. A long right tail of men is liked by 20%, 30%, more.

The tail falls off like a power law: the density drops with roughly the 2.2 power of the like rate.

For women the same chart is a broad hump centered on 38%. Men like broadly; the median woman is liked by more than a third of the men who see her.

On Hinge, the top 5% of men get 41% of all likes, the top 10% get 58%, and the bottom half of men share 4%.

About half of the inequality is taste; half is the feed showing popular profiles more. If every profile were shown equally, the top 5% of men would get 18% of likes. Offline, the same shape holds: among men 25-45, the top 5% by lifetime partners account for 40% of all partners (GSS).

Start from the end of the funnel. In 2022-23, 47% of straight single men aged 18-35 had no sexual partner at all in the past year, up from 28% in 2017-19.

Only 6.1% had sex with anyone they met online. Those who did averaged 3.4 partners, against about one for everyone else.

Work backwards and the funnel falls out. Each woman who is actively dating has about a dozen first dates a year and gives them to the best-looking of her matches. But only about 1 in 5 women on the apps is actually dating in a given year; the rest browse, swipe and chat.

The result: 60% of men on the apps get no first date in a year. For the bottom half of men by appeal it's 84%.

Calibrated so the model reproduces NSFG: women with any app sex average 2.8 partners, men 3.4.

The dates that do happen go to the top. The top 10% of men on the apps get 66% of all first dates, and the top 20% get 82%.

That's the experience of most men on the apps: not rejection after dates, but no dates to be rejected from.

For a sense of the best case: a man above average in looks, a founder with a name-brand degree, in New York. Over three years he received 2,031 likes and matched with 345 women. That became 57 meetup attempts, about 19 a year (Lana Li).

He is in the best market that exists for a man looking for educated women in their twenties, he is a top-tier guy, and he still got barely any dates. One every other week! Now adjust that down: most men are not in strongly gender-biased markets, and have less going for them. What is a realistic expectation there? Let's find out in the next section.

Desire for something casual scales directly with attractiveness. This is based on data from a real dating app, and it shows that the more attractive women rated a man, the more likely he was on the app for something casual: about 23% of the least attractive men, 52% of the most attractive.

You don't need hot men to be different people to get this. Serious men pair off when they meet someone they want; casual men mostly don't, and are the ones still on the app. The more matches a man gets, the faster a serious one gets taken off the market, and as we've seen, the most attractive men get most of the matches. That cashes out to the men on the app being fairly adversely selected in terms of what women want, and more adversely selected the more appealing they are.

Even the serious ones rarely commit, because abundance raises the bar. A man who can open the app and find someone new in an hour has every reason to wait for someone better. Chipped nail polish, doesn't go to the gym enough, doesn't know who Heidegger is: out.

A 90th-percentile man would commit to a median woman he has dated for months only about 0.4% of the time. A median man would, 47% of the time. And even when both would, it still has to work out (timing, chemistry, everything else), which in the model happens about 77% of the time.

People commit to partners near their own level. The model is calibrated so couples sort as tightly as real ones: husbands' status and their wives' families' status correlate 0.76 (PSID, 17,000 couple-years). It also matches who never marries (59% of the lowest-earning fifth of men in their forties have ever married, against 91% of the top fifth), how fast single women marry (census), and how often the most attractive women and men marry: about 90% by their late thirties, the same for both sexes (Add Health).

So he goes on first date after first date. If each woman had a 40% chance of being the one, a man would be off the market within a handful of dates. In the model, a first date ends in a relationship both people commit to about 4% of the time for a man between the 50th and 75th percentile of looks, and 1% for the top 5%.

After twenty first dates, most men are still single, and the more desirable the man, the likelier he is.

Men of every age, from 20 to 50, find women in their early twenties the most attractive (OkCupid). Interest in women peaks at 22. By 30 it's at about 65% of the peak, by 35 38%, by 40 23%.

This is the Wall, measured. Men's curve peaks later, at 26, and falls more slowly.

The two things men respond to first line up with fertility. Counting from 20, a woman has used about 69% of her lifetime fecundity by 30 and 87% by 35 (Geruso et al. 2023).

And waist-to-hip ratio, the other strong signal, goes the same way. Only about 5% of women at 20 have a WHR of 0.74 or lower; by 30 it's 2%, by 40 about 1%. Nearly half of women in their twenties are already at 0.85 or above.

For men the clock runs through selection. Each year, the men women want most get taken first. Among never-married men, the casual share rises from 23% at 25 to about 56% at 45, and their average value as partners falls from the 45th to the 27th percentile.

The census agrees: never-married men in their forties earn a median $36k against $67k for men who married, and fewer have degrees. A man in his forties who has never married usually has a reason, and women know it.

Put her clock and his selection together. A median woman's odds of a man who commits within five years fall from 41% if she starts at 22 to 28% at 30 and 21% at 36. Her odds of a man as good as her or better fall much faster: from 34% to 11% and 4%. The clock takes who even more than whether.

Five years on the apps, open to men up to two years older, two men properly dated a year. Fitted to the census rate at which single women marry from 25 to 30; in their thirties the model runs below the census, since real women that age also meet men through friends, work and remarriage.