Is Manhattan pricing out the young? What 14 years of data show

The Census Bureau says Manhattan gains more young people, not fewer

The Census Bureau's Population Estimates are not a survey. They start from the last census and roll it forward with birth certificates, death certificates, Medicare enrollment and IRS address changes. They give a count of Manhattan residents by age for every July. So I took everyone aged 20 to 24 in one July, found the same people aged 25 to 29 five Julys later, and treated the difference as net migration. Nobody dies at 24 in numbers that matter here.

The result pointed the wrong way. Manhattan's 20 to 24 year olds gained about 56,900 net arrivals over 2010 to 2015, 69,500 over 2015 to 2020 and 71,600 over 2020 to 2025. Citywide the same cohort went 137,100, 142,900, 167,400. Every five-year window in this series adds more young people than the one before.

Split by sex, the same thing. Manhattan's net gain is nearly equal for men and women, women outnumber men citywide, and neither line has a window with a smaller gain than the one before it.

Then I tried to join the 2010s series to the 2020s series and found the seam. The 2010-based estimates had walked Manhattan's 20 to 24 year olds down from 141,600 to 110,400 over the decade. The 2020 census counted them and found 135,900. The Bureau did not revise the old series. It just started a new one at the new number and let the two sit in different files.

There is no way to reconcile those lines. If the 2020 census is right, the 2010s series lost about 25,000 young adults somewhere, and every net migration figure from that decade is too low. If the 2010s series is right, the 2020 census found 25,000 people who were not there. The Bureau's own documentation acknowledges the break and offers nothing to bridge it. I kept the two series apart and stopped trusting any decade-long comparison of levels.

Net migration is arrivals minus departures. A borough where 100,000 young people arrive and 65,000 leave shows the same +35,000 as one where 40,000 arrive and 5,000 leave. This method cannot tell those apart, and my question is about arrivals. So the estimates could not prove my claim, but they could not kill it either.

The IRS says 20% fewer by 2020, then changes how it counts

The IRS knows where every tax filer lived last year and where they live now, because both addresses are on consecutive returns. The Statistics of Income division publishes those moves as county-to-county flows, with the number of returns, the number of people on them and their total adjusted gross income. It does not publish age below the state level. But it covers roughly nine in ten adults and asks nobody to fill in a survey, so it is the closest thing to a turnstile at the Lincoln Tunnel.

Returns arriving in Manhattan from outside the five boroughs fell from about 49,200 in the 2011 to 2012 filing year to 39,600 in 2019 to 2020. That is a 20% drop before the pandemic touched anything. Citywide, 18%. And in the state-level file, which does carry age, filers under 26 arriving anywhere in New York State fell 20% over the same years. Three numbers, one direction, and the claim was alive again.

Then the series turns straight up. 42,300 returns in 2020 to 2021, 59,300 the next year, 65,500 in 2022 to 2023, the highest in the file. For a while I believed all of it was the post-pandemic rush, and the 2021 to 2022 jump may be. The last point is different. Then I read the notes. For 2022 to 2023 the IRS writes that the data “begins a new series” built with enhanced address matching, and asks users not to compare it with earlier years.

You can see what the new matching did in the national numbers. Filers under 26 moving between states jumped about 40% in a single year, nationwide, with no event to explain it. Young people did not suddenly discover moving in 2022. The IRS discovered young people. So the IRS series is a clean 20% decline through 2020 and then a wall, with everything after the wall measured in different units.

The $50,000 line, and why it moves

The IRS does publish income bands, and the low band is under $50,000 of adjusted gross income. Among filers under 26 arriving in New York State, the share below that line fell from 89% in 2012 to 72% in 2020. Read quickly, that is the whole thesis: the young arrivals got rich.

Read slowly, it is inflation. The bands are in nominal dollars. $50,000 in 2012 buys what about $66,000 buys now, so a fixed line lets a steady stream of identical people appear to climb out of the bottom band. The IRS does not publish where inside a band people sit, so the line cannot be deflated. The only fix I could find was a control group.

Inflation and wage growth hit all four rows. The gap between rows is what is left. Young arrivals to New York fell 7 points further than the young people already living there. In every other state combined, arrivals fell only 2.5 points further than residents. That extra gap is specific to New York and does not come from the dollar. The count behind it: filers under 26 arriving in New York with AGI under $50,000 fell 35% from 2012 to 2020, against 20% for all young arrivals. Hold on to that 7-point gap. The survey in the next section, which looks at Manhattan rather than the whole state, does not find it.

Adjusted gross income is a tax return, not a person. A 24-year-old on a parent's return does not appear. Two roommates filing separately are two returns; a married couple is one. And the age split exists only for New York State, so this table says nothing about Manhattan specifically.

The survey nobody fills in

The American Community Survey is the only source with age, sex, income, last year's address and neighborhood in one record. It is also a survey, mailed to about one household in a hundred, legally mandatory, with no penalty anyone has ever paid. Housing-unit response ran around 97% in 2011 and about 85% by 2023, and the drop is concentrated in young renters in big cities. So the group I care about most is the group most likely to have thrown the envelope out.

I pulled every New York State respondent aged 20 to 29 for 2011 through 2024, except 2020, which the Bureau did not publish because too few people answered during the pandemic. A person counts as an arrival if they lived abroad a year ago, or lived in the United States outside the five boroughs. Each record comes with 80 replicate weights, which is how the Bureau lets you compute a margin of error on a slice this thin.

Citywide, young arrivals went from about 111,900 in 2011 to 109,000 in 2024. The margins of error are around 9,000 each, so that is no change. Manhattan went from 61,100 to 52,000, margins around 7,000 and 5,000, which is a 15% decline you can just about defend and a reviewer could just about attack. This does not match the IRS, which had citywide returns down 18% by 2020. Different years, different unit, different source; the survey runs four more years, through the post-pandemic rebound the IRS could not measure cleanly. The 2013 and 2018 dips are as large as the whole 2011 to 2024 change. This is the shape of a survey of a few hundred people per year, not of a trend.

By sex, women arrive in larger numbers than men every year, about 60% of arrivals in Manhattan. Women's arrivals fell into 2019 and came most of the way back by 2024. Men's barely moved. Manhattan's 2011 to 2024 decline sits almost entirely on the women's line, 38,700 to 31,800, and the margins on those two points just touch.

The income split is where the claim finally holds. Arrivals to Manhattan earning under $50,000 in 2024 dollars fell from about 36,700 to 25,900, a 29% drop, and the margins do not overlap. Citywide, 77,200 to 62,300, down 19%. Fewer broke young people arrive in Manhattan than in 2011. That sentence is true, and it survives the error bars.

What it does not survive is the grey line. Among Manhattan 20 to 29 year olds who lived in the same home a year earlier, the share under $50,000 fell from 60% to 47%. Among new arrivals, 60% to 50%. The people who moved in and the people who stayed put have almost the same income mix in 2011 and almost the same in 2024. Nothing at the door is filtering out poorer arrivals. Young Manhattan as a whole got richer, and the arrivals came along with it.

That is not what the IRS table said. There, young arrivals to New York State pulled 7 points ahead of young residents; here, Manhattan arrivals and residents move together. The two are measuring different things: the whole state against one borough, tax returns against people, adjusted gross income against personal income, under 26 against 20 to 29. I cannot make them agree, and I am not going to pick the one that fits the headline. The Manhattan question is the survey's to answer, because the IRS has no age below the state level.

$50,000 in constant dollars is still a line I drew. It corrects for prices, not for the fact that a 25-year-old at the 30th percentile earns more real money now than in 2011. The version of this analysis that draws the line at the national median income for 20 to 29 year olds each year is in the repo as a work in progress, and I am not quoting numbers from it until it is finished.

Where they went then, and where they go now

The finest geography the survey allows is the Public Use Microdata Area, which in New York maps onto one or two community districts. Not a ZIP code, not a corner. The boundaries were redrawn in 2022, so I averaged three years at each end and pooled the Midtown districts, which the two maps split differently.

Midtown and Chelsea take more young arrivals than the next two districts combined, in both periods, and the share of them earning under $50,000 fell 15 points, the largest drop in Manhattan. Williamsburg and Bushwick were not in the top ten in 2011 and are now, and Williamsburg's arrivals are the richest of any district on the list. Park Slope and Washington Heights fell off. Morningside and Hamilton Heights lost a quarter of their arrivals and stayed the poorest destination in the top ten, which suggests the students and the people around Columbia are still coming, and fewer other people are.

Outside the top ten the growth is in central Brooklyn and Harlem. Bedford-Stuyvesant roughly doubled its young arrivals, Crown Heights North roughly doubled, Harlem went from 1,500 a year to 2,700. Each of those rows has a margin of error that could swallow half the change. Together, the pattern across eight or nine districts is harder to wave away than any one of them.

What I would expect you to object to

• The survey undercounts exactly the people you are counting. Yes. Weighting pins the total number of young people to the Population Estimates, so the level is anchored. The share who moved comes from whoever answered, and if movers answer less than stayers, arrivals are understated, probably more in 2024 than in 2011. That bias makes the decline look bigger than it is, not smaller.
• Tax returns are not people. A return is a filing unit. The IRS files also count individuals on returns, and those move the same way, but a 23-year-old on a parent's return in Ohio who moves to Bushwick and files alone shows up as a new filer, not a migrant.
• $50,000 is arbitrary. It is. It is also the IRS band edge and a round number people recognize. The percentile version is the fix and it is not done.
• Manhattan is not New York. Correct, which is why both are in every chart. The story is different in Brooklyn, where arrivals grew.
• Net migration went up, so your premise is wrong. On the Census estimates, more young people are added to Manhattan each year than in the 2010s. That can be true at the same time as fewer arriving, if fewer are leaving. The three sources are measuring different doors.
• The title asks about pricing out and the ending says Manhattan is not turning the poor away. Both stand. Fewer low earners arrive, and the ones already here also earn more. Whether that is rents pushing people to Brooklyn or wages rising for everyone is the question the data leave open.

Where I ended up

The claim I started with was that fewer young people move to New York. The evidence for that is a 20% fall in IRS arrivals that stops dead in 2020 when the IRS changed its method, and a Manhattan survey decline that clears its margin of error by a hair. Citywide the 2011 and 2024 figures are 111,900 and 109,000, inside a 9,000 margin, and I would not print that headline.

The claim that fewer broke young people move to Manhattan is true on every source that can test it, and it means less than it sounds like it does. On the one source that can see Manhattan, the young people who were already here got richer by nearly the same amount; the state-level IRS file says arrivals got richer faster, and the two do not measure the same people. Manhattan did not start turning away the poor at the tunnel. It stopped producing them, or it priced the ones it had out to Bushwick and Bed-Stuy, where the arrivals are up and the incomes are lower. Which of those it is, no federal dataset I opened can say, and I think the people who talk about this most confidently are working from the same three anecdotes I was.

Every number, the code that produced it and the footnotes I tripped over are in the repository linked below, under a license that asks only for a link back here. The Census API key is free and the IRS files are public, so anyone can rerun it, and when the 2025 survey lands in October I will.

At a glance

Sources

Census PEP, IRS SOI, ACS PUMS

Years

2011 to 2024 (Census estimates run 2010 to 2025)

Who

Ages 20 to 29 in the survey; 20 to 24 in the Census cohort; under 26 in the IRS file

Geography

NYC, Manhattan, 54 community districts

Code

Python, pandas, matplotlib

Cost

$0 in data; a free Census API key