Every few months someone with a large salary tells a reporter they cannot afford New York, and the replies split into two camps that never meet. One camp says the housing crisis has reached the professional class. The other says a six-figure household complaining about rent has chosen the wrong neighbourhood. Both camps argue from anecdotes, because neither has the number they would need: how many of these households actually exist, and what they have in common.
This article counts them, and then follows the count into the places it leads. The short version is that the burdened six-figure renter is real, uncommon, and mostly a story about when the lease was signed. The numbers are below for you to try before the story starts; each question mark explains what a control or a figure means.
Rent-burdened share by move-in period, under the filters above
Loading the synthetic population…
Household type
The census adds up the income of everyone in an apartment, related or not. Families and singles are family households plus people living alone: the households where a person or a family earns six figures. Roommate households are unrelated adults sharing a unit, which also catches unmarried partners because the census cannot tell the two apart.
The article's main subject is families and singles; roommate households reach six figures by pooling, and only a third of the burdened ones contain anyone who personally earns $100,000.
Household income
Total income of everyone in the household, in constant 2024 dollars. "Six figures" means $100,000 and over. The bands are the ones the release's verification targets use, so each can be checked against the census.
Borough
Where the apartment is. The Bronx and Staten Island are combined because six-figure renters are few there. Two thirds of burdened six-figure families and singles live in Manhattan.
Moved in
How long ago the householder moved into the unit, in the census's seven bands. Rents are set at signing and then drift up more slowly than the market, so a household's rent is partly a record of when it arrived. This is the filter that moves the burdened share most.
If every rent changed by…
A what-if. Each household's rent is scaled by the chosen percentage with its income held fixed, and the burdened share is recomputed. It shows how much of the count is the 30% threshold itself: many households sit just under the line.
It is not a forecast. Real rent changes are uneven, and stabilised leases cannot rise this way.
Rent-burdened
The share of the selected households whose gross rent (contract rent plus separately paid utilities) is 30% or more of household income, the census's own definition. Half of income or more is "severely burdened" and is close to nonexistent above $100,000.
This is the synthetic population's figure. For the article's headline groups the census gives the same answer to within a point; the methods paper sets them side by side.
Households, scaled to the city
The synthetic population holds about one household for every 41 real NYC renter households. The count of synthetic households in the selection is multiplied by that ratio. Treat it as an order of magnitude when the selection is small.
Median gross rent
The middle household's monthly gross rent in the selection, in 2024 dollars, read off a histogram in $250 steps. For long-tenured households it lands near $1,500, which is the city's median rent-stabilised rent.
People per household
Average number of people living in the unit. Recent arrivals are mostly one or two people; households ten years in are typically families of three.
Median householder age
The age of the person the census records as the householder, in five-year steps. It rises from the early thirties among the newest arrivals to the late sixties among households thirty years in.
Pre-1970 buildings of 5+ units
The census does not record whether a lease is rent-stabilised. Stabilisation covers buildings of six or more units built before 1974, and the closest cut the census allows is five or more units built before 1970. This is a proxy for the stabilised stock, not a count of stabilised leases.
The staircase
Each bar is the rent-burdened share for one move-in period, under the type, income and borough filters. Choosing a move-in period highlights its bar rather than hiding the others, so the slope is always visible. The rent slider shifts every bar.
Too few households
Fewer than 30 synthetic households match, so the numbers would be noise. Widen a filter.
Start with the count
“Six figures” here means a household income of $100,000 or more, and the first thing to decide is whose income. The census adds up everyone living in an apartment, related or not, so three graduates earning $60,000 each count as one $180,000 household. That is not what most people mean by a six-figure renter. So this article follows the census’s own split: families and people living alone on one side, and Roommate householdThe census category “nonfamily household, householder not living alone”: two or more unrelated adults sharing a unit. Unmarried partners fall in here too, since the census cannot tell a partner from a roommate at this level, so the count is an upper bound on roommates in the everyday sense. on the other. The families and singles are the subject; the roommates get their own section near the end.
About 68,000 families and single people in New York City with six-figure incomes pay 30% or more of their income in rent. That is a census figure, tabulated from the anonymised survey records the Bureau publishes, and its margin, from the Bureau’s Margins of error from replicate weightsEach PUMS record carries 80 replicate weights alongside its main weight. A statistic is recomputed 80 times, once with each replicate, and the spread of those 80 answers around the main answer gives its standard error: variance = 4/80 × Σ(θr − θ)². The Census Bureau reports 90% margins, 1.645 standard errors, and every census margin in this article follows that convention.This is the Bureau’s own method (successive difference replication), so a margin here matches what the Bureau would publish for the same cut., is about 3,000 either way. No official table prints this cut; the number was always in the Census microdata (ACS PUMS)The Public Use Microdata Sample is the Census Bureau’s anonymised record-level release of the American Community Survey: one row per household and one per person, with a weight saying how many real households each record stands for. Geography is coarse (areas of about 100,000 people) and a few values are top-coded, but every variable for a household arrives together, so cross-tabulations the published tables never print can be computed directly.The 2020–24 five-year file blends five survey years; dollars are adjusted to 2024. and just had to be asked for.
The second thing to know is how small that is. The city has about 1.1 million Rent-burdenedGross rent as a share of household income, using the census’s own variable. Paying 30% or more is “burdened”; 50% or more is “severely burdened”. Households with zero or negative income have no defined ratio and are excluded, as the census does.The 30% line descends from the 25% rule in the 1969 Brooke Amendment for public housing, raised to 30% in 1981. It was never designed for households earning $120,000, which is why the article also reports what is left after rent. renter households, and the six-figure families and singles are 6% of them. Every other burdened renter earns less, and most earn much less. Figure 1 shows where they sit.
Reading guide
How to read it
Each grey bar is the number of renter households in that income band, in thousands, from the census microdata. The blue portion is how many of them pay 30% or more of income in rent. The label gives both numbers.
What to notice
The three bottom bars together hold about 93,000 burdened households, of which 68,000 are families and singles and 26,000 are roommate households. The top bar alone holds 563,000. The burdened share falls from 88% to 4% as you go down the chart.
What it cannot show
Income here is household income, everyone in the apartment added together. The article separates roommate households from families and singles in every later figure.
The dashboard above runs on the SparkyData The synthetic populationSparkyData’s New York metro release samples whole households from the census microdata, re-weights them to published totals by region, tenure, household size, income and age, and layers on attributes the census does not ask about. No record is a real person, but each household’s income, rent, size, borough, move-in year and members’ earnings came together from one real record, so those relationships are inherited, not modelled.This article uses only inherited variables. The modelled wealth layer was deliberately left out; the notebook explains why., which is built from the same census records, so it is fair to ask what it adds. For the count, nothing, and it should not pretend otherwise. Its job is to be the database the story is read from, with people and their earnings attached to households, their rents and their buildings, and to prove it can reproduce what the census says before it is trusted on anything else. It does: every one of the fourteen The held-out reproduction checkThe rent-burden statistics were computed from the census microdata after the synthetic population was built and were never used to calibrate it. Reproducing them shows that re-weighting to the published margins did not distort the relationship between rent and income.It is not an out-of-sample test: the synthetic households are sampled from the same microdata, so agreement is expected. It is a check that nothing broke, and it should not be read as more than that. rent-burden checks the release runs lands inside its What “within band” meansEach check has a band that combines two uncertainties: the census margin of error on the published value, and the sampling error a synthetic population of this size carries on the same statistic. A miss is a difference larger than the band. At 500,000 people the second term is small, so the bands are close to the census margins themselves.. Three are below; the full table ships with the article.
| Statistic (NYC renters) | Census | Synthetic | Band |
|---|---|---|---|
| Rent-burdened, all incomes | 52.6% | 52.7% | ±0.7 pts |
| Burdened, income under $35k | 87.7% | 87.4% | ±0.9 pts |
| Burdened, income $100k–150k | 21.3% | 21.6% | ±1.5 pts |
What burdened means at this income
Before asking who these households are, it helps to know what the 30% line measures for them. The median burdened six-figure family earns about $123,000 and pays about $3,900 a month in Gross rentContract rent plus the utilities and fuel the household pays separately, as the census defines it. It is the numerator of the burden ratio and is what “rent” means throughout this article. Medians are weighted medians of the households in the group.. Do the arithmetic and that is just over a third of its income.
Its unburdened counterpart is not so much a richer household as a cheaper apartment. The median unburdened six-figure family pays about $2,300, on an income only a third higher. Put the burdened household into that apartment and it is comfortably under the line. That is most of what is going on, and it is why the rest of this article is about rents rather than salaries.
It is also worth saying what the line does not measure. After paying rent, the burdened six-figure family has about $79,000 a year What is left after rentAnnual household income minus twelve months of gross rent. Housing researchers have long argued that this, not the ratio, is the better measure of hardship: a household is in trouble when what remains cannot cover everything else, whatever the percentage. At six figures the remainder is large even when the ratio is over the line., and two thirds of the city’s renter households have less than that left after paying their own rent. The 30% rule was written for households where the remainder was the problem. Burdened by the official yardstick, yes; in hardship, mostly no. Nonetheless the line is the one the argument uses, so this article uses it too.
It is a move-in date
If the burden is about the rent, the next question is why some six-figure families pay $3,900 and others $2,300 in the same city. The census asks one question that answers most of it: when did you Move-in yearThe census asks when the householder moved into the unit and reports the answer in seven bands, from “12 months or less” to “30 years or more”. Rent tends to fall with time in place because landlords raise rents on sitting tenants more slowly than the market moves, and because rent-stabilised leases cap the increase. The census cannot separate those two effects.? New York rents are set at signing and then drift up more slowly than the market, so a household’s rent is partly a record of when it arrived.
The answer is a staircase. Among six-figure families and singles who moved in within the last year, 22% are rent-burdened. Among those who have lived in the same home for thirty years or more, 3% are. Every step in between falls in order, and the median rent falls with it, from about $3,200 a month at the top of the staircase to about $1,500 at the bottom. Same income band, seven times the burden, depending on the date on the lease.
Reading guide
How to read it
Each row is how long ago the householder moved into the unit, from under a year at the top to thirty years or more at the bottom. The orange dot is the share of six-figure family and single-person renter households in that row who pay 30% or more of income in rent, with its 90% margin as a bar; the blue dot is the same statistic from the synthetic population. The dollar figure at the right of each row is the row’s median gross rent.
What to notice
Burden and rent fall together down the rows, at an income that does not change much across them. Households less than two years in the unit are about a quarter of six-figure families and singles but nearly half of the burdened ones.
What it cannot show
Two things are folded into the slope: the ordinary discount landlords give sitting tenants, and rent stabilisation, which caps increases on about a million of the city’s units. The census does not record which lease is stabilised. The table that follows uses the age and size of the building as a proxy.
Who is on each step matters as much as the rent, and the move-in filter in the dashboard walks you down it. The newest arrivals are the smallest households: about two people, nearly half of them living alone, in a studio or one-bedroom, with a householder in their early thirties. Ten years in, the typical household is a family of three with children and a householder around fifty, in a two-bedroom. Thirty years in, the children have left, the householder is about 68, and four in five of these households live in buildings of five or more units built before 1970, the The stock rent stabilisation covers, by proxyThe census does not record whether a lease is rent-stabilised. Stabilisation in New York covers buildings of six or more units built before 1974, plus some newer buildings with tax abatements. The closest cut the census allows is buildings with five or more apartments built before 1970, and that is what this article means by “the stock stabilisation covers”. It is a proxy: not every unit in those buildings is stabilised, and a few outside them are.About half of all the city’s renter households live in that stock. The 2023 Housing and Vacancy Survey puts the median stabilised rent at $1,500 against $2,000 for market rentals.. The full step-by-step table is in the methods paper.
Is $1,500 believable?
A six-figure household paying $1,500 a month in New York invites a fair question: why has the landlord not found a way to move them out? The answer is that the rent is typical of its kind and the landlord cannot. The city’s own Housing and Vacancy Survey puts the median rent-stabilised rent at $1,500 in 2023, against $2,000 for market rentals. A long-tenured household in a stabilised unit paying $1,500 is the median stabilised tenant, not an anomaly, and the census’s long-tenured six-figure renters live overwhelmingly in the buildings that stock is made of.
A stabilised tenant has a statutory Why a well-off tenant cannot be moved outA rent-stabilised tenant has a statutory right to renew the lease, and the allowed increase is set each year by the Rent Guidelines Board. Before June 2019 a unit could be deregulated if the tenants earned over $200,000 for two consecutive years and the rent was above a threshold; the Housing Stability and Tenant Protection Act of 2019 repealed that. Income is no longer a ground for deregulation or non-renewal, and a $150,000 household was never over the old line anyway., and the allowed increase is set each year by the Rent Guidelines Board. Income has nothing to do with it. Before 2019 a unit could be deregulated if the tenants earned over $200,000 for two years running and the rent was already high; the 2019 rent law repealed that, and a $150,000 household was never over the old line anyway. The Furman Center finds that stabilised tenants have lived in their units for about eight years on average, against three for unregulated renters, which is the same staircase seen from the other side.
None of this makes the low rents a loophole for the rich. The median stabilised tenant earns about $60,000, and the six-figure households on the bottom step of the staircase are a small minority of stabilised tenants who happened to sign decades ago and stayed. But it does mean the burdened six-figure renter is, above all, a recent one: 45% of burdened six-figure families and singles moved in within the last two years, against a quarter of the unburdened.
The roommates, separately
Now the households set aside at the start. About 153,000 of the city’s six-figure renter households, one in five, are roommate households in the census’s sense: unrelated adults sharing a unit, which includes unmarried partners because the census cannot tell the two apart. They are worth separating because “six-figure” means something different for them. Only a third of the burdened roommate households contain anyone who personally earns $100,000; the household clears the bar by adding people up.
They are also more often over the line. About 17% of six-figure roommate households are rent-burdened, against 12% of families and singles, and they are younger and more recently arrived. Put the two splits together in the dashboard and the range is wide: a six-figure family or single that has been in place more than two years is burdened about 9% of the time, and a roommate household that signed in the last two years about 21% of the time. The composite at the top of that range is not a mid-career professional priced out of the city. It is two or three people in their late twenties who pooled their incomes and signed a Manhattan lease last year.
The owner next door
The last place the count leads is next door. The census applies the same 30% line to owners, counting the mortgage payment, property taxes, insurance, utilities and condo fees as Owner costs, same 30% lineFor owners the census computes selected monthly owner costs as a share of income: the mortgage payment (principal and interest), property taxes, insurance, utilities and any condo or co-op fees. The same 30% threshold applies. Principal repayment is savings rather than spending, and the yardstick does not distinguish, so owner burden overstates hardship in the way renter burden does not.. So the same question can be asked of the six-figure household that bought instead of rented.
By that line, about 25% of the city’s owners with a mortgage earning six figures are cost-burdened, twice the rate for six-figure renters of any kind. In households the gap is wider still: roughly 102,000 burdened owners against 93,000 burdened renters, families and roommates combined. The caveat is real. Mortgage principal is savings, and a tax bill is not a rent cheque, and the yardstick does not care about either. Nonetheless, if the argument is about which six-figure households are paying more than the official share of their income for housing, the answer is the ones who bought.
Reading guide
How to read it
Each pair of bars is an income band. Orange is the share of owner households with a mortgage whose selected monthly owner costs are 30% or more of income; blue is the share of renter households whose gross rent is. The thin white ticks are the 90% margins, too small to matter here.
What to notice
The owner bar is about double the renter bar in every band, and the gap widens as income falls: at $100,000 to $150,000 more than half of owners with a mortgage are over the line.
What it cannot show
Owner costs include principal repayment, which builds equity, and property tax, which the renter pays indirectly. A burdened owner and a burdened renter are not in the same position, only over the same line. Owners without a mortgage are excluded; almost none of them are burdened.
What this adds up to
The published numbers were right. Half the city’s renters are burdened, a market-rate apartment does need a six-figure income, and the burden falls steeply as income rises. What the count adds is a size and a shape. About 68,000 six-figure families and single people in New York are rent-burdened, 6% of the city’s burdened renters, and most of them sit just over the line rather than far past it. They are burdened because they signed recently, in Manhattan more often than not, at rents roughly double what a family that signed a decade ago pays for a similar apartment. The households at the bottom of the staircase are not a puzzle: they are the city’s stabilised tenants, aged and settled, whose leases the landlord cannot decline to renew.
One more number belongs at the end, because it says how much of the count is the threshold itself. A further 9% of six-figure families and singles pay between 25% and 30% of income in rent. A rent increase of one tenth would carry them over the line and take the burdened share from 12% to about 14%; the slider in the dashboard does that arithmetic for any group. The count is real, and it is also a line drawn through a slope.
None of that settles the argument between the two camps. It does replace the anecdote with a population, and the population is smaller, more recent, more settled at the bottom and less distressed than the anecdote suggests.
What changed since the first version
- The line is now $100,000, which is what six figures means; the first two versions used $150,000. Families and singles are analysed separately from roommate households.
- The release grew from 100,000 to 500,000 people, and the census side now uses replicate weights directly.
- Two earlier findings did not survive the larger sample and are withdrawn: that burdened high-income renters were twice as likely to be headed by someone 65 or older, and that most had one earner or fewer. The second came from a census variable that is undefined for roommate households. Both are documented in the methods paper.
- The wealth section was removed. The modelled balance-sheet layer maps a household’s income rank in the metro onto national survey cells, so its dollar figures describe a poorer household than the one they are attached to. The notebook keeps the numbers and the reasoning.
- The long-tenure rents were checked against the city’s Housing and Vacancy Survey and the 2019 rent law, and the move-in staircase now shows who lives on each step.
Data and method
- Census side. ACS 2020–24 five-year microdata for the study region, in constant Income bands and 2024 dollarsAll incomes and rents are in constant 2024 dollars, using the Census Bureau’s own adjustment factors for the five survey years. The bands are the ones the release’s verification targets use: under $35,000; $35,000–75,000; $75,000–100,000; $100,000–150,000; $150,000–200,000; $200,000 and over. “Six figures” in this article means household income of $100,000 and over., with 90% margins from the replicate weights. Universe: NYC renter households with positive income. Families and singles are the census’s family households plus people living alone; roommate households are nonfamily households whose householder does not live alone.
- Synthetic side. SparkyData release 0.2, seed 20240101, with 90% Bootstrap intervals for the synthetic sideThe synthetic sample is unweighted, so its uncertainty is measured by resampling: the households in a group are drawn with replacement 2,000 times, the statistic recomputed each time, and the 5th and 95th percentiles reported as a 90% interval, the same level as the census margins. The draws are seeded, so the intervals reproduce exactly. intervals. Every variable used is inherited from the sampled census record; no modelled layer is used.
- Definitions. Burden is the census variable itself, gross rent as a share of household income, at 30%. Move-in period is the householder’s. “Pre-1970 buildings of five or more units” is the closest census cut to the rent-stabilised stock and is a proxy. Owner burden is selected monthly owner costs as a share of income, owners with a mortgage only.
- What this cannot show. Anything below the Microdata areasThe smallest geography in the census microdata is the Public Use Microdata Area, a cluster of one or two community districts with roughly 100,000 people. Manhattan has ten. Any claim below that scale is not supported by the data, which is why the article names areas but never neighbourhoods. scale. Which leases are stabilised. Whether a burdened household is burdened by choice. Anything about wealth.
- The dashboard. Runs in the browser on a 63 KB cube pre-aggregated from the 18,000 synthetic six-figure NYC renter households (one cell per type, income band, borough and move-in period, with small histograms of rent, rent-to-income ratio and householder age). Medians are read off the histograms and the rent slider shifts the ratio histogram; nothing is computed on a server. Built by
article_six_figure_renter_dashboard.py. - Where the detail is. The methods paper sets out the estimators, every margin, the reproduction checks, the full move-in table, the twelve-trait profile and the withdrawn wealth layer in four pages. The analysis script produces every figure and number from the census extract and the release database; its docstring is the long-form method and includes the sanity check on long-tenure rents. Alongside: the companion dataset of synthetic six-figure NYC renter households with fresh identifiers,
pums_direct.csvwith every census estimate and its margin,reproduction.csv,results.json, andfigS1_profile.svg.
Sources
- U.S. Census Bureau — ACS 2020–2024 5-Year PUMS (NY, NJ, CT), housing and person files with replicate weights — every count, share and margin labelled census; variables GRPIP, GRNTP, HINCP, PINCP, MV, HHT, NP, BDSP, BLD, YRBLT, OCPIP, TEN
- NYC HPD — 2023 New York City Housing and Vacancy Survey, Selected Initial Findings — citywide median rent $1,641; rent-stabilised median $1,500; market $2,000; rent-controlled $988 (2023 dollars)
- NYU Furman Center — Understanding Different Segments of New York City's Rent-Stabilized Housing Stock — average tenure eight years in stabilised units against three in unregulated; about 65% of the stabilised stock is pre-1974 buildings
- NYC Rent Guidelines Board — Rent Laws of 2019 (Housing Stability and Tenant Protection Act) — repeal of high-income deregulation; stabilised tenants' right to renew is not conditioned on income
- Office of the NYC Comptroller — Spotlight: New York City's Rental Housing Market (Jan 2024) — half of renters burdened (ACS 2022); asking rent $3,500; $140,000 income to avoid burden
- Office of the NYS Comptroller — Housing Cost Burdens for New Yorkers Among Nation's Highest (Feb 2024) — 9 in 10 renters under $35,000 burdened; 16% of households over $75,000 burdened (statewide)
- NYU Furman Center — State of Renters and Their Homes 2024 — just over half of renters burdened in 2023; 28.9% severely burdened