Along with Zohran Mamdani’s historic victory in last month’s elections, New York City also approved three housing-related ballot proposal. Together, these will make it somewhat easier to adjust land-use rules to allow for new housing development, by reducing the City Council’s ability to block zoning changes.
I am glad the proposals passed, for reasons similar to those laid out by Michael Kinnucan. While zoning changes are not a sufficient solution to the city’s housing problems, they are helpful — and more important, they are a necessary condition for a bigger program of public investment in housing.
Support for the proposals was shared by many, but far from all, housing and tenant advocates in the city. Debates over the proposals reflected differences on political principle — how big a voice should local as opposed to citywide officials have over land use? — as well as on economic theory — how well does the housing market fit a simple story of supply and demand? But there are also some background factual questions where the answers tend to get assumed rather than directly debated, about what kind of housing gets built in the city right now.
So in this post, I wanted to assemble some factual information about recent housing construction in New York. For convenience — and because that’s how much of the data is organized — I am defining recent as meaning the period since 2010. Some of this is assembled from various reports and publications, but the bulk of it is my own analysis of the New York Housing and Vacancy Survey (HVS).
*
The dimension of new housing construction that is probably most visible is how geographically concentrated it is. About one-third of all the new housing built since 2010 is in just four of the city’s 59 community districts, along the East River in Brooklyn and Queens.
You can see this clearly in this map from the Department of City Planning, as the strip of dark blue running from Brooklyn Heights to Astoria. (The dark blue area in Manhattan reflects some major projects on the far west side, including Hudson Yards.) Brooklyn Community District 1, including Williamsburg and Greenpoint, added 30,000 new housing units between 2010 and 2024. Half a dozen miles away at the south tip of Brooklyn, District 10, with a similar population, added only 500.
The concentration of new housing in a few areas reflects a number of factors, including zoning changes under the Bloomberg administration and the disappearance of manufacturing from former centers like Long Island City. This helps explain the association in many people’s minds of new housing development with gentrification and rising rents.
Less immediately visible is how much this newly-built housing costs, and who lives in it. I haven’t seen a report focusing on these questions — though I expect one exists — so I thought I would see what I could say using the Housing and Vacancy Survey.
*
For those who aren’t familiar with it, the HVS is a survey conducted every three years ago by the Census on behalf of the New York City Department of Housing Preservation and Development (HPD). Its primary purpose is to help administer the city’s rent regulations, but it’s a useful resource for all kinds of housing research. It’s a decently large sample — about 10,000 observations — but what it makes it especially nice is that it combines administrative data on things like building size, location and rent-regulation status, with survey data on things like occupant characteristics and the unit’s state of repair.
The HVS is a good tool if we want to answer questions like, what is the median household income of people living in housing built since 2010? ($73,500, it turns out — but we’ll come back to that a bit further down.) The most recent HVS was conducted in 2023; to get a reasonable sample for smaller subgroups I combined it with the 2021 survey, with appropriate adjustments to the monetary variables.1
Between 2010 and 2024, NY added just over 300,000 new units of housing, or a bit over 20,000 units a year. This is a respectable level of new building for the city by recent standards — comparable to the 2000s and 1970s, and faster than in the 1980s or 1990s — but less than in earlier periods of the 20th century. During the 1950s and 1960s, the city added over 30,000 units per year, and in the 1920s, over 70,000. A surprisingly large proportion of these houses are still here. For example, 729,000 housing units were built in the 1920s; according to the HVS, 718,000 of them were still present as of 2023. That housing lasts such a very long time is, to me, one of the central facts that makes it different from most commodities. (The other is that it’s located in a particular place.)
Of the housing units built between 2010 and 2023, about 10 percent are owner-occupied, a bit over 25 percent are unregulated market-rate rentals, and 60 percent are rent-regulated rentals. (There are also a small number of vacant units that are not for rent, and a very small number of new public housing units.)
It might be surprising that there are more rent-stabilized units than market-rate ones, given that rent regulations in New York by default apply only to large buildings built before 1974. There are two reasons for this.
The first reason is that a substantial fraction — 25 to 30 percent — of new housing built in New York in recent years has been subsidized affordable units. “Affordable” in this context is a term of art: It refers to housing that receives public subsidies, most importantly the federal Low-Income Housing Tax Credit, and in return is limited to renters (or occasionally purchasers) making below a certain income threshold — 80 percent of the area median income or some lower fraction.2 In New York, these subsidized units are also normally rent-stabilized. As the nearby figure from the Furman Center shows, the proportion of affordable-in-the-technical-sense units has fallen off somewhat in recent years, but is still substantial.
It’s important to note that while the figure shows “LIHTC” (Low Income Housing Tax Credit) units and “market rate units,” this is not a straightforward division. While most income-restricted units receive LIHTC subsidies, not all of them do; and units that do not receive operating subsidies or have income restrictions, and are thus counted in the market rate category here, may still be subject to rent regulation. In the rest of this post, I instead focus on rent-regulated versus unregulated units, where there is a sharper line. 3
The second reason for the high proportion of rent-regulated units is that most new housing built outside of Manhattan during this period was eligible for the 421-a property tax exemption.4 This gives long-term exemptions from property taxes — as long as 40 years in some cases — in return for certain conditions, including participation in rent stabilization. As a result of these programs, even though tent stabilization is not compulsory for any housing built since 1974, in practice newer housing in New York are more likely to be rent stabilized than older ones.
I personally agree with critics who argue that these tax exemptions are a wasteful and inefficient way to promote new housing construction. The problem for developers is financing the start of the project — a tax exemption decades from now is essentially worthless to them, while for the city, with its longer horizons, it is still costly. In effect, 421a is paying for housing in a currency that is worth much less to the recipient than to the payor.\efn_note]Put another way, the public sector ought to have, and in practice generally does have, a much lower discount rate than the private sector. This used to be a big part of debates on the economics of climate change. But it’s also relevant to housing. The common thread is the long time periods involved.[/efn_note] But be that as it may, it has resulted in a very large fraction of new housing being rent-stabilized.
The fact that perhaps a quarter of the new housing produced in New York is income-restricted affordable units — surely the highest proportion in any major US city — does not get much attention in discussions of housing, as far as I can tell. Nor does the fact that the large majority of new housing is rent-stabilized — I wasn’t aware of it myself until quite recently. But both of these seem like important facts.
*
Let’s move on to how much these recently-built apartments rent for, the question that got me started writing this post. The median rent for rent-regulated apartment built since 2010 is $1,800, while the median rent for an unregulated (i.e. market-rate) apartment built since 2010 is $3,200.5
To be sure, the comparison of rents in stabilized versus unregulated apartments is a bit tricky, because these are not the same types of apartments. As the figures nearby show, unregulated units are more likely to be in Manhattan, and are somewhat larger on average — studios and one-bedrooms make up 70 percent of recent rent-stabilized units, compared with 60 percent of recent unregulated ones. One thing that surprised me looking at these numbers was how few larger rental units are being built, market-rate or otherwise.
Since 421-a subsidies are not generally available in most of Manhattan, the rent-stabilized units there are mostly subsidized affordable units. So in Manhattan, recently-built market-rate apartments rent for almost twice as much as equal-size stabilized ones. Meanwhile, in Brooklyn regulated units rent for only about one-third more than unregulated ones, and in Queens and the Bronx rents for the two classes of apartments are essentially the same. (Staten Island has hardly any new housing of any kind.)
The distribution of rents by regulation status is shown in the figure below, which is perhaps the main thing you should take from this post.
Here we see that there are more 35,000 rent-regulated apartments built since 2010 that rent for less than $1,000, and barely 5,000 unregulated apartments renting for that little. But while most regulated apartments rent for less than $2,000, more than a quarter rent for over $3,000 and about 10 percent rent for over $4,000. Meanwhile, about 70 percent of unregulated units rent for between $2,000 and $4,000, while a quarter rent for less than $2,000 and 10 percent for more than $5,000.
Again, these differences are in part due to the fact that unregulated apartments are somewhat larger, and considerably more likely to be located in Manhattan, compared with rent-regulated apartments.
For recently-built rental units as a whole, the median rent is $2,000, with one-third renting for less than $1,100 and one-third for more $3,000; 10 percent rent for more than $4,500. This is somewhat higher than rents in older buildings — for the city as a whole the median rent is $1,670. (If we compare one-bedrooms only, the comparison looks similar.)
There are obviously many more ways one could slice this, but these numbers give a useful benchmark: If we are talking about a newly built market-rate apartment in New York, we should think about an apartment renting for around $3,200. If we want to get a bit more granular, we could think of one-bedroom apartment in Brooklyn renting for $2,200 a month, a 2-bedroom in Brooklyn renting for $3,800, or a one-bedroom in Manhattan renting for $4,700 — these would be typical examples of recently built market-rate apartments. (Though the sample size gets uncomfortably small as we slice the data on more dimensions.)
*
A nice thing about the HVS is that it lets us do the same analysis for incomes.
The short answer here is that median household income for residents of recently-built owner-occupied units, median income is $161,000. For rent-regulated apartments, median household income is $54,000; for unregulated apartments, it’s $117,000. For recently-built rental units as a whole, the median household income is $73,000.
As it happens, $73,000 is almost identical to median household income for the city as a whole. The $117,000 median income for residents of recently-built market rate rentals, meanwhile, is close to the 67th percentile for the city as a whole — in other words, two thirds of households have incomes below this, and one third have incomes above it.
The issues with geography and unit size are not as relevant here. 6 But for the half or so of rent-regulated units that are also subsidized and income-restricted, resident incomes will of course be lower. The median income in unregulated apartments is more than twice as high in Manhattan as Brooklyn — $205,000 versus $90,000 — while the median rent in rent-regulated apartments is only about 25 percent higher.
The figures nearby shows the distribution of recently-built regulated rentals, unregulated rentals, and owner-occupied units by household income and by per-capita income, which is arguably more relevant. (Note that the income categories are slightly different for the two figures.)
As you can see, the majority of recently-built rent-stabilized units — 78,000 out of 134,000 — are occupied by households with income below $75,000, approximately the city median. About 15,000 of them, however, are occupied by households with incomes above $250,000. The distribution of incomes in unregulated units is flatter — a bit over 10,000 have tenants with incomes under $40,000, and about the same number have tenants with incomes with incomes above $250,000. Incomes are much higher in owner-occupied units. Nearly half — 10,000 out of 22,000 — are occupied by households with incomes above $250,000.
The picture looks a bit different when we turn to per capita incomes. For comparison, the median per-capita household income in New York City is $36,000. The majority (about 55 percent) of rent-regulated new apartments are occupied by households with incomes below this. But only about one-third of unregulated apartments are. Interestingly, when we look at per-capita income, owner-occupied units are no longer so disproportionately likely to be occupied by households with very high incomes. In New York City, evidently, homeowners are much more likely to have larger families.7
*
How should one evaluate these numbers? My goal in this post is just to bring some facts into view. I’m not so much trying to make a substantive policy argument, as trying to make the debate more concrete and specific, at least in my own head. In some ways, the best case for this post would be that people would very different views about housing policy could find something in it they could use.
That said, what prompted me to start looking at these numbers were claims, in the runup to the election, that simply making it easier to build will not help with affordability, since private development won’t produce affordable housing, or will only produce luxury housing.
To be clear, these are two different claims. Or to put it another way, affordability in the everyday political sense is different from affordable as a term of art in housing policy.
Housing the is not affordable, in the technical sense, may still be helping with affordability in the broader sense, by offering better housing options for people who are not wealthy. A family of two New York public school teachers might have a combined income of $150,000 or so, putting them outside the income limits for subsidized affordable housing. But they may nonetheless have real problems finding reasonably-priced housing, especially if they have kids; and new construction might improve their situation even if it is not affordable in the technical sense.
What does this data say about these questions?
Perhaps unsurprisingly, the HVS data supports the claim that, in the absence of subsidies, private developers will not build much deeply affordable housing. One way of looking at this: About 20 percent of New York households have incomes below $20,000; but in unsubsidized units built since 2010, only about 6 percent of tenants have incomes below this level.
Another way of looking at it: The median New York household has an income of $73,000; for them not to be rent burdened, by conventional standards (30 percent of income going to rent), they should pay no more than $2,000 per month. But nearly 80 percent of the unregulated apartments (as well as 30 percent of rent-regulated apartments) built since 2010 rent for more than this. And many of the ones renting for less are studios or one-bedrooms, which will not be suitable for many households with incomes near the median.
So, the claim that allowing more private development will not by itself produce much housing affordable to lower-income New Yorkers, seems consistent with the data.
Now, any economists or abundistas reading this will want to jump up, and point out that even if newly-built housing is not affordable for many New Yorkers, it can still help them. The people who move into the newly built units are going to live somewhere, after all; and if these new ones weren’t available, they would be bidding up the price of the existing housing stock. Turning an old sugar refinery in Williamsburg into luxury apartments may not directly provide affordable housing in Williamsburg, but it takes the pressure off the rental market in other neighborhoods that the trust-fund hipsters might otherwise move to.
Ok, you guys can sit down, you’ve made your point. And it’s a valid one — there is definitely some truth to this. How much truth, and what factors might work in the other direction, is beyond the scope of this post. Here, I’m just trying to get my arms around the difficult-enough question of what rents and incomes look like in the newly-built housing itself.
Returning to the central question of how affordable newly-built housing is, it’s worth recalling that 20-25 percent of new housing is affordable in the sense of being income-restricted and receiving ongoing subsidies, and a majority of new housing opts into rent regulation. So focusing on the unregulated segment may be a bit misleading, especially in the context of the ballot proposals. A more sensible comparison might be between recently-built housing in the aggregate, and older housing. The next couple of figures do that.
Here we see the distribution of rents in newer and older buildings. Note that the vertical scale is share of units in that age group, as opposed to the absolute number of units as in earlier figures.
What we see is that while there are a substantial number of new units with moderate rents, there are many more high-rent units in the newer buildings. About 15 percent of units built since 2010 rent for more than $4,000, compared with just 3 percent of older units.
Of course, new units are different from older units in other ways — location, size and so on. But if we limit the analysis to, say, just one-bedroom apartments, the pattern is basically the same.
If anything, the excess of recent units at the high end is even clearer in this case.
Then again, one could look at the same numbers the other way. 15 percent of new units rent for over $4,000 and 30 percent rent for over $3,000, compared with just 3 and 8 percent, respectively, of older units. But that means that 70 percent of new units rent for under $3,000; and about 40 percent rent for less than $2,000 — which is, again, the threshold for rent burden for the median-income New York household.
So if we look at the housing that is being built in New York now, it is absolutely true that it is disproportionately luxury housing intended for the rich. Although not necessarily for the very rich — Andrew Cuomo’s $8,000-a-month Upper East Side apartment would be in the top 2 percent of rents among recently-built units. But disproportionately is not the same as exclusively. It is not true that recently-built housing is exclusively luxury units for the highest-income New Yorkers.
We can take this question on more directly by looking at household income among tenants in recently-built rental units as opposed to older ones. This is shown below.
Surprisingly, the distribution of incomes across newer and older apartments is much closer than the distribution of rents. High-rent apartments are much more overrepresented among newer apartments than high-income tenants are.
On reflection, this is not surprising. Thanks to rent regulations (and also to smaller landlords who don’t aggressively raise rent for current tenants) many current tenants are paying well below market rent. Remember, rent regulations in New York limit only rent increases. So one might even say, that if the rent regulation system is effective, it will inevitably result in newly-built apartments renting for significantly more than existing ones. And inevitably, many of those older rent-regulated buildings will be occupied by higher-income households.
Note, also, that newly-built apartment have a slightly higher proportion of very low income tenants than older ones do. This reflects the substantial fraction of subsidized affordable units, and is another reason to reject the “only luxury units are being built” claim.
*
What do we take from all this?
There are two things that surprised me the most, looking at these numbers. First was the large fraction of rent-regulated units — more than two-thirds of the units built since 2010. I had always thought of rent regulations in New York as applying almost exclusively to older buildings, but in fact, of the all the age categories in the HVS, post-2010 buildings have the highest proportion of regulated rentals.
Second was the preponderance of smaller apartments among recently built housing. 55 percent of the units built since 2010 are studios or one-bedrooms, compared with 38 percent of older units. Units with three bedrooms or more, meanwhile, account for only 10 percent of recently-built units, compared with a full third of older ones.
This second fact leads to the first of my policy takeaways: When we are talking about housing affordability, we need to think about what kind of housing, as well as its cost.
Most goods are fungible: If your family consumes more milk, or gas, or electricity, then you pay more for it, but the price of the next gallon or kilowatt is the same as the last. Buying a gallon of milk is essentially the same as buying two half-gallons. Housing is different: You can’t just rent some extra square feet when your family gets bigger, you need a whole new home. Building more SRO-type units, as some people advocate, would help address affordable housing at the low end; but it wouldn’t do anything to solve the problems of rent-burdened families.
This non-fungibility of housing was eloquently described by Sam Stein in a New York Review of Books piece a few years ago:
Housing will never be as elastic as households. This is not only because construction is complicated in a city as crowded as New York, but also because there is a fundamental difference between people and things. Households change shape over time and can recompose rapidly during an emergency like a pandemic. But despite the work of inventive architects, our housing tends to stay more or less the same. … There is nothing quite as concrete as concrete.
To be clear, the solution is not as simple as simply requiring developers to build more larger units. As this report from the Fiscal Policy Institute points out, this approach could be counterproductive, discouraging new housing construction of all kinds.8 But it is certainly something to consider in the design of subsidies or social housing programs.
My second policy conclusion was touched on a bit earlier: We need to be careful about what we mean by affordable. A lack of housing is an acute problem for the very poor. But many people with higher incomes also struggle with housing costs. The figure below shows the share of households paying over 30 percent of their income in rent — the conventional definition of rent-burdened.
As the figure shows, almost all low-income renters are rent-burdened, while almost no high-income households are. But a surprisingly high fraction of middle-income households are rent-burdened by the conventional standard. If we look at households in the middle third of the income distribution, from approximately $40,000 to $120,000, 45 percent of the renters pay more than 30 percent of their income in rent. (And in New York, the large majority of people in this income range rent.)
When we are talking about affordable housing, we should not just be talking about housing for very low-income people, with the implicit assumption that everyone else is adequately served by the existing housing market. We should be talking about a problem with the private provision of housing in general.
Two more points speak more directly to the ballot proposals. On the one hand, “build more housing” is a valid and important policy goal. Even if there were no gains to affordability, simply having more people living in New York (and other dense cities) is a win for humanity, for all sorts of reasons I do not need to go into here. But as the HVS data shows, new housing is helping with costs as well. A large fraction of the housing being built in recent years has been relatively affordable, and is occupied by households in the lower and middle parts of the income distribution.
A corollary of this: Rent-regulated housing rents for significantly less than market-rate housing, and houses people with significantly lower incomes. We can certainly ask whether our subsidy dollars could be spent more efficiently. I personally think that the long-term tax credits are not the right approach; if we want to trade future tax revenue for present housing production, we would do much better to issue bonds backed by that future revenue, and provide the subsidies up front. But for present purposes, the key point is that these subsidies do produce affordable housing.
On the other hand — my final policy point — the fact that recently-built unregulated apartments rent for so much more than existing apartments, and have such disproportionately higher-income tenants, should make us more skeptical of claims that land-use reform, by itself, will substantially reduce housing costs. It could be that rents in newer apartments are high because not many of them are being built, so that is what the market will bear. But it also could be that rents in newer apartments are so high because that’s what private developers require in order to build them.
There may be some truth to both of these views, of course; but I suspect there is more to the second. In which case, while land-use reforms like the three ballot proposals are desirable and important, they will need to be complemented with public interventions in the financing and development of new housing to have a real impact on affordability.
*
One final point, on the politics, and a final picture, not from the HVS data.
I started this post back before the election, before setting it aside for a while. A that point, I was concerned that misperceptions about what kind of housing is currently being built might be fueling opposition to the ballot proposals. People who care about affordable housing might oppose making it easier to develop housing if they thought that the only housing being built in the city was luxury apartments for the rich.
Now that the election is over, we can see who actually did oppose the measures, and who supported them. Below is the map for Proposal 2, with yes votes in green; the other two would look similar.
What do we see? Well, obviously, this looks like the map of the mayoral election. Not exactly — the proposals carried all of Manhattan, while the Upper East Side voted for Cuomo. But by and large, the areas that voted yes on the proposals are the areas that voted for Zohran Mamdani.
I think this tells us something important about the politics of housing. There’s an argument one often hears, that the politics of housing cut across conventional left-right lines — that arguments against new housing is often made on environmental or anti-gentrification grounds, and come from people who are, in other respects, on the left side of the political spectrum.
Now I would not say there is no truth to this idea. It’s probably most true in the Bay Area, but it’s not limited to there. During the fights over the Atlantic Yards development here in Brooklyn, I personally observed houses with both the iconic “In this house…” and “We love brownstone Brooklyn” signs; needless to say, most New Yorkers do not live in brownstones.
But it’s easy to exaggerate the importance of this combination of views. In the real world, the vast majority of opponents of higher-density housing are not liberals who fly rainbow flags and donate to the Sierra Club; they are conservative homeowners who, not to put too fine a point on it, don’t want Black people moving into their neighborhoods.
Of course there are sincere progressives and socialists who believe that building more housing will only raise rents; and it’s worth trying to persuade them that, in fact, more development, even private development, is an essential part of a broad public program for housing affordability.
But those people are not the main obstacle. The people who are against building more housing are, by and large, the same people who will oppose any program to raise living standards by redistributing income and power and expanding the role of the public sector. It’s the same old lines of left versus right.
UPDATE: I forgot to mention: I adjusted the total number of units built since 2010 in the HVS so it matched the total from the Department of City Planning for units built between 2010 and 2023. But I didn’t see an easy way to do this for subgroups; and while the HVS weights ensure that counts across various categories of buildings match the official totals, the weights are for the whole sample, not for building-age subgroups. So there is going to be some sampling error here — these are not exact counts. I feel reasonably confident that the picture is qualitatively correct, though.
