Airbnb and gentrification in New York

6 vulnerability index

For the last several months I’ve been working with Alexander Weisler (a recent MUP graduate from the School of Urban Planning at McGill) on a paper which explores the connection between short-term rentals and gentrification. We use a case study of Airbnb in New York City, based on a lot of number crunching, GIS, and interviews with community organizations and policymakers. The paper is nearly finished, and I’ll upload it here once it’s ready. But in the meantime (and recognizing that it will be a year or more before the paper makes it through peer review and the publishing process), I wanted to provide a quick tour of the arguments and evidence, using the near-final maps I’ve spent the last several weeks making.

The thesis of the paper is that Airbnb is systematically creating a new kind of rent gap. Following Neil Smith’s original argument, we normally think of rent gaps as emerging where localized disinvestment drives down the money landowners earn from their properties, even as overall city-wide growth increases the potential money they could earn if they were to renovate or redevelop. Once this gap between actual and potential profit gets big enough, developers take an interest, and reinvestment—and hence gentrification—is likely to occur.

What we see with Airbnb is also a rent gap, but it’s not one that relies on disinvestment or developers. Instead, the opportunity which the service provides to rent out an apartment to short-term tourists instead of long-term tenants is driving up potential landlord earnings without any disinvestment having occurred—and without any need for big expensive renovations to capitalize on the opportunity. All a landlord needs to do is evict current tenants or decide not to find new ones when a lease ends. So this is an enormous new opportunity for profit-making in urban housing markets where there is external tourist demand, and an enormous new risk of displacement and gentrification.

Building off these ideas, the paper makes three specific arguments: First, Airbnb has introduced a new potential investment flow into housing markets which is systematic but geographically uneven, creating a new form of rent gap in culturally desirable and internationally recognizable neighbourhoods which have generally already been subject to extensive gentrification. Second, Airbnb offers a means of exploiting its own rent gaps to a range of different housing actors, from developers to landlords, tenants and homeowners, and the actions these different actors take to exploit new rent gaps have very different impacts on urban housing markets. Third, even though Airbnb-induced gentrification frequently runs counter to entrenched urban governance interests, municipal regulators are severely hampered in their ability to effectively rein in short-term rentals, in large part due to scalar governance constraints and the demands of inter-urban competition.

How much rental housing has New York lost to Airbnb?

In order to estimate the impact Airbnb has had on New York housing, I’ve relied on data scraped from the public Airbnb website by the consulting firm Airdna, covering all Airbnb activity across the New York region from 2015. What follows is a series of maps using that data in combination with data from the US Census and the American Community Survey.

Figure 1 shows all Airbnb listings for 2015, aggregated by census tract. (There are approximately 110,000.) It reveals hotspots in Hell’s Kitchen and Chelsea (near the existing Manhattan hotel district, and an area with a long history of illegal hotels), the Lower East Side, and Williamsburg and Bushwick in Brooklyn.

1 all listings (updated).png

Figure 1: Total estimated number of Airbnb listings per census tract in the New York region.

(Incidentally, I looked at data from the entire tristate New York metropolitan region, but almost 90% of Airbnb listings are in the immediate vicinity of Manhattan shown on the map. The exceptions were the Hamptons and the Jersey Shore, where arguably Airbnb has simply made existing vacation rentals easier to accomplish rather than facilitate large-scale conversions of permanent rental housing to short-term rentals, so I felt justified mostly ignoring these areas.)

This map is a decent approximation of what you’ll see if you go to and search for listings in New York, but it doesn’t do a very good job of estimating the actual impact of the service on housing in the city. Many of these 110,000 listings were just rented once or twice, or even not at all.

So I narrowed in on whole-unit listings that were occupied more than two months a year. “Whole-unit listings” means I excluded cases where people were renting a spare bedroom; every time a whole-unit listing is rented, we know there’s no one else living in the unit. And two months a year of occupancy is a reasonable threshold for a “full-time” listing (used by other researchers as well), because it’s a period of time that isn’t compatible with the standard 12-month residential lease, even allowing for some transition time between tenants. There will surely be some false positives with apartments that are in the long-term rental market most of the year but just on Airbnb in the summer, but also many false negatives with all the private-room listings that would have formerly been ads for roommates on Craigslist—an important source of affordable housing that is hard to measure in a study like this. (In the paper I discuss how the results differ if we change the full-time assumptions.)

All told, around 21,000 housing units in New York were rented full-time on Airbnb in 2015. [Edit: following some pushback from Airbnb’s CEO, here are some additional details about this estimate. If I change the two-month threshold to three months, the estimate drops to 16,000, and if I change it to four months, the estimate drops to 12,000. There’s a lot of uncertainty in this parameter, but in both cases, the underlying patterns—depicted in the rest of the maps—remain the same, so I am reasonably confident about these estimates, based on the third-party data I have access to. Those who are interested can read a thorough discussion of my methodology.] If we compare this number with the amount of normal rental units in the region, we can estimate what portion of each neighbourhood’s rental housing stock has been lost to Airbnb. This is shown in Figure 2:

Figure 2: The percentage of long-term rental housing converted to full-time Airbnb activity.

The pattern is similar to Figure 1, although the clustering in Lower Manhattan and North Brooklyn is more pronounced. And the numbers themselves are shocking—many census tracts have seen five percent or more of their long-term rental housing converted into Airbnb hotels. It’s impossible to estimate how many tenants were forcibly evicted or harassed out of their apartments to free up units for Airbnb, and how many units were simply converted to short-term rentals after they “naturally” became vacant. But in either case, the result has been a huge and concentrated loss of rental housing in the city.

I consolidate these estimates of how much long-term rental housing has been directly lost to Airbnb, by neighbourhood, in Figure 3:

3 community boards.png

Figure 3: A neighbourhood-level summary of the rental apartments Airbnb has removed from New York’s housing system.

Two points to note here. First, Airbnb’s impact on housing in New York is highly uneven—it’s so far limited mostly to Lower Manhattan and to Northern Brooklyn. This means that any city- or region-wide aggregate statistics about Airbnb’s presence in New York are going to be highly misleading. No surprise that it is precisely these sorts of statistics that Airbnb leans on….

Second, this map arguably understates Airbnb’s impact on Williamsburg—which already looks quite high at 2.8% of the neighbourhood’s rental housing stock currently being rented full-time on the service. This is because the southern edge of Williamsburg has almost no Airbnb activity, thanks to the large and insular Hasidic Jewish population in the area. If you remove the 4,500 or so housing units from this area, the rest of Williamsburg ticks up over a 3% Airbnb-to-rental-housing ratio.

Airbnb’s rent gap

Beyond estimating how much long-term rental housing New York has lost to Airbnb, the paper also attempts to demonstrate the existence of an Airbnb rent gap in New York. Here the relevant metrics are less about housing units and more about money—revenue flows through the urban housing market. As Neil Smith explained nearly forty years ago, gentrification is “a back to the city movement by capital, not people”.

The existence of a rent gap means that, systematically across a neighbourhood, landowners can earn more money from some different use of their property than from the existing use, which creates an incentive to reinvestment and hence gentrification. As I discussed above, we normally think of rent gaps leading to new capital investment—renovations and redevelopments—but in the case of Airbnb this generally won’t be necessary. Property owners will just switch their units from residential leases to short-term rentals. So if there has been an Airbnb-induced rent gap, we shouldn’t expect to see big new capital expenditures; instead we should expect to see routine housing revenue flows (which are mostly composed of rent and mortgage payments) diverted into Airbnb. Figure 4 provides a rough estimate of this activity—an estimate of where Airbnb has created and already plugged a rent gap:

4 income ratio.png

Figure 4: The percentage of ongoing housing revenue flows (e.g. rents and mortgages) which Airbnb now accounts for—a measure of the rent gap which Airbnb has created and filled.

Unsurprisingly, the basic pattern is similar to the previous maps, although the clustering is even more acute. Airbnb as a new revenue stream from housing has been most consequential in Times Square, the Lower East Side, and Williamsburg. These are the areas where Airbnb created a rent gap, and where landlords have shifted housing into short-term rentals to capitalize on that rent gap. Importantly, these three neighbourhoods are all “post-gentrified”, in the sense that they saw massive increases in rents and massive displacement over the last several decades, and now have been to a greater or lesser extent transformed into wealthy neighbourhoods. Airbnb has had its biggest impact to date, in other words, not at the gentrification “frontier”, but in areas that have already been pervasively restructured by capital. It is further intensifying gentrification and displacement dynamics where these dynamics have already been acute.

However, we get a very different picture of Airbnb’s impact if we look at how much landlords can earn on the service when compared to prevailing rents in their neighbourhoods. In other words, leaving aside for the moment the question of where total Airbnb revenue flows have been highest, where are individual landlords making the most money on Airbnb relative to what they could have been making with traditional rentals? Figure 5 answers this question:

5 median rent.png

Figure 5: How much monthly revenue the average full-time Airbnb listing generates, compared to the median rent in each census tract—a measure of the potential profit landlords can make by converting housing from long-term to short-term rental, i.e. the rent gap Airbnb has created but not yet filled.

This is a completely different geography from the previous maps. While the Lower East Side remains a hotspot on this map, with average full-time Airbnb revenues in the range of 200% of median rents, the other major areas of Airbnb activity—Williamsburg and Hell’s Kitchen—have significantly receded in importance. Meanwhile, three new neighbourhoods have appeared: Harlem in North Manhattan, Bedford-Stuyvesant in Brooklyn, and Union City and its surrounding areas in New Jersey. These are areas where there isn’t yet a lot of Airbnb activity in absolute or even relative terms, but where the landlords who are using Airbnb are making a lot more money than they would have in the long-term rental market.

Put differently, Figure 5 shows the neighbourhoods which appear to have large and unfilled rent gaps—where there is money to be made but where landlords haven’t yet seized on the opportunity. These are the neighbourhoods at greatest risk for Airbnb-induced gentrification in the near future. And whereas current Airbnb impacts were concentrated in already-gentrified areas, these at-risk neighbourhoods are all still very clearly at the gentrification frontier.

Comparing these two patterns—the percentage of housing revenue that now flows through Airbnb, and the percentage of the median rent which an average full-time Airbnb property earns—allows us to see where Airbnb has already had a major impact on local housing and where it is likely to have an impact in the future. The first pattern indicates where Airbnb has already had a major impact on local housing—where it has created and filled a rent gap. The second pattern indicates where there is still money to be made for landlords by converting long-term rental housing to short-term rentals—where Airbnb has created a rent gap which hasn’t yet been filled.

I combined these two patterns by performing a cluster analysis on the two variables to identify the areas of New York which stand out either in terms of high current impact or high risk of future impact. The result is Figure 6, a “clustered vulnerability index” for Airbnb-induced gentrification in New York:

6 vulnerability index.png

Figure 6: A clustered vulnerability index of Airbnb-induced gentrification in the New York region, indicating that neighbourhoods now at risk are predominantly African-American ones.

There are three important types of neighbourhood which emerged here. First, shown in blue, are the areas which have had their housing supply heavily impacted by Airbnb, but which may be close to reaching an equilibrium (a closed rent gap). Most of lower Manhattan and Williamsburg fit this profile. Second, shown in red, are the areas which haven’t yet been seriously impacted by Airbnb, but are in real danger of it in the near future, because of how much more money landlords in these areas are making by using Airbnb (an open rent gap). Harlem and Bedford-Stuyvesant in New York fit this profile, as do parts of Hudson County in New Jersey. Last, shown in purple, are the areas which have already been heavily impacted by Airbnb, but where there appears to be more impact still to come (a not-yet closed rent gap). The Lower East Side and parts of Williamsburg fit this profile.

An important point which the map doesn’t communicate is that the blue already-gentrified areas are predominantly white neighbourhoods, while the red “high risk” areas are all heavily African-American neighbourhoods. When you combine this fact with research showing how prevalent racial discrimination is on Airbnb, this implies that a major new intensification of racialized gentrification is coming to these areas.

I am playing around with versions of the vulnerability index map to reflect this situation. But I’m also planning to use this index as a sort of template for looking at other cities, particularly where there are community organizations and policymakers interested in fighting back against gentrification and displacement—and against Airbnb’s increasingly obvious role in facilitating these destructive processes.