Airbnb has become a lightning rod for controversy about the so-called sharing economy. In tight housing markets – in particular, New York City and San Francisco, many have claimed that Airbnb contributes to the displacement of longtime, poorer residents.
While these reports are alarming, none have tried to map the major concern – concentrated neighborhood impact from Airbnb’s heavy concentration in small parts of cities. By leveraging US census data in CartoDB’s Data Observatory, it’s possible to isolate the impact of Airbnb to a the level of just a few city blocks.
Inside Airbnb data is incredibly detailed, to the point where visualizing every point can limit our ability to extract clear information. Our challenge was to visualize the listings in a way that would make clear what their impact was at the local level. A quick glance at a point-based Airbnb map shows that listings are concentrated in certain, generally central, parts of cities, but point-based maps don’t take into account neighborhood characteristics – all points look the same.
We wanted to understand the effect of Airbnb as it is experienced by residents at the level of a neighborhood or even just a block. Below we will work through several techniques for aggregating and exploring the Airbnb data in the context of the American Community Survey (ACS). The ACS provides us with information about the population across the US at the block group level.
We are excited to show the power of these 3rd party datasets and are slowly making them available to CartoDB users in exciting new ways. Watch out for announcements soon, or get in touch if you are interested in taking advantage of our work sooner.
Airbnb units per square mile
Our first analysis of the Airbnb data aggregates the point data into block groups and shows the concentration of listings per square mile. Next, we wanted to use a statistical method called Moran’s I to cluster and highlight areas of statistically significant high or low rates of Airbnb per square mile.
You can see the result in our first map below. While this still does not incorporate any neighborhood characteristics, it provides great visibility of high-density clusters of Airbnb running from the Lower East side up most of Central Park in Manhattan, with small clusters in Williamsburg and Clinton Hill in Brooklyn.
Airbnb compared to available housing units
A more nuanced analysis leverages the American Community Survey to look at what percentage of housing units in each block group are listed on Airbnb. Shown in the second map above, this approach reveals very different clusters than the naive analysis of Airbnb per square mile. While Airbnb is very dense in the Upper East and Upper West Side of Manhattan, it is actually a significantly “low” percentage of housing units in those neighborhoods. In contrast to that, the two Brooklyn clusters are much larger, with almost the entirety of Williamsburg and Greenpoint rented out on Airbnb at a high rate.
The reason for these differences is housing density. The Upper East and Upper West sides of Manhattan are very dense; even if they have a lot of Airbnbs, the neighborhood can more easily absorb these units into the huge existing stock of apartments.
Williamsburg and Greenpoint, however, are lower density neighborhoods. Even though the per square mile concentration of Airbnb may not be very high, as a percentage of the available housing it is much greater. Williamsburg and Greenpoint are therefore Brooklyn neighborhoods whose characters may be much more affected by Airbnb.
Manhattan also had several neighborhoods with a large percentage of apartments rented out on the platform: the Lower East Side, East and West Village, and Hell’s Kitchen are also most heavily affected by Airbnb rentals.
The sharing economy and the rental market
While a large percentage of housing units on Airbnb may change the character of a neighborhood, it does not give many clues about what effect Airbnb may have on rental prices in an area.
The simple solution might be to map Airbnb prices across the city. But mapping Airbnb prices in New York doesn’t reveal too many surprises (top map below). Apartments are much more expensive to rent in some of the most desirable residential neighborhoods, in Tribeca, Soho, Flatiron, and just south of Central Park. The cheapest Airbnbs were in more distant neighborhoods at the north end of Manhattan and a ring of Brooklyn from south end of Prospect Park through Crown Heights and Bed-Stuy to Bushwick.
To dig a bit deeper, we compare median Airbnb prices to median rent for every block. We find that clusters of high profitability in Manhattan center around Chinatown, East Harlem, and Manhattan Valley just west of Central Park on the west side. These are neighborhoods where the median Airbnb listing costs 25% of median rent – in other words, to be able to pay your rent by renting on Airbnb just four days of the month is easily possible! Williamsburg was the only large high-profitability cluster in Brooklyn, with a median Airbnb listing able to pay median rent in just five days.
These are neighborhoods where the median Airbnb listing costs 25% of median rent -- in other words, to be able to pay your rent by renting on Airbnb just four days of the month is easily possible!
Low profitability areas such as Crown Heights and Bushwick revealed an interesting pattern, perhaps a gap between desirability for residence and tourism. In very desirable parts of Brooklyn and Manhattan, like the Upper East Side, Park Slope, Cobble Hill, Ditmas Park, and Prospect Heights one would need to rent a median apartment at the median price for ten days or more to make rent. Considering the difficulty of renting the entirety of a month, this would mean it’s exceptionally hard in those areas to justify renting on Airbnb instead of to a long-term tenant.
Living with the sharing economy
For people interested in where Airbnb and its users are having the biggest impact on communities, these analyses provide good starting points. These maps allow us to focus on the relatively small clusters of blocks where high percentages of the available units are being rented on Airbnb, or where a financial incentive to rent out apartments may change the make-up of neighborhoods.
What is exciting about using these methods and data in CartoDB is that it allows us to work across the US all at once. In addition to New York City, we completed these same analyses for many other cities in the US, including San Francisco, Los Angeles, Nashville, Austin, Washington DC, Boston, Portland, Seattle, San Diego, and Chicago. Use the map below to search for any of those cities and see the Median Days to Make Rent with Airbnb.
If you are interested in learning about how location intelligence plays an essential role in the real estate sector and in data analysis, please join our webinar next Thursday, March 17.
Happy data mapping!