A new report alleges that Airbnb manipulated data to make itself look better to regulators

Airbnb’s efforts to become an “open and transparent” company might not quite line up with traditional definitions of transparency.

A report released Wednesday suggests that Airbnb manipulated listing data in New York City prior to publicly releasing that data to regulators in December, in an effort to quell concerns that the platform was enabling professional hosts who made a living off illegal short-term rentals. In the weeks leading up to the day Airbnb took its data sample, the report alleges, Airbnb deleted over 1,000 apartments belonging to hosts with multiple apartment listings in New York City from the site.

The report was published by Tom Slee and Murray Cox,  two technologists who have made a hobby of investigating the practices of Airbnb.

“The intervention was so specific, and the timing so close to the date of the New York City snapshot, that the conclusion is inescapable,” they wrote in the report. “Airbnb removed listings from its site so that its data set would paint a more attractive picture of its business, to better influence media and public opinion. It was a cover-up, not a move to transparency.”

In response to the report, Airbnb provided Fusion with the following statement:

“The facts are clear for all to see—the vast majority of our hosts are everyday people who have just one listing and share their space a few nights a month to help make ends meet. Airbnb is an open people-to-people platform where listings come on and go off throughout the year. We’ve also done significant work to educate our community about what is in the best interest of their city and we routinely review our listings to ensure guests are having the quality, local experience they expect and deserve.”

Airbnb also sent Fusion the current number of hosts with a single listing, which grew from 16,886 on the day in November to 18,010 this month. The percentage of hosts with a single listing decreased at the same time, from 95 to 94 percent, which means the number of hosts with multiple listings actually increased at a greater rate. An Airbnb spokesperson noted via e-mail that these numbers are “basically the same.” Airbnb did not respond to a request from Fusion to provide any numbers from earlier in November, before the alleged purging of listings.

When Airbnb released the data in December, it sought to debunk the idea that it was a platform for illicit apartment rentals, crowding out legitimate renters in the already overcrowded New York City market. Airbnb published a snapshot of data taken on November 17 that found 95 percent of entire home hosts only had one listing. Further, 75 percent of revenue from hosts with entire home listings from the previous year had come from people with only one or two rental listings on the platform. Over 2015 to 2016, Airbnb projected that number would rise to 93 percent. This seemed like a significant improvement over a report the previous year from the New York State attorney general’s office that claimed that nearly three-quarters of Airbnb rentals in the city were illegal.

“The vast majority of the community is doing this in the right way,” Chris Lehane, Airbnb’s head of global policy, told The New York Times.

Slee and Cox arrived at their own conclusions after separately scraping public data from Airbnb’s listings over a period of many months. Slee told me that when Airbnb released his public data, he was curious to compare it with his own data. It was then that he noticed the huge dip in November listings belonging to hosts with multiple listings on the site. Slee has never met Cox in person, but knew that he also regularly scraped Airbnb’s listings data. He shot him an e-mail. Cox had found the same thing. Cox, he told me, doesn’t even normally take data samples mid-month, but had because he had heard rumors that New York City hosts were being kicked off of Airbnb.

Digging deeper, they found that:

• While Airbnb said that only 10 percent of entire homes listings belonged to hosts with multiple listings, the number was probably closer to 19 percent.
• Where Airbnb claimed that 95 percent of entire home hosts only had one listing, that had actually only been true for two weeks of the year.
• The company’s future projections were based on a single day that appeared to be an anomaly in the data.

While Slee and Cox admit that it’s impossible from to tell why a listing disappeared from Airbnb’s platform, their data does show that in November their was a significant uptick in listings belonging to “multiple-lister hosts” that left:

I asked Slee whether something else could be responsible for such a data anomaly. He told me that it was unlikely, since Cox and Slee came to the same conclusions independently and had not found similar patterns in any other cities or at any other point in time.

“We are very confident that the data is accurate. It does look like Airbnb is being dishonest,” he told me. “Airbnb chooses to emphasize ‘percent of hosts’ because it looks good for them. It diverts from more important measures such as the percent of listings or the percent of revenue. The small percentage of multiple-listing hosts generates a much larger percentage of Airbnb’s business.”

Asked again about the accuracy of the Slee and Cox’s report and the allegations of data manipulation, Christopher Nulty, Airbnb’s public affairs head for Eastern North America, responded with another statement:

“Our community in New York has evolved to a point where 94 percent of hosts have just one listing and where there is no material presence of illegal hotels, which is why accusations from the same elected officials who called for there to be no illegal hotels on the platform and now want to fine middle class families $50,000 is akin to asking someone to walk on water and then, when they do, fining them for not swimming.”

Slee and Cox estimate that about 30 to 40 percent of revenue comes from hosts with multiple listings. The New York Attorney General’s report from 2014 similarly found that 38 percent of revenue came from hosts illegally renting units for more than half the year. Slee noted that the number of multiple-listing hosts had already started to creep upward again, a conclusion that was supported by the additional data that Airbnb shared with Fusion.

Airbnb’s data release, Slee and Cox suggest, was merely a clever numerical slight of hand, not an indication of transparency or openness.

“Airbnb is a $25 billion company, its success or failure depends on getting cities around the world to change their rules and regulations. If they’re going to do that they have to be a trustworthy,” Slee said. “To me this kind of representation is a very bad sign for Airbnb’s trustworthiness.”