In January, researchers released a paper regarding Airbnb and tourists visiting local neighborhoods.
Airbnb supports rigorous research and analysis to help create shared learnings about the impact of short-term rentals on neighborhoods and communities.
Unfortunately, this paper does not meet such a standard. Rather, it uses an unrepresentative sample within one city to make broad nationwide conclusions; applies a faulty methodology including a flawed regression analysis; and relies on inaccurate data. The result is a paper with inaccurate conclusions not supported by the evidence.
In fact, the authors’ own robustness check appears to indicate that a rise in crime could occur before or after an increase in Airbnb listings in a neighborhood, which raises the basic question as to how the paper could reach any such conclusion for these districts within Boston, let alone for the city of Boston, never mind serving as the empirical basis to support a nationwide finding.
Beyond the issue of the nature of the sample, there are a number of key issues related to the methodology that lead to inaccurate conclusions.
Generalizing findings from a single city
The study examines a handful of districts in Boston only, and then extrapolates that into nationwide generalizations without providing detailed evidence. This would be akin to saying that because Northeastern University’s enrollment rate has gone up, then surely the same is true for Harvard, Suffolk, Bentley, Emerson, Bunker Hill Community College, Boston University and MIT – in addition to Stanford, Georgetown and Texas A&M – without doing the work to see if that broader conclusion is actually true.
Using research extrapolated from trends in a small number of districts in Boston, the authors claim to find a general causal relationship between the presence of Airbnb listings and incidents of violent crime. However, the analysis on which their methodology relies fails to prove that the presence of Airbnb listings in a high number of buildings in a neighborhood (what the authors call “Airbnb penetration”) precedes a rise in violent crime.
Moreover, the authors’ methodological approach includes a series of errors, including:
- The variables that may be correlated with Airbnb growth at a tract level are not controlled for in this paper. These include (but are not limited to) new housing construction and overall economic conditions.
- Other control variables that the authors did include, such as median household income, are measured using survey data based on five-year estimates, even though the analysis is conducted at a yearly level. The standard of the peer review process applied to this paper was far below what is usually applied in academic social science, which raises significant concerns about the academic quality of the work.
- The authors relied on a method (“Granger Causality”) that is inappropriate for establishing actual causality. This approach is meant to establish whether one event or action sequentially precedes another — not whether one causes the other. Not only is the approach not truly causal (defined as causing a result), but it fails to prove that an increase in the presence of Airbnb listings across a neighborhood precedes the rise of violent crime over time. Indeed, as noted above, the authors’ own robustness check would appear to indicate that a rise in crime could occur before or after an increase in Airbnb listings in a neighborhood, raising fundamental questions about their findings.
The authors also rely on scraped data and make erroneous assumptions to arrive at the core data set on which their research depends.
In particular is the reliance on the field, “Joined in,” which appears on every Airbnb user’s profile page. The “Joined In” field shows when an individual began using Airbnb, as either a Host or a guest. It appears the authors used this information to presume when an individual began hosting on Airbnb. This is problematic and leads to flawed data.
To provide an example for why this is problematic:
- A person could have created an Airbnb account in 2013 to travel as a guest.
- That same person could have begun hosting on Airbnb in 2016.
- Their Airbnb profile on the listing page would therefore read, “Joined in 2013,” which the authors of this paper would have then used to presume the creation date of that Host’s listing, despite the fact that it was three years away from being in existence.
Reliance on the “Joined in” field is also flawed due to the fact that a Host could have had multiple listings, one of which was live in an earlier year in a different location, but not the listing being studied.
Over the last year-plus, various reports indicate the unfortunate trend that crime has risen in many American cities (even as travel and tourism, including short-term rentals, have suffered in urban markets). There are many complex issues that contribute to rises and falls in crime rates that require serious research to help inform and guide public policy work. As a society, we should be working to advance serious research. In this context, Airbnb will be formally reaching out to Northeastern University to express our concerns about the lack of academic rigor in this paper and learn more about the protocols the University applies to assure the quality of the research performed by those associated with the school.
While the approach taken by these authors resulted in conclusions not supported by the facts or the methodology, we continue to welcome the opportunity to work with those who want to approach these important issues with a seriousness of purpose.
Airbnb was born in 2007 when two Hosts welcomed three guests to their San Francisco home, and has since grown to 4 million Hosts who have welcomed more than 900 million guest arrivals across over 220 countries and regions. Travel on Airbnb keeps more of the financial benefits of tourism with the people and places that make it happen. Airbnb has generated billions of dollars in earnings for Hosts, 90 percent of whom are individuals listing their own homes, more than half of whom are women, and one in five employed Hosts are either teachers or healthcare workers. In 2019, Airbnb directly supported 300,000 jobs in just 30 destinations, averaging nine jobs for every 1,000 guest arrivals. Travel on Airbnb also has generated more than $3.4 billion in tax revenue for 29,000 jurisdictions around the world. Airbnb has helped advance more than 1,000 regulatory frameworks for short-term rentals, including in 70% of our top 200 geographies (pre-pandemic). In late 2020, to support our continued expansion and diversification, we launched the City Portal to provide governments with an automated one-stop shop that supports data sharing and compliance with local registration rules. We continue to invest in innovations and tools to support our ongoing work with governments around the world to advance travel that best serves communities.