How do "controllable" host behaviors affect host success?
This scatterplot addresses the question: how do host behaviors (like response rate and including a profile picture) impact host and listing success (represented by total number of reviews)?
From this scatterplot, we can see that there is some positive associative relationship between host response rate and number of reviews, which is to be expected based on previous literature. In other words, the higher the host response rate, the higher the number of reviews; of course, we should note that there are many exceptions to this relationship.
There are a cluster of data points around 0% host response rate with a significant number of reviews, perhaps suggesting that these listings are so appealing due to factors like value or ratings that consumers would book and review the listing without needing to contact the host. Furthermore, having an excellent host response rate of close to 100% does not always lead to a high number of reviews, as signified by the dense cluster of points around a rate of 1.0 with less than 200 reviews. This indicates that hosts being very responsive to potential renters may increase the popularity of their listing, but host response rate clearly does not fully explain the variation in listing and host success.
Additionally, from this scatterplot, it is clear that the vast majority of hosts included in this dataset included a profile picture in their profiles, with only a dozen or so blue points, indicating that the host did not include a profile picture. The distribution of blue points in a nearly horizontal line across the bottom of the graph indicates characteristically low reviews, and strongly suggests that not including a profile picture will significantly decrease the popularity of a host’s listings, regardless of whether the host is responsive to renters.
From this plot, we can see that the hosts that can potentially benefit most from AirBnb are those who can take advantage of the boost that these ideal controllable host behaviors provide. However, we should note that having a very high response rate may be difficult for hosts who have other employment apart from being a host, and that those who feel a strong desire or need to maintain privacy and anonymity make not feel comfortable to provide a host profile picture; therefore, even these effective branding techniques may not actually be possible for some hosts.
Although “controllable” host behaviors clearly contribute to varying host success, we are also interested in whether and to what extent host success depends on uncontrollable factors such as bias based on perceived race and gender.
How does host presentation affect host success? What might this reveal about bias in the Airbnb market?
Because our dataset does not offer demographic information for race, level of education, or other background information, we tested other available variables in an effort to seek variation in host success. Specifically, by grouping hosts names into categories of more common names (host’s first name is listed over 15 times) and less common names (host’s first name is listed 15 or less times), we can observe that hosts with more common names have an average nightly price of $232.55 whereas hosts with less common names have an average nightly price of $212.87, nearly a $20 difference. With over 15,000 hosts included in each name category, it seems difficult to argue that this difference in average nightly price is due to chance.
Examining the names in the more common category reveals that they primarily consist of hosts with more mainstream, “white-sounding” names. Some of these more common host names include Amy (105 listings), Brian (85 listings), Sara (89 listings), and Paul (86 listings). These findings parallel a study conducted on New York Airbnb listings, in which non-black hosts were charging about 12% more than black hosts for considerably equal listings (Edelman and Luca). Here, hosts with more common names are charging about 9% more than hosts with less-common names.
This finding is important and relevant to our research question regarding why some Airbnb hosts are more successful than others. We can observe that white hosts are typically seen earning and benefitting more than non-white hosts for comparable listings. This elevates the prevalence of discrimination in online marketplaces, specifically Airbnb, and demonstrates how AirBnb fails to “share the market” to which they boast.
In order to gain a deeper understanding of hosts presentations within Airbnb, we generated a word cloud of host names, filtered on unique host ID values, in order to eliminate observations that refer to the same host but different listings. Textual analysis reveals that the top 5 most common host names are generally male and “white-sounding”–“David”, “Michael”, “John”, “Alex”, and “Mark”–despite the fact that LA is one of the most diverse regions in the U.S.
This supports our above argument that white-presenting hosts perform best on AirBnb, and further suggests that presenting male may also confer an advantage. Both the bar plot and the word cloud reveal that host success is not simply a function of “controllable” host behaviors such as high response rates or adding a profile picture, but may also be affected by bias towards certain host presentations based on perceived race and gender.
What is the distribution of ratings and tone of keywords in reviews? Why are Airbnb reviews overwhelmingly positive?
This is a word cloud of textual data from Airbnb reviews of LA listings in 2020, and a histogram displaying the overall distribution of review scores for Rating. The histogram confirms speculation from existing literature that, on a 1-100 scale, ratings are highly positively skewed towards 100. For the word cloud, stop words were removed automatically and manually excluded in order to highlight keywords within the word cloud. From this word cloud of the 45 most used keywords, we can see that not only are the numerical ratings for the Airbnb listings skewed very close to the maximum value of 5, there is also an overwhelmingly positive tone to the content of the reviews themselves. Out of the 45 keywords, 28 were neutral: including “place”, “house”, “host”, “stay”, “space”; 17 were distinctly positive: including “great”, “clean”, “nice”, “comfortable”, “perfect”; and none were distinctly negative. This word distribution suggests that it is not simply that renters are incentivized to give unusually high ratings to listings while leaving less favorable reviews, but that the reviews themselves also appear to be unusually positive. Moreover, the manner in which the histogram of review scores is skewed to the right may be reflective of how the reputation of hosts and the type of their amenities may affect a guest’s willingness to review, as well as the overall ratings (Zervas et al.).
According to Zervas et al., Airbnb reviews are niche in terms of how travelers leave reviews in comparison to hotels, as Airbnb ratings are dramatically more positive than those on well-established platforms due to the personal motives of hosts to give guests a 5-star experience. The Airbnb platform is also able to flush out poorer-performing hosts and listings through its comprehensive review process, leaving behind predominantly satisfied customers and successful listings. However, the criterion behind what qualifies a host as potentially poor-performing may push out less wealthy, non-white hosts from the platform, thereby opening the door for more success for white hosts. Given these two visualizations, we argue that these high reviews and consistent guest satisfaction play a role in enabling Airbnb’s rapid, reckless expansion, as hosts can typically expect that they will receive high reviews regardless of how touristy their Airbnb’s location is.
Why do the price and density of listings vary throughout LA County?
The colors represent neighborhood groups of Los Angeles, Other Cities, and Unincorporated Areas, while the sizes of the individual points represent the prices of the listings. This visualization reveals that Los Angeles has the highest density of listings, followed by other cities and unincorporated areas, perhaps indicating that the Airbnb market is flourishing more in LA than in surrounding regions. Furthermore, we can see that there is a wide assortment of both more and less expensive listings within the City of LA and other cities, as indicated by point size, with some very large points toward the center of the City of LA and smaller ones towards the outskirts of LA, revealing that listings increase in both quantity and price as one approaches an urban center like LA. However, for unincorporated regions, the points are significantly smaller throughout, indicating that these areas may not be as affluent and may not have a strong tourism and Airbnb market.
Moreover, we can see based on the coloring of the map itself that the regions where both the listings and prices are less densely concentrated also tend to have lower median household incomes, as reflected in their lighter coloration. Supply and demand for listings seem to be higher in the more affluent areas of LA County due to tourism, crime rates, increasing development, and increasing desirability of neighborhoods, as indicated by the number of restaurants or opportunities for entertainment. Similarly, in a study conducted about Airbnbs in Boston, Airbnb listing density was found to have a positive statistically significant association with rental prices (Horn and Merante). This study is relevant to our project because it confirms that the prevalence of Airbnbs is indeed having an impact on the rental economy and urban development. Specifically, as Airbnb supply and demand increase, long-term rental locations are converted into short-term rentals, and the supply and demand for restaurants and entertainment increase.
How do gaps and silences in License numbers reveal the lack of regulation of the short-term rental market?
Around 25,000 license numbers of hosts in this dataset have not provided a license number to accompany their listing. Only 15% of hosts have obtained a license number by registering with the city of LA. Around 5% of hosts have claimed exemption, meaning that the city provided an exception indicating that this listing is not required to obtain a license number. Finally, we can note that around 80% of hosts have not completed the registration process with their corresponding cities, despite recent LA laws that require that the host register and provide a license number.
According to the AirBnb website, some cities require that hosts obtain a license number in order to list their homes on AirBnb, but many cities do not have this requirement imposed. As we observed from the literature, this is problematic because a lack of registration makes it difficult for cities to regulate the short-term rental market and to keep this market from negatively impacting surrounding communities and long-term rental markets. In 2019, LA rolled out new regulations stipulating that all short-term rentals must be registered within the city in order to operate legally, and that each host can only list one property, and that property must be their primary residence. Thousands of listings were found to have violated these new restrictions, even a year after the laws had been implemented, according to the New York Times.
This graph, in partnership with relevant reportings, argues that the methods that the city of LA is currently implementing to enforce short-term rental registrations are not working. The city of LA must consider alternative methods of regulation in order to mitigate the potentially harmful effects of the flourishing short-term rental market, that benefit hosts with certain ideal presentations at the cost of vulnerable communities.
To what extent does Airbnb allow hosts to "share the market"?
Map based on average of Longitude and average of Latitude broken down by First Review Year. Color shows average of Price. Details are shown for Id. The view is filtered on Exclusions (Id,YEAR(First Review)) and First Review Year. The Exclusions (Id,YEAR(First Review)) filter keeps 31,535 members. The First Review Year filter excludes Null.
Looking at the chart which takes the latitude and longitude of Airbnb rentals from the last 10 years including their price range shows the mass expansion of Airbnb throughout California. Starting in 2009, there were not many Airbnb locations throughout the map and the rental prices were relatively cheap. Year by year, Airbnbs were growing exponentially and spreading across the map. What is surprising is that with the influx of new Airbnb listings, prices rose substantially. Usually from economics when there is more supply, demand may go down as well as price. However from the map above, since there are so many listings available now, competition arises because customers are looking for the most convenient rental for the best price. However this does not stop Airbnb listers from having huge rent prices for beach homes, mountain-hill homes, and etc. Furthermore, Airbnb’s expansion has allowed for a diverse set of housing options. Customers can now rent houses, apartments, condos, beach homes, villas, hotel rooms, and more. When Airbnb first came out, not all of these options were available which led to lower prices.
Which cities are most popular, as shown by total reviews (area) and total rent (color)?
Excluded were cities with reviews less than 3000 for clarity. The tree map data was gathered from 2008 to 2020. The bigger the square’s size for a city, the more the total number of reviews since 2008. The darker the blue color, the higher the sum total rent. Hovering the mouse on any city shows the total reviews and total sum rent price. Venice has the most reviews at 94,752. This shows Venice is very popular, being a tourist destination known for its beaches. Long Beach and Santa Monica on the left part of the tree map further highlights California visited for its coastal area. Hollywood has the third highest total reviews at 49,565 and is visited for its entertainment and arts culture.
However, even though Venice has the most reviews, it is Malibu that has the highest total sum rent at $635,833 (hover to see sum rent). This also shows for Beverly Hills, abbreviated as Bev. H. in the tree map on the far-right side. Beverly Hills has a total sum rent of $432,506 which is slightly more than Venice’s $408,338 even though Beverly Hills has roughly 1/10th of the reviews Venice has. From the data, the grand total sum for the cities residing in Los Angeles County, including the exclusions, is $7,022,150 and the grand total sum of all the reviews is 1,113,802 (summed in Excel). This shows Airbnb’s business is thriving.
How may an increase in Airbnb listings be associated with gentrification?
The listings are grouped by neighborhood, separated by colors. Each dot represents the average price of a listing. Hovering the cursor over a dot shows the rental’s city location and average listing price. The bigger the dot, the higher the rental price. Rental listings greater than or equal to $5,000 are represented by the same dot size for clarity in viewing the map. Notably, the coastal areas have multiple listings. Malibu (dark blue) has larger dots in comparison to the rest of the coast and most of the map, showing that the cost of living is higher and that there is a big tourism market. The dots along nearby Santa Monica (light blue) also have multiple listings, yet they are more clustered in comparison to Malibu. Additionally, these dots are smaller in comparison to Malibu. Long Beach (brown) has numerous, smaller dots that extend to the inland.
Of interest is there are more Airbnb listings in more affluent neighborhoods and less in poverty-stricken, high crime areas. There are less rentals in certain inland areas such as in Cerritos (gray), having a crime index of 8, showing it is “Safer than 8% of U.S. Cities” according to NeighborhoodScout (“CERRITOS, CA CRIME RATES”). In these areas there are fewer rental spots possibly due to safety reasons for both hosts and clients. Moreover, these few listings have extremely small rental prices further showing less client desirability. Compton (gray) also has few listings related to poverty. The United States Census Bureau cited a study claiming Compton has a high poverty rate of 21.9% (“QuickFacts – Compton city, California”). This shows gentrification that there is an adverse affect on the long-term housing market, with more short-term rentals appearing in place. in which wealthier individuals move in and drive out poorer individuals to unsafe neighborhoods with poor opportunities and lack of resources.
However, in other inland areas such as Beverly Hills (orange Bev. H.) and Hollywood (brown), there are a lot of listings showing it is the social environment and entertainment industry that attracts visitors. Again, more desirable listings leads to higher rent prices. Branching off, cities such as Pasadena (orange) and Glendale (brown) may not be as popular as the coastal cities, yet many Airbnb clients choose these places for price concerns as indicated by the much smaller dots. These areas also have a negative effect on the housing market as metioned eaarlier.
