On May 13, 2021, the Financial Times published an article titled “Airbnb pricing algorithm led to increased racial disparities, study finds“. The initial version, now amended, led to believe that Airbnb’s free dynamic pricing tool, called Smart Pricing, was causing black hosts to make less money than white hosts. We looked for and read the actual research paper. For vacation rental managers and owners, the findings are thought-provoking and go beyond Airbnb. First, the paper shows that Smart Pricing increased host revenues, by lowering rates and boosting occupancy. Second, it demonstrates that the algorithm created benefits for black hosts who were using it. Yet, as too few black hosts adopted Smart Pricing, the prices of white-owned properties got better optimized. Last, the authors point out that the demand curve for white and black properties is different than a blanket pricing algorithm is not efficient: It is sub-optimal for everyone.
We’ll go beyond the conclusion of the paper to see what it means for vacation rental industry members. If you are using a dynamic pricing tool for your properties, you may now wonder whether the same algorithm can adapt to different demand curves among neighborhoods, property types, and luxury levels.
Airbnb and discrimination against black hosts and guests
It is not the first time that Airbnb’s platform has been found to enable discrimination. Because Airbnb is 2-sided marketplace where hosts and guests display their photo in their user profile, it is easier for researches to study it than on other platforms.
For instance, a 2017 Harvard research paper called “Racial Discrimination in the Sharing Economy: Evidence from a Field Experiment” found that booking requests on Airbnb from guests with distinctively African-American names were 16% less likely to be accepted relative to identical guests with distinctively white names.
In response, Airbnb has made some moves, such as enabling users not to display their profile photos before booking, if they wish. It has also created Project Lighthouse, an initiative to uncover, measure, and overcome discrimination when booking or hosting on Airbnb.
How low Smart Pricing adoption and diverging demand curves for black and white properties create revenue gaps
In March 2021, four researchers, Shunyuan Zhang (Harvard University), Nitin Mehta (University of Toronto), Param Vir Singh (Carnegie Mellon University), and Kannan Srinivasan (Carnegie Mellon University) published a paper called “Can an AI Algorithm Mitigate Racial Economic Inequality? An Analysis in the Context of Airbnb.”
Here’s our summary of their finings:
Airbnb’s Smart Pricing tool relies on an algorithm that prices nights up and down depending on demand.
A research paper on machine learning shows that Smart Pricing helps Airbnb hosts make more money. While the tool makes the average daily rate (ADR) go down by 5.7%, the average daily revenue increase by 8.6%. Occupancy and revenues go up thanks to prices more attuned to the shape of the demand curve.
Before Airbnb introduced the algorithm, black and white hosts charged similar prices for equivalent properties (in terms of observed host, property, and neighborhood characteristics), but white hosts earned $12.16 more in daily revenue than black hosts.
The revenue gap was caused by a difference in the occupancy rate (rental demand): 20% less for black hosts’ properties than for equivalent white hosts’ properties. The algorithm benefited black adopters more than white adopters, decreasing the revenue gap by 71.3%.
Yet, the arrival of the Smart Pricing tool increased the overall revenue gap between black and white hosts. Why? Because black hosts were 41% less likely than white hosts to adopt the algorithm. In other words: Because a lot of black hosts did not turn Smart Pricing on, the revenue gap of black hosts, as a population, increased compared with white hosts.
Yet, the paper findings do not stop here. It is not just an adoption problem. The researchers have established that the pricing algorithm is sub-optimal precisely because it is color-blind. Namely, the demand curve for white-owned properties and that for black-owned properties are different, according to the data they analyzed. They have found that black-owned properties are much more price sensitive. As a result, the color-blind pricing algorithm does not make their prices change fast enough, while it may be too sensitive to price white-owned properties in an optimal manner.
The paper opens the discussion: What can be done beyond helping everyone adopt tools and algorithms that should be beneficial to them? If the impact color-blind algorithm is to actually increase the gap between some racial groups because race is not factored into the parameters, what to do? Should race be included? Or socio-economical parameters that can be a proxy for races?
Food for thoughts for the vacation rental industry
Is the problem still happening? Note that the data set studied are from 2015-2017. Nowadays, when a user lists a property on Airbnb, Smart Pricing is turned on by default. It is actually quite a challenge to turn off, as Airbnb displays a series of warnings about potentially negative impacts if you try to do so. So, we assume that adoption of the Smart Pricing algorithm is more widespread.
Many of us in the vacation rental industry use dynamic pricing tools, which also rely on proprietary algorithms. This research paper mentions the issue of diverging demand curves between black and white-owned properties. As the Smart Pricing tool does not distinguish between them, its blanket algorithm optimizes prices in a way that not optimal. For neither blacks nor whites.
If you are a property manager with listings in both black and white neighborhoods, is the dynamic pricing tool you use good at pricing both areas at the same time? How about when you have a mix of low and high-end properties? This is a case when it is crucial that algorithms use data from the very property they are trying to optimize so that they can adapt to its booking patterns. It can be a good question to ask dynamic pricing software vendors. Share with them this article, ask them what they think of the findings, and get them to explain to you who their tool adapt to varying demand curves in your property portfolio.
If you want to read more about Airbnb’s response to discrimination against black hosts, Crain’s has a great article on the topic.