Does Ride-sharing Discriminate?
Xavier Lambin and Emil Palikot
The internet initially promised a utopia, a place of robust exchange of ideas and free of the prejudices of the physical world. That, unfortunately, has not turned out to be the case; new research shows that minorities who are drivers on BlaBlaCar, the largest ride-sharing marketplace in Europe, have less popular listings, sell fewer seats, and generate less revenue than non-minorities. These findings add to a fast expanding body of empirical evidence showing that prejudice manifests itself in disparate economic outcomes for ethnic minorities on several major online marketplaces. This new research brings, however, some good news: those effects diminish as drivers receive reviews, which means that platforms and users have tools to counter discrimination.
BlaBlaCar currently operates in 22 countries, with eight million active drivers and 50 million passengers. Unlike Uber or Lyft, BlaBlaCar focuses on longer trips (the average is 400 km (approx. 250 mi)) and caters not to professional drivers, but simply to those who are looking to cover costs on a trip from A to B. To advertise a ride, a driver must complete a profile, which includes the driver’s first name, age, photo, and brief biography. Prospective passengers can see all this information before booking a seat with that driver. After a ride, the passenger rates the driver on a scale of 1-5 and leaves comments.
The relative anonymity of viewing a profile and not booking a ride with a driver allows passengers to make choices about drivers that are not based on merit, but on prejudice. This prejudice translates into fewer seats sold and thus less revenue generated for minority drivers. The economic consequences of it are especially severe for drivers new to the platform.
Minority drivers with five or fewer reviews receive significantly fewer clicks on their posts for upcoming rides than non-minorities, sell fewer seats, and generate less revenue. These drivers receive 11.8% less revenue than non-minority drivers with the same number of reviews. This gap decreases quickly, as minority drivers with between six and fifteen reviews earn 6.9% less, and the gap drops to 1.6% for drivers with forty or more reviews. A similar progression occurs for clicks and seats sold. In other words, BlaBlaCar’s reputation system helps minority drivers counter discrimination they initially face. But it is costly for them to do so, since they must expend more effort and incur greater costs than non-minorities to accumulate those forty reviews.
For example, they initially charge less per kilometer than non-minorities. They also include options — like allowing pets — to make their trips more appealing. These options are costly to drivers (e.g., the driver would need to ensure the car is clean after any ride with a pet), but minority drivers often use them in the hope that they will appeal to a larger audience.
Despite minority drivers’ initial hurdles, BlaBlaCar’s ratings system ultimately allows them to reduce the gap with non-minorities once they have enough positive reviews to demonstrate their quality. Indeed, a driver’s reputation, as measured by the number of reviews, has a positive and statistically significant effect on popularity of posted rides, sold seats, and revenue.
The paper also exploits the French railway strikes of 2018 to compare drivers using the platform before, during, and after the strikes and measure the effect of the strikes on BlaBlaCar drivers’ reputation, and in consequence their ability to escape prejudice. French railroad workers struck for three months in the spring of 2018, leading to a surge in demand for BlaBlaCar. In April 2018, for example, five million passengers booked rides, up from the 1.5 million monthly average. Drivers who used the platform on the day of the strike were able to fill their cars regardless of the level of prejudice they faced, and as a result received reviews, which led to higher profits after the strike. In other words, the strikes suggest a causal link between the number of reviews and improvement of economic performance for minority drivers.
Although prejudice formed offline transmits to many online marketplaces, reputation systems create a possibility to alter discriminatory treatment. This possibility materializes in the economic outcomes converging across ethnic groups only when minorities work harder and bear the significant costs of initial discrimination. These new results highlight the importance of a robust rating system for platforms like BlaBlaCar.