Uber: A Platform for Discrimination?

As Uber tells it, Uber is not an employer but a technological platform connecting drivers and passengers.  This categorization – highly contested – has a number of well-known advantages for Uber: it allows the firm to avoid responsibility for minimum wage and overtime pay, for workers comp and unemployment insurance, and for providing healthy and safe working conditions. The categorization also allows Uber to avoid responsibility for discriminatory termination decisions.

Under Title VII of the Civil Rights Act of 1964, employers may not make hiring and firing decisions on the basis of race, sex, religion or national origin.  A key component of this anti-discrimination protection is that employers may not rely on customer preferences for employees of a certain race, sex, religion or national origin when making employment decisions.  So, for example, even if many or most customers of a particular restaurant would prefer white male waiters, the restaurant owner engages in illegal discrimination by hiring only white male waiters. And, of course, it is no answer to such a discrimination claim for the employer to assert, “it’s what the customers want.”  Indeed, to allow employers to discriminate because their customers prefer employees of a particular race or sex or religion or national origin would do enormous damage to Title VII.

The Uber platform, however, enables just this kind of customer preference discrimination.  As we’ve covered, Uber relies on a “star rating” system to determine when to terminate a driver’s access to the platform.  At the end of each ride, customers are asked to rate their drivers between one and five stars.  Uber compiles the driver’s overall star rating score, and if that score drops below a certain average (now about 4.6), the driver is terminated.  Thus, termination decisions at Uber are driven by customer preference.

We do not know, because we do not have access to the data, whether Uber riders discriminate on the basis of race, sex, religion or national origin when they rate their drivers, but as Nancy Leong has pointed out, we have some good reasons to expect that they do.  Indeed, there’s strong empirical evidence suggesting that consumers of services – including, specifically, taxi cab passengers – engage in race discrimination when they interact with service providers.  Ian Ayres, et al. studied the taxi cab industry in New Haven and found that the mean tip given to white taxi cab drivers was 20.3% while the mean tip given to black taxi cab drivers was 12.6%.  In an even more recent study of the restaurant industry, Brewster and Lynn find that customers who are served by a black waiter or waitress leave smaller tips (as a percentage of the bill) than those who are served by a white waiter or waitress.  If these dynamics are at play when Uber riders rate Uber drivers – and there is no reason to expect them not to be – then it’s likely that the star rating system is leading to discriminatory termination practices by Uber.  Put differently, the Uber platform provides a mechanism for translating customer bias (implicit or otherwise) into discriminatory termination decisions.

The battle over whether Uber meets the legal definition of “employer” is being fought out in courts and administrative agencies across the country.  If Uber is ultimately found to be an employer, then Title VII will apply and the firm will no longer be shielded from responsibility for discrimination even if that discrimination is a product of customer “preference.”  But Uber should not wait for this legal battle to play out and should instead take responsibility for discrimination now.  As Brishen Rogers rightly suggests, Uber’s access to star rating data positions the firm to figure out whether its customer review system is having a disparate impact on drivers and then to take steps to address any discrimination it finds.  I don’t claim that addressing bias will be easy; just that Uber should take responsibility for bias that its platform translates into discriminatory terminations.