Fighting discrimination with reputation: The case of online platforms

On a large French ridesharing platform, new minority drivers earn 11.6% less revenue than otherwise similar nonminority drivers; the gap nearly vanishes as they accumulate reviews.

Reviews drive the convergence: when a railway strike exogenously raised demand and sped up review accumulation, minority entrants gained the most.

We explain the pattern with an estimated model of passenger choice and driver career concerns.

Passengers hold overly pessimistic priors about minority entrants - expecting substantially lower quality before the ride than they report after it.

As a result, minority drivers cut introductory prices and exert extra effort to overturn those beliefs quickly.

Counterfactuals show the cost of incorrect priors is high, and the reputation system strictly benefits minority drivers.