Ulric B. and Evelyn L. Bray Social Sciences Seminar

This paper investigates the impact of algorithmic assignment on worker behavior and welfare within the ride-hailing industry. We demonstrate how algorithms can impose a exibility penalty on gig workers, despite their ostensible schedule autonomy. Utilizing rich transaction data from a leading ride-hailing company in Asia, we document a preferential assignment algorithm that favors drivers with longer working hours and consecutive hours during midday or late night. Drivers favored by the algorithm earn 8% more hourly than non-favored drivers. By constructing and estimating a two-sided market model, we quantify the welfare eects of such a preferential algorithm: Eliminating preferential assignment could raise ride fares by 7.79%, adversely aecting consumers and the platform. On the other hand, an additional 10% of drivers would switch to exible schedules, leading to a 3.51% surplus gain, especially bene ting young, male, and local drivers.

Joint with Yanyou Chen and Zhe Yuan