Workforce Scheduling in On-Demand Platforms
Team Information
Team Members
Raghav Singal, PhD Candidate, Industrial Engineering and Operations Research, School of Engineering and Applied Science, Columbia University
Faculty Advisors
Omar Besbes, Vikram S. Pandit Professor, Decision, Risk & Operations, Graduate School of Business, Columbia University
Vineet Goyal, Associate Professor, Industrial Engineering and Operations Research; and Member, Data Science Institute Columbia University
Garud Iyengar, Professor, Department of Industrial Engineering and Operations Research, Columbia Engineering
Abstract
Motivated by the discussion around drivers welfare in on-demand platforms, we provide a novel framework to understand the underlying forces at play, and prescribe operational changes to increase drivers welfare without hurting platform profit. We focus on a single-day horizon where the platform desires to gather a target supply level in each hour (“slots”). Too little on-road supply results in sub-optimal platform profit whereas too much supply results in lower effective wage for drivers. The platform employs an admission control policy to allocate the slots to the drivers and the drivers act strategically in order to maximize their expected utilities. The utility of a driver is driven by the allocation she receives, the time she spends on-road, and her private temporal preferences regarding when to drive. The platform does not know the drivers preferences a priori and hence, needs to implant the right incentives in the admission control mechanism. The mechanism design problem is to gather the target supply while maximizing the average effective wage. We use our framework to evaluate policies used in practice (first-come-first-serve (FCFS) and dynamic control (DC)). We establish that these policies can result in highly sub-optimal effective wage, even if the government imposes minimum wage regulation. Then, we propose a modification of the FCFS mechanism (sequential FCFS) and establish its optimality in markets with “enough” supply. We supplement our theory with simulations based on real data.
Contact this Team
Team Contact: Raghav Singal (use form to send email)