Rethink ride-hail: from understanding limits to reaching full potential

National Science Foundation Civil Infrastructure System Program, 2019-2022.

Transportation Network Companies (TNC) such as Uber and Lyft has reshaped traditional taxi services. Unlike traditional taxis where riders hail vehicles from street curbs, TNCs allow riders hail vehicles digitally through smart phones, or e-hail. It is however unclear how the performance of these two types of services (street hail vs. e-hail) compare with each other in terms of how well they are able to match riders with drivers and the wait times that riders must experience for the vehicles to arrive. This project will develop a general framework describing the matching between riders and drivers for both types of services, based on which, their respective performance will be compared. Policies and strategies will also be proposed in order to improve the performance of these services.  The project will develop educational materials and involve undergraduate and graduate students. Findings from this project will be published in academic journals and presented at international meetings.

The project proposes a novel and general model of the matching process between riders and drivers and validates it with empirical data.  This model describes spatial equilibrium between supply and demand in a mixed market that offers both e-hail and street-hail services and seeks to explain fundamental differences between the two types of services.  Parameters in the model will be calibrated from empirical data, allowing realism between the general model framework and the actual matching strategies implemented in the real world. The project will also analyze practical decision-making problems for service operators and regulators. Operators may be able to combine e-hail’s greater productivity with street-hail’s higher returns to scale.  The project results will also provide insights to regulators who are currently contemplating various policies such as entry cap and minimum wage. These decision problems, typically formulated as either Mathematical Programs constrained by Equilibrium Conditions or Markov Decision Processes, are large in both dimension and scale, highly nonlinear, and nonconvex.  Accordingly, the project also contributes to the analysis and solution of these challenging optimization problems.   Finally, the project develops a comprehensive evaluation framework that consists of model-based, simulation-based and application-based case studies.  These case studies can be used by other researchers and educators working in this field.