On the Empty Miles of Ride-Sourcing Services: Theory, Observation and Countermeasures

Abstract

The proliferation of smartphones in recent years has catalyzed the rapid growth of ride-sourcing services such as Uber, Lyft, and Didi Chuxing. Such on-demand e-hailing services significantly reduce the meeting frictions between drivers and riders and provide the platform with unprecedented flexibility and challenges in system management. A big issue that arises with service expansion is the empty miles produced by ride-sourcing vehicles. To overcome the physical and temporal frictions that separate drivers from customers and effectively reposition themselves towards desired destinations, ride-sourcing vehicles generate a significant number of vacant trips. These empty miles traveled result in inefficient use of the available fleet and increase traffic demand, posing substantial impacts on system operations. To tackle the issues, my dissertation is dedicated to deepening our understanding of the formation and the externalities of empty miles, and then proposing countermeasures to bolster system performance. There are two essential and interdependent contributors to empty miles generated by ride-sourcing vehicles: cruising in search of customers and deadheading to pick them up, which are markedly dictated by forces from riders, drivers, the platform, and policies imposed by regulators. In this dissertation, we structure our study of this complex process along three primary axes, respectively centered on the strategies of a platform, the behaviors of drivers, and the concerns of government agencies. In each axis, theoretical models are established to help understand the underlying physics and identify the trade-offs and potential issues that drive behind the empty miles. Massive data from Didi Chuxing, a dominant ride-sourcing company in China, are leveraged to evidence the presence of matters discussed in reality. Countermeasures are then investigated to strengthen management upon the empty miles, balance the interests of different stakeholders, and improve the system performance. Although this dissertation scopes out ride-sourcing services, the models, analyses, and solutions can be readily adapted to address related issues in other types of shared-use mobility services.