Three essays on ride-hailing markets

This dissertation consists of three chapters that study ride-hailing markets using an Uber price reduction experiment. The first chapter evaluates the congestion impacts of the price treatments. The second chapter focuses on the value of safety in transit mode decisions. The last chapter studies the heterogeneous Uber usage responses. In Chapter 1, I study the congestion demand response from a price reduction experiment. Despite the popularity of ride-hailing services, there is mixed empirical evidence on their congestion effects. In this chapter, I study the heterogeneous spatial and temporal demand responses related to congestion from an Uber experiment in Cairo, Egypt. The cost reduction experiment has a stronger volume response at congested hours and locations. Riders adjust the travel pattern with a higher probability of making trips in congested situations. Facing a 25 percent price reduction, the average rider makes 4.2 percent more trips to places 1 km/h slower. Conditional on a fixed location pair, the average rider makes 2.5 percent more trips in an hour which is 1 km/h slower. I then translate the trip volume increase into mobility benefit and congestion cost along the spatial and temporal dimensions. The comparison shows a large reduction in mobility benefits at busy locations during busy hours. In Chapter 2, co-authored with Adam Osman and Peter Christensen, we study how transit riders make transit mode decisions using an Uber experiment in Cairo, Egypt. While the cost reductions in Uber service increase Uber usage substantially, there is a strong substitution from Bus riders to Uber riders. To explore the underlying trade-offs, we build discrete choice models with the control function approach using multiple instrument variables to correct possible endogeneity problems. The analysis suggests a high value of time between 66 EGP to 76 EGP per hour and the value of safety harassment between 27 EGP to 31 EGP per unit increase in perceived level of safety. As the evidence suggests a higher value of safety for female riders, we suggest future transit policies to accommodate such preference heterogeneity and improve the safety level of public transit options. In Chapter 3, I exploit the experimental micro-level Uber data set to study transportation demand elasticity. Using the generic machine learning method, the chapter reveals a high level of heterogeneity in Uber usage facing significant price reductions. The most elastic riders show Uber responses twice as high as the average elasticity, while the least elastic riders show minimal to no effect. Being female, frequent Uber users and feeling unsafe on public transit creates substantial disparities among all characteristics. When the size of the discount increases, the effect of income becomes less critical. As the results indicate, price-related transit policies need to consider distributional effects. Better targeting policies could lead to a considerable increase in efficiency.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Economics

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2023

Author dc:creator

• Shen, Keshi

Contributors dc:contributor

• Christensen, Peter
• Kahn, Charles
• Lemus, Jorge
• Osman, Adam

• Ride-hailing
• Travel Demand
• Travel Safety
• Heterogeneous Demand

Statement dc:rights

• Copyright 2023 Keshi Shen

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/120388

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F120388