Abstract
We propose to investigate the economic and equity impacts of ride-hailing services like Uber and Lyft under normal and rare conditions, along with public policies that may enhance benefits and mitigate private and social costs and inequities. To do this, we will (1) leverage historical data to econometrically estimate the causal impact of Uber and Lyft market entry on wages and transit ridership during normal circumstances and the effect of rare changes in local policy and COVID-19 disruptions on operations and the distribution of riders served; (2) characterize demographic and geographic patterns of drivers and passengers that use ride-hailing services in Pittsburgh, Chicago and Austin, as well as shifts in other transit modes; (3) develop a deep understanding of driver, rider, and other stakeholder perspectives on the impacts of ride-hailing services using interviews and surveys; and (4) use simulation and optimization models to identify economic and policy incentives that could encourage efficient and equitable outcomes. Our holistic, equity-focused investigation will span normal operations as well as responses to rare events ranging from managing peak demand periods (e.g.: at sporting events) to managing disasters (including COVID-19 and city evacuations).
Description
MOBILITY 21 BIG IDEA THRUST AREAS ADDRESSED:
=========================================
7) Improved Transportation Access to Disadvantaged Neighborhoods
5) Novel Modes of Transport
4) Data Modeling and Analytical Tools
MOTIVATION
==========
Ride-hailing services from transportation network companies (TNCs), like Uber and Lyft, have revolutionized urban transportation. The benefits and costs of these changes are, however, distributed unevenly in society, and private ride-hailing firms have different incentives and procedures than public transit agencies for balancing economic and equity objectives in decision-making. For example, recent studies have found that some neighborhoods and individuals pay more for the same service, even when locations are as little as a few meters apart.[1,2] There is also inequality in driver wages by nature of origin and destination and whether the driver waits in place or cruises between rides.[3] Moreover, TNCs can both complement and displace public transportation, potentially improving or reducing access for groups that rely on transit.[4]
This prior work identifies some inequity impacts of TNCs on passengers and drivers on the TNC platform and substitute modes of transportation. What remains unknown is how the rise of TNCs has affected employment and wage opportunities across geographic areas and occupations, because access and pricing affect the degree to which some populations can use TNCs for gainful employment. Further, prior work typically does not leverage mixed methods, which integrate statistical assessment of trends with rigorous qualitative interviews and surveys of users, drivers, and stakeholders to understand perceptions and mechanisms underlying trends. Nor do these studies test possible interventions (whether imposed by regulators or proactively and voluntarily implemented by TNCs) that could help alleviate these issues, such as “equity pricing,” or various forms of rules, incentives and regulations. Even defining the appropriate equity objective for such interventions that balances system efficiency (net social welfare) with distributional impacts (who benefits and who is harmed) remains an open question. Finally, the role of ride-hailing in managing travel needs during rare events has been relatively unexplored. The impact of rare events, ranging from short-term (airport closure, earthquake) to medium-term (multi-day political convention, hurricane) to long-term (major bridge closure, COVID-19) on ride-hailing services, is expected to include changes on types of riders and rides served as well as drivers that will meet such demand.
RESEARCH TASKS
==============
To address these gaps in understanding the economic and equity effects of ride-hailing services and to identify economic, technological, and policy options for improving equity, we propose the following four tasks, spanning statistical analysis, qualitative methods, simulation, and optimization for normal operations as well as rare events.
TASK 1 US ECONOMETRIC ANALYSIS:
---------------------------------------------
Assess the economic and equity impacts of ride-hailing using modern econometric datasets and techniques. [Leads: Armanios, Michalek]
• Task 1a (normal operations): Exploit staggered entry of Uber and Lyft across US cities to perform a difference-in-difference regression to isolate the effect of TNC entry on employment and wages (both in the aggregate and broken down by occupation) as well as transit use.[18-20]
• Task 1b (rare events): Exploit (1) new regulations promoting ride pooling in downtown Chicago and (2) the onset of COVID-19 to perform a difference-in-difference and a regression discontinuity analysis to isolate impacts of policy and pandemic conditions on ride-hailing operations, transit use, and distribution of neighborhoods served. Here, the treatment onset includes (1) policy enforcement and (2) first confirmed COVID-19 case. For robustness we will also examine anticipatory changes.[13]
We will use wage and economic data from a variety of sources such as the Quarterly Census of Employment and Wages (QCEW)5, the Current Employment Statistics (CES)[6], as well as the Occupational Employment Statistics (OES)[7,8] databases. To ensure sharper comparisons between occupations, we will use standardized scales from the Occupation Information Network (O*NET) data to match similarly skilled occupations[9-12].
TASK 2 DEEP DIVE CITY DATA ANALYSIS:
------------------------------------------------
Leverage high-resolution origin-destination ride-hailing data in Pittsburgh, Chicago and Austin to characterize how TNC trips vary by neighborhood (i.e., census tract) geography, income, race, ethnicity, and major occupation. Assess the degree to which costs and benefits of TNCs identified in Task 1 are equitably distributed. [Leads: Nock, Michalek]
• Task 2a (normal operations): Identify trends in this high-resolution city data amidst normal operation periods
• Task 2b (rare events): Identify changes in trends during rare events, including the SXSW festival in Austin and the onset of COVID-19 in Pittsburgh and Chicago.
During the project period we will also explore the potential to expand the analysis to other cities via the We will use COVID-19 data by county and city from a variety of sources such as C3-AI14 and USAFACTS15. SharedStreets initiative (https://sharedstreets.io/) as data become available.
TASK 3 QUALITATIVE METHODS:
--------------------------------------
Conduct interviews and surveys to inform and ground validate our findings around economic (Task 1) and equity (Task 2) outcomes under normal operations and rare events. We will leverage previously-piloted methods and conduct interviews and surveys in-person and/or remote, depending on the status of COVID-19. [Leads: Armanios, Nock]
• Task 3a (normal operations): Conduct interviews and surveys to contextualize disparities in access, security, employment, and other factors as well as perceived reasons for such issues. Pending IRB approval and informed consent, we will also work with TNC divers on possible scenarios that could foreseeably occur when seeking to redress equity and employment disparities.
• Task 3b (rare events): We plan to conduct interviews and surveys with TNCs, drivers, and riders to gauge whether COVID-19 has changed dynamics and disparities in access, security, employment, and other factors as well as perceived reasons for such issues. Again, pending IRB approval and informed consent, we will also work with TNC divers on possible scenarios that could foreseeably occur when seeking to redress challenges that occur during rare events whereby stress is both on the user and the driver (e.g., evacuation of regions with low car ownership).
TASK 4 SIMULATION AND POLICY ASSESSMENT:
----------------------------------------------------------
Simulation modeling to test a variety of economic conditions, TNC policy, and public policy interventions that may help alleviate equity and employment disparity issues. These would incorporate the previously derived economic (Task 1) and equity (Task 2) model estimates as well as the ground validated relationships between equity and economic impact (Task 3). [Leads: Michalek, Nock]
• Task 4a (normal operations): Leverage and refine existing models at CMU[16] to simulate TNC operations and observe how alternative policy proposals affect both TNC operations and outcomes. Such possible interventions could include congestion and air pollution taxation, equity pricing (like surge pricing but for underserved areas), ride quotas, ride subsidies, or a dedicated portion of TNC fleets that purely operate in underserved areas. As part of these simulation models, we would need to consider (1) managing cost, environmental sustainability, and equality objectives, (2) undue burden on minority populations (e.g., greater wait times in low-income and minority neighborhoods), (3) system efficiency (including economic efficiency), and (4) unintended consequences.
• Task 4b (rare events): Expand aforementioned existing models at CMU16 to include current COVID essential worker transportation modeling to simulate TNC operations and observe how alternative policy proposals affect both TNC operations and outcomes amidst the pandemic. Such possible interventions could be subsidies or contract duties for drivers to serve users during emergencies or dedicated fleets of medical professionals taking users with potential symptoms (akin to Germany’s “corona taxis”[17]). Besides considering the four factors above in simulation models of normal operations, rare events have the added complexity of potential unwillingness of drivers to serve during rare events (i.e., driver compliance). We will capture driver compliance risks associated with rare events using compliance rate estimates from Task 3 as constraints in the model.
REFERENCES
===========
1 Pandey, A. & Caliskan, A. Iterative Effect-Size Bias in Ride-hailing: Measuring Social Bias in Dynamic Pricing of 100 Million Rides. ArXiv (2020).
2 Chen, L., Mislove, A. & Wilson, C. Peeking Beneath the Hood of Uber. Imc'15: Proceedings of the 2015 Acm Conference on Internet Measurement Conference, 495-508, doi:10.1145/2815675.2815681 (2015).
3 Bokanyi, E. & Hannak, A. Understanding Inequalities in Ride-Hailing Services Through Simulations. Sci Rep-Uk 10 (2020).
4 Ward, J. et al. Uber’s impact on vehicle ownership, fuel economy, and transit across US cities (Carnegie Mellon University, 2020).
5 U.S. Bureau of Labor Statistics, Washington, D.C., 2020
6 U.S. Bureau of Labor Statistics, Washington, D.C., 2020
7 Acemoglu, D. & Autor, D. Skills, Tasks and Technologies: Implications for Employment and Earnings. Hbk Econ 4, 1043-1171, doi:10.1016/S0169-7218(11)02410-5 (2011).
8 U.S. Bureau of Labor Statistics. Occupational Employment Statistics, <https://www.bls.gov/oes/home.htm> (2019).
9 Acemoglu, D. & Autor, D. Acemoglu and Autor (2011): HLE (Data Archive), <https://perma.cc/B7SK-VKUV> (2011).
10 Autor, D. H. The polarization of job opportunities in the US labor market: Implications for employment and earnings. Community Investments 23, 11-16 (2011).
11 Autor, D. H. The "task approach" to labor markets: an overview. Journal of Labour Market Research 46, 185-199 (2013).
12 Poliquin, C. The Wage and Inequality Impacts of Broadband Internet. Working Paper (2020).
13 Holshue, M. L. et al. First Case of 2019 Novel Coronavirus in the United States. New Engl J Med 382, 929-936 (2020).
14 ed C3.AI (2020).
15 ed USAFACTS
16 Bruchon, M., Michalek, J. & Azevedo, I. Effects of air emissions externalities on optimal ride-hailing fleet electrification and operations (Carnegie Mellon University, 2020).
17 Bennhold, K. in New York Times (2020).
18 Ward, J.W., J.J. Michalek, I.L. Azevedo, C. Samaras, P. Ferreira (2019) "Effects of on-demand ridesourcing on vehicle ownership, fuel consumption, vehicle miles traveled, and emisssions per capita in U.S. states," Transportation Research Part C: Emerging Technologies, v108 p289-301.
19 Ward, Michalek, Samaras, Azevedo, Henao, Rames, Wenzel, Gillingham “Uber’s Impact on Vehicle Ownership, Fuel Economy & Transit Across US Cities,” working paper, Carnegie Mellon University. [in review at iScience]
20 Ward, Michalek, Samaras “The air pollution, greenhouse gas, and traffic externality costs and benefits of shifting private vehicle travel to ride-hailing.” Working paper, Carnegie Mellon University. [in review at PNAS]
Timeline
We propose a 3-year plan, with the first 1.5 years funded by the Mobility21 Big Idea project funds. The attached Gantt chart summarizes the plan and identifies the PhD student lead for each subtask. Prof. Armanios and Prof. Michalek will serve as lead advisors for Adam Kolig, and Prof. Nock will serve as lead advisor for Lily Hanig. Prof. Davis will support both students, with a focus on survey and data analysis.
In the first year, Koling will begin the econometric studies of Task 1 while Hanig studies detailed data on changes to TNC operations and populations served following the COVID-19 outbreak in Chicago and begins to integrate knowledge into a simulation for rare event responses. Both students will work together to set up interviews and surveys to collect information from TNCs, drivers, riders, and other stakeholders that will inform modeling and simulation work.
In the second year, Koling will complete and publish the econometric tasks and the study of detailed data for Chicago and Austin and begin simulation and optimization modeling for normal TNC operations while Hanig completes her analysis of rare events in Chicago and Austin and continues interview work to support simulation/optimization for rare events.
In the final year, Koling and Hanig will complete interview work and use it to inform, finalize and publish the simulation and optimization studies. The team will plan and execute policy outreach to local, state, and/or federal policymakers, as appropriate, depending on findings.
Additional participating students Andreason, Bruchon, and Forsythe, funded on other projects, will also contribute to the proposed project in ways that enhance our capabilities by leveraging synergy other work without being reliant on students funded outside the Mobility21 Big Idea project for success. In particular, Andreason, Bruchon and Forsythe will all support Task 2b, and Bruchon and Forsythe will support Task 4a.
Strategic Description / RD&T
Deployment Plan
Our deployment partner, the Port Authority of Allegheny County, is interested in understanding the economic and equity issues of ride-hailing services as a piece of urban mobility, as they both complement and compete with transit. While public transit agencies like Port Authority have processes to explicitly consider and balance effectiveness, efficiency, and equity for any service changes considered, private for-profit ride-hailing services do not necessarily need to balance these factors in a similar way, potentially resulting in different outcomes and leaving gaps for disabled riders, lower income communities, and other groups. Port Authority is interested in using the results of this research to help inform strategies for how they interact with ride-hailing services in Pittsburgh and what kinds of policies could potentially improve equity outcomes without severe negative unintended consequences. Port Authority has also experienced massive changes in ridership with the COVID-19 pandemic, and understanding changes in ride-hailing trends during this period can help Port Authority gain a fuller picture of risks and opportunities for the transportation system and to inform planning of fleet schedules.
We are also in discussion with a firm that provides ride-hailing services and has agreed to support and participate in Mobility 21 research. Stan Caldwell is aware of the details.
Expected Outcomes/Impacts
Year 1 Milestones:
==============
1. Task 1: Data acquired and cleaned, model specified, initial estimation complete and shared with deployment partner for feedback. Submit Task 1a for publication.
2. Task 2: Obtain, clean, and conduct first analysis of city-level data.
3. Task 3: Scope and design first round interview and survey strategy. Design and obtain IRB approval. Implement first round of qualitative analysis targeted to inform year 1 tasks.
4. Task 4: Begin constructing simulation models for rare events.
Year 2 Milestones:
==============
1. Task 1: Finalize and publish econometric analyses. Disseminate to deployment partner, TNCs, media, and policymakers.
2. Task 2: Finalize and publish assessment of city-level trends.
3. Task 3: Scope and design second round interview and survey strategy. Design and obtain IRB approval. Implement second round of qualitative analysis targeted to inform year 2 tasks.
4. Task 4: Continue development of simulation models for rare events and begin constructing optimization and simulation models for normal operations leveraging prior CMU work in this area.
Year 3 Milestones:
==============
1. Task 3: Scope and design final round interview and survey strategy. Design and obtain IRB approval. Implement final round of qualitative analysis targeted to inform year 3 simulation and optimization tasks.
2. Task 4: Complete and publish simulation, optimization, and policy analysis tasks. Disseminate findings and recommendations to deployment partner, TNCs, media and policymakers.
3. Develop and implement a strategy to disseminate key findings and recommendations from the entire project to decision-makers, which may include policy briefs, videos, media outreach (e.g.: op-ed), and meetings with key decision-makers in industry and government.
Expected Outputs
TRID
Individuals Involved
| Email |
Name |
Affiliation |
Role |
Position |
| candreas@andrew.cmu.edu |
Andreasen, Charlotte |
Carnegie Mellon University |
Other |
Student - Undergrad |
| darmanios@cmu.edu |
Armanios, Daniel |
Carnegie Mellon University |
Co-PI |
Faculty - Untenured, Tenure Track |
| mbruchon@andrew.cmu.edu |
Bruchon, Matthew |
Carnegie Mellon University |
Other |
Student - PhD |
| alexdavis@cmu.edu |
Davis, Alex |
Carnegie Mellon University |
Co-PI |
Faculty - Untenured, Tenure Track |
| cforsyth@andrew.cmu.edu |
Forsythe, Connor |
Carnegie Mellon University |
Other |
Student - PhD |
| lhanig@andrew.cmu.edu |
Hanig, Lily |
Carnegie Mellon University |
Other |
Student - PhD |
| adam.koling@gmail.com |
Koling, Adam |
Carnegie Mellon University |
Other |
Student - PhD |
| jmichalek@cmu.edu |
Michalek, Jeremy |
Carnegie Mellon University |
PI |
Faculty - Tenured |
| dnock@andrew.cmu.edu |
Nock, Destenie |
Carnegie Mellon University |
Co-PI |
Faculty - Untenured, Tenure Track |
Budget
Amount of UTC Funds Awarded
$410988.00
Total Project Budget (from all funding sources)
$1025397.00
Documents
| Type |
Name |
Uploaded |
| Data Management Plan |
Data_Management_Plan_fBF9Bik.pdf |
Sept. 30, 2020, 10:36 a.m. |
| Presentation |
Slides-BigIdea-Michaleketal_aV8mLAs.pptx |
Oct. 1, 2020, 6:38 p.m. |
| Publication |
Plug-in hybrid electric vehicle LiFePO4 battery life implications of thermal management, driving conditions, and regional climate. |
Dec. 27, 2020, 10:07 p.m. |
| Publication |
Consistency and robustness of forecasting for emerging technologies: The case of Li-ion batteries for electric vehicles. |
Dec. 27, 2020, 10:08 p.m. |
| Publication |
A Tale of Two Policies: Interactions of US Federal and State Policies for Promoting Alternative Fuel Vehicles Increase in Greenhouse Gas Emissions |
Dec. 27, 2020, 10:09 p.m. |
| Publication |
On the implications of using composite vehicles in choice model prediction |
Dec. 27, 2020, 10:10 p.m. |
| Publication |
On-demand ridesourcing has reduced per-capita vehicle registrations and gasoline use in US States |
Dec. 27, 2020, 10:11 p.m. |
| Publication |
Effects of on-demand ridesourcing on vehicle ownership, fuel consumption, vehicle miles traveled, and emissions per capita in US States |
Dec. 27, 2020, 10:12 p.m. |
| Publication |
Alternative-fuel-vehicle policy interactions increase US greenhouse gas emissions |
Dec. 27, 2020, 10:12 p.m. |
| Publication |
Choice at the pump: measuring preferences for lower-carbon combustion fuels |
Dec. 27, 2020, 10:13 p.m. |
| Publication |
The economics of utility-scale portable energy storage systems in a high-renewable grid |
Dec. 27, 2020, 10:14 p.m. |
| Progress Report |
350_Progress_Report_2021-03-31 |
March 31, 2021, 7:47 a.m. |
| Progress Report |
350_Progress_Report_2021-09-30 |
Sept. 29, 2021, 2:56 p.m. |
| Publication |
The impact of Uber and Lyft on vehicle ownership, fuel economy, and transit across US cities |
Oct. 24, 2021, 8:36 p.m. |
| Publication |
Effects of Air Emission Externalities on Optimal Ridesourcing Fleet Electrification and Operations |
Oct. 24, 2021, 8:38 p.m. |
| Publication |
Air Pollution, Greenhouse Gas, and Traffic Externality Benefits and Costs of Shifting Private Vehicle Travel to Ridesourcing Services |
Oct. 24, 2021, 8:40 p.m. |
| Progress Report |
350_Progress_Report_2022-03-30 |
March 30, 2022, 7:08 a.m. |
| Final Report |
350_-_Mobility21_Final_Report.pdf |
Aug. 4, 2022, 8:42 a.m. |
| Progress Report |
350_Progress_Report_2022-09-30 |
Oct. 6, 2022, 3:51 a.m. |
| Publication |
Estimating global demand for land-based transportation services using the shared socioeconomic pathways scenario framework |
March 30, 2023, 6:06 a.m. |
| Publication |
COVID-19 public transit precautions: Trade-offs between risk reduction and costs |
April 10, 2023, 8:13 p.m. |
| Publication |
What Stay-At-Home Orders Reveal About Dependence on Transportation Network Companies |
April 10, 2023, 8:14 p.m. |
| Publication |
Analyzing and Optimizing Shared Mobility Fleet Impacts |
April 10, 2023, 8:15 p.m. |
Match Sources
No match sources!
Partners
| Name |
Type |
| Port Authority of Allegheny County |
Deployment Partner Deployment Partner |
| TNC |
Deployment Partner Deployment Partner |
