News | Research Spotlight: Modeling the Impacts of Ridehailing and Transit Fare Policy on Transit Ridership

Stop the Video



by Marlon Boarnet


When ridehailing (e.g., Uber, Lyft) became popular over ten years ago, scholars and transportation professionals wondered whether it would reduce or increase transit ridership. In short, is ridehailing a substitute or complement for public transit? Much more recently, a seemingly unrelated discussion has become popular: How can fare-free transit boost public transportation ridership compared to other policy tools? Our recent METRANS research sheds light on both questions.


Our research team (USC Professor and METRANS Director Marlon Boarnet and USC urban planning Ph.D. students Qifan Shao and Clemens Pilgram) estimated a mode choice model using travel diary data for the San Francisco Bay Area. The data includes 80,598 trips made by 3,732 persons in 2018 and 2019. We have data on the cost and travel time of each trip, using information on zone-to-zone travel times provided by the Bay Area Metropolitan Transportation Commission (MTC).


We fit a mixed-logit regression model to explain survey respondents’ travel mode choice from among five different trip modes: driving, ridehailing, transit, walking, and cycling. The results allow us to ask “what if” questions, examining what would happen to the shares of each travel mode if costs (prices) and travel times change. We used the model to answer two questions:


How does ridehailing compete with transit? The results show that if ridehailing costs increase by 10%, ridehailing trips would decrease by 6.97%. About half of the lost riders would switch to driving, and 20% of the lost ridehailing riders would switch to transit. Driving is the closest substitute for ridehailing, rather than transit. In our data sample, ridehailing competes with transit – but that effect is less than half as important as the way that ridehailing competes with driving for trips.


What happens when transit prices and travel time change? We modeled 10% reductions in transit travel time and transit cost (price, or fare). When each (travel time and cost) drop by 10%, the result is, respectively, a 24.63% and 5.03% increase in transit trips. If a transit agency could reduce travel times by 10%, the gain in ridership (almost 25%) would be five times as large as the ridership increase from the fare reduction. Of course, transit agencies must consider how easy it is to change fares and travel times. Reducing travel times might require running busses more frequently, possibly by hiring more drivers and buying more rolling stock. Our study did not examine those questions. But our results show that transit riders, in our data sample, are much more sensitive to changes in transit travel time than to changes in transit fare. The results suggest that focusing on improvements in transit travel time – more frequent and reliable service – might be a more effective way to increase ridership than focusing on fare reductions. Other issues, including equity, are crucially important, and might lead transit agencies to implement different policies.


This research was seeded by an earlier research project (Project No. PSR 19-11) funded by the Pacific Southwest Region (PSR) University Transportation Center, and while this work was unfunded, we benefited from the connections, support, and the intellectual milieu of the PSR UTC. Our results and interpretations do not reflect policy positions or opinions of any of the PSR UTC funders or of our colleagues at California metropolitan planning organizations who provided survey data (including the San Diego Association of Governments, who coordinated data access) or at the MTC who provided travel skim data. Our research was presented at the 2024 annual meetings of the Transportation Research Board and published in the journal Transportation Research Part D: Transport and the Environment. For the research article, see The full article can be accessed from most university libraries via that portal, and for questions contact the corresponding author, Marlon Boarnet, at [email protected].