TL;DR
This paper investigates the complex relationship between ride-hailing services and public transport, moving beyond a simple "compete or complement" dichotomy. By analyzing Uber trip data across six cities, the study reveals that ride-hailing offers significant time savings for many trips, especially in areas underserved by public transport, thereby increasing overall accessibility.
What Problem Does This Paper Address?
The rise of ride-hailing services like Uber has sparked a heated debate: do they primarily steal passengers from public transport (competing), or do they fill gaps and encourage car-free lifestyles, thus benefiting public transport (complementing)? Existing research has offered conflicting conclusions, often relying on surveys with limited scope or assumptions about travel times and costs. This paper aims to provide a more robust, data-driven understanding by quantitatively comparing ride-hailing and public transport across multiple dimensions (time, cost, accessibility) and exploring how these factors influence demand shares in different urban contexts.
Methods
The researchers employed a multi-faceted approach:
- Trip Matching: For each Uber trip, they identified the "fastest" public transport (PT) alternative using Open Trip Planner, considering origins, destinations, and request times. This involved building detailed PT schedules and walkable networks.
- Journey Metrics Calculation: They calculated three key metrics for both Uber and PT trips:
- Nominal Travel Time (T): The actual time spent on the trip (including waiting, walking, and in-vehicle time).
- Generalized Travel Time (G): Nominal travel time adjusted for the perceived discomfort of different stages (e.g., waiting is more uncomfortable than in-vehicle time), and including transfers for PT.
- Generalized Travel Cost (C): Generalized travel time converted into monetary cost using the local value-of-time, incorporating trip fares.
- Modal Accessibility Gap (MAG): This index quantifies the difference in accessibility between PT and ride-hailing, normalized to range from -1 (ride-hailing is much more competitive) to +1 (PT is much more competitive).
- Service Accessibility Analysis: They used a hexagonal hierarchical clustering approach to analyze spatial patterns of accessibility and the added value of ride-hailing in improving service accessibility, particularly in areas with limited PT.
- Demand Share Analysis: For Washington D.C., they fused smart card data for public transport with Uber trip data to analyze how relative attractiveness (measured by MAG) influences demand shares.
Datasets
The study utilized a substantial dataset of approximately 3.5 million completed Uber trips from October 2018 across six cities:
- US Cities: Houston, New York City, Washington D.C.
- European Cities: Amsterdam, Stockholm, Warsaw
This dataset was supplemented with public transport data for Washington D.C. (smart card transactions) to analyze demand shares.
Key Insights and Ideas
- Nuanced Reality: The "compete vs. complement" debate is too simplistic. Ride-hailing acts as both, depending on the specific trip and urban context.
- Time Savings are Real, but Costly: Ride-hailing consistently offers shorter travel times, especially out-of-vehicle time (waiting and walking), compared to the fastest PT alternative. However, this time saving comes at a higher monetary cost.
- "No Viable PT Alternative" is Significant: A substantial portion of ride-hailing trips (20-40% across the studied cities) either have no PT alternative within a reasonable walking distance or the PT alternative is significantly longer. This highlights ride-hailing's role in serving "transit deserts."
- Accessibility Varies Spatially: The added value of ride-hailing to overall service accessibility is not uniform. It is particularly pronounced in areas underserved by urban rail networks, especially in US cities.
- Demand Follows Competitiveness: When PT is time-competitive, ride-hailing demand share is lower, and vice-versa. This suggests that while time savings are a driver, cost remains a significant factor in mode choice.
Key Results
- For the majority of trips, ride-hailing offers shorter nominal travel times than the fastest PT alternative.
- Between 20% and 40% of ride-hailing trips have no viable PT alternative within a reasonable walking distance or are significantly longer by PT.
- The increased service accessibility attributed to ride-hailing is generally greater in US cities compared to European cities, particularly in areas lacking robust urban rail.
- In Washington D.C., a negative correlation (-0.492) was found between the Modal Accessibility Gap for nominal travel time and Uber's demand share, indicating that as PT becomes more competitive, Uber's share decreases.
- Despite time savings, ride-hailing is generally more expensive than public transport when considering generalized travel costs, with the cost difference often outweighing the time savings.
Limitations and Future Work
- Value of Time Assumptions: The study relies on assumed values of time for different countries, which might not perfectly capture individual preferences or trip-specific valuations.
- Single Rider Assumption: The cost analysis assumes a single traveler, which doesn't account for potential cost-sharing in ride-hailing (e.g., UberPool).
- Limited Scope of Demand: The analysis of demand shares was primarily focused on Washington D.C. and excluded other ride-hailing services, taxis, and private car usage.
- Future Work: The authors suggest exploring temporal factors, socioeconomic influences, and the distinction between pooled and private ride-hailing services. Further research could also incorporate more comprehensive urban travel demand data and a wider range of alternative modes.
Summary updated