Pham

Fubon Center Doctoral Fellow Research

Ridesharing and the Use of Public Transportation

Katherine Hoffmann Pham, Panos Ipeirotis, and Arun Sundararajan

Read the full paper

Ridesharing companies often highlight how their flexible digital business model complements public transit systems. For example, Lyft runs a “friends with transit” campaign that advertises “one gajillion new stops,” with services that take riders “the rest of the way”. For April Fool’s day in New York City, Uber announced a premature “expansion” of Manhattan’s long-awaited Second Avenue subway line, offering rides along Second Avenue for the price of a subway fare. In San Diego, Uber claims: “gaps in public transportation become hubs for Uber.”

In this paper, we empirically examine these claims. We use open data on Uber, Lyft, taxi, Citi Bike, and subway ridership to study how crowd-based capacity harnessed via a platform may serve as “invisible infrastructure” that absorbs the demand spikes caused by disruptions to centralized public mass transit in New York City. This is an important instance of a broader question – how digitally-enabled, decentralized systems can work alongside centralized infrastructure – whose answer is key to understanding the value of platform-based business models.

We answer three focused questions. First, do riders substitute to taxi, platform-based ridesharing, or Citi Bike when faced with subway disruptions? We find that subway service disruptions are associated with statistically significant increases in the use of Uber, Lyft, and taxi services, but at the city level, we find no evidence of a significant increase in Citi Bike use.

Second, do riders prefer the digital alternative – platform-based ridesharing – over the physical-world alternative – traditional street-hail taxis – when faced with a public transit disruption? In percentage terms, we find no evidence of a citywide preference for Uber and Lyft relative to taxis. While subway service disruptions are associated with 2.8% and 3.3% increases in the use of Uber and Lyft, respectively, they are associated with 8.2% and 7.0% increases in the use of yellow and green taxi services.

Third, how much of displaced subway ridership do the digitally-based and physical-world modes absorb, and which modes absorb the greatest number of displaced riders? We find that the digitally-enabled modes absorb only a small fraction of displaced riders, and that they likely have the ability to play a far greater role in the future. On average, we estimate that each disruption displaces over 1,500 rides, but that the corresponding increases in taxi, platform-based ridesharing, and Citi Bike use together account for less than 40 additional rides. In terms of the absolute number of riders accommodated, we estimate that ridesharing plays a more important public transit role relative to taxis in the “outer boroughs” of Brooklyn, Queens, and the Bronx (as compared with Manhattan), and that the significance of this role is growing over time.

Our findings suggest that the strategy adopted by digital transit players – i.e. competing for market share among consumers inconvenienced by public transit – makes sense; platform-based demand is increased by subway disruptions. However, these modes are not currently absorbing the majority of displaced riders. This suggests that the flexibility inherent in their crowd-based business model could be exploited further. Finally, although an increase in Uber, Lyft, and taxi rides during periods of subway service disruptions matches our expectations, the magnitude of this increase – and the variation in effects across modes, space, and time – is harder to predict, highlighting the value of quantitatively estimating the impact of digital platforms on centralized infrastructure systems.

Read the full paper here.