A study published today in Science Advances comparing pre- and post-rideshare boom traffic in San Fransisco found that the presence of Uber, Lyft, and similar companies has been an overall detriment for people who like getting where they’re going quickly.
That businesses which pay people to have their vehicles on the road would, well, increase the number of cars blocking up the transit grid might appear to be a forgone, perhaps even obvious conclusion. But the body of writings on Transportation Network Companies (TNCs) as they’re sometimes called is, surprisingly, mixed. Some studies found that Ubers and Lyfts were choking the streets of New York, Boston, and Chicago; a few claimed, conversely, that rideshares were alleviating traffic. Thus the team behind today’s paper—composed of two University of Kentucky staffers and members of San Francisco’s County Transportation Authority—had their work cut out for them.
To establish a baseline, the team looked at data signals from 2010, when Uber had yet to launch in San Francisco, Lyft’s predecessor Zimride was two years away, and a New York taxi medallion was still worth north of half a million. Those signals included data from TNCs themselves which had previously been processed by researchers at Northeastern and the California Public Utilities Commission, commercial speed data collected by private firm INRIX, and “activity-based travel demand microsimulation model” SF-CHAMP (fed with the city’s 2010-specific data.)
For contrasting analysis, the team landed on 2016 as a suitable benchmark year for when ridesharing had reached saturation. But using SF-CHAMP, they generated another simulated model as well: one fed with 2016 “population, employment, and network inputs” but “calibrated to 2010 conditions”—in essence, an alternate vision of the dumbest electoral year in American history where, for one reason or another, TNCs simply did not exist in San Francisco.
The team’s findings—comparing the recent past to either the version of 2016 that happened or the hypothetical, better one, even controlling for things like population growth, increased employment, and roadway improvements—are none too favorable to Uber and its ilk. The starkest examples were morning and evening rush hours when congestion was already high. A 2016 model without TNC’s saw VHD (vehicle hours of delay) jump 18 to 23-percent from 2010's numbers. With them? The estimate jumps to between 48 and 52-percent.
Even outside peak hours, the findings remain broadly consistent across a variety of metrics. Damning as it is, the researchers do point out that features like Uber Pool (which they refer to as “ridesplitting”) could help reduce traffic congestion. However, available information for other cities suggests less than a fifth of rides taken using the pooling option “with some of those trips carrying no additional passengers.”
Set to go into effect in 2021, New York recently became the first US city to adopt congestion pricing for vehicles in its busiest neighborhoods—something London did in 2003, Stockholm in 2007, and Milan in 2013. Washington, D.C. is exploring the possibility of a similar scheme, while Bostonians and Los Angelenos have less formally advocated for a congestion tax of some kind. New York, already one of the most restrictive places to drive for Uber or Lyft, also opted to cap the number of new rideshare drivers last summer. Both changes were preceded by a 2017 study “based on trip and mileage data that are uniquely available in New York City”—a study that these University of Kentucky researchers called “most similar to our own”—which reached similarly grim conclusion’s about TNCs’ impact on traffic. Whether this study will similarly help upset the status quo in Uber’s own backyard remains to be seen.