Economist: Big taxi-sharing study gets it wrong

by Sean Riley


Politico 

A prominent New York City transportation economist disagrees with a widely circulated M.I.T. study that found taxi-sharing apps could cut the size of the fleet by 40 percent and drastically reduce congestion on city streets.

“They got all these great algorithms and these equations and they did that very meticulously,” the economist, Charles Komanoff, told Capital. “What isn’t so meticulous is the conclusion that they tacked on to their paper.”

The study was published in September in the Proceedings of the National Academy of Sciences, and it said that taxi apps that enable cab sharing could reduce “cumulative trip length by 40 percent or more," and in so doing cut traffic and air pollution and allow the city’s taxi fleet to shrink by 40 percent. 

“Think of how much spare capacity you have in taxis in New York City,” Carlo Ratti, an Massachusetts Institute of Technology professor and one of the study’s authors, told the New York Times. “You’re at a hotel, you’re going to J.F.K. Airport and you take a taxi. And just minutes later, there’s somebody else taking another taxi, half empty, to J.F.K.”

According to Komanoff, it also overlooks several realities of how taxis and traffic work New York City.

Assuming the average taxi carries 1.3 passengers (an approximation based on various taxi regulator estimates), that would mean that to transport 100 New Yorkers today, you’d need 77 cabs, Komanoff said.

Assuming someone did create a system that could properly harness the latent power of taxi-sharing, like the study’s authors suggest, it would enable 80 percent of all taxi passengers to ride in shared cabs. In this new, idealized, taxi-sharing universe, the same 100 people would use 60 cabs (that's 1.7 people per cab). The taxi fleet could be cut 22 percent, not 40 percent, he said.

“But that’s the least of it,” Komanoff said. “What’s more important are that other things are going to happen. Let’s just list them.”

First, he said, since more people would share taxis, the average cost of a taxi trip would fall, which would, in turn, prompt even more people to use taxis.

The reduction in traffic congestion thanks to taxi sharing would free up street space in the central business district, enabling faster-moving cars and thereby attracting more private car drivers (and more taxi passengers) to abandon mass transit in favor of city streets. 

Finally, Komanoff’s model predicts the decline in taxi usage would inadvertently lead to a slight increase in the amount of time taxis cruise for fares, because demand is more fungible than the number of taxi medallions on the street.

“It’s not as if the number of shifts, all those 13,600 medallion taxis, is going to drop in exact proportion to the drop in the number of taxi fares,” Komanoff said. “In other words, I think there will be some drop in the number of shifts, but it’s not going to be as great proportionally as the drop in the number of fares."

In the end, Komanoff predicts the ridesharing scenario envisioned by the report's authors would lead to improvements in central business district travel speeds of about 9 percent between 6 a.m. and 6 p.m. weekdays. 

"That’s meaningful," he said. "But it is not by any stretch the end of gridlock.”

Another consequence of the cheap, shared care ride, Komanoff said, would be a drop-off in people using New York City’s subway and buses, causing the M.T.A. to lose some $45 million a year in fare revenue. Since every taxi ride includes a 50-cent surcharge for the M.T.A., Komanoff estimates surcharge revenue will drop another $15 million a year on top of that. 

Komanoff thinks a better solution for traffic congestion would be to implement tolls on East River bridges and across 60th Street, and lower them in the outer boroughs, part of a plan he advocates called MoveNY.

The report’s authors did not respond to a request for comment from Capital, though they did respond to Komanoff.

“[W]e usually approach these issues from the fundamental research side, where we analyze the issues in a more general, scientific context, and in most cases we tend to leave the practical considerations, that matter in the concrete implementation, to practitioners like you,” Michael Szell, one of the study’s authors, emailed him.