Did Subway Riders or Motorists Do More To Spread COVID-19 in New York City?

Transit wonks are debating which mode of transportation was most responsible for the country's worst COVID-19 outbreak.


Is the New York City subway system responsible for seeding the nation's worst COVID-19 outbreak? Or are the Big Apple's motorists to blame? The answer has major implications for our immediate response to the COVID-19 crisis, and the role cities will play in a post-pandemic America.

The debate started last week when Massachusetts Institute of Technology economist Jeffery Harris published a controversial working paper claiming that the nation's largest rail transit network was the major vector for spreading novel coronavirus in New York City, where it has killed over 10,000 people.

"New York City's multitentacled subway system was a major disseminator—if not the principal transmission vehicle –of coronavirus infection during the initial takeoff of the massive epidemic," wrote Harris in his paper, which was originally published on April 13.

Harris, in his initial paper, cites two pieces of evidence to make his point. The first is a correlation between a fall in subway ridership and the falling rate of new, reported COVID-19 cases. The number of daily turnstile entries fell from nearly 6 million to just over 500,000 by late March, his paper shows, which corresponded with a dropoff in new infections. New cases of the virus, Harris notes, fell most in areas of the city that saw the biggest decline in transit ridership.

"The near shutoff of subway ridership in Manhattan—down by over 90 percent at the end of March—correlates strongly with the substantial increase in the doubling time of new cases in this borough," writes Harris. Boroughs with a less dramatic fall in ridership saw shorter doubling times for new cases. ("Doubling time" is the period of time required for an observed phenomenon to double in number; in the case of a pandemic, the longer the doubling time, the better.)

Harris' first draft also includes a map of new COVID-19 cases broken down by ZIP code, which he suggests shows that one subway line stretching from Queens to Manhattan correlated with disease hotspots in the city.

In his original draft, Harris caveats all of this by saying correlation does not equal causation. But on April 19, he released an updated paper which pointed to the high death rate among the city's transit workers—79 of whom have reportedly died of COVID-19—as the smoking gun proving his case.

"How ironic it is that [the] unfathomable tragedy of these frontline workers turns out to be the clincher that transports us from correlation to causation," he said.

This paper and its conclusions have fueled calls by some officials to shut down New York's mass transit system, which carries nearly half of the country's transit riders, to help stem the spread of the virus. On Saturday, the New York Post reported that three New York City councilmembers sent a letter to Gov. Andrew Cuomo (D) urging him to temporarily close the subway and suspend bus service.

"We believe that the New York City transit system is a primary contributor to the spread of COVID-19, and we recommend a temporary closure of the system for at least one week for deep cleaning of trains, buses, and stations," these councilmembers wrote. They, like Harris, suggested that the city subsidize rideshare or cab rides for essential workers. So far, Cuomo has resisted these calls.

At the same time, a number of voices are pushing back on Harris' conclusions, and even suggesting the counterintuitive idea that motorists, not transit riders, are the ones most responsible for spreading COVID-19 throughout the city.

In an article over at Market Urbanism, Salim Furth, a researcher at George Mason University's Mercatus Center, pokes holes in Harris's case. While cratering subway ridership is correlated with a slowing rate of infection, says Furth, so are a lot of other activities.

"People take the subway, not for the heck of it, but to get places. When those destinations are places they no longer want to go or are no longer allowed to go" they ride less, Furth tells Reason. Thus, it becomes difficult to disentangle the effects of falling subway ridership from all the other social distancing measures that were happening simultaneously. Falling restaurant visits and declining use of bike-share also correlate with a declining infection rate, he says. But that correlation doesn't tell us anything specific about the role restaurants and bike-sharing played in spreading COVID-19 around New York.

While boroughs that saw a smaller percentage decline in transit ridership also saw a smaller decline in the rate of new cases, Furth says these same boroughs had lower numbers of transit riders to begin with.

"If subways (or ferries) are the primary vector, why is Staten Island, with a 67 percent automobile commute share, just as susceptible to COVID-19 case growth as the rest of the city?" Furth writes. "The change in transit usage is plausibly consistent with Harris' hypothesis; the level of transit usage is inconsistent with it."

The fact that transit workers have a higher death rate than transit riders could point to some sort of institutional spread within New York's transit agency, Furth argues. It is not proof that subway cars themselves are the primary disease vector.

The Metropolitan Transportation Authority, which runs bus and train service in New York City, also prevented its employees from wearing masks throughout early March.

In his response to Harris, Furth compares U.S. Census Bureau data on the share of people commuting via car to the number of new COVID-19 cases broken down by ZIP code. He found that rates of car travel are highly correlated with the number of new infections. Outer boroughs, where rates of car commuting are highest, are also showing the highest rates of COVID-19 infections.

Furth offers two theories for why motorists, not transit riders, might have been more likely to spread coronavirus. "Cars may give you a false sense of security," he says, suggesting drivers could have reduced their travel around town less than transit riders. "Somebody who only has the subway to get around may be only relying on delivery, only going to the corner store, not leaving their block for commerce," Furth says.

In addition, he says transit riders could be interacting with a more tightly-knit network of people; others who take the same subway line or frequent businesses near the same transit stops, for instance. That would reduce the number of new people they run the risk of infecting. Auto travelers, by comparison, could have a more random network of interpersonal interactions, helping them to spread infections farther and faster.

This is not just an academic debate. In the short term, knowing which mode of transportation is more likely to contribute to the spread of COVID-19 would help guide immediate policies for combating the virus.

If Harris is correct, then it would make more sense to shut down transit for deep cleaning or subsidize Lyft and Uber rides. But if cars are helping transport the virus, then raising bridge tolls or closing some bridges and highways in New York altogether might be the smarter move.

Getting this response wrong would mean actively encouraging the most dangerous modes of travel. Urbanites are also debating the longer-term impacts of the coronavirus pandemic on denser cities. Reason just published a video exploring this very topic.

Finding solid answers to these questions will determine how tens of millions of Americans get around each day, where they live, what kinds of developments their governments encourage and restrict, and how their tax dollars are spent.