On April 15, New York City Councilmen Robert Holden, Mark Gjonaj, Peter Koo, and Eric Ulrich sent a letter to Gov. Andrew Cuomo pleading him to shut down the city’s public transportation system, including the subway.
They noted that recent service cuts by the Metropolitan Transit Authority had led to crowded subway cars and buses. “This crowding [could] become a hot spot for COVID-19 transmission and presents an extreme danger to everyone on board,” they warned. “We believe that the New York City transit 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.”
On the surface, it makes sense. Subways are often crowded with people, and riders often touch surfaces such as handrails and seats that have been touched by other riders. They would seem to be efficient transmitters of the virus, and New York City has been the epicenter of the pandemic in the United States.
Subways as mass spreaders became common wisdom in part due to a paper written by Massachusetts Institute of Technology economist Jeffrey Harris. Titled “The subways seeded the massive coronavirus epidemic in New York City,” the paper claimed that the subway was “a major disseminator — if not the principal transmission vehicle — of coronavirus infection.” Indeed, the councilmen’s letter cited the research, noting how it showed that coronavirus infections in the Big Apple slowed only after subway ridership declined.
Yet that common wisdom may be received wisdom. New research suggests that automobiles were a far more potent transmitter of the coronavirus.
“The coronavirus has spread more in areas where most people drive than in areas where people walk or take the subway,” said Salim Furth, director of the Urbanity project at the Mercatus Center at George Mason University.
Furth found that an area in New York City with a roughly 30-point higher automobile commuting share would have one more case of coronavirus infection per 1,000 people. At present, there are about 19 infection per 1,000 people in New York City.
“It’s not like I came into this with some great insight,” Furth said. “Like everybody else, my intuition was that driving was safer than public transportation. But then, I looked at Harris’s paper, and the evidence in it was terrible. Indeed, his evidence pointed in the opposite direction.”
Harris did not respond to the Washington Examiner’s requests for an interview.
The strongest piece of evidence in his paper shows a leveling off of new cases of the coronavirus as subway ridership declined. But Furth noted that the leveling off could have been caused by a number of factors since all sorts of activities that bring people into close contact stopped at about the same time.
Harris’s paper also shows that the areas in New York City with higher rates of subway ridership tend to have lower rates of coronavirus infection. Furthermore, it did not control for other variables that could explain the spread of the infection — such as income or health insurance status.
Furth’s research did control for those other factors, and not only did it find that cars had a positive effect on infection rates, but also it found that subways had a negative effect. In other words, areas with higher subway ridership were likely to have lower rates of infection.
“I think the most obvious reason for that is that when you get on the subway, you know it is a risk,” said Furth. “I think people’s behavior changed more drastically on the subway when they knew they were at risk.”
By contrast, people who drive cars may have a false sense of security and may also see more people. Large gatherings are believed to be the quickest transmitters of the coronavirus.
Indeed, the coronavirus may have first spread among the wealthy before finding its way into the middle and lower classes. Jonathan Kay, an editor at the magazine Quillette, examined “superspreader events” from news reports related to the coronavirus. While there is no settled definition of SSEs, they tend to be events where dozens, hundreds, or even thousands of people are infected all at one time.
Among the 100 SSEs that he has been able to find, Kay said, “At the very beginning of the pandemic, you did see, outside of China, a disproportionate rate of infection not just among wealthy people but among extremely wealthy people. Some of the first outbreaks happened in very wealthy areas.”
Kay noted that the early SSEs “centered on weddings, birthday parties, and other events that were described in local media as glamorous or populated by ‘socialites.'” He said, “I think the reason [is] that the guests at these events often included people who had just flown in from some other part of the world, like New York or Las Vegas or Asia. As such, the velocity of the pathogen was that much quicker.”
Kay was skeptical that cars spread the coronavirus in New York City. “[The city] is full of extremely rich people who do not have cars. And subways, it is impossible to do contact tracing. Events like birthday parties and weddings may be overrepresented among SSEs because everyone remembers going to those events and many of these people contact each other via Facebook. When they start getting sick, it’s not hard to trace it back. However, there are no Facebook groups dedicated to people who traveled on the subway last Thursday.”
Furth responded that you also see highly concentrated outbreaks of the coronavirus in southwest Georgia, where there are no mass transit systems. He also said that Kay’s article “helps explain why ‘passive’ contact between people, like being on the subway or in a crowded store or office, might not be major risk factors. All the superspreader events identified are ones where people are relatively intimate and/or loud: singing, shouting, etc.”
None of these papers has been peer-reviewed, and both Kay and Furth warn against reading too much into them.
“The data on SSEs is biased,” said Kay. “People who are likely to end up in the news are the wealthy and powerful, not the lower class.”
“My research is not ironclad,” said Furth. “I know for a fact that we are missing a ton of coronavirus infections, so the data we have may not be representative of what is actually going on. This is all very early work.”
