The Subways Seeded the Massive Coronavirus Epidemic in New York City

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 that became evident throughout the city during March 2020. 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. Maps of subway station turnstile entries, superimposed upon zip code-level maps of reported coronavirus incidence, are strongly consistent with subway-facilitated disease propagation. Local train lines appear to have a higher propensity to transmit infection than express lines. Reciprocal seeding of infection appears to be the best explanation for the emergence of a single hotspot in Midtown West in Manhattan. Bus hubs may have served as secondary transmission routes out to the periphery of the city.

That is from a new NBER working paper by Jeffrey E. Harris.


Genuinely nice of you to mention this, but much like germ-filled reusable grocery bags, the Sacred God of Mass Transit is always going to be with us, no matter how many people it kills.

Far too many people have invested far too much Status in these things.

You could actually make the switch. You *could* use reusable bags in normal times and switch in pandemic.

Fwiw, my technique is no bag. The cart and checker have the exposure they have. I transfer to a "dirty box" in the back of my car.

If I were picking up less I might use my bag, and just wash it.

No, I can't make the switch, because the ruling class is proud to decide this for me.

No one makes you use a bag. Some jurisdictions might place a minimum charge. But this being an economics blog, you know that they were never free anyway. They were just hidden in retail price.

Is the "minimum charge" mandated by "some jurisdictions" less than a penny? Because that's the bulk cost of a plastic grocery bag that's "hidden in retail price".

I think around here it is 10 cents. Cry me a river.

That's exactly the sort of response I'd expect, this being an economics blog. Welcome to MR, where the priors get confirmed and the cents don't matter.

The store was paying more like 5 cents before:

And you have the *option* of paying nothing or 10 cents now.

Imagine complaining about the 5 cent difference, rather than sucking it up, one way or another.

Grocery chains don't buy .5 mil bags one carton at a time, my friend. Again, this is an economics blog, you ought to be able to figure that out even if you pride yourself on your E. coli box in your trunk. And since it's run by two paper libertarians (the blog, not the box), one would expect at least some lip service paid to consumer choice rather than top-down disincentives. Although, FWIW, I don't live in a communist shithole so I still get my dog poop bags at Publix for less than a penny apiece. Indeed, they do "feel" free.

ULine prices are not set as one-off for oddball shoppers.

That is a wholesale source and price.

The minimum order size you've linked at Uline is about 1,000. Big grocery stores might not lower than price all the way from $.05 to less than a penny, but it's crazy to think that's a reflection of a "wholesale" price for big retail.

Here's the blunt of it, "anonymous": Shark Lasers is basically correct, you're basically wrong, and if you can't/won't admit that then you are either a needlessly argumentative troll or an effing moron.

Honestly, if you guys are at a point where you believe that public transport is a conspiracy from the "elites", I don't have anything to say to you.

Isn't the idea of charging for bags to account up for the negative externalities associated with them--the costs borne by the public that aren't included in that "less than a penny"?

That was my point, as our anonymous friend apparently believes that the various nanny states created the surcharge to break out the (negligible) costs of the bags, instead of to arbitrarily virtue-signal.

Perhaps someone will do a study one day to determine what the surcharge should be for reusable bags due to the additional vector created for the Chinese bioweapon laboratory fever (name is a work in progress). On the other hand the people who come up with these surcharges haven't really been put out by shutting down the private sector, so I suppose to them it's nothing.

Oh no, I understand that is more than signalling.

I understand that bans reduce bags found in nature. Fewer in rivers etc.

I just think it's amazing that anyone is sweating a few cents *optional* cost.

If you carry your promotional reusable bag, it's cheaper and better for the environment.

You have to be super stubborn, or lazy, to choose to pay.

To be fair, if you are into the Behavioral Economics, they "felt" free.

Sometimes I grab 3 or 4 of the still single use plastic produce bags, then go through self checkout and bag everything up. I've never been yelled at, but our ban is less than a year old, I suppose there's still time.

There is no such thing as a "single use bag."

They don't explode once you take the groceries out of them.

Mass transit costs lives during a once-per-century scale pandemic event. The rest of the time, it is much safer than driving on a deaths per mile traveled basis.

Also, that is an underestimate of the difference in fatalities, since mass transit enables higher population density and reduces miles traveled. Plus, people who use mass transit tend to do more walking and have better health.

Presumably it causes flu deaths every year.

In San Francisco and NY City anyone who can avoid mass transit does so because your life is constantly at risk from the people who do use it. BART may be more dangerous than Covid-19.

How many people were killed by others on BART last year?

42 people were killed on the roads in SF alone in 2019.

So, you are a BART rider then???

But not a single drunk driving death.

The constant praise of mass traffic is undermined by the fact that as soon as it becomes possible to avoid using it, economically or otherwise, that is exactly what one does.

I don't think you live in an area with a mass transit infrastructure to speak of.

I would say 90% of people I have known living in Boston and DC over the last 15 years who could use mass transit in their daily lives does. This is especially true for commuting to work. Unless your commute includes very little traffic and you have free (i.e. subsidized by the taxpayer or your employer) parking, driving to work in these cities makes very little sense. I haven't lived in NYC, but add that one to the list.

True of bus systems, which are perceived as downscale, not so much of various sorts of commuter rail which often have a middle class cachet.

Even the bus stigma seems to be unique to the US, as far as I can tell.

The current problem with the New York subway system is because it's such a highly useful amenity most of the time, millions of New Yorkers have no Plan B for how to get around.

Los Angeles, in contrast, has a subway, but precisely because it's not very useful, few people are wholly reliant upon it. So, during an epidemic, most Angelenos have a Plan B: a private car.

it also creates a massive SPOF that can be attacked or otherwise leveraged to extort concessions.

If you want the new york city government to immediately cave to your demands, organize a hundred protestors to chain themselves to subway car doors and block stations. The existence of and dependence on public mass transit is what creates this. If no such transit system exited, no such SPOF would exist

Whether or not this risk is itself a justification for avoiding transit systems, vs whether the benefit of the transit system more than makes up for this risk, is an open question that depends on a lot of specifics.

This sounds like Robert Heinlein's 1940 short story "The Roads Must Roll" about a future in which instead of private vehicles driving on asphalt roads, the roads move like conveyor belts.

But, that means that a strike by the roadworkers union can shut down all of society.

But then, NY has the lowest traffic fatality rate in the country - 1/3 that of Florida, for example. Surely that has a lot to do with NYC subways, no?

It’s just plain silly to judge public transportation by its role in a pandemic. The NY subway has been around for over 100 years.
You could apply your logic to all kinds of institutions that helped spread the virus. Restaurants, horrible, people should cook at home. Forget the Sacred God of schools, homeschooling is safer. Sports events are dangerous, outlaw them ...

Silly? 100 years. Guns. Silly...

No, it isn’t silly to acknowledge the role public transportation plays in a pandemic particularly from the point of view as to how improve on such in the future (ie the ‘switch’, UV lights, masks [cultural different elsewhere] or my personal favorite, mimicking an egg carton crate) and particularly from the point of view, that compares similar metros with different public modes of transport.

Fair enough, but that wasn’t the point of the comment above that provoked this chain. The comment referred to the Sacred God of Mass Transit. It’s appropriate to analyze the role of the subway in the pandemic, but it is indeed silly to conclude that therefore mass transit is a failure.

The difference, of course, is that all of those things were, in fact, shut down because of the alleged pandemic.

Yeah, and public transit is, in most places, 90+% empty so probably safe, while still running to provide essential workers a way of getting to work. What's your point?

1. Add UV lights to train cars and/or have those UV emitting robots wonder car to car sterilizing.

2. Add more trains. Packed stations and trains add to transmission, less packed decreases.

1. The only UV devices known to kill Covid are not safe for use with humans in the room. In any event, having a pristine car is useless the moment you have an infectious individual emitting Covid with their cough.
2. Perhaps, on the flip side the better the ride, the more people who will use it.

I guess the questions are:

Is this effect real? Or does it just reflect the fact that people who can't work from home ride the subway to and from work and are picking it up everywhere.

If it is real are surfaces or air the problem? For air we could probably improve ventilation. Surfaces may be harder to address but you could have 'wipe down' crews doing frequent sanitizing.

A week ago, 59 NY Metro Transportation Authority workers had died in the epidemic, which sounds like a real problem.

Most likely. It seems to spread indoors. It seems to have very quickly erupted throughout NYC and communities directly connected by rail without a lot of other major networks in play. And from I understand subway employees are disproportioantely likely. Nothing definitive, but when time is live (i.e. right now), it is actionable enough for me.

We do not know if it is the air or not, nor do we know the optimum mechanisms of ventilation. I have seen decent evidence suggesting that low humidity worse for transmission, so it might make some sense to turn off the air conditioning. What sort of flow rate and venting is required, well I confess I still have not been able to track down the droplet size and dwell times so I cannot guess what best settings are.

Surfaces are another approach. The current rage is copper impregnated surfaces. The copper railings from 100 years ago are supposed to still be antiviral when tested so my guess is that somebody will make a cheap-enough-but-strong-enough mesh or plastic with copper impregnation that we will shortly be using on exactly these sorts of high traffic areas.

Ultimately, though, it is just a matter of time. Throughout all of human history, increased density has lead to increased disease. We made massive gains with sanitation, treated water, etc. but ultimately those are all linear improvements. Disease scales, roughly, with the square of density (as measured by social grid) so eventually something like this will happen again.

If we are lucky it will be something like Swine flu where people die in hard to spot numbers and the subway blip is offset by New Yorkers being much richer than national (let alone global) average. If we are unlucky, it will be worse next time. My over/under is another 30 years, provided the world does something about China.

OK so what's the deal then with those robots that wonder around hospitals supposedly 'disinfecting' by UV light or something?

They use UV-C, a very intense set of wavelengths that get filtered out of sunlight by the ozone layer. Humans are removed from the room while they operate.

The advantage is that they are quick, do not expose a human, and do not use consumable, caustic chemicals. I suspect they have a bright future of ahead of them.

Subways have the problem they fill with people. At best, you might be able to clean cars at the end of the line and lose an hour of service per cleaning.

Possibly there is some wavelength of light that is both safe for human exposure and disruptive to Covid but that has not been shown to be the case last I checked.

That's tragic. Subways are efficient in a lot of ways. It's just unfortunate that they are efficient in this way also.

I just drove a bit. My immediate thought after 2 weeks shut in was "driving is fun!" Less efficient though, more polluting (more fun) and in this instance less dangerous.

(My Cybertruck is still on order.)

With everyone wearing balaclavas/"head socks" and glasses, with face masks underneath, and washable gloves, there's no reason for subways to be significant sources of transmission.

Remember when trump order the subway system to run fewer trains worsening this?

Lol in a just world Cuomo and de Blasio would be executed for crimes against humanity. It’s hard to think of any two leaders with more blood on their hands.

Nope, don't remember that.

If only people had listened to the hoax line and kept riding the trains, they wouldn't have reduced service. Hannity/Trump were just deploying a noble lie to save lives in NY.

3 dimensional chess people....3 dimensional chess

"Water is Wet", next paper by Jeffrey Harris.

Everyone except New York officials know that the subways were the biggest factor.

Do the 7-Elevens in Tokyo sell dehydrated water?

➕ 1. After riding F, 2,34 for 25 yrs
File under “duh”

Yes! A zip code geographic analysis linked to subway. Network analysis.

Now, let's look at airport restaurants, bars, and TSA check ins.

Time to have the Japanese cleaning ladies greet the NYC subways and do their little thing. Here's where a little sanitary and design innovation could really pay off.

DeBlasio should have acted sooner.


Efficiency is not the motto of the MTA. But maybe if they throw enough manpower at it they can get close.

Yeah, I could imagine some sort of touchless electronic turnstiles or something. Seats made out of materials that virus doesn't like to live on, etc.

Many transit systems already have touchless turnstiles. See eg

The above link makes multiple interesting arguments as to why the analysis presented in the NBER paper is pretty flawed and based on very bad sampling.


TLDR of the linked piece:

“But when it comes to hard evidence, the paper makes two quantitative claims. The first is in figure 3: Manhattan had both the least increase in infections in the 3/13-4/7 period, equivalent to a doubling period of 20 days whereas the other boroughs ranged between 9.5 and 14, and also the largest decrease in subway entries in the 3/2-16 period, 65% whereas the other boroughs ranged between 33% and 56%.

The second is a series of maps showing per capita infection levels by zip code, similar to the one here. The paper also overlays a partial subway map and asserts that the map shows that there is correlation of infection rates along specific subway routes, for example the 7, as people spread the disease along the line.

I will address the second claim first, regarding line-level analysis, and then the first, regarding the borough-level difference-in-differences analysis; neither is even remotely correct.

The Paris Metro - should be a good source of data.

And does anyone know how many NY subways run out to somewhere like Westchester county which less than a month ago had NY guardsmen controlling access to one town.

The Tokyo Metro should also be a good source of data.

We can figure out in advance how it seeded the massive coronavirus epidemic there, while we sit around impatiently and wait for it to actually happen any day now.

The London Underground is stuffier and more crowded than NYC transit system and has miles of narrow corridors. That would also be an interesting comparison.

Of course every German city, as well as Vienna, is more reliant on public transport than NYC. Hearing libertarians bleat about how dangerous public transport is sounds a lot like mood affiliation.

No subway service to New Rochelle...Metro North to Grand Central

Great data point. Thanks. Not surprising. One question is if the transmission was via people touching the railings or by people being crowded into densely packed subway cars where they could breath on each other. I suspect the latter.
Maybe require mask usage in subways and restrict the number of passengers per car so that people can stand or sit further apart.

Wearing masks like in many Asian cities is a likely change. Restricting the number Of passengers though likely won’t work in NYC, it would cut usage for some major lines to the point where it isn’t worth it. I expect biking will become a much bigger mode of transportation, especially in Manhattan. Unfortunately though people can’t easily bike in from the outer boroughs. I’m not sure what can be done about that.

Ever heard of electric bikes?

Even by electric bike the trek from Ozone Park to midtown is, well, a trek. Not sure I can see many lawyers and analysts doing that in work attire every day.

Or maybe show you re under 65 ?

It occurs to me that this has a selection bias built into it. Since the virus spreads very fast in nursing homes you're going to overstate the over 65 death rate. You need to consider the infection rates among different age groups as well.

When I compared flu deaths by age from the CDC to NJ's COVID deaths by age, I noticed while the elderly were among the dead more often than young in both groups, they were more heavily when it came to the flu.

I see very little reason to believe that over-65's have a higher infection rate. In fact, I see quite a few reasons to think that it should be *lower*.

Before the lockdowns, who were the people going to work every day, going to bars and restaurants (as both customers and employees), going to concerts, etc.? Not the over-65's.

That's true of over 65 yr olds who live on their own. But nursing homes are a different story, it's spreading like wildfire through them. Early on in a breakout it will seem like older people predominate in deaths but that's because younger people will take longer to get it (except for healthcare workers) and will put up a longer fight before dying.

I'm not saying older people won't make up a large share of deaths when all the dust settles, they will.. Just relative to the flu, they are not as large. When you hear about 30K-60K flu deaths, almost all of that is older people or people with serious problems beforehand.

Is this the death rate among those infected or just percapita at large?

Because 800/100000 = 0.8%

Tyler, check out the recent Market Urbanist post disputing this study for a contrary perspective:

I lean toward subways being a genuine contributing factor, but it’s worth noting the data isn’t perfectly clear (e.g. Staten Island having a much higher case/population ratio despite being the least subway-oriented borough; Manhattan having the lowest ratio).

For whatever reason, public transport does not appear to be a major vector in Europe or Asia. Seems like more analysis is needed.

One problem with the refutation is he uses cases, which aren't necessarily the best measure. Car drivers might be getting tested more, especially since there are many drive-up testing operations. The fact that deaths are much higher in Manhattan than Staten Island might be a more significant number.

I read the paper. It's a worthwhile contribution but it honestly doesn't change my priors that much. The time series effect is full of confounds that the authors readily acknowledge. The geo data seems perhaps even weaker. Figure 8 shows the most data, and honestly it doesn't look like there's a strong relationship between subway lines and zip code infection counts. And that's despite the authors getting to choose which subway lines to show!

In summary, the subway hypothesis is plausible and the authors deserve credit for their valuable work, but the paper doesn't really change my beliefs much.

If you look at the maps in Harris' paper, or at the data as I did here (, you find the trend is the opposite of what he wrote. Nothing in Harris' paper comes close to justifying the certainty of his title & introduction. The rest of the paper is a hypothesis and some discussion. There's no correlation, let alone causation, shown.

I disagree. Look at the literal language of this obtuse gem of prose: "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" - it either says, by the literal language, that there was negative correlation, or, if you want to be charitable to the spirit of the author's conclusions, a positive correlation, between new Covid-19 cases and subway ridership. So the paper can say whatever you wish of it.

That's true, but he goes on to note this correlation is totally endogenous, which it clearly is, and to disclaim it as evidence. (I showed in my study that you can make the same chart with restaurant meals).

And what else was done at the same time that subways closed? A million other things that correlate.

I have little doubt that some people caught this in the subway, just like some people caught it at restaurants, bars and churches. Detaching effects is difficult.

The open secret about NYC WuFlu is that Ultra-Orthodox Jews were the super spreaders. Some of it is due to their religious beliefs, requiring 10- man quorum for prayer ("minyan"). But most of it is attributable to them being parasitic bumpkins, who don't believe in modern science or any obligations to the wider community. Their Purim celebrations then spread the infection to a much wider extent.

The US media vilified hapless Florida spring breakers and recent conservative protestors as being reckless. Meanwhile, they have said little about this issue. Hmm, I wonder why? The media in Israel -- where the Ultra-Orthadox are a serious, ongoing issue -- have been a little more open about it.

Not sure what media you're following but here in NJ both the Governor and the media have been calling out violations in Lakewood, many (not all) have revolved around the Orthodox community holding services and conducting weddings on the 'down-low'.

The difference is the police show up and disperse the gatherings with warnings initially and then citations. No such thing happened with Spring Breakers and Protesters because they are applauded by a political establishment that cannot process just how deeply their failed leadership on this appears to the larger public.

The open secret about NYC WuFlu is that Ultra-Orthodox Jews were the super spreaders. Some of it is due to their religious beliefs, requiring 10- man quorum for prayer ("minyan"). But most of it is attributable to them being parasitic bumpkins, who don't believe in modern science or any obligations to the wider community. Their Purim celebrations then spread the infection to a much wider extent.

Choice crowd the moderators attract.

You do have a point. But it's mainstream to target Christian pastors who... don't believe in modern science and persist with their religious observances.

I dunno dude. The "Jews who control the media" to whom you seem to refer are very different sorts of Jews. They are not conservatives, for one thing.

You are just lazy. Plenty of US media coverage at the height of the pandemic back in mid-March:

The abstract doesn't do the paper justice. My initial reaction was "Jeff Harris is a smart guy, how can he be confusing correlation and causation" but the paper analyzes that thoroughly, within the limitations of the data.

And the research leads to some potentially useful policy recommendations: "from the public health perspective, the optimal policy would have been to double – maybe even triple – the frequency of train service. The agency’s decision to convert multiple express lines into local service only enhanced the risk of contagion (Goldbaum 2020). How ironic it is that the preferred policy would have been to run even more express lines. We have not seen any public data on the incremental cost of the agency’s decision to begin to disinfect subway cars twice daily. Still, it is natural to inquire why the cars weren’t disinfected every time they emptied out of passengers at both ends of the line."

Also it was interesting to read about the epidemiological models of Gosce et al, which seem to combine a variation of the Schelling-Ising model of individual agents who meet and collide with each other possibly spreading an infection, with the Hotelling model of mobile vendors on a beach -- and he connects this to the pioneering work of John Snow taking the handle off the water pump that was spreading cholera in London: "In an innovative series of papers, Goscé and colleagues generalized this model to consider contagion when the S’s and I’s move along a corridor (Gosce, Barton, and Johansson 2014, Gosce and Johansson 2018). They applied their framework to the study of the spread of influenza-like illness in the London Underground, a vast network opened just nine years after Dr. John Snow got public officials to disable a pump at Broad (now Broadwick) and Lexington Streets, now about a five-minute walk from the Oxford Circus station."

His policy recommendations make me think that he does not understand the topic he purports to be studying. It is literally impossible to double or triple the frequency of trains, at least at rush hour. As for running more express trains, a passenger who boards an A train at 59th has to go all the way to 125th to get off. If his stop is in between, he has to take a second, local train back. Is taking two trains really less hazardous than simply taking the local in the first place? Seens unlikely.

If anyone is interested in learning instead of confirming their priors then here is a review of this study finding it flawed but suggesting improvements the MTA could make for reopening:

LOL. Learning? Where do you think you are?

It's always funny to watch mass transit get dissed by people that,,, have no need to use mass transit or live where there is no mass transit. I can pretty much guarantee that a decade from now myself and millions of others will still be using mass transit while internet curmudgeons will be ten years older and ten years more grumpy at the world...

Given that the lives of 330 million people have been disrupted for a month and a half now because people in the epicenter (conveniently full of media personnel and other Very Important People who think the country ends in Hoboken) won't stop using mass transit, I'd say it's a net negative for the curmudgeons.

Of course my preferred solution is to just build a wall around NYC and let them ride the subway as much as they like for the next ten years or however long they last.

There are about 400,000 cases outside of the NJ-NY-CT-MA corridor. Trump's swing states have 85,000.

"Won't stop using mass transit"

Oh, won't someone save us from these stubborn elites who have no alternative for getting to their job as a nurse, food worker, or cleaner!!!

Always remember:

-- A 2-week shutdown of the Government is a humanitarian disaster on par with the Ethiopian Famine

-- A 3-month shutdown of the Private Sector is... Shut The Fuck Up

It is blogging malpractice NOT to link to Alon Levy’s cogent rebuttal of this paper:

Quite remarkable to see strident discussions of "overreaction" and "opening the economy back up" yet there is not even basic, and I mean not even basic consensus about the scope and means of transmission.

One thing is certain, this virus has unleashed a cytokine storm of bullshit and ignorance.

I'm skeptical of this.

It's absolutely logical that the subway could be to blame. However, a lot of things in New York are correlated with subway routes. There have been studies of people's social networks from facebook etc. that show that people socialize along the same subway lines.

It would be natural that a correlation with subway lines would show up with anything that has a correlation with people's social networks.

The timing is also not surprising. People stopped taking the subway because of the lockdown. People stopped interacting because of the lockdown. Infections went down because people stopped interacting. Not just because they didn't take the subway.

Maybe people were getting infected on the subway. Maybe they weren't. I don't think the evidence here is sufficient to reach a conclusion.

The substantive basis of this analysis is almost laughably bad. See, for instance, the response that others have linked ( Also, several others have already commented on how subways are less transmissive than you would expect because no one is talking (primary form of transmission), e.g.:

I think the jury is far from out on the subway’s contribution. Regardless, this paper offers literally zero insight into the answer.

Just laughing at most of these comments. It is either the transit system, or it is all the people living like ants in big multi-unit apartment complexes. Make your choice which you believe.

If you don't want to blame on living/traveling conditions, then just blame it on bad fucking luck.

NYC is a shithole. Full of the worst of humanity living like animals in a zoo. Now they are dieing of Chinese Flu, and the elites are trying to make me care every night at 6pm, but I could give two shits. Open the country up already.

I'd rather take my chances in NYC than within 1,000 miles of wherever this delightful Anon lives.

Again, Seoul and Taipei are comparable in terms of density and transit usage to New York City and neither had outbreaks comparable to what New York is going through. Dense cities in general are more vulnerable to contagious diseases -- was this ever a controversial point? -- but we have just as much evidence that good policies can mitigate risk.

So far, Southeast Asian megacities like Bangkok, Manila, Hanoi, Jakarta and Kuala Lumpur also have avoided apocalyptic outcomes. Singapore's transit system is still running and doesn't appear implicated in the most recent outbreaks they have been experiencing, which are concentrated in foreign workers' dormitories.

And yet, we've seen nothing of the testing of surfaces in the subway or aerosol testing either.

How do they know that cleaning cars every 72 hrs was the best practice?


People always talk about cities and economies of scale. They seem to be diseconomies of scale.

Meanwhile people in NYC are horrified because the yokels walking the beach in Jacksonville are going to murder their grandmothers. Have they shut down the subway yet?

Comments for this post are closed