Neighborhoods With Excess Capacity

In New York, the tech industry has clustered in the Meatpacking District, around 14th Street and 8th Avenue. Google’s building (the company’s largest office outside the Googleplex) is there, Samsung’s New York offices are there, startup incubators are there with co-working spaces. Stephen Smith has called for commercial upzoning there (on YIMBY three years ago, and on Twitter just now), despite NIMBY objections. He argues not only that there is pent-up demand for office space, but also that there is excess subway capacity there: “the L train’s capacity west of Union Square is essentially unlimited, after the hordes from Brooklyn headed to destinations east of Broadway change for the 4/5/6 and N/Q/R.” While his other arguments for upzoning are solid, this one is incorrect, and I’d like to explain which areas have excess capacity and which don’t.

Two years ago, I wrote this post about modeling transit crowding. The model is primitive – it assumes a one-dimensional city, 100% mode share, and independent job and residence distributions. For the purposes of this post, cities A, B, and C from the model are not relevant (they have perfect mixture of jobs and residences); cities D, E, and F, with separation of residences and jobs, are more relevant, with city F, with partial mixture, the most useful.

The results of the model are fairly predictable. In the morning peak, transit vehicles (or roads!) fill up toward the center as they pass through residential areas, and then empty in the commercial core. This means that more residences outward of the point of greatest congestion, and more jobs inward of it, add more crowding; more jobs outward of the point, and more residences inward of it, do not. More jobs on the other side of city center add to crowding, because people still ride through the point of greatest crowding.

On the L, the point of greatest crowding is between Bedford Avenue (the last stop in Brooklyn) and First Avenue (the first in Manhattan). This means that more residential development on the L in Brooklyn and more commercial development in Manhattan would add crowding – even commercial development on the West Side would attract riders living in Brooklyn, who would ride through the overcrowded segment under the East River. The other subway lines serving the Meatpacking District suffer from the same problem: those are the 2 and 3 at 7th Avenue and 14th Street, and the A, C, and E at 8th Avenue. With Second Avenue Subway having taken some crowds off the 4 and 5 on the East Side, it’s likely the 2, 3, and E are the most crowded subway lines in New York today (the A has more room). Yes, most riders on those lines get off in Midtown, but it doesn’t matter, because riders from the Upper West Side and Queens, attracted to new jobs in the Meatpacking District, would still ride through the most crowded point, at the entry to Midtown.

So if not the Meatpacking District, where is it better to add jobs, purely from the perspective of subway crowding? Superficially, the answer is to mix them across the residential parts of the city. But here, my model runs into problems with mode share. The model says that adding jobs in (say) Downtown Brooklyn increases subway crowding, because of riders from Uptown Manhattan riding to the south. Per the model, it’s best to add jobs on the side with more crowding, which is the north and Queens sectors, not the Brooklyn sector, where only the L is very crowded. This means, more jobs on the Upper East and West Sides, and maybe also in Long Island City, near Queensboro Plaza.

But in reality, there is some travel segmentation in New York. People who work on the Upper East and West Sides probably live in those neighborhoods or in Harlem and the Bronx, and people who work in Downtown Brooklyn probably live elsewhere in Brooklyn. Yes, it’s possible to commute between the Upper East Side and Downtown Brooklyn, but people would not ordinarily choose to do so – the commute is long and crowded (because of all the Midtown-bound workers), and there isn’t much saving on rent. People might still do it for various reasons, like a two-body problem or moving frequently between jobs – this is why through-running is important – but it’s much less common than living and working on the same side of city center.

So most likely, office development in Downtown Brooklyn would mainly attract ridership from within Brooklyn. Extra ridership from Uptown Manhattan and the Bronx is likely to be small. The upshot is that locations outside the most crowded point on each inbound subway line are likely to lead to large gains in subway ridership without much additional crowding.

I bring up Downtown Brooklyn and not just the Upper West and East Sides because it is better-connected to more bedroom communities by subway. These include the Lower East Side and Chinatown, Long Island City, and nearly all of Brooklyn. Long Island City is also highly accessible, from much of Queens and the parts of Brooklyn on the G train. But the Upper West and East Sides aren’t so accessible because of the lack of good east-west subway options.

Of course, the situation on the ground is different. New York is desperate to add tech jobs in Downtown Brooklyn, but the tech industry insists on clustering in the Meatpacking District. There’s only so much a city can force developers to site themselves in the areas most convenient for infrastructure. But from a long-term capacity standpoint, it’s in New York’s interest to encourage commercial development outside the Manhattan core, especially in areas that get decent subway service from multiple directions, like Long Island City, Downtown Brooklyn, and maybe Jamaica.

It would be easier if there were more service targeted at off-core destinations. This is part of why I harp on regional rail all the time – the LIRR would be able to serve Downtown Brooklyn and Jamaica better if it didn’t exist just for the benefit of suburban salarymen working in Midtown. But this also includes Triboro, which would give multidirectional service to nodes including Jackson Heights, the Bronx Hub, and Brooklyn College. This would encourage developers to build commercial at these nodes, which suffer from poor access to workers today.

Note that opening circumferential transit, in this model, has the opposite of the expected effect on radial lines. Normally, a new transit line reduces demand on parallel lines and increases demand on intersecting lines, which runs the risk of overloading them. But if a circumferential line encourages office development at intersection points with radials, it will still encourage more ridership on the radials, but this ridership will completely miss the congested inner portions of the radials.


  1. newtonmarunner

    Thanks, Alon. Does this article also explain why NYCTA should have chosen the 10th Ave. stop on the 7 Extension over Hudson Yards?

    • Alon Levy

      Yes, and also why the city should’ve upzoned residential rather than commercial at Hudson Yards. The ridership generated would’ve been reverse-peak.

      • newtonmarunner

        Is there a way to correct for this? Shuttle from 10th & 42nd to Hudson Yards, & halt new development in Hudson Yards?

        Also, I think in addition to additional (super) express radial transit (aka regional rail) and additional local circumferential transit, doesn’t line reorganization also follow the thrust of your column for massive regional capacity increases?

        Thanks so much for answering.

          • adirondacker12800

            Developers are good at finagling anything they can squeeze out of naive governments.

        • Ed

          If there ever is the money for it, the correction will be an updown-downtown West Side mass transit line with a stop in that location, linking up with the 7 at the next stop.

  2. mdahmus

    I think another part of the model that oversimplifies reality is the presumption that everybody is equally likely to take transit. If you add employment in Brooklyn, isn’t it somewhat more likely a given employee will drive than if you add the same employment in the Meatpacking District?

    • Alon Levy

      Yes. I read once, around 10 years ago, in a writeup I spent years later looking for on the internet and not finding, that Downtown Brooklyn’s transit mode share is only 50%, whereas that of Jamaica, Long Island City, Flushing, and the Hub is 30%.

      That said, traffic’s bad enough that the marginal worker is probably a transit rider and not a driver.

      • Ian Mitchell

        “Traffic’s bad enough that the marginal worker is probably a transit rider and not a driver.”

        Is there a point at which this can be assumed in any given area? E.g. certain commercial/residential nodes in LA or the Bay Area?

        • Alon Levy

          In the Bay Area it’s probably true – evidently, there has been growth in mode share in the last few years even though there haven’t been any improvements to transit (there’s the Muni bus improvements, but that’s not where the big ridership increases are). In LA, I doubt it’s the case.

  3. krschultz

    Yesterday I was just thinking as I walked up Broadway between Union Square & Madison Square park that I have been working at different companies on this block for a decade. It’s kind of amazing how many small companies there are that bounce around the same (relatively crappy) office spaces. I do love that neighborhood though.

    The primary reason I haven’t seen tech companies move to downtown Brooklyn is that a pretty large group of workers live in New Jersey. Every startup I’ve been at had probably 20-30% of the workforce coming in through Penn Station or the PATH. Getting to Brooklyn is a pain for anyone coming in on an NJ Transit train. I’m not sure about the PATH.

          • adirondacker12800

            Atlantic Terminal is… a terminal….
            2035, 2040 they will need more tracks into Manhattan. Connecting Jamaica to the Meadowlands via Brooklyn and Wall Street means the Major Investment Study should have started in 2015.

  4. John

    I think you discount families, where say one spouce is going Upper East Side to Midtown while the other may be going to Downtown or Brooklyn.

    • Alon Levy

      I’m not discounting them at all! I’ve argued two-income families as a good reason to do through-running. But it’s less common than living based on commuting to your own workplace rather than your partner’s, so a job in Downtown Brooklyn is much likelier to generate a commuter from elsewhere in Brooklyn than from the Upper East Side.

  5. FlavaNation

    While the L train is less crowded west of Union Square, there are passenger circulation issues at 8th Avenue. Another staircase to the platform would certainly be helpful, as well as allowing the current west staircase to be a transfer opportunity within the paid fare-zone (Sometimes with my unlimited metrocard I’ll make this transfer via this route anyway. Surely they could reshuffle the space in a way to allow this to be a transfer within the fare zone.) Other issues that have been brought up before: Tail tracks at 8th Ave would certainly help, as well as additional substations on the L to get the TPH up to 30 trains per hour. The upcoming shutdown should be an opportunity for the MTA to bring the L train up to world-class standards.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.