Friday, December 2, 2011

Living Space, Working Space and Centralization

Reading Alon Levy's recent post on transit and employment centralization started me thinking about the urban consumption of space, at home and at work. Does an office district, during working hours, contain an even denser concentration of people than its tall buildings might outwardly suggest? And what implications might this have for commutes, and for proximity of homes to workplaces?

Census figures tell us that Manhattan, which Alon discusses, has 2.30 million jobs and 1.59 million residents. What we want to know here, however, is how much space each person occupies, on average, both at home and at work. This requires determining the total amount of residential and non-residential square footage in Manhattan.

For commercial space, the task is easy, as the New York Property Tax Report catalogs it by gross square footage, showing a total of 591.6 million square feet of office and other non- residential space. Residential property poses more of a challenge, since the Report only lists the number of housing units, rather than the total square footage of residential space. What we do know, from 2010 Census Data, ACS information and real estate records, is 1) the total number of housing units in Manhattan; 2) the distribution of these housing units among studios, 1 BRs, 2BRs, etc., and 3) the average square footage of each type of unit sold over the past several years.

In combination, these three factors permit a rough estimate of the total square footage of residential space on Manhattan: 975.3 million square feet (representing the average apartment of 1151 square feet multiplied by 847,090 total residential units). Adding so-called "non-yield" space – including a building's hallways, lobbies and utility rooms, estimated at between 15 and 30 percent of an apartment building in colder climates – results in a gross residential area of 1.258 billion square feet. To find the square footage per person, all that remains is to adjust for vacant offices and apartments, again using census data:

These figures show the gross square footage per person, per building, with non-yield space included in the equation. Private apartment space per person would be considerably less – 553 square feet. (For anyone curious, the numbers also permit calculating an approximate floor area ratio for all of Manhattan: 1,850 m.s.f./640 m.s.f = 2.89 FAR).

What significance do these numbers have for the distribution of offices and residences in a city?  They do suggest the difficulty of achieving high proportions of commutes by foot or bike in a city dominated by a compact central business district of office towers, as a single tower will require more than triple its total square footage (in fact, more than that, as population figures include non-working age persons) in residential space to achieve a 1:1 match of living to working space on a per person basis.  Yet, even in Manhattan, residential space is overwhelmingly mid-rise.  The very dense Upper East Side can supply only a fraction of the workers, within walking or biking distance, needed to fill the office buildings of Midtown.  At its best, it will be a transit city; far more often, in the United States, it will be a car-dependent city.

As urban living becomes increasingly desirable, however, this balance can shift.  Conversions of Class B or C office buildings into condos can swiftly change the resident/employment ratio, as has happened in lower Manhattan.  Could the same ever happen in Midtown?  A little employment decentralization, after all, might not be such a terrible thing.


  1. True, as long as further job nodes are transit-accessible (e.g. Shibuya, Shinjuku, Akihabara, etc., in Tokyo).

    The corollary of this is that, if (as?) autos cease to be viable, the city will need to redevelop as a transit city.

  2. Steve – I was thinking along the lines of an even dispersal of certain workplaces among dense residential sections, rather than discrete nodes, although a combination of both could work as well. It doesn't take much office space to serve a large residential area due to the modest spatial needs of office workers. In fact, the mismatch is even larger than I've shown here, since I initially forgot that something like 25-35% of the population is not of working age (plus unemployed, retired, etc). To fill its current employment needs from the population on the island alone, Manhattan would require something like 3.3 million inhabitants, double what it has now.

  3. @Charlie — "even dispersal" of employment is a poor mismatch for high-capacity transit like rail, though, and in a city the size we're talking about, high-capacity transit is almost certainly necessary. "Spiky" dispersal, on the other hand, works very well.

    I think "Spiky" also results in a more attractive urban environment: instead of a constant "medium" you have centers of high-intensity (= excitement) amidst larger areas of relative calm. High-density centers can support many uses that wouldn't really work well otherwise, and yet spiky distribution can easily support medium- and low-density uses as well. Variation is a also just a good thing generally—constant sameness is boring.

    [Business and residential uses actually seem much less segregated in Tokyo than in many big American cities — a large proportion of the big business/entertainment centers have significant amounts of housing of all sorts within an easy walk — but I don't really think it's workable to expect all workers to be within walking/biking distance of their employment, especially these days (people are used to more flexibility).]

  4. [er, obviously "poor mismatch" should read "poor match"!]


  5. Charlie, I agree with snogglethorpe here. At the scale we're talking about, it is impossible for a wholly pedestrian-oriented city to exist. I know you have the Paris core in mind, but remember that the historic urban cores in Europe are usually--though not quite always--complemented with near-suburban intense job centers: Paris' La Défense, London's Docklands, Amsterdam's Zuidas, Lisbon's Parque das Nações, etc. etc.

    While these examples are all centrally planned--didn't Midtown Manhattan first grow in the same way?

    You can also persuasively argue that every passenger rail innovation of the 19th century was spurred on by the need to provide more efficient movement patterns than what full pedestrian-centrism can allow.

    And in any event, remember Nathan Lewis recommends the use of rail transportation to link fabric together that is beyond pedestrian scale.

  6. Yeah I don't think a wholly pedestrian city would be very feasible. Even if you could housing all the workers of an office tower within walking distance, it wouldn't matter because people change jobs more often than they wish to change homes nowadays, and you have double income households to consider.

    I think what makes sense is to have buildings that generate the most trips per area closest to transit. I think in this sense, office towers have an even greater advantage that the stats posted in this article. First of all, the United States labour force only makes up 50% of the population, so the average person takes up almost 6 times more residential square footage than commercial. Next, offices and retail can both have larger floorplates, which allows for a higher FAR for buildings of the same height. I'm pretty sure offices take up less space per employee than retail too.

    I think what would make sense would be to have office towers around transit hubs, and then residential highrises surrounding the offices and around rapid transit stations, with residential midrises and lowrises around more local transit lines.

    That's more or less what exists in Toronto. The financial district which is almost entirely offices is located next to Union Station, which is the hub of the commuter rail network. A subway line also goes around the edge of the financial district, with streetcar lines going through it. This is the densest part of the city in terms of jobs+residents/acre. The next densest node of density is in the Northern part of downtown at the intersection of the Bloor and Yonge subway lines. Yonge & Bloor probably has a comparable amount of jobs and residents, which means most buildings are residential highrises, although with a few office towers. In between the two nodes, along the subway line are a mix of offices and residential highrises and within walking distance of these nodes are many areas that are predominantly residential highrises. Further out are dense midrise and lowrise neighbourhoods served by local transit (streetcar and bus).

    Having a dense, predominantly office district is not necessarily a bad thing. The centralization of employment in Calgary is often said to be a reason why Calgary's light rail system is so effective.

    Manhattan is an extreme case. A city the size of New York could support several office districts/transit hubs.

  7. @Snogglethorpe: Yes, sorry for the lack of an edit function. Blogger’s comment capabilities leave a lot to be desired (nested comments, in particular, would be nice to have, and don’t seem like such a technological leap).

    As for my comment on dispersal of employment, I was referring to a dispersal only of a relatively small number of offices (“a little decentralization”) within the context of an already-dense city. The goal would be to modestly increase the proportion of people able to walk (or bike) to work, without impairing transit efficiency elsewhere. This will involve increasing provision for pedestrian and bike-friendly infrastructure as well. I agree with Steve, though, that an entirely pedestrian city would not be possible at this scale, nor did I intend to suggest that it would.

    @Nicolas: Interesting information about Toronto. I’ll have to do more background on transit there. And you are right about the true ratio of residential to office space – I overlooked the labor force participation rate. I don’t think dense office districts are necessarily bad, either, although I’ve advocated for a more polycentric employment distribution in some earlier posts. What I was suggesting in the post was really intended to be only a slight modification of that idea.

  8. Nicolas,

    Toronto sounds a lot like Chicago. I will have to do some research on Toronto now.

  9. It seems that this discussion is tossing around walkable / bikable as if they are equivalent. But they are substantially different.

    While a quarter mile radius around a transit stop is taken to be a common walkable radius, that is to get to a high frequency local multi-use core and also to get to a transit stop from which you catch transit to complete a trip ... a pedestrian may prefer this as a commute to work (as a motorist may prefer to be a 10 minute drive from work), but a mile door to door is certainly walkable as a regular door to door commute.

    Whatever a reasonable pedestrian commute is, you can easily triple that for a cycle commute, and easily quintuple it for an ebike commute.

    For a residential fringe on both sides of a dense long axial CBD, a half a mile wide, would have a core pedestrian residential fringe 1/2 mile on both sides, for a 2:1 ratio, and a further 2 miles on both sides of bikeable fringe, which is an 8:1 ratio, giving a total 10:1 ratio. If the required floor space ratio is 6:1, that only allows the professional space to be 1.67 times the floor space per acre.

    A narrower dense long axial CBD, a quarter mile wide (essentially a continuous sequence of 1/4 mile walkable districts) would have a core pedestrian fringe 3/4 mile wide on both sides, for 5:1, and a further 2 1/4 bikable fringe on both sides, for 20:1. 25:1 allows office space to be 4 times the square feet per acre.

    Of course, transport along the long axial CBD with Active Transport as the primary commute would require a dedicated transit corridor, and two career households would have a strong preference to reside sufficiently close to that transit corridor for one or both to commute using that corridor.

    For a clustered geography, a dense CBD cluster of 1/2 mile radius would be located within a 1 mile radius pedestrian fringe, for a 3:1 ratio and a 3 mile radius bikable fringe, which brings it up to a 17:1 total ratio of fringe to dense CBD. That would allow for almost 3 times the square feet per acre to the professional space.

    In terms of the balance between commuting transit as a useful amenity and commuting transit as a structural necessity, a key dimension is the distance from the center of the dense CBD district to the nearest edge. A porous course grid of dense CBD districts laid out along separated main thoroughfares, or a grid of clusters of small dense districts has a lot of "surface area" to its footprint. Collapse the same professional office space into a compact CBD, and the average distance to the edge increases substantially.

    Of course, even there was a city with professional space distributed so that it is physically possible to rely entirely on active transport for the commute to work, that city is going to be a transit city, because a city of pedestrians and cyclists is going to demand and be willing to pay for longer distance transit as an essential urban amenity.

  10. Thanks for that comment, Bruce, and for looking at the implications of these numbers. You refer to "the balance between commuting transit as a useful amenity and commuting transit as a structural necessity," which is the essence of what I was trying to get at here. The basic concept is the city in which the greatest possible number of people at least have the option of biking or walking to work, while of course retaining transit as an amenity for less frequent (non-daily) trips.

    It does not mean everyone, or perhaps even a majority, but certainly more than 10 percent. I suspect most would prefer a 10-20 minute walk to a bus or subway ride of the same length, assuming a reasonably dense urban environment where basic services and shopping needs were available along the route. Biking, though, may have less mass appeal given weather and especially traffic conditions. Certainly it would require a much greater dedication to bicycle infrastructure than any American city has shown so far. Nathan Lewis has argued that if you really do need a bike for daily transportation, you may as well take transit:

  11. The other thing about biking is that it's not really suitable for people wearing dresses or dress pants (or nice shoes/heels). So if you work in a formal environment, you'll need a change of clothes, and possibly a shower.

    I definitely agree about a 10-20 walk being preferable to a 10-20 transit ride. I takes me about 25 minutes to walk to university and 15-25 minutes to take the bus and I almost always walk.

  12. Nicolas, perhaps you have not seen this, if you say "[bikes are] not really suitable for people wearing dresses or dress pants": [Dresses - Copenhagen] [Suits - Italy]
    Lot's of people ride in nice clothing, in the Netherlands and Denmark and Italy and Japan. Just like people walk to work, or walk to the train station, in those places.

  13. Re: "Nathan Lewis has argued that if you really do need a bike for daily transportation, you may as well take transit."

    This is somewhat true, especially since some people with disabilities cannot ride bikes, so you need transit as an alternative. But I would also argue that the inverse statement can be equally true for a healthy adult. "If you need transit for daily transportation, you may as well use a bike." With few exceptions, a bike is just as fast, or faster, than transit for most trips. The exception are long trips on express buses or commuter trains, or inter-city trips. Compared to an urban, local-stop bus or streetcar, which average 8 to 12 mph, a bike is almost always faster, door-to-door. And if you need to transfer once or twice for a transit trip, even a grade-separated subway will rarely win versus a bike:

    Making bikes possible does take a great deal of infrastructure and some space, just as good transit requires investment of political will, money and space.

    In the example above, Manhattan can best be served by transit because it is such a high-density employment center. There isn't enough room for all the bike parking that would be needed, even if every car parking space were turned into 10 or 12 bike parking spaces. But in most North American cities, which are currently car-dominated, bikes can be a good option if the right infrastructure is provided.

  14. I actually do bike to university in Ontario when the weather is nice, since it only takes about 8min and taking the bus would take 15-30min and walking about 25min. However, this year, there was a cool raining falling on every other day in October and by November it was around 40F most days so I switched to walking. In Toronto, only the very hardcore ones bike all year.

    And while I guess you can bike in a skirt if you're ok with it occasionally being blown up, you're not going to be going at 17mph like the girl on the racing bike in that race. I still agree that it makes sense to invest in biking infrastructure for medium distance trips, but most commutes are too long to bike, so rapid transit makes more sense imo.

  15. Hi, I'm a bit confused with the FSR/FAR numbers. So what is the floor area ratio of the whole of Manhattan, if there's such a number. Or, the FAR of zoning districts such as C5, C6 and the special zoning districts if I only wanted the data for Manhattan. Any help is appreciated. thanks!

    1. 2.89 was my best guess at total FAR for Manhattan (i.e. there is approximately 100 square miles of floor space on the island). For the particular zoning districts, it would require neighborhood-specific data, and I'm not sure that is readily available.

    2. following your calculations
      1.850.100.000 feet2 = 166.509.000 m2 = 166 km2 = 65 sm
      i think that the result set is 65 square miles of floor space.
      is it correct?