Monday, September 12, 2011

Skyscrapers: Cause and Effect

Last week, Planetizen ran an excellent piece on the role of skyscrapers in today's cities, contrasting the arguments of James Howard Kunstler and Ed Glaeser along with commentary from the engineers and architects responsible for designing and tall buildings and their infrastructure.  It examines arguments based on comparative energy use, maintenance and refurbishment and on the effect on urbanism and city life, ultimately concluding that whether skyscrapers are appropriate in a particular case is a complex question that will turn on a variety of contextual factors.

This is a topic that I've discussed before, here, here and here.  But I'd raise one more question: is it really possible to separate a discussion on the role of the skyscraper from the land use patterns and zoning constraints of the typical North American city?

Leon Krier has an illustration on this point in The Architecture of Community, where he depicts a recognizably American city of a handful of skyscrapers surrounded by a vast sea of detached single family homes next to his preferred model of the polycentric city with multiple nodes, none obviously larger in scale or importance to one another, under which he observes:

"Exactly like an individual who has reached maturity, a 'mature' city cannot grow bigger or spread out (vertically or horizontally) without losing its essential quality. Just like a family of individuals, a city can grow only by reproduction and multiplication, that is, by becoming polycentric and polynuclear."
The North American skyscraper, arguably, is only a symptom of the failure of cities to adhere to this growth model, a model that is inherently dense and mixed-used to enable the provision of convenient urban services and amenities to each quarter. 

Consider one of the best examples of the North American type: Vancouver, which avoided the construction of downtown freeways, and which has a large and efficient mass transit network.  As can easily be seen from an aerial map, Vancouver is a monocentric city, with commercial high-rises heavily concentrated in the downtown area outlined in red, representing just over 1% of the greater metro area:

The Vancouver city government confirms this, noting that over two-thirds of all city jobs (and 22 percent of regional jobs) are located in the immediate downtown area.  In the same document, the city tries to allay fears that jobs are "moving out to the suburbs," even to the extent of touting zoning exclusively for commercial use (i.e., subsidizing office space).  Unsurprisingly, the city notes that 102,800 commute in to the central business district each day, while 16,400 commute out.  The transit system itself reflects this monocentric arrangement, as all lines of the SkyTrain converge on the small downtown.  Outside the downtown, suburban-style single-family detached housing predominates, occupying the vast majority of city area -- perhaps in excess of 90%.

With this sort of development pattern, is it possible not to have a dense core of skyscrapers, regardless of what one may think of them?  When residential areas are 1) low density and 2) zone out commercial and industrial over large areas, the remaining office areas, confined to the original central business district by an ever-expanding belt of homes, have nowhere to go but up.  The exception, Washington D.C., its height constrained by law and its building footprints hemmed in by streets of excessive width, compensates by spreading its downtown horizontally and further outwards to several centers beyond its jurisdictional boundaries.

The mass transit system, in response to this arrangement, bends all lines toward the core, yet this only worsens the imbalance by further raising property values at the center.  At last, we arrive at the familiar arrangement of 500-foot office buildings surrounded by mile upon mile of low-density homes, until the bursting commercial downtown leapfrogs out to an edge city on the residential fringe, leading to hand-wringing in the city council about the "decline" of the downtown.  (Tyson's Corner lookalike Richmond, along with Burnaby, seem to fulfill that function in Vancouver, being as commercial-heavy, but even more auto-centric, than the downtown).  Can the skyscraper really be blamed for this state of affairs? 

On the other hand, an urbanism on the model Krier proposes, which exists in countless cities outside North America, including those of the United States' own southern neighbor, has no economic need for skyscrapers, and sprouts only a few, those driven by corporate status-seeking rather than necessity.  The issue becomes nearly moot, and height limits, where they are implemented, less contentious. 

A Vancouver built to a Parisian level of density, and imitating Paris' lack of Euclidean zoning, would occupy the area shown below -- barely 10% of the existing regional metro area, with no high-rises, and with the same population.  The key difference, though, being that while today's Vancouver (and dozens or hundreds of cities like it) is built out with a giant and immovable low-density residential buffer, forcing the city to shunt its residential growth into towers in the already overburdened downtown, or to engage in a difficult process of upzoning low density residential, in this model, the city has abundant room for growth within its own borders by "reproduction and multiplication." 


  1. Density itself is not a goal. The goal is quality of life. If people want to spread out, then the goal should be balancing their desires against other issues. Urbanists tend to get obsessed with the idea of packing 'em in rather than accomodating what people want.

  2. Los Angeles would seem to have a variation of that polycentric model that Leon Krier describes. Although there is a forest of skyscrapers downtown surrounded by low-rise development, there are a number of other nodes (such as Century City on the Westside, with a distinctly recognizable skyline) in the city, which was historically and remains one of the most decentralized with regards to employment in the United States.

    Incidentally, I also recently finished Robert Fogelson's "Downtown," which recounts how the distinctly American downtown, with its clump of high-value, low-residential, almost entirely commercial buildings, came to be. Very similarly to what you're saying here, he attributes it an early belief in separation of land use: Americans of the late 19th century believed that it was natural for residences to spread out and for high-value commercial enterprises to fill in the city core.

  3. Hi Andrew – thanks for commenting. Los Angeles has somehow become the poster child for 20th century sprawl, but as you allude to that's quite far from the truth. A mostly polycentric distribution is enabled by a relatively high population density throughout the city, far higher than in most American suburbs (thus fueling the ongoing debate as to which city is denser – NYC or LA).

    Other American cities are polycentric to some degree, or at least started as polycentric before the arrival of structural steel, elevators and mass transit made the concentration of huge amounts office space in a small area a physical possibility. Boston could be one such example.

  4. Funny thing about Boston (and nearby towns) is that it's becoming polycentric again. The core downtown is still occupied extensively by financial businesses, and there are efforts to redevelop the old Southie waterfront, but there are also an enormous number of high-tech companies spaced out around the periphery or near the universities.

  5. I have to strongly disagree with you that skyscrapers are the result only of suppression of density elsewhere. I think the most obviously counterpoint were American urban areas around the turn of the century, which were characterized by very strong and tall downtowns and relatively laissez-faire land use policies (standardized street widths notwithstanding). (Japanese cities today have relatively few skyscrapers, though as I understand it they also have zoning codes that are surprisingly hostile to height, so I'm not sure we can learn much about the skyscraper equilibrium there.)

    The reason is pretty simple geometry. Assuming that a metropolitan area is a circle (or even any reasonably compact shape – even a square), then the best place for businesses to locate is in the exact center, where they have access to the largest possible pool of workers and customers. That is to say, the problem with polycentric cities is that many people cannot access many of the jobs – surely they have less access than a monocentric city centered.

    And in fact, I would say that American cities today would be more polycentric than they were a century ago if density restrictions were loosened, because back then there were more people working in agriculture and industry, where locating downtown was not feasible, for a number of reasons.

  6. D'oh! *cities would be more monocentric.

  7. After studying the history of Chicago I have to conclude that the American model came about because of technology. Before the car and the expressway, the train carried commuters from the suburbs into downtown for work. And why would they want to live downtown? 100 years ago downtown Chicago was full of factories with pollution so thick it was often hard to see the sun. Zoning came later.

  8. Stephen -- there's a lot of economic literature out there comparing polycentric and monocentric models, and I've barely scratched the surface of it, so I can't claim to be speaking with the support of much authority here. My initial thoughts were that polycentric city is advantageous for several reasons:

    1) There is the potential for each person to live within walking, or at least biking distance, of their workplace, so long as there is residential mobility. This will be virtually impossible in a large monocentric city, which will rely on an extensive radial mass transit/highway system to permit the convergence of population on the center.

    2) As a corollary to 1), the transportation needs of the dense polycentric city will be less, and more efficiently used (less of the crowded in/empty out phenomenon of the monocentric city's commuter trains).

    3) Multiple centers provide a reliable release valve for commercial property demand. The monocentric American city center has some bizarre land use dynamics going on that are a function of extremely high demand, driven in part by the artificial scarcity of zoning, and reinforced by speculation abetted by property tax regimes that reward holding of vacant land. Worse, when the edge city does appear in response to this downtown situation, it is not a result of careful planning (often being in another county altogether) and mass transit therefore does not serve the new area, resulting in even greater car dependency than before.

    I also tend to doubt that this extreme concentration would persist if density/zoning were relaxed. The first-ring streetcar suburbs around many an American downtown ought to have been opened to commercial use decades ago -- zoning has massively suppressed land values there for many years. Cast zoning aside and you'll see land values decline downtown, reducing the demand for skyscrapers, and rise in the surrounding areas, producing a more generally mid-rise city (as New York mostly was as recently as 1900).

  9. Charlie, Paris is an exceptional case. Let's compare Vancouver to some other cities, density-wise:

    Paris, ex-Bois: 25,534
    Barcelona: 15,991
    Athens: 16,830
    Geneva: 12,058
    Moscow: 10,652
    New York, 2009 ACS: 10,630
    Naples: 8,215
    Tel Aviv: 7,868
    Milan: 7,247
    Brussels: 7,025
    Lisbon: 6,429
    Copenhagen: 6,142
    Madrid: 5,403
    Vancouver: 5,335
    London: 4,978
    Stockholm: 4,527
    Dublin: 4,398
    Munich: 4,359
    Vienna: 4,134
    Zurich: 4,049
    Berlin: 3,890
    Amsterdam: 3,506
    St. Petersburg: 3,369
    Helsinki: 2,755
    Frankfurt: 2,737
    Rome: 2,149
    Oslo: 1,333

    So, sure, Vancouver would be denser if it had Paris's density. So would all other Western cities, unless Manhattan seceded from the rest of New York and then it would not be much denser. It's a nice exercise or limiting case, but it's not a good criticism of Vancouver's policy.

    As an aside, Vancouver is much less monocentric than you think it is. It has a lot of skyscrapers downtown, but many are residential. And it's been pretty good about developing both residential and commercial buildings along the Expo Line. While the current Skytrain network is downtown-oriented, the next major expansion proposed (not counting the Evergreen Line) is a line on Broadway connecting to UBC and allowing non-downtown transfers between the Millennium and Canada Lines.

  10. Alon – good points and thanks for the info. In my defense I did state that Vancouver was one of the "best" North American examples – far denser than either Portland or Seattle for instance. On the other hand, Vancouver only has one neighborhood denser than 30,000/sq. mi. (the West End, which is very dense but not terribly urban, consisting mostly of mid-rise standalone apartments from the last 50 years with few stores and hardly any offices due to restrictive zoning). That is, in the entire Vancouver metro area, only 2 percent of the population lives at a density comparable to, say, Greenwich Village. Instead of Paris, consider Marseilles, a city smaller than Vancouver and with few high-rises where 13 percent of the population lives at densities of 50,000+.

    The downtown highrises are the most noticeable element of densification, but more significant in the long run is the upzoning of the single-family areas which constitute about 80% of the city area. There too with laneway houses and creative infill Vancouver has been running far ahead of most other comparable cities. I'll do a follow up post looking at the zoning in somewhat more detail to show why I'm still sticking to my guns on the fundamental point here.

  11. I got very surprised when I visited some north american cities this summer. They all seem to have small downtowns comparable to European cities of similar size.

    I visited Miami, Chicago, Montreal, Toronto & New York. When walking through downtown in most of these cities I felt that I had to be careful not to suddenly walk out of the busy downtown.

    I think that its kind of extraordinary that a city the size of Toronto can feel small.

    I also find the density comparison of some European cities and Vancouver very interesting. I'm not so very familiar to all those european cities. But I know stockholm a bit. And if you compare the way Stockholm spreads out with a typical american city grid with detached housing. You see that the american city has a very similar density trough out. While Stockholm is more varied with a blend of quite dense and nondense suburbs + a lot of fields & forests in between.

  12. Dear All,
    I believe that Charlie's point is that average density is not necessarily the best measure of the density of a given area because this measure quantifies density as equal to total population divided by area. While that is the raw definition of density, it does not take into account the idea of the bell curve, namely,North American metros' highest density will be more than a European metro's highest density. However,in the North American model the difference between high and average densities is far more dramatic than that of Europe where the average density is much closer to the highest density. In the case of European cities their highest densities tend to be rather dense. In summary, the North American metros have regions of much higher population density than those found in Europe by the same token these North American cities also have regions of much lower density than the average European city's density.

  13. Now, the monocentric versus polycentric debate is complex. The polycentric model does allow for a higher "walkability to density" ratio (i.e an area doesn't need to be as dense as it would normally be to achieve the same walkability/proximity)due to the mixed-use nature of polycentric cities. However, this does lead to the fragmentation of cities as the develop multiple "minds" or city centers and this ruins the cohesion of the city as a whole. This is where the advantage of a monocentric model comes to play. I submit that the optimum situation is to plan for a monocentric CBD (central business district) along with the many cultural,civic,and "tourist" amenities that come along with a CBD while at the same time encouraging a mixed-use (here I make a distinction between mixed-use and polycentric) approach in the outlying regions.

  14. Hi Anonymous:

    On the polycentric v. monocentric debate, I agree with your comments. My vision of a "polycentric" city was not one of dense nodes surrounded by low density areas, but a city of more or less consistent density (albeit with one CBD perhaps, a first among equals) composed of urban neighborhoods with their own identifiable centers. Thus each neighborhood exerts a local gravitational pull on its surrounding area which is equal to or greater than that of the center, rather than the monocentric city in which the gravitational pull of the center is felt at great distances. My own experience is that of living in Rome, where the Trastevere neighborhood was simultaneously its own self-contained city, centered on a public square, while seamlessly integrating into the broader urban fabric of Rome, but I can think of many other examples.