|The three classic urban ecology models. The Burgess Concentric Model (left), the Hoyt Sector Model (middle) and the Harris/Ullman Multiple Nuclei Model (right). With respect to the New Urban Transect Model, is it time for a new one? Source: csiss.org|
I’m going to do something I never before thought I’d do — I’m going to agree with something produced by Wendell Cox, perhaps one of the world’s leading apologists for conventional suburban-style development. He has a working model for a metropolitan area typology, and huge caveats aside, I think he’s onto something.
Cox is well known in some urbanist circles for his Demographia website and his crusade against Smart Growth and New Urbanist advocates and activists. His work can be described at times as libertarian in orientation (“let the people decide the kind of living environment they want”) and at other times conservative (“land use regulation is destroying the American middle class”). He’s been a leading proponent of the notion that the 2008 financial crisis and subsequent housing crash happened not because of the too-widespread availability of mortgages, but because strict land use regulations kept housing supply down. Whatever.
But he’s come out with something this week that rings more true. Cox has developed a sector model for U.S. metropolitan areas, looking at metros by development type rather than municipalities. Looking at population and housing data for zip codes in all of the metro areas with more than one million in population, Cox developed four functional classifications: 1) urban core, 2) early suburban areas, 3) later suburban areas, and 4) exurban areas. He’s developed a set of metro area maps that show how the four classifications are distributed in each metro area. Using his criteria, one can see how a metro area like New York or Chicago have rather expansive urban cores and nearly equal parts in the other three categories, while Atlanta or Charlotte have tiny or non-existent urban cores by his definition, and are almost exclusively suburban in orientation. it’s worth a look.
I’ve been pushing for this kind of development typology myself, but I’ve noted before where some of Cox’s work may have a blind spot:
Look at it this way. Boston may be 100% from the industrial era, and many of its adjacent suburbs may approach that figure. You may have to continue outward from Boston for some distance before you find a suburb outside of the city whose development makeup is more than 50% from the auto era. Meanwhile, in Phoenix the urban core may represent less than 10% of the innermost portions of the city, and the balance of the entire metropolitan area is comprised of auto era development.
Wendell Cox seems to recognize one part of this — he realizes that many Sun Belt cities are suburban in orientation. However, it seems he neglects the opposite of this, that Jersey City or Arlington, VA are urban suburban places.
And that is the core critique I have for his sector model for cities. Four points show where this model falls short:
- The “urban core” definition is too narrow. There are many urban cores that may approach or exceed Cox’s 7,500 per square mile density threshold, but lack the public transit or bike infrastructure to have 20 percent of residents utilize transit, walking or biking as a mode of transit. By expanding this definition, more areas would qualify as part of the urban core.
- The “early suburban” definition is too broad. Cox uses the median year of housing structure built data point as a proxy for development type, something I’ve done too. However, the broad range that he uses for “early suburban”, for areas with homes whose median year of construction is between 1946 and 1979, is too broad. That puts early “Levittown” style homes in the same genre as latter split-level homes, and that’s too broad.
- “Later suburbs” start later. I’d say the later suburban push was a phenomenon that started in the 1990’s, with “McMansions” emerging as a recognizable term and teardowns becoming a thing in rapidly growing suburbs.
- In fact, there should be post-urban core suburban phases, not two. I’d say that following the development of urban cores, there was a period of development that took place at the outlying areas of large cities as well as in the suburbs — what I call the “Levittown” era of development. This type of area is defined by small homes on small lots, creating moderately high densities, but with an overwhelmingly strong single-family home orientation. This area is generally an urban periphery/inner ring suburban phenomenon, and is deserving of its own classification. Similarly, the “McMansion” era development type that emerged after 1990 is deserving of its own classification, and the “split-level” development type that fit between the two is deserving of one itself.
- Redevelopment in urban cores is poorly accounted for. There are a few places — the Cleveland, Detroit and Milwaukee maps stick out — where there are “early suburban” areas located deep within cores. The only way I can explain this is that late stage urban renewal efforts and recent redevelopment projects that cleared lots of land of older housing and replaced it with something newer caused a shift in the typology. I don’t think that would necessarily be the case. I suspect this could be a flaw in Cox’s data that goes beyond the cities I’ve mentioned.