Our virtually instant analysis of 2000 census trends in metropolitan areas has the generated wide interest. The principal purpose is to chronicle the change in metropolitan area population and the extent to which that change occurred in the urban core as opposed to suburban areas.
From a policy perspective, this is especially timely because of the recurring report that suburbanites have been moving to the urban core over the last decade. We have dealt with this issue extensively, noting the lack of data for any such interpretation. As of this writing, with data for more than half of the major metropolitan areas (over 1,000,000 population) in, there remains virtually no evidence that people are "moving back to the city" (actually, most suburban growth came from outside metropolitan areas, not from the "cities").
The Policy Context: Urban Cores and Suburbs
This discussion is not new, and generally pits anti-automobile interests – including much of the urban planning community – who favor the urban development patterns of prewar America (generally the urban planning community) against those who would prefer allowing people to make their own choices about where they live or work..
Over the past 60 or more years, the data indicates that consumers have nearly exclusively chosen less dense and more suburban areas. This is not to suggest, however that many of us, including this author, automatically favor suburbs over urban cores. Indeed, I have enjoyed years of alternating between living in suburban America and the urban core of the (inner) ville de Paris (arrondissements I, II, V, VII and XI). But if you have a taste for urban living, that does not mean high-density cities are inherently superior to suburban living. People, after all, have different preferences.
Urban areas include both urban cores and suburbs. The delineation of urban cores and suburbs is subjective. There was for example a time – say around 1820 – when development to the north of New York’s Houston Street would have been considered suburban. More than two thirds of the present ville de Paris was suburban before the city limits were expanded in the 1860s. Now, no one would consider, for example, Washington Square or Herald Square to be suburban and the suburbs of Paris now extended to more than 80 times the land area of the 1860s ville de Paris.
One overlooked way to approach the current debate would be to look not at municipal boundaries but forms of development. Around 1950 we began the breakneck expansion of automobile oriented suburbanization which had proceeded more modestly for two or more decades before.
The Urban Core:
This analysis defines the urban core consistent with the criteria of the US Bureau of the Census in 1950. Metropolitan areas are organized around urban areas (urbanized areas). We use the "central cities" of the core urban areas in 1950 as the urban core in the analysis. Those portions outside the 1950 urban core are thus considered suburban. Where an urban area did not exist in 1950 (such as in Las Vegas and Tucson), the urban core is the central city of the urban area when it was first established.
No existing specification of the urban core is ideal, though the present one is appropriate for the policy purpose stated above. Clearly, the urban core would be far better defined at the census tract or even census block level based upon the characteristics of an urban core. This would include factors such as high residential population density, high transit usage, walkability and a high percentage of multiple unit residential buildings.
Such an ideal definition of the urban core cannot be measured with municipal boundaries. Yet, municipal boundaries have routinely been used by researchers to delineate the urban core, not least because the data is readily available. However there three notable difficulties with the use of municipal boundaries to define the urban core.
First; some areas with urban core characteristics are outside the core municipalities. As The Infrastructurist notes, municipalities like Jersey City or Hoboken have the characteristics of urban cores. However, since they are not a part of the core municipality (city of New York), they are classified as suburbs in our analysis. It is well to remember that both Hoboken and Jersey City represented suburban development, during their period of greatest growth, before 1930.
Second, other areas with postwar suburban characteristics are inside the core municipalities. For example, Richmond County (Staten Island), a part of the city of New York is principally suburban. Much of it was developed well after 1950 and consists largely of single family homes. The median construction date of owner occupied housing in Staten Island is 1970, which compares to 1965 in adjacent Middlesex County, New Jersey. It is newer than in Morris County New Jersey (1965), much of which is outside the urban area (all median house construction years from the 2000 census). Major portions of core municipalities such as Los Angeles, Houston, Dallas, Portland, Seattle, Denver and others are also postwar suburban.
Third, in a number of core municipalities, there is little, if any urban core, at least from a residential perspective. For example, one would be hard-pressed to identify an urban core in municipalities such as Phoenix or San Jose (despite the fact that the San Jose urban area is more dense than New York urban area). In metropolitan areas such as these, it might be preferable to define virtually all growth as suburban, though our analysis still defines these municipalities as the urban core.
Based upon the early results from the census it seems that if the more ideal census tract-based urban core definition were used, the urban cores would be shown to be capturing an even smaller share of growth, while suburban areas would be capturing more. But this analysis will have to wait until all the numbers are in.
Historical Core Municipality
The term "historical core municipality" is used to denote the urban cores using municipal boundaries. The term "city" is avoided because of its multiple definitions. Cities can be municipalities (such as in the city of New York), urban areas (such as the New York urban area), metropolitan areas (such as the New York metropolitan area) or multi-county regions or prefectures of countries like China (such as Wuhan or Shenyang).
This lack of clarity can be routinely seen in media reports that indiscriminately (and without comprehension) make comparisons between cities, using differing definitions. This can even extend even to more technical literature (see pages 12-14 of Urban Transportation Policy Requires Factual Foundations).
Principal Cities: Starting in 2003, the Census Bureau substituted the term "principal city" for the previous "central city" term. The use of principal city designations and the largest municipality as the principal name of a metropolitan area are appropriate for the purposes intended by the Census Bureau.
In its State of Metropolitan America, the Brookings Institution uses up to the three largest principal cities (which it calls "primary cities") and consider other parts of metropolitan areas as suburbs.
Neither approach, however, is appropriate in analyzing postwar suburbanization. Any municipality in a metropolitan area with more than 250,000 population is considered a principal city, regardless of its urban form. Any municipality with more than 50,000 population but which also has more jobs than resident workers is also a principal city, regardless of its actual on the ground reality.
This leads to a situation in which, for example, Los Angeles has 26 principal cities. Any postwar urban form definition would classify nearly all as suburban (and much of the historical core municipality of Los Angeles, notably the San Fernando Valley, itself is suburban). For example, the suburban city of Cerritos is a principal city, yet was largely filled by dairy farms well into the 1950s and was called Dairy Valley.
Other principal cities hardly existed in 1950. Virginia Beach has become the largest municipality in its metropolitan area, having displaced Norfolk. Yet, in 1950 Virginia Beach had a population of only 5,400, well below the 50,000 threshold that was required of central cities (smaller than Ponchatoula, Louisiana, doubtless an unfamiliar municipality to most readers). Arlington, Texas, the third municipality in the Dallas-Fort Worth-Arlington metropolitan area, had a population of 7,700 in 1950, again well below the central city threshold. Arlington is not an urban core, it is a suburban jurisdiction.
Virginia Beach is a good example of a suburban area that has become the largest municipality in a metropolitan area. Its greater size, however, does not make Virginia Beach the urban core. Otherwise, Contra Costa County in California could, by consolidating with its constituent municipalities (God forbid), replace San Francisco as the metropolitan area's urban core.
Perhaps the ultimate example of the problem of principal cities being confused with urban cores is Hemet, California, a principal city of the Riverside-San Bernardino metropolitan area that is, in fact an exurb and not in the primary urban area.
Toward the Future
An eventual more precise analysis of urban cores and suburban trends will be welcome. Yet, as our analysis of trends in New Jersey indicated, even the growth in more urban core oriented municipalities was minuscule compared to the state's suburban growth. Further, much of the urban core growth in the nation came from areas that, although formally located within “city limits” actually were on the suburban fringe. This was true, for example, in Kansas City, Oklahoma City and even Portland. This suggests that the small share of growth reported in urban cores would be even less if it were based on census tract data; and suburbanization, as a way of life, may indeed be even more prevalent than this year’s numbers suggest.
Wendell Cox is a Visiting Professor, Conservatoire National des Arts et Metiers, Paris and the author of “War on the Dream: How Anti-Sprawl Policy Threatens the Quality of Life”