Andrew Van Leuven and Dr. Ryan Jensen, Geography
One of the more widespread issues facing twenty-first century planners and policymakers is that of urban decline, especially in the midwest and northeastern United States.
Evidences of Urban Decline
Urban decline is evident in vacant and abandoned residential structures in the central downtown districts or a city or town (see photograph of abandoned commercial properties in Downtown Chester, PA, below left). It is illustrated by the displacement of economic activity from the central business district into the much less dense suburban periphery, such as the replacement of the downtown cafe or hardware store with big box retail and stripmall development. Most of all, urban decline is seen most easily in a clear and steady “shrinking” of a city’s population. Many of the nation’s most prominent metropolitan areas have lost over 50% of their peak population: St. Louis has declined by 62.7%, Detroit by 61.4%, Cleveland has lost 56.6.%, and Pittsburgh by 54.8%.
Contributors to Urban Decline
There are several contributors to this kind of decline. Most notable is the mid-twentieth century deindustrialization of the region, which was ignited when manufacturing and heavy industrial firms started to look to the southern United States (and overseas) for cheaper labor. This, coupled with the improvements in transportation that made certain geographic advantages (such as Pittsburgh’s location on three rivers or Cleveland’s vital connection to Lake Erie and her networks) obsolete, forced many mills and factories to close between 1960 and 1990.
The second most important contributor to the urban decline witnessed over the last several decades is suburbanization, or urban sprawl. As the entire country progressively shifted from walking and using trains and streetcars to the personal automobile, the built environment followed suit. Engineers and planners alike teamed up to mold old cities and build new ones that would accommodate the automobile’s every need, such as mobility (in place of accessibility) and parking. The kind of space needed to accommodate such a high number of cars could not be found in the city, leading the outskirts of America’s cities to become the new “center” for economic activity and the movement of people and ideas. However, due to suburbia’s inherent lack of an actual center, the form of urban vitality found in pre-automobile, industrial America was replaced by a more impersonal, low-density form of sprawl.
Anyone “left” in the urban core of a city during this time of flight to the suburbs was most likely left behind due to economic immobility or social standing. In other words, as the more prosperous middle and upper classes were able to leave the city and build up the suburbs, the poor, mostly black and hispanic working classes remained in the city, forced to cope with the removal and decline of their city’s services and benefits, such as transit.
The evidences and contributors of urban decline are essentially two sides of the same coin. Movements such as deindustrialization and suburbanization are the reasons for the patterns of vacant urban housing and lackluster downtown retail districts, while the dearth of urban vibrance and population is likewise the perpetuator of the movements themselves.
This study aims to identify a useful connection between urban aesthetic design tools and economic viability in towns whose once-prosperous industrial economic base has since disappeared or diminished. More specifically, this research compares the varying home values of 31 municipalities in Greater Philadelphia to examine the effect of parking and vacant lots on residential property values, with the assumption that cities with larger quantities of parking and vacant lots will exhibit lesser property values.
Several attempts have been made at studying both additional causes and new solutions for the ills of urban decline. A literature review on this matter is much too broad and vague to be relevant to this study in particular. This study is not a “comprehensive guidebook to cure the Rust Belt,” nor is it even a broad look at how to attract new business and beef up the economic base of the local economy. These two ideas are fantastic angles in which solutions to the pattern of economic decline may be considered, but their excessive breadth may be their undoing, as the processes at work are too complex to be dealt with in a wholisitc approach.
A major goal of this study was and is simplicity. The research question itself is very complex and open-ended—how aesthetic elements of a city’s urban character affect its present and future economic viability—yet the approach employed in this study is simple and concise. Much of the reviewed literature employed economic methods, such as hedonic pricing models, to pinpoint the financial impact of a certain urban element. These elements—ranging from greenways and parks to transit stops and bike lanes— definitely aided in narrowing the research question and focus, but did not contribute any sort of preliminary “answers.” Other paradigms, such as examining social and economic reasons for urban problems (i.e. white flight) and solutions (i.e. how to address the abundance of vacant and abandoned properties in Buffalo, NY) were similarly helpful and similarly incongruent to the specific approach that was sought after in this research.
Insofar as the literature review is concerned, the idea of comparing the quantity of land occupied by vacant and parking lots with corresponding home values appears to be a novel one. However, credit must be given to author James Howard Kunstler for both the genesis of this research idea and the phraseology of this paper’s title. In a 2008 podcast, Kunstler described the effects of empty lots on the urban character of Troy, NY, describing them as “missing teeth in the urban fabric.”1
This idea was taken and shaped into the current research theory: that parking and vacant lots are a form or aesthetic urban blight (see photograph of empty lot in Glassboro, NJ at right), capable of affecting the economic output of a municipality. Conversely, maintaining a dense, urban character—much like how the urban form appeared in the city’s peak—should yield positive economic results for a formerly-industrial city, despite the patterns of sprawl and suburbanization.
In a real-life urban planning context, this research aims to produce verifiable data to support the idea that eliminating massive parking lots and filling in vacant lots are economically viable enough to justify doing so.
Data & Methods
The Greater Philadelphia area was chosen for its high levels of economic variety among its boroughs, which have witnessed very diverse economic results since the shifts of deindustrialization and suburbanization of 1950 and 1960. Boroughs are the small, dense urban pockets that made up the periphery of the historic industrial region of the Philadelphia area. Their counterpart, “townships,” have more geometric boundaries and —prior to suburbanization—were only created to organize rural dwellers into a political community. This study will not focus on townships, as urbanism cannot be restored where it never existed.
However, boroughs such as Chester, Ambler, and Morrisville are very much in need of a study such as this. These and 28 other boroughs (a total of 31 data points) were selected from within the jurisdiction of the Delaware Valley Regional Planning Commission (DVRPC), which is the metropolitan planning organization of the Philadelphia-Camden area. The region encompasses nine counties from Pennsylvania and New Jersey and produces thorough GIS data every five years for several important categories.
Most important to this study were the DVRPC’s land use shapefiles, which included a comprehensive range of land use types as documented by the DVRPC in both 2005 and 2010. For use in this study, only the vacant and parking land uses were selected. Additionally, the “Parking: Row Home” land use was omitted from calculation, as the definition of “Residential: Row Home” in the metadata reads: “A series of connected single-family homes forming a continuous row, usually located in an urban area. Condominiums are identified as multi-family, not as row homes.” Parking associated with this land use did not seem to fit what this research aims to address.
Using ArcGIS the total square mileage of parking land use was calculated for each of the 31 municipality data points; this was repeated for vacant land use. These totals were each taken as a percentage of their municipality’s total land area. An additional measurement was recorded denoting the percentage of city land area dedicated to both land uses combined. This is the basis of this study’s primary independent variable. Additional variables, such as population density and euclidean distance from major metropolitan center (Philadelphia City Hall) were gathered for the purpose of using multiple regression to better explain the relationship between the two main variables.
The dependent variable (home values within borough limits) was drawn from Zillow, Inc., a company founded in 2005 for the facilitation of online home buying and selling. Their market research division had devised a measurement known as the Zillow Home Value Index (ZHVI). This measurement, due to its sound scientific methodology and quantitative standardization, is appropriate for measuring overall average home value at the city level. A ZHVI value was recorded for each of the 31 data points for both July 2010 and July 2005: the two dates corresponding to the DVRPC land use datasets, thus completing data collection.
I ran many iterations of statistical analysis for the data which I had collected. Most important and notable was the simple linear regression of parking/vacant lot quantities against ZHVI values. I attempted several combinations of this, only to find that not a single one was statistically signifiant. I ran a comparison for 2005 Parking & Vacant against 2005 ZHVI, 2010 Parking & Vacant against 2010 ZHVI, and again with only parking and only vacant for each year. Also, I tried to measure whether there was any statistical correlation between the change in parking/vacant lot quantities between 2005 and 2010 and the change of ZHVI between those years. Again, no luck.
Other statistics were collected, such as Moran’s I (which measures spatial autocorrelation, or the level at which near things are related) and geographically weighted regression. The idea behind GWR is to run a separate regression analysis for every data point, with a set bandwidth (either based on contiguity or nearness of neighbors). This again proved to be statistically insignificant.
As a last resort, a multiple regression analysis was run, using the additional independent variables (distance from major metro and population density). As with all previous tests, the resulting figures proved be statistically insignificant.
Results & Conclusion
The most important statistic in all of the test I ran was the adjusted R-squared, which measures the percentage of independent variable variation explained by the model (regression equation). In most, the R-squared value fell between 0 and 0.1%. In one or two attempts, the R-squared was above 1%, but still beneath 5%, meaning that hardly any of the variation in ZHVI could be explained by the model. This was extremely deflating and frustrating, but was a valuable lesson nonetheless.
In conclusion, the most important takeaway from this research is to keep trying. The cumulative knowledge from urban planning courses (coupled with a basic understanding of Donald Shoup’s “The High Cost of Free Parking”) still leads me to believe in my original hypothesis: that eliminating massive parking lots (see photo of parking lots in downtown Quakertown, PA, below right) and filling in vacant lots yield enough of an increase in economic viability to justify doing so.
I believe that where I went wrong was the use of Zillow’s home value index data as the independent, or response, variable. The relative ease at which it could be collected was quite alluring, and may have been responsible for the relative failure of this research attempt. Additionally, the faith at which I took the DVRPC’s land-use data may have been too high. In a situation where more time and resources were available for data collection, manual digitization of recent aerial imagery might serve more effectively than data taken from a government agency.
Despite any possible hiccups with the data, I still recognize that being wrong is not only an acceptable outcome, but is is an immense likelihood. After further examination, I realized data collection is not the only process that need refinement: my literature review is quite weak. Perhaps a more broad perspective on the issue, enhanced by increasing my reading of similar and relevant studies, might have helped me to refine my hypothesis to a more viable and likely prediction than the original.
As I embark on my MA in Urban Studies, this experience will most definitely help me in both selecting a topic for my thesis and improving my research methods. Not only would I like to revisit the idea of how “missing teeth” affect the economic viability of former industrial towns, but I will dig deeper into my main research question: how aesthetic elements of a city’s urban character affect its present and future economic viability, and how differences in aesthetic character among a region’s cities explain the differences in their varying economic statuses.
1 “Missing Teeth in the Urban Fabric.” Kunstler, James Howard. KunstlerCast #43. Duncan Crary. Podcast, 2008.