«A Journal of Policy Development and Research HoPe VI Volume 12, Number 1 • 2010 U.S. Department of Housing and Urban Development Office of Policy ...»
It seems then that total numbers of vacant lots and abandoned buildings are not being generated through the census counts. A closer consideration reveals otherwise: if a given census tract comprising 5 acres has 250 units of occupied housing units in 1990 and then has only 150 units of occupied housing units in 2000, a major physical change has occurred in this neighborhood (going from 50 to 30 occupied housing units per acre).
Four possible pitfalls with this approach are listed below, along with possible solutions for avoiding or addressing them.
1. Census tract boundaries change over time. One solution for avoiding this pitfall is the Geolytics Neighborhood Change Database, which features 1970–2000 census tract level data—all available at fixed 2000 tract boundaries—allowing for time series analysis.
2. The factors affecting the decrease in the number of occupied housing units may be unrelated to neighborhood decline; rather, they may reflect the construction of a new civic center or a highway. A solution for addressing this pitfall is to validate some of this quantitative data through direct observation of neighborhood conditions and interviews with long-term residents and community development and planning professionals.
3. Land use change conceptually is not interchangeable with housing density change; measuring one is problematic when planning for the other. The two terms are conceptually distinct, yet have much in common in terms of examining depopulation. For depopulating neighborhoods, a decrease in occupied-housing-unit density may indicate something other than vacant lots and abandoned buildings; it could mean a change in land use from multifamily homes into singlefamily homes, a conversion of homes into offices, or perhaps a consolidation of apartments within an apartment building. The problem might be that this single measure conflates changes in the number of structures, the number of units, and the land use. Fortunately, when used along with other census data (such as number of multifamily housing structures or number of business establishments), the occupied-housing-unit density variable can be dissected for meaning.
4. Household size and composition change over time, blurring the value of understanding housing-unit density. Some critics might suggest that looking at occupied-housing-unit density masks the changes at work within households. I defend this approach on the basis that it simply does not matter what changes happen within households from a physical planning perspective—what matters is how many structures remain when a neighborhood depopulates.
Conflating, in studying land use change, can actually be a good thing and aid in understanding broader changes occurring in a neighborhood. When considering an appropriate measure of physical change in depopulating neighborhoods, it is important to be aware of changing household compositions and the social dynamics at work. Those dynamics, however, are being captured by the occupied-housing-unit density measure, and this single measure reflects all the social, physical, environmental, and economic forces at work in a neighborhood that are generating a lower occupied-housing-unit density over time.4 The most important caveat here is that the measure reflects only residential housing conditions, excluding other major land uses such as commercial, industrial, or institutional. The results presented here can only be generalized to neighborhoods that are predominantly residential, where mixed-use or primarily commercial neighborhoods would be expected to function differently.
The value of a metric based on readily available national data is immense, but it is worth noting that some local governments already regularly collect their own land use, housing, and abandonment data. For such communities, the occupied-housing-unit metric could be useful as a check against their own data sources. For communities without the resources to collect local data, a metric based on free federal data sources is quite valuable.
How Neighborhoods Physically Change When They Lose Population Much is known within the urban geography and economics literature about how neighborhoods physically change when they lose population. When speaking of population decline, no single rationale explains why a place depopulates. Depopulation has been explained by everything from natural disasters (Vale and Campanella, 2005) to deindustrialization (Bluestone and Harrison, 1982), suburbanization (Clark, 1989; Jackson, 1985), globalization (Hall, 1997; Sassen, 1991), and, of course, the natural economic cycle of boom and bust (Rust, 1975). This article gives no attention to explaining why a place loses population, instead it focuses on the usefulness of one measure of loss—occupied-housing-unit density. This section of the article presents a cursory review of the extant literature that addresses how places physically change.
When employment declines in a territory, some people who lose their jobs might need to leave that territory and relocate to a place where new employment exists. The consequences for those who stay behind is that, just because some of their neighbors have departed (without being replaced by new neighbors), the physical form of the city does not naturally shrink. Glaeser and Gyourko (2005) studied the durability of housing in their time series sample of 321 U.S. cities and towns with at least 30,000 residents in 1970, showing how housing prices declined at a faster rate in depopulating cities than prices grew in growing cities. Their research suggests that the durability of housing poses a long-term threat to neighborhood stability. Others come to the same conclusion: if housing does not disappear as quickly as people do, then those abandoned structures may drag down neighborhoods by serving as a haven for criminal activity (Wallace, 1989). People losing their jobs and refusing to relocate for new employment can have huge implications for neighborhood conditions. Without income, a resident is less capable of caring for his or her home, which can lead to the deterioration of a neighborhood’s housing stock. When a bank forecloses on a resident’s home, the home, because of its unoccupied status, may bring further drag on the neighborhood’s quality.
Another problem resulting from population decline is that urban residents with means to relocate leave behind the poorest and most destitute residents. When fewer middle- and upper-income residents live in a neighborhood, fewer role models are available to youth, dimming prospects toward upward mobility (Sugrue, 1996; Wilson, 1987).
Over time, widespread racial discrimination, seen in hiring and in housing market trends, has systematically limited relocation options for African Americans (Massey and Denton, 1993; Sugrue, 1996). When a neighborhood loses jobs, African Americans have fewer housing choices, further increasing racial concentrations in ghettos.
As demand declines in depopulating residential neighborhoods, the housing demographic shifts from affluent residents paying higher rents to less affluent residents paying lower rents. Poor economic conditions decrease demand for housing through “filtering” economic classes of owners or renters (Hoyt, 1933; Temkin and Rohe, 1996). When demand ultimately sinks to certain threshold levels, owners tend to abandon their structures (Keenan and Spencer, 1999). Many abandoned structures become derelict over time and may become subject to arson. Thus, in a depopulating neighborhood, occupied housing units are replaced by unoccupied housing units, derelict structures, and, where fire consumed the unit(s), vacant lots. This process suggests it is appropriate to analyze physical change through the lens of occupied-housing-unit density.5 Occupied-housingunit density offers a clear picture of how a neighborhood’s physical form is changing and provides essential data to generate community development and planning strategies that respond directly to those changes. Flint, Michigan, is a true “poster child” for these very kinds of neighborhood changes, most notably through its depictions in Michael Moore’s infamous 1989 documentary, Roger and Me. As a place widely recognized as a victim of depopulation, Flint is an ideal location to test the usefulness of occupied-housing-unit density as a metric to be used in planning and policy practice. Field observations and interviews provide a check to the results of a quantitative calculation of occupied-housing-unit density in three of Flint’s depopulating neighborhoods.
Example of Using Occupied-Housing-Unit Density in Flint, Michigan First settled in 1818, Flint, Michigan, is located 60 miles northwest of Detroit along the Flint River.
The city was largely dependent on the timber industry until General Motors (GM) was founded there in 1908, turning the city into a world capital of the automobile industry in just three decades (Edsforth, 1982; Matthews, 1997; May, 1965). As GM and the American automobile industry shrank its workforce in the 1970s, so went Flint’s fortunes. Unemployment and reduced taxes translated to a reduction in city services—firefighters and police officers were laid off (Matthews, 1997). City officials responded with hundreds of millions of dollars in tax abatements and redevelopment financing in the 1980s and 1990s to encourage new industrial development and bolster the city’s central business district and to market the city as a tourist center (Matthews, 1997). At the same time, the United States government and the Michigan state government invested tens of millions of dollars in grants and loans while local philanthropists pushed vast sums of money into rebuilding downtown (Gilman, 1997). In his review of 14 redevelopment projects executed in Flint from 1970 to 1992, costing $568.5 million, Gilman (1997) found that 13 of these initiatives were explicitly intended to foster greater economic growth.
Although some benefits accrued to the city and its residents through these projects, the overwhelming evidence available shows that these efforts largely failed to reverse the city’s continuing At a regional level, the causal order is clear: fewer people will decrease demand for housing, resulting in fewer housing units. At the local level, Myers (1992) has argued that housing causes population because without the actual units there is no way for people to live in an area. This reverse causal order can be valuable in studying growing local areas, but offers little in helping to measure the physical effects of depopulation. Therefore, the more standard causal order of population causing housing will be used here.
economic decline (Gilman, 1997; Matthews, 1997). Flint’s total employment has gone down from 69,995 in 1970 to 40,213 in 2006 (42 percent) (U.S. Census Bureau, 2008). Although not all who cared to leave did, Flint’s population fell by almost one-third in the past half century, declining from 163,143 in 1950 to 112,524 in 2000 (U.S. Census Bureau, 2008). The city’s changing racial composition is harder to pin down because of the differing ways in which the Census Bureau characterized race and ethnicity between 1960 and 1980. From 1980 to 2006, when the definitions were being used consistently, the percentage of non-Hispanic African Americans in Flint increased from 41.1 to 56.3 percent.
Data and Methods for Flint, Michigan, Case Study For this study, I downloaded census data for Flint using the Geolytics software program for census tracts in Flint for 1970, 1980, 1990, and 2000.6 The key variables that I examined from the census were population loss and occupied-housing-unit density. I also examined socioeconomic variables, including income, poverty levels, race, and age.
I narrowed my analysis to three neighborhoods, each of which experienced severe drops in population and occupied-housing-unit density over the preceding three decades.7 Each of the three neighborhoods has a unique history, active community development organizations, and active residents groups. Carriage Town and Grand Traverse, which are in the city’s downtown area, are contiguous. Max Brandon Park is several miles outside the downtown core in a primarily residential section of town.
Working with a research assistant, I conducted background research on each neighborhood through electronic database searches.8 After completing the searches, we consulted with local experts to begin generating a list of potential interviewees.
From April through August of 2008, we conducted between two and four semistructured, in-person or telephone interviews with individuals in each of the following two categories for all three neighborhoods: (1) long-time residents and (2) professionals who work in the development, redevelopment, or planning fields in each neighborhood. In addition, we also conducted three interviews with individuals who are professionally involved in neighborhood development, redevelopment, or planning citywide but not necessarily in one of the study area neighborhoods.
Finally, I conducted onsite visits at each of the three neighborhoods in June 2008 and directly observed and recorded my observations about current land use and signs of historic land use.
I used the census tract as a unit of analysis because the Geolytics software has a special feature that normalizes the data to boundaries for the year 2000 across all four time periods, supporting time series analysis.
I also chose these neighborhoods for close study based on preliminary interviews with local community leaders who assured access to further interviewees.
We conducted searches in the following databases: Thomson Gale Expanded Academic ASAP and Academic OneFile, LexisNexis, ISI Web of Knowledge, ProQuest, Social Sciences Citation Index, Journal of Planning Literature, and CSA Illumina. We also conducted Google searches to identify relevant planning reports or news items for each neighborhood.
We limited articles to those printed from 1980 to the present. We present the results of the searches as background information for each neighborhood profile.