![Sample our Urban Studies journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110826/TOC-Subject-Sample-2021-Urban-Studies.jpg"})
Abstract
This article focuses on home ownership changes since 1990, particularly by minority or low-income populations—known colloquially as the American dream. This longtime centerpiece of U.S. policy has been primarily viewed in terms of national outcomes. Here, we address differences in a spatial context, focusing on the forty-nine Metropolitan Statistical Areas (MSA/CMSAs) with greater than 1 million population in 2000. The central question is how ownership change differed among urban areas. Following the annual State of the Nation's Housing report, we spotlight the percentage point change (PPC) of home ownership by all, white, black, Hispanic, Asian, and minority (as-a-group) households—as these vary among urban areas for 1990–2000, 2000–2007, and 1990–2007. Statistically, we find that PPCs for each racial and ethnic group tend to move in tandem; that there are considerable differences among MSAs in PPC performance; that these differences tend to cluster spatially in a manner that reflects regional dynamics but that, overall, the goal of reducing the minority–white gap in home ownership has not been realized. Regarding specific variables, metropolitan growth and, to a lesser extent, MSA size, best account for change in home ownership; subprime lending is not significant. In addition, consideration of unexplained variance leads us to conclude that, as a complement to the approach taken here, a more qualitative strategy would significantly increase understanding of this important issue—a strategy that focuses on institutional structures, supply-side actors, advocacy groups, financial organizational practices, community procedures, and the like—with key informant interviews as a central component.
Este artículo se centra en los cambios en el acceso a la casa propia desde 1990, particularmente por las minorías o poblaciones de bajos ingresos—coloquialmente conocido como el sueño americano. Esta eterna pieza central de la política de EE.UU. ha sido vista principalmente en función a los resultados nacionales. En este artículo tocamos las diferencias en un contexto espacial, centrándonos en las cuarenta y nueve Áreas Estadísticas Metropolitanas (MSA/CMSAs) con más de 1 millón de habitantes en el año 2000. La principal interrogante es cómo el cambio de propiedades difiere entre las zonas urbanas. A raíz del informe anual del Estado de la Vivienda en la Nación, destacamos la variación en puntos porcentuales (PPC) en el acceso a la casa propia para todos los hogares, de blancos, negros, hispanos, asiáticos, y de las minorías (como-grupo)—ya que estas varían entre las áreas urbanas en los periodos 1990–2000, 2000–2007 y 1990–2007. Estadísticamente, encontramos que las PPCs para cada grupo racial y étnico tienden a moverse en tándem, que hay considerables diferencias entre las MSAs en el funcionamiento de la PPC, que estas diferencias tienden a agruparse espacialmente de una manera que refleja la dinámica regional, pero que, en general, el objetivo de reducir la brecha minoría-blanco en el acceso a la casa propia no se ha concretado. En cuanto a las variables específicas, el crecimiento metropolitano y, en menor medida, el tamaño de MSA, representan mejor este cambio en el acceso a la casa propia, el préstamo hipotecario no es significativo. Además, al considerar las varianzas inexplicadas nos llevan a concluir que, como complemento al enfoque adoptado aquí, una estrategia más cualitativa aumentaría significativamente la comprensión de este importante tema—una estrategia que se centra en estructuras institucionales, agentes de oferta, grupos de apoyo, prácticas financieras organizacionales, procedimientos de la comunidad, y similares—con entrevistas a informantes clave como componente central.
Acknowledgments
Notes
*Earlier versions of this article were presented in 2009 at the meeting of the North American Regional Science Council and in 2010 at meetings of the Western Regional Science Association, Association of American Geographers, International Geographical Union, and European Regional Science Association. Comments from persons attending those meetings are appreciated, as are those of anonymous reviewers that significantly improved the final product. The article emanated from an idea by Jennifer Evans-Cowley of Ohio State's City and Regional Planning Program, while serving as a member of an MA thesis committee. Vital suggestions regarding relevant literature were provided by Mat Coleman and Kevin Cox of The Ohio State University, Geography, and an anonymous reviewer. Also of central importance were the efforts of Wenqin Chen and Ohio State's Center for Urban and Regional Analysis in assembling data and related materials.
1. Attention also is given to region (Northeast, Midwest, South, West) and age of householder (under thirty-five, thirty-five to forty-four, forty-five to fifty-four, fifty-five to sixty-four, and sixty-five and over).
2. Our concern with urban areas might readily extend to other scales. For example, to what degree are neighborhoods stabilized by higher rates of home ownership or alternatively, due to the demographics of new homeowners, made more dependent on community services?
3. In the 2000 U.S. Census, Consolidated Metropolitan Statistical Areas (CMSAs) consist of more than one Primary Metropolitan Statistical Area (PMSA), each of which is more or less equivalent to an MSA; see Frey et al. Citation(2004). Conurbations used here include eighteen CMSAs and thirty-one MSAs.
4. Examples of broad policy impacts with a distinct spatial dimension include Jackson's (1985) account of suburbanization in the United States and Kodras and Jones's (1990) Geographic Dimensions of United States Social Policy. The latter considers Aid to Families with Dependent Children, food stamps, health, homelessness, neighborhood rehabilitation, and public education. Also, whereas this article emphasizes differences in local conditions, an interesting counterpoint is J. Painter Citation(2006), who talks of the prosaic dimension of policy, calling attention to the “mundane practices through which something we label ‘the state’ becomes present in everyday life” (753); that is, the outcome of policies
depends on and proceeds through mundane practices undertaken by thousands of individual state officials and citizens … [providing] considerable scope for … qualitative and quantitative social and spatial variation … [implementation] necessarily proceeds unevenly … so [that] geographical variations in the provision of health care, policing, education and so on are not “aberrations” but integral to the operation of modern state institutions … The complex geographies of central-local relations contribute to the production of unintended state effects. (764) | |||||
5. MSAs mentioned as examples reflect actual computations of representative variables; that is, for life cycle, proportion of the population aged thirty to sixty; for user cost, 1990 to 2000 absolute gain in housing value; and for liquidity constraint, ratio of median housing value to median income in 2000.
6. K-means cluster is a method for identifying homogeneous groups of MSA/CMSAs (observations or sample points) based on their values for each of the six ownership change variables. This employs an iterative procedure that minimizes within-group variance and maximizes between-group variance. The authors most commonly use K-means clustering in connection with principal component scores (representing latent dimensions among variables) to group or classify observations (e.g., Brown, Mott, and Malecki Citation2007). This approach was bypassed here because the six PPC variables are meaningful in themselves, distinct, and comparable to one another when standardized.
7. Numbers reported here are actual PPCs, not the standardized.
8. Alternatively, this observation might simply be an aberration stemming from high levels of Hispanic in-migration that, as noted earlier, might artificially diminish (or distort) ownership levels.
9. Life cycle variables (Life Cycle 90,%35–60; Life Cycle 00,%35–60) are measured as the proportion of an MSA's population between ages thirty-five and sixty (i.e., Population 35–60/Total Population); change between 1990 and 2000 is in PPC terms (i.e., Life Cycle 00—Life Cycle 90). Liquidity constraint variables (Liquidity Constraint 90; Liquidity Constraint 00) are measured as the ratio of median value of owner-occupied housing to median family income; change between 1990 and 2000 is, again, in PPC terms. User cost is only calibrated in terms of its change from 1990 to 2000, calibrated as the difference between the median value of owner-occupied housing for each year (following Smith, Rosen, and Fallis Citation1988).
10. An important caution here is that our data are aggregated and not subdivided according to whether subprime use is for house purchase rather than maintenance, bill payment, lifestyle maintenance, and so on. Accordingly, our results do not contradict studies such as Chambers, Garriga, and Schlagenhauf Citation(2009), which find that home ownership increases are primarily due to mortgage innovations. Also relevant is Brooks and Ford Citation(2007) on “The United States of Subprime” (which also provided the subprime data used in our analyses).
11. The principal components analysis employs only a selection of the variables in . This was done on the basis of their significance in the zero-order correlation analyses, related discussion, or conceptual relevance.
12. The ACS replaces the long-form element of the decennial census with monthly data that, through aggregation, will ultimately provide information equivalent to decennial censuses. ACS data are currently available on a yearly basis for geographical units that are 65,000 or larger in population and three-year aggregations (beginning with 2005–2007, 2006–2008, etc.) for areas with more than 20,000 population. In 2010, the ACS plans to aggregate five years of data, which will provide information for all Census geographies, including block groups and Census tracts. For further details, see Mather, Rivers, and Jacobsen Citation(2005).
13. We will continue to use the term MSA/CMSA. Under ACS terminology, however, MSA could signify either a Metropolitan or Micropolitan Statistical Area (MetroSA, MicroSA). There also is a CSA (Combined Statistical Area) designation, a new unit comprised of “adjacent metropolitan and micropolitan statistical areas” (ProximityOne Citation2010). Regarding these distinctions and other ACS particulars, see Frey et al. Citation(2004), who, relevant to this article, noted, “For those interested in comparing metropolitan areas across the country, there is now really only one choice: the Metropolitan Statistical Area” (5).
14. New Orleans had PPC-All of 3.95, but this is likely distorted by the effects of Hurricane Katrina (August 2005).
15. The one instance is PPC-White-90–07 and PPC-Minority-00–07, which have a positive but nonsignificant correlation.
16. Three exceptions are New York, All-PPC, 1990–2000 and Las Vegas, All-/White-PPC, 2000–2007, low in high category; and Salt Lake City, All-PPC, 1990–2000, high in low category.
17. State of the Nation's Housing reports draw on a national sample, the U.S. Census National Housing Survey, which also would represent rural areas and urban places smaller than 1 million population. Further, tabulations of home ownership from decennial censuses and ACS, our data source, can be inaccurate as the result of in- or out-migration and immigration, which particularly pertain to minority groups. Given these drawbacks, our findings are taken with caution but are nevertheless relevant in charting directions for future inquiry and raising a flag of skepticism concerning progress elicited by American dream policies.
18. CitationFlippen (2010, 864) concluded somewhat differently that “the key contributor to minority home ownership is … a sizable coethnic base, and not how fast the base is expanding” but stated further that the role of growth is actually a function of “lower housing values, higher share[s] of new housing, and lower segregation in those areas.” She also found that “residential segregation has a clearly negative impact on minority homeownership … [that] lower homeownership propensities are not a general characteristic of highly segregated cities but rather that segregation affects minorities in particular” (863–64). Related to the latter, correlations between entropy indexes of intermixing for each of our forty-nine MSA/CMSAs, drawn from Brown and Sharma Citation(2010), and home ownership PPCs led us to conclude that a segregation index should not be included in further analyses.
19. The Berube et al. Citation(2010) groups are Border Growth, Diverse Giant, Industrial Core, Mid-Sized Magnet, New Heartland, Next Frontier, and Skilled Anchor.
