| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
a Department of Geography and Institute on Aging, University of Florida, Gainesville
Correspondence: Stephen M. Golant, PhD, University of Florida, Department of Geography and Institute on Aging, 3141 Turlington Hall, P.O. Box 117315, Gainesville, FL 32611-7315. E-mail: golant{at}geog.ufl.edu.
Decision Editor: Laurence G. Branch, PhD
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Poverty • Inequality • HUD facilities • Public housing
About one-third of the rental housing made affordable by various multifamily rental-assistance programs in the United States is now occupied by low-income older persons (Howard, Schless, and Edson 1999
). Even with the availability of these government-subsidized rental units, however, some 30% of elderly households pay more than they can afford for housing (U.S. Department of Housing and Urban Development 1999). Moreover, this national indicator often understates the extent of older persons' unmet shelter needs in locations where higher proportions of elders have housing-affordability problems (Golant and La Greca 1994). Such locational disparities are infrequently investigated by current housing need assessments, even as they would likely inform public policy initiatives designed to target poorly housed seniors (Pynoos and Golant 1996).
To improve our understanding of this locational influence, we investigated the extent to which government-subsidized affordable rental units available to older persons are unequally and unfairly distributed throughout Florida's counties. The findings of this study have public policy importance well beyond their immediate relevance to planners and policy makers in Florida who seek better information about their low-income seniors' unmet shelter needs. The available geographic assessments of unmet senior housing needs reported in the literature have two major weaknesses. First, they often analyze housing demand and housing supply in isolation of each other. Thus, there are analyses documenting the locations of low-income rural or innercity elderly people who would benefit from good-quality affordable housing (Golant and La Greca 1994); similarly, there are studies of where rent-subsidized affordable housing opportunities are located (Bach 1999). Analyses infrequently assess, however, whether the locational allocation of this affordable housing is appropriate given the locations of elderly residents with housing affordability problems. Second, most studies do not assess unmet locational needs using indicators that are sufficiently generalized to allow other states or municipalities to gauge the success (or lack of success) of their own local affordable housing initiatives. The literature is filled with simplistic mappings or locational summaries of low-income tenants and their affordable housing accommodations (Goering, Haghighi, Stebbins, and Siewert 1995; Schill and Scafidi 1999), but such descriptive renderings are almost always very place specific, and they offer findings that cannot easily be compared with those reported in other states or municipalities.
These analyses fail in a third important way. They rarely offer any absolute or relative standard by which a documented gap between affordable housing supply and senior-consumer need can be judged as unfair. Thus, there is no obvious basis by which representatives of state, county, or municipal governments can declare they have unmet location-specific needs for affordable housing of sufficient magnitude as to demand a public policy response. Rarely is the question raised as to how much geographic inequality is acceptable or unacceptable (Kane 1996; Sen 1992). Inevitably, any declaration that the current locations of affordable housing are inequitable represents an idiosyncratic assessment. This state of affairs should be contrasted with how we measure and evaluate the existence of racially segregated neighborhoods in this country, where we have both standardized locational inequality measures and inequality thresholds by which to declare communities and their neighborhoods unacceptably segregated.
This study seeks to address these past deficiencies in two linked ways. First, we judged the presence of unmet geographic need by older persons for affordable rental housing throughout Florida's counties by relying on several well-known location-inequality statistical measures that have mostly been used to study income, race, and ethnicity patterns (Massey and Denton 1989; White 1986). Second, we evaluated the locational unfairness in the availability of affordable housing for older persons by using three different locational standards: (a) the county location distributions of the community-wide low-income elderly population, (b) the county location distributions of the community-wide low-income nonelderly population, and (c) the extent of locational inequality of the government-subsidized rental housing available to low-income elderly persons compared with the extent of locational inequality of the government-subsidized rental housing available to low-income nonelderly persons.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Certain distinguishing features of these programs are relevant for understanding the locations of their low-income occupants. Two categories of tenants occupy facilities produced under the Public Housing Program. Rental units produced under the Public Housing Facility and Section 8 Moderate Rehabilitation Facility programs are project based. That is, they consist of buildings that were specifically constructed or upgraded for the purpose of accommodating a low-income tenant population. State-chartered Public Housing Authorities own and operate these facilities, and most of these organizations are administratively linked with either a county or a municipality. Section 8 Certificates and Vouchers, also administered by Public Housing Agencies, are tenant-based programs. Rental assistance is attached to persons (rather than facilities) who can seek out private-market housing accommodations as long as they meet minimum-quality and rental-cost standards, and the owners agree to participate in the program. Tenant-based programs are generally considered to offer tenants greater residential location flexibility. Although Multifamily Privately Owned Rental Housing, Rural Section 515 facilities, and Florida Housing Finance Corporation programs differ significantly as to their financing and operating features, they are all government-subsidized (both federal and state) project-based programs consisting of facilities mostly developed as a result of the initiatives of for-profit and nonprofit firms and organizations. The Rural Section 515 program is distinguished by its rural location developmental focus.
The number of rental units occupied by adults aged 62 and older (HUD's definition of elderly) and younger than 62 (nonelderly) households in each of Florida's 67 counties was computed for each of the above programs with the exception of the units administered by the Florida Housing Finance Corporation. Tenants' chronological age information was unavailable for these units, except for one small program subcategory, the State Apartment Incentive Loan Program (SAIL). The SAIL program units were included in the total number of units occupied by elderly people but were not analyzed separately. The Section 8 Moderate Rehabilitation Facility program is also small and its units were also not separately analyzed, although they were included in the totals for the Public Housing Agency programs and in the total number of units occupied by older adults.
County units were chosen for this analysis for several reasons: First, they offer a standardized and relatively unchanging administrative unit to summarize both the demand and the supply of affordable housing units; second, databases recording the availability of government-subsidized rental units almost always designate county locations, especially in the instance of Public Housing Authorities; and third, residential mobility data consistently show that when older people do move it is usually to an address within the same county.
The following data sets were used to collect the number of rent-subsidized units offered in these programs:
A computerized list of all Multifamily Privately Owned Rental HUD Housing in Florida was obtained from the Florida records of a national facility-based database (U.S. Department of Housing and Urban Development 1998).
To identify the number of low-income elderly (aged 65 and older) and nonelderly (younger than age 65) persons at risk of needing affordable rental housing by county, we obtained data from a special tabulation of the U.S. Census (U.S. Bureau of the Census 1994). Two age groups were defined: (a) elderly (aged 65 and older) low-income residents, not in institutions, who are under the 150% poverty level, and (b) nonelderly (younger than age 65) low-income residents not in institutions, who are under the 150% poverty level. U.S. Census 1989 poverty income thresholds were used to determine the poverty status of families and unrelated individuals. The number of persons below the poverty level was the sum of the number of persons in families with incomes below the poverty level and the number of unrelated individuals with incomes below the poverty level (U.S. Bureau of the Census 1994). The 150% poverty threshold was selected to capture most of the low income elderly and nonelderly tenants at risk of occupying the state's rent-subsidized programs (U.S. Department of Housing and Urban Development 1996).
Two data deficiencies should be acknowledged. First, county-specific population poverty data by chronological age was not available for 1997 to coincide with the collection date of the low-rent housing data. The potential for misinterpretation will exist if the 1997 and 1989 county distributions of the poverty-level older populations differ. Three analyses were conducted to assess this possibility. First, a correlation analysis of the 1990 and 1997 county distributions of the total population below the poverty level returned a coefficient of .86. Second, a correlation analysis of the 1990 and 1997 county distributions of the age 65 and older population (all income levels) returned a coefficient of .96. Third, a correlation analysis of the county distribution of the 1989 low-income older population with the county distribution of a projected year-2000 low-income older population, provided by a private vendor (using a somewhat different low-income measure), returned a coefficient of .96. All together, these analyses suggest a minimal possibility of the suspected misinterpretation. A second potential data problem results from the chronological age defining seniors in government-subsidized housing, aged 62 and older, differing from the age 65 and older age boundary used in U.S. Census tabulations. The potential for misinterpretation again will exist only if the county distribution of the population that is below poverty-level and aged 62 to 64 deviates significantly from the county distribution of the population below poverty-level and aged 65 and older (or, more correctly, aged 65–69). A simple correlation analysis of these county distributions returned a coefficient of .98, suggesting that this also was not a serious source of error.
The different units by which poverty and affordable-housing data are measured also had to be reconciled. Whereas the availability of government-subsidized rental units is measured by the number of occupied units or households, the poverty indicator is computed for the number of persons. To make the two variables comparable, we computed the number of persons in the rent-subsidized apartment units. This required assumptions about the average number of elderly and nonelderly persons per household for each of the government-subsidized housing programs. Household size information was drawn from HUD analyses (McGough 1997). Comparable data could not be found for the Rural Section 515 program; however, it was assumed that the elderly- and nonelderly-occupied units in this program had the same average household sizes as in the Multifamily Privately Owned Rental Housing program. It would be preferable for future studies to obtain average household size data specific to the rural housing program.
To avoid double accounting, I subtracted the number of elderly persons who occupied rent-subsidized units from the census-derived number of low-income elderly people in each of the 67 counties. Similarly, I subtracted the number of nonelderly persons who occupied rent-subsidized units from the census-derived number of low-income nonelderly. The two low-income populations living outside of government-subsidized rental housing are referred to respectively as the community-wide low-income elderly occupants and low-income nonelderly occupants.
County Location Measures of Affordable Housing Availability Relative to Populations in Need
County Inequality Quintiles.
These inequality measures are typically used to describe the extent to which the wealth of a population is concentrated in the hands of a relatively few, as in x% of the wealth is possessed by y% of the population (Schwartz and Winship 1979). Here these measures document the extent to which either affordable housing opportunities or persons at risk of needing affordable housing units are disproportionately concentrated in a relatively few counties. For any given measure of affordable housing or population in need, the 67 counties are ranked from high to low and then are divided into five groups (quintiles) on the basis of their ranked positions (four groups of 13 counties each and a fifth group of 15 counties). This allows an assessment, for example, of the percentage of low-income elderly persons that is found in the top 20% of the counties having the largest number of low-income elderly persons.
Index of Dissimilarity.
The Index of Dissimilarity—or segregation coefficient measure—has been predominantly used by sociologists and urban analysts when assessing racial and ethnic segregation patterns at the neighborhood (census tract or block) level (Massey and Denton 1989). Here it is designed to offer a global and generalizable measure of the extent to which two populations (e.g., elderly renters in low-income HUD housing and the low-income community-wide elderly population) are unequally distributed across a set of county geographic units (James and Taeuber 1985; White 1986).
The mathematical formula for this measure is computationally simple. Consider the r = 67 Florida counties. The ith county contains a population of Ei occupants in rent-subsidized elderly facilities and Ci persons in a comparison group (e.g., community-wide low-income elderly occupants). Summing over i,
 |
 |
It can assume values ranging from 0.0 to 1.0. Multiplied by 100, it can be interpreted as the percentage of one population that would have to change counties (from those counties where it is overrepresented) for both populations to display identical county location distributions (James and Taeuber 1985).
To test the validity of the Dissimilarity Index, I also computed Gini Coefficients (Coulter 1989; Crystal and Shea 1990; Osberg 1984). Although the statistical properties of the Index of Dissimilarity and Gini coefficient differ, several studies have demonstrated that these measures of inequality tend to be highly correlated (White 1986). No significant changes occurred in the patterns or interpretation of the results, although the Gini coefficients yielded consistently higher absolute measures of inequality. (Results are available from the author.)
Location Coefficient.
This measure assesses the degree of concentration of one population subgroup in each of a set of specific geographic areas relative to the degree of concentration exhibited by a comparison-population group. As an example, county X may be occupied by 5% of the state's community-wide low-income elderly population but may contain only 1% of the state's total supply of government-subsidized housing units. In contrast, county Y may be occupied by 5% of the state's community-wide low-income elderly population but may contain 9% of the state's total supply of government-subsidized units. In this example, county X has fewer affordable housing units than would be expected given its population in need, whereas county Y has more affordable units than expected. A location quotient of 1.0 would be returned if a county's percentage share of community-wide low-income elderly population equalled its percentage share of low-income tenants in HUD housing. The mathematical formula using the above symbols is Qi = (Ei/E)/(Ci/C), or, perhaps more easily interpreted, it is Qi = (Ei/Ci/(E/C). A unique Qi is computed for each of 67 counties. The latter formulation makes clear that the size of a county's computed location quotient depends on the similarity between the county and the state ratios of, for example, the number of rent-subsidized facilities to the number of community-wide low-income elderly persons. If a county has two rent-subsidized units for every ten low-income older persons (a ratio of 0.2), and in the state overall there is one rent-subsidized unit for every 10 low-income older persons (a ratio of 0.1), then the county's location quotient would equal 2.0. In this example, the county in question has two times the number of rent-subsidized units than would be expected given the availability of rent-subsidized units in the state overall. This particular interpretation makes clear a characteristic of most inequality/segregation measures. They measure how a subarea's concentration compares with the overall (state) concentration of a given population. Thus a location quotient of 1.0 would be found if county A had exactly the same ratio of its total low-income elder population in HUD housing as in Florida overall. The location quotient can return a value as low as 0 and as high as 1/(E/C).
The location-quotient values are categorized into five groups with equal intervals (with the exception of the last group having a variable upper limit): 0.0–0.49, 0.50–0.99, 1.00–1.49, 1.50–1.99, and 2.00 and over. By statistical definition, the first two intervals designate underserved counties. The analysis will refer to counties with this statistical range of location-quotient values as having less than their fair share of tenants in government-subsidized rental housing, whereas the other intervals designate overserved counties or those with at least or more than their fair share of tenants in government-subsidized rental housing. The analysis then computes the percentage of Florida's low-income community-wide population (either elderly or nonelderly residents) found in each group of counties defined by the five location-quotient intervals.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
The elderly tenants in rent-subsidized housing are also concentrated in fewer counties than are the nonelderly tenants in rent-subsidized housing. In every category of rent-assisted housing, nonelderly tenants display a less concentrated location pattern. This is particularly true for the nonelderly occupants in HUD Multifamily Rental Facilities and in Rural Section 515 facilities in which only 76% and 55% of the tenants, respectively, are concentrated in the top fifth of Florida's counties.
Dissimilarity Index.
The Index of Dissimilarity reveals that about 25% of the elderly tenants in the state's rent-subsidized housing programs would have to be reallocated to other counties if they were to be similarly distributed across the state as the community-wide elderly low-income population (Table 2 ). The highest dissimilarity index measure (.65) is found between the community-wide low-income elderly population and the elderly tenants in the Rural Section 515 rent-subsidized units. This reflects the predominantly urban locations of the state's community-wide low-income elderly population. The county location distributions of the elderly occupants in the Public Housing Agency programs and in the HUD Multifamily Rental Facilities also diverge substantially from the county locations of the community-wide low-income elderly population. About 34% and 32%, respectively, of the elderly tenants in these two programs would have to be relocated if they were to achieve residential evenness with the county distributions of the community-wide low-income elderly population. The elderly certificate/voucher holders (with more locational flexibility) are more likely to occupy county locations that are similar to those of the community-wide elderly low-income population (dissimilarity index of 0.29) than are the elderly occupants of the other two major project-based affordable-housing programs.
|
The county-location distributions of nonelderly tenants in affordable housing are much more similar to the community-wide low-income nonelderly population (Table 3 ). Only 14% of the nonelderly low-income tenants in all rent-subsidized facilities would have to move to achieve the same county location distribution as the community-wide nonelderly low-income population. The nonelderly tenants in the Public Housing Agency programs (both in Public Housing Facilities and certificate/voucher holders) are especially likely to be similarly distributed across counties as the community-wide nonelderly low-income population.
|
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
These findings emphasize that whatever the overall discrepancy between elder demand for affordable rental housing and its overall supply at the state (or national) level, the discrepancy between need and availability will be more pronounced in certain locations than others. They also confirm that judgments regarding the extent to which the elderly population at-risk for needing affordable housing is now being underserved depends on the standard used (in this study, the comparison group). The current county locations of affordable rental housing, for example, serve low-income nonelderly persons more fairly than they do low-income elderly persons. This was especially true for the locations of the affordable rental units available to low-income elderly persons that were produced under the Public Housing Programs; however, it was not true for the affordable housing units produced under the HUD Multifamily Rental Facility programs.
These findings raise two fundamental questions. First, why do these locational differences exist, and second, what are their effects on the social and economic well-being of the low-income elderly occupants who are living in these affordable rental facilities and on the community-wide elderly residents at risk of needing this affordable rental housing?
Although deserving of research, partial understanding of the basis for these locational differences comes from recognizing that a significant proportion of the elderly population in government-subsidized rental housing is occupying mixed-aged facilities. Thus, it is conceivable that the locations of these projects were driven more by the locations of the community-wide low-income nonelderly residents. This helps explain the greater residential evenness between the elderly tenants in affordable housing and the community-wide nonelderly low-income population. A second and related explanation is that when facilities now occupied by elderly tenants were first built, they often had larger shares of younger tenants as occupants, and it was only later that they came to be occupied by an increasingly larger population of elderly tenants A third explanation derives from recognizing that most affordable rental facilities now accommodating older persons were built in the 1960s and 1970s and, to a lesser extent, the 1980s (Golant 1999). These facilities may well have been constructed in counties that were then more dominated by a community-wide low-income elderly population. A fourth explanation draws on a frequently identified observation reported in the low-income housing literature. It is that the locations of affordable housing are often selected on the basis of where there is a perception (and reality) that the developer will confront low neighborhood (not in my backyard) resistance (Bach 1999). These are often nonelderly occupied poverty-dominated areas where neighborhood residents would be less likely to oppose the siting of yet another residential concentration of low-income residents (Thompson 1999). A final explanation is linked to the recognition that the development of affordable rental-housing facilities depends greatly on the initiatives of private developers, nonprofit organizations, and public housing authorities, all with distinctive locational preferences and biases (Schill 1997). In Florida, for example, public housing authorities are distributed widely throughout the state (117 public housing agencies in 67 counties), contributing to the greater locational availability of rent-subsidized units to accommodate nonelderly low-income populations (Golant 1999). Similarly, the more fairly located elderly occupied HUD multifamily rental units reflect the significant share of these facilities that are built specifically to accommodate low-income older persons and thus are more likely to be sited in counties with larger concentrations of this group.
It is unlikely that Florida is alone in this country regarding the location of its affordable housing supply that underserves a significant proportion of low-income elderly persons. Unfortunately, the current literature is not informative as how the older persons in these underserved counties cope with the lack of these affordable accommodations. They may be more likely, for example, to be living in poorer quality housing and in less supportive and attractive neighborhoods, to be experiencing higher and more stressful housing-expense burdens, or, alternatively, to be greater burdens on their families. These are clearly important questions for others to address.
There may be one positive outcome resulting from the current locational patterns whereby elderly occupied affordable rental housing is especially overconcentrated in a relatively small number of counties and, in turn, residential facilities. A significant number of policy analyses have pointed to the advantages of concentrating older persons in age-homogeneous settings. This literature emphasizes that the managements of these facilities are more likely to be knowledgeable about issues confronting an aging population and are better able to realize beneficial economies of scale in their service-delivery practices (Schulman 1996; Sheehan 1999; Struyk, Turner, and Ueno 1988; U.S. Select Committee on Aging 1987). This is especially important given that elder-occupied buildings, more than ever, are occupied by frail tenants who require supportive services to maintain their independent households (Golant 1999; U.S. Select Committee on Aging 1992). Thus, larger residential concentrations of low-income elderly people are more likely to be visible to human-service agencies that in turn can more effectively—organizationally and economically—target their residents' needs. Importantly, these residential concentrations of older people have seldom been linked to any significant negative consequences as they have with studies of nonelderly concentrations in affordable housing and their association with a host of "culture of poverty" social problems. Thus, even as the current county location pattern of affordable housing for seniors can be deemed locationally unfair, it also yields potentially important advantages for the current elderly occupants in these facilities.
At least one methodological construct decision in this study can be reasonably modified in future efforts. The county populations at risk of needing affordable housing can be profitably identified by more specific indicators. These would directly measure the presence of housing expense burdens and distinguish between renters and owners. It would also be desirable to express the income levels of individuals as a function of their counties' median income to capture statewide differences in the cost of living. Finally, it would be desirable to examine these patterns with up-to-date county poverty data available from the year 2000 U.S. Census.
Practice Concepts
The Forum
Book Reviews
|
Acknowledgments |
|---|
Received for publication May 17, 2001. Accepted for publication October 8, 2001.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. E. Northridge, G. N. Stover, J. E. Rosenthal, and D. Sherard Environmental Equity and Health: Understanding Complexity and Moving Forward Am J Public Health, February 1, 2003; 93(2): 209 - 214. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
