<%BANNER%>

Assisted Housing and the Geography of Opportunity

Permanent Link: http://ufdc.ufl.edu/UFE0042278/00001

Material Information

Title: Assisted Housing and the Geography of Opportunity A Comparative Suitability Analysis of Three Florida Counties
Physical Description: 1 online resource (171 p.)
Language: english
Creator: Seymour, Eric
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: Urban and Regional Planning -- Dissertations, Academic -- UF
Genre: Urban and Regional Planning thesis, M.A.U.R.P.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Privately-owned subsidized rental housing, which plays a critical role in providing affordable housing opportunities for lower-income families, seniors and disabled individuals, is increasingly at-risk of being permanently lost to the nation?s affordable housing stock. In response to this looming crisis, state and local entities have become engaged in initiatives to preserve these units and the affordability restrictions attached to them. Assisted housing receiving project-based rental assistance from the federal government has been prioritized for preservation, as it is generally able to house extremely low-income households. Although the affordability of assisted units is an important criterion for preservation, spatial considerations such as neighborhood quality and location efficiency, need to be incorporated into an effective prioritization scheme as well. In order to better inform preservation initiatives, this study assesses and compares the suitability of different generations of assisted housing production based on key spatial criteria. Three Florida Counties, Duval, Orange, and Pinellas, are taken as study areas for this research, as each contains a substantial share of both the state?s lower-income renter households and assisted housing units. The conceptual model for this research situates assisted housing preservation in the context of the ?geography of opportunity,? a concept that reflects key theories emphasizing the impact of neighborhood conditions on outcomes and opportunities for low-income households. This study employs a geographic information system (GIS)-based affordable housing suitability (AHS) model developed by the Department of Urban and Regional Planning and the Shimberg Center for Housing Studies at the University of Florida to determine the value of assisted properties in these counties based on key spatial criteria identified in the literature. These criteria are representative of key indicators of socioeconomic conditions and location efficiency. The statistical analysis of these suitability values is then used to determine whether there is a difference between the different generations of assisted housing for each of these criteria. The results of these tests show that younger, predominantly state-financed properties generally have significantly higher suitability values for variables measuring neighborhood quality, while older, predominantly federally-financed properties generally have significantly higher suitability values for variables measuring location efficiency. While newer properties may offer as many opportunities for the most needy, they may offer distinct advantages over the older stock in terms of individual outcome and opportunities for advancement. This study recommends that the decision to prioritize one generation of assisted housing over another should be informed by policy preferences and local housing conditions, such as poverty concentration and spatial isolation, and not on affordability alone.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Eric Seymour.
Thesis: Thesis (M.A.U.R.P.)--University of Florida, 2010.
Local: Adviser: Larsen, Kristin E.
Local: Co-adviser: Blanco, Andre.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042278:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042278/00001

Material Information

Title: Assisted Housing and the Geography of Opportunity A Comparative Suitability Analysis of Three Florida Counties
Physical Description: 1 online resource (171 p.)
Language: english
Creator: Seymour, Eric
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: Urban and Regional Planning -- Dissertations, Academic -- UF
Genre: Urban and Regional Planning thesis, M.A.U.R.P.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Privately-owned subsidized rental housing, which plays a critical role in providing affordable housing opportunities for lower-income families, seniors and disabled individuals, is increasingly at-risk of being permanently lost to the nation?s affordable housing stock. In response to this looming crisis, state and local entities have become engaged in initiatives to preserve these units and the affordability restrictions attached to them. Assisted housing receiving project-based rental assistance from the federal government has been prioritized for preservation, as it is generally able to house extremely low-income households. Although the affordability of assisted units is an important criterion for preservation, spatial considerations such as neighborhood quality and location efficiency, need to be incorporated into an effective prioritization scheme as well. In order to better inform preservation initiatives, this study assesses and compares the suitability of different generations of assisted housing production based on key spatial criteria. Three Florida Counties, Duval, Orange, and Pinellas, are taken as study areas for this research, as each contains a substantial share of both the state?s lower-income renter households and assisted housing units. The conceptual model for this research situates assisted housing preservation in the context of the ?geography of opportunity,? a concept that reflects key theories emphasizing the impact of neighborhood conditions on outcomes and opportunities for low-income households. This study employs a geographic information system (GIS)-based affordable housing suitability (AHS) model developed by the Department of Urban and Regional Planning and the Shimberg Center for Housing Studies at the University of Florida to determine the value of assisted properties in these counties based on key spatial criteria identified in the literature. These criteria are representative of key indicators of socioeconomic conditions and location efficiency. The statistical analysis of these suitability values is then used to determine whether there is a difference between the different generations of assisted housing for each of these criteria. The results of these tests show that younger, predominantly state-financed properties generally have significantly higher suitability values for variables measuring neighborhood quality, while older, predominantly federally-financed properties generally have significantly higher suitability values for variables measuring location efficiency. While newer properties may offer as many opportunities for the most needy, they may offer distinct advantages over the older stock in terms of individual outcome and opportunities for advancement. This study recommends that the decision to prioritize one generation of assisted housing over another should be informed by policy preferences and local housing conditions, such as poverty concentration and spatial isolation, and not on affordability alone.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Eric Seymour.
Thesis: Thesis (M.A.U.R.P.)--University of Florida, 2010.
Local: Adviser: Larsen, Kristin E.
Local: Co-adviser: Blanco, Andre.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042278:00001


This item has the following downloads:


Full Text





ASSISTED HOUSING AND THE GEOGRAPHY OF OPPORTUNITY: A
COMPARATIVE SUITABILITY ANALYSIS OF THREE FLORIDA COUNTIES


















By

ERIC COURTNEY SEYMOUR


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS IN URBAN AND REGIONAL PLANNING

UNIVERSITY OF FLORIDA

2010

































2010 Eric Courtney Seymour










ACKNOWLEDGMENTS

This research would not have been possible without the generous and substantial

contributions several people. First of all, I would like to thank my chair, Dr. Kristin

Larsen. Her insight, patience, and unflagging support were integral to the completion of

this work. I would also like to thank my co-chair, Professor Andres Blanco, for his

flexibility, practical advice, and methodological rigor. This research is indebted to the

work of the additional members of the Affordable Housing Suitability Model research

team in the Department of Urban and Regional Planning: Dr. Ilir Bejleri, Dr. Ruth

Steiner, Abdulnaser Arafat, and Eric Kramer. I am especially grateful to Eric for his

technical support and cartographic acumen. I would like to thank the Shimberg Center

for Housing Studies, especially Bill O'Dell and Anne Ray for their enthusiasm for the

project and assistance in shaping the direction of this study. In addition, I would like to

thank Nancy Muller at the Florida Housing Finance Corporation for taking the time to

meet with me. Of course, this list of acknowledgments would be incomplete without

thanking my family for their continued emotional and financial support.









TABLE OF CONTENTS

page

ACKNOW LEDGM ENTS ........ ......... .............................................. ........... 3

L IS T O F T A B L E S ........................................................................ ........... ........ ....... 7

LIS T O F F IG U R E S .................................................................. 9

LIST OF ABBREVIATIONS..................... .......... .............................. 10

ABSTRACT ...................................... ................... .............. ............... 12

CHAPTER

1 ASSISTED HOUSING PRESERVATION AND AFFORDABLE HOUSING
SUITABILTY MODELING .............. ..... ...................................... 14

2 ASSISTED HOUSING, THE GENERATIONAL SHIFT, AND PRESERVATION
STRATEGIES ............ ....... ................................. ............ 22

Affordable Housing Definitions and Trends.................................. .................... 23
Causes of the Affordability Problem ...................... .... .. ....... ................... 27
Evolution of Assisted Housing Production ........... ........... .................... 29
First Generation Programs ..................... ............. ...... ......... .. 30
Interest rate subsidy program s.............................................. .. .................. 3 1
Project-based rental assistance .............................................. ................. 34
Section 202 elderly housing program and Section 811 assisted housing.. 35
Second Generation Programs ......................_.. ......... .. ............... 36
The low-income housing tax credit program ............... ................ 37
Implications of the generational shift.................................... ............ 38
Preservation C challenges ...................... ....... ......... .. .. ............................ 40
O lder A assisted Stock ...................... ....... ......... .. .. ........................... 40
Newer Assisted Stock ...................... ............ .... ............... ......... 43
Low Income Housing Tax Credit Properties........................ ............... 44
Preservation Responses ............. .......................... ................ .............. 45
O lder A assisted Stock ............................. ........ .................... .............. 46
Section 8 Preservation Strategies ...................... ........... ....... ............. 47
Rent restructuring ....... .... ........ ...... ............................................... 47
Debt restructuring ....... .... ........ ...... ............................................... 48
State Preservation Initiatives ...... ......... ....... ............... ............... 49
P reservation Inventories ...................... .......................... .... ............................ 50
S u m m a ry ............. ......... .. .............. .. .................................................... 5 2

3 THE GEOGRAPHY OF OPPORTUNITY........................ ................... 54

Spatial Mismatch, Jobs-Housing Imbalance, and Location Efficiency .................... 54









Neighborhood Effects ..................................... ... ....................... 60
M mechanism s ............... ....... .. ............................ 60
Outcomes ..................... .................... 66
N e ighborhood C change ......................... ..... ................ ............... .. 73
Previous Studies on the Location of Assisted Housing .................... ......... ....... 75
Sum m ary ........... ......... ............................... 78

4 M E T H O D O LO G Y ......... ........... .......................... ....... ............... 80

Affordable Housing Suitability M odel .................. .......................... ............... 80
G geographic Inform ation System s ......... .......... ........... ......... .. .............. 81
Land-Use Suitability Analysis ........................... ........................ 82
Affordable Housing Suitability Model Structure and Methodology.................... 85
Methodology of This Study ................ ......... .. .............................. 89
Suitability Criteria Selected for This Study ...... ......... ..... .. ............... 89
Statistical Methods ................................ ....... ... ............... 92
Lim stations ....................................... .. ....................... ..........93
Sum m ary ........... ......... ............................... 94

5 DESCRIPTIONS OF STUDY AREAS: DUVAL, ORANGE, AND PINELLAS
C O U N T IE S .............. ..... ............ ................. ............................................ 9 8

Duval County .............. ......................................... .............. 98
Demographics, Income, and Housing Affordability................ .... .......... 99
Assisted Housing Inventory ................ ............. .. ............... 101
Loca l H housing P o licy .......................... ......... ............... ........... 10 1
Orange County ............... .................... ...... .. .... ..... ....... ... ......... 102
Demographics, Income, and Housing Affordability................... .... ........... 102
Assisted Housing Inventory ................ ............. .. ............... 104
Local H housing P policy .......................... ......... ............... ........... 104
Pinellas County ............... ................... ...... .. .... ..... ................... 105
Demographics, Income, and Housing Affordability................... .... ........... 106
Assisted Housing Inventory .............. ............................ ................. 107
Loca l H housing P o licy .......................................... ............... ........... 108

6 RESULTS FROM TESTS OF DIFFERENCES BETWEEN ASSISTED
HO USING G ENERATIO NS .............. ............ ...... ..... .......... ............... 118

Duval County ....................................................................................... ................... 118
Orange County ................... ... ......... ................... 121
Pinellas County .......................................................................... ......... ................... 124
S u m m a ry ................................................................................................. 1 2 5

7 CONCLUSIONS AND RECOMMENDATIONS .......... ............... 133

Policy Recommendations ............... ......................... 138
Future Research and Limitations .............. ............ ........ ................ 139









APPENDIX

A NEIGHBORHOOD CHANGE INDICATORS...................................... 141

Neighborhood Change as Gentrification............ ............ ..... ........ ........ 141
Neighborhood Change as Decline ...... ......... ....... .............. ............... 143

B TRANSPORTATION COST INDICATORS .................................................. 145

C MANN-WHITNEY U TEST RESULTS ................................................ 146

Duval County: Test between 1963-1979 and 1980-1994 .................................. 146
Duval County: Test between 1963-1995 and 1995-2008 .................................. 148
Orange County: Test between 1963-1979 and 1980-1994............................... 150
Orange County: Test between 1963-1995 and 1995-2008............................... 152
Pinellas County: Test between 1963-1979 and 1980-1994 ............................. 154
Pinellas County: Test between 1963-1995 and 1995-2008 ............................. 156

LIST OF REFERENCES .............. .. ....... ... ............. ........ ....... 158

B IO G RA P H ICA L S KETC H ............ .......... ...... .......... ......................... ............... 171









LIST OF TABLES


Table page

1-1 Expiration Dates by Funder and Program ........ ............. ........ .... .............. 21

5-1 Population projection (permanent residents) by age for 1990-2030, Duval
C o u n ty ...................................................................... ........... 1 1 1

5-2 Estimated employment by industry in Duval County, 2009............................. 111

5-3 Number of severely cost burdened renter households with income less than
80% AMI by tenure and income level, Duval County..................................... 111

5-4 Growth in severely cost burdened renter households with income less than
80% AMI by tenure and income level, Duval County..................................... 112

5-5 Population projection (permanent residents) by age for 1990-2030, Orange
C o u n ty ................. ..................................................................... 1 1 2

5-6 Estimated employment by industry in Orange County, 2009............................ 112

5-7 Number of severely cost burdened renter households with income less than
80% AMI by tenure and income level, Orange County ............. ... ............... 113

5-8 Growth in severely cost burdened renter households with income less than
80% AMI by tenure and income level, Orange County ............. ... ............... 113

5-9 Population projection (permanent residents) by age for 1990-2030, Pinellas
C o u nty ...................................................................... ........... 1 13

5-10 Employment by industry in Pinellas County, 2009....................................... 114

5-11 Number of severely cost burdened households with income less than 80%
AMI by tenure and income level, Pinellas County ........................ ............... 114

5-12 Growth in severely cost burdened households with income less than 80%
AMI by tenure and income level, Pinellas County ........................ ............... 114

5-13 Median age estimates and projections, 2006-2030..................................... 115

5-14 Median household income, 1989, 2008............................. 115

5-15 Renter households with cost burden above 30% and income below 50% AMI,
2 0 0 8 ..................................................... ................. ..... ... .... 1 16

5-16 Extremely low-income (<30 AMI), severely cost-burdened households, 2008. 116









5-17 Total properties, units, and assisted units in Assisted Housing Inventory by
county, 2010 .................... ..... ..... ......... ................ ........... 116

5-18 Properties and assisted units by founder (duplicated count), 2010 .................... 116

5-19 Assisted units by income limits, 2010..................... .... .............. 116

5-20 Assisted housing properties and units by age of property or year funded ........ 117

5-21 Median sales price for single-family homes (in thousands of dollars), 1995-
2 0 0 9 ............ .......... ................ ...... ......... ................................. 1 1 7

6-1 Comparison of assisted properties built/placed in service 1963-1979 and
1980-1994, Duval County ... .... ............. ................................... ......... 126

6-2 Comparison of assisted properties built/placed in service 1963-1994 and
1994-2008, Duval County................................. ............... 127

6-3 Comparison of assisted properties built/placed in service 1963-1979 and
1980-1994, Orange County ..................... ............... ...... ......... 128

6-4 Comparison of assisted properties built/placed in service 1963-1994 and
1994-2008, Orange County ......................... .... ............... 129

6-5 Comparison of assisted properties built/placed in service 1963-1979 and
1980-1994, Pinellas County ..... ............... .................. 131

6-6 Comparison of assisted properties built/placed in service 1963-1994 and
1994-2008, Pinellas County .......................... ... ............... 131

7-1 Incidence of significantly higher suitability values................ ... ........... 140









LIST OF FIGURES


Figci i9r
Figure page

4-1 Hierarchical structure of the A HS m odel.................................. ..................... 95

4-2 Goals and objectives of the AHS model. ....................................... ............... 95

4-3 Structure of the neighborhood change objective. ..................... ........ ......... 96

4-4 Neighborhood accessibility objective ................................... ............... 96

4-5 Local accessibility suitability layer, Orange County ........................ ............... 97

5-1 Study area counties ...................................... ........... 109

5-2 Location of properties in Florida's Assisted Housing Inventory (AHI)............. 110

5-3 Ratio of median existing single-family house prices to median household
incomes by metropolitan area, 1989-2009 ................................................ 115

6-1 Properties in the assisted housing inventory categorized by age for test
between HUD- and older FHFC-financed properties, Duval County................. 126

6-2 Properties in the assisted housing inventory categorized by age for test
between combined HUD- and older FHFC-financed properties and LIHTC
properties, Duval County ........... ........................._.......................... 127

6-3 Properties in the assisted housing inventory categorized by age for test
between HUD- and older FHFC-financed properties, Orange County.............. 128

6-4 Properties in the assisted housing inventory categorized by age for test
between combined HUD- and older FHFC-financed properties and LIHTC
properties, Orange County ..................... .. .................. ...... ........ 129

6-5 Properties in the assisted housing inventory categorized by age for test
between HUD- and older FHFC-financed properties, Pinellas County............. 130

6-6 Properties in the assisted housing inventory categorized by age for test
between combined HUD- and older FHFC-financed properties and LIHTC
properties, Pinellas County..................................... .. ............... 132









LIST OF ABBREVIATIONS

AHI Assisted Housing Inventory

AHP Analytic Hierarchy Process

AHS Affordable Housing Suitability Model

AMI Area Median Income

BMIR Below-Market Interest Rate

CNT Center for Neighborhood Technology

ELI Extremely Low Income

ELIHPA Emergency Low-Income Housing Preservation Act

FHFC Florida Housing Finance Corporation

FMR Fair Market Rent

GIS Geographic Information System(s)

HFA Housing Finance Agency

HMDA Home Mortgage Disclosure Act

IRP Interest Reduction Payment

HUD U.S. Department of Housing and Urban Development

LEM Location Efficient Mortgage

LHFA Local Housing Finance Agency

LI Low Income

MSA Metropolitan Statistical Area

NAP Neighborhood Action Plan Areas

NLIHC National Low-Income Housing Coalition

LHFA Local Housing Finance Agency

LIHTC Low-Income Housing Tax Credit

LIHPRA Low-Income Housing Preservation and Resident Homeownership
Act









PSS Planning Support System

RD U.S. Department of Agriculture, Rural Development

SES Socioeconomic Status

SMH Spatial Mismatch Hypothesis

USPS United States Postal Service

VLI Very Low Income

VMT Vehicle Miles Travelled









Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Arts in Urban and Regional Planning

ASSISTED HOUSING AND THE GEOGRAPHY OF OPPORTUNITY: A
COMPARATIVE SUITABILITY ANALYSIS OF THREE FLORIDA COUNTIES
By

Eric Courtney Seymour

August 2010

Chair: Kristin Larsen
Cochair: Andres Blanco
Major: Urban and Regional Planning

Privately-owned subsidized rental housing, which plays a critical role in providing

affordable housing opportunities for lower-income families, seniors and disabled

individuals, is increasingly at-risk of being permanently lost to the nation's affordable

housing stock. In response to this looming crisis, state and local entities have become

engaged in initiatives to preserve these units and the affordability restrictions attached

to them. Assisted housing receiving project-based rental assistance from the federal

government has been prioritized for preservation, as it is generally able to house

extremely low-income households. Although the affordability of assisted units is an

important criterion for preservation, spatial considerations such as neighborhood quality

and location efficiency, need to be incorporated into an effective prioritization scheme

as well. In order to better inform preservation initiatives, this study assesses and

compares the suitability of different generations of assisted housing production based

on key spatial criteria. Three Florida Counties, Duval, Orange, and Pinellas, are taken

as study areas for this research, as each contains a substantial share of both the state's

lower-income renter households and assisted housing units.









The conceptual model for this research situates assisted housing preservation in

the context of the "geography of opportunity," a concept that reflects key theories

emphasizing the impact of neighborhood conditions on outcomes and opportunities for

low-income households. This study employs a geographic information system (GIS)-

based affordable housing suitability (AHS) model developed by the Department of

Urban and Regional Planning and the Shimberg Center for Housing Studies at the

University of Florida to determine the value of assisted properties in these counties

based on key spatial criteria identified in the literature. These criteria are representative

of key indicators of socioeconomic conditions and location efficiency. The statistical

analysis of these suitability values is then used to determine whether there is a

difference between the different generations of assisted housing for each of these

criteria.

The results of these tests show that younger, predominantly state-financed

properties generally have significantly higher suitability values for variables measuring

neighborhood quality, while older, predominantly federally-financed properties generally

have significantly higher suitability values for variables measuring location efficiency.

While newer properties may offer as many opportunities for the most needy, they may

offer distinct advantages over the older stock in terms of individual outcome and

opportunities for advancement. This study recommends that the decision to prioritize

one generation of assisted housing over another should be informed by policy

preferences and local housing conditions, such as poverty concentration and spatial

isolation, and not on affordability alone.









CHAPTER 1
ASSISTED HOUSING PRESERVATION AND AFFORDABLE HOUSING SUITABILITY
MODELING

The nation's assisted housing stock plays a critical role in providing affordable

housing opportunities for lower-income families and individuals. In contrast to public

housing, assisted housing is privately owned and operated, and is held by both for-profit

and non-profit entities. In exchange for subsidies issued under various federal, state,

and local programs, owners of assisted housing agree to affordability restrictions

requiring the reservation of subsidized units for lower-income tenants. However, due to

the nature of these subsidy mechanisms, most assisted housing developments have

limited affordability periods. For example, as subsidized mortgages mature and rental

assistance contracts come to an end, owners of assisted housing have the option of

leaving subsidy programs and converting their properties to market-rate housing.

Nationwide, several hundred thousand properties have already opted-out of one the

older federal assistance programs, and many more remain at risk of leaving the

inventory at the expiration of their affordability periods (Achtenberg, 2002). Assisted

units are also at risk of being permanently lost through physical deterioration and

mortgage default. The combination of age and insufficient income has conspired to

make many older assisted properties particularly susceptible to this threat (see Table 1-

1) (Melendez, Schwartz, & de Montrichard, 2008; Schwartz, 2006).

The loss of older assisted housing developed under state and federal programs

has become a particular cause of concern for housing activists and agencies

(Affordable Housing Study Commission, 2006, National Housing Trust, 2008, Proscio,

2005). As federally-assisted housing developments are generally more deeply

subsidized than housing developed later under state and local programs, their









affordability restrictions are more likely to require them to set-aside units for extremely

low income (ELI) residents, who face the greatest difficulties in obtaining affordable

housing (Schwartz, 2006).1 Once assisted units serving ELI residents are lost, they are

difficult, if not impossible, to replace. Diminished federal assistance and heightened

construction costs have made the development of units affordable to lower-income

households especially difficult (Proscio, 2005). In response to the potential loss of such

a critical resource, preservation initiatives have emerged at the state and local level. In

particular, State Housing Finance Agencies (HFAs) have played a leading role in

preservation efforts. This largely stems from their responsibility for the allocation of the

Low-Income Housing Tax Credit (LIHTC), which has proven to be an effective

instrument in preserving assisted housing (National Housing Trust, 2008).

In order to efficiently target scarce state and local resources in preservation

initiatives, efforts have been made to collect and analyze data related to assisted

housing. Currently underway is the creation of a national preservation data

infrastructure, which is a standardized set of key variables useful for understanding

preservation needs (Shimberg Center, 2007). An important use of this data has been

the quantification of risk of assisted housing properties either opting-out or failing-out of

their subsidy programs (Ray et al., 2009; Roset-Zuppa, 2008). However, in addition to

the determination of risk, evaluations of assisted housing should be based on other

criteria indicative of their overall value to the affordable housing stock before being

prioritized for preservation, including neighborhood quality and transit accessibility.2 To


1 Extremely-low income is income at or below 30% of area median income (AMI).
2 While most urban analysts agree on the importance of neighborhood quality, there is no agreed upon
method for measuring it. The term is generally used to refer to the overall character of a neighborhood









this end, this study evaluates the relative suitability of different generations of properties

in Florida's assisted housing inventory for preservation initiatives based on such

criteria.3

An important distinction to be made between this study and those concerned with

performing risk assessments is that risk is here taken as the motivation and justification

for research, not as the object of study itself. In other words, a critical assumption of this

study is that the universe of assisted housing is situated in a general climate of risk. It is

not concerned with refining the picture of risk. Rather, this study is concerned with

identifying the relationship between the age and location of assisted housing and

considering the implication of this relationship for preservation initiatives. The

hypothesis of this study is that 1) older, federally-assisted properties are located in

areas with higher levels of accessibility than newer, state- and locally-assisted housing

units; and 2) newer properties are generally located in areas with higher levels of

neighborhood quality. In essence, it is testing whether the conventional wisdom

regarding the attributes of older versus newer housing in a metropolitan area-the

familiar urban/suburban dichotomy-holds true for assisted properties as well. However,

this is not an arbitrary exercise in testing an intuitive assumption. As Dreier, Mollenkopf,

and Swanstrom (2004) argue in their recent book, place matters for individual




based on a range of measures such as safety, environmental quality, level of public services, and school
quality (cf. Dubin & Sung, 1990; Greenburg, 1999; Newman & Schnare, 1997). Transit accessibility is a
function of the number of bus and rail stops/stations within a walkable distance of a particular location and
the frequency with which the buses and trains using these stations circulate.
3 The shift in housing production from federal, specifically, HUD administered and financed programs from
the 1960s through the mid-1980s to state and local responsibility for housing development is referred to in
the context of assisted housing as the "generational shift" (Ray, Nguyen, O'Dell, Roset-Zuppa, & White,
2009).









outcomes, and the spatial pattern of neighborhood quality and assisted housing need to

be jointly considered in making an informed preservation decision.

Spatial variables in addition to housing market and property-specific

characteristics are increasingly becoming recognized as having an influence on the cost

of housing, both narrowly and widely construed. For example, the relationship between

transportation and residential location can have a dramatic impact on household

expenditures. Housing units located far from employment, activity centers, and transit

stops increase the amount spent on transportation for households with cars, isolate

those dependent on public transportation, and limit the opportunities of both groups to

enjoy a high quality-of-life (Briggs, 2005; Lipman, 2006). Areas marked by high-poverty

levels, high crime rates, and poor school performance can have adverse outcomes for

residents of assisted housing, especially children, requiring additional outlays for

medical expenses and other services (Galster & Killen, 1995; Wilson, 1987). The cost in

terms of limited economic opportunities is especially important in the calculus of housing

affordability. Areas in which gentrification is occurring are likely to have few, if any,

opportunities for lower-income households to afford market-rate housing (Kennedy &

Leonard, 2001).

In order to evaluate the neighborhood quality of assisted housing along these

various dimensions, a land-use suitability model specifically designed for the purposes

of identifying areas suitable for affordable housing development and preservation will be

used in this study. This model, the Affordable Housing Suitability (AHS) model, is a joint

effort of the Shimberg Center for Housing Studies and the Department of Urban and

Regional Planning at the University of Florida. The AHS model provides a structured









system for the integration of relevant suitability criteria, community preferences, and

critical spatial data in a decision-making environment. Though weighted-overlay

analysis, the model produces a composite representation of suitability for affordable

housing. While the composite suitability values the model produces are invaluable and

will be included in this study, specific sub-components of the model indicative of

suitability along discrete dimensions will be selected for closer analysis.

As older properties are most at-risk for leaving the stock of assisted housing, this

study will compare the relative suitability of properties developed under U.S.

Department of Housing and Urban Development (HUD) assistance programs with older

properties developed by Florida Housing Finance Corporation (FHFC), which is the HFA

in Florida. In discussions of preservation, HUD-assisted properties have deservedly

received considerable attention due to the imminent termination of affordability

restrictions for a significant number of them. While older FHFC-assisted properties

generally have longer affordability restrictions and therefore are not at risk of converting

to market rate housing, their age places a number of them at risk of deterioration and

default. Thus, both of these generations of assisted housing are at risk, but for different

reasons. Consequently, it is important to ascertain whether HUD-financed properties

are more suitable than older FHFC-financed properties on a range of key indicators

before allocating scarce resources for preservation to one or the other population of

assisted housing. This study also compares the older properties of both types (HUD and

FHFC) with younger properties developed by FHFC and local housing finance agencies

(LHFAs). This comparison is intended to shed light on the relative suitability of older









assisted housing in general when compared to newer developments. The results of this

study will be useful in deciding which populations should be prioritized for preservation.

Duval, Orange, and Pinellas Counties serve as the case study areas for this

analysis. These Counties have been selected based primarily on their participation in

the development of the AHS model, but these areas are also ideal for this study as they

have large shares of Florida's assisted housing as well as large shares of the state's

lower-income households in need of affordable housing. These counties are located at

the core of some of the most populous metropolitan statistical areas (MSAs) in Florida:

Duval is the heart of the Jacksonville MSA; Orange is the core of the Orlando MSA, and

Pinellas is part of the Tampa-St. Petersburg-Clearwater MSA. Though these areas

have key similarities allowing for comparison, they are also in geographically diverse

regions, and are consequently marked by different development patterns and

socioeconomic conditions. This case study selection allows for a useful comparison of

assisted housing suitability taking the specific context of each area into consideration. In

addition, these counties privilege preservation over new development for meeting their

affordable housing needs. Therefore, the results of this study could be of particular

relevance to the preservation initiatives of these communities.

The next chapter supplies background on the specific programs responsible for

assisted housing production, the challenges assisted housing developments face in

terms of their continued presence in the state's stock of affordable housing, and

preservation responses. The concept of the "geography of opportunity" is presented in

Chapter 3. Chapter 4 describes the methodology used in this study, Chapter 5

describes the case study areas in greater detail, and Chapter 6 presents the findings of









the analysis. Finally, Chapter 7 evaluates these results and offers policy

recommendations and suggestions for future research.









Table 1-1. Expiration Dates by Funder and Program
Program Total Affordability Period By 2010 By 2020 By 2030 By 2030 By 2040 After
Units Expired 2040
HUD Programs 52,328 151 1,142 12,553 4,185 13,078 7,953 994
Section 202 24,510 0 205 3,997 3,518 10,310 5,703 777
Section 236 8,025 151 136 7,294 188 0 256 0
Section 811 745 0 0 0 0 0 528 217
Section 221 (d)(3) & (4) 7,471 0 801 1,262 479 2,768 1,466 0
Section 8 (project-based) 11,577 ,, I ,c nn.nr nr I I nrr r r, I .1r % n r ,r+ +,
\ J/ 1 1/ r/"K^Thi It l i ti i+ ll id dr i 1 / 5 t t t~ ~r\\t~t^\t\ <*l l- \t\^T \ ~~w^\rv ^ ^


ONLY I I U I l
Rental Assistance 16,845 properties
Rural Development 19,872 1
Section 514/516 3,934 1
Section 515 15,938
Section 521 11,171 This rental ass
Florida Housing Finance 155,769 13
Corporation
Source: Affordable Housing Study Commission, 2006


,945
,355
590
instance
,567


1 IUVV Ju IU ia ly pri Uviu via I -' yeai ,AI ILl i oL LU

4,131 3,738 5,540 3,814 704
865 445 221 816 232
3,266 3,293 5,319 2,998 472
is provided via 4-5 year contracts to properties
7,257 755 582 24,878 24,796


0
0
0

83,954









CHAPTER 2
ASSISTED HOUSING, THE GENERATIONAL SHIFT, AND PRESERVATION
STRATEGIES

This chapter places this study in the wider context of both affordable housing

needs and assisted housing development and preservation in the U.S. Of particular

relevance to this study is the character of assisted housing units produced under the

various subsidy programs including their level of assistance, the population they served,

and the specific preservation challenges they face. This chapter begins by providing an

overview of affordable housing trends both in Florida and in the nation as a whole, with

a particular emphasis on the unmet demand for affordable rental housing for low-

income households. This section answers the question of why preservation is a

worthwhile endeavor in the first place. Next, a review of the literature and history of

federal low-income rental housing assistance focuses on the HUD programs

responsible for the production of assisted rental housing. Then, federal devolution and

the increased role of state and local agencies in the production of affordable rental

housing will be considered. In this section, the "generational shift" from federal to state

assisted housing and its ramifications will be considered. The generational shift is a key

element of the present study's analysis. The assisted rental housing "preservation

crisis" will then be discussed, including the scale of the problem and the strategies that

have been devised to address it. Here, the implications of the generational shift on

preservation efforts will be examined. In conclusion, data collection efforts undertaken

to generate an accurate representation of assisted housing characteristics will be

discussed.









Affordable Housing Definitions and Trends

Inadequate conditions have historically been the greatest concern for housing

advocates and public officials. Tenement conditions in large industrial cities at the turn

of the twentieth century were the initial catalyst for significant housing interventions in

the U.S. (Hall, 2002). In recent decades, however, housing affordability has been the

greatest concern. While several measures of affordability exist, the most widely

accepted measure is the percentage of household income spent on housing, with 30%

being the accepted standard of affordability. Households spending more than 30% of

their pre-tax income are considered cost burdened, and those spending more than 50%

of their pre-tax income are considered severely cost burdened (Schwartz, 2006). For

renters, the cost of housing is the sum of rent and utility payments; for homeowners, the

cost of housing includes mortgage payments, property taxes, and insurance.

Affordable housing is out of reach for a significant number of U.S. households.

Illustrating the extent of the affordability problem, 30% of all homeowners and 45% of

all renters paid more than 30% of their income on housing in 2007 (Joint Center for

Housing Studies, 2009). Lower-income households are particularly likely to suffer from a

heavy housing cost burden. Fully 70% of renters and 67% of homeowners in the lowest

income quartile paid more than 50% of their income toward housing, while 50% of

renters and 42% of homeowners in the lowest income quartile paid as much. These

trends have worsened over time; between 2001 and 2007, the number of severely cost

burdened homeowners in the lowest income quartile increased nearly 18%, while the

number of severely cost burdened renters in the lowest income quartile increased 16%.

Although the greatest percent growth among severely cost burdened households

between 2001 and 2007 was among households in the upper-middle income quartile,









the overall percentage of higher-income households being severely cost burdened

remains relatively small compared to lower-income households, particularly renters. The

share of severely cost burdened renters was nearly twice that of homeowners in 2007

(Joint Center for Housing Studies, 2009). Further illustrating the affordability crisis for

low-income renter households, the most recent HUD (2007) worst-case housing needs

survey reports that 5.99 million very-low income renter households had worst case

needs in 2005, which is a 16% increase from 5.18 million in 2003.4 The report also

revealed that 72% of extremely low-income renters had worst case needs in 2005.

These national trends are equally manifest in the state of Florida. In 2008, nearly

53% of all renters paid more than 30% of their income toward housing, while nearly

38% of all homeowners did so (U.S. Bureau of the Census, 2008). Just as in the nation

as a whole, cost burden is disproportionally concentrated among low-income

households in Florida. In 2005, nearly two-thirds of low-income renter households in

Florida were cost burdened (Shimberg Center, 2007). In 2008, nearly 80% of renter

households earning less than $20,000 annually paid more than 30% of their income

toward housing (U.S. Bureau of the Census, 2006-2008). In absolute terms, 558,115

low-income renter households experienced cost burden in 2007 (Shimberg Center,

2007). Not only has the affordability problem in Florida been persistent, but it has also

worsened in recent years (Shimberg Center, 2007).

While households living in poverty are the largest group affected by the

inadequate provision of affordable housing, working households earning from one to



4 In 1990, HUD was directed by Congress to submit regular reports on households with worst case
housing needs, defined as unassisted and severely cost burdened renter households earning less than
50%of Area Median Income (AMI).









three times the minimum wage are increasingly subject to cost burden.5 In 2003, 37% of

households earning incomes in this range were severely cost burdened (Joint Center for

Housing Studies, 2003). Severe cost burden is largely a function of the confluence of

stagnant incomes and soaring housing costs. In order to highlight the difficulty of lower-

income working households securing affordable housing, the National Low Income

Housing Coalition (NLIHC) issues annual reports with the wage required to secure a

modest two-bedroom apartment at Fair Market Rent (FMR) while paying no more than

30% of income on rent and utilities.6 This measure is referred to as the "housing wage."

Fair Market Rent for a two-bedroom apartment in Florida is $1,055, meaning that a

household must earn more than $40,000 annually in order to pay no more than 30% of

income on rent and utilities combined; this translates to a housing wage of $20.29 per

hour. The mean wage for a renter, however, is only $13.23 per hour. Thus, the cost for

an apartment at FMR is 122% of mean renter household income. The rent for this

modest two-bedroom apartment is 236% of income for a person earning only minimum

wage (NLIHC, 2010). According to the NLIHC report, 59% of renters are unable to

afford a two-bedroom apartment at FMR, a significant increase from the 40% of renters

in 2000 (NLIHC, 2000; NLIHC, 2010). In 2003, the Joint Center for Housing Studies

reported that "households with one full-time minimum wage earner cannot afford to rent

even a one-bedroom apartment anywhere in the country" (p. 27). Even essential service

5 Poverty is a function of income and family size. In order to determine poverty, the U.S. Census Bureau
applies a set of "income thresholds that vary by family size and composition to determine who is in
poverty" (U.S. Census Bureau, 2009).
6 Fair Market Rents (FMRs) are gross rent estimates for different areas throughout the county, including
the cost of the rental unit and essential utilities, such as heat and water. FMRs are produced annually by
HUD in order to determine payment amounts for rental assistance programs. FMRs are not the same as
average rent in an area. They are set at a high enough level in order to allow for the selection of a range
of housing units in a variety of neighborhoods, but also low enough to accommodate a significant number
of program participants (HUD, 2007b).









personnel-such as teachers, police officers and nurses-are increasingly being priced

out of the communities they serve (National Association of Home Builders, 2004).

In order to secure affordable housing, many households move away from their

place of work as housing costs tend to fall with distance from employment centers.

While the resultant housing costs may be lower for these households, their

transportation costs are significantly higher than those living near their place of work.

Transportation costs are the second largest expense, behind housing, for U.S.

households, and in many cases affordability gains achieved by residing far from work

are negated by the increased costs of commuting (Lipman, 2006). Illustrating this

tradeoff, households spending 30% or less of their income on housing spend nearly

25% of their income on transportation costs, while severely cost burdened households

only about 8% of their income on transportation (Lipman, 2006). Thus, the Center for

Neighborhood Technology (CNT) (2010), a non-profit research institute, argues that the

conventional measure of affordability should be changed to reflect the influence of

transportation costs. The non-profit suggests a location may be considered affordable if

combined housing and transportation costs do not exceed 45% of area median income

(AMI).

Housing affordability problems affect diverse sections of society, but certain

demographic groups are particularly disadvantaged. Thus, while households of all ages

experience cost burden, the problem is most acutely felt by the youngest and oldest

groups of renters; 34% of renters under age 25 and 32% of renters over the age of 75

were severely cost burdened in 2005 (Joint Center for Housing Studies, 2008). In

Florida, nearly 60% of renter householders between 15 and 24 were cost burdened in









2008, while exactly 60% of renter householders over the age of 65 were cost burdened

in the same year (U.S. Bureau of the Census, 2008). Similarly, while households of all

races and ethnicities are subject to cost burden, minority renters are most adversely

affected. At the national level, more than 30% of black renters and 27% of Hispanic

renters were severely cost burdened, which in absolute terms is substantially more than

the 21% of white renters similarly burdened in 2006 (Joint Center for Housing Studies,

2008). Single-parent and female-headed households are also heavily affected by cost

burden (Joint Center for Housing Studies, 2003). In terms of geographic characteristics,

the majority of cost burdened households live in central cities, but a significant number

also live in the suburbs. Among households with worst case housing needs in 2005,

2.91 million resided in central cities, while 2.09 million resided in suburbs and 0.99

million resided in non-metropolitan areas (HUD, 2007a).

Causes of the Affordability Problem

The affordability crisis for low-income renter households is largely attributable to

the dearth of suitably priced housing units. Exacerbating this problem, rents have

increased much faster than the income of renter households, "pushing a growing portion

of the rental housing stock beyond the means of low-income renters" (Schwartz, 2006,

p. 34). These elevated rents also reflect the continued, simultaneous production of

newer, higher-quality, higher-cost units and destruction of older, lower-quality, lower-

cost units (Joint Center, 2008). This replacement of the affordable rental inventory by

more expensive units places low-income renters at a particular disadvantage. The

number of units affordable to renter households earning 30% or less of AMI declined by

19% in the 1990s, while the number of units affordable to renter households earning

between 50 and 80% of AMI declined by 5% during the same period (HUD, 2003). The









Joint Center for Housing Studies (2008) reports that "from 1995 to 2005, nearly 2.2

million of the 37 million initially available rental housing units (occupied and vacant)

were demolished or otherwise removed from the inventory" (p. 13). Among those lost,

the number of single-family and small multifamily rental units-those most frequently

inhabited by low-income renters-was especially high. In 2001, the number of renter

households in the bottom income quintile exceeded the stock of rental units affordable

to them by fully two million (Joint Center for Housing Studies, 2003).

The inadequate supply of rental housing affordable to low-income households is

primarily a function of the private housing market's inability to produce and maintain this

type of housing stock independent of public subsidies (Schwartz, 2006). Operating

costs, including repairs and other expenditures, as well as the rate of return demanded

by investors, skew the production of rental housing toward high-end submarkets in

order to ensure an adequate revenue stream and profit margin. When rent fails to cover

operation costs, as it frequently does in older rental properties, owners of existing

affordable stock may compensate by deferring maintenance and mortgage payments,

beginning a downward spiral that ends with the property eventually being removed from

the affordable housing stock (Joint Center for Housing Studies, 2003). Market forces in

recent years have further disincentivized the private production of modest rental

housing. The homeownership boom preceding the mortgage crisis steered multifamily

production toward condominiums and away from rental units; as a result, multifamily

rental construction fell off precipitously from 2002 to 2007 (Joint Center for Housing

Studies, 2008). While condominiums and housing units are being placed on the rental

market due to the foreclosure crisis, the rents they command are too high to be









affordable to low- and moderate-income households, leading to a mismatch in the rental

inventory (Joint Center for Housing Studies, 2009).

Rising development and land costs have also made it less attractive for

developers to produce modest rental properties, even with subsidies. These costs are

exacerbated by regulatory mechanisms, such as zoning, that limit the amount of land

available for multifamily rental construction and prescribe measures of size, quality and

density of housing units more amenable to high-end housing production. While these

regulations are intended to promote the worthy goals of environmental protection,

proper sanitation, and superior housing quality, the high cost of land per unit generated

by these regulatory requirements renders the production of affordable housing in some

communities altogether impractical (Downs, 1992; Schwartz, 2006).

Evolution of Assisted Housing Production

Since the first housing programs of the New Deal in 1933 and the landmark

Housing Act of 1949, which promoted the goal of "a decent home and a suitable

environment for every American family," the federal government has played an evolving

role in attempting to ameliorate the affordability crisis (Orlebeke, 2000). Orlebeke (2000)

divides the time since the passage of the 1949 act roughly into two segments: the first

of which, characterized by strong federal leadership in housing policy, ran from 1949

until 1973. The second period, characterized by increased federal devolution, the

emergence of state and local actors as key agents in the management of housing

programs, and a shift from supply-side to demand-side affordable housing solutions,

extends from 1973 to the present (Orlebeke, 2000).7 The 1973 moratorium imposed by


7 Of course, Orlebeke is writing before the most recent housing crisis, the occurrence of which may
necessitate a revision of this periodization scheme.









the Nixon administration on assisted housing production that Orlebeke (2000) locates

as the terminus of the "federal leadership model," however, did not altogether cease the

federal government's direct involvement in producing assisted housing, which

continued-albeit to an increasingly diminished extent-into the mid-1980s (p. 490).

Strictly in terms of the production of assisted housing, however, a more accurate

starting point for this devolution is the 1980s

This shift from the federal to state and local responsibility has been referred to as

the "national generational shift" in housing production, and it has important implications

for preservation initiatives (Ray, Nguyen, O'Dell, Roset-Zuppa, & White, 2009). In terms

of the larger trajectory of devolution, which Eisinger (1998) defines as the "reallocation

of specific responsibilities from Washington to subnational governments... primarily

involving] a shift from national to state government," the 1973 transition point is valid (p.

314). In that year, the Housing and Community Development Act created not only the

Section 8 program but inaugurated the use of block grants. Rather than continuing the

consolidation of housing policy authority, these programs place the responsibility for

administration and allocation on state and local governments. The production policies

and programs of these two periods, and the consequences of the generational shift will

be considered below.

First Generation Programs

The Housing Act of 1949 reauthorized the public housing program, which was the

federal government's first approach toward meeting the housing needs of low-income

households. This program is both publicly funded and managed, and the government, in

the form of local Public Housing Authorities (PHA), is the owner of these affordable

properties. In order to avoid competition with the private market, this program was









designed to house extremely low-income households, primarily those displaced as a

result of urban renewal (Listokin, 1991). The Housing Act of 1949 mandated a "20%

gap" between the highest rents charged for public housing and the lowest rents

commanded on the private market; as a result, public housing increasingly became the

"repository of the poor" (Listokin, 1991). Though the inventory of public housing

properties has been drastically reduced due to demolition, particularly as a result of

revitalization and mixed-income goals of the 1992 HOPE VI program, public housing

properties were designed to remain affordable in perpetuity. Beginning in the 1960s, the

federal government began to subsidize the production of privately owned affordable

rental housing properties in order to augment the public housing inventory and to offer a

more politically palatable alternative to the troubled public housing program (Listokin,

1991). The units produced under these federal programs through the 1980s constitute

the first generation of assisted housing. Unlike public housing, however, the continued

affordability of the properties produced under these programs was and remains

contingent upon the continuation of rental assistance contracts and other subsidy

mechanisms with a finite lifespan.

Interest rate subsidy programs

The Kennedy administration's engagement of the private sector reflects its

eagerness to develop less controversial assistance programs and to use public-private

partnerships in housing programs to stimulate the flagging economy of the early 1960s.

The Housing Act of 1961 created the Section 221(d)(3) Below Market Interest Rate

(BMIR) program, which was designed to supply housing for moderate-income families

whose earnings exceeded the limits for public housing but remained underserved by the

private rental market. Under this program, private and nonprofit developers could obtain









FHA-insured BMIR mortgages at 3% from private lenders, who in turn sold them to

Fannie Mae at market rate. These low-interest rate loans reduced the overall debt

service for the developers, allowing them to pass on the savings in the form of lower

rents to residents. When implemented. the Section 221(d)(3) program encountered a

host of difficulties and ultimately generated relatively little housing. The most critical

reason for the failure of the program was the perceived financial excesses of the

program, based on the large outlays in the annual federal budget for the purchase of

individual mortgages during the year they were acquired. The fiscal structure of the

program made it particularly vulnerable to criticism, though the actual size of the

subsidy was far less than these figures indicated as principal and interest would be

recouped over time (Hays, 1995).

As a consequence of these criticisms, the Section 221(d)(3) program was replaced

by the Section 236 interest rate subsidy program, which was established by the Housing

Act of 1968. As with the Section 221(d)(3) program, Section 236 was designed to allow

private developers to reduce their rents as a consequence of having a lower debt

service. However, rather than purchasing mortgages outright from private lenders, the

government provided monthly interest reduction payments (IRPs) to sponsors,

effectively reducing their loan payments from what they would pay at market rate

(Achtenberg, 2002).8 As Schwartz (2006) points out, "these annual subsidy payments

made Section 236 seem less costly from a budgetary standpoint than Section 221(d)(3),

although the level of public expenditure was actually greater under this subsequent

program" (p. 131). Due to the larger subsidy, properties developed under this program

8 Federal interest rate payments to owners of assisted properties reduced monthly mortgage payments
equal to what they would pay for a loan issued at one percent interest.









were able to charge lower rents and accommodate somewhat lower-income

households. In order to further attract investment, tax benefits were provided for

property owners in the form of rapid depreciation and mortgage interest deductions. As

a consequence, the Section 236 program was far more successful than Section

221(d)(3); more housing was produced under this new program within 3 years than had

been produced during the entirety of its predecessor. All told, the program produced

more than 544,000 units between 1968 and 1983 (Olsen, 2001).

A large number of the properties developed through Section 236, as well as

through Section 221(d)(3) received some form of rental assistance in order to make

units accessible to lower-income households. Income eligibility was set at 80% or less

of AMI; thus additional subsidies were needed to cover the rent for very-low-income

tenants. Older assisted properties developed with BMIR loans received this subsidy

through the Rent Supplement (RS) and Rental Assistance Payment (RAP) programs.

Though these programs were eliminated in the early 1970, many of the properties that

had received additional rental subsidies continued to do so under the Section 8 Loan

Mortgage Set-Aside program of 1974 (Achtenberg, 2002; Finkel, Hanson, Hilton, Lam,

& Vandawaller, 2006; Schwarz, 2006). As will be seen, layering of assistance for these

early projects lengthened their affordability periods.

These mortgage subsidy programs ultimately ran into serious difficulties.

Inadequate subsidies in the context of an inflationary economic climate led to operating

costs quickly surpassing revenue. As a result, numerous projects became delinquent on

their mortgage payments, and many eventually went into default. Furthermore, these

programs come under criticism for funding poorly-sited projects and projects undertaken









by inexperienced and unscrupulous sponsors (Listokin, 1991; Orlebeke, 2000). In

response to the poor performance of these programs, the Nixon administration imposed

a moratorium on the production of subsidized housing pending the development of more

effective mechanisms for assisting low-income households in acquiring affordable

housing.

Project-based rental assistance

In the 1970s, the federal government took a new approach toward the creation of

privately-owned, publicly-assisted affordable housing units serving low-income families.

The Section 8 New Construction and Substantial Rehabilitation program (NC/SR), the

supply-side component of the larger Section 8 program authorized by the Housing Act

of 1974, also provided a demand-side incentive. It augmented the rent property owners

charged, covering the difference between 25% (later 30%) of tenant income and FMR.

Income eligibility for the program was initially capped at 80% of AMI adjusted for family

size.9 Interest rate subsidies were no longer issued by the federal government, but in

some cases they were secured from state housing finance agencies. Developers were

able to allocate some or all of their projects for the program, making it particularly

flexible. As was the case with projects developed under Section 236, tax incentives

were extended to developers, permitting them to claim accelerated depreciation

allowances. "The combination of the deep subsidies and generous tax advantages

made the Section 8 program very attractive to developers and investors" (Schwartz,

2006, p. 133). As a result, the program ultimately subsidized more than 850,000 new or

rehabilitated housing units between 1974 and 1983 (Olsen, 2001).


9 In 1981 the criteria was adjusted so that only those earning under 50%of AMI became eligible, reflecting
a programmatic shift in emphasis from serving low- to very-low income households (Listokin, 1991).









Reflecting the shift from supply-side to demand-side housing assistance that

occurred during the Reagan administration, the Section 8 NC/SR program was

terminated under the Housing Act of 1983. This decision was largely informed by the

findings of the 1981 President's Commission on Housing, which concluded that as

affordable housing of both sufficient quantity and quality had been developed during the

1970s, it would be more efficient to direct resources toward tenant-based housing

assistance (Listokin, 1991; Orlebeke, 2000). With the repeal of the Section 8 NC/SR

production program, the Section 8 Existing Housing certificate program was left as the

largest remaining federal housing program (Orlebeke, 2000). This program gave

income-eligible households certificates allowing them to seek housing in the private

rental market. With the exception of programs designed to subsidize the production of

housing for the elderly and disabled, the federal government, guided by the Reagan

administration's "antiproduction, voucher-only housing policy," ceased its direct

involvement in rental housing production (Orlebeke, 2000, p. 509).

Section 202 elderly housing program and Section 811 assisted housing

Created by the Housing Act of 1959, the Section 202 Supportive Housing for the

Elderly program subsidizes the development of housing for low-income seniors by

nonprofit organizations (Schwartz, 2006). Initially, the program provided a 3% BMIR

loan for the costs of construction, rehabilitation, or acquisition, and the debt service for

these projects was covered by project-based Section 8 subsidies. Since 1992, capital

grants have been used in place of loans (Schwarz, 2006). Established under the

National Affordable Housing Act of 1990, Section 811 provided for the production of

assisted housing for the severely disabled using a subsidization mechanism similar to

that of Section 202. The Section 202 program has produced more than 260,000 units









and Section 811 more than 30,000 units (Schwartz, 2006). Unlike Sections 221(d)(3),

236, and 8, these programs continue to receive direct subsidies, but they have

produced far fewer total units than those of the terminated programs taken as a whole

(Ray et al., 2009).

Second Generation Programs

A defining characteristic of U.S. housing policy since the 1973 moratorium has

been the "formal transfer of most housing program control from the federal government

to state and local governments" (Orlebeke, 2000, p. 491). This transfer has largely been

accomplished through the replacement of categorical and centrally-administered federal

housing programs such as Section 8 NC/SR with block grants, which are allocated to

state and local governments for use in programs and initiatives they deem best suited to

their needs. In this context, HFAs have emerged as major players in the administration

and financing of assisted housing production (Basolo, 1999; Nenno, 1991; Scally,

2009). These state agencies first appeared in the 1960s, and had three important

functions: "a primary role as an administrator of other housing subsidies, a secondary

role as an administrator of other housing subsidies, and an emergent role as a

(re)developer of affordable housing" (Scally, 2009, p. 198). These state agencies were

granted the authority to issue tax-exempt bonds for financing the production of assisted

housing, and they served as administrators of federal housing programs, playing a

supporting role in the administration of Section 236 (Scally, 2009). With the devolution

of federal leadership, HFAs have become responsible for administering the HOME

Investment Partnership Program, authorized under the Cranston-Gonzalez National

Affordable Housing Act of 1990, the Low-Income Housing Tax Credit (LIHTC), created









under the Tax Reform Act of 1986, as well as a share of Community Development Block

Grants (CDBG) (Schwartz, 2006).10

The low-income housing tax credit program

The establishment of the Low-income Housing Tax Credit (LIHTC) program

reflects the larger context of devolution characteristic of federal policy since the advent

of the Reagan Administration and the "generational shift" in housing production more

specifically. The LIHTC allocates federal tax-credits to state agencies, usually HFAs,

which subsequently become responsible for managing their use in the production of

assisted rental housing. This program raises equity for the production of assisted

housing by offering federal income tax credits to investors, who are able to use them to

claim a dollar-for-dollar reduction in income-tax liability for a period of 10 years.

Developers apply to HFAs for tax credits, and then sell them to investors in order to

finance their projects. The amount of the credit depends on the cost of the project, its

location, and the proportion of total units reserved for low-income households.11 The

size of the tax credit increases when projects are located in difficult development areas

(DDAs), locations where the cost of housing is high relative to income, and in qualified

census tracts (QCTs), which are tracts in which at least half of all households earn 60%

or less of AMI. The maximum allowable rent for these projects ranges from 30 to 60% of


10 Devolution has placed a significant degree of responsibility for affordable housing financing on state
agencies and is characterized by a significant withdrawal of direct federal subsidies for housing. HUD's
budget of $34.3 billion for 2002 was only 41% of its $83.6 billion budget for 1976 (Dolbeare & Crowley,
2002). States have risen to this challenge through the establishment of housing trust funds, which in
many cases are also administered by state HFAs. This is the case with Florida's housing trust fund, the
State Housing Initiatives Partnership (SHIP) program, which was established under the William E.
Sadowski Act of 1992 with a dedicated revenue source (Larsen, 2009). SHIP had been the nation's
largest housing trust fund, until all its revenues were directed into the state's general fund during the
recessionary climate of 2009.
11 See Schwartz (2006) for a thorough explanation of the formula used to determine tax credits.









AMI, and is determined by the proportion of units reserved for very-low and low-income

households. Attesting to the significance of this program, Melendez, Schwartz, and

Montichard (2008) find that the LIHTC "is by far the most important source of subsidy

for low-income multifamily rental housing in the US, having produced nearly 2 million

units since its inception" (p. 67).

Implications of the generational shift

Devolution of responsibility for assisted housing production has affected far more

than the level of government responsible for these activities. As a consequence of the

programmatic changes marking the generational shift, a shift has occurred in the types

of households targeted and the actors involved in producing and owning assisted

housing. Though these trends may be equally manifested in other states, this section

will consider the implications of the generational shift in terms of the relationship

between the federal government and Florida only. Ray et al. (2009) have identified three

important effects of this shift. The first is the "long-term, growing emphasis on family

housing," in other words, the movement away from the production of assisted housing

intended for occupancy by target populations, such as the elderly and disabled, and

toward the production of units that serve the general population of low-income

households (Ray et al., 2009, p. 18). In the 1960s, HUD targeted assistance toward the

production of housing for the elderly and disabled through the Section 202 program, but

in the 1970s HUD began to subsidize the development of assisted housing for both

family and elderly households: "Only 13% of HUD-subsidized units in the 1970s

received funding from the Section 202 program" (Ray et al., 2009, p. 19). While HUD

retooled Sections 202 and 811 in the 1990s, renewing federal interest in the production

of assisted housing for populations with special needs, as a consequence of devolution









and state HFA's large-scale development of assisted family housing, the trend was not

reversed. Fully 87% of the HFA's inventory consists of family units (Ray et al., 2009).

The second effect of the generational shift is the transition from predominantly

non-profit ownership of assisted properties to for-profit ownership. In the arena of

assisted housing production, HUD has generally favored non-profit ownership and has

even required it, as in the cases of the Sections 202 and 811 programs. "In all, 63% of

HUD-assisted units are owned by non-profits... in contrast, 89% of Florida Housing-

assisted units are owned by for-profits" (Ray et al., 2009, p. 20). Consequently, older,

HUD-assisted housing is more likely to be under non-profit ownership than newer

FHFC-assisted properties. Furthermore, the number of for-profit developments assisted

by FHFC since the 1980s considerably outweighs the number of projects under non-

profit ownership developed from the 1970s onward.

The third effect of the generational shift identified by Ray et al. (2009) is the

movement away from the production of assisted properties capable of serving very-low-

income households. This effect is a consequence of the deep subsidies offered under

the earlier HUD programs being replaced with shallower subsidy mechanism of the

LIHTC and state-issued mortgage bonds. As Khadduri and Wilkins (2006) note

becauseue [LIHTC] has flat rents rather than percent-of-income rents, it does not easily

reach households with extremely low incomes or incomes below the poverty line" (p.

25). Similarly, in an examination of the use of the LIHTC and HOME resources, Mueller

and Schwartz (2008) find that "the poor are rarely the beneficiaries of state and locally

funded programs" (p. 131). Indeed, only four% of assisted units developed through use









of the LIHTC were affordable to households earning 30% or less of AMI (Mueller and

Schwartz, 2008).

Preservation Challenges

Unlike public housing, the affordability periods for assisted housing produced by

for-profit and nonprofit developers with direct federal subsidies are ultimately limited.

The amortization period of 40 years for the BMIR loans issued under Sections 221 (d)(3)

and 236 defines the window of affordability for these programs, and rental subsidies

provided under the project-based Section 8 program were issued on a contractual

basis, meaning that their continued affordability is contingent upon contract renewal.

Properties financed through the LIHTC face similarly finite affordability periods,

particularly the first cohort built. Physical deterioration, usually resulting from deferred

maintenance, also places assisted properties at risk of failing out of their subsidy

program. In addition, properties that default on their subsidized mortgage as a result of

their inability to cover their debt service are at risk of leaving the assisted inventory.

Thus, as Ray et al. (2009) find, older properties subsidized through the HUD programs

and the LIHTC program face "two countervailing pressures that may result in losses to

the subsidized housing inventory: 'opt-out' or 'time-out' risks, and 'fail-out' risks (p. 22).

Consistent with the literature, those properties of the earlier generation produced

through Sections 221(d)(3) and 236 will be referred to as the older federally-assisted

stock, and those produced under the Section 8 NC/SR program are referred to as

newer federally-assisted stock (Schwartz, 2006; Smith, 1999).

Older Assisted Stock

Although the subsidized mortgages given to developers of assisted housing under

Sections 221 (d)(3) and 236 were typically structured with a 40-year amortization period,









the length of which was intended to assure the long-term affordability of these projects,

owners were frequently extended the option of pre-paying their mortgage after 20 years

as an added incentive to participate in these programs.12 According to Achtenberg

(2002) "these incentives induced the construction of some 560,000 units of pre-payment

eligible housing" during the late 1960s and early 1970s (p. 2). Under certain conditions,

owners would have strong incentives to "opt-out" of the assisted housing inventory by

pre-paying their mortgage and converting their properties to market-rate. Schwartz

(2006) identifies two conditions under which this scenario is likely. These occur when

the property has appreciated significantly since its development, which is often the case

in desirable neighborhoods, resulting in the continued loss of the "cream of the older

assisted inventory" (Smith, 1999). As a consequence, "rents in the remaining older

assisted properties are 10 to 25% below HUD's estimated FMRs" (Smith, 1999, p. 151).

The second condition under which owners of the older assisted stock are likely to pre-

pay occurs when the tax-benefits of ownership become exhausted: "with depleted

depreciation and mortgage interest deductions no longer off-setting taxable income, the

typical pre-payment eligible project has become a tax liability for its owner" (Achtenberg,

2002, p. 2).

In addition to the more immediate threat to the assisted housing stock posed by

projects with pre-payment agreements, properties ineligible for pre-payment remain at

risk of "timing-out" with the maturation of their 40 year subsidized mortgages. When

mortgage maturity is reached, affordability restrictions are dissolved unless they are

12 It should be stressed that not every older assisted unit is eligible for pre-payment. These ineligible
properties include those that have accepted Section 8 rental assistance in addition to their subsidized
loans. Nearly 450,000 of the 650,000 units remaining in the older assisted stock receive assistance from
the Section 8 LMSA program (Schwartz, 2006).









attached to additional sources of funding for the property. A U.S. Government

Accountability Office (2004) report found that subsidized mortgages for 1,333 properties

(139,716 units) issued under Section 236 and for 502 properties (56,573 units) issued

under Section 221(d)(3) are scheduled to mature by 2013.

The expiration of rental assistance contracts attached to the older assisted stock

poses a further threat to their preservation. In order both to avoid defaulting on their

mortgage and to assist low-income tenants with inflationary budget-based rents during

the 1970s, many properties developed with interest rate subsidies entered into Section

8 Loan Management Set-Aside (LMSA) rental assistance contracts (Achtenberg,

2002).13 Indeed, nearly 80% of the older assisted stock receives Section 8 funding

(Schwartz, 2006). Schwartz (2006) observes that "unlike the newer assisted stock

[Section 8 NC/SR], rents in these properties tend to be relatively low. As a result, when

Section 8 subsidy contracts expire, owners have considerable incentive to leave the

program and convert to market-rate housing" (p. 135). These contracts were typically

short-term, consisting of an initial five-year period with the possibility of renewal for two

additional five-year periods (Achtenberg, 2002).

In addition to the contractual sources of concern raised above, many properties

developed under the interest rate subsidy programs are at risk of "failing-out" of the

assisted housing inventory as a result of physical deterioration. This situation is

primarily the result of the inability of some owners to cover both operating costs and to

maintain an adequate reserve fund for later capital investments, such as roofing and


13 This is in addition to the earlier stock of units developed under Sections 221 (d)(3) and 236 that had
accepted rental subsidies under the RP and RAP programs (which were subsequently renewed under
Section 8 LMSA) in order to house lower-income residents.









plumbing, while being required to keep rent at a predetermined level. "Below-market

rents were maintained though budget-based rent increases" that took only operating

costs into consideration-reflecting only a fraction of the total budget for an assisted

property (Smith, 1999, p. 146). This practice of "budget basing" had the net effect of

"slowly starving properties of capital," often desperately needed for capital reinvestment

(Smith, 1999, p. 146). As a consequence, the long-term viability of many of these

properties was seriously compromised. Poor physical conditions pose risk to properties

developed under Section 8 NC/SR as well. "These cases usually result from the owner's

failure to budget for capital improvements and invest rental income accordingly"

(Schwartz, 2006, p. 137). Those owners choosing to cover the costs of capital

investments in the face of limited income face the possibility of foreclosure, which is

another means whereby properties may fail-out of the assisted inventory (Pedone,

1991).

Newer Assisted Stock

Properties developed through the Section 8 NC/SR program were frequently

subject to contracts binding them to a long-term affordability period ranging from 20 to

40 years (Achtenberg, 2002). At the end of this term, owners of these properties had the

ability to "opt-out" of the program, allowing them to convert their properties to market-

rate. The costs of subsidizing project-based Section 8 properties, however, have proven

an obstacle to their preservation due to the rents in these properties rising

disproportionately relative to peer properties. "Although rents at individual Section 8

projects were increased every year by a HUD-calculated annual adjustment factor, rents

at surrounding apartment buildings did not necessarily grow at the same pace"

(Schwartz, 2006, p. 135). These properties placed an enormous strain on HUD's









budget, making them wary to renew affordability contracts. The agency's initial

response to the expiration of affordability periods was to substantially decrease the

length of the rental assistance contracts offered upon expiration of the original long-term

contracts, reducing them to five years and later to only one year (Schwartz, 2006).

However, as both long- and short-term contracts began to expire at the same time in the

mid-1990s, the cost of renewing all of these contracts placed HUD in a particularly

precarious situation, where the cost of renewal threatened to "consume the entire HUD

budget," while the cost of inaction would trigger "staggering claims against the HUD

mortgage fund" (Achtenberg, 2002, p. 4).

Low Income Housing Tax Credit Properties

At the time of the LIHTC program's creation, property owners were required to

preserve the affordability of units financed with the tax credit for a minimum of 15 years.

Melendez, Schwartz and de Montrichard (2008) identify three "year 15" challenges.

First, income and rent restrictions for LIHTC projects expire at this point unless they are

subject to further restrictions attached to additional funding sources. Owners have the

option of converting to market rate, which is especially likely when the potential to earn

higher rents presents itself. Second, year 15 signals the dissolution of the limited

partnership established to finance and develop these projects, "thus requiring the

sponsor to either acquire the limited partners' stake in the development or sell the

development altogether" (Melendez, Schwartz, & de Montrichard 2008, p. 68).14 Third,

many properties are also likely to require funding for major repairs and renovations. This



14 Sponsors are the for-profit and non-profit organizations with managerial responsibility for the LIHTC
properties. Limited partners are those that purchase interests in tax-credit properties and have no
managerial authority over them.









is especially a problem for the earliest LIHTC projects. The smaller amount of equity

generated for these projects from the sale of tax credits translated into modest

renovations compared to the "new construction and gut rehab that was to characterize

most tax credit developments from the mod-1990s onwards" (Schwartz, 2006, p. 97).

Thus, these more modest projects are more likely to be in need of repair at year 15 and

less likely to have the reserve funds required to cover these necessary capital

improvements. These conditions make older tax credits quite susceptible to failing out of

the assisted housing inventory.

An extended-affordability period of 15 years beyond the initial compliance period

was mandated for LIHTC projects through legislation in 1989 and 1990;15 however,

owners retain the ability to opt-out provided certain conditions are met. For instance, the

owner of a tax credit property seeking to sell it can opt-out if the state HFA

administering the program is unable to find a buyer able to both purchase the property

and retain affordability (Collignon, 1999). In 2002, the affordability restrictions for the

first 23,000 properties to be developed with the LIHTC expired, triggering concern

regarding the preservation of the rest of the LIHTC inventory (Collignon, 1999).

Preservation Responses

Different strategies have been employed at both the federal and state levels to

preserve the nation's stock of assisted housing. These strategies have taken different

forms based on the subsidy mechanism of the original program. While the federal





15 The affordability period for LIHTC projects in Florida is set at 50 years. In the case of projects with
longer affordability periods, the issue then becomes their physical and financial integrity. That is, while
they are less at risk of opt-out, they are still susceptible to failing out.









government initiated the preservation response in the 1980s, the responsibility for

preservation has increasingly devolved to the state and local levels.

Older Assisted Stock

The earliest preservation efforts emerged at the federal level as the older assisted

stock entered the period of eligibility for prepayment. "By the late 1980s, Congress

became concerned that a large number of owners might take advantage of the

prepayment clauses within a short period of time, thus dramatically reducing the supply

of low-income housing throughout the country" (Peiser, 1999, p. 372). The Emergency

Low-Income Housing Preservation Act (ELHIPA) of 1987 and the Low-Income Housing

Preservation and Resident Homeownership Act (LIHPRA) of 1990 were enacted by

Congress in order to prohibit prepayment by eligible owners. While these acts granted

owners financial compensation in the form of subsidies to provide them with the

equivalent of FMR, they generated a series of lawsuits and prepayment rights were

restored in 1996 (Peiser, 1999). In the wake of this failed effort to mandate the

continued affordability of the older assisted stock, "[t]he goal of federal policy shifted

dramatically from preserving the housing to protecting existing residents from

displacement" (Achtenberg, 2002, p. 2). These protections were accomplished through

the use of "enhanced vouchers," which cover the difference between 30% of the

tenant's income and FMR. If a tenant issued the enhanced voucher moves out, that unit

is no longer subject to affordability restrictions, meaning that the owner may charge as

much as the market may bear.

In order to retain Section 236 properties at risk of leaving the assisted housing

inventory through mortgage prepayment, the federal government allows owners and

purchasers to refinance their existing mortgage and continue to receive the same









interest reduction payments (IRPs) used to lower their debt service on their new

mortgage through a mechanism called Section 236 IRP decoupling (Achtenberg,

2002).16 The increased income from the subsidy may be used to rehabilitate the

structure and to establish a budget reserve for capital improvements. Owners who

refinance and receive this subsidy must continue to observe affordability restrictions for

a period of five years beyond the point at which the original mortgage would have

matured (Affordable Housing Study Commission, 2006; Schwartz, 2006).

Section 8 Preservation Strategies

In the late 1990s, Congress adopted two approaches for the preservation of at-risk

Section 8 properties. These programs differ based on the market in which properties are

located as well as whether or not they have additional rent restrictions, such as those

imposed by the older assisted stock that accepted Section 8 LMSA subsidies. In

essence, these two programs either increase rents on properties where they fall

appreciably below market or restructure the debt on properties where rents are

artificially high.

Rent restructuring

In 1999, Congress introduced the Mark up to Market program in order to prevent

properties with below-market rents from opting out of the assisted inventory, which is

especially likely for those located in strong rental markets. This program was particularly

influenced by the large number of Section 8 properties opting out as the rental market

was heating up at the same time that their affordability restrictions were expiring

(Achtenberg, 2002). Properties with below-market rents most commonly consist of the

16 "The decouplingg' program allows the IRP to be retained and continued after the Section 236 mortgage
is prepaid and refinanced" (Achtenberg, 2009, p. 1).









older projects receiving mortgage subsidies under Sections 221(d)(3) and 236 receiving

additional assistance through the Section 8 LMSA program (Schwartz, 2006). To

incentivize owners to renew their rental assistance contracts, the Mark up to Market

allows subsidized rents to be marked up to comparable market rents with a maximum

level of 150% of FMR. In exchange, owners must enter into a rental assistance contract

with a term of at least five years. An essentially identical program exists for non-profits

called Mark up to Budget, allowing them to increase rents to market level should a

demonstrated need exist.

Debt restructuring

Debt restructuring was the approach taken by the federal government to address

the preservation dilemma for Section 8 properties with above-market rents. This

problem resulted from owners being allowed to charge above-market rents in order to

cover development and operating costs for older Section 8 NC/SR properties receiving

an FHA-insured mortgage. As mentioned above, these properties continued to receive

an annual increase in rent based not on market conditions of peer properties, but on a

static adjustment factor. Renewing rental assistance contracts at these elevated levels

was untenable, but so was allowing them to default on their government-insured loans.

The response to this crisis was the Mark to Market (M2M) program, which "is a set of

financial incentives designed to encourage nonprofit and for profit owners to restructure

the debt on their properties, underwritten to a lower rent" (Affordable Housing Study

Commission, 2006, p. 12). In addition to a reduced mortgage capable of being serviced

by lower rents, the M2M program provides additional funds for rehabilitation and other

expenses. Properties restructured through M2M program are subject to a new rental

assistance contract preserving their affordability for 30 years.









State Preservation Initiatives

In the context of devolution, states have had to assume a great deal of

responsibility not only for the production of new assisted housing, but also for the

preservation of the federally assisted housing inventory (Nenno, 1991).

Since the termination of the LIHPRHA [Low-Income Housing Preservation
and Resident Homeownership Act] and ELIHPA [Emergency Low-Income
Housing Preservation Act] programs, responsibility for preserving HUD-
assisted housing has increasingly devolved to the state and local level.
Today's federal preservation tools... are not adequately designed or funded
to preserve all of the units that remain at risk (Achtenberg, 2002, p. 22).

As the projects in the federally assisted, first generation inventory have significantly

deeper subsidies attached to them than tax credit projects, states have increasingly

recognized the importance of preserving them for their lower-income residents

(Affordable Housing Study Commission, 2006). Furthermore, the combination of

diminished resources and soaring development costs make preservation a particularly

attractive endeavor as compared to new construction.

The primary organ through which states work to preserve the federally assisted

inventory is their HFAs, which are capable of allocating tax credits and using other

sources of financing, such as tax-exempt bonds and HOME grants, for preservation

activities (Affordable Housing Study Commission, 2006; National Housing Trust,

2004).17 In the past decade, many HFAs have expressed their concern for preservation

through the inclusion of preferences and set-asides for preservation-related activities in

their Qualified Allocation Plans (QAPs), which are federally mandated planning

documents required of all agencies responsible for the allocation of the LIHTC

17 While a host of regulatory mechanisms for preserving assisted housing have been developed at the
state and local level, the focus of this study is on the prioritization of properties for financial incentives by
HFAs. These regulatory mechanisms include such tools as right of first refusal and extended notice
requirements (cf. Achtenberg, 2002).









(Gustafson & Walker, 2002; National Housing Trust, 2004). According to a survey

conducted by the National Housing Trust (2008), 47 states prioritize preservation in their

QAPs, with 25 states having set-aside tax credits expressly for preservation.18 Through

their use of the LIHTC, HFAs have been able to preserve more than 280,000 units

(National Housing Trust, 2008). In 2008, Florida's HFA, the Florida Housing Finance

Corporation (FHFC) allocated funding for the creation of a pilot preservation bridge loan

program providing below-market short-term loans for preservation activities (Florida

Housing Finance Corporation, 2009).

Preservation Inventories

In order to provide an accurate representation of the quantity and quality of the

assisted housing stock and to guide the effective allocation of scarce resources in

preservation activities, state and local entities have begun to create "preservation

inventories," the purpose of which is to "collect available data on the existing affordable

rental housing stock, facilitating analysis of the portfolio and identification of at-risk

properties" (Center for Housing Policy, 2009, p. 1). When aggregated, these inventories

have the potential for providing a "national preservation data infrastructure," which

would allow comparative and collaborative efforts between localities and at the national

level (Shimberg Center, 2007). These inventories collect such information as:

* Project name
* Address
* Funding programs) (e.g., Section 236, Section 8 NC/SR)
* Target population (e.g., elderly, family)
* Total number of units
* Number of affordable units
* Years affordability restrictions begin and expire

18 Florida's 2009 QAP has a four million dollar set-aside of tax credits for qualifying preservation projects
(Florida Housing Finance Agency, 2009).









* Physical condition of the property (Roset-Zuppa, 2008).

The Shimberg Center for Affordable Housing (Shimberg Center) has worked to develop

a risk assessment method for analyzing the assisted housing inventory. To this end, it

has collected data on properties that have the potential to indicate whether a property is

at risk of opting- or failing-out, creating a distinct profile for each risk type (Ray et al.,

2009; Roset-Zuppa, 2008; Shimberg Center, 2008). Additional variables included in

these profiles include indicators of the financial condition of the project (e.g., loan to

value ratio, debt coverage ratio, and financial reserves), market conditions, and

ownership type (e.g., for-profit, non-profit) (Roset-Zuppa, 2008). These variables have

been used to create risk scores for projects in Florida's assisted housing inventory.

Quantifying the risk of a property leaving the assisted housing inventory is a critical

step in the on-going effort to preserve affordable housing. However, being at high risk of

leaving the assisted housing inventory is a necessary, but potentially insufficient

measure of worth for preservation efforts. According to the Shimberg Center (2008),

quantifyingig the risk to a set of properties is one step in setting preservation priorities,"

with the next step being to "prioritize properties by their value to the affordable housing

stock" (p. 20). The presence of units affordable to lower-income, especially ELI

households is an obvious measure of value for properties in the assisted housing

inventory, as these households face enormous difficulties in finding housing whose

costs do not impose a significant burden (HUD, 2003). Within the assisted housing

inventory, properties receiving project-based rental assistance under Section 8 and

related programs are far more likely to offer housing that is affordable to this

demographic. In recognition of this fact, the Affordable Housing Study Commission









(2006) suggests that the "highest priority" should be given to preserving these

properties (p. 4).

In addition to the affordability of their assisted units, the location of properties in

the inventory should also play a significant role in determining their value, and hence,

their relative prioritization for preservation initiatives. Indeed, as Briggs (2005a)

observes, "location, as... every realtor knows, helps define the real value of one's

housing" (p. 5). Hartman (1998), in making his famous case for a right to housing,

argues that "the social and physical characteristics of the neighborhood environment"

are just as necessary as the affordability and condition of the housing units themselves

(p. 237). While location is taken into account in assessing the assisted housing

inventory in so far as market conditions influence the likelihood of a property opting- or

failing-out, location, as a substantial body of literature shows, also structures the

opportunities and outcomes for low-income households living in these units. The

locations of employment opportunities, health care facilities, quality schools,

supermarkets, and other critical resources, far from being homogenously distributed

across metropolitan space, constitute a highly uneven "geography of opportunity" that

places low-income and minority households at a particular disadvantage (Briggs, 2005b;

Galster & Killen, 1995). Thus, properties in the assisted housing inventory occupying a

felicitous location within the geography of opportunity should be considered of high

value.

Summary

Assisted housing is a critical resource in meeting the nation's considerable

affordable housing needs, but the affordability periods for most of these properties is

limited. With the expiration of rental assistance contracts and the maturation of









subsidized mortgages, owners may decide to convert their properties to market-rate

housing. Assisted units may also be lost to the affordable housing stock through

physical deterioration and default. In order to prevent owners of assisted housing from

failing-out or opting-out, initiatives at all levels of government have been taken to

preserve the on-going affordability and functionality of assisted housing. In the context

of federal devolution, responsibility for preservation has primarily fallen to state and

local agencies. State agencies have expressed an interest in preserving older assisted

properties receiving federal subsidies, as they are better able to serve ELI residents.

This shift from deep to shallow assistance reflects a larger programmatic change in

assisted housing development, referred to as the generational shift. In order to inform

preservation initiatives, a national data infrastructure is being developed. At present, it is

able to provide an accurate picture of risk, but not of the overall value of assisted units

to the affordable housing stock. Data reflecting locational characteristics may assist in

developing a more robust picture of their value. Indicators of locational value for

assisted value may be found in literature of the "geography of opportunity," which

examines the impact of location on family and individual outcomes. This concept and

the empirical evidence that supports it are presented in the next chapter.









CHAPTER 3
THE GEOGRAPHY OF OPPORTUNITY

The concept of geography of opportunity emerged from research concerned with

the effects of continued urban sprawl and race- and class-based segregation on the

differential ability of families and individuals to achieve a high quality-of-life (Briggs,

2005b; Galster & Killen, 1995; Ihlanfeldt, 1999). Writers discussing the topic were

particularly concerned with debunking the existence of equal opportunity by placing the

concept in an explicitly spatial context (Dreier, Mollenkopf, & Swanstrom, 2004; Galster

& Killen, 1995). The results of this research have emphatically proven the hypothesis

that "where individuals live affects their opportunities and life outcomes" (Rosenbaum,

1995, p. 231). Indeed, "place shapes and constrains our opportunities not only to

acquire income, but also to become fully functioning members of the economy, society,

and polity" (Dreier et al., 2004, p. 28). Galster & Killen (1995) argued that geography not

only determines the quality and accessibility of key markets, institutions, and service

delivery systems, but also the strength of social networks and the normative values of a

community, both of which factor strongly into family and individual outcomes. Despite

the relative youth of the term, the concept behind the geography of opportunity is rooted

in earlier work by John Kain (1968) and Willam Julius Wilson (1987), who have

persuasively argued that place matters for individual outcome. This chapter presents a

number of key theories and empirical studies concerned with the linkages between

geography and opportunity.

Spatial Mismatch, Jobs-Housing Imbalance, and Location Efficiency

The spatial mismatch hypothesis (SMH) is one of the most extensively researched

dimensions of the problematic relationship between geography and opportunity in









metropolitan America. First advanced by John Kain in the 1960s, the SMH maintains

that post-war patterns of employment deconcentration and residential segregation have

conspired to constrain the employment opportunities of lower-skilled inner-city minorities

(Kain, 1968). As lower-skilled jobs, such those found in the manufacturing and retail

sector, began to move to the suburbs and beyond in the 1960s, minority workers have

found it difficult to follow them due to discriminatory practices in the housing market.

This has resulted in a "spatial mismatch" between the supply of low-skilled labor in the

central city and the demand for low-skilled labor at the metropolitan level, which is

exacerbated by the lack of adequate transportation options for workers seeking to

commute from the central city to the job-rich areas and by poor information about

distant job opportunities (Kain, 1968). Cumulatively, these conditions may result in

greater difficulties in finding and securing jobs, lower earnings, longer commute times,

and greater transportation costs for minority workers than similarly qualified white

workers (Ihlanfeldt & Sjoquist, 1998). Since the publication of Kain' seminal 1968 article,

"Housing Segregation, Negro Employment, and Metropolitan Decentralization," the

SMH has inspired a large research literature. Although interest in the subject waned in

the 1970s, Kasarda (1989) and Wilson (1987) renewed interest in the SMH during the

1980s and 1990s through their emphasis on job dencentralization and industrial

restructuring as key causative factors in the creation of an urban "underclass."19

Following Kain (1968), numerous studies have tried to test the relationship

between the spatial separation from employment opportunities and adverse labor-

market outcomes for minorities. Ihlanfeldt & Sjoquist (1998) found that most of these

19 Kasarda (1989) defines the underclass as "an immobilized subgroup of spatially isolated, persistently
poor ghetto dwellers" (p. 27).









studies have employed one of three methodological approaches: racial comparisons of

commuting times or distances, correlations of labor-market outcomes to job

accessibility, and comparisons of labor-market outcomes for inner-city and suburban

workers. In the early 1990s, six comprehensive reviews of SMH research were

published, all but one of which supported the hypothesis to varying degrees (Holzer,

1991; Ihlandfeldt, 19992; Jencks & Mayer, 1990; Kain, 1992; Moss & Tilly, 1991;

Wheeler, 1990). In a subsequent literature review, Ihlanfeldt and Sjoquist (1998) found

ample empirical evidence in support of the SMH. An important conclusion Ihlanfeldt and

Sjonquist (1998) reached from their review is that significant geographic variations in

mismatch exist, and in "areas with high levels of housing segregation and poor

transportation for reverse commuters, mismatch may play a more dominant role in

explaining the market problems of the inner-city poor" (Ihlanfeldt & Sjoquist, 1998, pp.

880-881).

Among recent studies, Raphael's (1998) test of the link between intra-metropolitan

accessibility to jobs and labor-market outcomes for black youth in San Francisco

provides strong evidence in support of the SMH. In contrast to previous research,

Raphael (1998) employed intra-metropolitan variations in job growth rather than

variations in employment levels as an accessibility measure. Raphael's (1998) results

show that "differential accessibility to areas of high employment growth is sufficient to

explain between 30 and 50% of the racial differential in neighborhood youth

employment rates" (p. 109). At the national level, Stoll (2005) found that metropolitan

areas with higher levels of employment deconcentration ("job sprawl"), measured as the

share of a metropolitan area's employment located outside a five-mile radius from the









central business district (CBD), exhibit a greater degree of spatial mismatch between

employment and black residents. In an analysis of the 150 largest metropolitan areas,

Holzer and Stoll (2007) found that while the share of blacks and Latinos in the suburbs

grew significantly during the 1990s, job growth occurred in newer, high-income

suburban regions. As a result, residents of lower-income suburbs must now commute to

other suburban regions for work, perpetuating the spatial mismatch. These results

challenge the idea of spatial mismatch as a strictly urban/suburban phenomenon, and

find support in Orfield's (1997) important work on the decline of older suburbs.

While Kain's (1968) original formulation of the SMH was specifically concerned

with employment outcomes for black men, subsequent research has extended the

hypothesis to countenance a wider range of the urban poor. Expanding the racial

dimension of the SMH, Ihlandfeldt and Sjoquist (1989) found that both white and black

low-income males were likely to suffer from reduced earnings as a result of job

decentralization. Kasarda and Ting (1996) found that spatial mismatch has a stronger

effect on women than men in terms of joblessness, regardless of race. They suggest

that this might be a consequence of the domestic duties that fall disproportionately upon

women, which make them less likely to take a job requiring longer commutes (Kasarda

& Ting, 1996).

Closely related to the concept of spatial mismatch is that of jobs-housing

imbalance, which is an important topic in transportation research. While the SMH is

specifically concerned with the adverse impact of employment decentralization on poor

central-city residents, the jobs-housing imbalance is concerned with the intra-

metropolitan mismatch between the location of jobs and affordable housing and its









consequences in terms of traffic congestion and air pollution (Cervero, 1989). Cervero

(1989) attributes this imbalance to several factors: fiscal and exclusionary zoning,

growth moratoria, mismatches between worker earnings and housing cost, the shift to

multiple wage-earner households, and job turnover. While the efficacy of a jobs-housing

balancing policy have been hotly debated in terms of trip-length and congestion

reduction (Levine, 1989), the jobs-housing balance literature has ably demonstrated the

need for affordable housing in job-rich areas (Cervero, 1989, 1996; Weitz, 2003). For

example, Cervero (1989) found that high housing costs and restrictive zoning in

suburban areas have contributed to lengthy commuting times in San Francisco and

Chicago. While workers may be able to secure affordable housing, the spatial mismatch

between the location of work and home may significantly increase the share of

household income spent on transportation. Thus, housing ostensibly affordable to

lower-income households may place a heavy cost-burden on residents when

commuting costs resulting from this mismatch are taken into consideration (Lipman,

2006).

As the term is generally used, location efficiency is a measure of the transportation

costs in a given area. In more precise terms, however, location efficiency refers to the

savings on transportation expenses achieved by households living in neighborhoods

with high geographic accessibility, which in turn refers to the "ease of reaching needed

or desired activities" (Handy & Clifton, 2001, p. 68). The concept of location efficiency

emerged from research seeking to quantify the relationship between new urbanist

design principles, specifically, compact, transit-oriented development patterns, and

automobile use. In their foundational study on the relationship between urban design,









transit accessibility, and auto use in Chicago, Los Angeles, and San Francisco,

Holtzclaw, Clear, Dittmar, Goldstein, and Haas (2002) found substantial evidence in

support of their hypothesis that auto ownership and driving decrease as destinations

become more accessible. Their results revealed that density (measured as households

per residential acre) and transit accessibility (measured as daily average number of

buses or trains available at transit stops within a walkable distance) were the strongest

determinants of both auto ownership and vehicle miles travelled (VMT). Pedestrian and

bicycle friendliness (the measure for which was derived from the density of the street

grid, age of housing, and traffic calming bonuses) was also found to be associated with

decreased VMT, though the relationship was weaker. Based on this research, it has

been concluded that homes situated in close proximity to daily destinations and transit

stops are highly location efficient. Given that transportation expenses are second only to

those for housing, location efficiency offers considerable benefits to low-income

households (Lipman, 2006). In recognition of the cost saving benefits of highly efficient

neighborhoods, entities in four US cities began to issue location-efficient mortgages

(LEMs) in the late 1990. LEMs are premised on the proposition that mortgage

underwriting guidelines may be relaxed for homeowners in location-efficient

neighborhoods because they have lower than average automobile-related

transportation expenses and more income available for mortgage payments (Blackman

& Krupnick, 2001).20


20 Some research has found that households financing their homes with LEMS may be equally, if not
more susceptible to default (Blackman & Krupnick, 2001). Though the cost-savings may not be sufficient
to allow lower-income households to afford homeownership (especially in tight housing markets), the
proven reduction in transportation-related expenses for households residing in location-efficient
neighborhoods nevertheless offers a significant opportunity for improving the economic well-being of
lower-income residents.









Neighborhood Effects

The study of neighborhood effects is principally concerned with understanding the

impact of neighborhood environments on outcomes for families and individuals in such

areas as educational attainment, labor-force participation, and criminal activity. In their

exhaustive review of the neighborhood effects literature, Ellen and Turner (1997)

identify six such mechanisms through which neighborhood effects operate: quality of

local services, socialization by adults, peer influences, social networks, exposure to

crime and violence, and physical distance and isolation. Additional mechanisms

identified in the literature of neighborhood effects are continued exposure to stressful

conditions (Dreier et al., 2004; Ellen, Mijanovich, & Dillman, 2001; Geronimus, 2006)

and exposure to environmental hazards (Bullard, 2000; Pastor, 2001). These

mechanisms are held to vary in their level of operation depending on gender, age, and

other personal characteristics (Ellen & Turner, 1997). The following review of the

neighborhood effects literature is divided into separate sections for mechanisms and

outcomes.

Mechanisms

The accessibility and quality of local services delivered at the neighborhood level

can have a significant impact on individual outcome (Ellen & Turner, 1997). The specific

public school an individual attends, particularly at the elementary school level, is usually

determined by neighborhood of residence. If the school is of poor quality, individuals

forced to attend them due to residential location are likely to suffer from diminished

educational attainment relative to individuals attending higher quality public schools.

Other services whose accessibility and quality vary with regard to residential location

include preschools and daycare centers. In lower-income neighborhoods, these









resources are likely to have reduced human and material resources, such as qualified

staff and learning materials. Consequently, "children may receive less attention and

supervision, be less challenged and stimulated, and ultimately be disadvantaged when

they attend school" (p. 837). For adolescents, access to afterschool programs may play

a role reducing the likelihood of involvement in dangerous and criminal activities. Of

course, access to quality medical care can have significant outcomes for individuals and

families regardless of age or personal characteristics. For example, increased

incidences in asthma morbidity among low-income children have been partially

attributed to deficiencies in access to quality medical care (Crain, Kercsmar, Weiss,

Mitchell, & Lynn, 1998).

Socialization by adults builds on Jencks and Mayer's (1990) concept of indigenous

adult influences, by which is meant the manner in which adults residing in a given

neighborhood may influence the outcomes of children and adolescents by serving as

role models. Adults communicate critical values to neighborhood youth, including the

relative importance of work and education, as well as information about community

norms and expectations. Neighborhoods with low levels of labor-force attachment

among adults are likely to produce a social context in which the operations of the labor

market and educational institutions are viewed with skepticism (Wilson, 1991). This

mechanism has received considerable attention as a result of Wilson's (1987)

contention that the absence of suitable role models has precipitated persistent social

pathologies among the underclass. Operating along similar lines, peer influences refer

to the effects that peers have on one another's behavior. Just as in the case of

socialization by adults, this mechanism too may result in either positive or negative









outcomes. Research that emphasizes this mechanism generally adopts an "epidemic"

model in their explanations of how peer influences spread (Jencks & Mayer, 1990).

Case and Katz's (1991) research supports this model; they found that the behavior of

neighborhood peers substantially youth behavior in a manner where "like begets like"

(p. 23).

Social networks refer to the spatially-contingent network of friends, family,

neighbors, and local organizations available to individuals residing in a particular

location (Ellen & Turner, 1997; Galster & Killen, 1995). The mechanism of social

networks is rooted in theories of social capital. According to Lang and Hornburg (1998),

"social capital commonly refers to the stocks of social trust, norms, and networks that

people can draw upon in order to solve common problems" (p 4). Coleman (1988), who

introduced the concept, identifies three distinct dimensions of social capital: obligations

and expectations, information channels, and social norms. Obligations and expectations

relate to the reciprocal nature of assistance offered through social networks. These

reciprocal relations are predicated on a structure of trustworthiness, without which there

would be little incentive for residents to participate. Information channels are critical for

learning about job opportunities and community news (Wilson, 1987), and social norms

influence behavior through the mechanisms of adult and peer influence. Speaking to the

importance of social capital, Coleman (1988) asserts that "social capital is productive,

making possible the achievement of certain ends that in its absence would not be

possible" (p. S98). Galster and Killen (1995) see local social networks as a key

component of the metropolitan opportunity structure, as they shape individual









perceptions regarding the ability of institutions and markets, such as education and the

labor-market, to provide advantageous individual outcomes.

Exposure to crime and violence has obvious implications for individual outcomes,

including physical and psychological harm and property theft. For example, the risk of

victimization is much greater in high-crime areas than others (Sampson & Lauritsen,

1994). This mechanism is capable of affecting all neighborhood residents, regardless of

age, but exposure to crime and violence has effects other than outright victimization for

neighborhood youths. "Simply witnessing crimes or knowing people who have been

victimized may also profoundly affect children's outlook, leading them to see the world

as fundamentally violent, dangerous, and unjust" (Ellen & Turner, 1997, p. 841). Certain

dimensions of this mechanism involve adult and peer influence as well; in

neighborhoods where crime and violence are commonplace, they may be accepted by

youths as normal and even expected activities (Case & Katz, 1991). In addition,

individuals living in high-crime neighborhoods are more likely to live sheltered

existences, precluding them from taking advantage of social networks and

neighborhood amenities, greatly reducing the quality-of-life for even non-victims (Dreier

et al., 2004; Ellen & Turner, 1997).

Neighborhood effects are also held to operate through the cumulative impact of

continued exposure to environmental stressors on health (Taylor, Repetti, & Seeman,

1997). These stressors include such conspicuous examples as crime and violence, but

also the more routine stressors of noise, poor public services, and degraded

environmental conditions. The relationship between stress and place is most frequently

discussed in the context of the poor health outcomes experienced by residents of









disadvantaged communities. In this regard, Geronimus's (2006) concept of weathering

has been particularly influential.21 As initially stated, the weathering hypothesis holds

that "blacks experience early health deterioration as a consequence of the cumulative

impact of repeated experience with social or economic adversity and political

marginalization" (Geronimus, Hicken, Keene, & Bound, 2006, p. 826). While weathering

was originally used to explain racial disparities in health issues related to birth

outcomes, the concept possesses significant explanatory in the context of health

outcomes for a range of low-income families and individuals residing in neighborhoods

with poor levels of service and high levels of hazard (Dreier et al, 2004; Ellen et al.,

2001). Illustrating the serious consequences of chronic stress, it has been found to

cause cardiovascular disease and contribute to premature mortality (Taylor, Repetti, &

Seeman, 1997).

The environmental justice/environmental racism literature has frequently

discussed the relationship between neighborhood characteristics and health outcomes

(Bullard, 2000).22 Research in this area primarily concerns examining whether and to

what degree land uses posing environmental hazards, such as garbage transfer

stations, power plants, medical incinerators, and power-plants are concentrated in

minority or low-income communities. A monumental work in the environmental justice

literature, Bullard's (2000) book Dumping in Dixie, found enormous racial disparities in

the siting of environmental hazards in the South. In their subsequent review of the

environmental justice literature, Pastor, Sadd, and Hipp (2001) found that "while the

21 The hypothesis that chronic stress may have a cumulative, adverse impact on both physical and mental
health was first advanced by Selye (1978) in his landmark book, The Stress of Life.
22 Thought the direct health effects of these hazardous uses have proven difficult to verify (Ellen et al.,
2001), they are no doubt related to neighborhood quality-of-life. In









evidence is more often mixed than many activists have believed, the bulk of the

research does seem to point to disproportionate exposures to hazards in minority

neighborhoods" (p. 3). In a frequently cited study in support of the environmental racism

hypothesis, Boer, Pastor, Sadd, & Snyder (1997) analyzed of the location of hazardous

waste treatment, storage, and disposal facilities (TSDFs) in Los Angeles County. They

found that while considerations external to race and class such as zoning and industrial

employment were associated with the siting of TSDFs, race and ethnicity were still

positively associated with their location. Traffic, truck traffic in particular, is also related

to poor environmental quality and adverse health outcomes through its impact on air

quality (Dreier et al., 2004).

The final mechanism identified by Ellen and Turner (1997) is physical distance and

isolation. They find that "the most straightforward impact of neighborhood is physical

proximity and accessibility to economic opportunities, particularly jobs" (p. 842). The

only literature they include under the rubric of this mechanism concerns the spatial

mismatch hypothesis, the importance of which concept merits its own section in this

literature review. However, recent studies on the absence of healthy and affordable food

options in low-income urban neighborhoods may also be included in discussions of the

effects of physical isolation on individual outcomes. Stemming from British research

conducted in the later 1990s, an increasing amount of literature has been produced

related to the existence and effects of food deserts, which are defined as "areas of poor

access to the provision of healthy affordable food where population is characterized by

deprivation and compound social exclusion" (Wrigley, Warm, & Margetts, 2003).

Research has shown that the type of food store available to neighborhood residents









varies with neighborhood characteristics (Moore & Roux, 2006). Supermarkets, which

offer healthy and affordable options, have been found to be more accessible in higher-

income areas, while convenience stores, which offer more expensive and less-nutritious

options, are found to be more accessible in lower-income neighborhoods.23 In addition,

poorer neighborhoods were found to be more likely to have fewer fruit and vegetable

stands, bakeries, and natural food stores, but more likely to have a greater number of

liquor stores, than more affluent neighborhoods (Moore & Roux, 2006). Reflecting the

impact of neighborhood racial characteristics on individual outcome in the area of food

access, a study of supermarket accessibility in Detroit found that predominantly black

neighborhoods with high levels of poverty were on average 1.1 miles further from

supermarkets than impoverished white neighborhoods (Zenck et al., 2005).

Neighborhoods located in food deserts can have adverse effects on individual health,

resulting in increased medical expenses and diminished quality-of-life. Individuals using

public transportation to access supermarkets may face time-consuming and

inconvenient trips.

Outcomes

While some of the literature of neighborhood effects discusses them principally in

terms of the theoretical mechanisms through which they operate, other works discuss

them in terms of specific outcomes for residents of varying age, gender, and other

personal characteristics. Unlike those that discuss mechanisms, these studies are

generally less concerned with causality than correlation and provide substantial


23 Even when supermarkets are located in urban markets, their prices are generally higher than those in
suburban markets. In an examination of 322 stores in 10 large metropolitan areas, MacDonald and
Nelson (1991) found that the price of fixed market basket of goods was on average four percent higher in
central cities stores than suburban stores.









empirical support for the existence of neighborhood effects. As this is a broad body of

literature, only the most salient threads will be presented here.

Empirical research strongly suggests that neighborhood characteristics have an

impact on childhood cognitive development. Perhaps the most frequently cited study in

support of this hypothesis is that by Brooks-Gunn, Duncan, Klebanov, and Sealand

(1993). These authors quantified the effects of neighborhoods (operationalized at the

census tract level) on childhood outcomes using data from the Panel Study of Income

the Infant Health and Development Program (IHDP). This program randomly selected

895 low-birth-weight children born in eight different sites across the US and observed

the effects of educational and family-support services and medical care on

development. Controlling for family background, Brooks-Gunn et al. (1993) found that

IQs were significantly higher for children living in neighborhoods with higher

concentrations of affluent households, defined as those with an annual income

exceeding $30,000. This study also evaluated outcomes for adolescents using data

from the Panel Study for Income Dynamics (PSID), which provides a lengthy time series

of data related to family and neighborhood characteristics for a randomized national

sample. Using a national sample of black and white women between the ages of 14 and

19, the researchers found that the presence of affluent neighbors is also associated

lower drop-out levels and teenage pregnancies.

As the results of Brooks et al.'s (1993) research suggests, impoverished

neighborhoods adversely affect educational attainment. This particular outcome is

supported by a substantial body of evidence. One such example is Garner and

Radenbush's (1991) study of educational attainment among 2,500 people who dropped









out of school between 1984 and 1986 in Scotland. After controlling for student ability,

family background, and school, they found a significant negative relationship between

neighborhood deprivation scores and educational attainment. Garner and Radenbush

(1991) calculated deprivation scores through a weighted combination of 12 census

variables, among which the most heavily weighted were unemployment, single-parent

families, low-earning socioeconomic groups, overcrowding, and the percentage of

permanently sick individuals. Testing a similar hypothesis, Crane (1991) studied the

relationship between neighborhood quality and drop-out rates. Crane (1991) used a

1970 Public Use Microdata Sample (PUMS) to examine teenagers living with their

parents, selecting 92,512 teenagers for a sample group. Controlling for individual

characteristics, Crane (1991) found the likelihood of dropping out of school for

adolescents of all races increases exponentially as the percentage of workers in the

neighborhood holding professional or managerial positions declined. While the pattern

of neighborhood effects is relatively linear for Hispanics, the risk of dropping-out for

blacks and whites rises precipitously once the local share of middle-class workers drops

below 3.5%.

Unlike Crane (1991), Duncan (1994) found that neighborhood effects on

educational attainment differ based on race and gender. Using PSID data to measure

the effects of neighborhood (census tract) and family characteristics on schooling

completed, Duncan (1994) found that higher levels of female employment reduced

college attendance rates for both black and white women and that a larger share of

female-headed households increased dropout rates for black females. In terms of the

effects produced by a larger share of affluent neighbors, this neighborhood









characteristic was found to increase the likelihood of college attendance only for white

males, but it was also found to be a predictor of completed schooling for dropouts and

college students.

In a study of outcomes for participants in the Chicago's Gatreaux program,

Rosenbaum (1995) found that there were significant differences between those that

moved to the suburbs and those that moved to another part of the urban area.

Rosenbaum (1995) conduced two separate studies to determine educational

attainment; one test was performed by selecting one school-aged child each from 114

families in 1982 (six years after the start of the program) and another was conducted on

the same individuals in 1989. Based on the second test, suburban-movers were shown

to have improved outcomes in terms of both educational attainment and employment.

Rosenbaum's (1995) results showed that more suburban movers than city movers had

attained a greater degree of academic achievement, were on a college track (40%

versus 24%), were enrolled in college (54% versus 21%), were in four-year colleges

rather than junior colleges (50% versus 20%), and had obtained jobs paying more than

$6.50 per hour (21% versus 5%). Also, fewer suburban movers dropped out of school

than city movers (20% versus 5%).

Teen pregnancy is another hypothesized outcome of neighborhood conditions

supported by substantial research. Crane's (1991) study also employed a sample of

44,466 females between 16 and 19 years taken from the 1970 PUMS for a childbearing

analysis. The results of this analysis are similar to those for his test of dropout rates;

Crane (1991) found that there is a sharp increase in the childbearing probability for both

black and white women when the share of professional or managerial workers drops









below 3.5 percent. The results of Brooks-Gunn et al.'s (1993) study also support the

argument that neighborhood characteristics are associated with out-of-wedlock

childbearing among adolescents. Brewster (1994) examined the impact of

neighborhood characteristics on racial differences in sexual activity among adolescent

women, using both individual level data on women between the ages of 15 and 19

taken from the National Survey of Family Growth (NSFG) and census data. Her results

showed that neighborhood socioeconomic statues and full-time female employment rate

are positively associated with risk of experiencing non-marital intercourse during

adolescence. Brewster (1994) concluded that racial differences in risk reflect racial

differences in access to economic resources and positive female role models.

Neighborhood conditions, in essence, alter the perceived incentive structure for black

women in a manner such that the cost of sexual activity seems low. Despite the

evidence supplied by these findings, however, other recent studies have failed to find a

link between neighborhood and adolescent sexual behavior, suggesting caution in

interpreting the results of previous studies (Ellen & Turner, 1997).

In terms of the link between neighborhood characteristics and crime, Case and

Katz's (1991) study provides the strongest evidence. Using National Bureau of

Economic Research Data on 1,200 disadvantaged youth between the ages of 17 and

24 in Boston, Case and Katz (1991) found that a 10 percent increase in the teenage

crime rate increased the likelihood of a youth committing a crime within the past year by

2.3%. This study also found that a given youth was 3.2% more likely to use illegal

drugs, 2.7% more likely to be friends with gang members, and 3.4% more likely to use









alcohol weekly if the neighborhood was subject to a 10% percent increase in other

youth exhibiting the same behavior.

Neighborhood effects also have a demonstrated impact on health. While the link

between individual wealth and health is now taken for granted, research has also found

that low neighborhood socioeconomic status is associated with both a range of health

problems and mortality from a number of causes (Dreier, et al., 2004). Using a sample

of 12,601 persons from four communities in the US and 1990 census tract data related

to socioeconomic status (SES) in a multi-level regression analysis, Diez-Roux et al.

(1997) found that neighborhoods with lower SES were associated with higher

incidences of coronary heart disease than more affluent communities. In a longitudinal

study of 1,129 adults in Alameda County, California Yen and Kaplan (1999) found that

lower quality social environments (measured in terms of per capital income, degree of

residential crowding, quantity of commercial stores, and ratio of homeowners to renters)

were associated with an increased risk of death. The results of this study were

significant even when controlling for individual characteristics such as age, income,

gender, smoking status, body mass index, and alcohol consumption (Yen & Kaplan,

1999). In addition, residents in communities with fewer healthy food options have been

shown to exhibit poor dietary habits, which may lead to diabetes, obesity and other

conditions. In a randomized control study of 22 communities nationwide, Cheadle et al.

(1991) found a significant association between the availability of healthy products and

their consumption. Studies also show that the density of fast-food restaurants is greater

in minority and lower-income neighborhoods, identifying this as a possible

environmental factor in the prevalence of obesity in disadvantaged neighborhoods. In a









study of all fast-food restaurants in New Orleans, Block, Scribner, & De Salvo (2004)

found that predominantly black neighborhoods have 2.4 fast-food restaurants per

square mile, while white neighborhoods have only 1.5.

Additional research performed in this area has been motivated by the desire to test

the purported benefits of housing mobility programs. In a review of the housing mobility

and health literature, Acevedo-Garcia et al. (2004) found that these studies support the

general conclusion that moving to lower-poverty neighborhoods may contribute to the

improvement of health and health-related behaviors for both adults and children. While

these effects have been shown to varying degrees along a host of health-related

dimensions, evidence in support of neighborhood effects on mental health is the

strongest. In a study of the short-term impact of the Moving to Opportunity (MTO)

program in New York City, Leventhal and Brooks-Gunn (2003) found that parents who

moved from high- to low-poverty neighborhoods experienced significantly less mental

distress than those who remained in impoverished neighborhoods. Leventhal and

Brooks-Gunn (2003) employed a randomized controlled design using three groups: an

experimental group of families participating in the MTO program, a comparison group of

families receiving Section 8 vouchers, and a control group of families receiving project-

based assistance However, the results were varied with regard to age and gender. Boys

in the experimental group were found to be significantly less likely to report problems

with anxiety or depression than the in-place control group, while no statistically

significant differences appeared for girls. Boys between the ages of 8 and 13 in the

Section 8 group were significantly less likely to exhibit headstrong behavior than the in-









place control group, but no significant differences were observed between the groups

for youths aged 14 to 18 years old.

Neighborhood Change

Although suburbanization remains the predominant housing trend in the United

States, many cities have witnessed the return of middle-class and professional

households. This process has generated acclaim and disdain alike, largely as result of

competing definitions of the process. Seeking to remedy the conceptual confusion

frequently generated by the term, Kennedy and Leonard (2001) define gentrification as

"the process by which higher income households displace lower income residents of a

neighborhood, changing its essential character and flavor" (p. 5). This is perhaps the

most comprehensive and useful formulation of gentrification as it takes both the social

and economic dimensions of the process into consideration. However, it is important to

note that gentrification is neither inherently good nor bad; rather, it is capable of

producing both positive and negative effects where it occurs. As a result, policy

recommendations suggest allowing the process to occur in a manner conducive to

realizing its benefits while simultaneously working to mitigate its negative consequences

(Kennedy & Leonard, 2001). Positive aspects of gentrification include deconcentration

of poverty, reduction in crime, neighborhood revitalization, and increased real estate

values and equity (Kennedy & Leonard, 2001; Sullivan, 2007). Negative aspects include

changes in the traditional character of the neighborhood and conflict between

community members; however, the most significant adverse impact of gentrification is

the displacement of lower-income residents through higher property values and their

concomitant rent level increases. In a climate in which the housing market is tight and

the availability of affordable housing low, gentrification exacerbates the already difficult









prospects of low-income households. In order to reflect the potential good that may

result from this process, as well as the fact that displacement does not always occur in

areas that witness investment, this study, following Mallach (2008) adopts the term

"neighborhood change" to refer to the complex totality of affects under review.

Until recently, two dominant theories accounted for the occurrence of

neighborhood change: demand-side conditions and supply-side conditions. Supply-side

arguments seek to explain neighborhood change through trends in the movement of

capital and its effect on urban space. In the wake of urban disinvestment after mid-

century, a widening margin has been created between actual and potential land values.

This margin is referred to in the literature as a "rent gap," which Smith (1987) defines as

"the gap between the actual capitalized ground rent (land value) of a plot of land given

its present use and the potential ground rent that might be gleaned under a 'higher and

better' use" (p. 462). Demand-side arguments focus on consumer preferences for urban

areas, which may be generated by such conditions as a growth in white-collar jobs; life-

style choices including deferred child-rearing, and the attraction of urban life to younger

professionals (NeighborWorks America, 2005). Bridging these competing schools of

thought, and reflecting the complexity of the causes of gentrification, Kennedy and

Leonard (2001) offer several conditions under which neighborhood change is likely to

occur, including rapid job growth, tight housing markets, consumer preference for city

amenities, increased traffic congestion and lengthening commutes, as well as public

policies facilitating the ingress of middle-class homebuyers.

Regardless of a study's methodology, data collection on neighborhood change is

considerably problematic. Several of these obstacles include the wide intervals between









data collection at the neighborhood level, the complex interaction of phenomenon

ostensibly contributing to neighborhood change, as well as the problem of using

regional data to account for specific localities (Kennedy & Leonard, 2001). As a result of

the complexity associated with indicators, methodology and data collection, research in

the areas of gentrification has produced contradictory results. However, two recent

studies suggest that displacement, while it necessitates policy attention, is not as

widespread and precipitous as previously thought (Freeman & Braconi, 2004; Vigdor,

2002). Counter-intuitively, Freeman and Braconi's (2004) study "indicates that rather

than speeding up the departure of low-income residents through displacement,

neighborhood gentrification was actually associated with a lower propensity of

disadvantaged households to move" (p.51). However, through residents may choose to

remain, they may be considerably more cost-burdened than before neighborhood

change occurred. Thus it remains critical to identify and monitor indicators for

neighborhood change in order to protect the existing stock of affordable housing.

Previous Studies on the Location of Assisted Housing

Despite the federal government's stated policy objective of providing affordable

housing in suitable neighborhoods, project-based assisted housing as a whole has been

found to be generally situated in relatively lower-quality neighborhoods (Newman &

Schnare, 1997). However, this statement should be tempered with the observation that

privately-owned assisted housing units have been found to be located in generally

better neighborhoods (measured in terms of racial segregation and poverty

concentration) than public housing units, which are frequently found in the most

disadvantaged areas (Massey & Kanaiaupuni, 1993). Relatively few studies directly

address the location of assisted housing in terms of neighborhood quality, but the ones









that do exist provide important context for this study.24 While an exhaustive review will

not here be attempted, a few key studies tracing the geography of opportunity available

to residents of assisted housing will be presented.

In a comparison of the neighborhoods surrounding assisted properties in the U.S.,

Newman and Schnare (1997) found that privately-owned assisted housing units

(including both HUD and LIHTC properties) are "significantly more concentrated in

lower-income neighborhoods and significantly less concentrated in upper-income

neighborhoods" when compared to the total universe of rental housing (p. 711). In an

examination of the differential neighborhood quality of properties developed under

separate privately-owned assisted programs, this study found that the proportions of

HUD- and LIHTC-assisted units located in highly segregated areas were similarly high.

Roughly one-third of the units developed under both the LIHTC program and other HUD

programs were located in census tracts where minorities constitute more than 40% of all

households. These results compare negatively with those for units occupied by

certificate and voucher holders (one quarter) and state-subsidized units (one fifth)

(Newman & Schnare, 1997). Newman and Schnare (1997) concluded that privately-

owned assisted housing programs do little to improve neighborhood quality relative to

welfare recipients, who are used as a proxy for properties in areas with poor

neighborhood quality, as they are most often found in neighborhoods with high poverty


24 However, a substantial body of literature examines whether or not assisted housing has been
developed in progressively less-impoverished and racially-segregated neighborhoods since the passage
of Civil Rights legislation and the introduction of housing programs designed to offer households greater
options in terms of location. The results are mixed, but there is evidence supporting the hypothesis that
assisted housing, especially under the LIHTC program, is being increasingly located in less-segregated
areas (Rohe & Freeman, 2001).









rates. Their study found that much of the assisted housing stock is located in

neighborhoods lacking suitable quality (expressed as economic status, housing quality,

concentration of assisted housing, and racial/ethnic mix). Of particular relevance to this

research, it suggests that there is little, if any, demonstrable difference in the

neighborhood quality of properties developed under HUD programs or the LIHTC

program.

Subsequent studies, however, have found that a significant number of LIHTC units

have been developed in areas of low-to-moderate poverty, rather than heavily-

concentrated poverty. Freeman (2004) analyzed the location and neighborhood

characteristics of LIHTC properties developed in the 1990s in the 100 metropolitan

areas with the largest numbers of units. Freeman (2004) found that approximately 42%

of all LIHTC units were located in the suburbs, compared to only 24% of all other

privately-owned federally assisted housing. Freeman (2004) also found that

neighborhoods in which LIHTC properties were located experienced larger declines in

poverty compared to median metropolitan values. On the other hand, Freeman (2004)

also found that neighborhoods with LIHTC properties have higher poverty rates, lower

median incomes, and lower median home values when compared to median

metropolitan values. A subsequent analysis of LIHTC developments in Atlanta, Chicago,

Los Angeles, and New York City by Oakley (2008) confirms Freeman's results. Oakley

(2008) found that "neighborhood characteristics associated with other assisted housing

programs like low income, poverty, and unemployment are not significant predictors of

the presence of LIHTC developments" (p. 624). Oakley attributes this finding to more

LIHTC units being located in the suburbs than assisted housing developed under other









federal programs. Tempering these results, Oakley (2008) also found that the single-

strongest predictor of the presence of LIHTC units are the presence of qualified census

tracts (QCTs), which are low-income areas eligible for assistance under certain

programs. These results, as Oakley (2008) commented, reflect the fact that under the

LIHTC program, bonuses are awarded to developers building in QCTs.

While many of the neighborhoods surrounding the assisted housing stock are

characterized by higher relative concentrations of poverty and minority households,

studies in some cities have found that privately-owned assisted housing performs much

better in terms of accessibility and location efficiency, which is an additional measure of

value useful for prioritizing preservation initiatives. A joint study by the National Housing

Trust and Reconnecting America (2008) found that a large share of federally assisted

units in eight major metropolitan areas are located within a half mile of existing or

proposed rail stations and bus stops, meaning that they are accessible to jobs and

services for low-income households. Their study found that Boston, Chicago, Cleveland,

Denver, New York City, Portland, St. Louis, and Seattle collectively contain more than

100,000 units of privately-owned federally-assisted housing. The percentage of units

near transit varied drastically between cities, however. In New York, 72% of federally-

assisted units are located within a half mile of an existing or proposed rail station. On

the opposite side of the spectrum, only 9% of St. Louis's units were located within the

same distance.

Summary

This chapter has presented the concept of the geography of opportunity, the

literature that frames the issue, and the evidence that supports the claim that location is

an important determinant of value for housing. Based on this literature review, several









key indicators of housing suitability measuring the relative quality of the opportunity

structure available to assisted housing residents have been selected for use in this

study. The next chapter presents these indicators and the methodology that will be used

in comparing different generations of the assisted housing stock.









CHAPTER 4
METHODOLOGY

The purpose of this study is to assess properties in the assisted housing inventory

(AHI) across a number of suitability criteria based on the generation in which they were

produced. Thus, the research design employed in this study compares the two different

generations of properties within each study area. The research design consists of

several phases. First, a geographic information system (GIS)-based land-use suitability

model is run for each of the study areas to find suitability for the purposes of affordable

housing across a number of criteria, each critical to determining the value of properties

in the AHI. Next, properties in the AHI are assigned the suitability values generated from

the model at their specific locations. After the suitability results have been joined to the

properties, they are subjected to statistical analysis. Descriptive statistics for the

suitability variables of the two different generations will be determined in each study

area, and statistical analyses will be performed to test for differences between the two

generations for each of the selected suitability criteria.

Affordable Housing Suitability Model

This study employs the Affordable Housing Suitability (AHS) model in determining

site suitability for properties in the AHI. The AHS is a GIS-based land-use suitability

model designed to assist community stakeholders in addressing their affordable

housing needs by identifying and assessing the suitability of sites for the production and

preservation of affordable housing. The model functions as a planning support system

(PSS) by using information technology to integrate community preferences, planning

expertise, and key spatial data in the pursuit of this critical community planning objective

(Harris & Beatty, 1993; Heikkila, 1998; Klosterman, 1997). The AHS model is a









collaborative effort of the Department of Urban and Regional Planning and the

Shimberg Center for Housing Studies at the University of Florida, and has been

developed with the support of the Wachovia Foundation and the MacArthur Foundation.

This project has been developed with the assistance of local planning agencies in three

Florida Counties: Duval (Jacksonville), Orange (Orlando) and Pinellas (Clearwater and

St. Petersburg in the Tampa Bay area). As the AHS model has so far been concerned

with assessing the suitability of these three communities, they constitute the study areas

selected for this study.

Geographic Information Systems

Many definitions of GIS exist, ranging from purely functional descriptions of

software applications and database capabilities, to more comprehensive definitions

situating the technology in the social context of its development and use (Chrisman,

1997; Heikkila, 1998). Burrough's (1986) classic functionalist definition states that a GIS

is "a powerful set of tools for storing and retrieving at will, transforming and displaying

spatial data from the real world for a particular set of purposes" (p. 6). At the most

fundamental level, all GIS "integrate a mapping function, which displays maps or

geographic features, with a database manager, which organizes the attribute data tied

to the various map features" (Levine & Landis, 1989, pp. 209-10). Five basic types of

functionality are conventionally cited in defining the capabilities of a GIS: the "capture,

storage, retrieval, analysis and display of spatial data" (Clarke, 1986, p. 175). In addition

to software, hardware and data, a critical component of a GIS is its "liveware," that is,

"the people responsible for designing, implementing and using GIS" (Maguire, 1991). A

GIS also encompasses the institutional frameworks and cultural practices structuring its

design and use (Chrisman, 1997; Maguire, 1991). Thus, a GIS is not simply a tool, but









is rather an integrated system of technical and social relations (Innes & Simpson, 1993).

In the context of the AHS model, Cowen's (1988) holistic and goal-oriented definition of

a GIS as a "decision support system involving the integration of spatially referenced

data in a problem solving environment" is particularly relevant (p. 1554).

Land-Use Suitability Analysis

While the storage and display of spatially-referenced data is at the core of GIS

functionality, the "greatest strength of GIS is in creating new information and combining

different sources of geographic data through the process of overlays" (Harris & Batty,

2001, p. 185). One of the most useful applications of these powerful overlay processes

is land-use suitability analysis (Collins, Steiner, & Rushman, 2001; Malczewski, 2004).

The purpose of a land-use suitability analysis is to identify the spatial pattern "of

requirements, preferences, or predictors of some use" (Hopkins, 1977, p. 386). In the

case of the AHS model, multiple layers of spatial information are combined in order to

produce a composite representation of suitability for affordable housing. Each of the

distinct layers included in the model have been selected based on their utility in

assessing the suitability of a particular site for affordable housing, such as

neighborhood conditions and distance from essential services. While each of these

layers individually provides critical information relevant in assessing suitability for this

purpose, the synthesis of this information through the overlay analysis generates

entirely new information useful in guiding intelligent community planning decisions.

The GIS-based land-use suitability analysis is rooted in overlay techniques

developed by landscape architects using hand-drawn sieve maps in the late nineteenth

and early twentieth centuries (Collins et al., 2001; Steinitz, Parker, & Jordan, 1976).

These overlay techniques for determining land-use suitability were significantly









advanced and widely popularized through lan McHarg's 1969 book Design with Nature

(Collins et al., 2001; McHarg, 1992). McHarg's (1992) procedure for overlay suitability

analysis began with recording the social values assigned to features in the built and

natural environments on separate transparent maps, which are shaded on a scale of

light to dark to represent varying degrees of value. These individual maps were

superimposed in order to produce a composite suitability map for a particular land use

category. The darkest areas on this composite map have the highest social value,

hence, the greatest cost associated with their conversion to another use. Conversely,

the lightest areas on the composite map have the lowest social value attached to them,

making them the most suitable (least costly) for conversion (McHarg, 1992). The

McHargian approach is useful for visualizing the social costs of different land-use

scenarios, especially the loss of ecologically important features.

While McHarg's overlay technique has had a profound influence on the

development of suitability analysis in GIS, computer-assisted overlay techniques were

developed in response to the limitation of hand-drawn overlay mapping (Collins, 2001).

Computer-assisted mapping techniques advanced at Harvard in the 1960s significantly

expanded the range and depth of land-use suitability analysis. An important milestone

was Harvard's development of SYMAP (synagraphic mapping system), which allowed

individual maps to be overprinted using gray scales in order to produce a composite

suitability map (Collins, 2001). Further advancements in computer-based suitability

analysis came through the formal development of GIS software and the introduction of

map algebra and cartographic modeling techniques pioneered by Dana Tomlin (Collins,

2001). The integration of multicriteria (or multiattribute) decision making (MCDM) and









multicriteria evaluation (MCE) methods with GIS has significantly advanced the

capabilities of overlay suitability analyses by allowing the incorporation of preferences in

a complex decision-making environment with multiple, and potentially conflicting, criteria

(Colllins Steiner, & Rushman, 2001; Jankowsi, 1995; Malczewski, 2004).25 Rather than

operating in a simple Boolean yes/no decision-making environment, these methods

allow decision-makers to determine the tradeoffs between criteria when conflict arises.

GIS-based land-use suitability models have been used in a variety of contexts, including

environmental studies of land for habitat suitability (Pereira & Duckstein, 1993; Store &

Kangas, 2001) and land conservation (Miller, Collins, Steiner, & Cook, 1998; Trust for

Public Land, 2005), as well as for more urban purposes such as public facility siting

(Higgs, 2006) and sustainable residential development (Sorrentino, Meenar, & Flamm,

2008). Suitability models have also been employed to predict future land-use scenarios

(Carr & Zwick, 2005).

While GIS-based land use suitability analysis has been applied in order to assess

sites for a variety of activities, few have been directly concerned with identifying sites

suitable for the location of affordable housing. Thomson and Hardin (2000) identify

potential sites for affordable housing in Bangkok, Thailand, employing criteria primarily

concerned with the physical characteristics of the study area, specifically, compatible





25Jankowski (1995) finds that the "general objective of MCDM is to assist the decision-maker (DM) is
selecting the 'best' alternative from the number of feasible-alternatives under the presence of multiple
choice criteria and diverse criterion priorities" (p. 252). According to Malczewksi (2004): "GIS-based
MCDA can be thought of as a process that combines and transforms spatial and aspatial data (inputs)
into a resultant decision (output). The MCDM procedures (or decision rules) define a relationship between
the input maps and the output map. The procedures involve the utilization of geographical data, the
decision maker's preferences and the manipulation of the data and preferences according to specified
decision rules" (p. 33).









land use, proximity to transportation routes, and flood risk.26 These criteria, while critical

for the purposes of affordable housing, are equally applicable for the siting of any

residential development, regardless of affordability. In order to incorporate additional

suitability criteria specifically related to affordable housing, Biermann (1999), created an

overlay suitability model incorporating socio-economic variables in addition to physical

and environmental conditions. Svatos and Doucette (2003) employed a similar

approach in a GIS-based suitability analysis for affordable housing in two Delaware

counties. Values of low, moderate, or high were assigned across the surface of the

maps for each criterion based on subjective thresholds. These layers were then

summed in order to represent the composite suitability of land for affordable housing.

Recognizing that the selection of sites suitable for housing (of all types) requires a

complex decision-making structure cutting across environmental, political, and social

dimensions, AI-Shalabi, Mansor, Ahmed, and Shiriff (2006) applied MCDM techniques

in a housing suitability analysis of Sana'a, Yemen.

Affordable Housing Suitability Model Structure and Methodology

Fundamental to the design of the AHS model is the Analytic Hierarchy Process

(AHP), an MCDM technique that decomposes complex suitability problems into a

comprehensive and logically-consistent hierarchical decision-making framework and

employs pairwise comparison to establish factor weights at each level within the

hierarchy (Banai-Kashini, 1989; Saaty, 2008). 27 Following the AHP technique, the AHS


26 This study, however, does not employ an overlay analysis, but instead considers each criterion
separately.
27 As all factors are assigned equal weights in this study, a more comprehensive explanation of AHP is
beyond the scope of this paper. Equal weights are used as the study area communities are still in the
process of assigning priorities to suitability criteria through pairwise comparison.









model is organized as a hierarchy of goals, objectives, and sub-objectives containing

sets of suitability criteria (see Figure 4-1). Due to the significant number of variables

needed for a comprehensive affordable housing suitability analysis, the AHS model is

comprised of three distinct goals: the identification of suitable sites based on

preferences expressed for "affordable housing" characteristics, 28 the identification of

suitable sites based on transportation costs, and the identification of suitable sites

based on the demand for affordable housing driven by employment (see Figure 4-2).

The generation of suitability values for each of the layers constitutes the first phase of

the AHS model. A second phase, based on Carr and Zwick's (2007) land-use conflict

identification strategy (LUCIS) model, is used to combine the separate goals in a

manner emphasizing potential points of both agreement and disagreement existing

between the suitability values of different goals. The third and final phase of the model

is an allocation procedure for selecting specific sites for suitable housing in each study

area based on policy criteria and community preference. This research, however, is only

concerned with first phase of the model, which is discussed below.

As currently structured, only the first goal ("affordable housing") of the AHS model

is designed to accommodate pairwise comparison, due to its complex hierarchy of

objectives and sub-objectives. The objectives contained in the first goal are:

* To identify property suitable for development or preservation of affordable housing
based on land and site characteristics.



28 The term "affordable housing," as expressed in the context of the first goal of the AHS model is
somewhat misleading. The entirety of the model is used for the purpose of identifying sites suitable for the
development and preservation of affordable housing based on local planning preference and expert
opinion, but the "affordable housing" goal is characterized by socioeconomic conditions, environmental
quality, and local accessibility. In general, the variables in this goal are more site-specific than those in
the other two.









* To identify property suitable for development or preservation of affordable housing
based on socio-economic conditions.

* To identify property suitable for development or preservation of affordable housing
based on local accessibility

* To identify property suitable for development or preservation of affordable housing
based on neighborhood change

Each one of these objectives consists of multiple sub-objectives, which consist of

distinct criteria. These criteria are integrated into the model in the form of map layers

representing specific features relevant to determining suitability for affordable housing,

such as the location of schools or areas with high crime rates. Scores are assigned

across the surface of these individual map layers based on the positive or negative

impact their attributes have on the suitability of different locations, transforming what

was before a simple representation of features into a suitability surface that functions as

a building block for the rest of the model. For example, in the case of the map layer

representing the location of hospitals, the closer a residential area is to a hospital, the

higher the suitability value it is assigned; conversely, values decrease with distance

from these hospitals. This process is oriented to the input of planners and experts on

housing policy but also incorporates statistical techniques to assign suitability values to

geographic locations. The raster model of spatial representation is employed throughout

the AHS, beginning with these initial suitability conversions.29 Rendering all of the

criteria layers into the same geometric grid system using raster analysis facilitates the

overlay operations used in the model (Chrisman, 1997). To illustrate this process,

Figure 4-5 presents the local accessibility suitability for Orange County.

29 Raster data models represent geographic space as a regular grid composed of cells aligned in rows
and columns. Raster models create a standard geographic unit for analysis among various spatial input
layers, allowing for overlay operations. The AHS model employs 30 X 30 meter cells for its raster
analysis, meaning that it is capable of a very fine-grained analysis.









Once the initial phase of creating these standardized suitability surfaces is

complete, the study communities and the model development team collaborate on

setting priorities for the different elements in the hierarchical structure of the model.

Thus, this stage of the model incorporates the determination of community preference.

Preference differs from suitability in that while suitability identifies potential land areas

consistent with the goals of the model, the purpose of preference is to "capture

community values while assessing locations that have already been identified" (Carr &

Zwick, 2007, p. 128). This work employs AHP to translate qualitative judgments about

the relative importance of criteria into quantitative weights assigned to layers. As

subordinate layers are combined while proceeding through the hierarchical structure of

the model, these weights determine the relative contribution of each layer to higher-

order elements of the model. The final weighted sum operation in the GIS combines

information on suitability contained in all of the model layers into a single output map

that indicates goal suitability, measured on a continuous ratio scale from 1 (low) to 9

(high). This final output map is referred to as a preference surface, as it expresses

community preference regarding the relative importance of multiple criteria in an

individual GIS map layer.

In sum, the first phase of the AHS model is comprised of four steps:

* Identifying objectives and sub-objectives as criteria for the ultimate goal of the
model and structuring them within a hierarchical framework.

* Obtaining and preparing relevant data sources for meeting these criteria.

* Determining the suitability of layers from these data sources through statistical
analysis and expert input.

* Deriving preference by combining suitability layers through AHP and weighted
overlay analysis in order to produce cumulative suitability layers for first sub-
objectives, then objectives, and finally the single output layer for individual goals.









Methodology of This Study

The suitability values of specific goals, objectives and sub-objectives were

selected from the AHS model to be used as criteria in evaluating properties in the AHI.

Several steps were undertaken in order to join the suitability values for each of the

selected suitability layers to properties in the AHI. As with the AHS model, this study

examines the specific Florida pilot communities of Duval, Orange, and Pinellas

Counties. Therefore, only AHI properties located within each of these counties were

selected for analysis. In order to bring the AHI into a GIS environment, properties in

each of the study areas were geocoded based on their address. Geocoding is a process

that finds the spatial coordinates of tabular data and allows for the visual representation

of them in a GIS as points on a map. These visual representations of AHI properties

were then converted into independent spatial data layers and overlaid on to raster

layers generated by the AHS for each of the selected suitability criteria. This process is

necessary in order to match AHI properties with the suitability values found at their

specific locations. The GIS tool "extract values to points" was used to join the suitability

values from individual AHS raster suitability layers to the properties in the AHI data

layer. This procedure was repeated for each of the selected suitability criteria. Once the

suitability values for each criterion had been joined to the AHI layer, the attribute table

containing these values was exported as a database file so that the results could be

subjected to analysis using statistical software.

Suitability Criteria Selected for This Study

The suitability values of selected goals, objectives and sub-objectives were

extracted from the AHS model to be used as suitability criteria in assessing properties in

the AHI. Some of the reclassification procedures may appear counter-intuitive, but they









are designed to generate values of suitability for affordable housing development and

preservation and are not pure measures of any given phenomenon. For example, high

suitability values for the neighborhood decline indicator are actually indicative of greater

demonstrable decline. This is because areas in decline may be suitable for affordable

housing intervention to stabilize the neighborhood. Thus, they receive higher suitability

scores. The variables selected from the AHS model for this study are:

* Crime risk: This is a sub-objective of the socioeconomic indicators objective. The
suitability values for this sub-objective are produced by overlaying violent crime
risk and property crime risk. This criterion was selected for its obvious significance
in determining suitability for affordable housing, including preservation initiatives.
Values above the mean for all block groups are reclassified with a suitability value
of 1, with suitability increasing one unit for every 1/4 standard deviation increment
below the mean.

* Poverty rate: This is one of several criteria layers used in finding suitability for the
sub-objective of positive neighborhood qualities. High poverty rates are associated
with a weak housing market, which, as mentioned above, is an indicator of risk for
the failing-out of an assisted property from the AHI. Values above the mean for
block groups were reclassified as a 1, with suitability values increasing one point
for every 1/4 standard deviation increment below the mean.

* FCAT (Florida Comprehensive Assessment Test) score: Reflecting neighborhood
school performance, this criterion layer is incorporated into the sub-objective of
positive neighborhood qualities. This layer is constructed by overlaying suitability
layers created from the FCAT scores of both elementary and high school students.
FCAT scores were reclassified as follows: A=9, B=7, C=5, D=3, F=1.This criterion
was selected for its use in indicating school quality, hence, suitability for
households with school-age children.30

* Neighborhood decline: Along with gentrification (discussed below), this is one of
two sub-objectives constituting the neighborhood change objective. Neighborhood
change considers changes in values among select indicators using two distinct
spatial scales: short-term change (2000-2007) and long-term change (1980-2000)
(see Figure 4-3. High measures of neighborhood decline may suggest that an
assisted property is located in a weak housing market. As discussed above, this is
an indicator used in assessing risk of a property failing-out of the AHI. Standard
deviation of values was used to determine suitability for each input layer. A subset

30 Florida assesses school performance based on FCAT scores and allocates greater levels of funding to
high-performance schools.









analysis was applied in the design of this component so that it only considers
areas in which median household income for a census tract was within the inter-
quartile range of median household incomes for the entire study area in 1980. The
purpose of the subset is to control for decline in areas either well above median
income (which might not manifest the problems and pressures conventionally
associated with neighborhood decline) and those in already depressed
communities. Reclassification occurs as follows: for a variable in which negative
change is associated with decline, all areas of positive change are assigned a
suitability value of 1, suitability increases one point for every 1/4 standard
deviation increment below the mean.

* Gentrification: As mentioned above, neighborhood change is one of two sub-
objectives constituting the neighborhood change objective. High measures of
gentrification suggest that an area has considerably appreciated over time,
increasing pressures on assisted properties capable of doing so to opt-out. A
subset analysis was applied in this suitability analysis in order to measure change
only in census tracts where the median household income was below the median
value for the entire study area. The purpose of this subset is to exclude already
affluent areas not subject to conventional gentrification processes, such as
displacement (Kennedy & Leonard, 2001). Reclassification occurs as follows: for a
variable in which negative change is associated with gentrification, all areas of
positive change are assigned a suitability value of 1, suitability increases one point
for every 1/4 standard deviation increment below the mean. For an indicator in
which positive change is associated with gentrification, the opposite holds true.

* Local accessibility: This objective consists of four sub-objectives: neighborhood
access to transit (transit stops), neighborhood access to services (elementary
schools, daycare centers, local health care, hospitals, fire and police stations),
neighborhood access to recreation (community centers, cultural centers, parks),
and neighborhood access to retail (restaurants, shopping). Accessibility for each of
these activities is assessed along two dimensions: opportunity (square footage)/
proximity (distance) and walking (four mile radius)/biking (half mile radius). Figure
4-4 provides an illustration of this organization scheme for the aggregated sub-
objective of accessibility to services and facilities. Suitability values are generated
for each surface by applying a raster analysis on a layer containing hospitals (for
example) in order to find the total square footage of hospitals located within four
miles (biking) or half a mile (walking) distance from multi-family housing. Raster
cells with values above the mean are reclassified with a suitability score of 9, with
suitability decreasing one unit for every 1/4 standard deviation increment of value
below the mean.

* Affordable housing goal: This is the final output layer generated through the
weighted overlay processes employed in the model. It is derived from the weighted
overlay of the following objectives: land and site characteristics, socio-economic
indicators, local accessibility, and neighborhood change. This provides a
composite suitability value useful for assessing the general value of properties in
the AHI.









Transportation: This is the reclassified output of the transportation cost surface
that constitutes the second goal of the AHS model. Transportation cost is
determined by evaluating the impact of land use on vehicle miles travelled (VMT)
by a household. This layer is generated through a regression analysis using VMT
as an independent variable and land use characteristics as the dependent
variables. These characteristics are the "4 D's" of land use, density, diversity,
design, and destinations, each of which have been found to have varying degrees
of influence on mode choice and VMT (Boarnat & Crane, 2001; Cervero, 2002;
Cervero & Kockelman, 1997). Density is operationalized in the variables of
developed area as percent of total neighborhood area and residential area as
percent of total neighborhood area. Diversity is operationalized in the square
footage of retail, commercial, and office buildings. Design is operationalized as
road miles per developed area and number of intersections per road mile. Finally,
distance is operationalized as distance to nearest residential center, distance to
nearest regional activity center, and range of distances to regional activity centers.
As transportation costs consume a significant portion of low-income household
budgets and determine the degree of accessibility for critical services and
amenities, this indicator is strongly suggestive of the value of assisted properties.

Statistical Methods

Differences in suitability values of properties in the AHI based on age group were

investigated through use of the Mann-Whitney U test, which is a non-parametric test

used to determine whether there is a statistically-significant difference between two

independent sample sets of data (Barber, 1988; Ebdon, 1977).31 Unlike parametric

tests, non-parametric tests are not restricted by assumptions about the nature of the

sample populations. As the level of measurement for the suitability values used in this

study is ordinal, normal population distribution may not be assumed, requiring the use of

a non-parametric test for analysis (Barber, 1988). The Mann-Whitney U operates by

testing the significance of the difference between the medians of the two samples. The

null hypothesis is that the two samples are drawn from the same population. Should the

test substantiate the null hypothesis, "any observed difference between the two

31 Statistical Package for Social Scientists (SPSS) 18 software was used to perform the analyses for this
study.









samples, such that one set of values is consistently larger than the other is due entirely

to chance in the sampling process" (Ebdon, 1977, p. 54). This test has been applied in

the fields of geography and urban planning for numerous purposes. Talen (1997), for

example, used the Mann-Whitney U test to evaluate differences in socioeconomic

characteristics based on whether or not an area possessed high or low access to public

parks.

Two separate Mann-Whitney U tests were run for each county. One test evaluated

the difference between properties built or funded between 1963 and 1979 with those

built or funded between 1980 and 1994. Here, age is used as a proxy for founder, with

the 1963-1979 category serving as a proxy for HUD-funded properties and the 1980-

1994 category serving as a proxy for older FHFC-funded properties. The second test

compares the difference between properties built or funded between 1963 and 1994

with those built or funded between 1995 and 2008. Age is again used as proxy, with the

older category of properties representing the universe of older federal and state

assisted properties and the more recent category representing properties developed

since the LIHTC was made permanent by Congress.

Limitations

One of the primary limitations of this study is that is was conducted prior to the

completion of pairwise comparison in the AHS model. As a result, the relative

preferences held by local communities for component suitability criteria were not

incorporated into the overlay analysis as they will be when the development of the AHS

model has been concluded. In this study, equal preference is assigned to criteria layers

during overlay procedures. For example, when the map layers for violent crime and

property crime are combined in order to create the composite crime suitability layer,









they will each have a 50% contribution to the outcome. The resultant suitability values

for this layer would be different than those obtained through pairwise comparison if

communities had expressed a stronger concern for one or the other of the two criteria.

Another limitation to this study is that it was conducted prior to the completion of the

housing demand goal. This goal incorporates the demand for housing generated by

commercial, office, and residential development. This critical suitability indicator would

have been useful in measuring the value of assisted properties based on their potential

to accommodate low-income workers attracted to employment opportunities in adjacent

areas. A final limitation is the absence of the transportation goal from the analysis of

properties in Pinellas County. This is a result of the fact that data collection and

statistical procedures required to produce the suitability surface for this goal have not

yet been completed.

Summary

The statistical analysis and comparison of the different generations of assisted

housing along key suitability variables should shed light on their relative value to the

affordable housing stock. As discussed in the previous chapter, risk alone may not be

enough to make an informed and efficient preservation response. Newman and

Schnare's (1997) study indicated that a number of privately-owned assisted housing

developments fare poorly in providing their residents with adequate neighborhood

conditions when compared to the entire universe of rental housing. In order to place the

results of the analysis in the context of the specific case study areas, the next chapter

will present demographic, economic, and political characteristics for each county. The

characteristics of each county's assisted housing will also be presented and situated

within the larger context for Florida's assisted housing inventory.



























Figure 4-1. Hierarchical structure of the AHS model.




Phase 1: Suitability modeling


Phase 2 : Opportunity
analysis based on
goal preferences


Phase 3: Allocation of land
for affordable housing
based on constraints


Figure 4-2. Goals and objectives of the AHS model.


w ulb





S


I


I



























Figure 4-3. Structure of the neighborhood change objective.


Figure 4-4. Neighborhood accessibility objective.













Orange County, Florida


Local accessibility
Value
i High: 9

m Low: 1
AHI Generation (Test 1)
* 1963-1979
0 1980-1994
* 1995-2008 (excluded]
- Major Highways
- Waterbodies


o i 12
Mi~s


Figure 4-5. Local accessibility suitability layer, Orange County









CHAPTER 5
DESCRIPTIONS OF STUDY AREAS: DUVAL, ORANGE, AND PINELLAS COUNTIES

Duval, Orange, and Pinellas Counties were selected as study areas for this

research based on the participation of the City of Jacksonville's Department of Housing

and Neighborhoods, the City of Orlando's Department of Housing and Community

Development, and Pinellas County's Department of Community Development in the

development of the Affordable Housing Suitability (AHS) model. Among other things,

these entities have provided crucial data sets and offered critical input regarding the

design of the model. Regardless of this relationship with the three counties, they remain

important areas for study in the context of affordable housing. They are all

demographically-diverse, substantially urbanized, high-growth areas, each also

experiencing significant housing needs. In addition to their similarities in terms of

affordable housing needs, these counties also offer a study of contrasts as they are

representative of different development patterns in Florida.

Duval County

Duval County, located in Northeast Florida, constitutes the core of the Jacksonville

metropolitan statistical area (MSA) and is the 7th most populous county in the state. The

City of Jacksonville and Duval County merged in 1968, from that time all of Duval

County has been governed by a single entity with the exception of Atlantic Beach,

Neptune Beach, Jacksonville Beach, and Baldwin. Duval County has a land area of 774

square miles and a population density of 1,006 persons per square mile.32 The St.

Johns River runs through the City of Jacksonville, where it turns east from its northerly



32 Population and other descriptive data from U.S. Census Bureau, State & County QuickFacts,
http://quickfacts.census.gov/qfd/states/12000.html









course to empty into the Atlantic Ocean. The river provides Jacksonville with the 14th

largest deep-water port in the U.S. This port, along with the area's airports and other

transportation assets make the county a major intermodal transportation hub.

Demographics, Income, and Housing Affordability

Duval County has experienced substantial growth as a result of its young

population, diverse economy, and high quality-of-life. Recent estimates place the

population of Duval County at 857,400, marking a 10% increase over its 2000

population of 778,866. Growth in the county was particularly strong in the 1990s, when

the population increased by 16 percent. The county's population is projected to top one

million by 2020 (see Table 5-1). The median age of Duval County is currently 35.9,

making it somewhat younger than the state as a whole, which has a median age of 40.5

(see Table 5-12). However, the median age of Duval County is projected to increase to

40.3 by 2030, by which time the share of persons age 65 or older is projected to have

risen from 11 to 20% of the county's population. Thus, a substantial share of the

county's future growth will come from senior citizens. In 2000, the county had 303,747

households with an average of 2.51 persons per household.

Median household income in 2008 was $50,660, which is higher than the median

household income of Florida as a whole (see Table 5-13). Employment is concentrated

in trade, transportation and utilities; professional and business services; education and

health services; and financial sectors of the economy (see Table 5-2). The county is

also home to numerous military facilities, including Naval Air Station Jacksonville, the









third-largest naval installation in the U.S. The largest private employers include Baptist

Health, Blue Cross & Blue Shield of Florida, the Mayo Clinic, and Citibank.33

The housing market in Duval County has remained affordable relative to Florida as

a whole. While home prices rose significantly during the middle of the first decade of the

twenty-first century, the movement of home prices relative to household income was not

as dramatic as that experienced by the other study areas and the nation as a whole

(see Figure 5-3). In recent years, home prices have continued to fall as a result of the

foreclosure crisis and economic recession (see Table 5-21). Nevertheless, housing

affordability remains a problem for many lower-income households in Duval County. In

2008, 24% of renter households earning less than 50% of AMI paid more than 30% of

their income toward housing (see Table 5-15). The NLIHC's 2010 "Out of Reach" report

finds that the housing wage for a one-bedroom apartment in Duval County is an

affordable 101% of the mean renter wage, but fully 206% of minimum wage.34 The

amount of severely cost burdened households earning 80% or less of AMI is projected

to increase 37% to more than 26,000 by 2030 (see Table 5-3). Among lower-income

households, those earning 30% or less of AMI are particularly susceptible to

experiencing cost burden. The growth in severely-cost burdened household earning up

to 30% of AMI is significantly higher than growth for other lower-income households

(see Table 5-4).





33 Jacksonville and Northeast Regional Economic Development,
http://www.expandinjax.com/About/Regional_Overview/Duval_Co.aspx
34 The housing wage is the full-time hourly wage required to pay HUD's established FMR without
spending more than 30% of income.


100









Assisted Housing Inventory

Duval County has the 4th largest number of assisted properties in the state with a

total of 139 properties in the AHI, and it has the 4th largest number of assisted units with

a total of 21,544 (see Table 5-17). Properties receiving assistance from HUD and FHFC

account for roughly even shares of the AHI in Duval County (see Table 5-18). The

majority of the county's assisted units are designated for renters earning between 55

and 60 percent, with only 3% designated for renters earning 35% or less of AMI (see

Table 5-19). For the assisted properties for which data is available, roughly half are

more than 30 years old and 30% are between 21 and 30 years old (see Table 5-20).

Local Housing Policy

The City of Jacksonville strongly supports the preservation of affordable rental

housing in order to meet the housing needs of its low-income residents. Jacksonville's

Consolidated Plan, which outlines the city's housing needs and strategies for meeting

them, explicitly priorities preservation activities, especially the preservation of

affordable housing in gentrifying areas:

The intent of the Consolidated Plan is to put in place policies that will help
and coordinate efforts to preserve and develop affordable housing
opportunities for low and moderate-income residents in a chain reaction
that will help preserve the social and historic character of low-income
neighborhoods threatened by gentrification (City of Jacksonville, 2005, p.
79).

Not only does the city emphasize preservation initiatives in gentrifying areas

characterized by rapid reinvestment, but also in distressed neighborhoods most in need

of assistance, which the city has designated as Neighborhood Action Plan Areas


101









(NAP).35 Jacksonville's Consolidated Plan encourages efforts to preserve existing

affordable rental housing in NAPs though various funding mechanisms, including the

provision of low-interest loans in exchange for affordable rental rates (City of

Jacksonville, 2005). These policies are to the benefit of properties at-risk of failing out of

the assisted housing stock due to physical deterioration and potential default. Another

important dimension of housing policy in Jacksonville is its emphasis on poverty

deconcentration and fair housing practices (Jacksonville, 2005).

Orange County

Located in Central Florida, Orange County forms the core of the Orlando-

Kissimmee MSA and is the 5th most populous county in the state. Like Duval and

Pinellas Counties, Orange is a charter county, meaning that it is self-governing. The

seat of government for Orange County is the City of Orlando. The cities in which

properties in the Assisted Housing Inventory are located are Apopka (pop. 35,563),

Orlando (pop. 220, 186), and Winter Park (pop. 28,083). Orange County has a land

area of 907 square miles and a population density of 988.3 persons per square mile.

Orange County is high-growth region and a major tourist destination, as it is the home of

Walt Disney World and Universal Studios Florida.

Demographics, Income, and Housing Affordability

Orange County is an especially high-growth region in Florida. Between 1990 and

2000, the population increased 35% and now stands at over one million (see Table 5-5).

Orange County is younger overall than the state of Florida taken as a whole, and is


35 "The NAP concept is a comprehensive long-term approach to community revitalization that focuses on
community assets as a means of stimulating market driven redevelopment" (City of Jacksonville, 2005, p.
3).


102









younger than the other study areas (see Table 5-12). The median age for the county is

only 33.1, which, unlike the rest of the state, is actually lower than its 2006 value

(Mdn=33.0). While the number of persons age 65 and older is projected to increase

over the next two decades, the share of this demographic is projected to reach just

15%. In 2000, the county had 336,286 households, with an average of 2.61 persons per

household.36

Median household income in 2008 was $50,674, which is higher than the

statewide median value of roughly $46,000 (see 5-13). Employment is concentrated in

the leisure and hospitality; professional and business services; and the trade,

transportation and utilities sectors of the economy (see 5-6). Many of the jobs created

as a consequence of the County's tourism-based economy are low-wage, contributing

to the increase in the number of lower-income households in the county and the

demand for affordable housing.

The housing market in Orange County is currently more expensive than the state

as a whole. Just as in the rest of the state, home prices rose precipitously during the

middle of the first decade of the twenty-first century, increasing as much as 41% in one

year alone (see 5-21). During this period, the ratio of median household income to

median single-family home sales price topped 5.0. (a ratio of 3.0 is used as a

benchmark of housing affordability) (see Figure 5-3). As in the rest of the state, housing

costs have dropped since the onset of the current recession, but, as in the other study

areas, the number of cost-burdened lower-income households is only projected to

increase. In 2008, 24% of renter households earning less than 50% of AMI paid more

36 Population and other descriptive data from U.S. Census Bureau, State & County QuickFacts,
http://quickfacts.census.gov/qfd/states/12000.html


103









than 30% of their income toward housing (see Table 5-15). The NLIHC's 2010 "Out of

Reach" report finds that the housing wage for a one-bedroom apartment in Orange

County is 123% of the mean renter wage and 244% of minimum wage. The number of

severely cost burdened households earning 80% or less of AMI is projected to increase

38% to nearly 40,000 by 2030 (see Table 5-7). By 2020, the number of severely cost

burdened lower-income households will have increased by more than 5,000 (see Table

5-8).

Assisted Housing Inventory

Orange County has the 2nd largest number assisted properties in the state with a

total of 170 properties in the AHI, and it has the 2nd largest number of assisted units with

a total of 31,181 (see Table 5-17). Roughly 65% of the county's assisted properties

were developed by FHFC, while only 15% were developed under HUD programs (see

Table 5-18). As in Duval County, the majority of the county's assisted units are

designated for renters earning between 55 and 60 percent, with only 3% designated for

renters earning 35% or less of AMI. However, far fewer units in Orange County are

designated for renters with incomes between 40 and 50% of AMI (see Table 5-19). For

the assisted properties for which data is available, 22% are more than 30 years old and

30% are between 21 and 30 years old. A larger share of properties (32%) was built in

Orange County within the past 10 years than in Duval County (see Table 5-20).

Local Housing Policy

Housing policy in the City of Orlando highlights the preservation of existing

affordable housing as a means of meeting the needs of low-income residents. Like

Duval County, it encourages balanced housing policies that do not privilege

homeownership and new development to the exclusion of preservation efforts, including


104









those for multifamily rental housing. Illustrating this point, the very first goal of Orlando's

Consolidated Plans is to "increase the availability of existing affordable housing options

within the City of Orlando for extremely low-, low-, and moderate-income residents"

(City of Orlando, 2005, p. SP 7). Under this goal are listed numerous strategies for

meeting this goal, such as, rehabilitating existing single-family and multifamily rental

housing, supporting nonprofits in obtaining additional funding for preservation efforts,

including Community Redevelopment Area Tax Increment Funds, and leveraging

additional funds to assist both for-profits and non-profits in acquiring and rehabilitating

affordable rental units (City of Orlando, 2005). The Consolidated Plan also indicates that

the city plans to continue allocating funds from its share of Community Development

Block Grants (CDBG), HOME grants, and State Housing Initiatives Partnership (SHIP)

funds toward preservation efforts (City of Orlando, 2005, p. SP 37). In addition,

Orlando's Housing Element contains a policy explicitly expressing the city's desire to

preserve assisted housing: "The city shall encourage preservation of units threatened

by expiring Section 8 contracts, condominium conversions, and foreclosures by working

with tenants, owners, and organizations who provide information about related issues"

(City of Orlando, 2010, p. H-5).

Pinellas County

Situated on Florida's Gulf Coast, Pinellas County occupies a peninsula whose

eastern shore constitutes the western boundary of Tampa Bay. Pinellas County,

together with Hillsborough, Hernando, and Pasco Counties comprise the Tampa-

St.Petersburg-Clearwater MSA. County properties in the AHI are located in the Cities of

Clearwater (pop. 105,774), Dunedin (pop. 35,988), Pinellas Park (pop. 47,173), Largo

(pop. 72,732), St. Petersburg (pop. 245, 314), South Pasadena (pop. 5,548), and


105









Tarpon Springs (pop. 23,359).37 Pinellas County is the 6th most populous in Florida with

a population of 938,459. Pinellas County has a land area of 279.92 square miles and

has a population density of 3,291 persons per square mile, making it significantly

denser than the other study areas. The county seat is Clearwater.

Demographics, Income, and Housing Affordability

While Pinellas County's population increased by 8% from 1990 to 2000, it has

dropped by 1.4% between 2000 and 2009. This loss is largely symptomatic of the

economic recession, and it suggests that the population projections of a year before

may be in need of revision. The estimated median age of the county (45.6), while

marginally lower than the state as a whole, is considerably higher than the other two

highly urbanized study areas (see Table 5-13). The share of households age 65 and

older is above 20%, and is projected to top 30% in 2030. In 2000, Pinellas County had

414,988 households, with an average of 2.17 persons per household.

Median household income was $45,650 for Pinellas County in 2008, which is

marginally lower than that for the state as a whole (see Table 5-14). Employment is

concentrated in the professional and business services; trade, transportation, and

utilities; and education and health services sectors (see Table 5-10). Pinellas County's

largest employers include Fidelity Information Services, the Home Shopping Network,

Nielsen Media Research, and Raymond James Financial.38

As in the other study areas, the housing market in Pinellas County rose

dramatically during the middle of the first decade of the twenty-first century and declined

37 Population estimates from U.S. Census Bureau, State & County QuickFacts,
http://quickfacts.census.gov/qfd/states/120001k.html
38 Pinellas County Economic Development, Pinellas County's Largest Private Employers,
http://www.pced.org/demographics_data/subpage.asp?TopEmployers


106









thereafter (see Table 5-21). As in Orange County, these increases significantly

outpaced increases in household income. As Figure 5-3 shows, the ratio of median

single-family home sales price to median household income rose as high as 5.0. As in

the other study areas, diminished home prices have not solved the county's affordable

housing problem. In 2008, 23% of renter households earning less 50% of AMI spent

more than 30% of their income on housing (see Table 5-15). The NLIHC's 2010 "Out of

Reach" report finds that the housing wage for a one-bedroom apartment in Pinellas

County is 110% of the mean renter wage and 210% of minimum wage. Unlike the other

study areas, growth in the number of severely cost burdened household is relatively

small. Following a decline between 2008 and 2010, the total number of severely cost

burdened households is projected to grow by only 676 through 2030 (see Table 5-12).

Assisted Housing Inventory

Pinellas County has the 6th largest number of assisted properties in the state with a total
of 101 properties in the AHI, and it has the 7th largest number of assisted units with a
total of 9,828 (see Table 5-17
Table). More than 50% of the county's assisted properties were funded by HUD

programs, while 33% were funded by FHFC (see Table 5-18). More than 70% of the

assisted units in the county are designated for renter households earning between 55

and 60% of AMI, but a larger share of units are designated for households earning 35%

or less of AMI than in other counties (see Table 5-19). In terms of age, assisted

properties are relatively younger on the whole in Pinellas County than in the other two

study areas, with nearly half of them being built or funded within the last 20 years (see

Table 5-20).


107









Local Housing Policy

As in the other two case study areas, Pinellas County is directly involved in the

preservation of affordable housing. Pinellas County annually allocates portions of its

federal and state housing assistance funds for preservation initiatives, including the

acquisition and rehabilitation of multifamily rental units. These preservation efforts take

place not only within unincorporated Pinellas County, but also within 20 participating

municipalities. In its 2009-2010 Action Plan, Pinellas County (2009) allocated $358,295

in HOME grants toward the preservation of affordable rental housing. Pinellas County

(n.d.) has also committed to allocating a share of a $17.8 million Neighborhood

Stabilization Program (NSP) grant toward the preservation of affordable rental housing.

Like Jacksonville, the City of St. Petersburg (2005) also has a Neighborhood

Revitalization Strategy Area (NRSA) for which it has requested federal and state

funding for a host of revitalization initiatives, including "the expansion and preservation

of affordable housing" (p. 151).


108












Florida Counties


Legend
Duval County
Orange County
Pinellas County


0 20 40 80 120 160
Miles


Figure 5-1. Study area counties


109














Florida's Assisted Housing


a.-


0 20 40 80 120 160
- -- Miles


Figure 5-2. Location of properties in Florida's Assisted Housing Inventory (AHI)


110









Table 5-1. Population projection (permanent residents) by age for 1990-2030, Duval
County
Age 0-19 20-64 65+ Total
1990 28.90% 60.42% 10.68% 672,514
2000 29.08% 60.41% 10.50% 778,446
2008 27.69% 61.33% 10.98% 904,407
2010 27.16% 61.36% 11.48% 916,938
2015 26.28% 60.28% 13.45% 974,938
2020 25.93% 58.34% 15.73% 1,040,436
2025 25.35% 56.34% 18.31% 1,104,138
2030 24.66% 54.72% 20.61% 1,164,443
Source: Florida Housing Data Clearinghouse, Population Projection by Age for 1990-
2030, http://flhousingdata.shimberg.ufl.edu/a/population

Table 5-2. Estimated employment by industry in Duval County, 2009
Industry Title Employment Share
Total, All Industries 498,848 100.00%
Natural Resources and Mining 372 0.07%
Construction 26,361 5.28%
Manufacturing 23,830 4.78%
Trade, Transportation, and Utilities 102,208 20.49%
Information 7,875 1.58%
Financial Activities 52,214 10.47%
Professional and Business Services 69,976 14.03%
Education and Health Services 65,326 13.10%
Leisure and Hospitality 41,764 8.37%
Other Services (Except Government) 18,890 3.79%
Government 51,566 10.34%
Source: Florida Agency for Workforce Innovation, Labor Market Statistics Center,
Employment Projections, http://www.labormarketinfo.com/Library/EP.htm

Table 5-3. Number of severely cost burdened renter households with income less than
80% AMI by tenure and income level, Duval County
Household Income 2008 2010 2015 2020 2025 2030 Percent
as% of AMI change
0-30% AM 14,119 14,373 15,334 16,455 17,608 18,733 33%
30.1-50% AMI 4,085 4,178 4,554 5,003 5,494 5,978 46%
50.1-80% AMI 851 876 994 1,135 1,301 1,463 72%
Total 19,055 19,427 20,882 22,593 24,403 26,174 37%
Source: Shimberg Center for Housing Studies, Affordable Housing Needs Summary,
http://flhousingdata.shimberg.ufl.edu/a/sum mary


111










Table 5-4. Growth in severely cost burdened renter households with income less than
80% AMI by tenure and income level, Duval County
Household Income 2008- 2010- 2015- 2020- 2025- Total
as% of AMI 2010 2015 2020 2025 2030
0-30% AM 254 961 1,121 1,153 1,125 4,614
30.1-50% AMI 93 376 449 491 484 1,893
50.1-80% AMI 25 118 141 166 162 612
Total 372 1,455 1,711 1,810 1,771 7,119
Source: Shimberg Center for Housing Studies, Affordable Housing Needs Summary,
http://flhousingdata.shimberg.ufl.edu/a/sum mary

Table 5-5. Population projection (permanent residents) by age for 1990-2030, Orange
County


Year 0-19 20-64 65+ Total
1990 27.44% 61.65% 10.91% 660,263
2000 28.38% 61.55% 10.07% 893,179
2008 28.91% 61.71% 9.37% 1,111,641
2010 28.59% 61.79% 9.63% 1,115,864
2015 27.96% 61.31% 10.73% 1,209,460
2020 27.84% 59.98% 12.18% 1,321,157
2025 27.50% 58.56% 13.94% 1,429,853
2030 26.93% 57.49% 15.58% 1,531,645
Source: Florida Housing Data Clearinghouse, Population Projection by Age for 1990-
2030, http://flhousingdata.shimberg.ufl.edu/a/population


Table 5-6. Estimated employment by industry in Orange County, 2009
Industry Title Employment Share
Total, All Industries 761,055 100.00%
Natural Resources and Mining 3,599 0.47%
Construction 31,833 4.18%
Manufacturing 26,426 3.47%
Trade, Transportation, and Utilities 121,878 16.01%
Information 16,234 2.13%
Financial Activities 40,942 5.38%
Professional and Business Services 126,877 16.67%
Education and Health Services 75,507 9.92%
Leisure and Hospitality 145,290 19.09%
Other Services (Except Government) 36,547 4.80%
Government 71,623 9.41%
Source: Florida Agency for Workforce Innovation, Labor Market Statistics (LMS) Center,
Employment Projections, http://www.labormarketinfo.com/Library/EP.htm


112










Table 5-7. Number of severely cost burdened renter households with income less than
80% AMI by tenure and income level, Orange County
Household Income 2008 2010 2015 2020 2025 2030 Percent
as% of AMI change
0-30% AM 16,611 16,724 18,085 19,718 21,381 22,980 38%
30.1-50% AMI 9,402 9,448 10,207 11,108 12,031 12,925 37%
50.1-80% AMI 2,193 2,211 2,401 2,634 2,878 3,114 42%
Total 28,206 28,383 30,693 33,460 36,290 39,019 38%
Source: Shimberg Center for Housing Studies, Affordable Housing Needs Summary,
http://flhousingdata.shimberg.ufl.edu/a/sum mary

Table 5-8. Growth in severely cost burdened renter households with income less than
80% AMI by tenure and income level, Orange County
Household Income 2008- 2010- 2015- 2020- 2025- Total
as% of AMI 2010 2015 2020 2025 2030
0-30% AM 113 1,361 1,633 1,663 1,599 6,369
30.1-50% AMI 46 759 901 923 894 3,523
50.1-80% AMI 18 190 233 244 236 921
Total 177 2,310 2,767 2,830 2,729 10,813
Source: Shimberg Center for Housing Studies, Affordable Housing Needs Summary,
http://flhousingdata.shimberg.ufl.edu/a/summary

Table 5-9. Population projection (permanent residents) by age for 1990-2030, Pinellas
County
Age 0-19 20-64 65+ Total
1990 19.90% 54.02% 26.08% 849,987
2000 21.21% 56.24% 22.55% 920,310
2008 20.86% 57.71% 21.43% 937,465
2010 20.45% 57.75% 21.81% 928,299
2015 19.62% 56.51% 23.87% 932,096
2020 19.28% 54.35% 26.37% 936,099
2025 18.88% 51.88% 29.24% 940,112
2030 18.50% 49.92% 31.58% 943,908
Source: Florida Housing Data Clearinghouse, Population Projection by Age for 1990-
2030, http://flhousingdata.shimberg.ufl.edu/a/population


113










Table 5-10. Employment by industry in Pinellas County, 2009
Industry Title Employment Share
Total, All Industries 453,449 100.00%
Natural Resources and Mining 161 0.04%
Construction 19,464 4.29%
Manufacturing 34,479 7.60%
Trade, Transportation, and Utilities 70,815 15.62%
Information 8,231 1.82%
Financial Activities 29,431 6.49%
Professional and Business Services 79,280 17.48%
Education and Health Services 70,626 15.58%
Leisure and Hospitality 42,990 9.48%
Other Services (Except Government) 17,583 3.88%
Government 45,742 10.09%
Source: Florida Agency for Workforce Innovation, Labor Market Statistics Center,
Employment Projections, http://www.labormarketinfo.com/Library/EP.htm

Table 5-11. Number of severely cost burdened households with income less than 80%
AMI by tenure and income level, Pinellas County
Household Income 2008 2010 2015 2020 2025 2030 Percent
as% of AMI change
0-30% AM 12,222 12,130 12,139 12,141 12,147 12,159 -1%
30.1-50% AMI 6,315 6,281 6,406 6,564 6,734 6,876 9%
50.1-80% AMI 1,508 1,503 1,541 1,589 1,639 1,686 12%
Total 20,045 19,914 20,086 20,294 20,520 20,721 3%
Source: Shimberg Center for Housing Studies, Affordable Housing Needs Summary,
http://flhousingdata.shimberg.ufl.edu/a/sum mary

Table 5-12. Growth in severely cost burdened households with income less than 80%
AMI by tenure and income level, Pinellas County
Household Income 2008- 2010- 2015- 2020- 2025- Total
as% of AMI 2010 2015 2020 2025 2030
0-30% AM -92 9 2 6 12 -63
30.1-50% AMI -34 125 158 170 142 561
50.1-80% AMI -5 38 48 50 47 178
Total -131 172 208 226 201 676
Source: Shimberg Center for Housing Studies, Affordable Housing Needs Summary,
http://flhousingdata. shimberg. ufl.edu/a/summary


114










Table 5-13. Median age estimates and projections, 2006-2030
Place 2006 2010 2020 2030
Florida 39.9 40.5 42.3 44.4
Duval 35.2 35.9 37.9 40.3
Orange 33 33.1 34.6 36.4
Pinellas 44.5 45.6 47.3 48.2
Source: Bureau of Economic and Business Research, Florida Statistical Abstract, 2007

Table 5-14. Median household income, 1989, 2008
Year Duval County Orange County Pinellas County Florida
1989 28,513 30,252 26,296 27,483
1999 40,703 41,311 37,111 38,819
2008 50,660 50,674 45,650 45,899
Source: U.S. Census Bureau, 1990 & 2000 SF3 Files, State and County QuickFacts.

6.0 1


---Jacksonville
--Orlando
--Tampa-St. Pete-Clearwater
-*-United States


U .U i I I I I I
0 0) N) '0) 0)) 10)) 0) 0 00 00000000
0)0)0)00 00000000000000
T-- T-- T- T-- T-- T-- T-- T- T- T- T- (" (" (" ( (" (" (N (N (N (

Figure 5-3. Ratio of median existing single-family house prices to median household
incomes by metropolitan area, 1989-2009. Source: Joint Center for Housing
Studies, 2009


115


-,WA










Table 5-15. Renter households with cost burden above 30% and income below 50%
AMI, 2008
Place Households %of all households Rank by number of
households
Duval County 31,149 24.1 6
Orange County 39,373 23.9 3
Pinellas 28,799 23.1 7
Source: Florida Housing Data Clearinghouse, Affordable Housing Needs

Table 5-16. Extremely low-income (<30 AMI), severely cost-burdened households, 2008
Place Households %of all households Rank by number of
households
Duval County 22,856 6.3 6
Orange County 24,350 5.8 5
Pinellas 22,484 5.3 7
Source: Florida Housing Data Clearinghouse, Affordable Housing Needs

Table 5-17. Total properties, units, and assisted units in Assisted Housing Inventory by
county, 2010
County Number of Properties Total Units Assisted Units
Duval 139 22,544 21,544
Orange 170 32,565 31,181
Pinellas 101 11,721 9,828
Source: Florida Housing Data Clearinghouse, Assisted Housing Inventory

Table 5-18. Properties and assisted units by founder (duplicated count), 2010
Duval Orange Pinellas
Funder Properties Units Properties Units Properties Units
HUD 70 9,041 41 5,552 61 4,648
RD 3 139 9 411 0 0
FHFC 68 12,523 114 24,276 37 4,445
LHFA 19 4,282 39 7,250 15 2,428
TOTAL 160 25,985 203 37,489 113 11,521
Source: Florida Housing Data Clearinghouse, Assisted Housing Inventory

Table 5-19. Assisted units by income limits, 2010
Number of Units Designated for Renters with Income:
County <=35% AMI 40-50% AMI 55-60% AMI 65-80% AMI >80% AMI
Duval 336 2,409 8,027 510 885
Orange 732 1,492 19,231 348 2,092
Pinellas 330 441 2,265 104 0
Source: Florida Housing Data Clearinghouse, Assisted Housing Inventory


116











Table 5-20. Assisted housing properties and units by age of property or year funded
0-10 Years 11-20 Years 21-30 Years > 30 Years Not available
County Prop. Units Prop. Units Prop. Units Prop. Units Prop. Units
Duval 9 1,939 6 463 24 3,559 41 5,929 59 9,654
Orange 24 3,994 12 2,052 23 3,419 17 2,810 94 18,956
Pinellas 15 916 17 444 21 1,547 12 2,371 36 4,550
Source: Florida Housing Data Clearinghouse, Assisted Housing Inventory


Table 5-21. Median sales price for single-family homes (in thousands of dollars), 1995-2009
Jacksonville Orlando- Tampa-St. Pete- Florida
MSA Kissimmee MSA Clearwater MSA
Year Median Change Median Change Median Change Median Change
sales sales sales sales
price price price price
1995 83.9 1% 86.1 -2% 78 4% 87.7 2%
1996 90 7% 90.7 5% 80.9 4% 92.3 5%
1997 88.6 -2% 94.4 4% 84 4% 95.8 4%
1998 97.3 10% 96 2% 88 5% 99.8 4%
1999 97.2 0% 102.8 7% 94.7 8% 108.4 9%
2000 103.9 7% 109.3 6% 105.8 12% 117.6 8%
2001 116.7 12% 120.1 10% 124.4 18% 126.6 8%
2002 124.4 7% 130.3 8% 133.3 7% 141.7 12%
2003 138.2 11% 143.5 10% 139.3 5% 155.8 10%
2004 158.6 15% 164.5 15% 159.9 15% 181.9 17%
2005 187.3 18% 231.4 41% 201.7 26% 235.2 29%
2006 200.6 7% 262.9 14% 224.8 11% 247.1 5%
2007 196.5 -2% 248.9 -5% 208.9 -7% 234.3 -5%
2008 180.4 -8% 201.5 -19% 169.5 -19% 187.7 -20%
2009 152.2 -16% 144.6 -28% 137.5 -19% 142.6 -24%
Source: Florida Association of Realtors. Places are metropolitan statistical areas (MSAs).









CHAPTER 6
RESULTS FROM TESTS OF DIFFERENCES BETWEEN ASSISTED HOUSING
GENERATIONS

In order to understand the relative suitability of properties in the assisted housing

inventory based on the generation of housing production to which they belong, the

Mann-Whitney U test was conducted to compare these groups based on key selected

criteria. This chapter examines the results of these tests in order to better understand

strategies for assisted housing preservation in Florida. This chapter begins by

presenting the results from the statistical analysis of suitability values for properties

belonging to the different generations. Results are presented for the two tests

performed for each county, the first being a comparison of properties either built or

funded between 1963 and 1979 with those either built or funded between 1980 and

1994 (HUD versus older-FHFC properties), the second being a comparison of

properties either built or funded between 1963 and 1994 with those either built or

funded between 1995 and 2008 (HUD and older-FHFC versus LIHTC properties).

Duval County

The results of the Mann-Whitney U test for differences between assisted

properties built/funded between 1963 and 1979 (proxy for HUD-financed properties) and

those built/funded between 1980 and 1994 (proxy for older FHFC-financed properties)

in Duval County are presented in Table 6-1. The former of the two groups has a sample

size of 43 and the latter a sample size of 40. The results of the Mann-Whitney U test

show a statistically significant difference (at a significance level of p=.05) between the

two groups for the suitability variables crime risk, FCAT scores, affordable housing

preference, poverty rate, and transportation. This finding indicates that older FHFC-

financed properties are in areas with significantly higher suitability values for crime risk,


118









FCAT scores, affordable housing preference, and poverty rates, but significantly lower

suitability related to transportation suitability values.39

These differences show that older FHFC-financed properties are located in areas

with significantly better neighborhood conditions than those developed under HUD

programs, but that these advantages potentially come at the expense of location

efficiency. The median poverty rate suitability value for the 1980-1994 group of

properties (Mdn=6) is substantially higher than the median value for the 1963-1979

group (Mdn=1), which would seem to strongly recommend the younger properties for

preservation in the context of a policy preference for poverty deconcentration.40 The test

showed no statistical difference between the two groups for the variables of

neighborhood decline and gentrification, indicating that properties in one group are no

more likely than those in the other to be found in areas marked by neighborhood

change. Finally, the test yielded no significant difference between the groups for local

accessibility. As the groups are similarly situated in terms of local accessibility, the

higher suitability values for variables indicating positive neighborhood conditions

exhibited by the older FHFC-financed properties strongly recommends their

consideration for prioritization in Duval County. The higher transportation values found

in areas in which HUF-financed properties are located suggest that they may promote

enhanced mobility for their residents, but without proximity to destinations and transit


39 Suitability values are reclassified so high numbers reflect greater suitability for housing. While high
actual crime risk and poverty rate suggests low suitability for housing, the suitability values for these
variables move in the opposite direction so as actual crime risk rises, suitability values for this variable
decline. In the case of variables where actual measures are directly related to suitability, they are
reclassified on the 1-9 scale, but in the same direction. That is, high measured accessibility means high
suitability accessibility.
40 As explained in the methodology, higher suitability values for the poverty rate variable indicate lower
actual poverty rate.


119









opportunities-key components of the local accessibility variable-it is perhaps not

enough to tip the balance in favor of HUD-financed properties based on this analysis.41

The results of the Mann-Whitney U test for differences between assisted

properties built/funded between 1963 and 1994 and those built/funded between 1995

and 2008 in Duval County are presented in Table 6-2. The former of the two groups has

a sample size of 85 and the latter a sample size of 51. The results of the Mann-Whitney

U test show a statistically significant difference between the groups for the variables of

decline, FCAT scores, gentrification, local accessibility, poverty rate, and transportation.

The younger group of properties (1995-2008) has significantly higher suitability values

for neighborhood decline, FCAT scores, and poverty rate, but significantly lower

suitability values for gentrification, local accessibility, and transportation. While the

median values are identical for the two groups for the transportation variable, the mean

ranks were considerably different, with the group 1963-1994 having a mean rank of

74.55 and 1995-2008 having a mean rank of 56.02.

The results of this test display an interesting and seemingly incongruous set of

relationships for the younger group. While they are better situated with regard to poverty

rate and FCAT scores, both signs of positive neighborhood quality, they are

simultaneously less well situated with regard to neighborhood decline (high

suitability=greater decline). The older of the two groups has significantly higher values

for gentrification, meaning that its properties are statistically more likely to be in areas

experiencing upward market pressure than those in the younger group. This, in

combination with higher local accessibility and transportation values suggests that these

41 Transportation scholars generally distinguish between accessibility and mobility by defining
accessibility as the "ease of reaching" as mobility as the "ease of moving" (Preston & Raj6, 2007, p. 154).


120









older properties are more likely to be in desirable urban neighborhoods at risk of

becoming unaffordable to lower-income households. In the context of Duval County's

stated objective of preserving affordable housing in gentrifying neighborhoods, the

older, HUD-financed properties seem to be of higher value to the affordable housing

stock. There is no statistical difference between the two groups for the variables crime

risk and affordable housing preference. This finding suggests that the overall tilt in favor

of the combined HUD- and older FHFC-financed properties is even stronger than that

exhibited in the first test (between HUD- and older FHFC-financed properties). Indeed,

the absence of difference between the groups for the affordable housing preference

variable does not mean that these two groups should be equally prioritized within the

AHI. The other, disaggregated variables included in the analysis (local accessibility,

gentrification, decline, poverty rate, school quality) are better able to shed light on the

manner in which these groups of properties score based on different dimensions of

suitability.

Orange County

The results of the Mann-Whitney U test for differences between HUD-financed

properties (those built/funded between 1963 and 1979) and older FHFC-financed

properties (those built/funded between 1980 and 1994) in Orange County are presented

in Table 6-3. The former of the two groups has a sample size of 16 and the latter a

sample size of 54. The results of the Mann-Whitney U test show a statistically significant

difference (at a significance level of p=.05) between the suitability variables of crime,

decline, gentrification, and local accessibility. This finding indicates that the younger

properties (1980-1994) are in areas with significantly higher suitability values for crime

risk and decline, and significantly lower values for gentrification and local accessibility.


121









There is no statistical difference between the two groups for the variables FCAT scores,

affordable housing preference, poverty rate, and transportation.

The combination of higher suitability for areas with the older FHFC-financed

properties related to both crime risk (lower measured crime risk) and neighborhood

decline (higher measured decline) is interesting, as these variables are generally

inversely related. This finding could be fortuitous in the context of preservation, as

investment in these properties could serve as a catalyst for neighborhood revitalization

in areas that seem particularly well-suited to take to such efforts based on lower crime

rates. However, the significantly higher suitability in areas with HUD-financed properties

related to gentrification and local accessibility suggest that they are better candidates

for prioritization in the AHI than the older FHFC-financed properties. This judgment is

informed by the knowledge that local accessibility is generally difficult to adjust for a

particular location (it involves numerous dimensions of the built environment), meaning

that it is a particularly important asset. In addition, the potential presence of

gentrification suggests that a limited window exists for the continued affordability of the

neighborhood and that these properties may be located in desirable, amenity-rich areas.

The results of the Mann-Whitney U test for differences between the combined

group of HUD- and older FHFC-financed properties (those built/funded between 1963

and 1994) and newer, predominantly LIHTC properties (those built/funded between

1995 and 2008) in Orange County are presented in Table 6-4. The former of the two

groups has a sample size of 75 and the latter a sample size of 79. The results of the

Mann-Whitney U test show a statistically significant difference between the groups for

the suitability variables of FCAT scores and decline. Areas where LIHTC properties are


122









located have significantly higher suitability values for decline, but significantly lower

values for FCAT scores, meaning that they are more likely to be situated in declining

areas with lower school quality.42 This result is noteworthy as it is incongruent with the

results of previous research finding that LIHTC properties are generally better situated

with regard to socioeconomic indicators than housing developed under other assistance

programs (Freeman, 2004; Oakley, 2008). Furthermore, as LIHTC properties are more

likely to be located in suburban areas than HUD-financed properties (Freeman, 2004),

these results suggest that LIHTC properties in Orange County may be located in

struggling inner-ring suburbs, not in the generally more prosperous newer suburbs.

There is no statistical difference between the two groups for the variables of crime

risk, gentrification, affordable housing preference, local accessibility, poverty rate, and

transportation. These results suggest that the two groups are not all that different,

making a clear decision about which should be prioritized in the AHI difficult. The higher

suitability values for decline found in the younger groups of properties may actually be

read as a negative attribute depending on the policy lens applied in making prioritization

decisions. Local policies related to affordable housing in some localities, such as

Duval's prioritization of preservation activities located in Neighborhood Action Plan

(NAP) areas, express a clear interest in undertaking preservation initiatives in

distressed areas as part of a larger neighborhood revitalization scheme. While Orange

County and the City of Orlando have expressed an interest in affordable housing

preservation (including assisted housing), their policies do not suggest that location in a


42 While the median values are identical for the groups on the FCAT score variable, the mean rank for the
group 1963-1995 is 84.95 while the mean rank for the group 1995-2008 is 70.43, meaning that the former
group has significantly higher suitability values.


123









declining area automatically makes assisted housing preferable for preservation. All

other things being equal, the older properties are significantly different since they are

more likely to be located in areas not experiencing decline, meaning that they should

have better neighborhood quality than those properties produced after the generational

shift.

Pinellas County

The results of the Mann-Whitney U test for differences between HUD-financed

properties (those built/funded between 1963 and 1979) and older FHFC-financed

properties (those built/funded between 1980 and 1994) in Pinellas County are

compared in Table 6-5. The former of the two groups has a sample size of 13 and the

latter a sample size of 29. The small sample size of these groups informed the decision

to use the Mann-Whitney U test rather than the parametric student's t-test. The results

of the Mann-Whitney U test show a statistically significant difference (at a significance

level of p=.05) between the two groups for none of the suitability variables. Similarly, no

statistically significant difference between assisted properties built/funded between 1963

and 1994 and those built/funded between 1995 and 2009 exists (see Table 6-6). In

terms of median suitability values, crime risk and local accessibility have relatively high

values for all four age categories. This finding suggests that assisted housing in Pinellas

may be well situated in terms of producing positive individual outcomes for residents,

regardless of funding program. In and of themselves, these results suggest that there is

no meaningful difference between the properties developed during the different

generations of housing production. This is most likely explained by the fact that more of

the county's assisted properties were developed recently, especially in comparison to

the other two study areas (see Table). Only 13 properties were developed during the


124









1963-1979 period, when the federal government was primarily responsible for housing

production. Therefore, no one group of properties in Pinellas County may be privileged

for preservation or given priority in the AHI based on this study.

Summary

The results of these tests yield considerably different results in each of the study

areas. In the comparison of HUD-financed properties with older FHFC-financed

properties, FHFC-financed properties appear to be of greater value in Duval County,

HUD-financed properties of greater value in Orange County, and neither group of

properties appears of greater value in Pinellas County. The results of the test comparing

both HUD- and older-FHFC financed properties with properties developed after

Congress made the LIHTC program permanent yield less definitive results, but the

degree of difference depends on the study area. In Duval County, the older group of

properties is clearly of greater value to the affordable housing stock than the newer

properties; in Orange County the results are less conclusive, with the older properties

appearing somewhat superior; and in Pinellas County, again, no statistically significant

difference was observed. The next chapter will discuss these results in the wider

context of preservation strategies and housing policy, and will relate the results of the

tests back to the preservation literature.


125












Duval County, Florida

AHI Generation (Test 1)
1963-1979
0 1980-1994
0 1995-2008 [excluded]
Water
Major Highways














0 1 2 4 6 a

Figure 6-1. Properties in the assisted housing inventory categorized by age for test
between HUD- and older FHFC-financed properties, Duval County


Table 6-1. Comparison of assisted properties built/placed in service 1963-1979 and
1980-1994, Duval County
Age Range
1963-1979 1980-1994
N=43 N=40 Mann-Whitney U test
N=43 N=40
Variable Median Median Z P(2-tailed)
Crime 7.20 9.00 -2.474 .013
Decline 1.00 1.33 -1.680 .093
FCAT 4.33 5.00 -2.299 .022
Gentrification 1.34 1.10 -1.351 .177
Affordable housing 4.95 5.39 -3.041 .002
preference
Local accessibility 7.33 6.78 -1.631 .103
Poverty rate 1.00 6.00 -3.213 .001
Transportation 9.00 8.50 -2.347 .019


126












Duval County, Florida


AHI Generation (Test 2)
1963-1994
0 1995-2008
Water
Major Highways














0 1 2 4 6 a

Figure 6-2. Properties in the assisted housing inventory categorized by age for test
between combined HUD- and older FHFC-financed properties and LIHTC
properties, Duval County


Table 6-2. Comparison of assisted properties built/placed in service 1963-1994 and
1994-2008, Duval County
Age Range
1963-1994 1994-2008
Mann-Whitney U test
N=85 N=51
Variable Median Median Z P(2-tailed)
Crime 8.20 9.00 -1.551 .121
Decline 1.00 1.83 -3.174 .002
FCAT 4.33 5.67 -2.942 .003
Gentrification 1.10 1.00 -2.407 .016
Affordable housing 5.04 5.25 -.730 .466
preference
Local accessibility 7.29 6.45 -3.529 .000
Poverty rate 2.00 8.00 -4.246 .000
Transportation 8.50 8.50 -2.786 .005


127










Orange County, Florida


AHI Generation (Test 1)
* 1963-1979
o 1980-1994
* 1995-2008 [excluded]
- Major Highways
- Waterbodies


U 3 6 12


Figure 6-3. Properties in the assisted housing inventory categorized by age for test
between HUD- and older FHFC-financed properties, Orange County


Table 6-3. Comparison of assisted properties built/placed in service 1963-1979 and
1980-1994, Orange County
Age Range
1963-1979 1980-1994 Mann- ney U test
N=16 N=54 Mann-Whitney U test
N=16 N=54
Variable Median Median Z P(2-tailed)
Crime 3.30 9.00 -1.981 .048
Decline 1.00 1.92 -1.993 .046
FCAT 7.33 6.33 -.392 .695
Gentrification 2.10 1.28 -2.473 .013
Affordable housing 5.25 4.96 -1.063 .288
preference
Local accessibility 7.26 6.23 -3.505 .000
Poverty rate 7.00 6.00 -.937 .349
Transportation 9.00 9.00 -1.116 .264


128










Orange County, Florida


AHI Generation (Test 2)
* 1963-1994
o 1995-2008
- Major Highways
1 Waterbodies


O 3 6 12


Figure 6-4. Properties in the assisted housing inventory categorized by age for test
between combined HUD- and older FHFC-financed properties and LIHTC
properties, Orange County


Table 6-4. Comparison of assisted properties built/placed in service 1963-1994 and
1994-2008, Orange County
Age Range
1963-1979 1980-1994 Many U
N=75 N=79 Mann-Whitney U test
N=75 N=79
Variable Median Median Z P(2-tailed)
Crime 9.00 9.00 -1.592 .111
Decline 1.58 2.17 -3.667 .000
FCAT 6.33 6.33 -2.034 .042
Gentrification 1.45 1.26 -1.624 .104
Affordable housing 4.98 4.96 -.330 .741
preference
Local accessibility 6.55 6.31 -1.520 .128
Poverty rate 6.00 6.00 -.739 .460
Transportation 9.00 8.50 -.579 .563


129










Pinellas County, Florida


AHI Generation (Test 1)
* 1963-1979
0 1980-1994
* 1995-2008 [excluded]
Waterbodies
- Major Highways


Figure 6-5. Properties in the assisted housing inventory categorized by age for test
between HUD- and older FHFC-financed properties, Pinellas County


130









Table 6-5. Comparison of assisted properties built/placed in service 1963-1979 and
1980-1994, Pinellas County
Age Range
1963-1979 1980-1994
N=13 N=29 Mann-Whitney U test
Variable Median Median Z P(2-tailed)
Crime 6.40 7.60 -.994 .345
Decline 1.00 1.00 -1.153 .249
FCAT 6.33 6.33 -.863 .388
Gentrification 1.54 1.11 -.616 .538
Affordable housing 5.17 4.91 -.986 .324
preference
Local accessibility 7.85 7.26 -1.388 .165
Poverty rate 3.00 2.00 -.597 .572

Table 6-6. Comparison of assisted properties built/placed in service 1963-1994 and
1994-2008, Pinellas County
Age Range
1963-1979 1980-1994
3-1979 -1994 Mann-Whitney U test
N=43 N=37
Variable Median Median Z P(2-tailed)
Crime 7.40 8.20 -1.645 1.00
Decline 1.00 1.00 -.286 .775
FCAT 6.33 5.67 -.518 .604
Gentrification 1.15 1.00 -1.179 .238
Affordable housing 4.92 5.00 -.163 .870
preference
Local accessibility 7.52 7.23 -1.650 .099
Poverty rate 3.00 5.00 -1.165 .244


131










Pinellas County, Florida


AHI Generation (Test 2)
* 1963-1994
o 1995-2008
Waterbodies
Major Highways


Figure 6-6. Properties in the assisted housing inventory categorized by age for test
between combined HUD- and older FHFC-financed properties and LIHTC
properties, Pinellas County


132









CHAPTER 7
CONCLUSIONS AND RECOMMENDATIONS

In making determinations about the value of assisted housing to the affordable

housing stock, neighborhood quality and location efficiency need to be taken into

consideration (Newman & Schnare, 1997; Holtzclaw et al., 2002). Indeed, as Briggs

(2005b) reminds us: "Location matters-for economic returns, quality of life, and many

other reasons" (p. 17). The discourse of preservation has continued to prioritize HUD-

financed properties above all others based on their deeper subsidy levels, but without

fully considering the geography of opportunity in which they are situated. Scarce

resources could be inefficiently allocated toward properties that exact unforeseen costs

on their residents. While younger properties may not offer the same level of subsidy as

offered through HUD assistance programs, the benefits accrued from residing in a

higher-quality area can be of even greater value. As discussed in Chapter 3,

neighborhood characteristics can affect individual outcomes along a number of

dimensions, including access to employment, educational attainment, and physical and

mental health. Living in an isolated area that requires significant transportation

expenditures can quickly erode the benefits gained from living in a residence with

assisted rent. Furthermore, properties in the AHI need to be evaluated based on their

ability to offer affordable housing in neighborhoods in which such housing is in short

supply or in danger of disappearance (Kennedy & Leonard, 2001). This chapter

evaluates the results of the statistical analyses, placing them in the context of the

geography of opportunity and research questions motivating this study, and offers

recommendations for preservation initiatives and future research based on these

conclusions.


133









The results of this study show that generational cohorts in the AHI do not always

perform consistently, either within or between study areas. This pattern of incoherence

is of great consequence, as it seems to refute the hypothesis stated at the outset of this

study that the locational characteristics of assisted housing will differ in predictable ways

based on the generation of housing production to which they belong. This is evidenced

in the internal inconsistencies wherein one cohort may have significantly higher or lower

values along one dimension of suitability, but also have significantly different values

along a different, contrasting dimension. This tendency is manifest in the results of the

test for differences in Duval County between HUD and older FHFC-financed properties

(1963-1994) on the one hand, and newer, predominantly LIHTC-financed properties

(1995-2008) on the other. In this instance, the newer properties have significantly higher

values for variables reflecting school quality (FCAT scores) and socioeconomic status

(poverty rate), while simultaneously having significantly higher suitability values for the

variable of neighborhood decline. Between counties, different generational cohorts

exhibit significantly higher or lower values for different variables. That is, while areas in

which properties developed under a specific program may be significantly more suitable

along a particular dimension than areas in which properties developed under another

program are located in one of the study areas, this superiority is not always replicated in

the others. For each test, there is only one variable for which the younger cohorts

possess significantly higher suitability values in more than one county. For example,

while the areas in which older FHFC-financed properties are located have significantly

higher suitability values for school quality than HUD-financed properties in Duval


134









County, they do not exhibit this relatively superior school performance over HUD-

financed properties in the other two counties.

Despite the variance of these results, they do provide some support for the

hypothesis that older properties are more likely to be located in the central city, while

newer units are more likely to be located on the outskirts. Indeed, if any predictable

pattern may be found in the results, it is that younger, predominantly state-financed

properties are generally located in areas possessing relatively higher neighborhood

quality, while older, predominantly federally-financed properties are generally located in

areas with relatively greater location efficiency. This distinction appears in the results of

both tests, that is, not only does it exist between age cohorts representing the larger

sweep of the generational shift (HUD and older-FHFC versus predominantly LIHTC

developments), but it also exists between groups specifically representing potentially at-

risk older properties (HUD versus older FHFC-financed properties). Younger cohorts in

both tests generally have significantly higher suitability values for variables measuring

neighborhood quality (i.e., crime risk, FCAT scores, poverty rate), while older cohorts in

both tests generally have significantly higher suitability values for variables measuring

location efficiency (i.e., local accessibility, transportation cost) (see Table 7-1). An

additional similarity found among study areas is the significantly higher suitability values

for gentrification for older cohorts. This appears for the HUD-financed cohort (1963-

1979) in the test between it and older FHFC-financed properties (1980-1994) in Orange

County, as well as for the combined HUD- and older FHFC-financed properties (1963-

1994) in the test between them and the cohort representing predominantly LIHTC

developments (1995-2008) in Duval County.


135









In essence, the prioritization of one cohort over another for preservation seems to

represent a trade-off between neighborhood quality and location efficiency, two

important dimensions shaping the overall geography of opportunity. While neighborhood

quality can have a significant impact on the quality-of-life for individuals, potentially

offering an environment with less environmental stressors and greater access to local

social networks, location efficiency is critical for many to access services and

employment. However, the degree of this trade-off appears to be variable. The

difference between median poverty rate suitability values (high suitability=low poverty)

for HUD-financed properties (Mdn=1) and older FHFC-financed properties (Mdn=6) in

Duval County is quite stark, but it is altogether absent from Orange and Pinellas

Counties. On the other hand, in both Duval and Orange Counties, older FHFC-financed

properties have significantly higher suitability values for crime risk than HUD-assisted

properties, suggesting that the trade-off between cohorts may be further distilled into

one between safety and accessibility. In Duval County, though, the aggregate suitability

of locations in which older FHFC-financed properties are located makes the trade-off

between location efficiency and neighborhood quality considerably more pronounced

than it is in Orange, not to mention Pinellas County. In the test of differences between

HUD- and older FHFC-financed properties, the latter has significantly higher suitability

values for four indicators of neighborhood quality, including overall affordable housing

preference, while the former has significantly higher transportation values.

The distinct historical patterns of development of these study areas, specifically

their principle cities, seem to account, at least in part, for both the consistencies and

inconsistencies observed in the results. While this study has heretofore discussed its


136









study areas solely in terms of county boundaries, the trajectories of the cities contained

within them may shed light on the differences between the results for Duval County and

the results for Orange and Pinellas Counties. As previously mentioned, Duval County

and the City of Jacksonville were consolidated in 1968, eliminating competing

governmental bodies and significantly expanding Jacksonville's city limits. It is to this

geographic expansion that Jacksonville owes its status as the largest city in Florida,

both in terms of land mass and population. In 2006, Jacksonville's population was

794,255, while Orlando's was only 220,186, and St. Petersburg's was only 248,098.43

David Rusk (2003) refers to this type of boundary expansion as an expression of

municipal "elasticity" (p. 12). An important dimension of elasticity is the "central city's

ability to exercise influence and control over the metropolitan area" (Smith et al., 1997,

p. 65). Jacksonville's high elasticity allowed it to capture its suburban growth and

prevent metropolitan governmental fragmentation, which was a genuine concern for the

city's leaders prior to consolidation (Crooks, 2001).

It seems likely that this metropolitan-wide sphere of influence exerted by

Jacksonville may also have influenced the spatial pattern of assisted housing

development in Duval County. As discussed in Chapter 5, Jacksonville has continued to

promote the development of affordable housing in its Neighborhood Action Plan (NAP)

areas, which are distressed neighborhoods located in or near its historic city limits. As

Jacksonville has not needed to compete with neighboring municipalities for resources

needed for assisted housing production, the spatial pattern of assisted housing in Duval

County is far more tightly knit than it is in Orange and Pinellas Counties. Orlando and

43 These figures are for cities, not metropolitan statistical areas. Population estimates from U.S. Census
Bureau, State & County QuickFacts, http://quickfacts.census.gov/qfd/states/120001k.html


137









St. Petersburg, the primary cities of Orange and Pinellas Counties, respectively, have

not been able to manifest nearly the same degree of gravitational pull on the location of

assisted housing units. Both of these counties contain numerous municipalities engaged

in assisted housing development-Orange County contains 15 incorporated

municipalities and Pinellas County contains 24-making the spatial pattern of assisted

housing appear more diffuse when viewed at the county scale. It should be noted that

the governmental fragmentation of Pinellas County is a clear reflection of its geographic

fragmentation (it is a relatively small peninsula with a number of barrier islands), which

likely also accounts for the failure of its assisted properties to exhibit a coherent shift in

locational attributes.

Policy Recommendations

As shown in this study, the decision to preserve one generation of assisted

housing over another reflects a trade-off between neighborhood quality and location

efficiency. Therefore, the decision to prioritize one cohort over another should be

decided by policy preferences and local housing conditions. The policy focus in many

urban areas is the deconcentration of poverty. In fact, the concentration of poverty is

"among the most urgent challenges confronting urban policy makers today" (Turner,

1998, p. 373). Duval County, specifically the City of Jacksonville, considers poverty

deconcentration an affordable housing goal due to the pronounced patterns of socio-

economic and minority segregation apparent in the city (Jacksonville, 2005).

Jacksonville may then want to consider prioritizing the preservation of older FHFC-

financed properties at-risk of failing out of the affordable housing inventory as they have

significantly higher suitability values for the poverty rate variable. On the other hand, the

city also holds a strong interest in preserving affordable housing opportunities in


138









gentrifying areas, so the preservation of the HUD-assisted stock may be a greater

priority to Jacksonville. Both Pinellas County and Jacksonville target resources toward

the preservation of affordable housing options as a redevelopment tool in distressed

neighborhoods. Given this policy climate, it may make even less sense for them to

prioritize younger properties.

Preserving assisted housing near transit and shopping can greatly improve the

geography of opportunity available to lower-income families and individuals. In order to

preserve the federally-assisted housing near transit, state and local housing entities

should pursue federal funding available through such programs as the joint sustainable

communities initiative for preservation resources, which promote the development of

walkable, transit-rich environments. In order to prepare for future funding applications,

state and local agencies should begin to collect data related to the location efficiency of

assisted housing, particularly those at-risk of failing out of the AHI. The Shimberg

Center for Housing Studies is especially positioned to conduct this research. A final

recommendation is for the inclusion of preservation activities for location efficient

assisted housing units in plans and policies addressing larger sustainable development

objectives such as transit-oriented development.

Future Research and Limitations

Future research could expand upon this study by conducting the statistical tests of

difference following the completion of the Affordable Housing Suitability (AHS) model.

This would allow for a more complete comparison of the counties based on

transportation suitability values and take into consideration the demand for housing

generated by commercial, industrial, and residential development. While local

accessibility is an importance measure of value for assisted housing, it does not provide


139









a robust measure for the relationship between employment demand and residence. The

inclusion of the linkage analysis would better enable the model to assess properties for

preservation based on the degree to which they address the job-housing balance. When

home and work are dislocated by lengthy distances, the added transportation

expenditures can be quite onerous. With the completion of the AHS mode, pairwise

comparison will have integrated community preferences into the suitability values

produced by the model. Thus, the scores produced by the completed model may vary

from those used in this study, which were produced by assigning equal weights to

layers during the weighted overlay procedures. Future studies could also compare a

greater number of counties to test for the consistency of this study's findings.


Table 7-1. Incidence of significantly higher suitability values
Test of differences between HUD- Test of differences between HUD-
and older FHFC-financed and FHFC-financed properties and
properties LIHTC properties
1963-1979 1980-1994 1963-1994 1995-2008
Duval Transportation Crime Gentrification Decline
County FCAT Local accessibility FCAT
AH preference Transportation Poverty rate
Poverty rate

Orange Gentrification Crime FCAT Decline
County Local Decline
accessibility

Pinellas
County


140









APPENDIX A
NEIGHBORHOOD CHANGE INDICATORS

Neighborhood Change as Gentrification


Long Term Change Indicators (Decennial census block groups)
Variable Explanation and Sources
Increase in this indicator may be indicative of
Non-Hispanic white displacement of minorities associated with gentrification.
c population, 1980-2000 (Freeman, 2005; Freeman & Braconi, 2004; Kennedy &
0 Leonard, 2001; Levy et al., 2006; Wyly & Hammel, 1999)
Young professionals and baby-boomers have been
identified in the literature as potential gentrifiers. Elements
125-34 yar age coh, of these cohorts may return to the city to take advantage
1980-2000
-6 of urban amenities (Hall & Ogden, 1992; Kennedy &
1980-2000 age Leonard, 2001; Lees, Slater, & Wyly, 2007; Levy et al.,
.c -2006; Ley, 1996; McKinnish, Walsh, & White, 2009; Wyly
0 & Hammel, 1999).
Gains in household income have been identified as
indicative of gentrification or decline depending on the
Median household degree of change. The influx of higher-income residents
Median household
is a key component of most definitions of gentrification
income, 1980-2000
(Freeman, 2005; Kennedy & Leonard, 2001; Lees et al.,
2007; Levy et al., 2006; Ley, 1996; Wyly & Hammel,
o 1999).
Gain indicates influx of the professional class, which has
Professional/white shown interest in urban amenities, possesses the wealth
ar emloyent to purchase and upgrade housing, participates in housing
collar employment
c1980-2000 speculation, and contributes to displacement pressures
._ (Lees et al., 2007; Ley, 1996; Smith, 1996; Wyly &
0 Hammel, 1999).
Increasing educational attainment indicates an influx of
High educational wealthy professional households, and is associated with
. attainment, 1980-2000 the cultural component of gentrification (Freeman, 2005;
0 Ley, 1996; Wyly & Hammel, 1999).
Areas with owner-occupied housing are typically wealthier
than those with predominantly renter-occupied housing.
Ownd Gains in owner-occupied units may indicate growing
investment in rehabilitation or new development and in
housing, 1980-2000
housing, 1980-2000 the case of a gentrifying area, increasing property values
(Lees et al., 2007; Levy et al., 2006; Wyly & Hammel,
o _1999).


141










Short Term Change indicators (Home Mortgage Disclosure Act [HMDA] data, County
property appraiser sales data)
Variable Explanation and Sources
An increase in loan origination signals increased
investment and is correlated with neighborhood change
Loan originations per
nini on n previously less-affluent urban neighborhoods
capital, 2 -2 (Hackworth, 2007; Lees et al., 2007; Pettit & Droesch,
0 2008).
Median household Recent gains in income may also indicate gentrification in
income of home some urban areas (Freeman, 2005; Kennedy & Leonard,
c purchase borrowers, 2001; Lees et al., 2007; Levy et al., 2006; Pettit &
0 2005-2007 Droesch, 2008; Wyly & Hammel, 1999).
Median home sales Sharply increasing property values, especially in lower
c price ($ per square income areas, may indicate a process of gentrification
0 foot), 2000-2008 (Turner & Snow, 2001).

Static Indicators of Gentrification potential
Variable Explanation and Sources
Historic district Historic district designation may raise property values
design n (l ad and create maintenance obligations proving onerous for
designation (local and
incumbent residents (Coulson & Leichenko, 2004; Turner
national & Snow, 2001).


142










Neighborhood Change as Decline


Long Term Change Indicators (Decennial census block groups)
Variable Explanation and Sources
Increase in minority population (loss of non-hispanic
Non-hi c we whites) suggests decline, as it captures processes of
Npon pn wt neighborhood succession and possibly, "tipping."
population, 1980-2000 (Freeman, 2005; Freeman & Braconi, 2004; Kennedy &
o Leonard, 2001; Levy et al., 2006; Wyly & Hammel, 1999)
.J
Sharp losses in neighborhood income are associated
Median household with decline (Freeman, 2005; Kennedy & Leonard, 2001;
Income, 1980-2000 Lees et al., 2007; Levy et al., 2006; Ley, 1996; Wyly &
__Hammel, 1999).
Professional/white
Professional/white Loss of professional class is indicative of decline (Lees et

collar eloy al., 2007; Ley, 1996; Smith, 1996; Wyly & Hammel, 1999)
0) 1980-2000
.J

High educational Decreasing educational attainment suggest
attainment, 1980-2000 neighborhood decline (Freeman, 2005; Ley, 1996; Smith,
attainment, 1980-2000 w
o 1996; Wyly & Hammel, 1999)

Ownd Decreases in home ownership are indicative of decline
Shouting, 1 0 (Lees et al., 2007; Levy et al., 2006; Wyly & Hammel,
housing, 1980-2000
o 1999)
An increase in the measure of this indicator is associated
Female-headed
CF helhd with decline (Beauregard, 1990; Galster, 2000; Newman
Households, 1980-2000
0 & Wyly, 2006)


143












Short Term Change Indicators (HMDA data, United States Postal Service [USPS] data)
Variable Explanation and Sources
U Loan originations per Decrease in loan originations may be correlated with
3 capital, 2005-2008 potential long term neighborhood decline
Median household Strong losses in neighborhood household income
income of home suggest decline (Freeman, 2005; Kennedy & Leonard,
u purchase borrowers, 2001; Lees, Slater, & Wyly, 2009; Levy et al., 2006; Pettit
3 2005-2007 & Droesch, 2008; Wyly & Hammel, 1999)
A concentration of high cost loans (associated with
High cost loan subprime mortgages) increases the likelihood of
originations, 2005-2008 foreclosure and decline (Immergluck & Smith, 2006;
SPettit & Droesch, 2008).
The Neighborhood Stabilization Program tracks the
USPS vacancy rate, USPS vacancy rate. Vacant properties are associated
c 2006-2009 with neighborhood decline (Beauregard, 1990; Cohen,
S__2001).

Static Indicators (HUD/Neighborhood Stabilization Program data)
Variable Explanation and Sources
This is an indicator of neighborhood decline taken from
Estimated foreclosure the Neighborhood Stabilization Program data set.
abandonment risk Foreclosure is closely associated with neighborhood
score, 2007-2008 decline. Like vacancy, it has many negative spillover
effects (Scheutz, Been, & Ellen, 2008).
Predicted 18 month
Predated 18 monh This is an indicator of neighborhood decline taken from
underlying problem the Neighborhood Stabilization Program data set
o oe r 2 (Immergluck & Smith, 2006; Scheutz et al., 2008).
2008


144









APPENDIX B
TRANSPORTATION COST INDICATORS


Variables


145


Indicator


Desi* Developed area as a % of total neighborhood area
Density
Residential area as a % of developed area


Building square feet (retail commercial)
Diversity Building square feet (office/service)
Building square feet (industrial)


Desn Road miles per developed area
Design
Number of intersections per road mile


Network distance to nearest regional residential center
Destination Network distance to nearest activity center
Range of network distances to regional residential center










APPENDIX C
MANN-WHITNEY U TEST RESULTS

Duval County: Test between 1963-1979 and 1980-1994


Ranks
Mean Sum of
AGERNG N Rank Ranks
CRIME 63-79 43 35.93 1545.00
80-94 40 48.53 1941.00
Total 83
DECLINE 63-79 43 38.15 1640.50
80-94 40 46.14 1845.50
Total 83
FCAT 63-79 43 36.20 1556.50
80-94 40 48.24 1929.50
Total 83
3ENT 63-79 43 45.29 1947.50
80-94 40 38.46 1538.50
Total 83
4FF HOUSING 63-79 42 32.63 1370.50
PREF 80-94 37 48.36 1789.50
Total 79
LOCAL ACCESS 63-79 43 46.16 1985.00


146


Descriptive Statistics
Std.
N Mean Deviation Minimum Maximum
CRIME 83 7.17 2.173 1 9
DECLINE 83 1.57229232 .8417046679 1.000002034.08334160
71 4
FCAT 83 4.81525623 1.555871124 1.66666496 8.33332539
99 51
GENT 83 1.58370892 .9689926386.99999827 5.41557598
42 4
AFF HOUSING 79 5.09481367 .6466975044 3.44125962 6.53876829
PREF 27 1
LOCAL ACCESS 83 6.92683409 1.3471510352.07291651 8.66666794
60 41
POVMASK 83 .51 .503 0 1
POVRATE 83 3.92 3.144 1 9
RANSP 83 8.289 1.1480 2.5 9.0
GERNG 83 1.48 .503 1 2


























Test Statisticsa
AFF
HOUSING LOCAL
CRIME DECLINE FCAT GENT PREF ACCESS
Mann-Whitney U 599.000 694.500 610.500 718.500 467.500 681.000
Wilcoxon W 1545.000 1640.500 1556.500 1538.500 1370.500 1501.000
S-2.474 -1.680 -2.299 -1.351 -3.041 -1.631
Asymp. Sig. (2- .013 .093 .022 .177 .002 .103
ailed) ______
a. Grouping Variable: AGERNG


Test Statisticsa
POVMAS POVRAT
K E TRANSP
Mann-Whitney U 579.500 524.500 618.000
Wilcoxon W 1525.500 1470.500 1438.000
Z -2.952 -3.213 -2.347
Asymp. Sig. (2- .003 .001 .019
tailed) ___
a. Grouping Variable: AGERNG


147


Ranks
Mean Sum of
AGERNG N Rank Ranks
80-94 40 37.53 1501.00
Total 83
POVMASK 63-79 43 35.48 1525.50
80-94 40 49.01 1960.50
Total 83
POVRATE 63-79 43 34.20 1470.50
80-94 40 50.39 2015.50
Total 83
TRANSP 63-79 43 47.63 2048.00
80-94 40 35.95 1438.00
Total 83










Duval County: Test between 1963-1995 and 1995-2008


148


Descriptive Statistics
Std.
N Mean Deviation Minimum Maximum
CRIME 134 7.32 2.179 1 9
DECLINE 134 1.74129703 .88871713881.000002034.08334160
78 5
FCAT 134 5.15919884 1.653690393 1.666664968.33332539
35 68
GENTRIF 134 1.45701449 .8486999068.99999827 5.41557598
72 8
RASTERVAL 125 5.12759935 .6049738168 3.44125962 6.53876829
U 37 2
GOAL01 125 5.12759935 .6049738168 3.44125962 6.53876829
37 2
LOCALACC 134 6.57151911 1.5392385332.072916518.66666794
E 78 08
POVMASK 134 .60 .491 0 1
POVRATE 134 4.86 3.263 1 9
RANSP 134 8.149 1.1389 2.5 9.0
AGERNG 134 1.38 .487 1 2


Ranks
Mean Sum of
AGERNG N Rank Ranks
CRIME 63-94 83 63.66 5284.00
95-08 51 73.75 3761.00
Total 134
DECLINE 63-94 83 59.60 4947.00
95-08 51 80.35 4098.00
Total 134
FCAT 63-94 83 59.85 4967.50
95-08 51 79.95 4077.50
Total 134
GENTRIF 63-94 83 73.38 6090.50
95-08 51 57.93 2954.50
Total 134
RASTERVAL 63-94 79 61.20 4834.50
U 95-08 46 36.10 3040.50
Total 125
GOAL01 63-94 79 31.20 4834.50
95-08 46 36.10 3040.50





























Test Statisticsa
RASTERVAL
CRIME DECLINE FCAT GENTRIF U GOAL01
Mann-Whitney U 1798.000 1461.000 1481.500 1628.500 1674.500 1674.500
Wilcoxon W 5284.000 4947.000 4967.500 2954.500 4834.500 4834.500
Z -1.551 -3.174 -2.942 -2.407 -.730 -.730
Asymp. Sig. (2- .121 .002 .003 .016 .466 .466
ailed) ____ _
a. Grouping Variable: AGERNG


Test Statisticsa
LOCALACCPOVMAS POVRAT
K E TRANSP
Mann-Whitney U 1346.500 1569.000 1212.500 1531.000
Wilcoxon W 2672.500 5055.000 4698.500 2857.000
Z -3.529 -2.962 -4.246 -2.786
Asymp. Sig. (2- .000 .003 .000 .005
ailed) ____
a. Grouping Variable: AGERNG


149


Ranks
Mean Sum of
AGERNG N Rank Ranks
Total 125
LOCALACC 63-94 83 76.78 6372.50
E 95-08 51 52.40 2672.50
Total 134
POVMASK 63-94 83 60.90 5055.00
95-08 51 78.24 3990.00
Total 134
POVRATE 63-94 83 56.61 4698.50
95-08 51 85.23 4346.50
Total 134
TRANSP 63-94 83 74.55 6188.00
95-08 51 56.02 2857.00
Total 134










Orange County: Test between 1963-1979 and 1980-1994


Ranks
Mean Sum of
AGERNG N Rank Ranks
CRIME 63-79 16 29.09 465.50
80-94 59 40.42 2384.50
Total 75
DECLINE 63-79 16 28.66 458.50
80-94 59 40.53 2391.50
Total 75
FCAT 63-79 16 39.88 638.00
80-94 59 37.49 2212.00
Total 75
3ENTRIF 63-79 16 49.75 796.00
80-94 59 34.81 2054.00
Total 75
O3ALO1 63-79 16 40.25 644.00
80-94 54 34.09 1841.00
Total 70
LOCALACC 63-79 16 54.94 879.00
E 80-94 59 33.41 1971.00
Total 75
POVMASK 63-79 16 36.72 587.50


150


Descriptive Statistics
Std.
N Mean Deviation Minimum Maximum
CRIME 75 6.65 3.291 1 9
DECLINE 75 1.87000376 .9660393063 1.000002034.41667557
68 5
FCAT 75 6.45777133 1.8910021262.333331118.99999142
87 51
GENTRIF 75 1.73709682 .8185544631.99999827 4.09739733
57 2
GOALO1 70 4.98443729 .5303706998 3.84675622 6.15086508
24 6
LOCALACC 75 6.43222380 1.121758340 3.60416794 8.58333492
E 69 65
POVMASK 75 .35 .479 0 1
POVRATE 75 5.05 3.110 1 9
RANSP 75 8.48 .690 7 9
GERNG 77 1.79 .408 1 2























Test Statisticsa
LOCALACC
CRIME DECLINE FCAT GENTRIFGOAL01 E
Mann-Whitney U 329.500 322.500 442.000 284.000 356.000 201.000
Wilcoxon W 465.500 458.500 2212.000 2054.000 1841.000 1971.000
Z -1.981 -1.993 -.392 -2.473 -1.063 -3.505
Asymp. Sig. (2- .048 .046 .695 .013 .288 .000
tailed) ______
a. Grouping Variable: AGERNG


Test Statisticsa
POVMAS POVRAT
K E TRANSP
Mann-Whitney U 451.500 400.500 393.500
Wilcoxon W 587.500 2170.500 2163.500
Z -.322 -.937 -1.116
Asymp. Sig. (2- .748 .349 .264
tailed) ___
a. Grouping Variable: AGERNG


151


Ranks
Mean Sum of
AGERNG N Rank Ranks
80-94 59 38.35 2262.50
Total 75
POVRATE 63-79 16 42.47 679.50
80-94 59 36.79 2170.50
Total 75
TRANSP 63-79 16 42.91 686.50
80-94 59 36.67 2163.50
Total 75










Orange County: Test between 1963-1995 and 1995-2008


Ranks
Mean Sum of
AGERNG N Rank Ranks
CRIME 63-94 75 72.11 5408.50
95-08 79 82.61 6526.50
Total 154
DECLINE 63-94 75 64.11 4808.00
95-08 79 90.22 7127.00
Total 154
FCAT 63-94 75 84.95 6371.00
95-08 79 70.43 5564.00
Total 154
GENTRIF 63-94 75 83.33 6249.50
95-08 79 71.97 5685.50
Total 154
GOAL01 63-94 70 69.07 4835.00
95-08 65 66.85 4345.00
Total 135
LOCALACC 63-94 75 83.11 6233.00
E 95-08 79 72.18 5702.00
Total 154


152


Descriptive Statistics
Std.
N Mean Deviation Minimum Maximum
CRIME 154 7.33766237 2.827365771 1.000000009.00000000
55 08
DECLINE 154 2.13420342 .96023667101.000002034.75000954
65 4
FCAT 154 6.18614099 1.8185610482.333331118.99999142
60 53
GENTRIF 154 1.61774487 .7130773887.99999827 4.09739733
81 8
GOAL01 135 4.96387961 .5008346064 3.66706848 6.15086508
56 3
LOCALACC 154 6.20630571 1.278419447 1.447917228.58333492
E 14 08
POVMASK 154 .37 .484 0 1
POVRATE 154 5.24 3.012 1 9
RANSP 154 8.45 .712 9
GERNG 156 1.51 .502 1 2
























Test Statisticsa
LOCALACC
CRIME DECLINE FCAT GENTRIFGOALO1 E
Mann-Whitney U 2558.500 1958.000 2404.000 2525.500 2200.000 2542.000
Wilcoxon W 5408.500 4808.000 5564.000 5685.500 4345.000 5702.000
Z -1.592 -3.667 -2.034 -1.624 -.330 -1.520
Asymp. Sig. (2- .111 .000 .042 .104 .741 .128
tailed) ____ _
a. Grouping Variable: AGERNG


Test Statisticsa
POVMAS POVRAT
K E TRANSP
Mann-Whitney U 2827.000 2760.500 2814.500
Wilcoxon W 5677.000 5610.500 5974.500
S-.586 -.739 -.579
Asymp. Sig. (2- .558 .460 .563
tailed) ___
a. Grouping Variable: AGERNG


153


Ranks
Mean Sum of
AGERNG N Rank Ranks
POVMASK 63-94 75 75.69 5677.00
95-08 79 79.22 6258.00
Total 154
POVRATE 63-94 75 74.81 5610.50
95-08 79 80.06 6324.50
Total 154
TRANSP 63-94 75 79.47 5960.50
95-08 79 75.63 5974.50
Total 154










Pinellas County: Test between 1963-1979 and 1980-1994

Descriptive Statistics
Std.
N Mean Deviation Minimum Maximum
CRIME 42 6.39047613 2.449072457 1.799999959.00000000
60 42
DECLINE 42 1.95040076 1.433494489 1.000002035.58334446
02 73
FCAT 42 5.96824796 1.7183977832.3333311 8.9999914
6 9
GENTRIF 42 1.87886218 1.404018611.99999827 5.65908194
21 01
GOALO1 35 5.00157527 .61433407653.912351136.19438791
97 8
LOCALACC 42 7.44667824 .8722803968 5.17708445 8.59375191
E 17 2
POVMASK 42 .31 .468 0 1
POVRATE 42 3.83 3.068 1 9
AGERNG 42 1.69 .468 1 2


Ranks
Mean Sum of
AGERNG N Rank Ranks
CRIME 63-79 13 18.85 245.00
80-94 29 22.69 658.00
Total 42
DECLINE 63-79 13 18.65 242.50
80-94 29 22.78 660.50
Total 42
FCAT 63-79 13 19.12 248.50
80-94 29 22.57 654.50
Total 42
3ENTRIF 63-79 13 23.19 301.50
80-94 29 20.74 601.50
Total 42
O3ALO1 63-79 10 20.70 207.00
80-94 25 16.92 423.00
Total 35
LOCALACC 63-79 13 25.42 330.50
E 80-94 29 19.74 572.50
Total 42
POVMASK 63-79 13 21.46 279.00


154



















Test Statisticsb
LOCALACC
CRIME DECLINE FCAT GENTRIFGOALO1 E
Mann-Whitney U 154.000 151.500 157.500 166.500 98.000 137.500
WilcoxonW 245.000 242.500 248.500 601.500 423.000 572.500
Z -.944 -1.153 -.863 -.616 -.986 -1.388
Asymp. Sig. (2-tailed) .345 .249 .388 .538 .324 .165
Exact Sig. [2*(1-tailed .360a .318a .404a .554a .339a .167a
Sig.)]
a. Not corrected for ties.
b. Grouping Variable: AGERNG


Test Statisticsb
POVMAS POVRAT
K E
Mann-Whitney U 188.000 167.500
Wilcoxon W 279.000 602.500
Z -.017 -.597
Asymp. Sig. (2-tailed) .986 .551
Exact Sig. [2*(1-tailed 1.000a .572a
Sig.)] _____
a. Not corrected for ties.
b. Grouping Variable: AGERNG


155


Ranks
Mean Sum of
AGERNG N Rank Ranks
80-94 29 21.52 624.00
Total 42
POVRATE 63-79 13 23.12 300.50
80-94 29 20.78 602.50
Total 42










Pinellas County: Test between 1963-1995 and 1995-2008

Descriptive Statistics
Std.
N Mean Deviation Minimum Maximum
CRIME 80 6.67 2.450 1 9
DECLINE 80 1.91146218 1.375096641 1.000002035.66667795
39 74
FCAT 80 5.92499404 1.7218483662.333331118.99999142
71 03
GENTRIF 80 1.73485921 1.253436301 .99999827 5.65908194
86 23
GOALO1 67 5.01027401 .58369922993.912351136.19438791
96 6
LOCALACC 80 7.18125159 1.169625814 2.79166794 8.59375191
E 78 69
POVMASK 80 .35 .480 0 1
POVRATE 80 4.37 3.188 1 9
AGERNG 80 1.46 .502 1 2


156


Ranks
Mean Sum of
AGERNG N Rank Ranks
CRIME 63-94 43 36.58 1573.00
95-08 37 45.05 1667.00
Total 80
DECLINE 63-94 43 41.10 1767.50
95-08 37 39.80 1472.50
Total 80
FCAT 63-94 43 41.73 1794.50
95-08 37 39.07 1445.50
Total 80
3ENTRIF 63-94 43 43.21 1858.00
95-08 37 37.35 1382.00
Total 80
3OALO1 63-94 36 33.64 1211.00
95-08 31 34.42 1067.00
Total 67
LOCALACC 63-94 43 44.48 1912.50
E 95-08 37 35.88 1327.50
Total 80
POVMASK 63-94 43 38.59 1659.50
95-08 37 42.72 1580.50


















Test Statisticsa
LOCALACC
CRIME DECLINE FCAT GENTRIFGOALO1 E
Mann-Whitney U 627.000 769.500 742.500 679.000 545.000 624.500
Wilcoxon W 1573.000 1472.500 1445.500 1382.000 1211.000 1327.500
Z -1.645 -.286 -.518 -1.179 -.163 -1.650
Asymp. Sig. (2- .100 .775 .604 .238 .870 .099
tailed) ____ _
a. Grouping Variable: AGERNG


Test Statisticsa
POVMAS POVRAT
K E
Mann-Whitney U 713.500 678.500
Wilcoxon W 1659.500 1624.500
Z -.958 -1.165
Asymp. Sig. (2- .338 .244
tailed) _
a. Grouping Variable: AGERNG


157


Ranks
Mean Sum of
AGERNG N Rank Ranks
Total 80
POVRATE 63-94 43 37.78 1624.50
95-08 37 43.66 1615.50
Total 80










LIST OF REFERENCES


Acevedo-Garcia, D., Osypuk, T. L., Werbel, R. E., Meara, E. R., Cutler, D. M., &
Berkman, L. F. (2004). Does housing mobility improve health? Housing Policy
Debate, 15(1), 49-98.

Achtenberg, E. P. (2009). Section 236 decoupling. Retrieved June 14, 2010, from
http://www.lisc.org/files/8207_file_section_236_decoupling.pdf

Achtenberg, E. P. (2002). Stemming the tide: A handbook on preserving multifamily
subsidized housing. New York: Local Initiatives Support Corporation. Retrieved
May 14, 2010, from http://www.lisc.org/files/893_file_asset_upload_file15_838.pdf

Affordable Housing Study Commission. (2006).A preservation strategy for Florida's
affordable multifamily housing. Tallahassee, FL: Author. Retrieved May 17, 2010,
from http://www.floridahousing.org/NR/rdonlyres/63332252-F3EA-486A-9F59-
B95E7F6CDE56/0/AHSCFinalReport2006.pdf

AI-Shalabi, M. A., Mansor, S. B., Ahmed, N. B., & Shiriff, R. (2006, October). GIS Based
Multicriteria Approaches to Site Suitability Analysis. Paper presented at the XXII
FIG Congress, Munich, Germany.

Barber,G. M. (1988). Elementary statistics for geographers. New York: The Guilford
Press.

Basolo, V. (1999). Passing the housing policy baton in the US: Will cities take the lead?
Housing Studies, 14(4), 433-452.

Beauregard, R. A. (1990). Trajectories of neighborhood change: the case of
gentrification. Environment and Planning A, 22 (7), 855-874.

Blackman, A., & Krupnick, A. (2001). Location-efficient mortgages: Is the rationale
sound? Journal of Policy Analysis and Management, 20(4), 633-649.

Block, J. P., Scribner, R. A., & De Salvo, K. B. (2004). Fast food, race/ethnicity, and
income: A geographic analysis. American Journal of Preventive Medicine, 27(3),
211-217.

Bnai-Kashani, R. (1989). A new method for site suitability analysis: The analytic
hierarchy process. Environmental Management, 13(6), 685-693.

Boarnet, M. G., & Crane, R. (2001). Travel by Design: The influence of urban form on
travel. New York: Oxford University Press.

Bodaken, M. (2002). The increasing shortage of affordable rental housing in America:
Action items for preservation. Housing Facts and Findings, 4(4), 1-8.


158









Boer, J. T., Pastor, M., Jr., Sadd, J. L., Snyder, L. D. (1997). Is there environmental
racism? The demographics of hazardous waste in Los Angeles County. Social
Science Quarterly, 78(4), 793-810.

Brewster, K. L. (1994). Race differences in sexual activity among women: The role of
neighborhood characteristics. American Sociological Review, 59(3), 408-424.

Briggs, X. S. (2005a). Introduction. In X. D. Briggs (Ed.), The Geography of Opportunity:
Race and Housing Choice in Metropolitan America (pp. 1-16). Washington, DC:
Brookings Institution Press.

Briggs, X. S. (2005b). More pluribus, less unum?. In X. D. Briggs (Ed.), The Geography
of Opportunity: Race and Housing Choice in Metropolitan America (pp. 17-41).
Washington, DC: Brookings Institution Press.

Brooks-Gunn, J., Duncan, G. J., Klebanov, P. K., & Seland, N. (1993). Do
neighborhoods influence child and adolescent development? The American
Journal of Sociology, 99(2), 353-395.

Bullard, R. D. (2000). Dumping in Dixie: Race, class, and environmental quality.
Boulder, CO: Westview Press.

Burrough, P. A. (1986). Principles of geographical information systems for land
resources assessment. Oxford: Clarendon Press.

Carr, M. H., & Zwick, P. (2005). Using GIS suitability analysis to identify potential future
land use conflicts in North Central Florida. Journal of Conservation Planning, 1,
58-73.

Carr, M. H, & Zwick, P. D. (2007). In Smart land-use analysis: The LUCIS model land-
use conflict identification strategy. Redlands, CA: ESRI Press.

Case, A. C., & Katz, L. F. (1991). The company you keep: The effects of family and
neighborhood on disadvantaged youths. NBER Working Paper No. 3705.
Cambridge, MA: National Bureau of Economic Research.

Center for Housing Policy. (2009). Insights from housing policy research: Taking stock -
the role of "preservation inventories" in preserving affordable rental housing.
Washington, DC: Author. Retrieved May 19, 2010, from
http://www.nhc.org/pdf/takingstockfinalreorder.pdf

Center for Neighborhood Technology. (2010). Penny wise, pound fuelish: New
measures of housing and transportation affordability. Retrieved May 12, 2010,
from http://www.cnt.org/publications

Cervero, R. (2002). Built environments and mode choice: toward a normative
framework. Transportation Research Part D, 7, 265-284.


159









Cervero, R., & Kockelman, K. (1997). Travel demand and the 3Ds: Density, diversity,
and design. Transportation Research Part D, 2(3), 199-219.

Chrisman, N. R. (2002). Exploring geographic information systems (2nd ed.). New York:
Wiley

City of Jacksonville, Planning and Development Department. (2005, October 1). 2005-
2010 consolidated plan. Retrieved June 21, 2010, from
http://www.coj.net/Departments/Housing+and+Neighborhoods/Community+Develo
pment/default.htm

City of Orlando. (2010). Growth management plan: housing element. Retrieved July 4,
2010, from
http://www.cityoforlando.net/planning/cityplanning/Policy%20Document/GMP%202
010/05%20-%20Housing%20GOPs_Supp2o%0No%202.pdf

City of Orlando, Housing Department. (2005). Consolidated plan for housing and
community development programs. Retrieved June 21, 2010, from
http://www.cityoforlando.net/housing/report.htm

City of St. Petersburg. (2005). Five year strategic plan. Retrieved June 22, 2010, from
http://www.stpete.org/pdf/2005housestrategicplan.pdf

Clarke, K. C. (1986). Advances in geographic information systems. Computers,
Environment and Urban Systems, 10(3), 175-184.

Coleman, J. S. (1998). Social capital in the creation of human capital. The American
Journal of Sociology, 94, S95-S920.

Collignon, K. (1999). Expiring affordability of Low-Income Housing Tax Credit
properties: The next era in preservation. Cambridge, MA: Neighborhood
Reinvestment, Harvard Joint Center for Housing Studies.

Collins, M. G., Steiner, F. R., & Rushman, M. J. (2001). Land use suitability analysis in
the United States: Historical development and promising technological
achievements. Environmental Management, 28(5), 611-621.

Coulson, E., & Leichenko, R. (2004). Historic preservation and neighbourhood change.
Urban Studies, 41 (8), 14.

Cowen, D. J. (1988). GIS versus CAD versus DBMS: What are the differences?
Photogrammetric Engineering and Remote Sensing, 54(11), 1551-1555.

Crane, J. (1991). The epidemic theory of ghettos and neighborhood effects on dropping
out and teenage childbearing. The American Journal of Sociology, 96(5), 1226-
1259.


160









Crooks, J. B. (2001). Jacksonville's Consolidation Mayor: Hans G. Tanzler Jr. The
Florida Historical Quarterly, 80(2), 198-224.

Diez-Roux, A V., Nieto, F. J., Muntaner, C., Tyroler, H. A., Comstock, G. W., Shahar, E.,
Cooper, L. S., Watson, R. L., & Szklo, M. (1997). Neighborhood environmental
conditions and coronary heart disease: A multilevel analysis. American Journal of
Epidemiology, 146(1), 48-63.

Dolbeare, C. N., & Crowley, S. (2002). Changing priorities: The federal budget and
housing assistance, 1976-2007. Washington, DC: National Low Income Housing
Coalition. Retrieved May 16, 2010, from
http://www.knowledgeplex.org/kp/report/report/relfiles/nlihc_changing_priorities. pdf

Downs, A. (1992). Regulatory barriers to affordable housing. Journal of the American
Planning Association, 58(4), 419. Retrieved from Business Source Premier
database.

Dubin, R. A., & Sung, C-H. (1990). Specification of hedonic regression: Non-nested
tests on measures of neighborhood quality. Journal of Urban Economics, 27, 97-
110.

Duncan, G. J. (1994). Families and neighbors as sources of disadvantage in the
schooling decisions of white and black adolescents. American Journal of
Education, 103(1), 20-53.

Ebdon, D. (1977). Statistics in geography. Oxford: Basil Blackwell.

Eisinger, P. (1998). City politics in an era of federal devolution. Urban Affairs Review,
33(3), 308-325.

Ellen, I. G., Mijanovich, T., & Dillman, K-N. (2001). Neighborhood effects on health:
Exploring the links and assessing the evidence. Journal of Urban Affairs, 23(3-4),
391-408.

Ellen, I. G., & Turner, M. A. (1997). Does neighborhood matter?: Assessing recent
evidence. Housing Policy Debate, 8(4), 833-866.

Finkel, M., Hanson, C., Hilton, R., Lam, K., & Vandawalker, M. (2006). Multifamily
properties: Opting in, opting out and remaining affordable. Bethesda, MD, and
Cambridge, MA: Econometrica, Inc., and Abt Associates.

Florida Department of Community Affairs (2004) State of Florida consolidated plan,
federal fiscal years 2005-2010. Retrieved May 10, 2010, from
http://www.dca.state.fl.us/fhcd/cdbg/ConsolidatedPlan. cfm


161









Florida Housing Finance Corporation. (2010). Discussion points for consideration for the
2011 Universal Application Cycle. Retrieved June 22, 2010, from
http://www.floridahousing.org/FH-
ImageWebDocs/UniversalApps/2011/ImportantAnnouncements/020-Notices/042-
2011_Universal_Cycle_Discussion_Pointsfor_4-29-
10_First_Rule_Development_Workshop.pdf

Florida Housing Finance Corporation. (2009a). Strategic plan for Florida Housing
Finance Corporation. Retrieved May 6, 2010, from http://floridahousing.org/

Florida Housing Finance Corporation (2009b). 2009 Qualified Allocation Plan. Retrieved
May 6, 2010, from http://floridahousing.org

Florida Housing Finance Corporation and The Shimberg Center for Housing Studies.
(2007). Creation of a national data infrastructure. Retrieved May 10, 2010, from
http://www.preservation.shimberg.ufl.edu/ResearchAndReports. html

Freeman, L. (2005). Displacement or Succession? Urban Affairs Review, 40 (4), 463-
491.

Freeman, L. (2004). Siting affordable housing: Location and neighborhood trends of
low-income housing tax credit developments in the 1990s. Washington, DC: The
Brookings Institution.

Freeman, L., & Braconi, F. (2004). Gentrification and displacement: New York City in
the 1990s. Journal of the American Planning Association, 32(3).

Garner, C. L., & Raudenbush, S. W. (1991). Neighborhood effects on educational
attainment: A multilevel analysis. Sociology of Education, 64(4), 251-262.

Geronimus, A. T., Hicken, M., Keene, D., & Bound, J. (2006). "Weathering" and age
patterns of allostatic load scores among blacks and whites in the Unites States.
American Journal of Public Health, 96(5), 826-833.

Galster, G. C., & Killen, S. P. (1995). The geography of metropolitan opportunity: A
reconnaissance and conceptual framework. Housing Policy Debate, 6(1), 7-43.

Greenburg, M. R. (1999). Improving neighborhood quality: A hierarchy of needs.
Housing Policy Debate, 10(3), 601-624.

Gustafson, J., & Walker, C. J. (2002). Analysis of State Qualified Allocation Plans for
the Low-Income Housing Tax Credit program. Washington, DC: The Urban
Institute.

Hackworth, J. R. (2007). The neoliberal city: governance, ideology, and development in
American urbanism. Ithaca: Cornell University Press.


162









Hall, R., & Ogden, P. E. (1992). The Social Structure of New Migrants to London's
Docklands: Recent Evidence from Wapping. London Journal, 17, 153-169.

Handy, S. L., & Clifton, K. J. (2001). Evaluating neighborhood accessibility: Possibilities
and practicalities. Journal of Transportation and Statistics, 4, 67-78.

Harris, B., & Batty, M. (1993). Locational models, geographic information and planning
support systems. Journal of Planning Education and Research, 12, 184-198.

Hartman, C. (1998). The case for a right to housing. Housing Policy Debate, 9(2), 223-
246.

Hays, R. A. (1995). The federal government and urban housing: Ideology and change in
public policy. Albany, NY: State University of New York Press.

Heikkila, E. (1998). GIS is dead; long live GIS! Journal of the American Planning
Association, 64(3), 350-360.

Higgs, G. (2006). Integrating multicriteria techniques with geographical information
systems in waste facility location to enhance public participation. Waste
Management & Research, 24, 105-117.

Hopkins, L. D. (1977). Methods for generating land use suitability maps: A comparative
evaluation. Journal of the American Institute of Planners, 43(4), 386-400.

Holtzclaw, J., Clear, R., Dittmar, H., Goldstein, D., & Haas, P. (2002). Location
efficiency: Neighborhood and socio-economic characteristics determine auto
ownership and use-studies in Chicago, Los Angeles and San Francisco.
Transportation Planning and Technology, 25, 1-27.

Holzer, H. J., & Stoll, M. A. (2007). Where workers go, do jobs follow? Washington, DC:
Brookings Institution.

Ihlanfeldt, K. R., & Sjoquist, D. L. (1998). The spatial mismatch hypothesis: A review of
recent studies and their implications for welfare reform. Housing Policy Debate,
9(4), 849-892.

Innes, J., & Simpson, D. (1993). Implementing GIS for planning. Journal of the
American Planning Association, 59(2), 230-236.

Jencks, C., & Mayer, S. E. (1990). The social consequences of growing up in a poor
neighborhood. In L. E. Lynn & McGeary, M. G. H. (Eds.), Inner-city poverty in the
United States (pp. 111-186). Washington, DC: National Academy Press.

Joint Center for Housing Studies of Harvard University. (2009). State of the nation's
housing 2009. Cambridge, MA: Author. Retrieved May 13, 2010, from
http://www.jchs.harvard.edu/ publications/markets/son2009/son2009.pdf


163









Joint Center for Housing Studies of Harvard University (2008). America's rental housing:
The key to a balanced national policy. Cambridge, MA: Author. Retrieved May 13,
2010, from
http://www.jchs. harvard.edu/publications/rental/rh08_americas_rental_housing/rh0
8_americas_rental_housing.pdf

Joint Center for Housing Studies of Harvard University. (2003). State of the nation's
housing 2003. Cambridge, MA: Author. Retrieved May 13, 2010, from
http://www.jchs.harvard.edu/publications/markets/son2003.pdf

Kain, J. F. (1992). The spatial mismatch hypothesis: Three decades later. Housing
Policy Debate, 3(2), 371-392.

Kain, J. F. (1968). Housing segregation, negro employment, and metropolitan
decentralization. The Quarterly Journal of Economics, 82(1), 175-197.

Kasarda, J. D. (1989). Urban industrial transition and the underclass. Annals of the
American Academy of Political and Social Sciences, 501, 26-47.

Kasarda, J. D., & Ting, K. (1996). Joblessness and poverty in America's central cities:
Causes and policy prescriptions. Housing Policy Debate, 7(2), 387-419.

Kennedy, M., & Leonard, P. (2001). Dealing with neighborhood change: A primer on
gentrification and policy choices. Montclair, NJ: Brookings Institution Center on
Urban and Metropolitan Policy.

Khadduri, J., & Wilkins, C. (2007). Designing subsidized rental housing programs: What
have we learned? Paper presented at Revisiting Rental housing: A national policy
summit.

Klosterman, R. E. (1997). Planning support systems: A new perspective on computer-
aided planning. Journal of Planning Education and Research, 17, 45-54.

Lang, R., & Horburg, S. P. (1998). What is social capital and why is it important to
public policy? Housing Policy Debate, 9(1), 1-16.

Larsen, K. (2009). Reassessing state housing trust funds: Results from a Florida
survey. Housing Studies, 24(2), 173-201.

Lees, L., Slater, T., & Wyly, E. K. (2007). Gentrification. New York: Routledge/Taylor
and Francis Group.

Leventhal, T., & Brooks-Gunn, J. (2003). Moving to opportunity: An experimental study
of neighborhood effects on mental health. American Journal of Public Health,
93(9), 1576-1582.

Levine, J., & Landis, J. D. (1989). Geographic information systems for planning. Journal
of the American Planning Association, 55(2), 209-220.


164









Levy, D. K., Comey, J., & Padilla, S. (2006). In the face of gentrification: Case studies of
local efforts to mitigate displacement. Washington, DC: The Urban Institute.

Ley, D. (1996). The new middle class and the remaking of the central city. Oxford:
Oxford University Press

Lipman, B. J. (2006). A heavy load: The combined housing and transportation burdens
of working families. Washington, DC: Center for Housing Policy

Listokin, D. (1991). Federal housing policy and preservation: Historical evolution,
patterns, and implications. Housing Policy Debate, 2(2), 157-185.

MacDonald, J. M., & Nelson, P. E., Jr., Do the poor still pay more? Food price variations
in large metropolitan areas. Journal of Urban Economics, 30, 344-359.

Maguire, D. J., (1991). An overview and definition of GIS. In D. J. Maguire, M. F.
Goodchild, & D. W. Rhind (Eds.), Geographical Information Systems Principles
and Applications (pp. 9-20). New York: John Wiley and Sons, Inc.

Massey, D. S., & Kanaiaupuni, S. M. (1993). Public housing and the concentration of
poverty. Social Science Quarterly, 74(1), 109-122.

Malczewski, J. (2004). GIS-based land-use suitability analysis: a critical overview.
Progress in Planning, 62, 3-65.

McHarg, I. L. (1992). Design with nature. New York: J. Wiley.

McKinnish, T., Walsh, R., & White, K. (2009). Who gentrifies low-income
neighborhoods? Journal of Urban Economics, 180-193.

Melendez, E., Schwartz, A., & de Montrichard, A. (2008). Year 15 and Preservation of
Tax-Credit Housing for Low-Income Households: An Assessment of Risk. Housing
Studies, 23(1), 67-87.

Miller, W., Collins, M. G., Steiner, F. R., & Cook, E. (1998). An approach to greenway
suitability analysis. Landscape and Urban Planning, 42(2-4), 85-105.

Mills, D. E. (1981). Growth, speculation and sprawl in a monocentric city. Journal of
Urban Economics, 10, 201-226.

Mollenkopf, J. H., Swanstrom, T., & Dreier, P. (2004). Place matters: Metropolitics for
the twenty-first century. Lawrence, KA: University Press of Kansas.

Moore, L. V., & Roux, A. V. D. (2006). Associations of neighborhood characteristics with
the location and type of food store. American Journal of Public Health, 96(2), 325-
331.


165









Mueller, E. J., & Schwartz, A. (2008). Reversing the tide: Will state and local
governments house the poor as direct federal subsidies decline? Journal of the
American Planning Association, 74(1), 122-135.

National Association of Home Builders. (2004). Where is workforce housing located?: A
study of the geography of housing affordability. Retrieved May 13, 2010, from
http://www.nahb.com/generic.aspx?sectionlD=784&genericContentlD=33617

National Housing Trust. (2008). State and local housing preservation initiatives.
Washington, DC: National Housing Trust. Retrieved May 19, 2010, from
http://www.nhtinc.org/downloads/il_preservation.pdf

National Housing Trust. (2004). State housing finance agencies: Developing programs
to preserve affordable housing. Retrieved May 5, 2010, from
http://content.knowledgeplex.org/kp2/cache/documents/66651 .pdf

National Housing Trust, Reconnecting America. (2008). Preserving opportunities:
Saving affordable homes near transit. Retrieved May 5, 2010, from
http://www.reconnectingamerica.org/public/display_asset/nhtraaff

Nenno, M. K. (1991) State and local governments: New initiatives in low-income
housing preservation. Housing Policy Debate, 2(2), 467-497.

Newman, K., & Wyly, E. (2006). The right to stay put, revisited: Gentrification and
resistance to displacement in New York City. Urban Studies, 43(1), 35.

Newman, S. J., & Schnare, A. B. (1997). "...And a suitable living environment": The
failure of housing programs to deliver on neighborhood quality. Housing Policy
Debate, 8(4), 703-741.

Oakley, D. (2008). Locational patterns of low-income housing tax credit developments:
A sociospatial analysis of four metropolitan areas. Urban Affairs Review, 43(5),
599-628.

Olsen, E. 0. (2001). Housing programs forlow-income households (Working Paper
8208). Cambridge, MA: National Bureau of Economic Research. Retrieved May
17, 2010, from http://www.nber.org/papers/w8208.pdf

Orlebeke, C. J. (2000). The evolution of low-income housing policy, 1949 to 1999.
Housing Policy Debate, 11(2), 489-520.

Paston, M., Jr., Sadd, J., Hipp, J. (2001). Which came first: Toxic facilities, minority
move-in, and environmental justice. Journal of Urban Affairs, 23(1), 1-21.

Pedone, C. I. (1991). Estimating mortgage prepayments and defaults in older federally
assisted rental housing and the possible costs of preventing them. Housing Policy
Debate, 2(2), 245-273.


166









Peiser, R. B. (1999). The fallout from federal low-income housing preservation
programs: A case study in estimating damages. Housing Policy Debate, 10(2),
371-393.

Pereira, J. M. C., & Duckstein, L. (1993). A multiple criteria decision-making approach to
GIS-based land suitability evaluation. International Journal of Geographic
Information Science, 7(5), 407-424.

Pettit, K. L. S., & Droesch, A. E. (2008). A guide to home mortgage disclosure act data.
Washington, DC: The Urban Institute.

Pinellas County, Community Development Planning Division (2009). Description of
projects: 2009-2010 action plan. Retrieved June 22, 2010, from
http://www.pinellascounty.org/community/pdf/action-plan.pdf

Pinellas County, Community Development (n.d.) Welcome to the Community
Development Department. Retrieved June 22, 2010, from
http://www.pinellascounty.org/community/default.htm

Preston, J., & Raje, F. (2007). Accessibility, mobility and transport-related social
exclusion. Journal of Transport Geography, 15, 151-160.

Proscio, T. (2005). Preserving America's affordable housing: Retooling a 20th Century
asset for 21st Century needs. Washington, DC: Local Initiatives Support Coalition.

Raphael, S. (1998). The spatial mismatch hypothesis and black youth joblessness:
Evidence from the San Francisco Bay area. Journal of Urban Economics, 43, 79-
111.

Ray, A., Nhuyen, D., O'Dell, W., Roset-Zuppa, P., & White, D. (2009). The state of
Florida's assisted rental housing. Gainesville, FL: Florida Housing Data
Clearinghouse, Shimberg Center for Housing Studies.

Rohe, W. M., & Freeman, L. (2001). Assisted housing and residential segregation: The
role of race and ethnicity in the siting of affordable housing. Journal of the
American Planning Association, 67(3), 279-292.

Rosenbaum, J., DeLuca, S., & Tuck, T. New capabilities in new places: Low-income
black families in suburbia. In X. D. Briggs (Ed.), The Geography of Opportunity:
Race and Housing in Metropolitan America (pp. 150-175). Washington, DC:
Brookings Institution Press.

Roset-Zuppa, P. (2008). Curbing the loss of affordable rental housing in Florida: A risk
assessment approach. Cornell Real Estate Review, 6(1), 54-69.

Rusk, D. (2003). Cities without suburbs: A Census 2000 update. Woodrow Wilson
Center special studies. Washington, D.C: Woodrow Wilson Center Press.


167









Saaty, T. L. (2008). Decision making with the analytic hierarchy process. International
Journal of Services Sciences, 1(1), 83-98.

Sampson, R. J., & Lauritsen, J. L. (1994). Violent victimization and offending: Individual-
situational-, and community-level risk factors. In A.J. Reiss & J.A. Roth (Eds.),
Understanding and preventing violence: Volume 3 (pp. 1-115). Washington, D.C:
National Academy Press.

Scally, C. P. (2009). State housing finance agencies forty years later: Major or minor
players in affordable housing? Journal of Planning Education and Research, 29(2),
194-212.

Scheutz, J., Been, V., & Ellen, I. E. (2008). Neighborhood effects of concentrated
mortgage foreclosures. Journal of Housing Economics, 17(4), 306-319.

Schwartz, A. F. (2006). Housing policy in the United States: An introduction. New York:
Routledge.

Selye, H. (1978). The stress of life. New York: McGraw-Hill.

Shimberg Center for Affordable Housing. (2008). A risk assessment method for
preservation of assisted rental housing. Retrieved May 7, 2010, from
http://flhousingdata.shimberg.ufl.edu/AHI_introduction. html

Shimberg Center for Affordable Housing. (2007). 2007 rental market study. Retrieved
May 10, 2010, from
http://flhousingdata.shimberg.ufl.edu/shim berg_pub_presentation. html

Sorrentino, J. A., Meenar, M. R., & Flamm, B. J. (2008). Suitable housing placement: A
GIS-based approach. Environmental Management, 42, 803-820.

Smith, D. A. (1999). Mark-to-market: A fundamental shift in affordable housing policy.
Housing Policy Debate, 10(1), 143-182.

Smith, J. P., Edmonston, B., National Research Council (U.S.)., National Research
Council (U.S.)., & National Research Council (U.S.). (1997). The new Americans:
Economic, demographic, and fiscal effects of immigration. Washington, D.C:
National Academy Press.

Smith, N. (1996). The new urban frontier: gentrification and the revanchist city. New
York: Routledge.

Stoll, M. A. (2005). Job sprawl and the spatial mismatch between blacks and jobs.
Washington, DC: Brookings Institution.

Store, R., Kangas, J. (2001). Integrating spatial multi-criteria evaluation and expert
knowledge for GIS-based habitat suitability modeling. Landscape and Urban
Planning, 55, 79-93.


168









Svatos, V. C., & Doucette, J. S. (2003). Analyzing Suitability of Land for Affordable
Housing. Paper presented at the ESRO International User Conference, July 7-11,
2003, San Diego, CA. Retrieved June 4, 2010, from
http://proceedings.esri.com/library/userconf/proc03/p0339.pdf

Talen, E. (1997). The social equity of urban service distribution: An exploration of park
access in Pueblo, Colorado and Macon, Georgia. Urban Geography, 18(6), 521-
544.

Taylor, S. E., Repetti, R. L., & Seeman, T. (1997). Heath psychology: What is an
unhealthy environment and how does it get under the skin. Annual Review of
Psychology, 48, 411-447.

Thomson, C. N., & Hardin, P. (2000). Remote sensing/GIS integration to identify
potential low-income housing sites. Cities, 17(2), 97-109.

Trust for Public Land. (2005). Greenprint for King County. Retrieved June 4, 2010, from
http://www.tpl.org/tier3_cd.cfm?content_itemid= 18178&folderid=262

Turner, M. A. (1998). Moving out of poverty: Expanding mobility and choice through
tenant-based housing assistance. Housing Policy Debate, 9(2), 373-394.

Turner, M. A., & Snow, C. (2001). Leading indicators of gentrification in D.C.
Neighborhoods. Washington, DC: The Urban Institute.

University of Florida. Bureau of Economic and Business Research. (2007). Florida
statistical abstract. Gainesville, FL: University of Florida Press.

U.S. Census Bureau. (2009, September 29). Poverty-definitions. Retrieved June 14,
2010, from http://www.census.gov/hhes/www/poverty/definitions.html

U.S. Department of Housing and Urban Development. (2007a). Affordable housing
needs 2005: A report to Congress. Washington, DC: Author.

U.S. Department of Housing and Urban Development (2007b). Fair Market Rents for the
Section 8 Housing Assistance Payments Program. Retrieved June 21, 2010, from
http://www.huduser.org/portal/datasets/fmr/fmrover_071707R2.doc

U.S. Department of Housing and Urban Development. (2003). Trends in worst case
needs for housing, 1978-1999. Washington, DC: Author.

U.S. Government Accountability Office. (2004). Multifamily housing: More accessible
HUD data could help efforts to preserve housing for low-income tenants.
Washington, DC: Author. Report GAO-04-02. Retrieved May 17, 2010, from
http://www.gao.gov/new.items/d0420.pdf

Wallace, J. E. (1995). Financing affordable housing in the United States. Housing Policy
Debate, 6(4), 785-814.


169









Williamson, A. R., Smith, M. T., & Strambi-Kramer, M. (2009). Housing choice vouchers,
the Low-Income Housing Tax Credit, and the federal poverty deconcentration goal.
Urban Affairs Review, 45(1), 119-132.

Wilson, W. J. (1987). The truly disadvantaged: The inner city, the underclass, and public
policy. Chicago: University of Chicago Press.

Wrigley, N., Warm, D., & Margetts, B. (2003). Deprivation, diet, and food-retail access:
findings from the Leeds 'food deserts' study. Environment and Planning A, 35,
151-188.

Wyly, E. K., & Hammel, D. J. (1999). Islands of decay in seas of renewal: housing policy
and the resurgence of gentrification. Housing Policy Debate, 10(4), 711-771.

Zenck, S. N., Schulz, A. J., Israel, B. A., James, S. A., Boa, S., & Wilson, M. L. (2005).
Neighborhood racial composition, neighborhood poverty, and the spatial
accessibility of supermarkets in metropolitan Detroit. American Journal of Public
Health, 95(4), 660-667.


170










BIOGRAPHICAL SKETCH

Eric Seymour was born in Buffalo, New York in 1979. He earned a Bachelor of

Arts from the Harriet L. Wilkes Honors College of Florida Atlantic University, with a

major concentration in English literature and a minor concentration in history. Before

entering the graduate program in urban and regional planning at the University of

Florida, he pursued graduate studies in early American literature and worked as a

teaching assistant at the University of Florida. While pursuing his master's degree in

urban and regional planning at the University of Florida, he served as president of the

Student Planning Association. He received the American Institute of Certified Planners

Outstanding Student Award in 2010.

This fall, Eric is entering the doctoral program in urban and regional planning at

the University of Michigan, where he plans to study the shrinking cities phenomenon in

Detroit and other cities in the Great Lakes region. Eric hopes to contribute to the

sensitive redevelopment of shrinking cities and assist in the preservation of America's

industrial landscapes and working-class neighborhoods.


171





PAGE 1

1 ASSI STED HOUSING AND THE GEOGRAPHY OF OPPORTUNITY: A COMPARATIVE SUITABILITY ANALYSIS OF THREE FLORIDA COUNTIES By ERIC COURTNEY SEYMOUR A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS IN URBAN AND REGIONAL PLANNING UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Eric Courtney Seymour

PAGE 3

3 ACKNOWLEDGMENTS This research would not have been possible without the generous and substantial contributions several people. First of all, I would like to thank my chair, Dr. Kristin Larsen. Her insight, patience, and unflagging support were integral to the completion of this work. I would also like to thank my cochair, Professor Andres Blanco, for his flexibility, practical advice, and methodological rigor. This research is indebted to the work of the additional members of the Affordable Housing Suitability Model resear ch team in the Department of Urban and Regional Planning: Dr. Ilir Bejleri, Dr. Ruth Steiner, Abdulnaser Arafat, and Eric Kramer. I am especially grateful to Eric for his technical support and cartographic acumen. I would like to thank the Shimberg Center for Housing Studies, especially Bill ODell and Anne Ray for their enthusiasm for the project and assistance in shaping the direction of this study. In addition, I would like to thank Nancy Muller at the Florida Housing Finance Corporation for taking the t ime to meet with me. Of course, this list of acknowledgments would be incomplete without thanking my family for their continued emotional and financial support.

PAGE 4

4 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 3 LIST OF TABLES ............................................................................................................ 7 LIST OF FIGURES .......................................................................................................... 9 LIST OF ABBREVIATI ONS ........................................................................................... 10 ABSTRACT ................................................................................................................... 12 CHAPTER 1 ASSISTED HOUSING PRESERVATION AND AFFORDABLE HOUSING SUITABILTY MODELING ....................................................................................... 14 2 ASSISTED HOUSING, THE GENERATIONAL SHIFT, AND PRESERVATION STRATEGIES ......................................................................................................... 22 Affordable Housing Definitions and Trends ............................................................. 23 Causes of the Affordability Problem ........................................................................ 27 Evolution of Assisted Housing Production .............................................................. 29 First Generation Programs ............................................................................... 30 Interest rate subsidy programs ................................................................... 31 Project based rental assistance ................................................................. 34 Section 20 2 elderly housing program and Section 811 assisted housing .. 35 Second Generation Programs .......................................................................... 36 The low income housing tax credit program .............................................. 37 Implications of the generational shift .......................................................... 38 Preservation Challenges ......................................................................................... 40 Older Assisted Stock ........................................................................................ 40 Newer Assisted Stock ...................................................................................... 43 Low Income Housing Tax Credit Properties ..................................................... 44 Preservation Responses ......................................................................................... 4 5 Older Assisted Stock ........................................................................................ 46 Section 8 Preservation Strategies .................................................................... 47 Rent restructuring ...................................................................................... 47 Debt restructuring ...................................................................................... 48 State Preservation Initiatives ............................................................................ 49 Preservation Inventories ......................................................................................... 50 Summary ................................................................................................................ 52 3 THE GEOGRAPHY OF OPPORTUNITY ................................................................ 54 Sp atial Mismatch, Jobs Housing Imbalance, and Location Efficiency .................... 54

PAGE 5

5 Neighborhood Effects ............................................................................................. 60 Mechanisms ..................................................................................................... 60 Outcomes ......................................................................................................... 66 Neighborhood Change ............................................................................................ 73 Previous Studies on the Location of Assisted Housing ........................................... 75 Summary ................................................................................................................ 78 4 METHODOLOGY ................................................................................................... 80 Affordable Housing Suitability Model ...................................................................... 80 Geographic Information Systems ..................................................................... 81 LandUse Suitability Analysis ........................................................................... 82 Affordable Housing Suitability Model Structure and Methodology .................... 85 Methodology of This Study ..................................................................................... 89 Suitability Criteria Selected for This Study ........................................................ 89 Statistical Methods ........................................................................................... 92 Limitations ........................................................................................................ 93 Summary ................................................................................................................ 94 5 DESCRIPTIONS OF STUDY AREAS: DUVAL, ORANGE, AND PINELLAS COUNTIES ............................................................................................................. 98 Duval County .......................................................................................................... 98 Demographics, Income, and Housing Affordability ........................................... 99 Assisted Housing Inventory ............................................................................ 101 Local Housing Policy ...................................................................................... 101 Orange County ..................................................................................................... 102 Demographics, Income, and Housing Affordability ......................................... 102 Assisted Housing Inventory ............................................................................ 104 Local Housing Policy ...................................................................................... 104 Pinellas County ..................................................................................................... 105 Demographics, Income, and Housing Affordability ......................................... 106 Assisted Housing Inventory ............................................................................ 107 Local Housing Policy ...................................................................................... 108 6 RESULTS FROM TESTS OF DIFFERENCES BETWEEN ASSISTED HOUSING GENERATIONS .................................................................................. 118 Duval County ........................................................................................................ 118 Orange County ..................................................................................................... 121 Pinellas County ..................................................................................................... 124 Summary .............................................................................................................. 125 7 CONCLUSIONS AND RECOMMENDATIONS ..................................................... 133 Policy Recommendations ..................................................................................... 138 Future R esearch and Limitations .......................................................................... 139

PAGE 6

6 APPENDIX A NEIGHBORHOOD CHANGE INDICATORS ......................................................... 141 Neighborhood Change as Gentrification ............................................................... 141 Neighborhood Change as Decline ........................................................................ 143 B TRANSPORTATION COST INDICATORS ........................................................... 145 C MANN WHITNEY U TEST RESULTS .................................................................. 146 Duval County: Test between 19631979 and 19801994 ...................................... 146 Duval County: Test between 19631995 and 19952008 ...................................... 148 Orange County: Test between 19631979 and 1980 1994 ................................... 150 Orange County: Test between 19631995 and 1995 2008 ................................... 152 Pinellas County: Test between 19631979 and 19801994 .................................. 154 Pinellas County: Test between 19631995 and 1995200 8 .................................. 156 LIST OF REFERENCES ............................................................................................. 158 BIOGRAPHICAL SKETCH .......................................................................................... 171

PAGE 7

7 LIST OF TABLES Table page 1 1 Expiration Dates by Funder and Program .......................................................... 21 5 1 Population projection (permanent residents) by age for 19902030, Duval County .............................................................................................................. 111 5 2 Estimated employment by industry in Duval County, 2009 ............................... 111 5 3 Number of severely cost burdened renter households with income less than 80% AMI by tenure and income level, Duval County ........................................ 111 5 4 Growth in severely cost burdened renter households with income less than 80% AMI by tenure and income level, Duval County ........................................ 112 5 5 Population projection (permanent residents) by age for 19902030, Orange County .............................................................................................................. 112 5 6 Estimated em ployment by industry in Orange County, 2009 ............................ 112 5 7 Number of severely cost burdened renter households with income less than 80% AMI by tenure and income level, Orange County ..................................... 113 5 8 Growth in severely cost burdened renter households with income less than 80% AMI by tenure and income level, Orange County ..................................... 113 5 9 Population projection (permanent residents) by age for 19902030, Pinellas Count y .............................................................................................................. 113 5 10 Employment by industry in Pinellas County, 2009 ............................................ 114 5 11 Number of severely cost burdened households with income less than 80% AMI by tenure and income level, Pinellas County ............................................. 114 5 12 Growth in severely cost burdened households with income less than 80% AMI by tenure and income level, Pinellas County ............................................. 114 5 13 Median age estimates and projections, 20062030 .......................................... 115 5 14 Median household income, 1989, 2008 ............................................................ 115 5 15 Renter households with cost burden above 30% and income bel ow 50% AMI, 2008 ................................................................................................................. 116 5 16 Extremely low income (<30 AMI), severely cost burdened households, 2008 116

PAGE 8

8 5 17 Total properties, units, and assisted units in Assisted Housing Inventory by county, 2010 ..................................................................................................... 116 5 18 Properties and assisted units by funder (duplicated count), 2010 .................... 116 5 19 Assisted units by income limits, 2010 ............................................................... 116 5 20 Assisted housing properties and units by age of property or year funded ........ 117 5 21 Median sales price for singlefamily homes (in thousands of dollars), 19952009 ................................................................................................................. 117 6 1 Comparison of assisted properties built/placed in service 19631979 and 19801994, Duval County ................................................................................. 126 6 2 Comparison of assisted properties built/placed in service 19631994 and 19942008, Duval County ................................................................................. 127 6 3 Comparison of assisted properties built/placed in service 19631979 and 19801994, Orange County .............................................................................. 128 6 4 Comparison of assisted properties built/placed in service 19631994 and 19942008, Orange County .............................................................................. 129 6 5 Comparison of assi sted properties built/placed in service 19631979 and 19801994, Pinellas County ............................................................................. 131 6 6 Comparison of assisted properties built/pl aced in service 19631994 and 19942008, Pinellas County ............................................................................. 131 7 1 Incidence of significantly higher suitability values ............................................. 140

PAGE 9

9 LIST OF FIGURES Figure page 4 1 Hierarchical structure of the AHS model. ............................................................ 95 4 2 Goals and objectives of the AHS model. ............................................................ 95 4 3 Structure of the neighborhood change objective. ............................................... 96 4 4 Neighborhood accessibility obj ective. ................................................................. 96 4 5 Local accessibility suitability layer, Orange County ............................................ 97 5 1 Study area counties .......................................................................................... 109 5 2 Location of properties in Florida's Assisted Housing Inventory (AHI) ............... 110 5 3 Ratio of median existing singlefamily house prices to median household incomes by metropolitan area, 1989 2009 ....................................................... 115 6 1 Properties in the assisted housing inventory categorized by age for test between HUD and older FHFC financed properties, Duval County ................. 1 26 6 2 Properties in the assisted housing inventory categorized by age for test between combined HUD and older FHFC financed properties and LIHTC properties, Duval County .................................................................................. 127 6 3 Properties in the assisted housing inventory categorized by age for test between HUD and older FHFC financed prop erties, Orange County .............. 128 6 4 Properties in the assisted housing inventory categorized by age for test between combined HUD and older FHFC financed properties and LIHTC properties, Orange County ............................................................................... 129 6 5 Properties in the assisted housing inventory categorized by age for test between HUD and older FHFC financed properties, Pinellas County ............. 130 6 6 Properties in the assisted housing inventory categorized by age for test between combined HUD and older FHFC financed properties and LIHTC properties, Pinellas County ............................................................................... 132

PAGE 10

10 LIST OF ABBREVIATION S AHI Assisted Housing Inventory AHP Analytic Hierarchy Process AHS Affordable Housing Suitability Model AMI Area Median Income BMIR Below Market Interest Rate CNT Center for Neighborhood Technology ELI Extremely Low Income ELIHPA Emergency Low Income Housing Preservation Act FHFC Florida Housing Finance Corporation FMR Fair Market Rent GIS Geographic Information System(s) HFA Housing Finance Agency HMDA Home Mortgage Disclosure Act IRP Interest Reduction Payment HUD U.S. Department of Housing and Urban Development LEM Location Efficient Mortgage LHFA Local Housing Finance Agency LI Low Income MSA Metropolitan Statistical Area NAP Neighborhood Action Plan Areas NLIHC National Low Income Housing Coalition LHFA Local Housing Finance Agency LIHTC Low Income Housing Tax Credit LIHPRA Low Income Housing Preservation and Resident Homeownership Act

PAGE 11

11 PSS Planning Support System RD U.S. Department of Agriculture, Rural Development SES Socioeconomic Status SMH Spatial Mismatch Hypothesis USPS United States Postal Service VLI Very Low Income VMT Vehicle Miles Travelled

PAGE 12

12 Abstract of Thesis Presented to the Graduate School of the Univ ersity of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts in Urban and Regional Planning ASSIS TED HOUSING AND THE GEOGRAPHY OF OPPORTUNITY: A COMPARATIVE SUITABILITY ANALYSIS OF THREE FLORIDA COUNTIES By Eric Courtney Seymour August 2010 Chair: Kristin Larsen Cochair: Andres Blanco Major: Urban and Regional Planning Privately owned subsidized rental housing, which plays a critical role in providing affordable housing opportunities for lower income fam ilies, seniors and disabled individuals, is increasingly at risk of being permanently lost to the nations affordable housing stock. In response to this looming crisis, state and local entities have become engaged in initiatives to preserve these units and the affordability restrictions attached to them. Assisted housing receiving project based rental assistance from the federal government has been prioritized for preservation, as it is generally able to house extremely low income households. Although the a ffordability of assisted units is an important criterion for preservation, spatial considerations such as neighborhood quality and location efficiency, need to be incorporated into an effective prioritization scheme as well. In order to better inform preservation initiatives, this study assesses and compares the suitability of different generations of assisted housing production based on key spatial criteria. Three Florida Counties, Duval, Orange, and Pinellas, are taken as study areas for this research, as each contains a substantial share of both the states lower income renter households and assisted housing units.

PAGE 13

13 The conceptual model for this research situates assisted housing preservation in the context of the geography of opportunity, a concept that reflects key theories emphasizing the impact of neighborhood conditions on outcomes and opportunities for low income households. This study employs a geographic information system (GIS) based affordable housing suitability (AHS) model developed by the Department of Urban and Regional Planning and the Shimberg Center for Housing Studies at the University of Florida to determine the value of assisted properties in these counties based on key spatial criteria identified in the literature. These criteria are representative of key indicators of socioeconomic conditions and location efficiency. The statistical analysis of these suitability values is then used to determine whether there is a difference between the different generations of assisted housing for eac h of these criteria. The results of these tests show that younger, predominantly statefinanced properties generally have significantly higher suitability values for variables measuring neighborhood quality, while older, predominantly federally financed properties generally have significantly higher suitability values for variables measuring location efficiency. While newer properties may offer as many opportunities for the most needy, they may offer distinct advantages over the older stock in terms of individual outcome and opportunities for advancement. This study recommends that t he decision to prioritize one generation of assisted housing over another should be informed by policy preferences and local housing conditions such as poverty concentration and spatial isolation, and not on affordability alone.

PAGE 14

14 CHAPTER 1 ASSISTED HOUSING PRESERVATION AND AFFORDABLE HOUSING SUITABILTY MODELING The nations assisted housing stock plays a critical role in providing affordable housing opportunities for lower incom e families and individuals. In contrast to public housing, assisted housing is privately owned and operated, and is held by both for profit and nonprofit entities. In exchange for subsidies issued under various federal, state, and local programs, owners of assisted housing agree to affordability restrictions requiring the reservation of subsidized units for lower income tenants. However, due to the nature of these subsidy mechanisms, most assisted housing developments have limited affordability periods. For example, as subsidized mortgages mature and rental assistance contracts come to an end, owners of assisted housing have the option of leaving subsidy programs and converting their properties to market rate housing. Nationwide, several hundred thousand pr operties have already optedout of one the older federal assistance programs, and many more remain at risk of leaving the inventory at the expiration of their affordability periods (Achtenberg, 2002). Assisted units are also at risk of being permanently lost through physical deterioration and mortgage default. The combination of age and insufficient income has conspired to make many older assisted properties particularly susceptible to this threat (see Table 11) (Melndez, Schwartz, & de Montrichard, 2008; Schwartz, 2006). The loss of older assisted housing developed under state and federal programs has become a particular cause of concern for housing activists and agencies (Affordable Housing Study Commission, 2006, National Housing Trust, 2008, Proscio, 2005). As federally assisted housing developments are generally more deeply subsidized than housing developed later under state and local programs, their

PAGE 15

15 affordability restrictions are more likely to require them to set aside units for extremely low income (ELI) residents, who face the greatest difficulties in obtaining affordable housing (Schwartz, 2006).1In order to efficiently target scarce state and local resources in preservation initiatives, efforts have been made to collect and analyze data related to assisted housing. Currently underway is the creation of a national preservation data infrastructure, which is a standardized set of key variables useful for understanding preservation needs (Shimberg Center, 2007). An important use of this data has been the quantification of risk of as sisted housing properties either opting out or failing out of their subsidy programs (Ray et al., 2009; Roset Zuppa, 2008). However, in addition to the determination of risk, evaluations of assisted housing should be based on other criteria indicative of t heir overall value to the affordable housing stock before being prioritized for preservation, including neighborhood quality and transit accessibility. Once assisted units serving ELI residents are lost, they are difficult, if not impossible, to replace. Diminished federal assistance and heightened cons truction costs have made the development of units affordable to lower income households especially difficult (Proscio, 2005). In response to the potential loss of such a critical resource, preservation initiatives have emerged at the state and local level. In particular, State Housing Finance Agencies (HFAs) have played a leading role in preservation efforts. This largely stems from their responsibility for the allocation of the Low Income Housing Tax Credit (LIHTC), which has proven to be an effective inst rument in preserving assisted housing (National Housing Trust, 2008). 2 1 Extremely low income is income at or below 30% of area median income (AMI). To 2 While most urban analysts agree on the importance of neighborhood quality, there is no agreed upon method for measuring it. The term is generally used to refer to the overall character of a neighborhood

PAGE 16

16 this end, this study evaluates the relative suitability of different generations of properties in Floridas assisted housing inventory for preservation initiatives based on such criteria.3An important distinction to be made between this study and those concerned with performing risk assessments is that risk is here taken as the motivation and justificat ion for research, not as the object of study itself. In other words, a critical assumption of this study is that the universe of assisted housing is situated in a general climate of risk. It is not concerned with refining the picture of risk. Rather, this study is concerned with identifying the relationship between the age and location of assisted housing and considering the implication of this relationship for preservation initiatives. The hypothesis of this study is that 1) older, federally assisted properties are located in areas with higher levels of accessibility than newer, stateand locally assisted housing units; and 2) newer properties are generally located in areas with higher levels of neighborhood quality. In essence, it is testing whether the c onventional wisdom regarding the attributes of older versus newer housing in a metropolitan areathe familiar urban/suburban dichotomy holds true for assisted properties as well. However, this is not an arbitrary exercise in testing an intuitive assumption. As Dreier, Mollenkopf, and Swanstrom (2004) argue in their recent book, place mat t ers for individual based on a range of measures such as safety, environmental quality, level of public services, and school quality (cf. Dubin & Sung, 1990; Greenburg, 1999; Newman & Schnare, 1997). Transit accessibility is a function of the number of bus and rail stops/stations within a walkable distance of a particular loc ation and the frequency with which the buses and trains using these stations circulate. 3 The shift in housing production from federal, specifically, HUD administered and financed programs from the 1960s through the mid1980s to state and local responsibility for housing development is referred to in the context of assisted housing as the generational shift (Ray, Nguyen, ODell, Roset Zuppa, & White, 2009).

PAGE 17

17 outcomes, and the spatial pattern of neighborhood quality and assisted housing need to be jointly considered in making an informed preservation decision. Spatial variables in addition to housing market and property specific characteristics are increasingly becoming recognized as having an influence on the cost of housing, both narrowly and widely construed. For example, the relationship between transportat ion and residential location can have a dramatic impact on household expenditures. Housing units located far from employment, activity centers, and transit stops increase the amount spent on transportation for households with cars, isolate those dependent on public transportation, and limit the opportunities of both groups to enjoy a high quality of life (Briggs, 2005; Lipman, 2006). Areas marked by highpoverty levels, high crime rates, and poor school performance can have adverse outcomes for residents of assisted housing, especially children, requiring additional outlays for medical expenses and other services (Galster & Killen, 1995; Wilson, 1987). The cost in terms of limited economic opportunities is especially important in the calculus of housing affo rdability. Areas in which gentrification is occurring are likely to have few, if any, opportunities for lower income households to afford market rate housing (Kennedy & Leonard, 2001). In order to evaluate the neighborhood quality of assisted housing along these various dimensions, a landuse suitability model specifically designed for the purposes of identifying areas suitable for affordable housing development and preservation will be used in this study. This model, the Affordable Housing Suitability (AH S) model, is a joint effort of the Shimberg Center for Housing Studies and the Department of Urban and Regional Planning at the University of Florida. The AHS model provides a structured

PAGE 18

18 system for the integration of relevant suitability criteria, communit y preferences, and critical spatial data in a decisionmaking environment. Though weightedoverlay analysis, the model produces a composite representation of suitability for affordable housing. While the composite suitability values the model produces are invaluable and will be included in this study, specific subcomponents of the model indicative of suitability along discrete dimensions will be selected for closer analysis. As older properties are most at risk for leaving the stock of assisted housing, t his study will compare the relative suitability of properties developed under U.S. Department of Housing and Urban Development (HUD) assistance programs with older properties developed by Florida Housing Finance Corporation (FHFC), which is the HFA in Flor ida. In discussions of preservation, HUD assisted properties have deservedly received considerable attention due to the imminent termination of affordability restrictions for a significant number of them. While older FHFC assisted properties generally have longer affordability restrictions and therefore are not at risk of converting to market rate housing, their age places a number of them at risk of deterioration and default. Thus, both of these generations of assisted housing are at risk, but for different reasons. Consequently, it is important to ascertain whether HUD financed properties are more suitable than older FHFC financed properties on a range of key indicators before allocating scarce resources for preservation to one or the other population of a ssisted housing. This study also compares the older properties of both types (HUD and FHFC) with younger properties developed by FHFC and local housing finance agencies (LHFAs). This comparison is intended to shed light on the relative suitability of older

PAGE 19

19 assisted housing in general when compared to newer developments. The results of this study will be useful in deciding which populations should be prioritized for preservation. Duval, Orange, and Pinellas Counties serve as the case study areas for this a nalysis. These Counties have been selected based primarily on their participation in the development of the AHS model, but these areas are also ideal for this study as they have large shares of Floridas assisted housing as well as large shares of the stat es lower income households in need of affordable housing. These counties are located at the core of some of the most populous metropolitan statistical areas (MSAs) in Florida: Duval is the heart of the Jacksonville MSA; Orange is the core of the Orlando M SA, and Pinellas is part of the TampaSt. Petersburg Clearwater MSA. Though these areas have key similarities allowing for comparison, they are also in geographically diverse regions, and are consequently marked by different development patterns and socio economic conditions. This case study selection allows for a useful comparison of assisted housing suitability taking the specific context of each area into consideration. In addition, these counties privilege preservation over new development for meeting t heir affordable housing needs. Therefore, the results of this study could be of particular relevance to the preservation initiatives of these communities. The next chapter supplies background on the specific programs responsible for assisted housing pr oduction, the challenges assisted housing developments face in terms of their continued presence in the states stock of affordable housing, and preservation responses. The concept of the geography of opportunity is presented in Chapter 3. Chapter 4 desc ribes the methodology used in this study, Chapter 5 describes the case study areas in greater detail and Chapter 6 presents the findings of

PAGE 20

20 the analysis. Finally, Chapter 7 evaluates these results and offers policy recommendations and suggestions for fut ure research.

PAGE 21

21 Table 11. Expiration Dates by Funder and Program Program Total Units Affordability Period Expired By 2010 By 2020 By 2030 By 2030 By 2040 After 2040 HUD Programs 52,328 151 1,142 12,553 4,185 13,078 7,953 994 Section 202 24,510 0 205 3,997 3,518 10,310 5,703 777 Section 236 8,025 151 136 7,294 188 0 256 0 Section 811 745 0 0 0 0 0 528 217 Section 221(d)(3) & (4) 7,471 0 801 1,262 479 2,768 1,466 0 Section 8 (project based) ONLY 11,577 This rental assistance is now generally provided via 15 year contracts to properties Rental Assistance 16,845 Rural Development 19,872 1,945 4,131 3,738 5,540 3,814 704 0 Section 514/516 3,934 1,355 865 445 221 816 232 0 Section 515 15,938 590 3,266 3,293 5,319 2,998 472 0 Section 521 11,171 This rental assistance is provided via 4 5 year contracts to properties Florida Housing Finance Corporation 155,769 13,567 7,257 755 582 24,878 24,796 83,954 Source: Affordable Housing Study Commission, 2006

PAGE 22

22 CHAPTER 2 ASSISTED HOUSING, THE GENERATIONAL SHIFT, AND PRESERVATION STRATEGIES This chapter places this study in the wider context of both affordable housing needs and assisted housing development and preservation in the U.S. Of particular relevance to this study is the character of assisted housing units produced under the various s ubsidy programs including their level of assistance, the population they served, and the specific preservation challenges they face. This chapter begins by providing an overview of affordable housing trends both in Florida and in the nation as a whole, wit h a particular emphasis on the unmet demand for affordable rental housing for low income households. This section answers the question of why preservation is a worthwhile endeavor in the first place. Next, a review of the literature and history of federal lowi ncome rental housing assistance focuses on the HUD programs responsible for the production of assisted rental housing. Then federal devolution and the increased role of state and local agencies in the production of affordable rental housing will be considered. In this section, the generational shift from federal to state assisted housing and its ramifications will be considered. The generational shift is a key element of the present studys analysis. The assisted rental housing preservation crisis will then be discussed, including the scale of the problem and the strategies that have been devised to address it. Here, the implications of the generational shift on preservation efforts will be examined. In conclusion, data collection efforts undertake n to generate an accurate representation of assisted housing characteristics will be discussed.

PAGE 23

23 Affordable Housing Definitions and Trends Inadequate conditions have historically been the greatest concern for housing advocates and public officials. Tenem ent conditions in large industrial cities at the turn of the twentieth century were the initial catalyst for significant housing interventions in the U.S. (Hall, 2002). In recent decades, however, housing affordability has been the greatest concern. While several measures of affordability exist, the most widely accepted measure is the percentage of household income spent on housing, with 30% being the accepted standard of affordability. Households spending more than 30% of their pretax income are considered cost burdened, and those spending more than 50 % of their pre tax income are considered severely cost burdened (Schwartz, 2006). For renters, the cost of housing is the sum of rent and utility payments; for homeowners, t he cost of housing includes mortgage payments, property tax es and insurance. Affordable housing is out of reach for a significant number of U.S. households. Illustrating the extent of the affordability problem, 30% of all homeowners and 45% of all rent ers paid more than 30% of their income on housing in 2007 (Joint Center for Housing Studies, 2009). Lower income households are particularly likely to suffer from a heavy housing cost burden. Fully 70% of renters and 67% of homeowners in the lowest income quartile paid more than 50% of their income toward housing, while 50% of renters and 42% of homeowners in the lowest income quartile paid as much. These trends have worsened over time; between 2001 and 2007, the number of severely cost burdened homeowners in the lowest income quartile increased nearly 18% while the number of severely cost burdened renters in the lowest income quartile increased 16% Although the greatest percent growth among severely cost burdened households between 2001 and 2007 was among households in the upper middle income quartile,

PAGE 24

24 the overall percentage of higher income households being severely cost burdened remains relatively small compared to lower income households, particularly renters. The share of severely cost burdened renter s was nearly twice that of homeowners in 2007 (Joint Center for Housing Studies, 2009). Further illustrating the affordability crisis for low income renter households, the most recent HUD (2007) w orst case housing needs survey reports that 5.99 million ver y low income renter households had worst case needs in 2005, which is a 16% increase from 5.18 million in 2003.4These national trends are equally manifest in the state of Florida. In 2008, nearly 53% of all renters paid more than 30% of their income toward housing, while nearly 38% of all homeowners did so ( U.S. Bureau of the Census, 2008). Just as in the nation as a whole, cost burden is disproportionally conc entrated among low income households in Florida. In 2005, nearly twothirds of low income renter households in Florida were cost burdened (Shimberg Center, 2007). In 2008, nearly 80% of renter households earning less than $20,000 annually paid more than 30% of their income toward housing (U.S. Bureau of the Census, 20062008). In absolute terms, 558,115 low income renter households experienced cost burden in 2007 (Shimberg Center, 2007). Not only has the affordability problem in Florida been persistent, but it has also worsened in recent years (Shimberg Center, 2007). The report also revealed that 72% of extremely low income renters had worst case needs in 2005. While households living in poverty are the largest group affected by the inadequate provision of affordable housing, working households earning from one to 4 In 1990, HUD was directed by Congress to submit regular reports on households with worst case h ousing needs, defined as unassisted and severely cost burdened renter households earning less than 50%of Area Median Income (AMI).

PAGE 25

25 three times the minimum wage are incr easingly subject to cost burden.5 In 2003, 37% of households earning incomes in this range were severely cost burdened (Joint Center for Housing Studies, 2003). Severe cost burden is largely a function of the confluence of stagnant incomes and soaring hous ing costs. In order to highlight the difficulty of lower income working households securing affordable housing, the National Low Income Housing Coalition (NLIHC) issues annual reports with the wage required to secure a modest two bedroom apartment at Fair Market Rent (FMR) while paying no more than 30% of income on rent and utilities.6 5 Poverty is a function of income and family size. In order to determine poverty, the U.S. Census Bureau applies a set of i ncome thresholds that vary by family size and composition to determine who is in poverty (U.S. Census Bureau, 2009). This measure is referred to as the housing wage. F air M arket R ent for a two bedroom apartment in Florida is $1,055, meaning that a household must earn more than $40,000 annually in order to pay no more than 30% of income on rent and utilities combined; this translates to a housing wage of $20.29 per hour. The mean wage for a renter, however, is only $13.23 per hour. Thus, the cost for an apartment at FMR is 122% of mean rent er household income. The rent for this modest two bedroom apartment is 236% of income for a person earning only minimum wage (NLIHC, 2010). According to the NLIHC report, 59% of renters are unable to afford a twobedroom apartment at F MR, a significant inc rease from the 40% of renters in 2000 (NLIHC, 2000; NLIHC, 2010). In 2003, the Joint Center for Housing Studies reported that households with one full time minimum wage earner cannot afford to rent even a onebedroom apartment anywhere in the country (p. 27). Even essential service 6 Fair Market Rents (FMRs) are gross rent estimates for different areas throughout the county, including the cost of the rental unit and essential utilities, such as heat and water. FMRs are produced annually by HUD in order to determine payment amounts for rental assistance programs. FMRs are not the same as average rent in an area. They are set at a high enough level in order to allow f or the selection of a range of housing units in a variety of neighborhoods, but also low enough to accommodate a significant number of program participants (HUD, 2007b).

PAGE 26

26 p ersonnel such as teachers, police officers and nurses are increasingly being priced out of the communities they serve (National Association of Home Builders, 2004). In order to secure affordable housing, many households move away from their place of work as housing costs tend to fall with distance from employment centers. While the resultant housing costs may be lower for these households, their transportation costs are significantly higher than those living near their place of work. Transportation costs are the second largest expense, behind housing, for U.S. households, and in many cases affordability gains achieved by residing far from work are negated by the increased costs of commuting (Lipman, 2006). Illustrating this tradeoff, households spending 30% or less of their income on housing spend nearly 25% of their income on transportation costs, while severely cost burdened households only about 8% of their income on transportation (Lipman, 2006). Thus, the Center for Neighborhood Technology (CNT) (2010) a nonprofit research institute, argues that the conventional measure of affordability should be changed to reflect the influence of transportation costs. The nonprofit suggests a location may be considered affordable if combined housing and transportat ion cost s do not exceed 45% of area median income (AMI). Housing affordability problems affect diverse sections of society, but certain demographic groups are particularly disadvantaged. Thus, while households of all ages experience cost burden, the probl em is most acutely felt by the youngest and oldest groups of renters; 34% of renters under age 25 and 32% of renters over the age of 75 were severely cost burdened in 2005 (Joint Center for Housing Studies, 2008). In Florida, nearly 60% of renter household ers between 15 and 24 were cost burdened in

PAGE 27

27 2008, while exactly 60% of renter householders over the age of 65 were cost burdened in the same year (U.S. Bureau of the Census, 2008). Similarly, while households of all races and ethnicities are subject to cos t burden, minority renters are most adversely affected. At the national level, more than 30 % of black renters and 27% of Hispanic renters were severely cost burdened, which in absolute terms is substantially more than the 21% of white renters similarly bur dened in 2006 (Joint Center for Housing Studies, 2008). Singleparent and femaleheaded households are also heavily affected by cost burden (Joint Center for Housing Studies, 2003). In terms of geographic characteristics, the majority of cost burdened hous eholds live in central cities, but a significant number also live in the suburbs. Among households with worst case housing needs in 2005, 2.91 million resided in central cities, while 2.09 million resided in suburbs and 0.99 million resided in nonmetropolitan areas (HUD, 2007 a ). Causes of the Affordability Problem The affordability crisis for low income renter households is largely attributable to the dearth of suitably priced housing units. Exacerbating this problem, rents have increased much faster than the income of renter households, pushing a growing portion of the rental housing stock beyond the means of low income renters (Schwartz, 2006, p. 34). These elevated rents also reflect the continued, simultaneous production of newer, higher quality, hi gher cost units and destruction of older, lower quality, lower cost units (Joint Center, 2008). This replacement of the affordable rental inventory by more expensive units places low income renters at a particular disadvantage. The number of units affordable to renter households earning 30% or less of AMI declined by 19% in the 1990s, while the number of units affordable to renter households earning between 50 and 80% of AMI declined by 5% during the same period (HUD, 2003). The

PAGE 28

28 Joint Center for Housing Studies (2008) reports that from 1995 to 2005, nearly 2.2 million of the 37 million initially available rental housing units (occupied and vacant) were demolished or otherwise removed from the inventory (p. 13). Among those lost, the number of singlefamily and small multifamily rental units those most frequently inhabited by low income renters was especially high. In 2001, the number of renter households in the bottom income quintile exceeded the stock of rental units affordable to them by fully two million (Joint Center for Housing Studies, 2003). The inadequate supply of rental housing affordable to low income households is primarily a function of the private housing markets inability to produce and maintain this type of housing stock independent of publ ic subsidies (Schwartz, 2006). Operating costs, including repairs and other expenditures, as well as the rate of return demanded by investors, skew the production of rental housing toward highend submarkets in order to ensure an adequate revenue stream and profit margin. When rent fails to cover operation costs, as it frequently does in older rental properties, owners of existing affordable stock may compensate by deferring maintenance and mortgage payments, beginning a downward spiral that ends with the property eventually being removed from the affordable housing stock (Joint Center for Housing Studies, 2003). Market forces in recent years have further disincentivized the private production of modest rental housing. The homeownership boom preceding the mortgage crisis steered multifamily production toward condominiums and away from rental units; as a result, multifamily rental construction fell off precipitously from 2002 to 2007 (Joint Center for Housing Studies, 2008). While condominiums and housing unit s are being placed on the rental market due to the foreclosure crisis, the rents they command are too high to be

PAGE 29

29 affordable to low and moderateincome households, leading to a mismatch in the rental inventory (Joint Center for Housing Studies, 2009). Rising development and land costs have also made it less attractive for developers to produce modest rental properties, even with subsidies. These costs are exacerbated by regulatory mechanisms, such as zoning, that limit the amount of land available for mul tifamily rental construction and prescribe measures of size, quality and density of housing units more amenable to highend housing production. While these regulations are intended to promote the worthy goals of environmental protection, proper sanitation, and superior housing quality, the high cost of land per unit generated by these regulatory requirements renders the production of affordable housing in some comm unities altogether impractical ( Downs, 1992; Schwartz, 2006). Evolution of Assisted Housing Pr oduction Since the first housing programs of the New Deal in 1933 and the landmark Housing Act of 1949, which promoted the goal of a decent home and a suitable environment for every American family, the federal government has played an evolving role in attempting to ameliorate the affordability crisis (Orlebeke, 2000). Orlebeke (2000) divides the time since the passage of the 1949 act ro ughly into two segments: t he first of which characterized by strong federal leadership in housing policy, ran from 1949 until 1973. T he second period, characterized by increased federal devolution, the emergence of state and local actors as key agents in the management of housing programs, and a shift from supply side to demandside affordable housing solutions extends fr om 1973 to the present (Orlebeke, 2000).7 7 Of course, Orlebeke is writing before the most recent housing crisis, the occurrenc e of which may necessitate a revision of this periodization scheme. The 1973 moratorium imposed by

PAGE 30

30 the Nixon administration on assisted housing production that Orlebeke (2000) locates as the terminus of the federal leadership model, however, did not altogether cease the federal g overnments direct involvement in producing assisted housing, which continued albeit to an increasingly diminished extent into the mid1980s (p. 490). Strictly in terms of the production of assisted housing, however, a more accurate starting point for this devolution is the 1980s This shift from the federal to state and local responsibility has been referred to as the national generational shift in housing production, and it has important implications for preservation initiatives (Ray, Nguyen, ODell, Ros et Zuppa, & White, 2009). In terms of the larger trajectory of devolution, which Eisinger (1998) defines as the reallocation of specific responsibilities from Washington to subnational governmentsprimarily involv[ing] a shift from national to state gover nment, the 1973 transition point is valid (p. 314). In that year, the Housing and Community Development Act created not only the Section 8 program but inaugurated the use of block grants. Rather than continuing the consolidation of housing policy authori ty, these programs place the responsibility for administration and allocation on state and local governments. The production policies and programs of these two periods, and the consequences of the generational shift will be considered below. First Generat ion Programs The Housing Act of 1949 reauthorized the public housing program, which was the federal governments first approach toward meeting the housing needs of low income households. This program is both publicly funded and managed, and the government, in the form of local Publi c Housing Authorities (PHA), is the owner of these affordable properties. In order to avoid competition with the private market, this program was

PAGE 31

31 designed to house extremely low income households, primarily those displaced as a r esult of urban renewal (Listokin, 1991). The Housing Act of 1949 mandated a 20% gap between the highest rents charged for public housing and the lowest rents commanded on the private market; as a result, public housing increasingly became the repository of the poor (Listokin, 1991). Though the inventory of public housing properties has been drastically reduced due to demolition, particularly as a result of revitalization and mixedincome goals of the 1992 HOPE VI program, public housing properties were designed to remain affordable in perpetuity. Beginning in the 1960s the federal government began to subsidize the production of privately owned affordable rental housing properties in order to augment the public housing inventory and to offer a more politically palatable alternative to the troubled public housing program (Listokin, 1991). The units produced under these federal programs through the 1980s constitute the first generation of assisted housing. Unlike public housing, however, the continued affordability of the properties produced under these programs was and remains contingent upon the continuation of rental assistance contracts and other subsidy mechanisms with a finite lifespan. Interest rate subsidy programs T he Kennedy administrations engagement of the private sector reflects its eager ness to develop less con troversial assistance programs and to use public private partnerships in housing programs to stimulate the flagging economy of the early 1960s. The Housing Act of 1961 created the Section 221(d)(3) Below Market Interest Rate (BMIR) program, which was designed to supply housing for moderateincome families whose earnings exceeded the limits for public housing but remained underserved by the private rental market. Under this program, private and nonprofit developers could obtain

PAGE 32

32 FHA insured BMIR mortgages at 3% from private lenders, who in turn sold them to Fannie Mae at market rate. These low interest rate loans reduced the overall debt service for the developers, allowing them to pas s on the savings in the form of lower rents to residents. When implemented. the Section 221(d)(3) program encountered a host of difficulties and ultimately generated relatively little housing. The most critical reason for the failure of the program was the perceived financial excesses of the program, based on the large outlays in the annual federal budget for the purchase of individual mortgages during the year they were acquired. T he fiscal structure of the program made it particularly vulnerable to critic ism though the actual size of the subsidy was far less than these figures indicated as principal and inter est would be recouped over time (Hays, 1995). As a consequence of these criticisms the Section 221(d)(3) program was replaced by the Section 236 interest rate subsidy program, which was established by the Housing Act of 1968. As with the Section 221(d)(3) program, Section 236 was designed to allow private developers to reduce their rents as a consequence of having a lower debt service. However, rather than purchasing mortgages outright from private lenders, the government provided monthly interest reduction payment s (IRP s) to sponsors, effectively reducing their loan payments from what they would pay at market rate (Achtenberg, 2002).8 8 Federal interest rate payments to owners of assisted properties reduced monthly mortgage payments equal to what they would pay for a loan issued at one percent interest As Schwartz (2006) points out, these annual subsidy payments made Section 236 seem less costly from a budgetary standpoint than Section 221(d)(3), although the level of public expenditure was actually greater under this subsequent program (p. 131). Due to the larger su bsidy, properties developed under this program

PAGE 33

33 were able to charge lower rents and accommodate somewhat lower income households. In order to further attract investment, tax benefits were provided for property owners in the form of rapid depreciation and mortgage interest deductions. As a consequence, t he Section 236 program was far more successful than Section 221(d)(3); more housing was produced under this new program within 3 years than had been produced during the entirety of its predecessor. All told, t he program produced more than 544,000 units between 1968 and 1983 (Olsen, 2001). A large number of the properties developed through Section 236, as well as through Section 221(d)(3) received some form of rental assistance in order to make units accessible to lower income households. Income eligibility was set at 80% or less of AMI; thus additional subsidies were needed to cover the rent for very lowincome tenants. Older assisted properties developed with BMIR loans received this subsidy through the Rent Supplement (RS) and Rental Assistance Payment (RAP) programs. Though these programs were eliminated in the early 1970, many of the properties that had received additional rental subsidies continued to do so under the Section 8 Loan Mortgage Set Aside program of 1974 (Achtenberg, 2002; Finkel, Hanson, Hilton, Lam, & Vandawaller, 2006; Schwarz, 2006). As will be seen, layering of assistance for these early projects lengthened their affordability periods. These mortgage subsidy programs ultimately ran into seri ous difficulties. Inadequate subsidies in the context of an inflationary economic climate led to operating costs quickly surpassing revenue. As a result, numerous projects became delinquent on their mortgage payments, and many eventually went into default. Furthermore, these programs come under criticism for funding poorly sited projects and projects undertaken

PAGE 34

34 by inexperienced and unscrupulous sponsors (Listokin, 1991; Orlebeke, 2000). In response to the poor performance of these programs, the Nixon admini stration imposed a moratorium on the production of subsidized housing pending the development of more effective mechanisms for assisting low income households in acquiring affordable housing. Project based rental assistance In the 1970s, the federal government took a new approach toward the creation of privately owned, publicly assisted affordable housing units serving low income families. The Section 8 New Construction and Substantial Rehabilitation program (NC/SR), the supply side component of the larger Section 8 program authorized by the Housing Act of 1974, also provided a demandside incentive. It augmented the rent property owners charged, covering the difference between 25% (later 30%) of tenant income and FMR. Income eligibility for the program was initially capped at 80% of AMI adjusted for family size.9 9 I n 1981 the cr iteria was adjusted so that only those earning under 50 % of AMI became eligible, reflecting a programmatic shift in emphasis from serving low to very low income households (Listokin, 1991). Interest rate subsidies were no longer issued by the federal government, but in some cases they were secured from state housing finance agencies. Developers were able to allocate some or all of their projects for the program, making it particularly flexible. As was the case with projects developed under Section 236, tax incentives were extended to developers, permitting them to claim accelerated depreciation allowances. The combination of the deep subsidies and generous tax advantages made the Section 8 program very attractive to developers and investors (Schwartz, 2006, p. 133). As a result, the program ultimately subsidized more than 850,000 new or rehabilitated housing units between 1974 and 1983 (Olsen, 2001).

PAGE 35

35 Reflecting the shift from supply side to demandside housing assistance that occurred during the Reagan administration, the Section 8 NC/SR program was terminated under the Housing Act of 1983. This decision was largely informed by the fi ndings of the 1981 Presidents Commission on Housing, which concluded that as affordable housing of both sufficient quantity and quality had been developed during the 1970s, it would be more efficient to direct resources toward tenant based housing assista nce (Listokin, 1991; Orlebeke, 2000). With the repeal of the Section 8 NC/SR production program, the Section 8 Existing Housing certificate program was left as the largest remaining federal housing program (Orlebeke, 2000). This program gave incomeeligible households certificates allowing them to seek housing in the private rental market With the exception of programs designed to subsidize the production of housing for the elderly and disabled, the federal government, guided by the Reagan administrations antiproduction, voucher only housing policy, ceased its direct involvement in rental housing production (Orlebeke, 2000, p. 509). Section 202 elderly housing program and Section 811 assisted housing Created by the Housing Act of 1959, the Section 202 Supportive Housing for the Elderly program subsidizes the development of housing for low income seniors by nonprofit organizations (Schwartz, 2006). Initially, the program provided a 3% BMIR loan for the costs of construction, rehabilitation, or acquisition, and the debt service for these projects was covered by project based Section 8 subsidies. Since 1992, capital grants have been used in place of loans (Schwarz, 2006). Established under the National Aff ordable Housing Act of 1990, Section 811 provided for the production of assisted housing for the severely disabled using a subsidization mechanism similar to that of Section 202. The Section 202 program has produced more than 260,000 units

PAGE 36

36 and Section 811 more than 30,000 units (Schwartz, 2006). Unlike Sections 221(d)(3), 236, and 8, these programs continue to receive direct subsidies, but they have produced far fewer total units than those of the terminated programs taken as a whole (Ray et al., 2009). Second Generation Programs A defining characteristic of U.S. housing policy since the 1973 moratorium has been the formal transfer of most housing program control from the federal government to state and local governments (Orlebeke, 2000, p. 491). This t ransfer has largely been accomplished through the replacement of categorical and centrally administered federal housing programs such as Section 8 NC/SR with block grants, which are allocated to state and local governments for use in programs and initiativ es they deem best suited to their needs. In this context, HFAs have emerged as major players in the administration and financing of assisted housing production (Basolo, 1999; Nenno, 1991; Scally, 2009). These state agencies first appeared in the 1960s, and had three important functions: a primary role as an administrator of other housing subsidies, a secondary role as an administrator of other housing subsidies, and an emergent role as a (re)developer of affordable housing (Scally, 2009, p. 198). These st ate agencies were granted the authority to issue tax exempt bonds for financing the production of assisted housing, and they served as administrators of federal housing programs, playing a supporting role in the administration of Section 236 (Scally, 2009) With the devolution of federal leadership, HFAs have become responsible for administering the HOME Investment Partnership Program, authorized under the CranstonGonzalez National Affordable Housing Act of 1990, the Low Income Housing Tax Credit (LIHTC), created

PAGE 37

37 under the Tax Reform Act of 1986, as well as a share of Community Development Block Grants (CDBG) (Schwartz, 2006).10The low income housing tax credit program The establishment of the Low income Housing Tax Credit (LIHTC) program reflects the larg er context of devolution characteristic of federal policy since the advent of the Reagan Administration and the generational shift in housing production more specifically. The LIHTC allocates federal tax credits to state agencies, usually HFAs, which subsequently become responsible for managing their use in the production of assisted rental housing. This program raises equity for the production of assisted housing by offering federal income tax credits to investors, who are able to use them to claim a dol lar for dollar reduction in incometax liability for a period of 10 years. Developers apply to HFAs for tax credits, and then sell them to investors in order to finance their projects. The amount of the credit depends on the cost of the project, its locati on, and the proportion of total units reserved for low income households.11 10 Devolution has placed a significant degree of responsibility for affordable housing financing on state agencies and is characterized by a significant withdrawal of direct federal subsidies for housing. HUDs budget of $34.3 billion for 2002 was only 41% of its $83.6 billion budget for 1976 (Dolbeare & Crowley, 2002). S tates have risen to this challenge through the establishment of housing trust funds, which in many cases are also administered by state HFAs. This is the case with Floridas housing trust fund, the State Housing Initiatives Partnership (SHIP) program whic h was established under the William E. Sadowski Act of 1992 with a dedicated revenue source (Larsen, 2009). SHIP had been the nations largest housing trust fund, until all its revenues were directed into the states general fund during the recessionary cl imate of 2009. The size of the tax credit increases when projects are located in difficult development areas (DDAs), locations where the cost of housing is high relative to income, and in qualified census tracts (QCTs), which are tracts in which at least half of all households earn 60% or less of AMI. The maximum allowable rent for these projects ranges from 30 to 60% of 11 See Schwartz (2006) for a thorough explanation of the formula used to determine tax credits.

PAGE 38

38 AMI, and is determined by the proportion of units reserved for very low and low income households. Attesting to the significance of this program, Melndez, Schwartz, and Montichard (2008) find that the LIHTC is by far the most important source of subsidy for low income multifamily rental housing in the US, having produced nearly 2 million units since its inception (p. 67). Implications of the generational shift Devolution of responsibility for assisted housing production has affected far more than the level of government responsible for these activities. As a consequence of the p rogrammatic changes marking the generational shift, a shift has occurred in the types of households targeted and the actors involved in producing and owning assisted housing. Though these trends may be equally manifested in other states, this section will consider the implications of the generational shift in terms of the relationship between the federal government and Florida only. Ray et al. (2009) have identified three important effects of this shift. The first is the long term, growing emphasis on fami ly housing, in other words, the movement away from the production of assisted housing intended for occupancy by target populations, such as the elderly and disabled, and toward the production of units that serve the general population of low income households (Ray et al., 2009, p. 18). In the 1960s, HUD targeted assistance toward the production of housing for the elderly and disabled through the Section 202 program, but in the 1970s HUD began to subsidize the development of assisted housing for both family and elderly households: Only 13% of HUD subsidized units in the 1970s received funding from the Section 202 program (Ray et al., 2009, p. 19). While HUD retooled Sections 202 and 811 in the 1990s, renewing federal interest in the production of assisted housing for populations with special needs, as a consequence of devolution

PAGE 39

39 and state HFAs largescale development of assisted family housing, the trend was not reversed. Fully 87% of the HFAs inventory consists of family units (Ray et al., 2009). The se cond effect of the generational shift is the transition from predominantly nonprofit ownership of assisted properties to for profit ownership. In the arena of assisted housing production, HUD has generally favored nonprofit ownership and has even required it, as in the cases of the Sections 202 and 811 programs. In all, 63% of HUDassisted units are owned by nonprofitsin contrast, 89% of Florida Housing assisted units are owned by for profits (Ray et al., 2009, p. 20). Consequently, older, HUDassiste d housing is more likely to be under nonprofit ownership than newer FHFC assisted properties. Furthermore, the number of for profit developments assisted by FHFC since the 1980s considerably outweighs the number of projects under nonprofit ownership developed from the 1970s onward. The third effect of the generational shift identified by Ray et al. (2009) is the movement away from the production of assisted properties capable of serving very low income households. This effect is a consequence of the deep subsidies offered under the earlier HUD programs being replaced with shallower subsidy mechanism of the LIHTC and stateissued mortgage bonds. As Khadduri and Wilkins (2006) note [b]ecause [LIHTC] has flat rents rather than percent of income rents, it does not easily reach households with extremely low incomes or incomes below the poverty line (p. 25). Similarly, in an examination of the use of the LIHTC and HOME resources, Mueller and Schwartz (2008) find that the poor are rarely the beneficiaries of st ate and locally funded programs (p. 131). Indeed, only four% of assisted units developed through use

PAGE 40

40 of the LIHTC were affordable to households earning 30% or less of AMI (Mueller and Schwartz, 2008). Preservation Challenges Unlike public housing, the affordability periods for assisted housing produced by for profit and nonprofit developers with direct federal subsidies are ultimately limited. The amortization period of 40 years for the BMIR loans issued under Sections 221(d)(3) and 236 defines the window of affordability for these programs, and rental subsidies provided under the project based Section 8 program were issued on a contractual basis, meaning that their continued affordability is contingent upon contract renewal. Properties financed through the LIHTC face similarly finite affordability periods, particularly the first cohort built. Physical deterioration, usually resulting from deferred maintenance, also places assisted properties at risk of failing out of their subsidy program. In addition, p roperties that default on their subsidized mortgage as a result of their inability to cover their debt service are at risk of leaving the assisted inventory. Thus, as Ray et al. (2009) find, older properties subsidized through the HUD programs and the LIHT C program face two countervailing pressures that may result in losses to the subsidized housing inventory: opt out or time out risks, and fail out risks (p. 22). Consistent with the literature, those properties of the earlier generation produced thr ough Sections 221(d)(3) and 236 will be referred to as the older federally assisted stock, and those produced under the Section 8 NC/SR program are referred to as newer federally assisted stock (Schwartz, 2006; Smith, 1999). Older Assisted Stock Although the subsidized mortgages given to developers of assisted housing under Sections 221(d)(3) and 236 were typically structured with a 40year amortization period,

PAGE 41

41 the length of which was intended to assure the long term affordability of these projects, owners were frequently extended the option of prepaying their mortgage after 20 years as an added incentive to participate in these programs.12In addition to the more immediate threat to the assisted housing stock posed by projects with prepayment agreements, properties ineligible for prepayment remain at risk of timing out with the maturation of their 40 year subsidized mortgages. When mortgage maturity is reached, affordability restrictions are dissolved unless they are According to Achtenberg (2002) these incentives induced the construction of some 560,000 units of prepayment eligible housing during the late 1960s and early 1970s (p. 2). Under certain conditions, owners would have strong incentives to opt out of the assisted housing inventory by pre paying their mortgage and converting their properties to market rate. Schwartz (2006) identifies two conditions under which this scenario is likely. These occur when the property has appreciated significantly since its development, which is often the case in desirable neighborhoods, resulting in the continued loss of the cream of the ol der assisted inventory (Smith, 1999). As a consequence, rents in the remaining older assisted properties are 10 to 25% below HUDs estimated FMRs (Smith, 1999, p. 151). The second condition under which owners of the older assisted stock are likely to pr e pay occurs when the tax benefits of ownership become exhausted: with depleted depreciation and mortgage interest deductions no longer off setting taxable income, the typical prepayment eligible project has become a tax liability for its owner (Achtenb erg, 2002, p. 2). 12 It should be stressed that not every older assisted unit is eligible for prepayment. These ineligible properties include those that have accepted Section 8 rental assistance in addition to their subsidized loans. Nearly 450,000 of the 650,000 units remaining in the older assisted stock receive assistance from the Section 8 LMSA program (Schwartz, 2006).

PAGE 42

42 attached to additional sources of funding for the property. A U.S. Government Accountability Office (2004) report found that subsidized mortgages for 1,333 properties (139,716 units) issued under Section 236 and for 502 properties (56,573 units) issued under Section 221(d)(3) are scheduled to mature by 2013. The expiration of rental assistance contracts attached to the older assisted stock poses a further t hreat to their preservation. In order both to avoid defaulting on their mortgage and to assist low income tenants with inflationary budget based rents during the 1970s, many properties developed with interest rate subsidies entered into Section 8 Loan Management Set Aside (LMSA) rental assistance contracts (Achtenberg, 2002).13In addition to the contractual sources of concern raised above, many properties developed under the interest rate subsidy programs are a t risk of failing out of the assisted housing inventory as a result of physical deterioration. This situation is primarily the result of the inability of some owners to cover both operating costs and to maintain an adequate reserve fund for later capital investments, such as roofing and Indeed, nearly 80% of the older assisted stock receives Section 8 funding (Schwartz, 2006). Schwartz (2006) observes that unlike the newer assisted stock [Section 8 NC/SR], rents in these properties tend to be relatively low. As a result, when Section 8 subsidy contracts expire, owners have considerable incentive to leave the program and convert to market rate housing (p. 135). These contracts were typically short term, consisting of an initial five year period with the possibility of renewal for two additional fiveyear periods (Achtenberg, 2002). 13 This is in addition to the earlier stoc k of units developed under Sections 221(d)(3) and 236 that had accepted rental subsidies under the RP and RAP programs (which were subsequently renewed under Section 8 LMSA) in order to house lower income residents.

PAGE 43

43 plumbing, while being required to keep rent at a predetermined level. Below market rents were maintained though budget based rent increases that took only operating costs into considerationreflecting only a fraction of the total budget for an assisted property (Smith, 1999, p. 146). This practice of budget basing had the net effect of slowly starving properties of capital, often desperately needed for capital reinvestment (Smith, 1999, p. 146). As a consequence, the long term viability of many of these properties was seriously compromised. Poor physical conditions pose risk to properties developed under Section 8 NC/SR as well. These cases usually result from the owners failure to budget for capital improvements and invest rental income accordingly (Schwartz, 2006, p. 137). Those owners choosing to cover the costs of capital investments in the face of limited income face the possibility of foreclosure, which is another means whereby properties may fail out of the as sisted inventory (Pedone, 1991). Newer Assisted Stock Properties developed through the Section 8 NC/SR program were frequently subject to contracts binding them to a long term affordability period ranging from 20 to 40 years (Achtenberg, 2002). At the end of this term, owners of these properties had the ability to opt out of the program, allowing them to convert their properties to market rate. The costs of subsidizing project based Section 8 properties, however, have proven an obstacle to their preser vation due to the rents in these properties rising disproportionately relative to peer properties. Although rents at individual Section 8 projects were increased every year by a HUD calculated annual adjustment factor, rents at surrounding apartment buildings did not necessarily grow at the same pace (Schwartz, 2006, p. 135). These properties placed an enormous strain on HUDs

PAGE 44

44 budget, making them wary to renew affordability contracts. The agencys initial response to the expiration of affordability periods was to substantially decrease the length of the rental assistance contracts offered upon expiration of the original long term contracts, reducing them to five years and later to only one year (Schwartz, 2006). However, as both long and short term contra cts began to expire at the same time in the mid 1990s, the cost of renewing all of these contracts placed HUD in a particularly precarious situation, where the cost of renewal threatened to consume the entire HUD budget, while the cost of inaction would trigger staggering claims against the HUD mortgage fund (Achtenberg, 2002, p. 4). L ow I ncome H ousing T ax C redit Properties At the time of the LIHTC programs creation, property owners were required to preserve the affordability of units financed with th e tax credit for a minimum of 15 years. Melendz, Schwartz and de Montrichard (2008) identify three year 15 challenges. First, income and rent restrictions for LIHTC projects expire at this point unless they are subject to further restrictions attached t o additional funding sources. Owners have the option of converting to market rate, which is especially likely when the potential to earn higher rents presents itself. Second, year 15 signals the dissolution of the limited partnership established to finance and develop these projects, thus requiring the sponsor to either acquire the limited partners stake in the development or sell the development altogether (Melendz, Schwartz, & de Montrichard 2008, p. 68).14 14 Sponsors are the for profit and nonpr ofit organizations with managerial responsibility for the LIHTC properties. Limited partners are those that purchase interests in tax credit properties and have no managerial authority over them. Third, many properties are also likely to require funding for major repairs and renovations. This

PAGE 45

4 5 is especially a problem for the earliest LIHTC projects. The smaller amount of equity generated for these projects from the sale of tax credits translated into modest renovations compared to the new construction and gut rehab that was to characterize most tax credit developments from the mod1990s onwards (Schwartz, 2006, p. 97). Thus, these more modest projects are more likely to be in need of repair at year 15 and less likely to have the reserve funds required to cover these necessary capital improvements. These conditions make older tax credits quite susceptible to failing out of the assisted housing inventory. An extendedaffordability period of 15 years beyond the initial compliance period was mandated for LIHTC projects through legislation in 1989 and 1990;15Preservation Responses however, owners retain the ability to opt out provided certain conditions are met. For instance, the owner of a tax credit property seeking to sell it can opt out if the state HFA administering the program is unable to find a buyer able to both purchase the property and retain affordability (Collignon, 1999). In 2002, the affordability restrictions for the first 23,000 properties to be developed wit h the LIHTC expired, triggering concern regarding the preservation of the rest of the LIHTC inventory (Collignon, 1999). Different strategies have been employed at both the federal and state levels to preserve the nations stock of assisted housing. These strategies have taken different forms based on the subsidy mechanism of the original program. While the federal 15 The affordability period for LIHTC projects in Florida is set at 50 years. In the case of projects with longer affordability periods, the issue then becomes their physical and financial integrity. That is, while they are less at risk of opt out, they are still susceptible to failing out.

PAGE 46

46 government initiated the preservation response in the 1980s, the responsibility for preservation has increasingly devolv ed to the state and local levels. Older Assisted Stock The earliest preservation efforts emerged at the federal level as the older assisted stock entered the period of eligibility for prepayment. By the late 1980s, Congress became concerned that a large number of owners might take advantage of the prepayment clauses within a short period of time, thus dramatically reducing the supply of low income housing throughout the country (Peiser, 1999, p. 372). The Emergency Low Income Housing Preservation Act (ELHIPA) of 1987 and the Low Income Housing Preservation and Resident Homeownership Act (LIHPRA) of 1990 were enacted by Congress in order to prohibit prepayment by eligible owners. While these acts granted owners financial compensation in the form of subsidi es to provide them with the equivalent of FMR, they generated a series of lawsuits and prepayment rights were restored in 1996 (Peiser, 1999). In the wake of this failed effort to mandate the continued affordability of the older assisted stock, [t]he goal of federal policy shifted dramatically from preserving the housing to protecting existing residents from displacement (Achtenberg, 2002, p. 2). These protections were accomplished through the use of enhanced vouchers, which cover the difference between 30% of the tenants income and FMR. If a tenant issued the enhanced voucher moves out, that unit is no longer subject to affordability restrictions, meaning that the owner may charge as much as the market may bear. In order to retain Section 236 properties at risk of leaving the assisted housing inventory through mortgage prepayment, the federal government allows owners and purchasers to refinance their existing mortgage and continue to receive the same

PAGE 47

47 interest reduction payments (IRPs) used to lower thei r debt service on their new mortgage through a mechanism called Section 236 IRP decoupling (Achtenberg, 2002).16Section 8 Preservation Strategies The increased income from the subsidy may be used to rehabilitate the structure and to establish a budget reserve for capital improvements. Owners who refinance and receive this subsidy must continue to observe affordability restrictions for a period of five years beyond the point at which the original mortgage would have matured (Affordable Housing Study Commission, 2006; Schwartz, 2006). In the late 1990s, Congress adopted two approaches for the preservation of at risk Section 8 properties. These programs differ based on the market in which properties are located as well as whether or not they have additional rent restrictions, such as those imposed by the older assisted stock that accepted Section 8 LMSA subsidies. In essence, these two programs either increase rents on properties where they fall appreciably below market or restructure the debt on properties wher e rents are artificially high. Rent restructuring In 1999, Congress introduced the Mark up to Market program in order to prevent properties with below market rents from opting out of the assisted inventory, which is especially likely for those located in strong rental markets. This program was particularly influenced by the large number of Section 8 properties opting out as the rental market was heating up at the same time that their affordability restrictions were expiring (Achtenberg, 2002). Properties w ith below market rents most commonly consist of the 16 The decoupling program allows the IRP to be retained and continued after the Section 236 mortgage is prepaid and refinanced (Achtenberg, 2009, p. 1).

PAGE 48

48 older projects receiving mortgage subsidies under Sections 221(d)(3) and 236 receiving additional assistance through the Section 8 LMSA program (Schwartz, 2006). To incentivize owners to renew their rental assistance contracts, the Mark up to Market allows subsidized rents to be marked up to comparable market rents with a maximum level of 150% of FMR. In exchange, owners must enter into a rental assistance contract with a term of at least five years. An es sentially identical program exists for nonprofits called Mark up to Budget, allowing them to increase rents to market level should a demonstrated need exist. Debt restructuring Debt restructuring was the approach taken by the federal government to address the preservation dilemma for Section 8 properties with abovemarket rents. This problem resulted from owners being allowed to charge abovemarket rents in order to cover development and operating costs for older Section 8 NC/SR properties receiving an FHA insured mortgage. As mentioned above, these properties continued to receive an annual increase in rent based not on market conditions of peer properties, but on a static adjustment factor. Renewing rental assistance contracts at these elevated levels was untenable, but so was allowing them to default on their government insured loans. The response to this crisis was the Mark to Market (M2M) program, which is a set of financial incentives designed to encourage nonprofit and for profit owners to restructure the debt on their properties, underwritten to a lower rent (Affordable Housing Study Commission, 2006, p. 12). In addition to a reduced mortgage capable of being serviced by lower rents, the M2M program provides additional funds for rehabilitation and ot her expenses. Properties restructured through M2M program are subject to a new rental assistance contract preserving their affordability for 30 years.

PAGE 49

49 State Preservation Initiatives In the context of devolution, states have had to assume a great deal of re sponsibility not only for the production of new assisted housing, but also for the preservation of the federally assisted housing inventory (Nenno, 1991). Since the termination of the LIHPRHA [Low Income Housing Preservation and Resident Homeownership Act ] and ELIHPA [Emergency Low Income Housing Preservation Act] programs, responsibility for preserving HUD assisted housing has increasingly devolved to the state and local level. Todays federal preservation toolsare not adequately designed or funded to pr eserve all of the units that remain at risk (Achtenberg, 2002, p. 22). As the projects in the federally assisted, first generation inventory have significantly deeper subsidies attached to them than tax credit projects, states have increasingly recognized the importance of preserving them for their lower income residents (Affordable Housing Study Commission, 2006). Furthermore, the combination of diminished resources and soaring development costs make preservation a particularly attractive endeavor as compared to new construction. The primary organ through which states work to preserve the federally assisted inventory is their HFAs, which are capable of allocating tax credits and using other sources of financing, such as tax exempt bonds and HOME grants, for preservation activities (Affordable Housing Study Commission, 2006; National Housing Trust, 2004).17 17 While a host of regulatory mechanisms for preserving assisted housing have been developed at the state and local level, the focus of this study is on the prioritization of properties for financial incentives by HFAs. These regulatory mechanisms include such tools as right of first refusal and extended notice requirements (cf. Achtenberg, 2002). In the past decade, many HFAs have expressed their concern for preservation through the inclusion of preferences and set asides for preservation related activities in their Qualified Allocation Plans (QAPs), which are federally mandated planning documents required of all agencies responsible for the allocation of the LIHTC

PAGE 50

50 (Gustafson & Walker, 2002; National Housing Trust, 2004). According to a survey conducted by the National Housing Trust (2008), 47 states prioritize preservation in their QAPs, with 25 states having set aside tax credits expressly for preservation.18Preservation Inventories Through their use of the LIHTC, HFAs have been able to preserve more than 280,000 units (Na tional Housing Trust, 2008). In 2008, Floridas HFA, the Florida Housing Finance Corporation (FHFC) allocated funding for the creation of a pilot preservation bridge loan program providing below market short term loans for preservation activities (Florida Housing Finance Corporation, 2009). In order to provide an accurate representation of the quantity and quality of the assisted housing stock and to guide the effective allocation of scarce resources in preservation activities, state and local entities have begun to create preservation inventories, the purpose of which is to collect available data on the existing affordable rental housing stock, facilitating analysis of the portfolio and identification of at risk properties (Cen ter for Housing Policy, 2009, p. 1). When aggregated, these inventories have the potential for providing a national preservation data infrastructure, which would allow comparative and collaborative efforts between localities and at the national level (Sh imberg Center, 2007). These inventories collect such information as: Project name Address Funding program(s) (e.g., Section 236, Section 8 NC/SR) T arget population (e.g., elderly, family) Total number of units Number of affordable units Years affordability restrictions begin and expire 18 Floridas 2009 QAP has a four million dollar set aside of tax credits for qualifying preservation projects (Florida Housing Finance Agency, 2009).

PAGE 51

51 Physical condition of the property (Roset Zuppa, 2008). The Shimberg Center for Affordable Housing (Shimberg Center) has worked to develop a risk assessment method for analyzing the assisted housing inventory. To this end, it has collected data on properties that have the potential to indicate whether a property is at risk of opting or failing out, creating a distinct profile for each risk type (Ray et al., 2009; Roset Zuppa, 2008; Shimberg Center, 2008). Additional variables included in these profiles include indicators of the financial condition of the project (e.g., loan to value ratio, debt coverage ratio, and financial reserves), market conditions, and ownership type (e.g., for profit, nonprofit) (Roset Zuppa, 2008). These variables have been used to create risk scores for projects in Floridas assisted housing inventory. Quantifying the risk of a property leaving the assisted housing inventory is a critical step in the on going effort to preserve affordable housing. However, being at high risk of leaving the assisted housing inventory is a necessary, but potentially insufficient measure of worth for preserv ation efforts. According to the Shimberg Center (2008), [q]uantifying the risk to a set of properties is one step in setting preservation priorities, with the next step being to prioritize properties by their value to the affordable housing stock (p. 2 0). The presence of units affordable to lower income, especially ELI households is an obvious measure of value for properties in the assisted housing inventory, as these households face enormous difficulties in finding housing whose costs do not impose a significant burden (HUD, 2003). Within the assisted housing inventory, properties receiving project based rental assistance under Section 8 and related programs are far more likely to offer housing that is affordable to this demographic. In recognition of this fact, the Affordable Housing Study Commission

PAGE 52

52 (2006) suggests that the highest priority should be given to preserving these properties (p. 4). In addition to the affordability of their assisted units, the location of properties in the inventory should also play a significant role in determining their value, and hence, their relative prioritization for preservation initiatives. Indeed, as Briggs (2005a) observes, location, asevery realtor knows, helps define the real value of ones housing (p. 5). Hartman (1998), in making his famous case for a right to housing, argues that the social and physical characteristics of the neighborhood environment are just as necessary as the affordability and condition of the housing units themselves (p. 237). While location is taken into account in assessing the assisted housing inventory in so far as market conditions influence the likelihood of a property opting or failing out, location, as a substantial body of literature shows, also structures the opportunities and outcomes for low income households living in these units. The locations of employment opportunities, health care facilities, quality schools, supermarkets, and other critical resources, far from being homogenously distributed across metropolitan space, constitute a highly uneven geography of opportunity that places low income and minority households at a particular disadvantage (Briggs, 2005b; Galster & Killen, 1995). Thus, properties in the assisted housing inventory occupying a felicitous location within the geography of opportunity should be considered of high value. Summary Assisted housing is a critical resource in meeting the nations considerable affordable housing needs, but the affordability periods for most of these properties is limited. W ith the expiration of rental assistance contracts and the maturation of

PAGE 53

53 subsidized mortgages, owners may decide to convert their properties to market rate housing. Assisted units may also be lost to the affordable housing stock through physical deteriorati on and default. In order to prevent owners of assisted housing from failing out or opting out, initiatives at all levels of government have been taken to preserve the ongoing affordability and functionality of assisted housing. In the context of federal devolution, responsibility for preservation has primarily fallen to state and local agencies. State agencies have expressed an interest in preserving older assisted properties receiving federal subsidies, as they are better able to serve ELI residents. This shift from deep to shallow assistance reflects a larger programmatic change in assisted housing development, referred to as the generational shift. In order to inform preservation initiatives, a national data infrastructure is being developed. At present, it is able to provide an accurate picture of risk, but not of the overall value of assisted units to the affordable housing stock. Data reflecting locational characteristics may assist in developing a more robust picture of their value. Indicators of locational value for assisted value may be found in literature of the geography of opportunity, which examines the impact of location on family and individual outcomes. This concept and the empirical evidence that supports it are presented in the next chapter.

PAGE 54

54 CHAPTER 3 THE GEOGRAPHY OF OPPORTUNITY The concept of geography of opportunity emerged from research concerned with the effects of continued urban sprawl and raceand class based segregation on the differential ability of families and individuals to achieve a high quality of life (Briggs, 2005b; Galster & Killen, 1995; Ihlanfeldt, 1999). Writers discussing the topic were particularly concerned with debunking the existence of equal opportunity by placing the concept in an explicitly spatial context (Dreier, Mollenkopf, & Swanstrom, 2004; Galster & Killen, 1995). The results of this research have emphatically proven the hypothesis that where individuals live affects their opportunities and life outcomes (Rosenbaum, 1995, p. 231). Indeed, place shapes and constrains our opportunities not only to acquire income, but also to become fully functioning members of the economy, society, and polity (Dreier et al., 2004, p. 28). Galster & Killen (1995) argued that geography not only determines the quality and ac cessibility of key markets, institutions, and service delivery systems, but also the strength of social networks and the normative values of a community, both of which factor strongly into family and individual outcomes. Despite the relative youth of the t erm, the concept behind the geography of opportunity is rooted in earlier work by John Kain (1968) and Willam Julius Wilson (1987), who have persuasively argued that place matters for individual outcome. This chapter presents a number of key theories and e mpirical studies concerned with the linkages between geography and opportunity. Spatial Mismatch, JobsHousing Imbalance, and Location Efficiency The spatial mismatch hypothesis (SMH) is one of the most extensively researched dimensions of the problemati c relationship between geography and opportunity in

PAGE 55

55 metropolitan America. First advanced by John Kain in the 1960s, the SMH maintains that post war patterns of employment deconcentration and residential segregation have conspired to constrain the employment opportunities of lower skilled inner city minorities (Kain, 1968). As lower skilled jobs, such those found in the manufacturing and retail sector, began to move to the suburbs and beyond in the 1960s, minority workers have found it difficult to follow them due to discriminatory practices in the housing market. This has resulted in a spatial mismatch between the supply of low skilled labor in the central city and the demand for low skilled labor at the metropolitan level, which is exacerbated by the lack of adequate transportation options for workers seeking to commute from the central city to the jobrich areas and by poor information about distant job opportunities (Kain, 1968). Cumulatively, these conditions may result in greater difficulties in finding and securing jobs, lower earnings, longer commute times, and greater transportation costs for minority workers than similarly qualified white workers (Ihlanfeldt & Sjoquist, 1998). Since the publication of Kain seminal 1968 article, Housing Segregation, Negro Employment, and Metropolitan Decentralization, the SMH has inspired a large research literature. Although interest in the subject waned in the 1970s, Kasarda (1989) and Wilson (1987) renewed interest in the SMH during the 1980s and 1990s through t heir emphasis on job dencentralization and industrial restructuring as key causative factors in the creation of an urban underclass.19Following Kain (1968), numerous studies have tried to test the relationship between the spatial separation from employm ent opportunities and adverse labor market outcomes for minorities. Ihlanfeldt & Sjoquist (1998) found that most of these 19 Kasarda (1989) defines the underclass as an immobilized subgroup of spatially isolated, persistently poor ghetto dwellers (p. 27).

PAGE 56

56 studies have employed one of three methodological approaches: racial comparisons of commuting times or distances, correlations of labor market outcomes to job accessibility, and comparisons of labor market outcomes for inner city and suburban workers. In the early 1990s, six comprehensive reviews of SMH research were published, all but one of which supported the hypothesis to varying deg rees (Holzer, 1991; Ihlandfeldt, 19992; Jencks & Mayer, 1990; Kain, 1992; Moss & Tilly, 1991; Wheeler, 1990). In a subsequent literature review, Ihlanfeldt and Sjoquist (1998) found ample empirical evidence in support of the SMH. An important conclusion Ihlanfeldt and Sjonquist (1998) reached from their review is that significant geographic variations in mismatch exist, and in areas with high levels of housing segregation and poor transportation for reverse commuters, mismatch may play a more dominant role in explaining the market problems of the inner city poor (Ihlanfeldt & Sjoquist, 1998, pp. 880 881). Among recent studies, Raphaels (1998) test of the link between intrametropolitan accessibility to jobs and labor market outcomes for black youth in San Francisco provides strong evidence in support of the SMH. In contrast to previous research, Raphael (1998) employed intrametropolitan variations in job growth rather than variations in employment levels as an accessibility measure. Raphaels (1998) results show that differential accessibility to areas of high employment growth is sufficient to explain between 30 and 50% of the racial differential in neighborhood youth employment rates (p. 109). At the national level, Stoll (2005) found that metropolitan areas with higher levels of employment deconcentration (job sprawl), measured as the share of a metropolitan areas employment located outside a fivemile radius from the

PAGE 57

57 central business district (CBD), exhibit a greater degree of spatial mismatch bet ween employment and black residents. In an analysis of the 150 largest metropolitan areas, Holzer and Stoll (2007) found that while the share of blacks and Latinos in the suburbs grew significantly during the 1990s, job growth occurred in newer, highincom e suburban regions. As a result, residents of lower income suburbs must now commute to other suburban regions for work, perpetuating the spatial mismatch. These results challenge the idea of spatial mismatch as a strictly urban/suburban phenomenon, and fin d support in Orfields (1997) important work on the decline of older suburbs. While Kains (1968) original formulation of the SMH was specifically concerned with employment outcomes for black men, subsequent research has extended the hypothesis to countenance a wider range of the urban poor. Expanding the racial dimension of the SMH, Ihlandfeldt and Sjoquist (1989) found that both white and black low income males were likely to suffer from reduced earnings as a result of job decentralization. Kasarda and Ting (1996) found that spatial mismatch has a stronger effect on women than men in terms of joblessness, regardless of race. They suggest that this might be a consequence of the domestic duties that fall disproportionately upon women, which make them less likely to take a job requiring longer commutes (Kasarda & Ting, 1996). Closely related to the concept of spatial mismatch is that of jobs housing imbalance, which is an important topic in transportation research. While the SMH is specifically concerned wi th the adverse impact of employment decentralization on poor central city residents, the jobs housing imbalance is concerned with the intrametropolitan mismatch between the location of jobs and affordable housing and its

PAGE 58

58 consequences in terms of traffic c ongestion and air pollution (Cervero, 1989). Cervero (1989) attributes this imbalance to several factors: fiscal and exclusionary zoning, growth moratoria, mismatches between worker earnings and housing cost, the shift to multiple wage earner households, and job turnover. While the efficacy of a jobs housing balancing policy have been hotly debated in terms of triplength and congestion reduction (Levine, 1989), the jobs housing balance literature has ably demonstrated the need for affordable housing in jobrich areas (Cervero, 1989, 1996; Weitz, 2003). For example, Cervero (1989) found that high housing costs and restrictive zoning in suburban areas have contributed to lengthy commuting times in San Francisco and Chicago. While workers may be able to secure affordable housing, the spatial mismatch between the location of work and home may significantly increase the share of household income spent on transportation. Thus, housing ostensibly affordable to lower income households may place a heavy cost burden o n residents when commuting costs resulting from this mismatch are taken into consideration (Lipman, 2006). As the term is generally used, location efficiency is a measure of the transportation costs in a given area. In more precise terms, however, location efficiency refers to the savings on transportation expenses achieved by households living in neighborhoods with high geographic accessibility, which in turn refers to the ease of reaching needed or desired activities (Handy & Clifton, 2001, p. 68). The concept of location efficiency emerged from research seeking to quantify the relationship between new urbanist design principles, specifically, compact, transit oriented development patterns, and automobile use. In their foundational study on the relationship between urban design,

PAGE 59

59 transit accessibility, and auto use in Chicago, Los Angeles, and San Francisco, Holtzclaw, Clear, Dittmar, Goldstein, and Haas (2002) found substantial evidence in support of their hypothesis that auto ownership and driving decrease as destinations become more accessible. Their results revealed that density (measured as households per residential acre) and transit accessibility (measured as daily average number of buses or trains available at transit stops within a walkable distance) were the strongest determinants of both auto ownership and vehicle miles travelled (VMT). Pedestrian and bicycle friendliness (the measure for which was derived from the density of the street grid, age of housing, and traffic calming bonuses) was also found to be associated with decreased VMT, though the relationship was weaker. Based on this research, it has been concluded that homes situated in close proximity to daily destinations and transit stops are highly location efficient. Given that transportation expenses are second only to those for housing, location efficiency offers considerable benefits to low income households (Lipman, 2006). In recognition of the cost saving benefits of highly efficient neighborhoods, entities in four US cities began to issue location efficient mortgages (LEMs) in the late 1990. LEMs are premised on the proposition that mortgage underwriting guidelines may be relaxed for homeowners in locationefficient neighborhoods because they have lower than average automobilerelated transportation expenses and more income available for mortgage payments (Blackman & Krupnick, 2001).20 20 Some research has found that households financing their homes with LEMS may be equally, if not more susceptible to default (Blackman & Krupnick, 2001). Though the cost savings may not be sufficient to allow lower income households to afford homeownership ( especially in tight housing markets), the proven reduction in transportationrelated expenses for households residing in locationefficient neighborhoods nevertheless offers a significant opportunity for improving the economic well being of lower income residents.

PAGE 60

60 Neighborhood Effects The study of neighborhood effects is principally concerned with understanding the impact of neighborhood environments on outcomes for families and individuals in such areas as educational attainment, labor force participation, and criminal activity. In their exhaustive review of the neighborhood effects literature, Ellen and Turner (1997) identify six such mechanisms through which neig hborhood effects operate: quality of local services, socialization by adults, peer influences, social networks, exposure to crime and violence, and physical distance and isolation. Additional mechanisms identified in the literature of neighborhood effects are continued exposure to stressful conditions (Dreier et al., 2004; Ellen, Mijanovich, & Dillman, 2001; Geronimus, 2006) and exposure to environmental hazards (Bullard, 2000; Pastor, 2001). These mechanisms are held to vary in their level of operation dep ending on gender, age, and other personal characteristics (Ellen & Turner, 1997). The following review of the neighborhood effects literature is divided into separate sections for mechanisms and outcomes. Mechanisms The accessibility and quality of local services delivered at the neighborhood level can have a significant impact on individual outcome (Ellen & Turner, 1997). The specific public school an individual attends, particularly at the elementary school level, is usually determined by neighborhood of residence. If the school is of poor quality, individuals forced to attend them due to residential location are likely to suffer from diminished educational attainment relative to individuals attending higher quality public schools. Other services whose ac cessibility and quality vary with regard to residential location include preschools and daycare centers. In lower income neighborhoods, these

PAGE 61

61 resources are likely to have reduced human and material resources, such as qualified staff and learning materials. Consequently, children may receive less attention and supervision, be less challenged and stimulated, and ultimately be disadvantaged when they attend school (p. 837). For adolescents, access to afterschool programs may play a role reducing the likelihood of involvement in dangerous and criminal activities. Of course, access to quality medical care can have significant outcomes for individuals and families regardless of age or personal characteristics. For example, increased incidences in asthma morbidit y among low income children have been partially attributed to deficiencies in access to quality medical care (Crain, Kercsmar, Weiss, Mitchell, & Lynn, 1998). Socialization by adults builds on Jencks and Mayers (1990) concept of indigenous adult influences, by which is meant the manner in which adults residing in a given neighborhood may influence the outcomes of children and adolescents by serving as role models. Adults communicate critical values to neighborhood youth, including the relative importance o f work and education, as well as information about community norms and expectations. Neighborhoods with low levels of labor force attachment among adults are likely to produce a social context in which the operations of the labor market and educational ins titutions are viewed with skepticism (Wilson, 1991). This mechanism has received considerable attention as a result of Wilsons (1987) contention that the absence of suitable role models has precipitated persistent social pathologies among the underclass. Operating along similar lines, peer influences refer to the effects that peers have on one anothers behavior. Just as in the case of socialization by adults, this mechanism too may result in either positive or negative

PAGE 62

62 outcomes. Research that emphasizes t his mechanism generally adopts an epidemic model in their explanations of how peer influences spread (Jencks & Mayer, 1990). Case and Katzs (1991) research supports this model; they found that the behavior of neighborhood peers substantially youth behav ior in a manner where like begets like (p. 23). Social networks refer to the spatially contingent network of friends, family, neighbors, and local organizations available to individuals residing in a particular location (Ellen & Turner, 1997; Galster & Killen, 1995). The mechanism of s ocial networks is rooted in theories of social capital. Accord ing to Lang and Hornburg (1998), social capital commonly refers to the stocks of social trust, norms, and networks that people can draw upon in order to solve common problems (p 4). Coleman (1988), who introduced the concept identifies three distinct dimensions of social capital : obligations and expectations, information channels, and social norms. Obligations and expectations relate to the reciprocal nature o f assistance offered through social networks. These reciprocal relations are predicated on a structure of trustworthiness, without which there would be little incentive for residents to participate. Information channels are critical for learning about job opportunities and community news (Wilson, 1987), and social norms influence behavior through the mechanisms of adult and peer influence. Speaking to the importance of social capital, Coleman (1988) asserts that social capital is productive, making possibl e the achievement of certain ends that in its absence would not be possible (p. S98). Galster and Killen (1995) see local social networks as a key component of the metropolitan opportunity structure, as they shape individual

PAGE 63

63 perceptions regarding the abil ity of institutions and markets, such as education and the labor market, to provide advantageous individual outcomes. Exposure to crime and violence has obvious implications for individual outcomes, including physical and psychological harm and property t heft. For example, the risk of victimization is much greater in highcrime areas than others (Sampson & Lauritsen, 1994). This mechanism is capable of affecting all neighborhood residents, regardless of age, but exposure to crime and violence has effects other than outright victimization for neighborhood youths. Simply witnessing crimes or knowing people who have been victimized may also profoundly affect childrens outlook, leading them to see the world as fundamentally violent, dangerous, and unjust (El len & Turner, 1997, p. 841). Certain dimensions of this mechanism involve adult and peer influence as well; in neighborhoods where crime and violence are commonplace, they may be accepted by youths as normal and even expected activities (Case & Katz, 1991) In addition, individuals living in high crime neighborhoods are more likely to live sheltered existences, precluding them from taking advantage of social networks and neighborhood amenities, greatly reducing the quality of life for even nonvictims (Dreier et al., 2004; Ellen & Turner, 1997). Neighborhood effects are also held to operate through the cumulative impact of continued exposure to environmental stressors on health (Taylor, Repetti, & Seeman, 1997). These stressors include such conspicuous exam ples as crime and violence, but also the more routine stressors of noise, poor public services, and degraded environmental conditions. The relationship between stress and place is most frequently discussed in the context of the poor health outcomes experienced by residents of

PAGE 64

64 disadvantaged communities. In this regard, Geronimuss (2006) concept of weathering has been particularly influential.21The environmental justice/environmental racism literat ure has frequently discussed the relationship between neighborhood characteristics and health outcomes (Bullard, 2000). As initially stated, the weathering hypothesis holds that blacks experience early health deterioration as a conseq uence of the cumulative impact of repeated experience with social or economic adversity and political marginalization (Geronimus, Hicken, Keene, & Bound, 2006, p. 826). While weathering was originally used to explain racial disparities in health issues related to birth outcomes, the concept possesses significant explanatory in the context of health outcomes for a range of low income families and individuals residing in neighborhoods with poor levels of service and high levels of hazard (Dreier et al, 2004; Ellen et al., 2001). Illustrating the serious consequences of chronic stress, it has been found to cause cardiovascular disease and contribute to premature mortality (Taylor, Repetti, & Seeman, 1997). 22 21 The hypothesis that chronic stress may have a cumulative, adverse impact on both physical and mental health was first advanced by Selye (1978) in his landmark book, The Stress of Life Research in this area primarily concerns examining whether and to what degree land uses posing environmental hazards, such as garbage transfer stations, power plants, medical incinerators, and power plants are concentrated in minority or low income communities. A monumental work in the environmental justice literature, Bullards (2000) book Dumping in Dixie found enormous racial disparities in the siting of environmental hazards in the South. In their subsequent review of the environmental justice literat ure, Pastor, Sadd, and Hipp (2001) found that while the 22 Thought the direct health effects of these hazardous uses have proven difficult to verify (Ellen et al. 2001), they are no doubt related to neighborhood quality of life. In

PAGE 65

65 evidence is more often mixed than many activists have believed, the bulk of the research does seem to point to disproportionate exposures to hazards in minority neighborhoods (p. 3). In a frequently cited study in support of the environmental racism hypothesis, Boer, Pastor, Sadd, & Snyder (1997) analyzed of the location of hazardous waste treatment, storage, and disposal facilities (TSDFs) in Los Angeles County. They found that while considerations external to race and class such as zoning and industrial employment were associated with the siting of TSDFs, race and ethnicity were still positively associated with their location. Traffic, truck traffic in particular, is also related to poor environment al quality and adverse health outcomes through its impact on air quality (Dreier et al., 2004). The final mechanism identified by Ellen and Turner (1997) is physical distance and isolation. They find that the most straightforward impact of neighborhood is physical proximity and accessibility to economic opportunities, particularly jobs (p. 842). The only literature they include under the rubric of this mechanism concerns the spatial mismatch hypothesis, the importance of which concept merits its own sec tion in this literature review. However, recent studies on the absence of healthy and affordable food options in low income urban neighborhoods may also be included in discussions of the effects of physical isolation on individual outcomes. Stemming from B ritish research conducted in the later 1990s, an increasing amount of literature has been produced related to the existence and effects of food deserts which are defined as areas of poor access to the provision of healthy affordable food where population is characterized by deprivation and compound social exclusion (Wrigley, Warm, & Margetts, 2003). Research has shown that the type of food store available to neighborhood residents

PAGE 66

66 varies with neighborhood characteristics (Moore & Roux, 2006). Supermarket s, which offer healthy and affordable options, have been found to be more accessible in higher income areas, while convenience stores, which offer more expensive and less nutritious options, are found to be more accessible in lower income neighborhoods.23Outcomes I n addition, poorer neighborhoods were found to be more likely to have fewer fruit and vegetable stands, bakeries, and natural food stores, but more likely to have a greater number of liquor stores, than more affluent neighborhoods (Moore & Roux, 2006). Ref lecting the impact of neighborhood racial characteristics on individual outcome in the area of food access, a study of supermarket accessibility in Detroit found that predominantly black neighborhoods with high levels of poverty were on average 1.1 miles f urther from supermarkets than impoverished white neighborhoods (Zenck et al., 2005). Neighborhoods located in food deserts can have adverse effects on individual health, resulting in increased medical expenses and diminished quality of life. Individuals us ing public transportation to access supermarkets may face timeconsuming and inconvenient trips. While some of the literature of neighborhood effects discusses them principally in terms of the theoretical mechanisms through which they operate, ot her works discuss them in terms of specific outcomes for residents of varying age, gender, and other personal characteristics. Unlike those that discuss mechanisms, these studies are generally less concerned with causality than correlation and provide subs tantial 23 Even when supermarkets are located in urban markets, their prices are generally higher than those in suburban markets. In an examinat ion of 322 stores in 10 large metropolitan areas, MacDonald and Nelson (1991) found that the price of fixed market basket of goods was on average four percent higher in central cities stores than suburban stores.

PAGE 67

67 empirical support for the existence of neighborhood effects. As this is a broad body of literature, only the most salient threads will be presented here. Empirical research strongly suggests that neighborhood characteristics have an impact on childhood cognitive development. Perhaps the most frequently cited study in support of this hypothesis is that by Brooks Gunn, Duncan, Klebanov, and Sealand (1993). These authors quantified the effects of neighborhoods (operationalized at the census tract level ) on childhood outcomes using data from the Panel Study of Income the Infant Health and Development Program (IHDP). This program randomly selected 895 low birth weight children born in eight different sites across the US and observed the effects of educati onal and family support services and medical care on development. Controlling for family background, Brooks Gunn et al. (1993) found that IQs were significantly higher for children living in neighborhoods with higher concentrations of affluent households, defined as those with an annual income exceeding $30,000. This study also evaluated outcomes for adolescents using data from the Panel Study for Income Dynamics (PSID), which provides a lengthy time series of data related to family and neighborhood charact eristics for a randomized national sample. Using a national sample of black and white women between the ages of 14 and 19, the researchers found that the presence of affluent neighbors is also associated lower dropout levels and teenage pregnancies. As t he results of Brooks et al.s (1993) research suggests, impoverished neighborhoods adversely affect educational attainment. This particular outcome is supported by a substantial body of evidence. One such example is Garner and Radenbushs (1991) study of e ducational attainment among 2,500 people who dropped

PAGE 68

68 out of school between 1984 and 1986 in Scotland. After controlling for student ability, family background, and school, they found a significant negative relationship between neighborhood deprivation scor es and educational attainment. Garner and Radenbush (1991) calculated deprivation scores through a weighted combination of 12 census variables, among which the most heavily weighted were unemployment, singleparent families, low earning socioeconomic groups, overcrowding, and the percentage of permanently sick individuals. Testing a similar hypothesis, Crane (1991) studied the relationship between neighborhood quality and dropout rates. Crane (1991) used a 1970 Public Use Microdata Sample (PUMS) to examine teenagers living with their parents, selecting 92,512 teenagers for a sample group. Controlling for individual characteristics, Crane (1991) found the likelihood of dropping out of school for adolescents of all races increases exponentially as the percent age of workers in the neighborhood holding professional or managerial positions declined. While the pattern of neighborhood effects is relatively linear for Hispanics, the risk of dropping out for blacks and whites rises precipitously once the local share of middle class workers drops below 3.5%. Unlike Crane (1991), Duncan (1994) found that neighborhood effects on educational attainment differ based on race and gender. Using PSID data to measure the effects of neighborhood (census tract) and family charac teristics on schooling completed, Duncan (1994) found that higher levels of female employment reduced college attendance rates for both black and white women and that a larger share of femaleheaded households increased dropout rates for black females. In terms of the effects produced by a larger share of affluent neighbors, this neighborhood

PAGE 69

69 characteristic was found to increase the likelihood of college attendance only for white males, but it was also found to be a predictor of completed schooling for dropouts and college students. In a study of outcomes for participants in the Chicagos Gatreaux program, Rosenbaum (1995) found that there were significant differences between those that moved to the suburbs and those that moved to another part of the urban area. Rosenbaum (1995) conduced two separate studies to determine educational attainment; one test was performed by selecting one school aged child each from 114 families in 1982 (six years after the start of the program) and another was conducted on the same individuals in 1989. Based on the second test, suburbanmovers were shown to have improved outcomes in terms of both educational attainment and employment. Rosenbaums (1995) results showed that more suburban movers than city movers had attained a greater degree of academic achievement, were on a college track (40% versus 24%), were enrolled in college (54% versus 21%), were in four year colleges rather than junior colleges (50% versus 20%), and had obtained jobs paying more than $6.50 per hour (21% versus 5%). Also, fewer suburban movers dropped out of school than city movers (20% versus 5%). Teen pregnancy is another hypothesized outcome of neighborhood conditions supported by substantial research. Cranes (1991) study also employed a sample of 44,466 females between 16 and 19 years taken from the 1970 PUMS for a childbearing analysis. The results of this analysis are similar to those for his test of dropout rates; Crane (1991) found that there is a sharp increase in the childbearing probability for both black and white women when the share of professional or managerial workers drops

PAGE 70

70 below 3.5 percent. The results of Brooks Gunn et al.s (1993) study also support the argument that neighborhood characteristics are associated with out of wedlock childbearing among adolescents. Brewster (1994) examined the impact of neighborhood characteristics on racial differences in sexual activity among adolescent women, using both individual level data on women between the ages of 15 and 19 taken from the National Sur vey of Family Growth (NSFG) and census data. Her results showed that neighborhood socioeconomic statues and full time female employment rate are positively associated with risk of experiencing nonmarital intercourse during adolescence. Brewster (1994) concluded that racial differences in risk reflect racial differences in access to economic resources and positive female role models. Neighborhood conditions, in essence, alter the perceived incentive structure for black women in a manner such that the cost o f sexual activity seems low. Despite the evidence supplied by these findings, however, other recent studies have failed to find a link between neighborhood and adolescent sexual behavior, suggesting caution in interpreting the results of previous studies ( Ellen & Turner, 1997). In terms of the link between neighborhood characteristics and crime, Case and Katzs (1991) study provides the strongest evidence. Using National Bureau of Economic Research Data on 1,200 disadvantaged youth between the ages of 17 and 24 in Boston, Case and Katz (1991) found that a 10 percent increase in the teenage crime rate increased the likelihood of a youth committing a crime within the past year by 2.3%. This study also found that a given youth was 3.2% more likely to use illeg al drugs, 2.7% more likely to be friends with gang members, and 3.4% more likely to use

PAGE 71

71 alcohol weekly if the neighborhood was subject to a 10% percent increase in other youth exhibiting the same behavior. Neighborhood effects also have a demonstrated impact on health. While the link between individual wealth and health is now taken for granted, research has also found that low neighborhood socioeconomic status is associated with both a range of health problems and mortality from a number of causes (Dreier et al., 2004). Using a sample of 12,601 persons from four communities in the US and 1990 census tract data related to socioeconomic status (SES) in a multi level regression analysis, Diez Roux et al. (1997) found that neighborhoods with lower SES were as sociated with higher incidences of coronary heart disease than more affluent communities. In a longitudinal study of 1,129 adults in Alameda County, California Yen and Kaplan (1999) found that lower quality social environments (measured in terms of per capita income, degree of residential crowding, quantity of commercial stores, and ratio of homeowners to renters) were associated with an increased risk of death. The results of this study were significant even when controlling for individual characteristics such as age, income, gender, smoking status, body mass index, and alcohol consumption (Yen & Kaplan, 1999). In addition, residents in communities with fewer healthy food options have been shown to exhibit poor dietary habits, which may lead to diabetes, ob esity and other conditions. In a randomized control study of 22 communities nationwide, Cheadle et al. (1991) found a significant association between the availability of healthy products and their consumption. Studies also show that the density of fast foo d restaurants is greater in minority and lower income neighborhoods, identifying this as a possible environmental factor in the prevalence of obesity in disadvantaged neighborhoods. In a

PAGE 72

72 study of all fast food restaurants in New Orleans, Block, Scribner, & De Salvo (2004) found that predominantly black neighborhoods have 2.4 fast food restaurants per square mile, while white neighborhoods have only 1.5. Additional research performed in this area has been motivated by the desire to test the purported bene fits of housing mobility programs. In a review of the housing mobility and health literature, AcevedoGarcia et al. (2004) found that these studies support the general conclusion that moving to lower poverty neighborhoods may contribute to the improvement of health and healthrelated behaviors for both adults and children. While these effects have been shown to varying degrees along a host of healthrelated dimensions, evidence in support of neighborhood effects on mental health is the strongest. In a study of the short term impact of the Moving to Opportunity (MTO) program in New York City, Leventhal and Brooks Gunn (2003) found that parents who moved from highto low poverty neighborhoods experienced significantly less mental distress than those who remai ned in impoverished neighborhoods. Leventhal and Brooks Gunn (2003) employed a randomized controlled design using three groups: an experimental group of families participating in the MTO program, a comparison group of families receiving Section 8 vouchers, and a control group of families receiving project based assistance However, the results were varied with regard to age and gender. Boys in the experimental group were found to be significantly less likely to report problems with anxiety or depression than the in place control group, while no statistically significant differences appeared for girls. Boys between the ages of 8 and 13 in the Section 8 group were significantly less likely to exhibit headstrong behavior than the in-

PAGE 73

73 place control group, but no s ignificant differences were observed between the groups for youths aged 14 to 18 years old. Neighborhood Change Although suburbanization remains the predominant housing trend in the United States, many cities have witnessed the return of middleclass and professional households. This process has generated acclaim and disdain alike, largely as result of competing definitions of the process. Seeking to remedy the conceptual confusion frequently generated by the term, Kennedy and Leonard (2001) define gentri fication as the process by which higher income households displace lower income residents of a neighborhood, changing its essential character and flavor (p. 5). This is perhaps the most comprehensive and useful formulation of gentrification as it takes both the social and economic dimensions of the process into consideration. However, it is important to note that gentrification is neither inherently good nor bad; rather, it is capable of producing both positive and negative effects where it occurs. As a r esult, policy recommendations suggest allowing the process to occur in a manner conducive to realizing its benefits while simultaneously working to mitigate its negative consequences (Kennedy & Leonard, 2001). Positive aspects of gentrification include dec oncentration of poverty, reduction in crime, neighborhood revitalization, and increased real estate values and equity (Kennedy & Leonard, 2001; Sullivan, 2007). Negative aspects include changes in the traditional character of the neighborhood and conflict between community members; however, the most significant adverse impact of gentrification is the displacement of lower income residents through higher property values and their concomitant rent level increases. In a climate in which the housing market is t ight and the availability of affordable housing low, gentrification exacerbates the already difficult

PAGE 74

74 prospects of low income households. In order to reflect the potential good that may result from this process, as well as the fact that displacement does not always occur in areas that witness investment, this study, following Mallach (2008) adopts the term neighborhood change to refer to the complex totality of affects under review. Until recently, two dominant theories accounted for the occurrence of nei ghborhood change: demandside conditions and supply side conditions. Supply side arguments seek to explain neighborhood change through trends in the movement of capital and its effect on urban space. In the wake of urban disinvestment after midcentury, a widening margin has been created between actual and potential land values. This margin is referred to in the literature as a rent gap, which Smith (1987) defines as the gap between the actual capitalized ground rent (land value) of a plot of land given its present use and the potential ground rent that might be gleaned under a higher and better use (p. 462). Demandside arguments focus on consumer preferences for urban areas, which may be generated by such conditions as a growth in whitecollar jobs; life style choices including deferred childrearing, and the attraction of urban life to younger professionals (NeighborWorks America, 2005). Bridging these competing schools of thought, and reflecting the complexity of the causes of gentrification, Kennedy and Leonard (2001) offer several conditions under which neighborhood change is likely to occur, including rapid job growth, tight housing markets, consumer preference for city amenities, increased traffic congestion and lengthening commutes, as well as public policies facilitating the ingress of middleclass homebuyers. Regardless of a studys methodology, data collection on neighborhood change is considerably problematic. Several of these obstacles include the wide intervals between

PAGE 75

75 data collection at th e neighborhood level, the complex interaction of phenomenon ostensibly contributing to neighborhood change, as well as the problem of using regional data to account for specific localities (Kennedy & Leonard, 2001). As a result of the complexity associated with indicators, methodology and data collection, research in the areas of gentrification has produced contradictory results. However, two recent studies suggest that displacement, while it necessitates policy attention, is not as widespread and precipitous as previously thought (Freeman & Braconi, 2004; Vigdor, 2002). Counter intuitively, Freeman and Braconis (2004) study indicates that rather than speeding up the departure of low income residents through displacement, neighborhood gentrification was ac tually associated with a lower propensity of disadvantaged households to move (p.51). However, through residents may choose to remain, they may be considerably more cost burdened than before neighborhood change occurred. Thus it remains critical to identi fy and monitor indicators for neighborhood change in order to protect the existing stock of affordable housing Previous Studies on the Location of Assisted Housing Despite the federal governments stated policy objective of providing affordable housing in suitable neighborhoods, project based assisted housing as a whole has been found to be generally situated in relatively lower quality neighborhoods (Newman & Schnar e, 1997). However, this statement should be tempered with the observation that privately owned assisted housing units have been found to be located in generally better neighborhoods (measured in terms of racial segregation and poverty concentration) than p ublic housing units, which are frequently found in the most disadvantaged areas (Massey & Kanaiaupuni, 1993). Relatively few studies directly address the location of assisted housing in terms of neighborhood quality, but the ones

PAGE 76

76 that do exist provide important context for this study.24In a comparison of the neighborhoods surrounding assi sted properties in the U.S., Newman and Schnare (1997) found that privately owned assisted housing units (including both HUD and LIHTC properties) are significantly more concentrated in lower income neighborhoods and significantly less concentrated in upper income neighborhoods when compared to the total universe of rental housing (p. 711). In an examination of the differential neighborhood quality of properties developed under separate privately owned assisted programs, this study found that the proporti ons of HUDand LIHTC assisted units located in highly segregated areas were similarly high. Roughly onethird of the units developed under both the LIHTC program and other HUD programs were located in census tracts where minorities constitute more than 40% of all households. These results compare negatively with those for units occupied by certificate and voucher holders (one quarter) and statesubsidized units (one fifth) (Newman & Schnare, 1997). Newman and Schnare (1997) concluded that privately owned assisted housing programs do little to improve neighborhood quality relative to welfare recipients, who are used as a proxy for properties in areas with poor neighborhood quality, as they are most often found in neighborhoods with high poverty While an exhaustive review will not here be attempted, a few key studies tracing the geography of opportunity available to residents of assisted housing will be presented. 24 However, a substantial body of literature examines whether or not assisted housing has been developed in progressively less impoverished and racially segregated neighborhoods since the passage of Civil Rights legislation and the introduction of housing programs des igned to offer households greater options in terms of location. The results are mixed, but there is evidence supporting the hypothesis that assisted housing, especially under the LIHTC program, is being increasingly located in less segregated areas (Rohe & Freeman, 2001)

PAGE 77

77 rates. Their study found that much of the assisted housing stock is located in neighborhoods lacking suitable quality (expressed as economic status, housing quality, concentration of assisted housing, and racial/ethnic mix). Of particular relevance to this research, it suggests that there is little, if any, demonstrable difference in the neighborhood quality of properties developed under HUD programs or the LIHTC program. Subsequent studies, however, have found that a significant number of LIHTC units have been developed in areas of low to moderate poverty, rather than heavily concentrated poverty. Freeman (2004) analyzed the location and neighborhood characteristics of LIHTC properties developed in the 1990s in the 100 metropolitan areas with the largest numbers of uni ts. Freeman (2004) found that approximately 42% of all LIHTC units were located in the suburbs, compared to only 24% of all other privately owned federally assisted housing. Freeman (2004) also found that neighborhoods in which LIHTC properties were locate d experienced larger declines in poverty compared to median metropolitan values. On the other hand, Freeman (2004) also found that neighborhoods with LIHTC properties have higher poverty rates, lower median incomes, and lower median home values when compar ed to median metropolitan values. A subsequent analysis of LIHTC developments in Atlanta, Chicago, Los Angeles, and New York City by Oakley (2008) confirms Freemans results. Oakley (2008) found that neighborhood characteristics associated with other assi sted housing programs like low income, poverty, and unemployment are not significant predictors of the presence of LIHTC developments (p. 624). Oakley attributes this finding to more LIHTC units being located in the suburbs than assisted housing developed under other

PAGE 78

78 federal programs. Tempering these results, Oakley (2008) also found that the singlestrongest predictor of the presence of LIHTC units are the presence of qualified census tracts (QCTs), which are low income areas eligible for assistance under certain programs. These results, as Oakley (2008) commented, reflect the fact that under the LIHTC program, bonuses are awarded to developers building in QCTs. While many of the neighborhoods surrounding the assisted housing stock are characterized by hi gher relative concentrations of poverty and minority households, studies in some cities have found that privately owned assisted housing performs much better in terms of accessibility and location efficiency, which is an additional measure of value useful for prioritizing preservation initiatives. A joint study by the National Housing Trust and Reconnecting America (2008) found that a large share of federally assisted units in eight major metropolitan areas are located within a half mile of existing or proposed rail stations and bus stops, meaning that they are accessible to jobs and services for low income households. Their study found that Boston, Chicago, Cleveland, Denver, New York City, Portland, St. Louis, and Seattle collectively contain more than 100,000 units of privately owned federally assisted housing. The percentage of units near transit varied drastically between cities, however. In New York, 72% of federally assisted units are located within a half mile of an existing or proposed rail station. On the opposite side of the spectrum, only 9% of St. Louiss units were located within the same distance. Summary This chapter has presented the concept of the geography of opportunity, the literature that frames the issue, and the evidence that supports the claim that location is an important determinant of value for housing. Based on this literature review, several

PAGE 79

79 key indicators of housing suitability measuring the relative quality of the opportunity structure available to assisted housing residents hav e been selected for use in this study. The next chapter presents these indicators and the methodology that will be used in comparing different generations of the assisted housing stock.

PAGE 80

80 CHAPTER 4 METHODOLOGY The purpose of this study is to assess properties in the assisted housing inventory (AHI) across a number of suitability criteria based on the generation in which they were produced. Thus, the research design employed in this study compares the two different generations of properties within each study area. The research design consists of several phases. First, a geographic information system ( GIS ) based landuse suitability model is run for each of the study areas to find suitability for the purposes of af fordable housing across a number of criteria, each critical to determining the value of properties in the AHI. Next, properties in the AHI are assigned the suitability values generated from the model at their specific locations. After the suitability resul ts have been joined to the properties, they are subjected to statistical analysis. Descriptive statistics for the suitability variables of the two different generations will be determined in each study area, and statistical analyses will be performed to test for differences between the two generations for each of the selected suitability criteria. Affordable Housing Suitability Model This study employs the Affordable Housing Suitability (AHS) model in determining site suitability for properties in the AHI. The AHS is a GIS based land use suitability model designed to assist community stakeholders in addressing their affordable housing needs by identifying and assessing the suitability of sites for the production and preservation of affordable housing. The mo del functions as a planning support system (PSS) by using information technology to integrate community preferences, planning expertise, and key spatial data in the pursuit of this critical community planning objective (Harris & Beatty, 1993; Heikkila, 1998; Klosterman, 1997). The AHS model is a

PAGE 81

81 collaborative effort of the Department of Urban and Regional Planning and the Shimberg Center for Housing Studies at the University of Florida, and has been developed with the support of the Wachovia Foundation and the MacArthur Foundation. This project has been developed with the assistance of local planning agencies in three Florida Counties: Duval (Jacksonville), Orange (Orlando) and Pinellas (Clearwater and St. Petersburg in the Tampa Bay area). As the AHS model has so far been concerned with assessing the suitability of these three communities, they constitute the study areas selected for this study. Geographic Information Systems Many definitions of GIS exist, ranging from purely functional descriptions of soft ware applications and database capabilities, to more comprehensive definitions situating the technology in the social context of its development and use (Chrisman, 1997; Heikkila, 1998). Burroughs (1986) classic functionalist definition states that a GIS is a powerful set of tools for storing and retrieving at will, transforming and displaying spatial data from the real world for a particular set of purposes (p. 6). At the most fundamental level, all GIS integrate a mapping function, which displays maps or geographic features, with a database manager which organizes the attribute data tied to the various map features (Levine & Landis, 1989, pp. 209 10). Five basic types of functionality are conventionally cited in defining the capabilities of a GIS: the capture, storage, retrieval, analysis and display of spatial data (Clarke, 1986, p. 175). In addition to software, hardware and data, a critical component of a GIS is its liveware, that is, the people responsible for designing, implementing and using GIS (Maguire, 1991). A GIS also encompasses the institutional frameworks and cultural practices structuring its design and use (Chrisman, 1997; Maguire, 1991). Thus, a GIS is not simply a tool, but

PAGE 82

82 is rather an integrated system of technical and social relations (Innes & Simpson, 1993). In the context of the AHS model, Cowens (1988) holistic and goal oriented definition of a GIS as a decision support system involving the integration of spatially referenced data in a problem solving environment is part icularly relevant (p. 1554). LandUse Suitability Analysis While the storage and display of spatially referenced data is at the core of GIS functionality, the greatest strength of GIS is in creating new information and combining different sources of geog raphic data through the process of overlays (Harris & Batty, 2001, p. 185). One of the most useful applications of these powerful overlay processes is land use suitability analysis (Collins, Steiner, & Rushman, 2001; Malczewski, 2004). The purpose of a la nduse suitability analysis is to identify the spatial pattern of requirements, preferences, or predictors of some use (Hopkins, 1977, p. 386). In the case of the AHS model, multiple layers of spatial information are combined in order to produce a compos ite representation of suitability for affordable housing. Each of the distinct layers included in the model have been selected based on their utility in assessing the suitability of a particular site for affordable housing such as neighborhood conditions and distance from essential services. While each of these layers individually provides critical information relevant in assessing suitability for this purpose, the synthesis of this information through the overlay analysis generates entirely new informatio n useful in guiding intelligent community planning decisions. The GIS based landuse suitability analysis is rooted in overlay techniques developed by landscape architects using handdrawn sieve maps in the late nineteenth and early twentieth centuries (Co llins et al., 2001; Steinitz, Parker, & Jordan, 1976). These overlay techniques for determining landuse suitability were significantly

PAGE 83

83 advanced and widely popularized through Ian McHargs 1969 book Design with Nature (Collins et al., 2001; McHarg, 1992). McHargs (1992) procedure for overlay suitability analysis began with recording the social values assigned to features in the built and natural environment s on separate transparent maps, which are shaded on a scale of light to dark to represent varying deg rees of value. These individual maps were superimposed in order to produce a composite suitability map for a particular land use category. The darkest areas on this composite map have the highest social value, hence, the greatest cost associated with their conversion to another use. Conversely, the lightest areas on the composite map have the lowest social value attached to them, making them the most suitable (least costly) for conversion (McHarg, 1992). The McHargian approach is useful for visualizing the social costs of different landuse scenarios, especially the loss of ecologically impor tant features. While McHargs overlay technique has had a profound influence on the development of suitability analysis in GIS, computer assisted overlay techniques wer e developed in response to the limitation of handdrawn overlay mapping (Collins, 2001). Computer assisted mapping techniques advanced at Harvard in the 1960s significantly expanded the range and depth of landuse suitability analysis. An important milestone was Harvards development of SYMAP (synagraphic mapping system), which allowed individual maps to be overprinted using gray scales in order to produce a composite suitability map (Collins, 2001). Further advancements in computer based suitability analys is came through the formal development of GIS software and the introduction of map algebra and cartographic modeling techniques pioneered by Dana Tomlin (Collins, 2001). The integration of multicriteria (or multiattribute) decision making (MCDM) and

PAGE 84

84 multic riteria evaluation (MCE) methods with GIS has significantly advanced the capabilities of overlay suitability analyses by allowing the incorporation of preferences in a complex decisionmaking environment with multiple, and potentially conflicting, criteria (Colllins Steiner, & Rushman, 2001; Jankowsi, 1995; Malczewski, 2004).25 While GIS based land use suitability analysis has been applied in order to assess sites for a variety of activities, few have been directly concerned with identifying sites suitable for the location of affordable housing. Thomson and Hardin (2000) identify potential sites for affordable housing in Bangkok, Thailand, employing criteria primarily concerned with the physical characteristics of the study area, specifically, compatible Rather than operating in a simple Boolean yes/no decision making environment, these methods allow decisionmakers to determine the tradeoffs between criteria when conflict arises. G IS based landuse suitability models have been used in a variety of contexts, including environmental studies of land for habitat suitability (Pereira & Duckstein, 1993; Store & Kangas, 2001) and land conservation (Miller, Collins, Steiner, & Cook, 1998; T rust for Public Land, 2005), as well as for more urban purposes such as public facility siting (Higgs, 2006) and sustainable residential development (Sorrentino, Meenar, & Flamm, 2008). Suitability models have also been employed to predict future landuse scenarios (Carr & Zwick, 2005). 25Jankowski (1995) finds that the general objective of MCDM is to assist the decisionmaker (DM) is selecting the best alternative from the number of feasiblealternatives under the presence of multiple choice criteria and diverse criterion priorities (p. 252). According to Malczewksi (2004): GIS based MCDA can be thought of as a process that combines and transforms spatial and aspatial data (inputs) into a resultant decision (output). The MCDM procedures (or decision rules) define a r elationship between the input maps and the output map. The procedures involve the utilization of geographical data, the decision makers preferences and the manipulation of the data and preferences according to specified decision rules (p. 33).

PAGE 85

85 land use, proximity to transportation routes, and flood risk.26A ffordable H ousing S uitability Model Structure and Methodology These criteria, while critical for the purposes of affordable housing, are equally applicable for the siting of any residential development, regardless of affordability. In order to incorporate additional suitability criteria specifical ly related to affordable housing, Biermann (1999) created an overlay suitability model incorporating socioeconomic variables in addition to physical and environmental conditions Svatos and Doucette (2003) employed a similar approach in a GIS based suita bility analysis for affordable housing in two Delaware counties. Values of low, moderate, or high were assigned across the surface of the maps for each criterion based on subjective thresholds. These layers were then summed in order to represent the compos ite suitability of land for affordable housing. Recognizing that the selection of sites suitable for housing (of all types) requires a complex decisionmaking structure cutting across environmental, political, and social dimensions, Al Shalabi, Mansor, Ahm ed, and Shiriff (2006) applied MCDM techniques in a housing suitability analysis of Sanaa, Yemen. Fundamental to the design of the AHS model is the Analytic Hierarchy Process (AHP), an MCDM technique that decomposes complex suitability problems into a comprehensive and logically consistent hierarchical decisionmaking framework and employs pairwise comparison to establish factor weights at each level within the hierarchy (Banai Kashini, 1989; Saaty, 2008). 27 26 This st udy, however, does not employ an overlay analysis, but instead considers each criterion separately. Following the AHP technique, t he AHS 27 As all factors are assigned equal weights in this study, a more comprehensive explanation of AHP is beyond the scope of this paper. Equal weights are use d as the study area communities are still in the process of assigning priorities to suitability criteria through pairwise comparison.

PAGE 86

86 model is organized as a hierarchy of goals, objectives, and subobjectives containing sets of suitability criteria (see Figure 41). Due to the significant number of variables needed for a comprehensive affordable housing suitability analysis, the AHS model is comprised of three distinct goals: the identification of suitable sites based on preferences expressed for affordable housing characteristics, 28As currently structured, only the first goal (affordable housing) of the AHS model is designed to accommodate pairwise comparison, due to its complex hierarchy of objectives and subobjectives. The object ives contained in the first goal are: the identification of suitable sites based o n transportation costs, and the identification of suitable sites based on the demand for affordable housing driven by employment (see Figure 42). The generation of suitability values for each of the layers constitutes the first phase of the AHS model. A s econd phase, based on Carr and Zwicks (2007) landuse conflict identification strategy (LUCIS) model, is used to combine the separate goals in a manner emphasizing potential points of both agreement and disagreement existing between the suitability values of different goals. The third and final phase of the model is an allocation procedure for selecting specific sites for suitable housing in each study area based on policy criteria and community preference. This research, however, is only concerned with fi rst phase of the model, which is discussed below. To identify property suitable for development or preservation of affordable housing based on land and site characteristics. 28 The term affordable housing, as expressed in the context of the first goal of the AHS model is somewhat misleading. The entirety of the model is used for the purpose of identifying sites suitable for the development and preservation of affordable housing based on local planning preference and expert opinion, but the affordable housing goal is characterized by socioeconomic conditions, environmental quality, and local accessibility. In general, the variables in this goal are more sitespecific than those in the other two.

PAGE 87

87 To identify property suitable for development or preservation of affordable housing based on s ocio economic conditions. To identify property suitable for development or preservation of affordable housing based on local accessibility To identify property suitable for development or preservation of affordable housing based on neighborhood change Eac h one of these objectives consists of multiple subobjectives, which consist of distinct criteria. These criteria are integrated into the model in the form of map layers representing specific features relevant to determining suitability for affordable hous ing, such as the location of schools or areas with high crime rates. Scores are assigned across the surface of these individual map layers based on the positive or negative impact their attributes have on the suitability of different locations, transforming what was before a simple representation of features into a suitability surface that functions as a building block for the rest of the model. For example, in the case of the map layer representing the location of hospitals, the closer a residential area i s to a hospital the higher the suitability value it is assigned; conversely, values decrease with distance from these hospitals This process is oriented to the input of planners and experts on housing policy but also incorporates statistical techniques t o assign suitability values to geographic locations. The raster model of spatial representation is employed throughout the AHS, beginning with these initial suitability conversions.29 29 Raster data models represent geographic space as a regular grid composed of cells aligned in rows and columns. Raster models create a standard geographic unit for analysis among various spatial input layers, allowing for overlay operations. The AHS model employs 30 X 30 meter cells for its raster analysis, meaning that it is capable of a very finegrai ned analysis. Rendering all of the criteria layers into the same geometric grid system using raster analysis facilitates the overlay operations used in the model (Chrisman, 1997). To illustrate this process, Figure 45 presents the local accessibility suitability for Orange County.

PAGE 88

88 Once the initial phase of creating these standardized suitability surfaces is complete, the study communities and the model development team collaborate on setting priorities for the different elements in the hierarchical structure of the model. Thus, this stage of the model incorporates the determination of community preference. Preference differs from suitability in that while suitability identifies potential land areas consistent with the goals of the model, the purpose of preference is to capture community values while assessing locations that have already been identified (Carr & Zwick, 2007, p. 128). This work employs AHP to translate qualitative judgments about the relative importance of criteria into quantitative weights assigned to layers. As subordinate layers are combined while proceeding through the hierarchical structure of the model, these weights determine the relative contribution of each layer to higher order elements of the model. The final weighted sum operation in the GIS combines information on suitability contained in all of the model layers into a single output map that indicates goal suitability, measured on a continuous ratio scale from 1 (low) to 9 (high). This final output map is referred to as a preference surface, as it expresses community preference regarding the relative importance of m ultiple criteria in an individual GIS map layer. In sum, the first phase of the AHS model is comprised of four steps: Identifying objectives and subobjectives as criteria for the ultimate goal of the model and structuring them within a hierarchical framew ork Obtaining and preparing relevant data sources for meeting these criteria. Determining the suitability of layers from these data sources through statistical analysis and expert input Deriving preference by combining suitability layers through AHP and weighted overlay analysis in order to produce cumulative suitability layers for first subobjectives, then objectives, and finally the single output layer for individual goals.

PAGE 89

89 Methodology of This Study The suitability values of specific goals, objectives and subobjectives were selected from the AHS model to be used as criteria in evaluating properties in the AHI. Several steps were undertaken in order to join the suitability values for each of the selected suitability layers to properties in the AHI. As w ith the AHS model, this study examines the specific Florida pilot communities of Duval, Orange, and Pinellas Counties. Therefore, only AHI properties located within each of these counties were selected for analysis. In order to bring the AHI into a GIS env ironment, properties in each of the study areas were geocoded based on their address. Geocoding is a process that finds the spatial coordinates of tabular data and allows for the visual representation of them in a GIS as points on a map. These visual representations of AHI properties were then converted into independent spatial data layers and overlaid on to raster layers generated by the AHS for each of the selected suitability criteria. This process is necessary in order to match AHI properties with the s uitability values found at their specific locations. The GIS tool extract values to points was used to join the suitability values from individual AHS raster suitability layers to the properties in the AHI data layer. This procedure was repeated for each of the selected suitability criteria. Once the suitability values for each criterion had been joined to the AHI layer, the attribute table containing these values was exported as a database file so that the results could be subjected to analysis using statistical software. Suitability Criteria Selected for This Study The suitability values of selected goals, objectives and subobjectives were extracted from the AHS model to be used as suitability criteria in assessing properties in the AHI. Some of the reclassification procedures may appear counter intuitive, but they

PAGE 90

90 are designed to generate values of suitability for affordable housing development and preservation and are not pure measures of any given phenomenon. For example, high suitability values for the neighborhood decline indicator are actually indicative of greater demonstrable decline. This is because areas in decline may be suitable for affordable housing intervention to stabilize the neighborhood. Thus, they receive higher suitability scores. T he variables selected from the AHS model for this study are: Crime risk: This is a sub objective of the socioeconomic indicators objective. The suitability values for this subobjective are produced by overlaying violent crime risk and property crime risk. This criterion was selected for its obvious significance in determining suitability for affordable housing, including preservation initiatives. Values above the mean for all block groups are reclassified with a suitability value of 1, with suitability increasing one unit for every 1/4 standard deviation increment below the mean. Poverty rate: This is one of several criteria layers used in finding suitability for the sub objective of positive neighborhood qualities. High poverty rates are associated with a weak housing market, which, as mentioned above, is an indicator of risk for the failing out of an assisted property from the AHI. Values above t he mean for block groups were reclassified as a 1, with suitability values increasing one point for every 1/4 standard deviation increment below the mean. FCAT (Florida Comprehensive Assessment Test) score: Reflecting neighborhood school performance, this criterion layer is incorporated into the subobjective of positive neighborhood qualities. This layer is constructed by overlaying suitability layers created from the FCAT scores of both elementary and high school students. FCAT scores were reclassified as follows: A=9, B=7, C=5, D=3, F=1.This criterion was selected for its use in indicating school quality, hence, suitability for households with school age children.30 Neighborhood decline: Along with gentrification (discussed below), this is one of two subobjectives constituting the neighborhood change objective. Neighborhood change considers changes in values among select indicators using two distinct spatial scales: short term change (20002007) and long term change (19802000) (see Figure 43. High meas ures of neighborhood decline may suggest that an assisted property is located in a weak housing market. As discussed above, this is an indicator used in assessing risk of a property failing out of the AHI. Standard deviation of values was used to determine suitability for each input layer. A subset 30 Florida assesses school performance based on FCAT scores and allocates greater levels of funding to high performance schools.

PAGE 91

91 analysis was applied in the design of this component so that it only considers areas in which median household income for a census tract was within the inter quartile range of median household incomes for the ent ire study area in 1980. The purpose of the subset is to control for decline in areas either well above median income (which might not manifest the problems and pressures conventionally associated with neighborhood decline) and those in already depressed co mmunities. Reclassification occurs as follows: for a variable in which negative change is associated with decline, all areas of positive change are assigned a suitability value of 1, suitability increases one point for every 1/4 standard deviation increment below the mean. Gentrification: As mentioned above, neighborhood change is one of two subobjectives constituting the neighborhood change objective. High measures of gentrification suggest that an area has considerably appreciated over time, increasing pressures on assisted properties capable of doing so to opt out. A subset analysis was applied in this suitability analysis in order to measure change only in census tracts where the median household income was below the median value for the entire study ar ea. The purpose of this subset is to exclude already affluent areas not subject to conventional gentrification processes, such as displacement (Kennedy & Leonard, 2001). Reclassification occurs as follows: for a variable in which negative change is associa ted with gentrification, all areas of positive change are assigned a suitability value of 1, suitability increases one point for every 1/4 standard deviation increment below the mean. For an indicator in which positive change is associated with gentrificat ion, the opposite holds true. Local accessibility: This objective consists of four subobjectives: neighborhood access to transit (transit stops), neighborhood access to services (elementary schools, daycare centers, local health care, hospitals, fire and police stations), neighborhood access to recreation (community centers, cultural centers, parks), and neighborhood access to retail (restaurants, shopping). Accessibility for each of these activities is assessed along two dimensions: opportunity (square footage)/ proximity (distance) and walking (four mile radius)/biking (half mile radius). Figure 4 4 provides an illustration of this organization scheme for the aggregated subobjective of accessibility to services and facilities. Suitability values are generated for each surface by applying a raster analysis on a layer containing hospitals (for example) in order to find the total square footage of hospitals located within four miles (biking) or half a mile (walking) distance from multi family housing. Raster cells with values above the mean are reclassified with a suitability score of 9, with suitability decreasing one unit for every 1/4 standard deviation increment of value below the mean. Affordable housing goal: This is the final output layer generated thr ough the weighted overlay processes employed in the model. It is derived from the weighted overlay of the following objectives: land and site characteristics, socioeconomic indicators, local accessibility, and neighborhood change. This provides a composit e suitability value useful for assessing the general value of properties in the AHI.

PAGE 92

92 Transportation: This is the reclassified output of the transportation cost surface that constitutes the second goal of the AHS model. Transportation cost is determined by evaluating the impact of land use on vehicle miles travelled (VMT) by a household. This layer is generated through a regression analysis using VMT as an independent variable and land use characteristics as the dependent variables. These characteristics ar e the 4 Ds of land use, density, diversity, design, and destinations, each of which have been found to have varying degrees of influence on mode choice and VMT (Boarnat & Crane, 2001; Cervero, 2002; Cervero & Kockelman, 1997). Density is operationalized in the variables of developed area as percent of total neighborhood area and residential area as percent of total neighborhood area. Diversity is operationalized in the square footage of retail, commercial, and office buildings. Design is operationalized as road miles per developed area and number of intersections per road mile. Finally, distance is operationalized as distance to nearest residential center, distance to nearest regional activity center, and range of distances to regional activity centers. A s transportation costs consume a significant portion of low income household budgets and determine the degree of accessibility for critical services and amenities, this indicator is strongly suggestive of the value of assisted properties. Statistical Meth ods Differences in suitability values of properties in the AHI based on age group were investigated through use of the MannWhitney U test, which is a nonparametric test used to determine whether there is a statistically significant difference between two independent sample sets of data (Barber, 1988; Ebdon, 1977).31 31 Statistical Package for Social Scientists (SPSS) 18 software was used to perform the analyses for this study. Unlike parametric tests, non parametric tests are not restricted by assumptions about the nature of the sample populations. As the level of measurement for the suitability values used in this study is ordinal, normal population distribution may not be assumed, requiring the use of a nonparametric test for analysis (Barber, 1988). The MannWhitney U operates by testing the significance of the difference between the medians of the two samples. Th e null hypothesis is that the two samples are drawn from the same population. Should the test substantiate the null hypothesis, any observed difference between the two

PAGE 93

93 samples, such that one set of values is consistently larger than the other is due entir ely to chance in the sampling process (Ebdon, 1977, p. 54). This test has been applied in the fields of geography and urban planning for numerous purposes. Talen (1997), for example, used the MannWhitney U test to evaluate differences in socioeconomic ch aracteristics based on whether or not an area possessed high or low access to public parks. Two separate MannWhitney U tests were run for each county. One test evaluated the difference between properties built or funded between 1963 and 1979 with those b uilt or funded between 1980 and 1994. Here, age is used as a proxy for funder, with the 19631979 category serving as a proxy for HUD funded properties and the 19801994 category serving as a proxy for older FHFC funded properties. The second test compares the difference between properties built or funded between 1963 and 1994 with those built or funded between 1995 and 2008. Age is again used as proxy, with the older category of properties representing the universe of older federal and state assisted properties and the more recent category representing properties developed since the LIHTC was made permanent by Congress. Limitations One of the primary limitations of this study is that is was conducted prior to the completion of pairwise comparison in the AH S model. As a result, the relative preferences held by local communities for component suitability criteria were not incorporated into the overlay analysis as they will be when the development of the AHS model has been concluded. In this study, equal preference is assigned to criteria layers during overlay procedures. For example, when the map layers for violent crime and property crime are combined in order to create the composite crime suitability layer,

PAGE 94

94 they will each have a 50% contribution to the outco me. The resultant suitability values for this layer would be different than those obtained through pairwise comparison if communities had expressed a stronger concern for one or the other of the two criteria. Another limitation to this study is that it was conducted prior to the completion of the housing demand goal. This goal incorporates the demand for housing generated by commercial, office, and residential development. This critical suitability indicator would have been useful in measuring the value of assisted properties based on their potential to accommodate low income workers attracted to employment opportunities in adjacent areas. A final limitation is the absence of the transportation goal from the analysis of properties in Pinellas County. This is a result of the fact that data collection and statistical procedures required to produce the suitability surface for this goal have not yet been completed. Summary The statistical analysis and comparison of the different generations of assisted housing al ong key suitability variables should shed light on their relative value to the affordable housing stock. As discussed in the previous chapter, risk alone may not be enough to make an informed and efficient preservation response. Newman and Schnares (1997) study indicated that a number of privately owned assisted housing developments fare poorly in providing their residents with adequate neighborhood conditions when compared to the entire universe of rental housing. In order to place the results of the anal ysis in the context of the specific case study areas, the next chapter will present demographic, economic, and political characteristics for each county. The characteristics of each countys assisted housing will also be presented and situated within the l arger context for Floridas assisted housing inventory.

PAGE 95

95 Figure 41. Hierarchical structure of the AHS model. Figure 42. Goals and objectives of the AHS model.

PAGE 96

96 Figure 43. Structure of the neighborhood change objective. Figure 44. Neighborhood accessibility objective.

PAGE 97

97 Figure 45. Local accessibility suitability layer, Orange County

PAGE 98

98 CHAPTER 5 DESCRIPTIONS OF STUDY AREAS: DUVAL, ORAN GE, AND PINELLAS COU NTIES Duval, Orange, and Pinellas Counties were selected as study areas for this research based on the participation of the City of Jacksonvilles Department of Housing and Neighborhoods, the City of Orlandos Department of Housing and Community Development, and Pinellas Countys Department of Community Development in the development of the Affor dable Housing Suitability (AHS) model. Among other things, these entities have provided crucial data sets and offered critical input regarding the design of the model. Regardless of this relationship with the three counties, they remain important areas for study in the context of affordable housing. They are all demographically diverse, substantially urbanized, highgrowth areas, each also experiencing significant housing needs. In addition to their similarities in terms of affordable housing needs, these c ounties also offer a study of contrasts as they are representative of different development patterns in Florida. Duval County Duval County, located in Northeast Florida, constitutes the core of the Jacksonville metropolitan statistical area (MSA) and is t he 7th most populous county in the state. The City of Jacksonville and Duval County merged in 1968, from that time all of Duval County has been governed by a single entity with the exception of Atlantic Beach, Neptune Beach, Jacksonville Beach, and Baldwin. Duval County has a land area of 774 square miles and a population density of 1,006 persons per square mile.32 32 Population and other descriptive data from U.S. Census Bureau, State & County QuickFacts, The St. Johns River runs through the City of Jacksonville, where it turns east from its northerly http://quickfacts.census.gov/qfd/states/12000.html

PAGE 99

99 course to empty into the Atlantic Ocean. The ri ver provides Jacksonville with the 14th largest deepwater port in the U.S. This port, along with the areas airports and other transportation assets make the county a major intermodal transportation hub. Demographics, Income, and Housing Affordability Du val County has experienced substantial growth as a result of its young population, diverse economy, and high quality of life. Recent estimates place the population of Duval County at 857,400, marking a 10% increase over its 2000 population of 778,866. Grow th in the county was particularly strong in the 1990s, when the population increased by 16 percent. The countys population is projected to top one million by 2020 (see Table 51) The median age of Duval County is currently 35.9, making it somewhat younger than the state as a whole, which has a median age of 40.5 (see Table 512 ). However, the median age of Duval County is projected to increase to 40.3 by 2030, by which time the share of persons age 65 or older is projected to have risen from 11 to 20% of the countys population. Thus, a substantial share of the countys future growth will come from senior citizens. In 2000, the county had 303,747 households with an average of 2.51 persons per household. Median household income in 2008 was $50,660, which i s higher than the median household income of Florida as a whole (see Table 513). Employment is concentrated in trade, transportation and utilities; professional and business services; education and health services; and financial sectors of the economy (se e Table 52) The county is also home to numerous military facilities, including Naval Air Station Jacksonville, the

PAGE 100

100 third largest naval installation in the U.S. The largest private employers include Baptist Health, Blue Cross & Blue Shield of Florida, the Mayo Clinic, and Citibank.33The housing market in Duval County has remained affordable relative to Florida as a whole. While home prices rose significantly during the middle of the first decade of the twenty first century, the movement of home prices relative to household income was not as dramatic as that experienced by the other study areas and the nation as a whole ( see Figure 53 ). In recent years, home prices have continued to fall as a result of the foreclosure crisis and economic recession (see Tabl e 521) Nevertheless, housing affordability remains a problem for many lower income households in Duval County. In 2008, 24% of renter households earning less than 50% of AMI paid more than 30% of their income toward housing (see Table 515 ). The NLIHCs 2010 Out of Reach report finds that the housing wage for a onebedroom apartment in Duval County is an affordable 101% of the mean renter wage, but fully 206% of minimum wage. 34 33 Jacksonville and Northeast Regional Economic Development, The amount of severely cost burdened households earning 80% or less of AMI is projected to increase 37% to more than 26,000 by 2030 (see Table 53). Among lower income households, those earning 30% or less of AMI are particularly susceptible to experiencing cost burden. The growth in severely cost burdened household earning up to 30% of AMI is significantly higher than growth for other lower income households (see Table 54). http://www.expandinjax.com/About/Regional_Overview/Duval_Co.aspx 34 The housing wage is the full time hourly wage required to pay HUDs established FMR without spending more than 30% of income.

PAGE 101

101 Assisted Housing Inventory Duval County has the 4th largest number of assisted properties in the state with a total of 139 properties in the AHI, and it has the 4th largest number of assisted units with a total of 21,544 (see Table 517). Properties receiving assistance from HUD and FHFC account for roughly even shares of the AHI in Duval County (see Table 518). The majority of the countys assisted units are designated for renters earning between 55 and 60 percent, with only 3% designated for renters earning 35% or less of AMI (see Table 519). For the assisted properties for which data is available, roughly half are more than 30 year s old and 30% are between 21 and 30 years old (see Table 520). Local Housing Policy The City of Jacksonville strongly supports the preservation of affordable rental housing in order to meet the housing needs of its low income residents. Jacksonvilles Consolidated Plan, which outlines the citys housing needs and strategies for meeting them, explicitly prioritizes preservation activities, especially the preservation of affordable housing in gentrifying areas: The intent of the Consolidated Plan is to put in place policies that will help and coordinate efforts to preserve and develop affordable housing opportunities for low and moderateincome residents in a chain reaction that will help preserve the social and historic character of low income neighborhoods threatened by gentrification (City of Jacksonville, 2005, p. 79). Not only does the city emphasize preservation initiatives in gentrifying areas characterized by rapid reinvestment, but also in distressed neighborhoods most in need of assistance, which the city has designated as Neighborhood Action Plan Areas

PAGE 102

102 (NAP).35Orange County Jacksonvilles Consolidated Plan encourages efforts to preserve existing affordable rental housing in NAPs though various funding mechanisms, including the provision of low interest loans in exchange for affordable rental rates (City of Jacksonville, 2005). These policies are to the benefit of properties at risk of failing out of the assisted housing stock due to physical deterioration and potential default. Another important dimension of housing policy in Jacksonville is its emphasis on poverty deconcentration and fair housing practices (Jacksonville, 2005). Located in Central Florida, Orange County forms the core of the OrlandoKissimmee MSA and is the 5th most populous county in the state. Like Duval and Pinellas Counties, Orange is a charter county, meaning that it is self governing. The seat of government for Orange County is the City of Orlando. The cities in which properties in the Assisted Housing Inventory are located ar e Apopka (pop. 35,563), Orlando (pop. 220, 186), and Winter Park (pop. 28,083). Orange County has a land area of 907 square miles and a population density of 988.3 persons per square mile. Orange County is highgrowth region and a major tourist destination, as it is the home of Walt Disney World and Universal Studios Florida. Demographics, Income, and Housing Affordability Orange County is an especially highgrowth region in Florida. Between 1990 and 2000, the population increased 35% and now stands at over one million (see Table 55). Orange County is younger overall than the state of Florida taken as a whole, and is 35 T he NAP concept is a comprehensive longterm approach to community revitalization that focuses on community assets as a means of stimulating market driven redevelopment (City of Jacksonville, 2005, p. 3)

PAGE 103

103 younger than the other study areas (see Table 512). The median age for the county is only 33.1, which, unlike the rest of the state, is ac tually lower than its 2006 value ( Mdn =33.0). While the number of persons age 65 and older is projected to increase over the next two decades, the share of this demographic is projected to reach just 15%. In 2000, the county had 336,286 households, with an average of 2.61 persons per household.36Median household income in 2008 was $50,674, which is higher than the statewide median value of roughly $46,000 (see 513). Employment is concentrated in the leisure and hospitality; professional and business servic es; and the trade, transportation and utilities sectors of the economy (see 56). Many of the jobs created as a consequence of the Countys tourism based economy are low wage, contributing to the increase in the number of lower income households in the cou nty and the demand for affordable housing. The housing market in Orange County is currently more expensive than the state as a whole. Just as in the rest of the state, home prices rose precipitously during the middle of the first decade of the twenty f irst century, increasing as much as 41% in one year alone (see 521). During this period, the ratio of median household income to median singlefamily home sales price topped 5.0. (a ratio of 3.0 is used as a benchmark of housing affordability) (see Figure 5 3). As in the rest of the state, housing costs have dropped since the onset of the current recession, but, as in the other study areas, the number of cost burdened lower income households is only projected to increase. In 2008, 24% of renter households earning less than 50% of AMI paid more 36 Population and other descriptive data from U.S. Census Bureau, State & County QuickFacts, http://quickfacts.census.gov/qfd/states/12000.html

PAGE 104

104 than 30% of their income toward housing (see Table 515). The NLIHCs 2010 Out of Reach report finds that the housing wage for a onebedroom apartment in Orange County is 123% of the mean renter wage and 244% of minimum wage. The number of severely cost burdened households earning 80% or less of AMI is projected to increase 38% to nearly 40,000 by 2030 (see Table 57 ). By 2020, the number of severely cost burdened lower income households will have increased by more t han 5,000 (see Table 5 8) Assisted Housing Inventory Orange County has the 2nd largest number assisted properties in the state with a total of 170 properties in the AHI, and it has the 2nd largest number of a ssisted units with a total of 31,181 (see Table 5 17). Roughly 65% of the countys assisted properties were developed by FHFC, while only 15% were developed under HUD programs (see Table 518). As in Duval County, t he majority of the countys assisted units are designated for renters earning between 55 and 60 percent, with only 3% designated for renters earning 35% or less of AMI However, far fewer units in Orange County are designated for renters with incomes between 40 and 50% of AMI (see Table 519). For the assisted properties for which data is available, 22% are more than 30 years old and 30% are between 21 and 30 years old. A larger share of properties (32%) was built in Orange County within the past 10 years than in Duval County (see Table 520). Local Housing Policy Housing policy in the City of Orlando highlights the preservation of existing affordable housing as a means of meeting the needs of low income residents. Like Duval County, it encourages balanced housing policies that do not privilege homeownership and new development to the exclusi on of preservation efforts, including

PAGE 105

105 those for multifamily rental housing. Illustrating this point, the very first goal of Orlandos Consolidated Plans is to i ncrease the availability of existing affordable housing options within the City of Orlando for extremely low low and moderateincome residents (City of Orlando, 2005, p. SP 7) Under this goal are listed numerous strategies for meeting this goal, such as, rehabilitating existing singlefamily and multifamily rental housing, supporting nonprofits in obtaining additional funding for preservation efforts, including Community Redevelopment Area Tax Increment Funds, and leveraging additional funds to assist both for profits and nonprofits in acquiring and rehabilitating affordable rental units (City of Orlando, 2005). The Consolidated Plan also indicates that the city plans to conti nue allocating funds from its share of Community Development Block Grants (CDBG), HOME grants, and State Housing Initiatives Partnership (SHIP) funds toward preservation efforts (City of Orlando, 2005, p. SP 37). In addition, Orlandos Housing Element cont ains a policy explicitly expressing the citys desire to preserve assisted housing: The city shall encourage preservation of units threatened by expiring Section 8 contracts, condominium conversions, and foreclosures by working with tenants, owners, and organizations who provide information about related issues (City of Orlando, 2010, p. H 5). Pinellas County Situated on Floridas Gulf Coast, Pinellas County occupies a peninsula whose eastern shore constitutes the western boundary of Tampa Bay. Pinellas County, together with Hillsborough, Hernando, and Pasco Counties comprise the TampaSt.Petersburg Clearwater MSA. County properties in the AHI are located in the Cities of Clearwater (pop. 105,774), Dunedin (pop. 35,988), Pinellas Park (pop. 47,173), Largo (pop. 72,732), St. Petersburg (pop. 245, 314), South Pasadena (pop. 5,548), and

PAGE 106

106 Tarpon Springs (pop. 23,359).37Demographics, Income, and Housing Affordability Pinellas County is the 6th most populous in Florida with a population of 938,459. Pinellas County has a land area of 279.92 square miles and has a population density of 3,291 persons per square mile, making it significantly denser than the other study areas. The county seat is Clearwater. While Pinellas Countys population increased by 8% from 1990 t o 2000, it has dropped by 1.4% between 2000 and 2009. This loss is largely symptomatic of the economic recession, and it suggests that the population projections of a year before may be in need of revision. The estimated median age of the county (45.6), w hile marginally lower than the state as a whole, is considerably higher than the other two highly urbanized study areas (see Table 513). The share of households age 65 and older is above 20%, and is projected to top 30% in 2030. In 2000, Pinellas County had 414,988 households, with an average of 2.17 persons per household. Median household income was $45,650 for Pinellas County in 2008, which is marginally lower than that for the state as a whole (see Table 514). Employment is concentrated in the profess ional and business services; trade, transportation, and utilities; and education and health services sectors (see Table 510). Pinellas Countys largest employers include Fidelity Information Services, the Home Shopping Network, Nielsen Media Research, and Raymond James Financial.38As in the other study areas, the housing market in Pinellas County rose dramatically during the middle of the first decade of the twenty first century and declined 37 Population estimates from U.S. Census Bureau, State & Count y QuickFacts, http://quickfacts.census.gov/qfd/states/12000lk.html 38 Pinellas County Economic Development, Pinellas Countys Largest Private Employers, http://www.pced.org/demographics_data/subpage.asp?TopEmployers

PAGE 107

107 thereafter (see Table 5 21). As in Orange County, these increases significantly outpaced increases in household income. As Figure 53 shows, the ratio of median single family home sales price to median household income rose as high as 5.0. As in the other study areas, diminished home prices have not solved the countys affordable housing problem. In 2008, 23% of renter households earning less 50% of AMI spent more than 30% of their income on housing (see Table 515). The NLIHCs 2010 Out of Reach report finds that the housing wage for a onebedroom apartment in Pinellas County is 110% of the mean renter wage and 210% of minimum wage. Unlike the other study areas, growth in the number of severely cost burdened household is relatively small. Following a decline between 2008 and 2010, the total number of severely cost burdened households is projected to grow by only 676 through 2030 (see Table 512). Assisted Housing Inventory Pinellas County has the 6th largest number of assisted properties in the state with a total of 101 properties in the AHI, and it has the 7th larges t number of assisted units with a total of 9,828 (see Table 517 Table ). More than 50% of the countys assisted properties were funded by HUD programs, while 33% were funded by FHFC (see Table 518). More than 70% of the assisted units in the county are designated for renter households earning between 55 and 60% of AMI, but a larger share of units are designated for households earning 35% or less of AMI than in other c ounties (see Table 519). In terms of age, assisted properties are relatively younger on the whole in Pinellas County than in the other two study areas, with nearly half of them being built or funded within the last 20 years (see Table 520).

PAGE 108

108 Local Housing Policy As in the other two case study areas, Pinellas County is directly involved in the preservation of affordable housing. Pinellas County annually allocates portions of its federal and state housing assistance funds for preservation initiatives, including the acquisition and rehabilitation of multifamily rental units. These preservation efforts take place not only within unincorporated Pinellas County, but also within 20 participating municipalities. In its 20092010 Action Plan, Pinellas County ( 2009) allocated $358,295 in HOME grants toward the preservation of affordable rental housing. Pinellas County (n.d.) has also committed to allocating a share of a $17.8 million Neighborhood Stabilization Program (NSP) grant toward the preservation of affor dable rental housing. Like Jacksonville, the City of St. Petersburg (2005) also has a Neighborhood Revitalization Strategy Area (NRSA) for which it has requested federal and state funding for a host of revitalization initiatives, including the expansion and preservation of affordable housing (p. 151).

PAGE 109

109 Figure 51. Study area counties

PAGE 110

110 Figure 52. Location of properties in Florida's Assisted Housing Inventory (AHI)

PAGE 111

111 Table 51. Population projection (permanent residents) by age for 19902030, Duval Cou nty Age 0 19 20 64 65+ Total 1990 28.90% 60.42% 10.68% 672,514 2000 29.08% 60.41% 10.50% 778,446 2008 27.69% 61.33% 10.98% 904,407 2010 27.16% 61.36% 11.48% 916,938 2015 26.28% 60.28% 13.45% 974,938 2020 25.93% 58.34% 15.73% 1,040,436 2025 25.35% 56.34% 18.31% 1,104,138 2030 24.66% 54.72% 20.61% 1,164,443 Source: Florida Housing Data Clearinghouse, Population Projection by Age for 1990 2030, http://flhousingdata.shimberg.ufl.edu/a/population Table 52. Estimated employment by industry in Duval County, 2009 Industry Title Employment Share Total, All Industries 498,848 100.00% Natural Resources and Mining 3 72 0.07% Construction 26,361 5.28% Manufacturing 23,830 4.78% Trade, Transportation, and Utilities 102,208 20.49% Information 7,875 1.58% Financial Activities 52,214 10.47% Professional and Business Services 69,976 14.03% Education and Health Services 65,326 13.10% Leisure and Hospitality 41,764 8.37% Other Services (Except Government) 18,890 3.79% Government 51,566 10.34% Source: Florida Agency for Workforce Innovation, Labor Market Statistics Center, Employment Projections, http://www.labormarketinfo.com/Library/EP.htm Table 53. Number of severely cost burdened renter households with income less than 80% AMI by tenure and income level Duval County Household Income as % of AMI 2008 2010 2015 2020 2025 2030 Percent change 0 30% AM 14 119 14 373 15 334 16 455 17 608 18 733 33% 30.1 50% AMI 4 085 4 178 4 554 5 003 5 494 5 978 46% 50.1 80% AMI 851 876 994 1 135 1 301 1 463 72% Total 19 055 19 427 20 882 22 593 24 403 26 174 37% Source: Shimberg Center for Housing Studies, Affordable Housing Needs Summary, http://flhousingdata.shimberg.ufl.edu/a/summary

PAGE 112

112 Table 54. Growth in severely cost burdened renter households with income less than 80% AMI by tenure and income level Duval County Household Income as % of AMI 2008 2010 2010 2015 2015 2020 2020 2025 2025 2030 Total 0 30% AM 254 961 1 121 1 153 1 125 4 614 30.1 50% AMI 93 376 449 491 484 1 893 50.1 80% AMI 25 118 141 166 162 612 Total 372 1 455 1 711 1 810 1 771 7 119 Source: Shimberg Center for Housing Studies, Affordable Housing Needs Summary, http://flhousingdata.shimberg.ufl.edu/a/summary Table 55. Population projection (permanent residents) by age for 19902030, Orange County Year 0 19 20 64 65+ Total 1990 27.44% 61.65% 10.91% 660,263 2000 28.38% 61.55% 10.07% 893,179 2008 28.91% 61.71% 9.37% 1,111,641 2010 28.59% 61.79% 9.63% 1,115,864 2015 27.96% 61.31% 10.73% 1,209,460 2020 27.84% 59.98% 12.18% 1,321,157 2025 27.50% 58.56% 13.94% 1,429,853 2030 26.93% 57.49% 15.58% 1,531,645 Source: Florida Housing Data Clearinghouse, Population Projection by Age for 1990 2030, http://flhousingdata.shimberg.ufl.edu/a/population Table 5 6. Estimated employment by industry in Orange County, 2009 Industry Title Employment Share Total, All Industries 761,055 100.00% Natural Resources and Mining 3,599 0.47% Construction 31,833 4.18% Manufacturing 26,426 3.47% Trade, Transportation, and Utilities 121,878 16.01% Information 16,234 2.13% Financial Activities 40,942 5.38% Professional and Business Services 126,877 16.67% Education and Health Services 75,507 9.92% Leisure and Hospitality 145,290 19.09% Other Services (Except Government) 36,547 4.80% Government 71,623 9.41% Source: Florida Agency for Workforce Innovation, Labor Market Statistics (LMS) Center, Employment Projections http://www.labormarketinfo.com/Library/EP.htm

PAGE 113

113 Table 57. Number of severely cost burdened renter households with income less than 80% AMI by tenure and income level Orange County Household Income as % of AMI 2008 2010 2015 2020 2025 2030 Percent change 0 30% AM 16,611 16,724 18,085 19,718 21,381 22,980 38% 30.1 50% AMI 9,402 9,448 10,207 11,108 12,031 12,925 37% 50.1 80% AMI 2,193 2,211 2,401 2,634 2,878 3,114 42% Total 28,206 28,383 30,693 33,460 36,290 39,019 38% Source: Shimberg Center for Housing Studies, Affordable Housing Needs Summary, http://flhousingdata.shimberg.ufl.edu/a/summary Table 58. Growth in severely cost burdened renter households with income less than 80% AMI by tenure and income level Orange County Household Income as % of AMI 2008 2010 2010 2015 2015 2020 2020 2025 2025 2030 Total 0 30% AM 113 1,361 1,633 1,663 1,599 6,369 30.1 50% AMI 46 759 901 923 894 3,523 50.1 80% AMI 18 190 233 244 236 921 Total 177 2,310 2,767 2,830 2,729 10,813 Source: Shimberg Center for Housing Studies, Affordable Housing Needs Summary, http://flhousingdata.shimberg.ufl.edu/a/summary Table 59. Population projection (permanent residents) by age for 19902030, Pinellas County Age 0 19 20 64 65+ Total 1990 19.90% 54.02% 26.08% 849 987 2000 21.21% 56.24% 22.55% 920 310 2008 20.86% 57.71% 21.43% 937 465 2010 20.45% 57.75% 21.81% 928 299 2015 19.62% 56.51% 23.87% 932 096 2020 19.28% 54.35% 26.37% 936 099 2025 18.88% 51.88% 29.24% 940 112 2030 18.50% 49.92% 31.58% 943 908 Source: Florida Housing Data Clearinghouse, Population Projection by Age for 1990 2030, http://flhousingdata.shimberg.ufl.edu/a/population

PAGE 114

114 Table 5 10. Employment by industry in Pinellas County, 2009 Industry Title Employment Share Total, All Industries 453,449 100.00% Natural Resources and Mining 161 0.04% Construction 19,464 4.29% Manufacturing 34,479 7.60% Trade, Transportation, and Utilities 70,815 15.62% Information 8,231 1.82% Financial Activities 29,431 6.49% Professional and Business Services 79,280 17.48% Education and Health Services 70,626 15.58% Leisure and Hospitality 42,990 9.48% Other Services (Except Government) 17,583 3.88% Government 45,742 10.09% Source: Florida Agency for Workforce Innovation, Labor Market Statistics Center, Employment Projections, http://www.labormarketinfo.com/Library/EP.htm Table 511. Number o f severely cost burdened households with income less than 80% AMI by tenure and income level Pinellas County Household Income as % of AMI 2008 2010 2015 2020 2025 2030 Percent change 0 30% AM 12,222 12,130 12,139 12,141 12,147 12,159 1% 30.1 50% AMI 6,315 6,281 6,406 6,564 6,734 6,876 9% 50.1 80% AMI 1,508 1,503 1,541 1,589 1,639 1,686 12% Total 20,045 19,914 20,086 20,294 20,520 20,721 3% Source: Shimberg Center for Housing Studies, Affordable Housing Needs Summary, http://flhousingdata.shimberg.ufl.edu/a/summary Table 512. Growth in severely cost burdened households with income less than 80% AMI by tenure and income level Pinellas County Household Income as % of AMI 2008 2010 2010 2015 2015 2020 2020 2025 2025 2030 Total 0 30% AM 92 9 2 6 12 63 30.1 50% AMI 34 125 158 170 142 561 50.1 80% AMI 5 38 48 50 47 178 Total 131 172 208 226 201 676 Source: Shimberg Center for Housing Studies, Affordable Housing Needs Summary, http://flhousingdata.shimberg.ufl.edu/a/summary

PAGE 115

115 Table 5 13. Median age estimates and projections, 20062030 Place 2006 2010 2020 2030 Florida 39.9 40.5 42.3 44.4 Duval 35.2 35.9 37.9 40.3 Orange 33 33.1 34.6 36.4 Pinellas 44.5 45.6 47.3 48.2 Source: Bureau of Economic and Business Research, Florida Statistical Abstract, 2007 Table 514. Median household income, 1989, 2008 Year Duval County Orange County Pinellas County Florida 1989 28 513 30 252 26 296 27 483 1999 40 703 41 311 37 111 38 819 2008 50 ,660 50,674 45 650 45,899 Source: U.S. Census Bureau, 1990 & 2000 SF3 Files, State and County QuickFacts. Figure 53. Ratio of m edian e xisting single f amily h ouse p rices to m edian h ousehold incomes by metropolitan area, 1989 2009. Source: Joint Center for Housing Studies, 2009

PAGE 116

116 Table 515. Renter households with cost burden above 30% and income below 50% AMI, 2008 Place Households %of all households Rank by number of households Duval County 31,149 24.1 6 Orange County 39,373 23.9 3 Pinellas 28,799 23.1 7 Source: Florida Housing Data Clearinghouse Affordable Housing Needs Table 516. Extremely low income (<30 AMI), severely cost burdened households, 2008 Place Households %of all households Rank by number of households Duval County 22,856 6.3 6 Orange County 24,350 5.8 5 Pinellas 22,484 5.3 7 Source: Florida Housing Data Clearinghouse Affordable Housing Needs Table 517. Total properties, units, and assisted units in Assisted Housing Inventory by county, 2010 County Number of Properties Total Units Assisted Units Duval 139 22 ,544 21 544 Orange 170 32 565 31 181 Pinellas 101 11 721 9 828 Source: Florida Housing Data Clearinghouse, Assisted Housing Inventory Table 518. Properties and assisted units by funder (duplicated count), 2010 Duval Orange Pinellas Funder Properties Units Properties Units Properties Units HUD 70 9 041 41 5 552 61 4 648 RD 3 139 9 411 0 0 FHFC 68 12 523 114 24 276 37 4 445 LHFA 19 4 282 39 7 250 15 2 428 TOTAL 160 25 985 203 37 489 113 11 521 Source: Florida Housing Data Clearinghouse, Assisted Housing Inventory Table 519. Assisted units by income limits, 2010 Number of Units Designated for Renters with Income: County <=35% AMI 40 50% AMI 55 60% AMI 65 80% AMI >80% AMI Duval 336 2 409 8 027 510 885 Orange 732 1 492 19 231 348 2 092 Pinellas 330 441 2 265 104 0 Source: Florida Housing Data Clearinghouse, Assisted Housing Inventory

PAGE 117

117 Table 520. Assisted housing properties and units by age of property or year funded 0 10 Years 11 20 Years 21 30 Years > 30 Years Not available County Prop. Units Prop. Units Prop. Units Prop. Units Prop. Units Duval 9 1,939 6 463 24 3,559 41 5,929 59 9,654 Orange 24 3,994 12 2,052 23 3,419 17 2,810 94 18,956 Pinellas 15 916 17 444 21 1,547 12 2,371 36 4,550 Source: Florida Housing Data Clearinghouse, Assisted Housing Inventory Table 521. Median sales price for singlefamily homes (in thousands of dollars), 19952009 Jacksonville MSA Orlando Kissimmee MSA Tampa St. Pete Clearwater MSA Florida Year Median sales price Change Median sales price Change Median sales price Change Median sales price Change 1995 83.9 1% 86.1 2% 78 4% 87.7 2% 1996 90 7% 90.7 5% 80.9 4% 92.3 5% 1997 88.6 2% 94.4 4% 84 4% 95.8 4% 1998 97.3 10% 96 2% 88 5% 99.8 4% 1999 97.2 0% 102.8 7% 94.7 8% 108.4 9% 2000 103.9 7% 109.3 6% 105.8 12% 117.6 8% 2001 116.7 12% 120.1 10% 124.4 18% 126.6 8% 2002 124.4 7% 130.3 8% 133.3 7% 141.7 12% 2003 138.2 11% 143.5 10% 139.3 5% 155.8 10% 2004 158.6 15% 164.5 15% 159.9 15% 181.9 17% 2005 187.3 18% 231.4 41% 201.7 26% 235.2 29% 2006 200.6 7% 262.9 14% 224.8 11% 247.1 5% 2007 196.5 2% 248.9 5% 208.9 7% 234.3 5% 2008 180.4 8% 201.5 19% 169.5 19% 187.7 20% 2009 152.2 16% 144.6 28% 137.5 19% 142.6 24% Source: Florida Association of Realtors. Places are metropolitan statistical areas (MSAs).

PAGE 118

118 CHAPTER 6 RESULTS FROM TESTS OF DIFFERENCES BETWEEN ASSISTED HOUSING GENERATIONS In order to understand the relative suitability of properties in the assisted housing inventory based on the generation of housing production to which they belong, the MannWhitney U test was conducted to compare these groups based on key selected criteria This chapter examines the results of these tests in order to better understand strategies for assisted housing preservation in Florida. This chapter begins by presenting the results from the statistical analysis of suitability values for properties belonging to the different generations. Results are presented for the two tests performed for each county, the first being a comparison of properties either built or funded between 1963 and 1979 with those either built or funded between 1980 and 1994 (HUD versus older FHFC properties), the second being a comparison of properties either built or funded between 1963 and 1994 with those either built or funded between 1995 and 2008 (HUD and older FHFC versus LIHTC properties). Duval County The results of the MannWhitney U test for differences between assisted properties built/funded between 1963 and 1979 (proxy for HUD financed properties) and those built/funded between 1980 and 1994 (proxy for older FHFC financed properties) in Duval County are presented in Table 6 1. The former of the two groups has a sample size of 43 and the latter a sample size of 40. The results of the MannWhitney U test show a statistically significant difference (at a significance level of p =.05) between the two groups for the suitability variables crime risk, FCAT scores, affordable housing preference, poverty rate, and transportation. This finding indicates that older FHFC financed properties are in areas with significantly higher suitability values for crime risk,

PAGE 119

119 FCAT scores, affordable housing preference, and poverty rates, but significantly lower suitability related to transportation suitability values.39These differences show that older FHFC financed properties are located in areas with significantly better neighborhood conditions than those developed under HUD programs, but that these advantages potentially come at the expense of location efficiency. The median poverty rate suitability value for the 19801994 group of properties ( Mdn =6) is substantially higher than the median value for the 19631979 group ( Mdn =1), which would seem to strongly recommend the younger properties for preservation in the context of a policy preference for poverty deconcentration. 40 39 Suitability values are reclassified so high numbers reflect greater suitability for housing. While high actual crime risk and poverty rate suggests low suitability for housing, the suitability values for these variables move in the opposite direction so as actual crime risk rises, suitability values for this variable decline. In the case of variables where actual measur es are directly related to suitability, they are reclassified on the 19 scale, but in the same direction. That is, high measured accessibility means high suitability accessibility. The test showed no statistical difference between the two groups for the variables of neighborhood decline and gentrification, indicating that properties in one group are no more likely than those in the other to be found in areas marked by neighborhood ch ange. Finally, the test yielded no significant difference between the groups for local accessibility. As the groups are similarly situated in terms of local accessibility, the higher suitability values for variables indicating positive neighborhood conditi ons exhibited by the older FHFC financed properties strongly recommends their consideration for prioritization in Duval County. The higher transportation values found in areas in which HUF financed properties are located suggest that they may promote enhanced mobility for their residents, but without proximity to destinations and transit 40 As explained in the methodology, higher suitability values for the pover ty rate variable indicate lower actual poverty rate.

PAGE 120

120 opportunities key components of the local accessibility variable it is perhaps not enough to tip the balance in favor of HUD financed properties based on this analysis.41T he results of the MannWhitney U test for differences between assisted properties built/funded between 1963 and 1994 and those built/funded between 1995 and 2008 in Duval County are presented in Table 62. The former of the two groups has a sample size of 85 and the latter a sample size of 51. The results of the MannWhitney U test show a statistically significant difference between the groups for the variables of decline, FCAT scores, gentrification, local accessibility, poverty rate, and transportation. T he younger group of properties (19952008) has significantly higher suitability values for neighborhood decline, FCAT scores, and poverty rate, but significantly lower suitability values for gentrification, local accessibility, and transportation. While the median values are identical for the two groups for the transportation variable, the mean ranks were considerably different, with the group 19631994 having a mean rank of 74.55 and 19952008 having a mean rank of 56.02. The results of this test display an interesting and seemingly incongruous set of relationships for the younger group. While they are better situated with regard to poverty rate and FCAT scores, both signs of positive neighborhood quality, they are simultaneously less well situated with regard to neighborhood decline (high suitability=greater decline). The older of the two groups has significantly higher values for gentrification, meaning that its properties are statistically more likely to be in areas experiencing upward market pressure th an those in the younger group. This, in combination with higher local accessibility and transportation values suggests that these 41 Transportation scholars generally distinguish between accessibility and mobility by defining accessibility as the ease of reaching as mobility as the ease of moving (Preston & Raj, 2007, p. 154).

PAGE 121

121 older properties are more likely to be in desirable urban neighborhoods at risk of becoming unaffordable to lower income households. In the context of Duval Countys stated objective of preserving affordable housing in gentrifying neighborhoods, the older, HUDfinanced properties seem to be of higher value to the affordable housing stock. There is no statistical difference between the two groups for the variables crime risk and affordable housing preference. This finding suggests that the overall tilt in favor of the combined HUD and older FHFC financed properties is even stronger than that exhibited in the first test (between HU D and older FHFC financed properties). Indeed, the absence of difference between the groups for the affordable housing preference variable does not mean that these two groups should be equally prioritized within the AHI. The other, disaggregated variables included in the analysis (local accessibility, gentrification, decline, poverty rate, school quality) are better able to shed light on the manner in which these groups of properties score based on different dimensions of suitability. Orange County The r esults of the MannWhitney U test for differences between HUD financed properties (those built/funded between 1963 and 1979) and older FHFC financed properties (those built/funded between 1980 and 1994) in Orange County are presented in Table 63. The form er of the two groups has a sample size of 16 and the latter a sample size of 54. The results of the MannWhitney U test show a statistically significant difference (at a significance level of p =.05) between the suitability variables of crime, decline, gent rification, and local accessibility. This finding indicates that the younger properties (19801994) are in areas with significantly higher suitability values for crime risk and decline, and significantly lower values for gentrification and local accessibil ity.

PAGE 122

122 There is no statistical difference between the two groups for the variables FCAT scores, affordable housing preference, poverty rate, and transportation. The combination of higher suitability for areas with the older FHFC financed properties related to both crime risk (lower measured crime risk) and neighborhood decline (higher measured decline) is interesting, as these variables are generally inversely related. This finding could be fortuitous in the context of preservation, as investment in these properties could serve as a catalyst for neighborhood revitalization in areas that seem particularly well suited to take to such efforts based on lower crime rates. However, the significantly higher suitability in areas with HUD financed properties related to gentrification and local accessibility suggest that they are better candidates for prioritization in the AHI than the older FHFC financed properties. This judgment is informed by the knowledge that local accessibility is generally difficult to adjust for a particular location (it involves numerous dimensions of the built environment), meaning that it is a particularly important asset. In addition, the potential presence of gentrification suggests that a limited window exists for the continued affordabili ty of the neighborhood and that these properties may be located in desirable, amenity rich areas. The results of the MannWhitney U test for differences between the combined group of HUD and older FHFC financed properties (those built/funded between 1963 and 1994) and newer, predominantly LIHTC properties (those built/funded between 1995 and 2008) in Orange County are presented in Table 64. The former of the two groups has a sample size of 75 and the latter a sample size of 79. The results of the MannWhitney U test show a statistically significant difference between the groups for the suitability variables of FCAT scores and decline. Areas where LIHTC properties are

PAGE 123

123 located have significantly higher suitability values for decline, but significantly low er values for FCAT scores, meaning that they are more likely to be situated in declining areas with lower school quality.42 There is no statistical difference between the two groups for the variables of crim e risk, gentrification, affordable housing preference, local accessibility, poverty rate, and transportation. These results suggest that the two groups are not all that different, making a clear decision about which should be prioritized in the AHI difficult. The higher suitability values for decline found in the younger groups of properties may actually be read as a negative attribute depending on the policy lens applied in making prioritization decisions. Local policies related to affordable housing in some localities, such as Duvals prioritization of preservation activities located in Neighborhood Action Plan (NAP) areas, express a clear interest in undertaking preservation initiatives in distressed areas as part of a larger neighborhood revitalization s cheme. While Orange County and the City of Orlando have expressed an interest in affordable housing preservation (including assisted housing), their policies do not suggest that location in a This result is noteworthy as it is incongruent with the results of previous research finding that LIHTC properties are generally bett er situated with regard to socioeconomic indicators than housing developed under other assistance programs (Freeman, 2004; Oakley, 2008). Furthermore, as LIHTC properties are more likely to be located in suburban areas than HUD financed properties (Freeman, 2004), these results suggest that LIHTC properties in Orange County may be located in struggling inner ring suburbs, not in the generally more prosperous newer suburbs. 42 While the median values are identical for the groups on the FCAT score variable, the mean rank for the group 19631995 is 84.95 while the mean rank for the group 19952008 is 70.43, meaning that the former group has significantly higher suitability values.

PAGE 124

124 declining area automatically makes assisted housing preferable f or preservation. All other things being equal, the older properties are significantly different since they are more likely to be located in areas not experiencing decline, meaning that they should have better neighborhood quality than those properties prod uced after the generational shift. Pinellas County The results of the MannWhitney U test for differences between HUD financed properties (those built/funded between 1963 and 1979) and older FHFC financed properties (those built/funded between 1980 and 1994) in Pinellas County are compared in Table 65. Th e former of the two groups has a sample size of 13 and the latter a sample size of 29. The small sample size of these groups informed the decision to use the MannWhitney U test rather than the parametric students t test. The results of the Mann Whitney U test show a statistically significant difference (at a significance level of p =.05) between the two groups for none of the suitability variables. Similarly, no statistically significant difference between assisted properties built/funded between 1963 and 1994 and those built/funded between 1995 and 2009 exists (see Table 66). In terms of median suitability values, crime risk and local accessibility have relatively high values for all four age categories. This finding suggests that assisted housing in Pinellas may be well situated in terms of producing positive individual outcomes for residents, regardless of funding program. In and of themselves, these results suggest that there is no meaningful difference between the properties developed during the differ ent generations of housing production. This is most likely explained by the fact that more of the countys assisted properties were developed recently, especially in comparison to the other two study areas (see Table). Only 13 properties were developed during the

PAGE 125

125 19631979 period, when the federal government was primarily responsible for housing production. Therefore, no one group of properties in Pinellas County may be privileged for preservation or given priority in the AHI based on this study. Summary The results of these tests yield considerably different results in each of the study areas. In the comparison of HUD financed properties with older FHFC financed properties, FHFCfinanced properties appear to be of greater value in Duval County, HUDfinanced properties of greater value in Orange County, and neither group of properties appears of greater value in Pinellas County. The results of the test comparing both H UDand older FHFC financed properties with properties developed after Congress made the LIHTC program permanent yield less definitive results, but the degree of difference depends on the study area. In Duval County, the older group of properties is clearl y of greater value to the affordable housing stock than the newer properties; in Orange County the results are less conclusive, with the older properties appearing somewhat superior; and in Pinellas County, again, no statistically significant difference was observed. The next chapter will discuss these results in the wider context of preservation strategies and housing policy, and will relate the results of the tests back to the preservation literature.

PAGE 126

126 Figure 61. Properties in the assisted housing inventory categorized by age for test between HUD and older FHFC financed properties, Duval County Table 61. Comparison of assisted properties built/placed in service 19631979 and 19801994, Duval County Age Range 1963 1979 N =43 1980 1994 N =40 MannWhitney U test Variable Median Median Z P (2 tailed) Crime 7.20 9.00 2.474 .013 Decline 1.00 1.33 1.680 .093 FCAT 4.33 5.00 2.299 .022 Gentrification 1.34 1.10 1.351 .177 Affordable housing preference 4.95 5.39 3.041 .002 Local accessibility 7.33 6.78 1.631 .103 Poverty rate 1.00 6.00 3.213 .001 Transportation 9.00 8.50 2.347 .019

PAGE 127

127 Figure 62. Properties in the assisted housing inventory categorized by age for test between combined HUDand older FHFC financed properties and LIHTC properties, Duval County Table 62. Comparison of assisted properties built/placed in service 19631994 and 19942008, Duval County Age Range 1963 1994 N =85 1994 2008 N =51 MannWhitney U test Variable Median Median Z P (2 tailed) Crime 8.20 9.00 1.551 .121 Decline 1.00 1.83 3.174 .002 FCAT 4.33 5.67 2.942 .003 Gentrification 1.10 1.00 2.407 .016 Affordable housing preference 5.04 5.25 .730 .466 Local accessibility 7.29 6.45 3.529 .000 Poverty rate 2.00 8.00 4.246 .000 Transportation 8.50 8.50 2.786 .005

PAGE 128

128 Figure 63. Properties in the assisted housing inventory categorized by age for test between HUD and older FHFC financed properties, Orange County Table 63. Comparison of assisted properties built/placed in service 19631979 and 19801994, Orange County Age Range 1963 1979 N =16 1980 1994 N =54 MannWhitney U test Variable Median Median Z P (2 tailed) Crime 3.30 9.00 1.981 .048 Decline 1.00 1.92 1.993 .046 FCAT 7.33 6.33 .392 .695 Gentrification 2.10 1.28 2.473 .013 Affordable housing preference 5.25 4.96 1.063 .288 Local accessibility 7.26 6.23 3.505 .000 Poverty rate 7.00 6.00 .937 .349 Transportation 9.00 9.00 1.116 .264

PAGE 129

129 Figure 64. Properties in the assisted housing inventory categorized by age for test between combined HUDand older FHFC financed properties and LIHTC properties, Orange County Table 64. Comparison of assisted properties built/placed in service 19631994 and 19942 008 Orange County Age Range 1963 1979 N =75 1980 1994 N =79 MannWhitney U test Variable Median Median Z P (2 tailed) Crime 9.00 9.00 1.592 .111 Decline 1.58 2.17 3.667 .000 FCAT 6.33 6.33 2.034 .042 Gentrification 1.45 1.26 1.624 .104 Affordable housing preference 4.98 4.96 .330 .741 Local accessibility 6.55 6.31 1.520 .128 Poverty rate 6.00 6.00 .739 .460 Transportation 9.00 8.50 .579 .563

PAGE 130

130 Figure 65. Properties in the assisted housing inventory categorized by age for test between HUD and older FHFC financed properties, Pinellas County

PAGE 131

131 Table 65. Comparison of assisted properties built/placed in service 19631979 and 19801994, Pinellas County Age Ra nge 1963 1979 N =13 1980 1994 N =29 MannWhitney U test Variable Median Median Z P (2 tailed) Crime 6.40 7.60 .994 .345 Decline 1.00 1.00 1.153 .249 FCAT 6.33 6.33 .863 .388 Gentrification 1.54 1.11 .616 .538 Affordable housing preference 5.17 4.91 .986 .324 Local accessibility 7.85 7.26 1.388 .165 Poverty rate 3.00 2.00 .597 .572 Table 66. Comparison of assisted properties built/placed in service 19631994 and 19942008, Pinellas County Age Range 1963 1979 N =43 1980 1994 N =37 MannWhitney U test Variable Median Median Z P (2 tailed) Crime 7.40 8.20 1.645 1.00 Decline 1.00 1.00 .286 .775 FCAT 6.33 5.67 .518 .604 Gentrification 1.15 1.00 1.179 .238 Affordable housing preference 4.92 5.00 .163 .870 Local accessibility 7.52 7.23 1.650 .099 Poverty rate 3.00 5.00 1.165 .244

PAGE 132

132 Figure 66. Properties in the assisted housing inventory categorized by age for test between combined HUDand older FHFC financed properties and LIHTC properties, Pinellas County

PAGE 133

133 CHAPTER 7 CONCLUSIONS AND RECOMMENDATIONS In making determinations about the value of assisted housing to the affordable housing stock, neighborhood quality and location efficiency need to be taken into consideration (Newman & Schnare, 1997; Holtzclaw et al., 2002). Indeed, as Briggs (2005b) reminds us: Location matters for economic returns, quality of life, and many other reasons (p. 17). The discourse of preservation has continued to prioritize HUD financed properties above all others based on their deeper subsidy levels, but without fully considering the geography of opportunity in which they are situated. Scarce resources could be inefficiently allocated toward properties that exact unforeseen costs on their residents. While younger properties may not offer the same level of subsidy as offered through HUD assistance programs, the benefits accrued from residing in a higher quality area can be of even greater value. As discussed in Chapter 3, neighborhood characteristics can affect individual outcomes along a number of dimensions, including access to employment, educational attainment, and physical and mental health. Living in an isolated area that requires significant transportation expenditures can quickly erode the benefits gained from living in a residence with assisted rent. Furthermore, properties in the AHI need to be evaluated based on their ability to offer affordable housing in neighborhoods in which such housing is in short supply or in danger of disappearance (Kennedy & Leonard, 2001). This chapter evaluates the results of the statistical analyses, placing them in the context of the geography of opportunity and research questions motivating this study, and offers recommendations for preservation initiatives and future research based on these conclusions.

PAGE 134

134 The results of this study show that generational cohorts in the AHI do not always perform consistently, either within or between study areas. This pattern of incoherence is of great consequence, as it seems to refute the hypothesis stated at the outset of this study that the locational characteristics of assisted housing will differ in predictable ways based on the generation of housing production to which they belong. This is evidenced in the internal inconsistencies wherein one cohort may have significant ly higher or lower values along one dimension of suitability, but also have significantly different values along a different, contrasting dimension. This tendency is manifest in the results of the test for differences in Duval County between HUD and older FHFC financed properties (19631994) on the one hand, and newer, predominantly LIHTC financed properties (19952008) on the other. In this instance, the newer properties have significantly higher values for variables reflecting school quality (FCAT scores) and socioeconomic status (poverty rate), while simultaneously having significantly higher suitability values for the variable of neighborhood decline. Between counties, different generational cohorts exhibit significantly higher or lower values for differ ent variables. That is, while areas in which properties developed under a specific program may be significantly more suitable along a particular dimension than areas in which properties developed under another program are located in one of the study areas, this superiority is not always replicated in the others. For each test, there is only one variable for which the younger cohorts possess significantly higher suitability values in more than one county. For example, while the areas in which older FHFC fina nced properties are located have significantly higher suitability values for school quality than HUD financed properties in Duval

PAGE 135

135 County, they do not exhibit this relatively superior school performance over HUD financed properties in the other two counties Despite the variance of these results, they do provide some support for the hypothesis that older properties are more likely to be located in the central city, while newer units are more likely to be located on the outskirts. Indeed, if any predictable pattern may be found in the results, it is that younger, predominantly statefinanced properties are generally located in areas possessing relatively higher neighborhood quality while older, predominantly federally financed properties are generally located in areas with relatively greater location efficiency This distinction appears in the results of both tests, that is, not only does it exist between age cohorts representing the larger sweep of the generational shift (HUD and older FHFC versus predominantly LIHTC developments), but it also exists between groups specifically representing potentially at risk older properties (HUD versus older FHFC financed properties). Younger cohorts in both tests generally have significantly higher suitability values for variables measuring neighborhood quality (i.e., crime risk, FCAT scores, poverty rate), while older cohorts in both tests generally have significantly higher suitability values for variables measuring location efficiency (i.e., local accessibility, transportation cost) (see Table 71). An additional similarity found among study areas is the significantly higher suitability values for gentrification for older cohorts. This appears for the HUD financed cohort (19631979) in the test between it and older FHFC financed properties (19801994) in Orange County, as well as for the combined HUD and older FHFC financed properties (19631994) in the test between them and the cohort representing predominantly LIHTC developments (19952008) in Duval County.

PAGE 136

136 In essence the prioritization of one cohort over another for preservation seems to represent a tradeoff between neighborhood quality and location efficiency, two important dimensions shaping the overall geography of opportunity. While neighborhood quality can have a significant impact on the quality of life for individuals, potentially offering an environment with less environmental stressors and greater access to local social networks, location efficiency is critical for many to access services and employment. How ever, the degree of this tradeoff appears to be variable. The difference between median poverty rate suitability values (high suitability=low poverty) for HUDfinanced properties ( Mdn =1) and older FHFC financed properties ( Mdn =6) in Duval County is quite stark, but it is altogether absent from Orange and Pinellas Counties. On the other hand, in both Duval and Orange Counties, older FHFC financed properties have significantly higher suitability values for crime risk than HUD assisted properties, suggesting that the trade off between cohorts may be further distilled into one between safety and accessibility. In Duval County, though, the aggregate suitability of locations in which older FHFC financed properties are located makes the tradeoff between location efficiency and neighborhood quality considerably more pronounced than it is in Orange, not to mention Pinellas County. In the test of differences between HUDand older FHFC financed properties, the latter has significantly higher suitability values for fo ur indicators of neighborhood quality, including overall affordable housing preference, while the former has significantly higher transportation values. The distinct historical patterns of development of these study areas, specifically their principle cit ies, seem to account, at least in part, for both the consistencies and inconsistencies observed in the results. While this study has heretofore discussed its

PAGE 137

137 study areas solely in terms of county boundaries, the trajectories of the cities contained within them may shed light on the differences between the results for Duval County and the results for Orange and Pinellas Counties. As previously mentioned, Duval County and the City of Jacksonville were consolidated in 1968, eliminating competing governmental bodies and significantly expanding Jacksonvilles city limits. It is to this geographic expansion that Jacksonville owes its status as the largest city in Florida, both in terms of land mass and population. In 2006, Jacksonvilles population was 794,255, while Orlandos was only 220,186, and St. Petersburgs was only 248,098.43It seems likely that this metropolitanwide sphere of influence exerted by Jacksonville may also have influenced the spatial pattern of assisted housing development in Duval County. As discussed in Chapter 5, Jacksonville has continued to promote the development of affordable housing in its Neighborhood Action Plan (NAP) areas, which are distressed neighborhoods located in or near its historic city limits. As Jacksonville has not needed to compete with neighboring municipalities for resources needed for assisted housing production, the spatial pattern of assisted housing in Duval County is far more tightly knit than it is in Orange and Pinellas Counties. Orlando and David Rusk (2003) refers to this type of boundary expansion as an expression of municipal elasticity (p. 12). An important dimension of elasticity is the central citys ability to e xercise influence and control over the metropolitan area (Smith et al., 1997, p. 65). Jacksonvilles high elasticity allowed it to capture its suburban growth and prevent metropolitan governmental fragmentation, which was a genuine concern for the citys leaders prior to consolidation (Crooks, 2001). 43 These figures are for cities, not metropolitan statistical areas. Population estimates from U.S. Census Bureau, State & County QuickFacts, http://quickfacts.census.gov/qfd/states/12000 lk.html

PAGE 138

138 St. Petersburg, the primary cities of Orange and Pinellas Counties, respectively, have not been able to manifest nearly the same degree of gravitational pull on the location of assisted housing units. Both of these counties contain numerous municipalities engaged in assisted housing development Orang e County contains 15 incorporated municipalities and Pinellas County contains 24making the spatial pattern of assisted housing appear more diffuse when viewed at the county scale. It should be noted that the governmental fragmentation of Pinellas County i s a clear reflection of its geographic fragmentation (it is a relatively small peninsula with a number of barrier islands), which likely also accounts for the failure of its assisted properties to exhibit a coherent shift in locational attributes. Policy R ecommendations As shown in this study, the decision to preserve one generation of assisted housing over another reflects a tradeoff between neighborhood quality and location efficiency. Therefore, the decision to prioritize one cohort over another should be decided by policy preferences and local housing conditions. The policy focus in many urban areas is the deconcentration of poverty. In fact, the concentration of poverty is among the most urgent challenges confronting urban policy makers today (Turner 1998, p. 373). Duval County, specifically the City of Jacksonville, considers poverty deconcentration an affordable housing goal due to the pronounced patterns of socioeconomic and minority segregation apparent in the city (Jacksonville, 2005). Jacksonv ille may then want to consider prioritizing the preservation of older FHFC financed properties at risk of failing out of the affordable housing inventory as they have significantly higher suitability values for the poverty rate variable. On the other hand, the city also holds a strong interest in preserving affordable housing opportunities in

PAGE 139

139 gentrifying areas, so the preservation of the HUD assisted stock may be a greater priority to Jacksonville. Both Pinellas County and Jacksonville target resources toward the preservation of affordable housing options as a redevelopment tool in distressed neighborhoods. Given this policy climate, it may make even less sense for them to prioritize younger properties. Preserving assisted housing near transit and shopping c an greatly improve the geography of opportunity available to lower income families and individuals. In order to preserve the federally assisted housing near transit, state and local housing entities should pursue federal funding available through such prog rams as the joint sustainable communities initiative for preservation resources, which promote the development of walkable, transit rich environments. In order to prepare for future funding applications, state and local agencies should begin to collect dat a related to the location efficiency of assisted housing, particularly those at risk of failing out of the AHI. The Shimberg Center for Housing Studies is especially positioned to conduct this research. A final recommendation is for the inclusion of preser vation activities for location efficient assisted housing units in plans and policies addressing larger sustainable development objectives such as transit oriented development. Future Research and Limitations Future research could expand upon this study by conducting the statistical tests of difference following the completion of the Affordable Housing Suitability (AHS) model. This would allow for a more complete comparison of the counties based on transportation suitability values and take into considerati on the demand for housing generated by commercial, industrial, and residential development. While local accessibility is an importance measure of value for assisted housing, it does not provide

PAGE 140

140 a robust measure for the relationship between employment deman d and residence. The inclusion of the linkage analysis would better enable the model to assess properties for preservation based on the degree to which they address the jobhousing balance. When home and work are dislocated by lengthy distances, the added transportation expenditures can be quite onerous. With the completion of the AHS mode, pairwise comparison will have integrated community preferences into the suitability values produced by the model. Thus, the scores produced by the completed model may vary from those used in this study, which were produced by assigning equal weights to layers during the weighted overlay procedures. Future studies could also compare a greater number of counties to test for the consistency of this studys findings. Table 7 1. Incidence of significantly higher suitability values Test of differences between HUD and older FHFC financed properties Test of differences between HUD and FHFC financed properties and LIHTC properties 1963 1979 1980 1994 1963 1994 1995 2008 Duval County Transportation Crime FCAT AH preference Poverty rate Gentrification Local accessibility Transportation Decline FCAT Poverty rate Orange County Gentrification Local accessibility Crime Decline FCAT Decline Pinellas County

PAGE 141

141 APPENDIX A NEIGHBORHOOD CHANGE INDICATORS Neighborhood Change as Gentrification Long Term Change Indicators (Decennial census block groups) Variable Explanation and Sources Gain Non Hispanic white population, 19802000 Increase in this indicator may be indicative of displacement of minorities associated with gentrification. (Freeman, 2005; Freeman & Braconi, 2004; Kennedy & Leonard, 2001; Levy et al., 2006; Wyly & Hammel, 199 9) Gain 2534 year age cohort, 19802000 5564 year age cohort, 19802000 Young professionals and baby boomers have been identified in the literature as potential gentrifiers Elements of these cohorts may return to the city to take advantage of urban amenities (Hall & Ogden, 1992; Kennedy & Leonard, 2001; Lees, Slater, & Wyly, 2007; Levy et al., 2006; Ley, 1996; McKinnish, Walsh, & White, 2009; Wyly & Hammel, 1999) Gain Median household income, 1980 2000 Gains in household income have been identified as indicative of gentrification or decline depending on the degree of change. The influx of higher income residents is a key component of most definitions of gentrification (Freeman, 2005; Kennedy & Leonard, 2001; Lees et al., 2007; Levy et al., 2006; Ley, 1996; Wyly & Hammel, 1999) Gain Professional/white collar employment, 19802000 Gain indicates influx of the professional class, w hich has shown interest in urban amenities, possesses the wealth to purchase and upgrade housing, participates in housing speculation, and contributes to displacement pressures (Lees et al., 2007; Ley, 1996; Smith, 1996; Wyly & Hammel, 1999) Gain High educational attainment, 19802000 Increasing educational attainment indicates an influx of wealthy professional households, and is associated with the cultural component of gentrification (Freeman, 2005; Ley, 1996; Wyly & Hammel, 1999) Gain Owner occupied housing, 19802000 Areas with owner occupied housing are typically wealthier than those with predominantly renter occupied housing. Gains in owner occupied units may indicate growing investment in rehabilitation or new development and in the case of a gentrifying area, increasing property values (Lees et al., 2007; Levy et al., 2006; Wyly & Hammel, 1999)

PAGE 142

142 Short Term Change indicators (H ome M ortgage D isclosure A ct [HMDA] data, County property appraiser sales data) Variable Explanation and Sources Gain Loan originations per capita, 20052008 An increase in loan origination signals increased investment and is correlated with neighborhood change in previously less affluent urban neighborhoods (Hackworth, 2007; Lees et al., 2007; Pettit & Droesch, 2008) Gain Median household income of home purchase borrowers, 2005 2007 Recent gains in income may also indicate gentrif ication in some urban areas (Freeman, 2005; Kennedy & Leonard, 2001; Lees et al., 2007; Levy et al., 2006; Pettit & Droesch, 2008; Wyly & Hammel, 1999) Gain Median home sales price ($ per square foot), 2000 2008 Sharply increasing p roperty values, especially in lower income areas, may indicate a process of gentrification (Turner & Snow, 2001) Static Indicators of Gentrification potential Variable Explanation and Sources Historic district designation (local and national) Historic district designation may raise property values and create maintenance obligations proving onerous for incumbent residents (Coulson & Leichenko, 2004; Turner & Snow, 200 1)

PAGE 143

143 Neighborhood Change as Decline Long Term Change Indicators (Decennial census block groups) Variable Explanation and Sources Loss Non hispanic white population, 19802000 Increase in minority population (loss of non hispanic whites) suggests decline, as it captures processes of neighborhood succession and possibly, tipping. (Freeman, 2005; Freeman & Braconi, 2004; Kennedy & Leonard, 2001; Levy et al., 2006; Wyly & Hammel, 1999) Loss Median household income, 19802000 Sharp losses in neighborhood income are associated with decline (Freeman, 2005; Kennedy & Leonard, 2001; Lees et al., 2007; Levy et al., 2006; Ley, 1996; Wyly & Hammel, 1999) Loss Professional/white collar employment, 19802000 Loss of professional class is indicative of decline (Lees et al., 2007; Ley, 1996; Smith, 1996; Wyly & Hammel, 1999) Loss High educational attainment, 19802000 Decreasing educational attainment suggest neighborhood decline (Freeman, 2005; Ley, 1996; Smith, 1996; Wyly & Hammel, 1999) Loss Owner occupied housing, 19802000 Decreases in home ownership are indicative of decline (Lees et al., 2007; Levy et al., 2006; Wyly & Hammel, 1999) Gain Femaleheaded households, 19802000 An increase in the measure of this indicator is associated with decline (Beauregard, 1990; Galster, 2000; Newman & Wyly, 2006)

PAGE 144

144 Short Term Change Indicators (HMDA data, U nited S tates P ostal S ervice [USPS] data) Variable Explanation and Sources Loss Loan originations per capita, 2005 2008 Decrease in loan originations may be correlated with potential long term neighborhood decline Loss Median household income of home purchase borrowers, 2005 2007 Strong losses in neighborhood household income suggest decline (Freeman, 2005; Kennedy & Leonard, 2001; Lees, Slater, & Wyly, 2009; Levy et al., 2006; Pettit & Droesch, 2008; Wyly & Hammel, 1999) Gain High cost loan originations, 20052008 A concentration of high cost loans (associated with subprime mortgages) increases the likelihood of foreclosure and decline (Immergluck & Smith, 2006; Pet tit & Droesch, 2008) Gain USPS vacancy rate, 20062009 The Neighborhood Stabilization Program tracks the USPS vacancy rate. Vacant properties are associated with neighborhood decline (Beauregard, 1990; Cohen, 2001) Static Indicators (HUD/Neighborhood Stabilization Program data) Variable Explanation and Sources Estimated foreclosure abandonment risk score, 20072008 This is an indicator of neighborhood decline taken from the Neighborhood Stabilization Program data set. Foreclosure is closely associated with neighborhood decline. Like vac ancy, it has many negative spillover effects ( Scheutz, Been, & Ellen, 2008) Predicted 18 month underlying problem foreclosure rate, 20072008 This is an indicator of neighborhood decline taken from the Neighborhood Stabilization Program data set (Immergluck & Smith, 2006; Scheutz et al., 2008)

PAGE 145

145 APPENDIX B TRANSPORTAT ION COST INDICATORS Indicator Variables Density Developed area as a % of total neighborhood area Residential area as a % of developed area Diversity Building square feet (retail commercial) Building square feet (office/service) Building square feet (industrial) Design Road miles per developed area Number of intersections per road mile Destination Network distance to nearest regional residential center Network distance to nearest activity center Range of network distances to regional residential center

PAGE 146

146 APPENDIX C MANN WHITNEY U TEST RESULTS Duval County: Test between 19631979 and 19801994 Descriptive Statistics N Mean Std. Deviation Minimum Maximum CRIME 83 7.17 2.173 1 9 DECLINE 83 1.57229232 71 .8417046679 4 1.00000203 4.08334160 FCAT 83 4.81525623 99 1.555871124 51 1.66666496 8.33332539 GENT 83 1.58370892 42 .9689926386 4 .99999827 5.41557598 AFF HOUSING PREF 79 5.09481367 27 .6466975044 1 3.44125962 6.53876829 LOCAL ACCESS 83 6.92683409 60 1.347151035 41 2.07291651 8.66666794 POVMASK 83 .51 .503 0 1 POVRATE 83 3.92 3.144 1 9 TRANSP 83 8.289 1.1480 2.5 9.0 AGERNG 83 1.48 .503 1 2 Ranks AGERNG N Mean Rank Sum of Ranks CRIME 63 79 43 35.93 1545.00 80 94 40 48.53 1941.00 Total 83 DECLINE 63 79 43 38.15 1640.50 80 94 40 46.14 1845.50 Total 83 FCAT 63 79 43 36.20 1556.50 80 94 40 48.24 1929.50 Total 83 GENT 63 79 43 45.29 1947.50 80 94 40 38.46 1538.50 Total 83 AFF HOUSING PREF 63 79 42 32.63 1370.50 80 94 37 48.36 1789.50 Total 79 LOCAL ACCESS 63 79 43 46.16 1985.00

PAGE 147

147 Ranks AGERNG N Mean Rank Sum of Ranks 80 94 40 37.53 1501.00 Total 83 POVMASK 63 79 43 35.48 1525.50 80 94 40 49.01 1960.50 Total 83 POVRATE 63 79 43 34.20 1470.50 80 94 40 50.39 2015.50 Total 83 TRANSP 63 79 43 47.63 2048.00 80 94 40 35.95 1438.00 Total 83 Test Statisticsa CRIME DECLINE FCAT GENT AFF HOUSING PREF LOCAL ACCESS Mann Whitney U 599.000 694.500 610.500 718.500 467.500 681.000 Wilcoxon W 1545.000 1640.500 1556.500 1538.500 1370.500 1501.000 Z 2.474 1.680 2.299 1.351 3.041 1.631 Asymp. Sig. (2 tailed) .013 .093 .022 .177 .002 .103 a. Grouping Variable: AGERNG Test Statisticsa POVMAS K POVRAT E TRANSP Mann Whitney U 579.500 524.500 618.000 Wilcoxon W 1525.500 1470.500 1438.000 Z 2.952 3.213 2.347 Asymp. Sig. (2 tailed) .003 .001 .019 a. Grouping Variable: AGERNG

PAGE 148

148 Duval County: Test between 19631995 and 19952008 Descriptive Statistics N Mean Std. Deviation Minimum Maximum CRIME 134 7.32 2.179 1 9 DECLINE 134 1.74129703 78 .8887171388 5 1.00000203 4.08334160 FCAT 134 5.15919884 35 1.653690393 68 1.66666496 8.33332539 GENTRIF 134 1.45701449 72 .8486999068 8 .99999827 5.41557598 RASTERVAL U 125 5.12759935 37 .6049738168 2 3.44125962 6.53876829 GOAL01 125 5.12759935 37 .6049738168 2 3.44125962 6.53876829 LOCAL_ACC E 134 6.57151911 78 1.539238533 08 2.07291651 8.66666794 POVMASK 134 .60 .491 0 1 POVRATE 134 4.86 3.263 1 9 TRANSP 134 8.149 1.1389 2.5 9.0 AGERNG 134 1.38 .487 1 2 Ranks AGERNG N Mean Rank Sum of Ranks CRIME 63 94 83 63.66 5284.00 95 08 51 73.75 3761.00 Total 134 DECLINE 63 94 83 59.60 4947.00 95 08 51 80.35 4098.00 Total 134 FCAT 63 94 83 59.85 4967.50 95 08 51 79.95 4077.50 Total 134 GENTRIF 63 94 83 73.38 6090.50 95 08 51 57.93 2954.50 Total 134 RASTERVAL U 63 94 79 61.20 4834.50 95 08 46 66.10 3040.50 Total 125 GOAL01 63 94 79 61.20 4834.50 95 08 46 66.10 3040.50

PAGE 149

149 Ranks AGERNG N Mean Rank Sum of Ranks Total 125 LOCAL_ACC E 63 94 83 76.78 6372.50 95 08 51 52.40 2672.50 Total 134 POVMASK 63 94 83 60.90 5055.00 95 08 51 78.24 3990.00 Total 134 POVRATE 63 94 83 56.61 4698.50 95 08 51 85.23 4346.50 Total 134 TRANSP 63 94 83 74.55 6188.00 95 08 51 56.02 2857.00 Total 134 Test Statisticsa CRIME DECLINE FCAT GENTRIF RASTERVAL U GOAL01 Mann Whitney U 1798.000 1461.000 1481.500 1628.500 1674.500 1674.500 Wilcoxon W 5284.000 4947.000 4967.500 2954.500 4834.500 4834.500 Z 1.551 3.174 2.942 2.407 .730 .730 Asymp. Sig. (2 tailed) .121 .002 .003 .016 .466 .466 a. Grouping Variable: AGERNG Test Statisticsa LOCAL_ACC E POVMAS K POVRAT E TRANSP Mann Whitney U 1346.500 1569.000 1212.500 1531.000 Wilcoxon W 2672.500 5055.000 4698.500 2857.000 Z 3.529 2.962 4.246 2.786 Asymp. Sig. (2 tailed) .000 .003 .000 .005 a. Grouping Variable: AGERNG

PAGE 150

150 Orange County: Test between 1963 1979 and 19801994 Descriptive Statistics N Mean Std. Deviation Minimum Maximum CRIME 75 6.65 3.291 1 9 DECLINE 75 1.87000376 68 .9660393063 5 1.00000203 4.41667557 FCAT 75 6.45777133 87 1.891002126 51 2.33333111 8.99999142 GENTRIF 75 1.73709682 57 .8185544631 2 .99999827 4.09739733 GOAL01 70 4.98443729 24 .5303706998 6 3.84675622 6.15086508 LOCAL_ACC E 75 6.43222380 69 1.121758340 65 3.60416794 8.58333492 POVMASK 75 .35 .479 0 1 POVRATE 75 5.05 3.110 1 9 TRANSP 75 8.48 .690 7 9 AGERNG 77 1.79 .408 1 2 Ranks AGERNG N Mean Rank Sum of Ranks CRIME 63 79 16 29.09 465.50 80 94 59 40.42 2384.50 Total 75 DECLINE 63 79 16 28.66 458.50 80 94 59 40.53 2391.50 Total 75 FCAT 63 79 16 39.88 638.00 80 94 59 37.49 2212.00 Total 75 GENTRIF 63 79 16 49.75 796.00 80 94 59 34.81 2054.00 Total 75 GOAL01 63 79 16 40.25 644.00 80 94 54 34.09 1841.00 Total 70 LOCAL_ACC E 63 79 16 54.94 879.00 80 94 59 33.41 1971.00 Total 75 POVMASK 63 79 16 36.72 587.50

PAGE 151

151 Ranks AGERNG N Mean Rank Sum of Ranks 80 94 59 38.35 2262.50 Total 75 POVRATE 63 79 16 42.47 679.50 80 94 59 36.79 2170.50 Total 75 TRANSP 63 79 16 42.91 686.50 80 94 59 36.67 2163.50 Total 75 Test Statisticsa CRIME DECLINE FCAT GENTRIF GOAL01 LOCAL_ACC E Mann Whitney U 329.500 322.500 442.000 284.000 356.000 201.000 Wilcoxon W 465.500 458.500 2212.000 2054.000 1841.000 1971.000 Z 1.981 1.993 .392 2.473 1.063 3.505 Asymp. Sig. (2 tailed) .048 .046 .695 .013 .288 .000 a. Grouping Variable: AGERNG Test Statisticsa POVMAS K POVRAT E TRANSP Mann Whitney U 451.500 400.500 393.500 Wilcoxon W 587.500 2170.500 2163.500 Z .322 .937 1.116 Asymp. Sig. (2 tailed) .748 .349 .264 a. Grouping Variable: AGERNG

PAGE 152

152 Orange County: Test between 1963 1995 and 1995 2008 Descriptive Statistics N Mean Std. Deviation Minimum Maximum CRIME 154 7.33766237 55 2.827365771 08 1.00000000 9.00000000 DECLINE 154 2.13420342 65 .9602366710 4 1.00000203 4.75000954 FCAT 154 6.18614099 60 1.818561048 53 2.33333111 8.99999142 GENTRIF 154 1.61774487 81 .7130773887 8 .99999827 4.09739733 GOAL01 135 4.96387961 56 .5008346064 3 3.66706848 6.15086508 LOCAL_ACC E 154 6.20630571 14 1.278419447 08 1.44791722 8.58333492 POVMASK 154 .37 .484 0 1 POVRATE 154 5.24 3.012 1 9 TRANSP 154 8.45 .712 6 9 AGERNG 156 1.51 .502 1 2 Ranks AGERNG N Mean Rank Sum of Ranks CRIME 63 94 75 72.11 5408.50 95 08 79 82.61 6526.50 Total 154 DECLINE 63 94 75 64.11 4808.00 95 08 79 90.22 7127.00 Total 154 FCAT 63 94 75 84.95 6371.00 95 08 79 70.43 5564.00 Total 154 GENTRIF 63 94 75 83.33 6249.50 95 08 79 71.97 5685.50 Total 154 GOAL01 63 94 70 69.07 4835.00 95 08 65 66.85 4345.00 Total 135 LOCAL_ACC E 63 94 75 83.11 6233.00 95 08 79 72.18 5702.00 Total 154

PAGE 153

153 Ranks AGERNG N Mean Rank Sum of Ranks POVMASK 63 94 75 75.69 5677.00 95 08 79 79.22 6258.00 Total 154 POVRATE 63 94 75 74.81 5610.50 95 08 79 80.06 6324.50 Total 154 TRANSP 63 94 75 79.47 5960.50 95 08 79 75.63 5974.50 Total 154 Test Statisticsa CRIME DECLINE FCAT GENTRIF GOAL01 LOCAL_ACC E Mann Whitney U 2558.500 1958.000 2404.000 2525.500 2200.000 2542.000 Wilcoxon W 5408.500 4808.000 5564.000 5685.500 4345.000 5702.000 Z 1.592 3.667 2.034 1.624 .330 1.520 Asymp. Sig. (2 tailed) .111 .000 .042 .104 .741 .128 a. Grouping Variable: AGERNG Test Statisticsa POVMAS K POVRAT E TRANSP Mann Whitney U 2827.000 2760.500 2814.500 Wilcoxon W 5677.000 5610.500 5974.500 Z .586 .739 .579 Asymp. Sig. (2 tailed) .558 .460 .563 a. Grouping Variable: AGERNG

PAGE 154

154 Pinellas County: Test between 19631979 and 19801994 Descriptive Statistics N Mean Std. Deviation Minimum Maximum CRIME 42 6.39047613 60 2.449072457 42 1.79999995 9.00000000 DECLINE 42 1.95040076 02 1.433494489 73 1.00000203 5.58334446 FCAT 42 5.96824796 6 1.718397783 9 2.3333311 8.9999914 GENTRIF 42 1.87886218 21 1.404018611 01 .99999827 5.65908194 GOAL01 35 5.00157527 97 .6143340765 8 3.91235113 6.19438791 LOCAL_ACC E 42 7.44667824 17 .8722803968 2 5.17708445 8.59375191 POVMASK 42 .31 .468 0 1 POVRATE 42 3.83 3.068 1 9 AGERNG 42 1.69 .468 1 2 Ranks AGERNG N Mean Rank Sum of Ranks CRIME 63 79 13 18.85 245.00 80 94 29 22.69 658.00 Total 42 DECLINE 63 79 13 18.65 242.50 80 94 29 22.78 660.50 Total 42 FCAT 63 79 13 19.12 248.50 80 94 29 22.57 654.50 Total 42 GENTRIF 63 79 13 23.19 301.50 80 94 29 20.74 601.50 Total 42 GOAL01 63 79 10 20.70 207.00 80 94 25 16.92 423.00 Total 35 LOCAL_ACC E 63 79 13 25.42 330.50 80 94 29 19.74 572.50 Total 42 POVMASK 63 79 13 21.46 279.00

PAGE 155

155 Ranks AGERNG N Mean Rank Sum of Ranks 80 94 29 21.52 624.00 Total 42 POVRATE 63 79 13 23.12 300.50 80 94 29 20.78 602.50 Total 42 Test Statisticsb CRIME DECLINE FCAT GENTRIF GOAL01 LOCAL_ACC E Mann Whitney U 154.000 151.500 157.500 166.500 98.000 137.500 Wilcoxon W 245.000 242.500 248.500 601.500 423.000 572.500 Z .944 1.153 .863 .616 .986 1.388 Asymp. Sig. (2 tailed) .345 .249 .388 .538 .324 .165 Exact Sig. [2*(1 tailed Sig.)] .360a .318a .404a .554a .339a .167a a. Not corrected for ties. b. Grouping Variable: AGERNG Test Statisticsb POVMAS K POVRAT E Mann Whitney U 188.000 167.500 Wilcoxon W 279.000 602.500 Z .017 .597 Asymp. Sig. (2 tailed) .986 .551 Exact Sig. [2*(1 tailed Sig.)] 1.000a .572a a. Not corrected for ties. b. Grouping Variable: AGERNG

PAGE 156

156 Pinellas County: Test between 19631995 and 19952008 Descriptive Statistics N Mean Std. Deviation Minimum Maximum CRIME 80 6.67 2.450 1 9 DECLINE 80 1.91146218 39 1.375096641 74 1.00000203 5.66667795 FCAT 80 5.92499404 71 1.721848366 03 2.33333111 8.99999142 GENTRIF 80 1.73485921 86 1.253436301 23 .99999827 5.65908194 GOAL01 67 5.01027401 96 .5836992299 6 3.91235113 6.19438791 LOCAL_ACC E 80 7.18125159 78 1.169625814 69 2.79166794 8.59375191 POVMASK 80 .35 .480 0 1 POVRATE 80 4.37 3.188 1 9 AGERNG 80 1.46 .502 1 2 Ranks AGERNG N Mean Rank Sum of Ranks CRIME 63 94 43 36.58 1573.00 95 08 37 45.05 1667.00 Total 80 DECLINE 63 94 43 41.10 1767.50 95 08 37 39.80 1472.50 Total 80 FCAT 63 94 43 41.73 1794.50 95 08 37 39.07 1445.50 Total 80 GENTRIF 63 94 43 43.21 1858.00 95 08 37 37.35 1382.00 Total 80 GOAL01 63 94 36 33.64 1211.00 95 08 31 34.42 1067.00 Total 67 LOCAL_ACC E 63 94 43 44.48 1912.50 95 08 37 35.88 1327.50 Total 80 POVMASK 63 94 43 38.59 1659.50 95 08 37 42.72 1580.50

PAGE 157

157 Ranks AGERNG N Mean Rank Sum of Ranks Total 80 POVRATE 63 94 43 37.78 1624.50 95 08 37 43.66 1615.50 Total 80 Test Statisticsa CRIME DECLINE FCAT GENTRIF GOAL01 LOCAL_ACC E Mann Whitney U 627.000 769.500 742.500 679.000 545.000 624.500 Wilcoxon W 1573.000 1472.500 1445.500 1382.000 1211.000 1327.500 Z 1.645 .286 .518 1.179 .163 1.650 Asymp. Sig. (2 tailed) .100 .775 .604 .238 .870 .099 a. Grouping Variable: AGERNG Test Statisticsa POVMAS K POVRAT E Mann Whitney U 713.500 678.500 Wilcoxon W 1659.500 1624.500 Z .958 1.165 Asymp. Sig. (2 tailed) .338 .244 a. Grouping Variable: AGERNG

PAGE 158

158 LIST OF REFERENCES AcevedoGarcia, D., Osypuk, T. L., Werbel, R. E., Meara, E. R., Cutler, D. M., & Berkman, L. F. (2004). Does housing mobility improve health? Housing Policy Debate, 15 (1), 49 98. Achtenberg, E. P. (2009). Section 236 decoupling. Retrieved June 14, 2010, from http://www.lisc.org/fil es/8207_file_section_236_decoupling.pdf Achtenberg, E. P. (2002). Stemming the tide: A handbook on preserving multifamily subsidized housing New York: Local Initiatives Support Corporation. Retrieved May 14, 2010, from http://www.lisc.org/files/893_file_asset_upload_file15_838.pdf Affordable Housing Study Commission. (2006). A preservation strategy for Floridas affordable multifamily housing. Tallahassee, FL: Author. Retrieved May 17, 2010, from http://www.floridahousing.org/NR/rdonlyres/63332252F3EA 486A 9F59 B95E7F6CDE56/0/AHSCFinalReport2006.pdf Al Shalabi, M A., Mansor, S. B., Ahmed, N. B., & Shiriff, R. (2006, October). GIS Based Multicriteria Approaches to Site Suitability Analysis Paper presented at the XXII FIG Congress, Munich, Germany. Barber,G. M. (1988). Elementary statistics for geographers New Y ork: The Guilford Press. Basolo, V. (1999). Passing the housing policy baton in the US: Will cities take the lead? Housing Studies 14(4), 433 452. Beauregard, R. A. (1990). Trajectories of neighborhood change: the case of gentrification. Environment and Planning A 22 (7), 855 874. Blackman, A., & Krupnick, A. (2001). Location efficient mortgages: Is the rationale sound? Journal of Policy Analysis and Management 20 (4), 633 649. Block, J. P., Scribner, R. A., & De Salvo, K. B. (2004). Fast food, race/et hnicity, and income: A geographic analysis. American Journal of Preventive Medicine, 27(3), 211 217. Bnai Kashani, R. (1989). A new method for site suitability analysis: The analytic hierarchy process. Environmental Management 13 (6), 685 693. Boarnet, M. G., & Crane, R. (2001). Travel by Design: The influence of urban form on travel New York: Oxford University Press. Bodaken, M. (2002). The increasing shortage of affordable rental housing in America: Action items for preservation. Housing Facts and Findi ngs 4 (4), 1 8.

PAGE 159

159 Boer, J. T., Pastor, M., Jr., Sadd, J. L., Snyder, L. D. (1997). Is there environmental racism? The demographics of hazardous waste in Los Angeles County. Social Science Quarterly 78(4), 793 810. Brewster, K. L. (1994). Race differences in sexual activity among women: The role of neighborhood characteristics. American Sociological Review 59(3), 408 424. Briggs, X. S. (2005a). Introduction. In X. D. Briggs (Ed.), The Geography of Opportunity: Race and Housing Choice in Metropolitan America (pp. 1 16). Washington, DC: Brookings Institution Press. Briggs, X. S. (2005b). More pluribus, less unum?. In X. D. Briggs (Ed.), The Geography of Opportunity: Race and Housing Choice in Metropolitan America (pp.17 41). Washington, DC: Brookings Institutio n Press. Brooks Gunn, J., Duncan, G. J., Klebanov, P. K., & Seland, N. (1993). Do neighborhoods influence child and adolescent development? The American Journal of Sociology 99 (2), 353 395. Bullard, R. D. (2000). Dumping in Dixie: Race, class, and environmental quality Boulder, CO: Westview Press. Burrough, P. A. (1986). Principles of geographical information systems for land resources assessment Oxford: Clarendon Press. Carr, M. H., & Zwick, P. (2005). Using GIS suitability analysis to identify potential future land use conflicts in North Central Florida. Journal of Conservation Planning, 1 58 73. Carr, M. H & Zwick, P. D. (2007). In Smart landuse analysis: The LUCIS model landuse conflict identification strategy Redlands, CA: ESRI Press. C ase, A. C., & Katz, L. F. (1991). The company you keep: The effects of family and neighborhood on disadvantaged youths NBER Working Paper No. 3705. Cambridge, MA: National Bureau of Economic Research. Center for Housing Policy. (2009). Insights from housi ng policy research: Taking stock the role of preservation inventories in preserving affordable rental housing Washington, DC: Author. Retrieved May 19, 2010, from http://www.nhc.org/pdf/ takingstockfinalreorder.pdf Center for Neighborhood Technology. (2010). Penny wise, pound fuelish: New measures of housing and transportation affordability Retrieved May 12, 2010, from http://www.cnt.org/publi cations Cervero, R. (2002). Built environments and mode choice: toward a normative framework. Transportation Research Part D 7 265 284.

PAGE 160

160 Cervero, R., & Kockelman, K. (1997). Travel demand and the 3Ds: Density, diversity, and design. Transportation Researc h Part D 2 (3), 199 219. Chrisman, N. R. (2002). Exploring geographic information systems (2nd ed.). New York: Wiley City of Jacksonville, Planning and Development Department. (2005, October 1). 2005 2010 consolidated plan. Retrieved June 21, 2010, from http://www.coj.net/Departments/Housing+and+Neighborhoods/Community+Develo pment/default.htm City of Orlando. (2010). Growth management plan: housing element Retrieved July 4, 2010, from http://www.cityoforlando.net/planning/cityplanning/Policy%20Document/GMP%202 010/05%20%20Housing%20GOPs_Supp%20No%202.pdf City of Orlando, Housing Department. (2005). Consolidated plan for housing and community development programs Retrieved June 21, 2010, from http://www.cit yoforlando.net/housing/report.htm City of St. Petersburg. (2005). Five year strategic plan. Retrieved June 22, 2010, from http://www.stpete.org/pdf/2005housestrategicplan.pdf Clarke, K. C. (1986). Advances in geographic information systems. Computers, Environment and Urban Systems 10(3), 175 184. Coleman, J. S. (1998). Social capital in the creation of human capital. The American Journal of Sociology 94 S95 S920. Collignon, K. (1999). Ex piring affordability of Low Income Housing Tax Credit properties: The next era in preservation. Cambridge, MA: Neighborhood Reinvestment, Harvard Joint Center for Housing Studies. Collins, M. G., Steiner, F. R., & Rushman, M. J. (2001). Land use suitabilit y analysis in the United States: Historical development and promising technological achievements. Environmental Management 28 (5), 611 621. Coulson, E., & Leichenko, R. (2004). Historic preservation and neighbourhood change. Urban Studies 41 (8), 14. Cowen, D. J. (1988). GIS versus CAD versus DBMS: What are the differences? Photogrammetric Engineering and Remote Sensing 54(11), 1551 1555. Crane, J. (1991). The epidemic theory of ghettos and neighborhood effects on dropping out and teenage childbearing The American Journal of Sociology 96 (5), 1226 1259.

PAGE 161

161 Crooks, J. B. (2001). Jacksonvilles Consolidation Mayor: Hans G. Tanzler Jr. The Florida Historical Quarterly 80(2), 198 224. Diez Roux, A V., Nieto, F. J., Muntaner, C., Tyroler, H. A., Comstock, G. W., Shahar, E., Cooper, L. S., Watson, R. L., & Szklo, M. (1997). Neighborhood environmental conditions and coronary heart disease: A multilevel analysis. American Journal of Epidemiology 146 (1), 48 63. Dolbeare, C. N., & Crowley, S. (2002). Changing pr iorities: The federal budget and housing assistance, 19762007. Washington, DC: National Low Income Housing Coalition. Retrieved May 16, 2010, from http://ww w.knowledgeplex.org/kp/report/report/relfiles/nlihc_changing_priorities.pdf Downs, A. (1992). Regulatory barriers to affordable housing. Journal of the American Planning Association, 58(4), 419. Retrieved from Business Source Premier database Dubin, R. A. & Sung, C H. (1990). Specification of hedonic regression: Nonnested tests on measures of neighborhood quality. Journal of Urban Economics 27, 97 110. Duncan, G. J. (1994). Families and neighbors as sources of disadvantage in the schooling decisions o f white and black adolescents. American Journal of Education, 103 (1), 20 53. Ebdon, D. (1977). Statistics in geography Oxford: Basil Blackwell. Eisinger, P. (1998). City politics in an era of federal devolution. Urban Affairs Review, 33(3), 308 325. Ellen, I. G., Mijanovich, T., & Dillman, K N. (2001). Neighborhood effects on health: Exploring the links and assessing the evidence. Journal of Urban Affairs 23(3 4), 391 408. Ellen, I. G., & Turner, M. A. (1997). Does neighborhood matter?: Assessing rec ent evidence. Housing Policy Debate, 8 (4), 833 866. Finkel, M., Hanson, C., Hilton, R., Lam, K., & Vandawalker, M. (2006). Multifamily properties: Opting in, opting out and remaining affordable. Bethesda, MD, and Cambridge, MA: Econometrica, Inc., and Abt Associates. Florida Department of Community Affairs (2004) State of Florida consolidated plan, federal fiscal years 20052010. Retrieved May 10, 2010, from http://www.dca.state.fl.us /fhcd/cdbg/ConsolidatedPlan.cfm

PAGE 162

162 Florida Housing Finance Corporation. (2010). Discussion points for consideration for the 2011 Universal Application Cycle. Retrieved June 22, 2010, from http://www.floridahousing.org/FH ImageWebDocs/UniversalApps/2011/ImportantAnnouncements/020 Notices/0422011_Universal_Cyc le_Discussion_Points_for_42910_First_Rule_Development_Workshop.pdf Florida Housing Finance Corporation. (2009a). Strategic plan for Florida Housing Finance Corporation. Retrieved May 6, 2010, from http://floridah ousing.org/ Florida Housing Finance Corporation (2009b). 2009 Qualified Allocation Plan. Retrieved May 6, 2010, from http://floridahousing.org Florida Housing Finance Corporation and The Shimberg Center for Housing Studies. (2007). Creation of a national data infrastructure. Retrieved May 10, 2010, from http://www.preservation.shimberg.ufl.edu/ResearchAndReports.html Freeman, L. (2005). Di splacement or Succession? Urban Affairs Review 40 (4), 463 491. Freeman, L. (2004). Siting affordable housing: Location and neighborhood trends of low income housing tax credit developments in the 1990s Washington, DC: The Brookings Institution. Freeman, L., & Braconi, F. (2004). Gentrification and displacement: New York City in the 1990s. Journal of the American Planning Association, 32 (3). Garner, C. L., & Raudenbush, S. W. (1991). Neighborhood effects on educational attainment: A multilevel analysis. Sociology of Education 64 (4), 251 262. Geronimus, A. T., Hicken, M., Keene, D., & Bound, J. (2006). Weathering and age patterns of allostatic load scores among blacks and whites in the Unites States. American Journal of Public Health 96 (5), 826 833. Galster, G. C., & Killen, S. P. (1995). The geography of metropolitan opportunity: A reconnaissance and conceptual framework. Housing Policy Debate 6 (1), 7 43. Greenburg, M. R. (1999). Improving neighborhood quality: A hierarchy of needs. Housing P olicy Debate 10(3), 601 624. Gustafson, J., & Walker, C. J. (2002). Analysis of State Qualified Allocation Plans for the Low Income Housing Tax Credit program Washington, DC: The Urban Institute. Hackworth, J. R. (2007). The neoliberal city: governance, ideology, and development in American urbanism Ithaca: Cornell University Press.

PAGE 163

163 Hall, R., & Ogden, P. E. (1992). The Social Structure of New Migrants to London's Docklands: Recent Evidence from Wapping. London Journal 17 153 169. Handy, S. L., & Clif ton, K. J. (2001). Evaluating neighborhood accessibility: Possibilities and practicalities. Journal of Transportation and Statistics 4 67 78. Harris, B., & Batty, M. (1993). Locational models, geographic information and planning support systems. Journal of Planning Education and Research 12 184 198. Hartman, C. (1998). The case for a right to housing. Housing Policy Debate, 9 (2), 223 246. Hays, R. A. (1995). The federal government and urban housing: Ideology and change in public policy Albany, NY: State University of New York Press. Heikkila, E. (1998). GIS is dead; long live GIS! Journal of the American Planning Association 64 (3), 350 360. Higgs, G. (2006). Integrating multicriteria techniques with geographical information systems in waste facilit y location to enhance public participation. Waste Management & Research, 24, 105 117. Hopkins, L. D. (1977). Methods for generating land use suitability maps: A comparative evaluation. Journal of the American Institute of Planners 43(4), 386 400. Holtzclaw, J., Clear, R., Dittmar, H., Goldstein, D., & Haas, P. (2002). Location efficiency: Neighborhood and socioeconomic characteristics determine auto ownership and usestudies in Chicago, Los Angeles and San Francisco. Transportation Planning and Technology 25, 1 27. Holzer, H. J., & Stoll, M. A. (2007). Where workers go, do jobs follow? Washington, DC: Brookings Institution. Ihlanfeldt, K. R., & Sjoquist, D. L. (1998). The spatial mismatch hypothesis: A review of recent studies and their implicati ons for welfare reform. Housing Policy Debate, 9 (4), 849 892. Innes, J., & Simpson, D. (1993). Implementing GIS for planning. Journal of the American Planning Association 59(2), 230 236 Jencks, C., & Mayer, S. E. (1990). The social consequences of growi ng up in a poor neighborhood. In L. E. Lynn & McGeary, M. G. H. (Eds.), Inner city poverty in the United States (pp. 111 186) Washington, DC: National Academy Press. Joint Center for Housing Studies of Harvard University. (2009). State of the nations housing 2009 Cambridge, MA: Author. Retrieved May 13, 2010, from http://www.jchs.harvard.edu/ publications/markets/son2009/son2009.pdf

PAGE 164

164 Joint Center for Housing Studies of Harvard University (2008). Americas rental housing: The key to a balanced national policy Cambridge, MA: Author. Retrieved May 13, 2010, from http://www.jchs.harvard.edu/publications/rental/rh08_americas_rental_housing/rh0 8_americas_rental_housing.pdf Joint Center for Housing Studies of Harvard University. (2003). State of the nations housing 2003 Cambridge, MA: Author. Retrieved May 13, 2010, from http://www.jchs.harvard.edu/publications/markets/son2003.pdf Kain, J. F. (1992). The spatial mismatch hypothesis: Three decades later. Housing Policy Debate 3 (2), 371 392. Kain, J. F. (1968). Housing segregation, negro employment, and metropolitan decentralization. The Quarterly Journal of Economics 82(1), 175 197. Kasarda, J. D. (1989). Urban industrial transition and the under class. Annals of the American Academy of Political and Social Sciences 501 26 47. Kasarda, J. D., & Ting, K. (1996). Joblessness and poverty in Americas central cities: Causes and policy prescriptions. Housing Policy Debate, 7 (2), 387 419. Kennedy, M., & Leonard, P. (2001). Dealing with neighborhood change: A primer on gentrification and policy choices Montclair, NJ: Brookings Institution Center on Urban and Metropolitan Policy. Khadduri, J., & Wilkins, C. (2007). Designing subsidized rental housing programs: What have we learned? Paper presented at Revisiting Rental housing: A national policy summit. Klosterman, R. E. (1997). Planning support systems: A new perspective on computer aided planning. Journal of Planning Education and Research, 17 45 54. L ang, R., & Hornburg, S. P. (1998). What is social capital and why is it important to public policy? Housing Policy Debate, 9 (1), 1 16. Larsen, K. (2009). Reassessing state housing trust funds: Results from a Florida survey. Housing Studies 24 (2), 173 201. Lees, L., Slater, T., & Wyly, E. K. (2007). Gentrification New York: Routledge/Taylor and Francis Group. Leventhal, T., & Brooks Gunn, J. (2003). Moving to opportunity: An experimental study of neighborhood effects on mental health. American Journal of P ublic Health 93(9), 1576 1582. Levine, J., & Landis, J. D. (1989). Geographic information systems for planning. Journal of the American Planning Association, 55 (2), 209 220.

PAGE 165

165 Levy, D. K., Comey, J., & Padilla, S. (2006). In the face of gentrification: Case studies of local efforts to mitigate displacement Washington, DC: The Urban Institute. Ley, D. (1996). The new middle class and the remaking of the central city Oxford: Oxford University Press Lipman, B. J. (2006). A heavy load: The combined housing and transportation burdens of working families Washington, DC: Center for Housing Policy Listokin, D. (1991). Federal housing policy and preservation: Historical evolution, patterns, and implications. Housing Policy Debate, 2 (2), 157 185. MacDonald, J. M., & Nelson, P. E., Jr., Do the poor still pay more? Food price variations in large metropolitan areas. Journal of Urban Economics 30, 344 359. Maguire, D. J., (1991). An overview and definition of GIS. In D. J. Maguire, M. F. Goodchild, & D. W. Rhind (Eds.), Geographical Information Systems Principles and Applications (pp. 9 20) New York: John Wiley and Sons, Inc. Massey, D. S., & Kanaiaupuni, S. M. (1993). Public housing and the concentration of poverty. Social Science Quarterly 74 (1), 109 122. M alczewski, J. (2004). GIS based landuse suitability analysis: a critical overview. Progress in Planning, 62, 3 65. McHarg, I. L. (1992). Design with nature. New York: J. Wiley McKinnish, T., Walsh, R., & White, K. (2009). Who gentrifies low income neighborhoods? Journal of Urban Economics 180 193. Melndez, E., Schwartz, A., & de Montrichard, A. (2008). Year 15 and Preservation of Tax Credit Housing for Low Income Households: An Assessment of Risk. Housing Studies 23 (1), 67 87. Miller, W., Collins, M. G., Steiner, F. R., & Cook, E. (1998). An approach to greenway suitability analysis. Landscape and Urban Planning 42(2 4), 85 105. Mills, D. E. (1981). Growth, speculation and sprawl in a monocentric city. Journal of Urban Economics 10, 201 226. Mollenkopf, J. H., Swanstrom, T., & Dreier, P. (2004). Place matters: Metropolitics for the twenty first century Lawrence, KA: University Press of Kansas. Moore, L. V., & Roux, A. V. D. (2006). Associations of neighborhood characteristics with the location and ty pe of food store. American Journal of Public Health, 96 (2), 325 331.

PAGE 166

166 Mueller, E. J., & Schwartz, A. (2008). Reversing the tide: Will state and local governments house the poor as direct federal subsidies decline? Journal of the American Planning Association 74(1), 122 135. National Association of Home Builders. (2004). Where is workforce housing located?: A study of the geography of housing affordability Retrieved May 13, 2010, from http://www.nahb.com/generic.aspx?sectionID=784&genericContentID=33617 National Housing Trust. (2008). State and local housing preservation initiatives Washington, DC: National Housing Trust. Retrieved May 19, 2010, from http://www.nhtinc.org/downloads/il_preservation.pdf National Housing Trust. (2004). State housing finance agencies: Developing programs to preserve affordable housing. Retrieved May 5, 2010, from http://content.knowledgeplex.org/kp2/cache/documents/66651.pdf National Housing Trust, Reconnecting America. (2008). Preserving opportunities: Saving affordable homes near transit Retrieved May 5, 2010, from http://www.reconnectingamerica.org/public/display_asset/nhtraaff Nenno, M. K. (1991) State and local governments: New initiatives in low income housing preserv ation. Housing Policy Debate 2 (2), 467 497. Newman, K., & Wyly, E. (2006). The right to stay put, revisited: Gentrification and resistance to displacement in New York City. Urban Studies 43 (1), 35. Newman, S. J., & Schnare, A. B. (1997). And a suitable living environment: The failure of housing programs to deliver on neighborhood quality. Housing Policy Debate, 8 (4), 703 741. Oakley, D. (2008). Locational patterns of low income housing tax credit developments: A sociospatial analysis of four metropolitan areas. Urban Affairs Review 43(5), 599 628. Olsen, E. O. (2001). Housing programs for low income households (Working Paper 8208). Cambridge, MA: National Bureau of Economic Research. Retrieved May 17, 2010, from http://www.nber.org/papers/w8208.pdf Orlebeke, C. J. (2000). The evolution of low income housing policy, 1949 to 1999. Housing Policy Debate 11(2), 489 520. Paston, M., Jr., Sadd, J., Hipp, J. (2001). Which came first: Toxic facilities, minority move in, and environmental justice. Journal of Urban Affairs 23 (1), 1 21. Pedone, C. I. (1991). Estimating mortgage prepayments and defaults in older federally assisted rental housing and the possible costs of preventing them. Housing Policy Debate, 2 (2), 245 273.

PAGE 167

167 Peiser, R. B. (1999). The fallout from federal low income housing preservation programs: A case study in estimating damages. Housing Policy Debate 10 (2), 371 393. Pereira, J. M. C., & Duckstein, L. (1993). A multiple criteria decisionmaking approach to GIS based land suitability evaluation. International Journal of Geographic Information Science, 7 (5), 407 424. Pettit, K. L. S., & Droesch, A. E. (2008). A guide to home mortgage disclosure act data. Washington, DC: The Urban Institute. Pinellas County, Community Development Planning Division (2009). Description of projects: 20092010 action plan. Retrieved June 22, 2010, from http://www.pinellascounty.org/community/pdf/actionplan.pdf Pinellas County, Community Development (n.d.) Welcome to the Community Development Department Retrieved June 22, 2010, from http://www.pinellascounty.org/community/default.htm Preston, J., & Raj, F. (2007). Accessibility, mobility and transport related social exclusion. Journal of Transport Geography 15 151 160. Proscio, T. (2005). Preserving Americas affordable housing: Retooling a 20th Century asset for 21st Century needs Washington, DC: Local Initi atives Support Coalition. Raphael, S. (1998). The spatial mismatch hypothesis and black youth joblessness: Evidence from the San Francisco Bay area. Journal of Urban Economics 43, 79 111. Ray, A., Nhuyen, D., ODell, W., Roset Zuppa, P., & White, D. (2009 ). The state of Floridas assisted rental housing. Gainesville, FL: Florida Housing Data Clearinghouse, Shimberg Center for Housing Studies. Rohe, W. M., & Freeman, L. (2001). Assisted housing and residential segregation: The role of race and ethnicity in the siting of affordable housing. Journal of the American Planning Association 67(3), 279 292. Rosenbaum, J., DeLuca, S., & Tuck, T. New capabilities in new places: Low income black families in suburbia. In X. D. Briggs (Ed.), The Geography of Opportunit y: Race and Housing in Metropolitan America (pp. 150 175). Washington, DC: Brookings Institution Press. Roset Zuppa, P. (2008). Curbing the loss of affordable rental housing in Florida: A risk assessment approach. Cornell Real Estate Review 6 (1), 54 69. R usk, D. (2003). Cities without suburbs: A Census 2000 update. Woodrow Wilson Center special studies. Washington, D.C: Woodrow Wilson Center Press.

PAGE 168

168 Saaty, T. L. (2008). Decision making with the analytic hierarchy process. International Journal of Services Sciences 1 (1), 83 98. Sampson, R. J., & Lauritsen, J. L. (1994). Violent victimization and offending: Individual situational and community level risk factors. In A.J. Reiss & J.A. Roth (Eds.), Understanding and preventing violence: Volume 3 (pp. 1 115 ) Washington, D.C: National Academy Press. Scally, C. P. (2009). State housing finance agencies forty years later: Major or minor players in affordable housing? Journal of Planning Education and Research, 29(2), 1 94 212. Scheutz, J., Been, V., & Ellen, I. E. (2008). Neighborhood effects of concentrated mortgage foreclosures. Journal of Housing Economics 17(4), 306 319. Schwartz, A. F. (2006). Housing policy in the United States: An introduction. New York: Routledge. Selye, H. (1978). The stress of life N ew York: McGraw Hill. Shimberg Center for Affordable Housing. (2008). A risk assessment method for preservation of assisted rental housing Retrieved May 7, 2010, from http://flhousi ngdata.shimberg.ufl.edu/AHI_introduction.html Shimberg Center for Affordable Housing. (2007). 2007 rental market study Retrieved May 10, 2010, from http://flhousingdata.shi mberg.ufl.edu/shimberg_pub_presentation.html Sorrentino, J. A., Meenar, M. R., & Flamm, B. J. (2008). Suitable housing placement: A GIS based approach. Environmental Management 42, 803 820. Smith, D. A. (1999). Mark to market: A fundamental shift in affor dable housing policy. Housing Policy Debate 10(1), 143 182. Smith, J. P., Edmonston, B., National Research Council (U.S.)., National Research Council (U.S.)., & National Research Council (U.S.). (1997). The new Americans: Economic, demographic, and fisca l effects of immigration. Washington, D.C: National Academy Press. Smith, N. (1996). The new urban frontier: gentrification and the revanchist city New York: Routledge. Stoll, M. A. (2005). Job sprawl and the spatial mismatch between blacks and jobs Washington, DC: Brookings Institution. Store, R., Kangas, J. (2001). Integrating spatial multi criteria evaluation and expert knowledge for GIS based habitat suitability modeling. Landscape and Urban Planning, 55, 79 93.

PAGE 169

169 Svatos, V. C., & Doucette, J. S. ( 2003). Analyzing Suitability of Land for Affordable Housing. Paper presented at the ESRO Intern ational User Conference, July 7 11, 2003, San Diego, CA. Retrieved June 4, 2010, from http://proceedings.esri.com/library/userconf/proc03/p0339.pdf Talen, E. (1997). The social equity of urban service distribution: An exploration of park access in Pueblo, Colorado and Macon, Georgia. Urban Geography 18(6), 521 544. Taylor, S. E., Repetti, R. L., & Seeman, T. (1997). Heath psychology: What is an unhealthy environment and how does it get under the skin. Annual Review of Psychology, 48 411 447. Thomson, C. N., & Hardin, P. (2000). Remote sensing/GIS integration to identify potential low income housing sites. Cities 17 (2), 97 109. Trust for Public Land. (2005). Greenprint for King County Retrieved June 4, 2010, from http://www.tpl.org/tier3_cd.cfm?content_item_id=18178&folder_id=262 Turner, M. A. (1998). Moving out of poverty: Expanding mobility and choice through tenant based housing assistance. Housing Policy Debate, 9 (2), 373 394. Turner, M. A., & Snow, C. (2001). Leading indicators of gentrification in D.C. Neighborhoods Washington, DC: The Urban Institute. University of Florida. Bureau of Economic and Business Research. (2007). Florida statistical abstract Gainesville, FL: University of Florida Press. U.S. Cens us Bureau. (2009, September 29). Poverty definitions Retrieved June 14, 2010, from http://www.census.gov/hhes/www/poverty/definitions.html U.S. Department of Housing and Urban Development. (2007a). Affordable housing needs 2005: A report to Congress Washington, DC: Author. U.S. Department of Housing and Urban Development (2007b). Fair Market Rents for the Section 8 Housing Assistance Payments Program Retrieved June 21, 2010, from http ://www.huduser.org/portal/datasets/fmr/fmrover_071707R2.doc U.S. Department of Housing and Urban Development. (2003). Trends in worst case needs for housing, 1978 1999. Washington, DC: Author. U.S. Government Accountability Office. (2004). Multifamily hous ing: More accessible HUD data could help efforts to preserve housing for low income tenants Washington, DC: Author. Report GAO 04 02. Retrieved May 17, 2010, from http://www.gao.gov/new.items/d0420.pdf Wallace, J. E. (1995). Financing affordable housing i n the United States. Housing Policy Debate, 6 (4), 785 814.

PAGE 170

170 Williamson, A. R., Smith, M. T., & Strambi Kramer, M. (2009). Housing choice vouchers, the Low Income Housing Tax Credit, and the federal poverty deconcentration goal. Urban Affairs Review 45(1), 119 132. Wilson, W. J. (1987). The truly disadvantaged: The inner city, the underclass, and public policy Chicago: University of Chicago Press. Wrigley, N., Warm, D., & Margetts B. (2003). Deprivation, diet, and foodretail access: findings from the Leeds food deserts study. Environment and Planning A 35 151 188. Wyly, E. K., & Hammel, D. J. (1999). Islands of decay in seas of renewal: housing policy and the resurgence of gentrification. Housing Policy Debate, 10(4), 711 771. Zenck, S. N., Schulz, A. J., Israel, B. A., James, S. A., Boa, S., & Wilson, M. L. (2005). Neighborhood racial composition, neighborhood poverty, and the spatial accessibility of supermarkets in metropol itan Detroit. American Journal of Public Health 95(4), 660 667.

PAGE 171

171 BIOGRAPHICAL SKETCH Eric Seymour was born in Buffalo, New York in 1979. He earned a Bachelor of Arts from the Harriet L. Wilkes Honors College of Florida Atlantic University, with a major c oncentration in English l iterature and a minor concentration in h istory. Before entering the graduate program in u rban and r egional p lanning at the University of Florida, he pursued graduate studies in e arly American l iterature and worked as a teaching ass istant at the University of Florida. While pursuing his masters degree in u rban and r egional p lanning at the University of Florida, he served as president of the Student Planning Association. He received the American Institute of Certified Planners Outsta nding Student Award in 2010. This fall, Eric is entering the doctoral program in u rban and r egional p lanning at the University of Michigan, where he plans to study the shrinking cities phenomenon in Detroit and other cities in the Great Lakes region. Eric hopes to contribute to the sensitive redevelopment of shrinking cities and assist in the preservation of Americas industrial landscapes and working class neighborhoods.