x :1... x? .. A

mumafié ‘ i .... e.

.xuamv $.33...
“‘1 1“" fl

.
J

E

g ..
11m. :3

.7.
.E :c.

blitzzzxi.
Ji'-*E‘.~‘I~ it I
\O .5 Soil}!
. ‘ tug"-
! :3...

.. . .
I . Iwww
, II It.
.11.. it : “53¢

15... 14...:
ii!!! I
‘o .: yslwhuurihzx .

.l 1;.
weiél
II .. 151:.
iatiinrw. 1.3? ..
.

7: .
fiftiill vtfih v.0?
. Run... .
333;». 3.5? a
3 . s. : W h.
{adduzavi 33.4%..
.. y . 2.... : is: ..
1....1,§..2..v..2!., a,
33.1.6... .32.: .
3:135; ......z. - .i.
.2331. ..:..:-.:.l

 

 

It 40. 9.7.. o I
,vatttcns...‘

 

 

 

 

 

 

.

c... 11...!) 343! g

. V sfialspr! :0.
$7531.11- .

 

 

act 11.01:..03 Err...

 

 

 

“a?” LIBRARY #-

' - .MIICI'II‘SJI‘: State
2039 Uni*=/ers'.ty #4

”r

This is to certify that the
dissertation entitled

MEASURING INDIRECT BENEFITS OF BROWNFIELD

REDEVELOPMENT USING THE HEDONIC PRICE METHOD:

THE CASES OF LANSING, MICHIGAN

presented by

YOUNG-TAE KIM

has been accepted towards fulfillment
of the requirements for the

PHD. degree in Resource Development-Urban
' Studies

 

 

 

Major Professor’s Signature

Date

MSU is an Affirmative Action/Equal Opportunity Employer

/
..--—--—.--.--.-.-.---.-.-.-.—.—.—.--—

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K lProj/Acca-Pres/CIRC/Dateoue Indd

 

 

 

MEASURING INDIRECT BENEFITS OF BROWNFIELD REDEVELOPMENT
USING THE HEDONIC PRICE METHOD:
THE CASES OF LANSING, MICHIGAN
By

Young-Tae Kim

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Resource Development-Urban Studies

2009

ABSTRACT
MEASURING INDIRECT BENEFITS OF BROWNFIELD REDEVELOPMENT
USING THE HEDONIC PRICE METHOD:
THE CASES OF LANSING, MICHIGAN
By
Young-Tae Kim

The main purpose of this study is to measure the indirect economic benefits of
brownfield redevelopment by using the hedonic price method. The study analyzed
impacts of brownfield redevelopment on the price of surrounding properties, using three
brownfield redevelopment cases in Lansing, Michigan. The empirical findings of the
hedonic price analysis confirmed the existing literature observing that the proximity to
brownfield sites or environmental hazards was negatively related to the property values in
nearby neighborhoods. The empirical findings of the analysis also supported the
hypothesized relationship between the proximity to brownfield sites and nearby housing
values before and after redevelopment. Redevelopment of brownfields was expected to
provide positive impacts on the housing values in the surrounding neighborhood by
eliminating or substantially reducing the negative extemality associated with brownfields.

The analysis employed two different hedonic functional models to estimate the
indirect benefits of three brownfield redevelopment projects: a linear functional model
and a log-linear functional model. The results of the two analyses were consistent with
each other, showing that the estimated total benefit of three redevelopment projects was
$78,007,696 for the log-linear functional model and $85,962,590 for the linear functional
model, in addition to the direct benefits of the projects such as the number of jobs created

and increases in the tax base after redevelopment.

As this study confirms, the cleanup or redevelopment of brownfields provides
indirect benefits that cannot be measured easily in the current market system. However,
this kind of indirect benefit of brownfield redevelopment has not been an important
consideration in the public decision-making framework. Without acknowledgement of the
full potential benefits and costs associated with brownfield redevelopment, it is not
possible to make an efficient allocation of resources in the public policy area. The
empirical findings of this study can be used to justify public financial assistance for the
cleanup or redevelopment of brownfields. The findings can also help government
officials prioritize competing projects to use public resources more efficiently. This study
also suggests that the estimated hedonic function can be used to determine the benefits of
brownfield cleanup or redevelopment in the decision-making process.

However, some limitations were inherent in this study due to the small number of
redeveloped brownfield sites. First, its empirical findings cannot be generalized as the
indirect economic benefits of brownfield redevelopment and can only represent the
impacts of three brownfield redevelopment case studies in Lansing, MI. Second, even
though the study compared the benefits of three brownfield redevelopment cases based
on their different redevelopment scenarios, the comparison should not be generalized due
to the limited number of brownfield redevelopment cases in the study. The limitations of
the study suggest that future researchers might consider analyzing the indirect economic
impacts of brownfield redevelopment using a large number of redeveloped brownfields in
various geographical locations, and contrasting hedonic price methods to other more
descriptive approaches to measuring the indirect economic benefit of brownfield

redevelopment.

Copyn'ght by
Young Tae Kim
2009

DEDICATION

To my wife Jung Hee Choi, daughter, Bo Ra, and son Hyong Uk Kim

ACKNOWLEDGEMENT

I would like to express my profound gratitude to each member of my dissertation
committee who has guided me to the completion of my Ph. D. program with great
patience. I would like to thank Dr. Rene Rosenbaum, my advisor and committee chair, for
his warm guidance and continuous support throughout my Ph.D. program. Dr. Gerhardus
Schultink provided me with invaluable advice and feedback on my dissertation research,
Dr. George Rowan showed warm friendship and care throughout my program, and Dr.
Rex LaMore continuously challenged me to become a true scholar and practitioner.

I would like to extend my sincere gratitude to Mr. Douglas Stover, Director of the
Ingham County Equalization Department and Mr. Rob Francis, City Official in the
Lansing Assessor’s Office, who provided the property transaction data. I also thank Mr.
Sam Quan, GIS Administrator, the City of Lansing, who allowed me to use the GIS maps
for my research. Special thanks go to Mr. Michael Miller, who hired me as an
international program coordinator, for his continuous challenges for the completion of my
study. I am also grateful to Ms. Sandy Holy, who proofread my dissertation.

Finally, I am especially grateful to my family for their ongoing support and
encouragement for my study. My wife, Junghee Choi, lost her father and a younger
brother during my study but was not even able to attend their funerals due to various
reasons resulting from my study. My daughter, Bo Ra, and my son, Hyong Uk, have

brought joy and love to my life. I could not complete my study without them.

vi

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES ........................................................................................................... ix
1. INTRODUCTION ...................................................................................................... 1
1.1 Background .......................................................................................................... 1
1.2 The Purposes and Limitations of the Study ....................................................... 10

II. LITERATURE REVIEW: THE ECONOMIC IMPACTS OF BROWNFIELD
REDEVELOPMENT & THE HEDONIC PRICE METHOD .................................. 15
2.1 The Economic Impacts of Brownfield Redevelopment ..................................... 15
2.2 The Hedonic Price Method & Brownfield Redevelopment ............................... 18
III. STUDY AREA, DATA AND EMPIRICAL MODEL .............................................. 30
3.1 Study Area .......................................................................................................... 30
3.2 Data .................................................................................................................... 37
3.3 Empirical Model ................................................................................................. 41
IV. EMPERICAL RESULTS .......................................................................................... 51
4.1 Descriptive Summary of Variables ..................................................................... 51
4.2 Results of the Hedonic Price Analysis ............................................................... 54
4.3 Benefit Estimates of Redevelopment ................................................................. 70
4.4 Comparison with a Semi—Log Functional Analysis ........................................... 74
V. CONCLUSION ......................................................................................................... 77
5.1 Summary of Findings ......................................................................................... 77
5.2 Policy Implications of the Study ........................................................................ 80
5.3 Limitations and Recommendations .................................................................... 81
APPENDICES .................................................................................................................. 83
APPENDIX A Regression Analysis for PRUDDEN ................................................. 83
APPENDIX B Regression Analysis for NEOGEN ................................................... 87
APPENDIX C Regression Analysis for BUILDERS ................................................ 91
APPENDIX D Regression Analysis for ALL SITES ................................................ 95
APPENDIX E Log-Linear Functional for ALL SITES ............................................. 99
BIBLIOGRAPHY ........................................................................................................... 101

vii

LIST OF TABLES

Table I-l Michigan’s Programs on Brownfield Redevelopment ........................................ 5

Table 111-1 Comparison of Selected Socio—economic Characteristics in 1999 and 2007 . 30

Table 111-2 Description of Variables for the Hedonic Analysis ......................................... 46
Table IV-l Summary of Variables for All Sites (N of observations: 8458) ...................... 51
Table IV-2 Summary of Variables for Individual Sites ..................................................... 53
Table IV-3 Estimated Coefficients (Prudden) ................................................................... 55
Table IV-4 Estimated Coefficients (N EOGEN) ................................................................ 60
Table IV-5 Estimated Coefficients (BUILDERS) ............................................................. 64
Table IV-6 Coefficient Estimates (All 3 Brownfield Sites) .............................................. 67
Table IV—7 The Coefficient Estimates of Difference”HR .................................................. 72
Table IV-8 Benefit Estimates of Redevelopment of Brownfield Sites ............................. 73

Table IV—9 Results of the Linear Functional & the Log-Linear Functional Analyses ...... 75

viii

LIST OF FIGURES

Figure III-1 Location of Brownfields ............................................................................... 33
Figure III-2 Motor Wheel Lofis (left) and Prudden Place Apartments (right) .................. 34
Figure III-3 Neo gen Corporation ...................................................................................... 35
Figure III-4 Builders Plumbing & Heating Supply ........................................................... 36
Figure 111-5 The Use of a GIS Map to Construct the Database for the Study .................. 39

Figure 111-6 The Calculation of Distance Variables Using “Near” Tool of the ArcINFO. 40

Figure III-7 Scatterplot Matrix for Selected Variables ..................................................... 43

ix

I. INTRODUCTION

1 .1 Background

A brownfield, defined as “an abandoned, idle, or under used industrial or
commercial property where expansion or redevelopment is complicated by real or
perceived environmental contamination (EPA, 1996),” is a symbol of deterioration in
many urban communities in the United States. Contamination in these brownfields may
be found in various forms and degrees ranging from extremely hazardous contamination
with heavy metals or chemicals to zero or suspected contamination with debris or
eyesores of old buildings. These brownfields limit economic development opportunities
and restricts urban revitalization while raising a variety of environmental and public
health concerns. The Environmental Protection Agency (EPA) estimates that there are
more than 450,000 brownfields nationwide.

The reuse or redevelopment of brownfields has not been an attractive option for
property owners or prospective deve10pers due to unanticipated effects of federal or state
environmental legislation that regulates the management of hazardous wastes and
contaminated sites. The important legislation that has major impact on the reuse of the
contaminated sites includes the Comprehensive Environmental Response, Compensation,
and Liability Act of 1980 (CERCLA), and state statues that regulate the management and
cleanup of contaminated properties. The federal and state laws pose several barriers to the
reuse or redevelopment of the contaminated sites, such as the uncertain cleanup standards,
unknown or very high cleanup costs, potential liability risk to prospective purchasers or

developers, and uncertain timeline of redevelopment.

CERCLA, which is commonly known as “Superfund”, was enacted in 1980 as a
congressional response to the “Love Canal” tragedy1 and the growing concerns over
closed or abandoned hazardous waste sites. CERCLA contains two main purposes. First,
CERCLA establishes a tool to finance the govemment’s cleanup of hazardous waste sites
when no responsible party can be identified. Second, it provides a mechanism to recover
the cleanup cost from all identifiable responsible parties. Since its enactment, CERCLA
has had success in reducing human health risks and improving the environmental quality
across the country. According to EPA, more than 900 sites were cleaned up and cleanup
activities are underway at additional 422 sites in 2005 (EPA, 2005).

However, CERCLA’s imposition of liability on all potential responsible parties for
cleanup made it difficult to reuse the contaminated sites. The enactment of CERCLA
dramatically changed the way of imposing legal liability for contamination or
environmental injuries. First, CERCLA’S liability is “retroactive and strict” since property
owners can be held liable for contamination on their property even though their activities
were legal at the time of contamination and were undertaken with all due care. CERCLA
also imposes liability for cleanup on new property owners who purchased contaminated
sites but had nothing to do with contamination. In addition, CERCLA’s liability is “joint
and several” because an individual or single party who is partly responsible for
contamination can be liable for the full cleanup cost if there are no other responsible
parties identified.

As a result, property owners, new investors, and developers have stepped away

from the reuse or redevelopment of the contaminated sites in fear of potential liability for

 

1 Love Canal is one of the biggest modern environmental disasters caused by hazardous chemical wastes.
For detail information, see Beck, Eckardt, “The Love Canal Tragedy” in EPA Journal, January 1979.

2

contamination and the cleanup costs. Under CERCLA, all potential responsible parties
include: the current owner or operator of the site, the past owner or operator, a person
who arranged for the disposal of waste at a site, and a person who transported waste to a
site.2 CERCLA’s strict and joint and several liability on these potential responsible parties
made it difficult to transfer or redevelop the contaminated sites.

Many states also began to introduce their own laws similar to CERCLA because a
lot of contaminated sites within their jurisdictions did not meet the criteria for being listed
on the National Priority List (NPL)3 of CERCLA. Even though these contaminated sites
were not regulated under CERCLA, many of them were still considered as
environmentally hazardous sites that imposed human health risk. The basic purpose of
the state laws was to regulate these contaminated sites by imposing liability of cleanup on
property owners or other responsible parties. Approximately forty five states4 introduced
their own state superfund laws that are very similar with CERCLA. In general, they
contain the following features: procedures for emergency response and permanent
cleanup, provision of financing sources for cleanup, enforcement authority to regulate the
contaminated sites, and provision for public participation in the cleanup process (Geltman,
2000269)

The so-called state superfund laws along with CERCLA made it more difficult to
reuse the contaminated properties by imposing stringent cleanup standards and liability

for property owners and prospective developers. Acknowledging the barriers to the reuse

 

2 See CERCLA Section 107 (2) for details.
3 The NPL is a list of contaminated sites, published by USEPA, which pose an immediate or significant
pubic health threat. NLP sites are eligible for extensive, long-term cleanup action under CERCLA.

“Forty one states have strict liability, and thirty six have several and joint liability to allocate
responsibility for costs among responsible parties. Forty-three states impose retroactive liability. Thirty-two
states have authority to recover for damages to natural resources.” (Hula, 2003: 4)

3

of the contaminated properties, several states took a lead in encouraging the reuse or
redevelopment of these sites by establishing their own voluntary cleanup program (VCP).
Approximately twenty one states enacted state voluntary cleanup programs and fifteen
states introduced brownfield economic redevelopment programs to facilitate the reuse or
redevelopment of the contaminated sites (Geltman, 2000: 81). The state programs focus
on relieving the cleanup liability, introducing flexible cleanup standards, and providing
financial incentives for site assessment and cleanup activities.

Michigan is one of the leading states in promoting brownfield redevelopment.
Michigan along with some other states in the Midwest region pioneered innovative and
effective approaches to encourage and facilitate the redevelopment of properties that were
environmentally degraded and contaminated with heavy metals, chemicals, and
petroleum constituents due to previous manufacturing and other old industrial activitiess.
Michigan established its own voluntary cleanup program and provided several grant and
loan programs that are available to local units of government along with other public
incentives to attract private investment on brownfields. Table I-1 shows Michigan’s

programs for brownfield cleanup and redevelopment.

 

5 See “Chapter 8. Region V: Illinois, Indiana, Michigan, Minnesota, Ohio, and Wisconsin” in Recycling
Land: Understanding the Legal Landscape 0 f Brownfield Development for the brownfield redevelopment
efforts of the Midwest states (Geltman, 2000: 199-233).

4

Table I-l Michigan’s Programs on Brownfield Redevelopment

 

 

Financing Programs Voluntary Cleanup Other Incentives
Programs
0 Site Reclamation/Site Assessment 0 Part 201 of Natural - 10% Single Business
Grant Program Resources Tax Credits ($1 million
0 Cleanup and Redevelopment Fund Environmental cap) for innocent
o Revitalization Loan Fund Protection Act (1994, party’s development
. Brownfield Redevelopment amended in 1995) costs on a property.
Authorities (Tax Increment
Financing)
0 Clean Michigan Initiatives

 

 

 

 

 

Source: Modified from Northeast-Midwest Institute. 1999.

Michigan first began to regulate the cleanup process of brownfield sites by
enacting a state superfund law, the Michigan Environmental Response Act (MERA) of
1982, known as Public Act 307. Like most state superfund programs, MERA established
a process for assessing risks and providing for response activity at environmentally
contaminated sites. MERA imposed liability for cleanup on owners or operators of the
contaminated sites or facilities and provided a means for public financing of remedial
actions at the contaminated sites. MERA also set standards for cleanup of the
contaminated sites. The Michigan superfund law was amended to introduce strict joint
and several liability for potentially responsible parties in 1992.

As CERCLA and most state superfund programs were criticized in the mid 19905
due to their unanticipated effects on the reuse of the contaminated sites, Michigan began
to find more innovative approaches to govern the contaminated sites. MERA was
amended in 1995 and became Part 201 of the Natural Resources and Environmental
Protection Act known as Public Act 451. The amendment aimed at promoting the reuse of
brownfields by relieving liability of property owners and operators, providing flexible

cleanup standards, and reducing uncertainty and remediation costs.

First, Part 201 of PA 451 changed liability scheme significantly to facilitate
transactions of the potentially contaminated properties and to promote brownfield
redevelopment. Under Part 201, new property owners or operators of the contaminated
properties are no longer liable for existing contamination as long as they did not cause the
contamination. New owners who acquired the potentially contaminated properties after
1995 will not be liable for the existing contamination if they complete a Baseline
Environmental Assessment (BEA). A BEA is an assessment of the existing environmental
conditions of a property at the time of property transaction. The BEA must be conducted
no later than 45 days after purchase, occupancy, or foreclosure of the property. The
properly conducted BEA will provide new property owners with liability protection for
existing contamination. In exchange for liability protection, new owners are obligated to
exercise “due care” with regard the existing contamination.

Part 201 also introduced flexible cleanup standards, by taking into account future
use of the contaminated properties. The Michigan Department of Environmental Quality
(MDEQ) is authorized to establish cleanup standards based on land use-based categories.
MDEQ set three main categories of land use-based cleanup standards: residential,
commercial, and industrial.6 There are also three additional categories of cleanup
standards in case there is a need to restrict property use beyond the current land use
regulations. The three additional categories are limited residential, limited commercial,
and limited industrial. Under this land use-based cleanup standard provision, property
owners can reduce the cleanup costs significantly by choosing an appropriate category of

cleanup standard based on the intended future use of the properties.

 

6 Michigan also allows cleanups based on recreational land use but the standard has not been developed.

6

In addition to the state VCP, Michigan also deve10ped various public financing
programs to attract private investment on brownfields. Michigan is one of the pioneering
states that developed grant programs to facilitate brownfield redevelopment. In 1988,
Michigan introduced two grant programs: the Site Reclamation Grant (SRG) and the Site
Assessment Grant (SAG). Michigan legislature appropriated a total of $45 million in the
Environmental Protection Bond Fund of 1988 for the two brownfield grant programs.
The SRG program provides funding up to $2 million to local units of govemment to
investigate and remediate known sites of environmental contamination, which can be
used for identified economic redevelopment projects. The SAG program provides grants
up to $1 million to eligible local units of government to assess the nature and extent of
contamination at properties with economic development potential. Michigan voters also
approved a $675 million bond, the Clean Michigan Initiative (CMI), in 1998. As part of
the CMI, a total of $335 million was earmarked for various brownfield programs
including the Environmental Cleanup and Redevelopment Program, the Brownfield
Redevelopment Grant and Loan Programs, and the Landfill Grant Program.

Another important financing source for brownfield cleanup or redevelopment is
local brownfield authorities (BRAs), which are created by municipal and county
governments under the 1996 Brownfield Redevelopment Financing Act. The Brownfield
Redevelopment Financing Act creates a state Brownfield Redevelopment Board within
MDEQ to oversee brownfield redevelopment and authorizes local governments to create
their own BRAs, which are allowed to adopt brownfield redevelopment plans, and to
capture tax increment revenues from functionally obsolete or blighted properties in order

to pay for eligible environmental costs necessary for safe redevelopment of the properties.

Michigan’s brownfield policies and programs have assisted in transforming a
number of abandoned, idle, or underutilized properties into productive use. According to
MDEQ (2008), Michigan has been provided about $927 million of state funding for
cleanup or redevelopment activities of nearly 1,800 sites. Michigan’s brownfield grant
and loan program alone awarded about $143 million for 300 projects statewide. MDEQ
(2008) estimates that the brownfield grants and loans have leveraged about $3 billion in
private investment while creating more than 20,000 new jobs.

Hula (2003) shows the success of Michigan’s brownfield programs after
analyzing a state database of Baseline Environmental Assessments (BEAS). The BEA
program is expected to facilitate transactions of the potentially contaminated properties
by eliminating new owners’ liability for existing contamination. Hula explored whether
Michigan’s BEA program attracts private investment for the reuse of those sites. His
study analyzed a sample of 1,505 BEA sites selected from the state BEA database. His
field visits to the sample sites identified that many BEA sites are fully redeveloped and
approximately 79% of the sample sites are engaged in at least minimal economic
activities. The study concluded that Michigan’s brownfield programs help attract private
investment on the potentially contaminated properties to some degree.

Moyhamed and Dancik (2007) also provide some evidence that shows the
effectiveness of the Michigan’s brownfield programs. They evaluated the investment
impacts of the Site Assessment Grant (SAG) program by selecting a sample of 30
projects funded by the grant program. The study aimed to make suggestions for the future
grant program based on analysis of the investment impact of the sample projects. The

study employed several economic indicators to evaluate the economic impacts of the

program. These indicators included the status and types of redevelopment, the number of
new jobs, and increased property tax revenues. The study found that the 30 projects
generated over $6.5 million in gross property tax revenues and created 124 new jobs as of
2005. The study also pointed out that the SAG program was helpful for transferring
property ownership from local governments to private parties. Based on their findings,
Moyhamed and Dancik concluded that the grant program is worth continuing due to its
positive impacts on brownfield redevelopment.

As these studies briefly indicate, Michigan’s brownfield programs play a positive
role in developing and reusing potentially contaminated properties. The reuse or
redevelopment of these sites provides numerous direct or indirect benefits to not only
individual property owners but also an entire community considering the pervasive
problems of brownfields throughout the entire community. These benefits include
retention/recovery of the local tax base, job creation, re-use of existing infrastructure, and
health and environmental benefits to the community.

Despite the direct and indirect benefits anticipated, it is not easy to assess all
possible impacts of individual brownfield redevelopment projects. As Moyhamed and
Dancik (2007) showed, the most common approach is to develop simple indicators,
which measure the possible direct outcome of brownfield redevelopment. Some examples
of indicators for measuring direct impacts of brownfield redevelopment include: income
from new jobs created during or after redevelopment; sales from new business
establishments on sites; new sales, income, and business taxes generated from

redeveloped sites; changes in property values; and increased local property taxes. The

direct impacts of brownfield redevelopment can be measured relatively easily using a set
of these indicators.

However, these simple indicators cannot fully reflect the environmental and
socioeconomic impacts of brownfield redevelopment projects. Beyond the direct benefits,
brownfield redevelopment can have indirect impacts on the nearby neighborhood or the
community by reducing health risks or improving the local environment or amenity due
to cleanup of contamination or removal of disamenities. In theory, these indirect impacts
are types of extemalities, which arise when the actions of individuals produce incidental

effects to others without intention.

1.2 The Purposes and Limitations of the Study

This study is an attempt to measure the indirect impacts of brownfield
redevelopment projects in Michigan using the hedonic price method. Since the indirect
impacts cannot be captured in the market system directly, it is necessary to use either
related goods or a survey based on a hypothetical scenario in order to quantify these
effects economically. The former method is the hedonic price method; the latter is the
contingent valuation method. This study uses the hedonic price method, which has been
wisely used to assess the value of environmental goods such as amenities or improvement
of environmental qualities.

The hedonic price method was introduced comprehensively for the first time in
the 19608 and 19705 by Griliches (1971) and Rosen (1974). After Griliches attempted to
measure the price changes of commodities as price index, Rosen formulated the

underlying theory of the hedonic price method, which determines the bid prices or

10

implicit value of the attributes, or characteristics of a commodity in competitive markets.
In principle, the hedonic price method assumes that certain commodities such as housing
or land are made up of different bundles of attributes that can have an impact on the
commodity values. Using this assumption, the hedonic price method seeks to determine
the implicit price of each attribute by analyzing the relationship between different
attributes and the price of the commodity.

After Rosen’s comprehensive formulation of the theory, many studies7 used the
hedonic price method to measure the values of environmental amenities or disamenities.
In general brownfields are considered as disamenities and the benefit of clean-up of
brownfields has been documented in the existing literature by analyzing the impacts of
brownfields on surrounding property values. It can be assumed that the proximity to
brownfields is negatively related to the nearby property values and property values
decrease as the proximity to a brownfield increases. Extending the assumption
furthermore, the nearby property values once declined, would rebound if brownfields are
cleaned up or redeveloped. Even though there are not many empirical studies that
assessed the impacts of brownfield remediation or redevelopment on nearby property

values, Grigelis (2005), Ihlanfeld and Taylor (2004), and Kaufinan and Cloutier (2006)

 

7 Some examples of the environmental valuations with the hedonic price method include: the effects of
hazardous waste disposal (Adler et a1. 1982, Smith and Desvousges, 1986), benefits or costs of noise
reduction (O’Bryne et al. 1985, Espey and Lopez, 2000, Becker and Lavee, 2003), the benefits of air
quality (Kim et al. 2003), the effects of water quality (Leggett and Bockstael, 2000), the value of scenic
beauty or amenity (Smith and Palmquist, 1994, Cheshire and Sheppard. 1995, Bishop, 1996, Tyrvainen
1997, Tyrvainen and Miettinen, 2000), the impact of open space or green space (McPherson, 1992, Riddel,
2001, Smith et al. 2002, Morancho, 2003, Tajima, 2003), the demand function of public school quality
(Brasington, 2002, Downes and label, 2002.), the impact of nuclear facilities (Clark et al. 1997, Gawande
and Jenkins-Smith, 2001), and the impact of hazardous waste sites or brownfields (Michaels and Smith,
1990, Kohlhase, 1991, Patchin, 1991, Kiel, 1995, McCluskey and Rausser, 1999 & 2003, Kiel and label,
2001, Deaton, 2002, Corona, 2004, Ihlanfeldt and Taylor, 2004, Grigelis, 2005, Longo and Alberini, 2005,
Kaufman and Cloutier, 2006)

11

concluded that the effects of brownfields are substantial and significant economic
benefits will be expected from the remediation or redevelopment of brownfields.

However, there is another argument that property values may not be firlly
recovered even though brownfields are cleaned up or redeveloped. Patchin (1991), Kiel
(1995), and McCluskey and Rausser (2003) provide some evidence that remediation of
contaminated commercial and industrial properties does not always lead to a full recovery
of the property’s own value due to “stigma” effects. Some studies show that there is no
evidence that the removal of negative extemalities is substantially capitalizing into the
real estate market (Corona, 2004, Longo and Alberini, 2005) and brownfield remediation
or redevelopment does not always lead to a rebound in property values in surrounding
communities.

This study reexamined whether redevelopment of brownfields provides positive
impacts on surrounding property values. The impacts of three brownfield redevelopment
cases on their surrounding residential property values were assessed by estimating
hedonic functions for 8,458 housing sales observations fi'om 1998 to 2007 in Lansing,
Michigan. The study first hypothesized that the brownfields had negative impacts on their
surrounding property values before redevelopment due to negative extemalities
associated with the brownfields. The analysis also tested to see if the redevelopment of
brownfields recovers the loss of surrounding property values by eliminating or
substantially reducing the negative extemalities. The study was able to provide empirical
evidence for indirect economic benefits of brownfield redevelopment, finding positive
impacts of brownfield redevelopment on the surrounding property values. In addition, the

study was able to provide a relative comparison on the benefits of three brownfield

12

redevelopment cases based on different redevelopment scenarios including different end
uses and sizes of three brownfield sites.

This study expected that its empirical findings would help state and local
governments have a better understanding of benefits of brownfield redevelopment. In
general, the indirect benefit has not been considered as an important factor in the decision
making process partly because it is not easy to quantify the indirect benefit. This stems
from perhaps due to the nonmarket nature of “commodity goods”. This study as well as
the existing literature proved that such indirect benefits can be measured by using the
hedonic price method. The ability to measure the indirect benefits from brownfield
redevelopment provides justification for including these benefits in the decision-making
framework of public policy.

However, this study contains the following limitations. First, since this study used
a small number of brownfield redevelopment cases in a single geographical location, its
empirical findings cannot be generalized. Second, even though the study compared the
benefits of three brownfield redevelopment cases based on their different redevelopment
scenarios, the comparison should not be generalized due to the limited number of
brownfield redevelopment cases of the study. The limitation of the study suggest that
future researchers might consider analyzing the indirect economic impacts of brownfield
redevelopment using a large number of redeveloped brownfields in various geographical
locations.

The remainder of the study is organized as follows. Chapter II introduces the
theoretical background of the hedonic price method as well as reviews in detail the

literature on the impact of brownfields on surrounding property values. Chapter III

13

describes the methodology of this study, including study areas, data, and hedonic price
models used for the empirical analysis. Chapter IV presents the empirical results of the
study and key findings. Chapter V provides a summary of key findings and what the

findings mean for research and policy.

14

II. LITERATURE REVIEW: THE ECONOMIC IMPACTS OF BROWNFIELD
REDEVELOPMENT & THE HEDONIC PRICE METHOD

2.1 The Economic Impacts of Brownfield Redevelopment

Brownfield redevelopment, which aims at revitalizing urban communities as well
as improving environmental qualities, has become an important policy for federal and
state governments. Due to the additional costs and financial risks associated with
brownfield redevelopment, public financing and other assistance are essential to
encourage the conversion of brownfields to productive properties. Federal and state
brownfield programs focus on reducing financial burdens and relieving other risks for
property owners or developers ranging from direct financial assistance to other incentives
including technical assistance.

According to the EPA (2005), more than 20 federal agencies maintain financing
and other service programs available for brownfield remediation and redevelopment. The
EPA alone has provided a total of $454.8 million to support assessment, clean-up, or
redevelopment of brownfields since the beginning of its brownfield program in 1995. The
EPA’s direct funding assistance has leveraged more than $8.2 billion in private
investment for 8,467 brownfield sites over the same time period.

State governments also offer a variety of financial and technical assistance to
promote remediation and redevelopment of brownfields (EPA, 2007). According to one
study (Bartsch and Wells, 2003), 23 states provide some form of tax incentives and 22
states offer financial assistance to promote brownfield remediation and redevelopment. In

addition to the efforts of federal and state governments, local governments complement

15

the federal and state programs by providing direct financial assistance8 and other tax
incentives.

Due to supports from all levels of government, communities throughout the US.
have been able to document their success stories of brownfield redevelopment since the
19905. One survey revealed that 154 cities had developed 1,409 sites for a total of 10,905
acres. In addition, a total of 10,256 acres at 1,189 sites were in various stages of
redevelopment (The US. Conference of Mayors, 2006). The remediation or
redevelopment of brownfields is already providing economic benefits to the communities.
According to the survey of the US. Conference of Mayors (2006), 62 cities reported
$233 million of actual tax revenues generated from redeveloped brownfield sites. Also,
71 cities reported that 83,171 new jobs have been created from redevelopment of
brownfield sites.

Despite widespread brownfield success stories throughout the US, it is not easy
to quantify the effects of brownfield clean-up or redevelopment entirely. One reason is
the lack of systematic data collection efforts. As Wemstedt (2004) addressed, “data
collection and analysis are inconsistent across federal and state agencies, across programs
in a single agency, and even within individual programs” (p.1). Another challenge is that
some effects of brownfield redevelopment cannot be observed in the market system
directly. As Wemstedt’s survey showed, most direct economic benefits from brownfield
redevelopment can be measured with such common economic indicators as the number of
new jobs, increased income, the increased value of the sites, and the increased tax

revenues from the redeveloped site. However, brownfield clean-up or redevelopment

 

8 . . . . .
The main sources of local financral assrstance for brownfield redevelopment Include tax Increment

financing (TIF), general obligation bonds, and local revolving loan funds. (See Bartsch and Wells, 2003 for
further information)

16

generates some other effects such as improvement of environmental amenities and
reduction of health risks that cannot be captured in the market system directly. Since
these effects have no market value, it is necessary to use either related goods or a survey
based on a hypothetical scenario in order to quantify these effects economically. The
former method is the hedonic price method; the latter is the contingent valuation method.
The contingent valuation method measures values of non-market environmental
goods by analyzing individuals’ responses to hypothetical questions regarding their
willingness to pay for environmental goods. The hypothetical questions in general take
the form of “How much would you pay if . . .?”. Despite its usefulness for measuring
value of environmental goods, the contingent valuation method has some issues and
problems due to its nature of formulating the questions based on a hypothetical scenario.
According to Goodstein (1999), there are at least four sources of possible errors identified
from the previous contingent evaluation studies: free riding, strategic bias, hypothetical
bias, and embedding bias. Free riding and strategic bias often occur when the actual
outcome of individuals’ responses can affect the provision of environmental quality or
their actual payment for the environmental goods in question. In this case, the
respondents may strategically understate or overstate their true willingness to pay.
Hypothetical bias arises when respondents provide poorly thought out or meaningless
answers to hypothetical questions. Embedding bias can be observed when individuals’
responses are strongly affected by the amount of information about the environmental
good in question. Despite these problems, as Goodstein (1999) pointed out, the
contingent valuation method has been increasingly used “because it provides the only

available means for estimating non-market benefits based primarily on existence value,

17

such as the benefits of preserving the striped shiner” (p. 138). Goodstein (1999) also
mentioned that much of the potential bias associated with the contingent valuation
method can be overcome if hypothetical surveys are carefully designed.

Unlike the hypothetical nature of the contingent valuation method, the hedonic
price method measures values of nonmarket environmental goods from observed market
behavior using “the change in prices of related, or complementary goods to infer a
willingness to pay for a healthier environment” (Goodstein: 140). As the word “hedonic”
indicates, the hedonic price method has been used to measure the pleasure or utility
associated with an improved environment. This study used the hedonic price method
since its main purpose is to measure the indirect value of an improved environment, the

redevelopment of brownfields.

2.2 The Hedonic Price Method & Brownfield Redevelopment

Even though the history of the hedonic price method goes back to the 19205, it
was Rosen (1974) who developed its theoretical framework comprehensively. According
to Rosen (1974), “goods are valued for their utility bearing attributes” and “hedonic
prices are defined as the implicit prices of attributes and are revealed to economic agents
from observed prices of differentiated products and the specific amounts of

characteristics associated with them” (p. 13). In his theory, a commodity (A) can be
described by various attributes (Z,) and expressed as A = (21, Z2, Z3, ..., 2,). The

hedonic price method estimates the implicit price of the attribute by using the following

general form of the hedonic price function:

PA =P(ZI,Zz, Z3, Zr) (1)

18

Where P A is the price of the commodity A and (21, 22, Z3, ..., 2,) are the various

attributes associated with the commodity A.

Once the hedonic function of the commodity price and its attributes is established,
the implicit marginal price of the attributes can be estimated after regressing the price of
the commodity on its attributes. In other words, the implicit marginal price of the
attributes is the partial derivative of the hedonic function with respect to each attribute

and can be expressed as:

P(Z,~) = a P Ana 2,- (2)9

The final stage of the hedonic price method is to estimate the inverse demand

function for the attribute Z, by combining the implicit marginal price and the quantity

 

9 To explain this in more detail, consider an individual who purchases the commodity A, with which all

attributes (Zi ) are associated. If X is all other goods consumed, the individual’s utility (U) is given by

U = U(Zi, X)
or
U = U(M - PA, 21')
since the individual maximizes utility subject to the income constraint:

M=X+PA

Then, a typical first order condition for the choice of attributes is Z,-

(a we z,)/(a U/a X) = a P A/a z,- = P(Z,-)

Thus, the partial derivative of the hedonic function represents the implicit marginal price of the attributes.

l9

along with other exogenous variables. The demand function can be obtained by inverting
the individual’s utility function and holding all but attributes i constant.

The hedonic price method uses primarily the housing market to quantify the
implicit values of environmental goods or services. As Freeman (2003) discussed, some
environmental goods and services such as amenities, proximity to hazard, and air and
water qualities are important attributes of a residential property and affect the
productivity of the property. In this case, the market value of the property will reflect the
productivity differentials caused by those environmental goods or services. The hedonic
price method measures the value of environmental goods or services by determining the
buyer’s willingness to pay for such attributes in the housing market.

There are many studies that measure the disamenity effect of hazardous waste
sites or contaminated sites in the literature. Many studies10 assume that the existence of
the hazardous waste sites can be considered disamenity factors to surrounding properties
due to environmental harms, aesthetic disruption, and other social problems associated
with these sites.

It can be further assumed that due to the disamenity effects, the proximity to these
sites is negatively related to the nearby property values and property values decrease as

the proximity to a hazardous waste site increases as Figure II-l shows. In Figure 11-1 , the

area P1P2D1 represents the aggregate loss of housing value because of the existence of a

hazardous waste site and is equivalent to the aggregated benefit of cleanup of the site.

 

‘0 See footnote 7 (p 11) for some examples.

20

A

Housing
Price

P2

 

 

A

Distance from
hazardous sites

Figure 11-1 Basic Principle of Hedonic Price Function (Ph)

Most previous studies showed that the assumption is true (Michaels and Smith,
1990, Kohlhase, 1991, Ketkar, 1992, and Kiel, 1995, Farber 1998, and Boyle and Kiel,
2001). Michaels and Smith (1990) used the hedonic price model to estimate the impact of
hazardous waste sites including 4 NPL sites in Boston on the equilibrium price of the
housing market. The study argued that a single hedonic price model would not be
adequate to describe the relationship between the housing equilibrium price and different
attributes of the homes in a large and complex market and characterized the housing
market into different submarkets using submarket definitions provided by local realtors.
The study used three variables to estimate the effect of the hazardous waste sites in
addition to the structural characteristics of the houses, neighborhood characteristics, and a
measure of access to employment areas: the distance between each home and the nearest

hazardous waste sites and two qualitative variables for sales in the six months after

21

discovery of the sites and after six months of announcements of hazardous waste sites.
The study estimated that the annual benefits from removing the specific hazardous waste
sites would be $115/mile for the sample analysis and an average of $139/mi1e for the
submarket analysis.

Kohlhase (1991) analyzed the impacts of toxic waste sites in the Huston area
before and after their designation as NPL sites using house sales data in 1976, 1980, and
1985. The study used three types of explanatory variables: structural characteristics of
properties, neighborhood characteristics from the census tract, and environmental
attributes such as distance between the house and the nearest waste sites. The study found
a significant impact of the toxic waste sites on the property value only after
announcement of NPL sites. The study also revealed that the house sales price increased
by $2,364 for one mile from the NPL sites. The distance effect could be observed up to
6.2 miles from the sites.

Ketkar (1992) analyzed the impact of hazardous waste sites on property values in
New Jersey using mainly 1980 Census data. The study employed three types of
explanatory variables such as the structural attributes of the properties, the neighborhood
attributes including the locational characteristics, and the environmental attributes such as
the number of hazardous waste sites in a municipality. The study found that the value of
the coefficient of the number of hazardous sites was negative and statistically significant.
The study estimated that the median property value would increase by $1,255 if a
municipality cleaned up one hazardous waste site.

Kiel (1995) measured the impact of two superfund sites on the property values in

a suburb of Boston using the single-family home sales data from 1975 to 1992. The two

22

hazardous waste sites were listed as NPL sites in the early 19805. The study used
proximity to the NPL sites as an environmental attribute in addition to other explanatory
variables. The study divided the time period under observation (1975-1992) into six
different phases to distinguish the effect of the hazardous waste sites based on the
availability of new information on the hazardous waste sites such as EPA’s announcement
of NPL sites and the cleanup plan. The study revealed that the effects of the hazardous
waste sites were negative and statistically significant for all time periods except the first
phase when the sites were neither deemed severely harmful by the residents nor listed in
the NPL. It was estimated that the property sales prices increased by various amounts
ranging from $1,377 to $6,468 during the five periods for one mile increase in the
distance of the property from the sites. The empirical study also revealed that the
announcement of the cleanup plan and initiation of the cleanup did not lead to a rebound
in nearby residential property values due to possible stigma effects.

These studies clearly show that the hazardous waste sites are considered
disamenities and the proximity to these sites is negatively related to the property values in
the surrounding neighborhoods. While these studies focused on the impact of severely
contaminated sites such as NPL sites, some other studies analyzed the impact of small-
scale hazardous waste sites, less contaminated sites, or sites that may be perceived as
contaminated (Ihlanfeldt and Taylor, 2004, Grigelis, 2005, and Kaufman and Cloutier,
2006). These studies provided meaningful information on the impacts of brownfields on
the surrounding neighborhoods since the definition of brownfields includes properties

with both actual and perceived contamination and most brownfields are not severely

23

contaminated. In general, these studies revealed that the effects of brownfields would be
substantial and statistically significant.

Ihlanfeldt and Taylor (2004) analyzed the effects of small-scale hazardous waste
sites on the values of surrounding commercial and industrial properties in Fulton County,
Georgia. The study analyzed the effects of the hazardous waste sites both before and after
the sites appear on various government lists of hazardous waste sites and tests whether
the effects are statistically different from one another. The study categorized the
commercial and industrial properties into five different land uses and analyzes separate
property price gradients for the properties’ proximity to hazardous waste sites. The five
land uses included apartment buildings, industrial facilities, office buildings, retail
buildings, and vacant land. The study also examined tax increment financing as an option
for funding for the cleanup of the hazardous waste sites by comparing the aggregated loss
in values of nearby commercial and industrial properties to estimated cleanup costs. The
study showed that the sites had negative extemality effects on all of five different land-
uses both before and after the sites were listed but the effects were statistically significant
only after the sites were listed. The study also revealed that the total loss of property
value resulting from the contaminated sites could be estimated as large as $1 billion for
the entire study area.

Grigelis (2005) investigated how real and perceived contaminated sites affect the
value of commercial and industrial properties in Fulton County, Georgia. The study used
two sets of hedonic price models to estimate the impact of both real and perceived
contaminated properties. For the first set of hedonic models, various governments’ lists of

hazardous sites were used to analyze the economic impact of sites with known

24

contamination. For the second set of hedonic models, the study developed the probability
of contamination model to identify potentially contaminated sites. The probability of
contamination model was developed with factors that could affect the investor’s
perception on the contamination of commercial and industrial properties. The study
identified that the negative economic impacts of both actual and perceived contaminated
sites on nearby commercial and industrial properties were substantial. According to the
study, property value losses resulting from known contamination could be estimated as
much as $1.07 billion and estimated losses resulting from perceived contamination could
be $663.09 million. The study concluded that significant gains could be expected from
the clean-up of the contaminated properties considering the impacts of actual or
perceived environmentally contaminated sites on the value of surrounding commercial
and industrial properties.

Kaufman and Cloutier (2006) estimated the impacts of both brownfields and
green spaces on residential property values in the Lincoln Neighborhood of Kenosha,
Wisconsin. The study identified two local brownfields and a neighborhood park to
measure the effects of their impact on the property values in the neighborhood. The study
formulated two versions of the basic semi-log model since the effects of the
environmental amenities and disamenities on a residential property value decrease as
distance increases. The study constructed property value gradient fimctions of residential
properties with distance from the brownfields and neighborhood park and found
statistically significant impacts of three sites on the residential property value in the
neighborhood. The study also estimated the impacts on the residential property values of

remediation and redevelopment of two brownfields into green spaces. According to the

25

study, remediation of the brownfields would increase the total value of the 890 residential
properties in the neighborhood by between $1.19 and $4.31 million and redevelopment of
the sites into green spaces would increase the property values by between $2.40 and

$7.01 million.

The studies about the impacts of hazardous waste sites or small scaled
brownfields on nearby property values using the hedonic price method show that the
disamenity effects of these sites are measurable and significant. Some studies also
suggest the importance of public policies regarding remediation and redevelopment of
these sites since significant economic benefits would be expected from the remediation or
redevelopment of these sites. However, there is another argument that property values
may not be fully recovered even though brownfields are cleaned up or redeveloped. Some
studies show that even though the negative effects of contaminated sites are substantial
and significant, the remediation of these sites does not always lead to a full recovery of
the properties’ own values or nearby property values due to stigma effects (Patchin, 1991 ,
Kiel, 1995, and McCluskey and Rausser, 2003). There is also an argument that failure to
account for the spatial correlation in the hedonic price model overemphasizes the
negative impacts of hazardous waste on surrounding property values and inflates benefit
estimates of remediation of these sites (Deaton and Hoehn, 2004). Furthermore, some
other studies show that the removal of negative extemalities is not capitalized into the
real estate markets and brownfield remediation or brownfield redevelopment does not
always lead to a rebound in property values in surrounding communities (Corona, 2004

and Longo and Alberini, 2005).

26

Patchin (1991) provided some evidence that property values of known
contaminated commercial and industrial properties are not fully recovered after
remediation due to stigma effects. He defined the term, stigma as “a loss in value beyond
the cost to cure the contamination itself” (p. 167). According to him, stigma has various
causes including fear of hidden cleanup cost, the trouble factor, fear of public liability,
and the inability to obtain financing. Kiel (1995) also revealed that the announcement and
the initiation of cleanup plans did not lead to a rebound in nearby property values due to
possible stigma effects as introduced before. McCluskey and Rausser (2003) also pointed
out that “both temporary and long-term stigma are possible equilibrium outcomes after
the discovery and cleanup of a hazardous waste site” (p. 276). According to them, two
extemalities contribute to stigma effects: an environmental extemality of contaminated
sites and a neighborhood extemality associated with contaminated sites. In their
theoretical model, the environmental extemality is the source of temporary stigma and
the neighborhood extemality results in long-term stigma. The study tested the existence
of both temporary and long-term stigma by estimating the impact of a lead smelter on
housing sales prices in Dallas County, Texas. The study analyzed the impact of the
smelter over four time periods: before identification of health risks, the cleanup stage and
the first ruling of completion of cleanup, second ruling of cleanup, and the additional
cleanup stage. The study also estimated bid functions for two attributes such as the
distance from the smelter site and the level of poverty in order to find empirical supports
for long-term stigma caused by the neighborhood extemality. The study concluded that

long-term stigma existed in a very limited area: within 1.2 mile from the smelter. The

27

study also noted that households from higher-income neighborhoods required a larger
discount to live near the remediated hazardous waste sites.

Deaton and Hoehn (2004) showed that standard hedonic procedure that uses a
distance variable might overestimate the effect of a hazardous waste site on nearby
property values due to omitted variable bias. The study examined two hedonic regression
models to estimate the marginal effect of reduced proximity to Superfund sites in Lansing,
Michigan. The first regression model omitted a measure of industrial activity while the
second model included the variable in the analysis. The results of the first regression
model indicates that a 10% increase in distance from a Superfund site increases house
prices by 0.32%. The result of the second regression model shows that the distance effect
is very low and no longer statistically significant. On the other hand, the effect of the
industrial activity variable is similar in size to the distance effect in the first model and
statistically significant. The study concluded that failure to account for the spatial
correlation overemphasizes the negative impacts of Superfund sites on surrounding
property values and inflates benefit estimates of hazardous waste clean-up.

Corona (2004) and Longo and Alberini (2005) further showed that the removal of
the negative effects of brownfields is not always capitalized in surrounding real estate
markets. Corona (2004), first, estimated the distance effects of brownfield sites on
residential properties in Connecticut stemming from proximity to brownfield sites and
found that the brownfield distance effect was positive and statistically significant. The
study also tested whether surrounding property values are subject to a rebound after the
brownfield is redeveloped using seven brownfields that were developed during the study

period. The study found that prior to redevelopment the brownfield distance effect was

28

similar to the previous analysis. However, the study suggests that a premium continues to
exist for distance from brownfields after redevelopment even though the premium is
lower than before. Longo and Alberini (2005) used the hedonic price model to
investigate the effects of proximity to contaminated sites and the effects of policies to
stimulate grth in neglected areas on the values of commercial and industrial properties
in Baltimore, Maryland. The study found that commercial and industrial property values
in Baltimore city were virtually unaffected by the proximity to sites either listed on or de-
listed from registries of contaminated sites.

The hedonic literature confirms that the existence of brownfields negatively
affects the value of surrounding properties. All studies revealed the negative relationship
between the proximity to brownfields and surrounding property values. However, the
literature does not provide an agreement on whether removal of negative amenities fi'om
brownfields can fully or substantially recover the property values in the surrounding
community. This study reexamines whether redevelopment of brownfields provides
positive impacts on surrounding property values, and thus measures the recovery of
surrounding property values. The following chapter discusses in detail the data and

empirical models for the study.

29

III.

3.1 Study Area

STUDY AREA, DATA AND EMPIRICAL MODEL

The study area is Lansing, Michigan, which is located in the southern central part

of Michigan’s Lower Peninsula. Lansing is the sixth largest city in Michigan and located

in Ingham County even though a small part of the City is in Baton County. Lansing,

which is the capital of Michigan, contains an estimated total population of 119,128

people and is comprised of an area of 33.8 square miles (U .8. Census Bureau, 2000).

Lansing is a typical urban area suffering from lower socio-economic status as

compared to the rest of the State. Table III-l shows a brief comparison of socio-economic

characteristics of the City of Lansing, Ingham County, and the State of Michigan in 1999

and 2007.

Table III-1 Comparison of Selected Socio-economic Characteristics in 1999 and 2007

 

 

 

 

 

 

 

 

Selected Socio-Economic US Census 2000 2007 Community Survey
Characteristics Lansing Ingham Michigan Lansing Ingham Michigan
Total population 1 19,128 279,320 9,938,444 118,123 279,295 10,071,822
(-O.84%) (-0.01%) (1.34%)
Median household
income ($) 34,833 40,774 44,667 36,550 45,204 47,950
(4.93%) (10.86%) (7.35%)
% of population with 16.9 14.6 10.5 23.20 18.30 14.00
Joverty (37.28%) (25.34%) (33.33%)
% of unemployed 6.4 5.7 3.7 8.6 5.9 6.1
population (16+) (34.38%) (3.51%) (64.86%)
% of population (25+)
with Bachelor degree or 21.2 33 . 21.8 24.3 35.1 24.7
flgher (14.62%) (6.36%) (13.30%)
Median Value of Housing 73,500 98,400 115,600 111,900 145,400 153,100
($) (52.24%) (47.76%) (32.33%)

 

 

 

 

 

 

 

Source: US Census 2000 and 2007 American Community Survey (US. Census Bureau)

30

In 1999 the median household income in Lansing was $34,833 and 16.9 % of the
total population was in poverty while the median household income and the percentage of
people below poverty level of the State were $44,667 and 10.5 %, respectively. The
median value of owner occupied housing in Lansing was $73,500 while the State’s
median value of housing was $115,600. The socio-economic characteristics of Lansing
were not significantly changed in 2007 as Table III-1 shows. In 2007 the estimated
median household income in Lansing was still well below the estimation of the State
median household income. These socio-economic indicators show that Lansing is a
typical urban area suffering from the flight of the middle class and a deteriorating urban
core as compared to the surrounding suburban areas.

As Lansing has experienced a decline in socio-economic conditions, many
commercial and industrial properties that once contributed to the local economy began to
lose their productivity and were eventually abandoned. In 1999 the City identified 19
brownfield sites, which consume approximately 75 acres of land within its limit (The
United States Conference of Mayors, 2000). Besides the City’s own identification of the
19 brownfield sites, a state database shows that between 1992 and 2006 Baseline
Environmental Assessments (BEAs) were conducted on more than 200 sites (MEDQ,
2007). Since BEAs tend to be conducted for a site that is planned or considered for reuse
or redevelopment, the list accounts for only a portion of all brownfields in Lansing,
indicating that the number of brownfields is even greater. The presence of many
brownfields has reinforced and magnified the problems associated with the declining
local economy. However, as the number of BEAs filed in Lansing indicated, there have

been ongoing public and private efforts to reuse or redevelop brownfields since Michigan

31

began to implement its voluntary cleanup program progressively by amending the
Michigan Environmental Response Act (MERA) as Part 210 of the Natural Resources
and Environmental Protection Act known as Public Act 451.

This study analyzed the effects of four brownfield redevelopment cases on
property values in surrounding neighborhoods as an attempt to quantify indirect benefits
of brownfield redevelopment. The four cases identified by Lansing Economic
Development Corporation included Prudden Place Apartments, Motor Wheel Lofts,
Neogen Corporation, and Builders Plumbing & Heating Supply. These sites were
previously obsolete and blighted properties and successfully redeveloped with public
financial assistance between 2004 and 2006. Figure III-1 provides a map that shows the
location of the sites.

Prudden Placement Apartments and Motor Wheel Lofts are two different multi-
family residential redevelopment projects on a 15-acre former wheel manufacturing site
of Motor Wheel. Motor Wheel was founded in 1898 by Georgia-born William Prudden
and became the world’s largest producer of both wood and steel wheels (Grater Lansing
Business Monthly, 2007). Motor Wheel became a subsidiary of Good Tire and Rubber in
1964 and operated the facility in Lansing until 1974. The property was sold to Haynes
Wheels in 1974 but was vacated the following year. The property was abandoned until
Harry Helper purchased the property in 1998, with plans to develop Motor Wheel Lofts.
The western portion of the property was resold to another developer named Pat Gillespie

who built Prudden Place Apartments.

32

 

 

Prudden Place & Motor Wheel

Neogen

Builders Plumbing & Heating
Supply

 

0 0.2 0.4 0.8 1.2
c—Zwfiles

 

Figure III-1 Location of Brownfields

Motor Wheel Lofts, a 120 unit loft apartment complex, was developed by
renovating the historic four-story Prudden Motor Wheel factory in 2005. Prudden Place
Apartments is another 120 unit multi-family residential development on the western

portion of the Prudden site. The two redevelopment projects were helped and facilitated

33

 

by many federal and state incentive programs such as the Brownfield Single Business Tax
(SBT), Brownfield Tax Inclement Financing, Neighborhood Enterprise Zone credits, and
Federal and State Historic Preservation Tax Credits. Figure III-2 shows the picture of the
two sites after redevelopment. The two independent brownfield redevelopment projects

will be considered as one case for this study since the two projects are located in the same

area and have developed in the same time period.

    

Figure III-2 Motor Wheel Lofts (left) and Prudden Place Apartments (right)
Source: The City of Lansing

Another brownfield redevelopment case for this study is Neogen Corporation.
Neogen Corporation, founded in 1982, is a developer and manufacturer of diverse
products related to food and animal safety. The Lansing-based company, which has been
repeatedly named one of the 200 best small companies in America by Forbes, has been
expanding its business by acquiring several facilities in Lansing. The company purchased

the building on 720 East Shiawassee Street to open its fifth facility in Lansing in 2002.

34

The old industrial building was renovated in 2003 with support from state and
local brownfield financing programs such as the Michigan Brownfield Single Business
Tax Credit and the Brownfield Incremental Financing Program. The company also
decided to expand and consolidate two out—of-state facilities in Lansing after considering
locating the facility at a competing site in Chicago. The company renovated an
abandoned building adjacent to its recently renovated building on Shiawassee Street to
house the new consolidated manufacturing facility in 2005. The State and City

governments provided financial assistance to help the company’s expansion project.

 

Figure III-3 Neogen Corporation
Source: The City of Lansing

The last brownfield redevelopment case examined for this study is an expansion
project of Builders Plumbing and Heating Supply near the intersection of Homer and
Kalamazoo streets in Lansing. Builders Plumbing and Heating Supply is one of the

companies of the Crawford group, which is a wholesale supplier of plumbing and heating

35

products in the Midwest. The company planned the partial demolition of an existing
structure and the construction of a new 40,000-square—foot expansion in 2004. The

expansion project was completed with financial assistance fiom the State Brownfield

Single Business Tax Credit as well as the City’s tax increment financing.

5;

    

Figure III-4 Builders Plumbing Heating Supply
Source: The City of Lansing

These brownfield redevelopment projects have helped revitalize the City’s
economy in addition to bringing the old and obsolete properties into productive use. The
availability of financial assistance has leveraged millions of dollars in private investments
on these properties, creating new jobs and raising the City’s tax base. Motor Wheel Loft
created 331 new jobs after attracting about'$1.32 million in private investment. Prudden
Place Apartments was expected to spur more than $6.5 million in private investment.
Neogen Corporation invested about $2 million for the renovation of two buildings and
created 42 new jobs. Builders Plumbing and Heating Supply also created 35 high—paying

jobs with $1.9 million in private investment.

36

In addition to the direct impacts of these redevelopment projects, they provide
other indirect impacts such as helping revitalize the City, improving the local
environment by removing eyesores, and increasing property values in the surrounding
neighborhoods. This study measures the indirect impacts of these brownfield

redevelopment projects on local property values.

3.2 Data

This study collected available data necessary for designing hedonic price
fiinctions to measure impacts of brownfield redevelopment projects on the surrounding
properties. The primary data source of the study is property transaction data and
residential building structural data on each parcel obtained from the Equalization
Department of Ingham County and the Assessor’s Office of Lansing. The US. Census
2000 was another important data source to identify the characteristics of neighborhoods
in which each house is located. The digital parcel map and land use map of Lansing was
obtained from the Office of Lansing Information and Technology and used to map all
house sales data on a GIS (Geographic Information System) map with other attributes
that came from the US. Census 2000. Building characteristics data were also
incorporated into the GIS map in order to determine the distance between each sales
observation and the brownfield redevelopment sites.

The property transactions data include all property sales that were registered to
the Office of the Ingham County Register of Deeds between 1998 and 2007. This study
used only single-family home sales in Lansing and took several elimination processes to

exclude recording errors and “non-arms length” sales. First, the study eliminated single

37

family housing transactions that did not use Warranty Deeds as their sales instrument.
The eliminated transactions include sales categorized as Quit Claim Deeds, Sherriff’s
Deeds, Tax Deeds, Court Deeds, etc. The property transactions with these types of
transaction instruments are not normally considered as “arms length” sales. The next step
was to remove all other sales that were not readily considered as normal sales. The
Ingham County property sales data include records and explanations on unusual sales,
including transactions between family members, adding a spouse, estate actions, non-
profit transactions, etc. The next elimination process gave consideration to missing
property attributes. All properties with no sale amount, zero lot size, or no bathroom,
were eliminated in this process. After these elimination processes, the study identified
19,489 housing sales from 1998 to 2007 for this hedonic price analysis.

The housing sale observations were further reduced in the process of the hedonic
price analysis to minimize the distortion of a hedonic price function. Considering the
relatively lower contamination levels of the brownfield sites, negative extemalities of the
brownfields are assumed to be localized to a relatively small area as Deaton (2002)
addressed. This study limited the study area within 1.5 miles from each brownfield site
for the hedonic price analysis resulting in 8,458 housing sales observations in the analysis.

The structural data of the buildings obtained from the Equalization Department of
Ingham County contains many physical attributes of houses used for the hedonic analysis.
The structural attributes include lot size, floor area, garage size, basement area, finished
basement area, number of stories, exterior type, etc. The US Census 2000 Summary File
1, 2, 3, 4 produced by the US. Census Bureau provided important neighborhood

attributes that have been considered as affecting the housing sales price. These additional

38

variables were incorporated with all selected housing sales observations in the digitized
parcel map. The ArcINFO, a GIS modeling and mapping software, was used to combine
different sets of data into a database as Figure HI-5 shows. The GIS map was also used to
determine the distance variables between sales observations and each brownfield site.
Figure III-6 shows a calculation of the distance variable using the “NEAR” tool of the
ArcINFO. Another distance variable between housing observations and the perimeter
boundary of the nearest industrial area was also determined using the GIS system. The
land use map and parcel map were obtained from the Ofiice of Lansing Information and

Technology.

. ~va . '. v.1
5... run {new War: but Wm 19m mm" hdp
D IS 9 O 1. n Q “"515! ‘ " "

(dflej' * =' '

   
 
   
  
 
 
   
     
   
  
  
 

.Bu-Hen PlrmD-ngJW-ng new
. mm
I
- E Fmdenggwxwmum
I was. vim Mom \‘l‘nw 1dr:

    

. i ' u
. a Impass- . II I I
- El “my. slim _ 1
Mill"-
- B Mann-I an
'-’ B l-MmlI-lgflomjumtcd
C
= Cl

. [j vacuum-

»Ermm

 

 

,.._-_ WW4.» ......... 4 .
Ripggfiaml

W9" 0”“

 

   

 

39

 

Ei- idt In Mind: hm: Won Iwi mm— Help
Drifll H n s hulf'mm ”7'

 

     
 
 
  
  
 
 
 
 
 
 
  
 

 

64.51%- R? O -=«i‘”"3 U Iang—QA’I‘IZQGR

 

 

. . .
IiJm-"'“rr.-" '7 " W " " } "unites; " ' ' ‘l

 

 

 

 

 

 

: B has. I —. . WI,“ rot-n
undo-warm"; Nut”, $099., I l ,4 ‘ hum d
_ NW ' ;. - .
B - 1‘ Nu: ,h 159.3?“
. _ _ . . . 3
I vmdam PI-u _Mulnr wl TEQTIQ‘ _ 3 ‘Ji ; N."
-. 8 "my: _ run-- I
, , ...H-.. .. .., i l Datum-nos m. arr-nu
.12 5' 1 MN ugh tutu:- In 10.
; B WNW" . - .- ... Input Futuurarni
I ”’9‘" I'll < nun: hmm m m
-. E m I ’4. 1 row Farm. Mhm rm
‘5; v XJI i Sum. m" -
, D melmamgmm : fr ' x
w " I
Ax. ) .
.-. C1 vacuumi- . l
v “1.1

.-' B mun

  

 

 

 

 

 

 

Figure III-6 The Calculation of Distance Variables Using “Near” Tool of the ArclNFO

4O

3.3 Empirical Model

In the housing market, there are basically three categories of attributes that atl‘eet

housing sales prices: structural. neighborhood, and environmental attributes. The

structural attributes (S,) are the characteristics that reflect size and other structural

qualities of the property itself. The neighborhood attributes (N ,-) reflect the socio-

economic characteristics of the neighborhood where the property is located.

Environmental attributes (E,-) refer to various site-specific environmental amenities or

disamenities that affect the quality of life in the property.

The hedonic price method expresses the ith housing price (Pi) as a function of

these attributes:

P,- = P(Si, Ni, Er) (3)

The literature suggests that there are various functional forms that can be used to
explain the relationship between housing prices and proximity to brownfields: the linear
function, the quadratic function, the logarithmic function, and the inverse function. In the
early stage of the hedonic literature, the linear function was commonly used to explain
the relationship. The recent studies preferred to assume other relationships such as an
inverse relationship and a log-linear relationship because it could be assumed that the
disamenity effects of brownfields decay with distance. However, these functional fomrs

provide interpretative difficulties, as Corona (2004) pointed out.

41

This study used the linear relationship between housing prices and proximity to
brownfields as a base functional form due to its relatively easy interpretation of the
coefficient estimates of independent variables. Many previous studies including Wisinger
(2006)ll have also used the linear relationship to measure the impacts of environmental
hazards because of the ease of interpretation.

The study also confirmed the linear relationship between housing prices and other
independent variables by drawing a scatterplot matrix. Figure III-7 shows the relationship
between selected variables. The scatterplot matrix showed a modest linear relationship
between these variables exists within a 1.5 mile radius of each brownfield. The scatterplot
matrix also showed that a log-linear function can be also used for the hedonic analysis
since the natural log of price (LnPrice) appeared to be linearly related to the selected

independent variables including the distance variable.

 

” Wisinger (2006) used two functional models for measuring the irrrpact of chemical hazardous sites on

residential values: liner and log-linear functional models. This study used the same functional forms as
Wisinger’s models.

42

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m JsAAJL
We fidddh
a»; anew
'Wres new
sense JL
wmrvrrr u
MMLiiiL$

 

 

 

Figure III-7 Scatterplot Matrix for Selected Variables

The study is also limited to sales observations within 1.5 miles from each
brownfield site in the hedonic price model since impacts of brownfields were expected to
be highly localized. The existing literature shows little guidance in determining the exact
point where brownfields have no impact on housing value. According to Deaton (2002),
determining the exact point is a judgment call. To determine a reasonable range of impact,
Ihlanfeldt and Taylor (2004) continued to expand the assumed impact area surrounding a
site using quarter-mile increments until a decline in the precision of the estimated

gradients was observed. In their research, this occurred between 1.5 and 2.0 miles for

43

each of the land uses. Grigelis (2005) used the same approach and identified 1.25 ~ 1.5
miles from contaminated sites as a range of impact for different categories of land uses.
This study followed the method used by Ihlanfeldt and Taylor to determine the impact
area and observed that a decline in the precision of the estimated gradient occurred
between 1.5 and 2 miles from brownfield sites. Based on the result, this study assumed
that the range of impact of brownfields would be 1.5 miles from each brownfield site.
With the above assumptions, this study used the following linear functional form

as a base hedonic model:

T J K
price” 2 a ..th1), +chsy. +deN, +mIND, +
k=1

z=1 j=1

,6, Brownfield B + ,BzBrownfield D + ,83Brownfield A + e”

(4)

where Price“ is the sale price of house i at the time of t, a is constant, D, are dummy
variables indicating a year when the house was sold, Si}- are structural variables of house i,

Nik are neighborhood variables, and IND,- is the distance between the house i and the

nearest industrial area in feet. Brownfield B is the distance between the house i and
brownfields in feet, if the sale occurred before development, otherwise 0. Brownfield D is
the distance between the house i and brownfields in feet, if the sale occurred during the
development phase, otherwise 0. Brownfield A is the distance between the house i and

brownfields in feet, if the sale occurred after development, otherwise 0. Finally, e

represents a random error.

44

While the linear functional form is easy to interpret, the log-linear functional form
has some advantages over the linear functional form. According to Wooldridge (as cited
in Hwang, 2003), the log-linear functional form usually reduces the likelihood of
heteroskedasticity, makes estimates less sensitive to outliers by narrowing the range of
the dependent variable, and provides two interpretations for the coefficient estimates such
as the implicit hedonic price and the percent of the average house price. Because of the
advantages from using the log—linear function, this study conducted an additional log-
linear functional analysis to compare the overall benefit estimate of the three brownfield
redevelopment cases with the results of the linear functional analysis.

For the log-linear fimctional analysis, all housing sales prices were converted to
natural log values (LnPriceit). The new hedonic price analysis used the following log-

linear functional model:

T J K
Ln Price, = a + 212,1), + chs, + deN,, + eIND, +
[:1 j=1 k=1
(5)

,6, Brownfield B + ,BzBrownfield D + ,83Brownfield A + 6,,
Table III-2 shows the full set of variables including structural variables 8), the

neighborhood variables N,-, and two environmental variables E. In general, the study

followed the hedonic price literature in selecting the independent variables, but the

availability of data was also an important factor.

45

Table III-2 Description of Variables for the Hedonic Analysis

 

 

Variable Description
Dependent Variable
Price Housing sales price in 2007 dollars
LnPrice The natural log value of Price

Environmental Variables

 

IND

Brownfield B
Brownfield D

Brownfield A

Distance from each house to the nearest industrial area in feet

Distance (feet) from each house to brownfield site if sale occurred before
development, otherwise 0

Distance (feet) from each house to brownfield site if sale occurred during
the development phase, otherwise 0

Distance (feet) from each house to brownfield site if sale occurred after
development, otherwise 0

Housing Structural Variables

Acre

Age

Area
Garage
Basement
FINBasement
Update
styl

styI .25
sty] . 5
styl . 75
sty2
dumsty

alum_vinyl
block_brick
pine_lap

Dumexterior

Total acreage of the house

Age of the house when sold

Total floor area of the house in square feet

The total area of garage in square feet

The total area of basement in square feet

The total area of finished basement in square feet
Dummy variable indicating remodel or update of the property after 1975
Dummy variable indicating 1 story building
Dummy variable indicating 1 1/4 story building
Dummy variable indicating 1 1/2 story building
Dummy variable indicating 1 3/4 story building
Dummy variable indicating 2 story building

Dummy variable indicating bi-level, tri-level or 2 1/5 story house

Dummy variable indicating the exterior of the house is aluminum or vinyl
siding

Dummy variable indicating the exterior of the house is block, brick, stone, or
stucco siding

Dummy variable indicating the exterior of the house is pine/cedar, masonite,
or lap siding

Dummy variable indicating the exterior of the house is other materials except
wood siding

 

46

Table III-2 (cont’d).

 

Neighborhood (Census Block Group) Variables

Minority] Percentage of the population that is Afiican American

MinorityZ Percentage of the population that is Hispanic origin

Income Median household income in 1999 dollar

Education Percentage of the population 25+ with at least an associate degree

Rent Percentage of rental houses in the census tract

Sale Years

D, Dummy variables indicating the year (l997~2007) when the house was sold

 

Dependent Variables

The study used Price as the dependent variable for the linear functional analysis.
Price is the sale price of single family housing in Lansing from 1997 to 2007. The
housing price was converted to 2007 dollars using the Consumer Price Index (CPI) in

order to capture inflation effects in housing prices for each year. The model also included

dummy variables (D,) of the sales year as explanatory variables to reflect local trends in

the housing market. The study then converted all housing sales prices to natural log
values (LnPrice) and used it as the dependent variable for the log-linear functional

analysis.

Environmental Variables
The study included two environmental variables basically: Brownfield and IND.
Brownfield is the distance between each house to a brownfield site. Since the study

focused on the benefits of brownfield redevelopment, the housing sales observations were

47

divided into three different time phases of brownfield project redevelopment: pre-
development, development, and post-development. Several previous studies such as
Kohlhase (1991), Kiel (1995), and Dale, et al. (1999) used the same approach to

distinguish the effect of the hazardous waste sites based on the different status of the sites

such as before and after cleanup. The three variables, Brownfield B , Brownfield D ,

and Brownfield A represent the distance from each house to a brownfield site if property

sale occurs before, during, and after redevelopment of the brownfield, respectively.

This study distinguished the development phase because it can be assumed that
the public funding announcement of the development projects and actual development
activities may affect the property values of the surrounding neighborhoods. Since all
three brownfield sites of this study were named based on the same redevelopment period,
the three sites showed the same development time phases. The brownfield sites were not
developed in the pre-development phase (1998~2003). Even though the Neogen site was
partly remodeled in 2003, its expansion and consolidation project started right after the
state government announced its financial support for the project in 2004. Because of this,
this study considered 2004 as the beginning of the development phase for this site. In the
development time phase (2004»2005), various public financial supports were announced
for the development of these sites and various development activities were implemented
accordingly.

The study included an additional environmental variable (IND), the distance
between housing observations and the perimeter boundary of the nearest industrial area to
avoid over-estimation of the effect of negative extemalities resulting from the existence

of other brownfield sites. As Deaton. (2002) shows, failure to take into account areas of

48

high industrial activity as well as other potential brownfields will result in inflating the

impacts of hazardous sites on the surrounding neighborhood.

Housing Structural Variables: Environmental Variables

In the housing market, the structural attributes are important factors that affect
housing sales price. The study used housing structural variables commonly included in
hedonic analyses such as the total acreage of the house (Acre), the age of the house (Age),
the total floor area (Area), the total area of garage (Garage), the total area of basement
(Basement), and the total area of finished basement (FINBasement). The study also
included a series of dummy variables indicating remodel or update (Update), the story of
the house (styl, 31321.25, sty]. 75, 51322, dumsty)”, and the exterior of the house

(alum_vinyl, block_brick, pine_lap, Dumexterior)”.

Neighborhood Variables

The neighborhood variables containing socio-economic characteristics are also
important factors that affect housing sales price. The study obtained the neighborhood
variables from the US. Census 2000. These variables include racial composition
(Minority! and Minority2), the median household income (Income), the level of
education (Education), and the percentage of rental houses (Rent). These neighborhood
variables were obtained from the US. Census 2000 Summary File 1, 2, 3, 4 produced by

the US. Census Bureau.

 

‘2 For the detailed descriptions of the dummy variables, see Table III-2.
‘3 For the detailed descriptions of the dummy variables, see Table III-2.

49

This study, first, analyzed all three brownfield sites individually in order to
examine the impact of redevelopment of the individual site on surrounding property
values by using the base hedonic functional form, the linear functional form. Then, this
study combined all three sites in an analysis to assess the aggregated impact of
redevelopment of all three sites. Finally, this study conducted an additional log-linear
functional analysis to compare the aggregated benefit estimate of the three brownfield
redevelopment cases with the result of the linear functional analysis.

The main hypothesis of the study was that the redevelopment of brownfields
provides a positive impact on surrounding property values by eliminating or substantially
reducing the negative extemality associated with each brownfield site. The following
chapter shows the empirical estimations of the effects of these brownfield redevelopment

projects.

50

IV.

4.1 Descriptive Summary of Variables

EMPERICAL RESULTS

This study is limited to housing sales observations within 1.5 miles from each

brownfield site. The completed data set contained a total of 8,458 single-family housing

sales that were observed within 1.5 miles from all three sites between 1998 and 2007. As

explained in Chapter III, a total of 37 independent variables were identified to explain the

housing sales price (Price). The housing sales price was adjusted to 2007 dollars using

the Consumer Price Index (CPI) and the average sale price was $82,912. The average

housing sales price and the average household income of the study area was a bit lower

than the City’s average housing value and household income described in Chapter III.

Table lV-l shows the summary statistics of all variables in detail.

Table IV-l Summar)I of Variables for All Sites (N of observations: 8458)

 

 

 

 

 

 

 

Variables Mean SD Minimum Maximum

Price 82912.56 37848.798 6000 500282
Acre .13165 .062010 .028 1.130
Age 78.10 20.204 1 152
Area 1147.67 412.148 400 5082
Base 682.34 241.177 0 2307
FINBase 58.77 159.847 0 1457
Garage 223 .65 196.523 0 1660
Update .02 .136 0 1
Black 17.372 12.8498 3.5 66.7
Hispanic 12.48 5.685 1 34
Rent 34.7491 17.06897 4.57 86.82
Income 33423.02 9211.719 10323 60063

 

51

 

Table IV-l (cont’d).

 

 

 

 

 

 

Education 29.219 14.8873 7.3 69.2
STYI .39 .488 0 1
$771.25 .12 .319 0 1
5771.5 .08 .264 O 1
STYI. 75 .10 .294 0 1
STY2.0 .31 .465 0 1
dumSTY .01 .083 0 l
alum_vinyl .30 .457 0 l
block_brick .02 . 149 0 1
pine_lap .00 .043 0 1
wood .67 .471 0 1
dumexterior .01 .091 O l
dum98 .09 .287 0 l
dum99 .10 .300 0 l
dim00 .10 .300 O 1
dumOI .10 .306 0 l
dum02 .11 .319 0 1
dum03 .11 .317 0 1
dum04 .12 .328 0 l
dum05 .11 .307 O l
dum06 .09 .291 0 l
dum07 .06 .228 0 1
Industry 1326.9808 918.43984 .00 4983.83
Brownfield B 3189.9175 2914.87682 .00 7919.65
Brownfield D 1161.17 2329.618 0 7918
Brownfield " 751.13 1957.574 0 7918
Brownfield” 5102.2176 1951.80243 96.86 7919.65

 

 

4 . . . . . . .
l Brownfield variable was not used In the hedonic analysrs. Instead drstance variables before, during, after
development were used to estimate the effects of brownfield during the different time phases.

52

 

The study also separated all housing sales observations into three different groups,
which were used to assess the effects of individual brownfield sites on the surrounding
property values. The mean values of housing price and household income of the Builders
group were higher than the other two groups. Table IV-2 provides summary statistics of

variables of all three groups.

Table IV-2 Summary of Variables for Individual Sites

 

 

 

 

PRUDDEN NEOGEN BUILDERS
(N=5933) (N=5 345) (N=6543)

Variables Mean SD Mean SD Mean SD

Price 79975.38 38493.55 77929.60 37604.14 81515.86 34810.00
Acre .132 .067 .126 .058 .124 .053
Age 80.27 20.957 82.94 19.839 81.12 19.396
Area 1170.13 433.758 1188.27 436.151 1172.27 416.152
Base 674.61 247.452 680.21 230.625 688.94 224.227
FINBase 46.41 148.468 41.22 134.016 53.77 149.723
Garage 217.99 201.272 203.90 198.312 212.75 193.592
Update .02 . 136 .02 .142 .02 .141
Black 19.085 12.697 18.962 12.2916 16.834 12.517
Hispanic 13.95 5.915 13.71 6.041 11.62 5.398
Rent 38.6999 16.01126 40.753 15.906 36.691 18.136
Income 31724.32 8641.046 31160.54 8063.546 33260.09 8526.255
Education 27.427 13.5054 27.564 13.2909 30.362 14.2917
S TYI .35 .477 .32 .466 .36 .480
STYI.25 .10 .298 .09 .290 .11 .310
STYI.5 .07 .259 .07 .263 .08 .264
STYI.75 .11 .311 .12 .323 .11 .307
STY2.0 .36 .481 .39 .488 .35 .476
dumSTY .01 .088 .01 .086 .01 .078
alum_viny1 .21 .410 .20 .398 .30 .456

 

 

 

 

 

 

 

53

 

Table IV-2 (cont’d).

 

 

 

 

 

 

 

 

block_brick .01 .111 .01 .112 .02 .155
pine_lap .00 .026 .00 .027 .00 .049
wood .76 .425 .78 .414 .67 .471
dumexterior .01 .092 .01 .094 .01 .095
dum98 .09 .288 .09 .291 .09 .291
dum99 .10 .300 .10 .301 .10 .300
dim00 .10 .297 .10 .293 .10 .300
dumOI .10 .299 .10 .295 .10 .306
dum02 .11 .319 .12 .321 .12 .320
dum03 .11 .317 .11 .318 .11 .317
dum04 .12 .330 .13 .332 .12 .327
dum05 .11 .308 .11 .309 .10 .305
dum06 .10 .293 .09 .291 .09 .292
dum07 .06 .234 .06 .232 .05 .224
Industry 1521.850 979.982 1521.907 983.164 1326.113 899.775
Brownfield” 3339.74 2970.693 3345.03 2920.405 3464.7455 3035.155
Brownfield D 1227.04 2406.490 1272.09 2439.069 1257.80 2481.261
Brownfield ‘ 827.09 2066.716 814.71 2037.276 800.76 2064.967
Brownfield 5393.874 1770.476 5431.824 1609.903 5523.30 1830.648

 

4.2 Results of the Hedonic Price Analysis

The first step in the analysis was to analyze the impact of the redevelopment of

individual brownfields on property values in the surrounding neighborhood by using the

base hedonic firnctional form, the linear functional form as stated in Chapter III.

54

 

The Case of Prudden Place Apartments and Motor Wheel Lofts (Prudden)

For the group of sales observations within 1.5 miles from Prudden Place
Apartments and Motor Wheel Lofts (Prudden), the mean distance between all sales
observations and Prudden was 5,393.874 ft and the average sales price was $ 79,975 as
converted to 2007 US Dollars. As formulated in the empirical hedonic price model, a
total of 37 independent variables were used to explain the sales price of single-family

housing. Table IV-3 provides the estimated coefficients of the linear hedonic price

fimction.

Table IV-3 Estimated Coefficients (Prudden)

 

 

 

 

 

Variables B Sig.

(Constant) 45 798.555 .000*
Acre 34842.593 .000*
Age -328.242 .000*
Area 32.786 .000*
Base 13.930 .000*
FINBase 18.464 .000*
Garage 19.190 .000*
Update 12853072 .000*
Black -167.995 .000*
Hispanic -55 .445 .516
Rent -249.361 .000*
Income .209 .002*
Education 631.608 .000*
STYI -5734.537 .000*
STYI.25 -2948.795 .028*
STYI.5 -132.550 .923

 

55

 

Table IV-3 (cont’d).

 

 

 

 

STYI. 75 322.825 .782
dumSTY -6263.894 .097“
alum_vinyl -1079. 134 .185
block_brick -1606.380 .584
pine_lap 7992.735 .519
dumexterior -6408.754 .068”
dum98 -39254.313 .OOO*
dum99 -34193.550 .000*
dum00 -32192. 123 .000*
dum01 -21870.819 .000*
dum02 -18041.544 .OOO*
dum03 -1 1094.049 .000*
dum05 2064.431 .130
dum06 -2229. 173 .519
dum07 -13539.781 .000*
Industry 2.142 .000*
Brownfield 3 1.816 000*
Brownfield D 1.019 .016*
Brownfield A 1.267 013*
DifferenceD‘B '5 -.797 .069**
Difference“? 16 -.549 .290
N=5933 R square = 0.589 F: 248.786

Excluded Variables: STY2, wood, dum04

 

 

* Statistically significant at the 5% level
** Statistically significant at the 10% level

 

 

'5 DifferenceD’B is the difference between Brownfield D and Brownfield B . Ihlanfeldt and Taylor

(2004) used a dummy variable to measure statistical significance of the difference between the coefficients
for two distance variables. This study used their approach to measure the statistical significance of the

difference between Brownfield D and Brownfield B .

'6 Difference’H’ is the difference between Brownfield A and Brownfield B

56

 

The estimated coefficients of the housing structural variables were consistent with
the prior literature. The positive coefficient estimates of lot size (Acreage), floor area
(Area), basement area (Base), finished basement area (FINBase), and garage area
(Garage) showed that these variables were positively related to the housing price (Price).
For example, the estimated coefficient of the floor area indicates that the housing price is
expected to go up by $32.786 per a one square foot increase in the floor area, all else
constant. The negative sign of the age of house (Age) indicates that the age of the house
is negatively related to the housing price and the housing price is expected to decrease by
$328242 per a one year increase in the age of the house, holding all else constant. These
coefficient estimates were statistically significant at the 5% level.

The estimated coefficients of the neighborhood variables were also consistent
with the prior expectations. The percentages of the minority populations (Black,
Hispanic) and the percentage of rental housing units (Rent) were negatively related to the
housing price (Price) while the relationship between the housing price and other two
neighborhood variables such as the median household income (Income) and the level of
education (Education) were positive. The coefficient estimates of all neighborhood
variables except Hispanic were statistically significant at the 5% level.

The negative sign of the coefficient estimate of Industry shows that housing price
would be lower when a house is located closer to an industrial area. As Deaton (2002)
pointed out, proximity to industrial areas affects the benefit estimation of brownfield
redevelopment negatively. Omitting this variable would result in an over-estimation of

impacts of brownfields on the surrounding property values.

57

The coefficient estimate of Brownfield D confirmed the previous literature that the
proximity to brownfield sites or environmental hazards is negatively related to the
property values in nearby neighborhoods. The estimated coefficient of Brownfield D was
positive and statistically significant at the 5% level. The positive sign of the coefficient
estimate of Brownfield D means that the distance between Prudden and properties within

a 1.5 mile radius of Prudden is positively related to the housing price. The housing price
was expected to increase by $1.816 as a house was located one foot further away from
Prudden, holding all else constant. Conversely, before redevelopment of Prudden, the
housing price in the surrounding neighborhood was expected to decrease by $1.816 as its

proximity to Prudden was increased by one foot, holding all else constant.
The coefficient estimates of Brownfield D and Brownfield A represent the effects

of redevelopment of Prudden on property values in the surrounding neighborhood. Both
coefficient estimates were positive and statistically significant at the 5% level, indicating
that redevelopment of Prudden did not lead to a full rebound in property values in the
surrounding neighborhood. This can be explained by the fact that a portion of the Motor
Wheel site still remained undeveloped. Therefore, it is reasonable to expect that the

Motor Wheel site still generated negative extemalities on residential properties in the

surrounding neighborhood. However, the coefficient estimates of both Brownfield D

and Brownfield ” were lower than the coefficient estimate of Brownfield D . Thus,

redevelopment of Prudden had a positive impact on housing values in the surrounding
neighborhood. Especially, the housing price was affected greatly during the development
period. It means that the announcement of the State’s financial assistance and the wide

publicity of the residential development projects provided a sign that the negative

58

extemality associated with Prudden would be eliminated. After redevelopment of
Prudden, the housing price was estimated to recover $0.549 or 30% out of the housing
value 1055 of $1.816 per one foot from Prudden.

However, a separate regression analysis showed that the difference between the

coefficient estimates of Brownfield D and Brownfield D was not statistically significant.
In order to determine the difference between the coefficient estimates of Brownfield D ,
Brownfield D , and Brownfield D , new coefficient estimates of DifferenceD‘D and
DzflerenceD'D were determined from a separate regression analysis in which
Brownfield D , Brownfield D , and Brownfield D were combined to form

Brownfield (defined as the distance to the brownfield site). The regression analysis
estimated Brownfield , Brownfield D , and Brownfield D all together in the model. The
coefficient of DifferenceD‘D shows the difference between the coefficients of
Brownfield D and Brownfield D and was statistically significant at the 10% level. The
coefficient of DiflerenceD‘D shows the differende between the coefficients of

Brownfield D and Brownfield D was not statistically significant.

The Case of Neogen Corporation (Neogen)

In the case of Neogen Corporation (Neogen), a total of 5,345 housing sales
observations entered the hedonic model. The mean distance between all sales
observations and Neogen was 5431.824 ft and the average sales price was $ 77,9296 as
converted to 2007 US Dollar. Table IV-4 provides the estimated coefficients of 37

independent variables, which were used to explain housing prices.

59

Table IV-4 Estimated Coefficients (NEOGEN)

 

 

 

 

 

Variables B Sig.

(Constant) 45255.137 .000*
Acre 48531.874 .000*
Age -305.497 .000*
Area 30.381 .000*
Base 13.615 .000*
FINBase 19.150 .000*
Garage 17.666 .000*
Update 12657.339 .000*
Black -240.778 .000*
Hispanic -40.023 .620
Rent -232.876 .000*
Income .112 .117
Education 665.052 .000"'
STYI -6847.707 .000*
STYI.25 -2122.386 .134
STYI.5 895.241 .525
STYI. 75 504.693 .665
dumSTY -2035.365 .611
alum_vinyl -2021.958 .018*
block_brick 1133.498 .709
pine_lap 12185.087 .320
dumexterior -7891.549 .027*
dum98 -37565.647 .000*
dum99 -32030. 141 .000*
dum00 -30512.982 .000*
dumOI -21274.617 .000*
dum02 -16658.057 .000*
dum03 -9343.866 .002"‘

 

60

 

Table IV-4 (cont’d).

 

 

 

 

dum05 3298.492 .020*
dum06 -1367.242 .726
dum07 -12052.347 .003*
Industry 2.415 .000*
Brownfield D 1.810 .000“
Brownfield D 1.211 .008*
Brownfield " 1.572 005*
DifferenceD’D -.599 .234
Difi'erenceD‘D -.238 .686
N=5345 R square = 0.581 F= 216.320

Excluded Variables: ST Y2, wood, dum04

 

 

* Statistically significant at the 5% level
** Statistically significant at the 10% level

 

Like Prudden, the estimated coefficients of the housing structural variables and
neighborhood variables were consistent with prior expectations. The coefficient estimates
of lot size (Acreage), floor area (Area), basement area (Base), finished basement area
(FINBase), and garage area (Garage) were positive and statistically significant at the 5%
level, while the estimated coefficient of the age of the house (Age) was negative and
significant. For the neighborhood variables, the coefficient estimates of the percentage of
the minority population (Black, Hispanic) and the percentage of rental housing units
(Rent) were negative, while the household income (Income) and the level of education
(Education) were positive. Among the neighborhood variables were Income and
Education that were not statistically significant at the 10% level. The positive Sign of the
coefficient estimate of Industry suggests that the proximity to the industrial area was
negatively related to the housing value.

61

 

The coefficient estimates of Brownfield D , Brownfield D , and Brownfield D

shows the same pattern as those of Prudden. The estimated coefficient of Brownfield D

was positive and significant at the 5% level. Before redevelopment of Neogen, the
housing value in the surrounding neighborhood was expected to decrease by $1.81 as its

proximity to Neogen increased by one foot, all else constant. Both coefficient estimates
of Brownfield D and Brownfield A were positive and statistically significant at the 5%

level. Like Prudden, redevelopment of Neogen did not lead to a full rebound in property
values in the surrounding neighborhood, which can be explained by the fact that the
property still remained in industrial use after redevelopment. As the coefficient estimate
of Industry shows above, the industrial use of Neo gen still negatively affected property
values on the surrounding neighborhood.

Nevertheless, redevelopment of Neogen provided positive impacts on housing

values in the surrounding neighborhood even though it did not lead to a full recovery of

housing values as the coefficient estimates of Brownfield D and Brownfield A show. The
coefficient estimates of both Brownfield D and Brownfield D were significantly lower

than the coefficient estimate of Brownfield D . Like Prudden, the housing price was

affected greatly during the development period. The announcement of the State’s
financial assistance for the redevelopment project and the publicity of the project played
an important role in raising expectations of an increase in housing values resulting from
removing the negative extemality associated with the Neogen site. After redevelopment
of Neogen, the housing value was estimated to recover $0.238 or 13.1% out of the

housing value 1055 of $1 .81 per one foot from Neogen. However, a new separate hedonic

62

analysis shows that the difference between the coefficient estimates of Brownfield D and
Brownfield D was not statistically significant. The coefficient estimates of both

DifferenceD‘D and DifferenceDD were not statistically significant.

The Case of the Builders Plumbing and Heating Supply (Builderg

For the Builders Plumbing and Heating Supply case (Builders), a total of 6,543
housing sales observations entered the hedonic model. The mean distance between all
sales observations and Neogen was 5,523.3 ft and the average sales price was $81,515.86
as converted to 2007 US Dollars. Table IV-5 provides the estimated coefficients of 37
independent variables, which were used to explain the housing price located within 1.5
mile radius of Builders.

The Builders case is a little bit different from the other two redevelopment cases
considering the nature of development. The Builders case was an expansion project of the
existing use of the property, while the other two projects transformed abandoned sites
into the new site uses. It was expected that the difference in the nature of development of
Builders would make the pattern of the coefficient estimates of distance variables
different from those of the other two cases. The expansion of the existing light industrial
use might affect property values in the surrounding neighborhood more negatively while
the improvement of the Builders site might provide positive impacts on the surrounding

properties.

63

Table IV-5 Estimated Coefficients (BUILDERS)

 

 

 

 

 

Variables B Sig.

(Constant) 49410.21 5 .000*
Acre 46222.489 .000*
Age -292.180 .000*
Area 30.430 .000*
Base 12.790 .000*
FINBase 14.252 .000*
Garage 19.493 .000*
Update 13575.949 .000*
Black -159.759 .000*
Hispanic ~552.075 .000*
Rent -225.507 .000*
Income .103 .097**
Education 466.565 .000*
STY1.25 5067.550 .000*
S TY] . 5 6468.453 .000*
STY1.75 6188.108 .000*
STYZ 5141.655 .000*
dumSTY -5537.072 .137
alum_vinyl -716.206 .272
block_brick 4235 .545 .023*
pine_lap -1502.035 .789
dumexterior -5969. 166 .041 *
dum98 -30269.998 .000*
dum99 -23951.685 .000*
dum00 -22248.254 .000*
dumOI ~15204.307 .000*
dum02 -9041.400 .000*
dum03 -2724.464 .250

 

64

 

Table IV-5 (cont’d).

 

 

 

 

dum05 2263.631 .055**
dum06 4002.916 .176
dum07 -5819.920 .057**
Industry .534 .152
Brownfield D .991 .000*
BrownfieldD 1.756 .000*
Brownfield D .832 .030*
DiflerenceD'D .766 .042*
Diflerence’” -.159 .704
N=6543 R square = 0.592 F= 277.476

Excluded Variables: STYI, wood, dum04

 

 

* Statistically significant at the 5% level
** Statistically significant at the 10% level

 

First, as with the two other cases, the estimated coefficients of housing structural
variables and neighborhood variables were consistent with prior expectations. The signs
of the coefficient estimates of all structural and neighborhoods variables were consistent
with the other two cases. The coefficient estimates of the variables were all statistically
significant at the 5% level except Income variable. The coefficient estimate of Industry
variable was also consistent with the other two cases, suggesting the proximity to
industrial areas was negatively related to housing values.

The coefficient estimates of Brownfield D , Brownfield D , and Brownfield A
show the different pattern from those of Prudden and Neogen, as expected. The estimated

coefficient of Brownfield D was positive and significant at the 5% level. Before the

expansion of Builders, the housing value in the surrounding neighborhood was expected

65

 

to decrease by $ 0.991 as its proximity to Builders is increased by one foot, holding all
else constant. Both coefficient estimates of Brownfield D and Brownfield D were also
positive and statistically significant at the 5% level. Like the other two cases, the
expansion of Builders did not lead to a full rebound in property values in the surrounding
neighborhood. The lack of a full rebound can be explained by the fact that the property
still remained as an industrial use after expansion, which still negatively affected property

values on the surrounding neighborhood.

The coefficient estimate of Brownfield D was different from those of the two other
cases. The coefficient estimate of Brownfield D was higher than the coefficient estimate

of Brownfield D . This suggests that the expansion of the property for the existing use
contributed to more negative extemality during the development phase. The construction
phase of the project might also affect the surrounding housing values negatively.
However, the lower coefficient estimate of Brownfield D suggests that the improvement
of the Builders site after development provided positive impacts on the surrounding
properties. After the expansion of Builders, the housing value was estimated to recover
$0.159 or 16% out of the housing value loss of $ 0.991 per one foot from Builders.

However, a new separate hedonic analysis shows that the difference between the

coefficient estimates of Brownfield D and Brownfield D was not statistically significant.

As seen above, the coefficient estimate of Dijj’erenceD'D was not statistically significant,

B

while the coefficient estimate of DiflerenceD‘ was statistically significant at the 5%

level.

66

The Overall Impacts of the Three Brownfield Redevelopment Cases
Finally, this study combined all three groups of data in order to estimate the

overall impacts of the three brownfield redevelopment cases. A total of 8,458 sales

observations within 1.5 miles from all three sites between 1998 and 2007 were identified

for the analysis. Table IV-6 provides the estimated coefficients of 37 independent

variables, which were used to explain the housing price.

Table IV-6 Coefficient Estimates (All 3 Brownfield Sites)

 

 

 

 

 

Variables B Sig.

(Constant) 31406776 .000*
Acre 32006.452 .000*
Age -328.529 .000*
Area 32.840 .000*
Base 14.701 .000*
FINBase 14.948 .000"‘
Garage 19.354 .000*
Update 13644.708 .000*
Black -167.270 .000*
Hispanic -79.393 .237
Rent -141.045 .000"‘
Income .304 .000*
Education 582.281 .000*
STYI.25 3263.780 .000*
STYI. 5 5567.551 .000*
STYI. 75 6312.080 .000*
STYZ 5184.351 .000"‘
dumSTY -1852.983 .559
alum_vinyl -1621.470 .006*

 

67

 

Table IV-6 (cont’d).

 

 

 

 

block_brick 708.062 .688
pine_lap -3969.895 .497
dumexterior -5202.279 .063**
dum98 -35507.496 .000*
dum99 -30224.136 .000*
dum00 -28294.259 .000*
dumOI -19343.178 .000*
dum02 -13936.576 .000*
dum03 -7973.187 .000*
dum05 1619.622 .132
dum06 2164.144 .369
dum07 -7838.495 .002*
Industry 1.261 000*
Brownfield D 2.343 000*
Brownfield” 2.103 000*
Brownfield D 1.242 000*
DifferenceD‘D -.240 .452
DifferenceD'D -1 . 102 .003*
N=8458 R square = 0.623 F: 409.107

Excluded Variables: STYI, wood, dum04

 

 

* Statistically significant at the 5% level
** Statistically significant at the 10% level

 

The coefficient estimates of the structural and neighborhood variables were

consistent with prior expectations as well as all three separate analyses for individual

sites. The coefficient estimate of the Industry variable was also consistent with the

previous study as well as all three separate analyses. The positive sign of the coefficient

68

 

estimate of Industry suggests that the proximity to the industrial area was negatively
related to the housing value.

The coefficient estimates of Brownfield D , Brownfield D , and Brownfield D
support the hypothesized relationship between proximity to the brownfield sites and
housing values. The estimated coefficient of Brownfield D was positive and significant at
the 5% level, suggesting that before redevelopment of all three sites, the housing value in
the surrounding neighborhood was expected to decrease by $2.343 as its proximity to the
nearest brownfield increased by one foot, all else constant. Both coefficient estimates of
Brownfield D and Brownfield D were also positive and statistically significant at the 5%
level. As shown in the separate analyses, the three sites were still negatively affecting
housing values in the surrounding neighborhood. However, the coefficient estimates of
both variables were lower than the coefficient estimate of Brownfield D . Redevelopment
of these sites had positive impacts on housing values in the surrounding neighborhood.
After redevelopment of the three brownfield sites, the housing value was estimated to
recover $1.102 or 47.2% out of the housing value less of $2.343 per one foot from the
nearest brownfield site.

Unlike the previous hedonic analyses for individual brownfield sites, a new

separate hedonic analysis shows that the difference between the coefficient estimates of

Brownfield D and Brownfield D was statistically significant. As seen above, the coefficient

8

estimate of DiflerenceD‘ was statistically significant at the 5% level while the

coefficient estimate of DifferenceD‘D was not statistically significant. The empirical

finding supports the hypothesized relationship between the proximity to brownfield sites

69

and housing values in the surrounding neighborhood before and after redevelopment.
Redevelopment of brownfields provides positive impacts on the housing values in the
surrounding neighborhood by reducing the negative extemality associated with
brownfields. The following section will discuss the total benefits of brownfield

redevelopment.

4.3 Benefit Estimates of Redevelopment
The coefficient estimates of all variables are determinants of housing values as

shown in Equation (4).

T ‘ J K
price” 2 a + Zap, + Eels, + Zd,N,, + mIND, +

1:1 j=l k=1

,8, Brownfield D + flzBrownfield D + B3Brownfield D + e,

(4)

The hedonic price functions can be used to predict the expected price of a house
both before and after redevelopment of brownfield sites. Since the empirical findings
from the hedonic price analyses suggest that redevelopment of brownfields provides
positive impacts on housing values in the surrounding neighborhoods, the benefits of
redevelopment can be estimated by comparing the expected prices of a house before and

after redevelopment. The coefficient estimates of the hedonic function suggest that the

A
change ( A P,- ) in the price of the ith house can be estimated with the following equation:

70

AA

AP,=PD P. — I9:D:(,B3— ,Bl)Brownfield,

= (DifferenceD _ B)Brownfield, (6)

Since the coefficient estimates of all structural, neighborhood variables, dummy

variables, and the proximity to industrial areas were the same regardless of different

phases of redevelopment, the coefficient estimates (,61, (,63) of Brownfield D

and Brownfield D will determine the change in the housing value before and after
redevelopment of brownfield sites. The price change equals the difference in the
coefficient estimates for Brownfield D and Brownfield D , weighted by the distance of

house i to the brownfield site as shown in Equation (6). Based on Equation (6), the total

impact of brownfield redevelopment can be estimated with the following equation:

AP = (,63 — ,6, )Mean_Brownfield * Num 2

7
(DifferenceD‘D )Mean _ Brownfield * Num ( )

where Mean _ Brownfield is the mean distance between all houses and the brownfield

site and Num is the number of all single family housing within 1.5 mile radius of the

brownfield site. In Equation (7), A P represents the total estimation of the change in
property value 1055 in the neighborhood since the coefficient estimate of Brownfield D

represents the estimated decrease in housing prices as proximity to Prudden is increased

71

by one foot, all else constant. Table IV-7 summarizes the coefficient estimates of

DifferenceD'D for all four hedonic price functions in this study.

Table IV-7 The Coefficient Estimates of DiflerenceD"D

 

 

 

 

 

 

 

 

 

Prudden Neogen Builders All Sites
Variable B Sig B Sig B Sig B Sig
Brownfield D 1.816 000* 1.810 000* .991 000* 2.343 000*
Brownfield D 1.267 013* 1.572 005* .832 030* 1.242 000*
DiflerenceD'D -.549 .290 -.238 .686 -.159 .704 -1.102 003*

 

* Statistically significant at the 5% level
** Statistically significant at the 10% level

 

In this study, benefit estimates were derived for all houses within a 1.5 mile radius

of the brownfield site as the study was conducted under an assumption that the negative

extemality associated with brownfield sites is expected to affect the property values

within 1.5 miles from the brownfield sites.

Table IV-8 shows the number of properties affected by the existence or

redevelopment of brownfields and the mean distance between all properties and the

brownfield site for all four hedonic price functions of this study. The table also provides

the benefit estimations for redevelopment of individual brownfield sites and the

estimation of aggregate benefits resulting from redevelopment of all three brownfield

sites.

72

Table IV -8 Benefit Estimates of Redevelopment of Brownfield Sites

 

 

 

 

 

 

Variable Prudden Neogen Builders All Sites
Difference‘DD -.549 -.238 -.159 -1.102
(Sig) (.290) (.686) (.704) (.003)*
N of Properties 11,526 10,469 11,939 15,761
Mean Distance 5,256.291 5,377.224 5,438.146 4,949.304
Estimated Benefits 33,260,622 13,398,010 10,323,238 85,962,590

 

The benefit estimates of redevelopment of Prudden, Neogen, and Builders were
approximately $33,260,622, $13,398,010, and $10,323,238 respectively. The study may
compare the benefit estimates of brownfield redevelopment based on different
redevelopment scenarios since the three cases represent different redevelopment
scenarios such as a multi-family residential development for Prudden, renovation of an
old industrial building for Neogen, and expansion of a warehouse for Builders. According
to the study, the multi-family residential redevelopment generated the greatest benefit
while the expansion of the warehouse provided the least amount of benefit. However, this
comparison cannot be generalized due to the limited number of brownfield
redevelopment cases of the study.

In addition, the study could not conclude that the benefit estimates of these

individual brownfield redevelopment projects were statistically meaningful because the
coefficient of DiflerenceD‘D was not statistically significant for all individual brownfield
cases. However, the coefficient of DifferenceD’D for all sites was statistically significant

at the 5% level and thus the study was able to provide a meaningful estimation of the

overall benefits of the redevelopment of all three brownfield sites. The estimation of

73

aggregate indirect benefits of the three brownfield redevelopment projects was

approximately $85,962,590 in addition to the direct benefits of these projects.

4.4 Comparison with a Semi-Log Functional Analysis
The linear functional analysis shows that the estimation of indirect benefits of the

three brownfield redevelopment projects was substantial. However, according to Figure

III-7, the natural log of price (LnPriceit) was also appeared to be linearly related to the

distance variable, a log-linear function could be also used for assessing the impacts of
brownfield redevelopment. The log-linear functional form has some advantages over the
linear functional form as explained in Chapter 111.

Because of the advantages from using the log-linear function, this study
conducted an additional log-linear functional analysis to compare the overall benefit
estimate of the three brownfield redevelopment cases with the result of the linear

functional analysis. For the log-linear functional analysis, all housing sales prices were

converted to natural log values (LnPriceit). The new hedonic price analysis used the

following log-linear functional model:

T J K
Ln Price, = a + 2150+ Eels,j + deN, + eIND, +
k=l

t=l j=l

. (5)
,8, Brownfield D + BZBrownfield D + ,8, Brownfield D + e1,

74

The coefficient estimatesnof the structural and neighborhood variables were

consistent with the results of the linear functional analysis, showing the same signs of the

independent variables. Table IV—9 presents the summary results of the log-linear model in

comparison with the results of the linear functional model.

Table IV-9 Results of the Linear Functional & the Log-Linear Functional Analyses

 

Log-Linear Functional Analysis

Linear Functional Analysis

 

 

 

 

 

 

 

 

 

 

 

F=317.055 F= 409.107
R square = 0.554 R square = 0.623

Variable B B B Sig
Brownfield D 4.03 lE-5 .000* 2.343 _000*
Brownfield D 2.017E-5 .000* 2.103 .0004:
Brownfield D 1.856E-5 .000* 1.242 .0004:
DijferenceD'D -2.01‘5E-5 .000* -.240 .452
DiflerenceD‘D -2.176E-5 .000* -1 . 102 .003*
DfierenoeD‘D -2.176E-5 -1.102
Transformation -1.000022

N of Properties 15,761 15,761
Mean Distance 4,949.304 4,949.304
Estimated Benefits $78,007,696 $85,962,590

 

 

 

* Statistically significant at the 5% level
** Statistically significant at the 10% level

 

The coefficient estimates of Brownfield D , Brownfield D , and Brownfield D in the

log-linear model had the expected signs and are statistically significant at the 1% level.

The coefficient estimate of Brownfield D was lower than the coefficient estimate of

 

17 For the result of the log-linear functional analysis, see Appendix E.

75

Brownfield D , indicating that redevelopment of the three brownfield sites had positive
impacts on housing values in the surrounding neighborhood. The difference
(DiflerenceD‘D ) between the coefficient estimates of Brownfield D and Brownfield D

was statistically significant at the 1% level.

In order to determine the benefits of the redevelopment of three brownfields, the

B

natural log value of the coefficient estimate of Di'fierenceD‘ was converted to the

original normal value using the following procedure.

ln(x) = 000002176

eO'DOOOZ‘ 76 = 1.000022

The natural value of the coefficient estimate of DiflerenceD’D in the log-linear

model was slightly lower than the coefficient estimate of the same variable in the linear
functional model, resulting in a slightly lower estimated benefit of redevelopment. The
benefit estimated by the log-linear model was $78,007,696. Even though the estimated
benefit was a bit different, the results of the log-linear model analysis were almost similar
to the results of the linear model analysis. The values of R square of both functional
analyses show that the linear model better explained the variation in housing values than

the log~linear model.

76

V. CONCLUSION

The main purpose of this study was to assess the indirect benefits of brownfield
redevelopment by using the hedonic price method. The hedonic price method has been
widely used to measure environmental goods and services that cannot be assessed in the
existing market system. The hedonic price method uses the changes in complementary
goods such as housing to assess a willingness to pay for an improved environment. The
existing literature measured the negative impacts of brownfields on property values in the
surrounding neighborhood as an attempt to measure the benefits of brownfield
redevelopment. Many studies hypothesized that the existence of brownfields could be
considered disamenity factors to the surrounding properties. These studies revealed that
the proximity to brownfields is negatively related to nearby property values. The previous
studies further assumed that the negative extemality associated with brownfields would
be eliminated after cleanup or redevelopment of brownfields, concluding that the
negative impacts of the existence of brownfields on the surrounding property values
would become the expected benefits of cleanup or redevelopment of brownfields.
However, the existing literature does not provide an agreement on whether the cleanup or
redevelopment of brownfields can fully or substantially recover the loss of property

values in the surrounding neighborhood.

5.1 Summary of Findings

This study attempted to analyze the impacts of brownfield redevelopment on the
surrounding properties, using three brownfield redevelopment cases in Lansing,
Michigan. The study constructed hedonic functions to explain housing values with

77

housing structural variables, neighborhood variables, and two environmental variables.
One environmental variable measured the negative extemality associated with proximity
to industrial areas. The main environmental variable was the distance between housing
observations and brownfield sites. The distance variable was constructed to help measure
the impacts of brownfields on the surrounding property values before, during, and after
redevelopment.

The empirical findings of the hedonic price analysis confirmed findings from the
existing literature arguing that the proximity to brownfield sites or environmental hazards
is negatively related to the property values in nearby neighborhoods. Before the
redevelopment of brownfields, the proximity to the brownfield sites and housing prices
were negatively related to each other. The negative relationship between the proximity
and housing prices suggests that there was a substantial loss in property values in the
surrounding neighborhood due to the negative extemality associated with brownfields.
Redevelopment of brownfields was expected to eliminate such negative extemality and
lead to recovering the loss of property values.

This study was able to analyze the effects of brownfield redevelopment on the
surrounding property values. The empirical findings of the analyses support the
hypothesized relationship between the proximity to brownfield sites and housing values
in the surrounding neighborhood before and after redevelopment. Redevelopment of
brownfields was expected to provide positive impacts on the housing values in the
surrounding neighborhood by eliminating or substantially reducing the negative
extemality associated with brownfields. Even though the difference between the

coefficient estimates of the distance variables before and after redevelopment were not

78

statistically significant for all individual brownfield redevelopment projects, the
coefficient estimate of the distance variable after redevelopment was consistently lower
than the coefficient estimate of the distance variable before redevelopment. The lower
coefficient suggests that the redevelopment of individual brownfields may have positive
impacts on the nearby surrounding property values, but the analyses could not conclude
that it is statistically meaningfirl.

The study was also able to compare the benefit estimates of brownfield
redevelopment based on different redevelopment scenarios. The three cases represent
different redevelopment scenarios such as a multi-family residential development,
renovation of an old industrial building, and expansion of a warehouse. The empirical
findings of the study suggest that the multi-family residential redevelopment generated
the greatest benefit while the expansion of the warehouse provided the least amount of
benefit. However, this comparison cannot be generalized due to the limited number of
brownfield redevelopment cases in the study.

Finally, the final analysis that estimated the overall impact of three brownfield
redevelopment projects provided statistically meaningfirl results on the impacts of
brownfield redevelopment on the surrounding property values. The analysis showed that
the difference between the coefficient estimates of the distance variables before and after
redevelopment were statistically significant at the 5% level for combined brownfield
redevelopment projects. It was estimated that redevelopment of the three brownfield sites
provided a total benefit of $85,962,590 ($78,007,696 if the log-linear function was
applied) in addition to the direct benefit of the projects such as the number of jobs created

or an increase in the tax base after redevelopment.

79

5.2 Policy Implications of the Study

Brownfield redevelopment has been an important policy for the federal and state
governments. Due to several barriers associated with cleanup or reuse of brownfields
such as liability and additional costs, aggressive public financing and other technical
assistance are necessary to facilitate brownfield redevelopment. Federal and state
governments have responded very actively to these challenges by providing a variety of
financial and technical assistance programs to promote cleanup and redevelopment of
brownfields. Due to support from governments, a large number of brownfield sites have
been converted to productive uses throughout the US. Many studies reported economic
benefits of brownfield cleanup and redevelopment by using a set of indicators such as the
number of jobs created, increased income, increased tax revenues generated by newly
redeveloped sites, etc. However, these studies did not successfully report all the possible
benefits of brownfield redevelopment because some environmental and social benefits
were not easy to quantify.

The study analyzed impacts of brownfield redevelopment on the price of
surrounding properties in order to quantify the indirect benefits of brownfield
redevelopment using three brownfield redevelopment cases in Lansing, Michigan.
Redevelopment of brownfields was expected to provide positive impacts on the housing
values in the surrounding neighborhood by eliminating or substantially reducing the
negative extemality associated with brownfields.

As this study confirms, the cleanup or redevelopment of brownfields provides

indirect benefits that cannot be measured easily in the current market system. However,

80

this kind of indirect benefit of brownfield redevelopment has not been an important
consideration in the public decision-making framework. Without acknowledgement of the
full potential benefits and costs associated with brownfield redevelopment, it is not
possible to make an efficient allocation of resources in the public policy area. The
empirical findings of this study can justify public financial assistance for the cleanup or
redevelopment of brownfields. The findings can also help government officials prioritize
competing projects to use public resources more efficiently. This study suggests that the
estimated hedonic function can be used to determine the benefits of brownfield cleanup

or redevelopment in the decision-making process.

5.3 Limitations and Recommendations

Even with its meaningful empirical findings, this study contained certain
limitations. First, since this study used a small number of brownfield redevelopment
cases in a single geographical location, its empirical findings cannot be generalized and
can only represent the impacts of three brownfield redevelopment projects in Lansing, MI.
In addition, even though the study compared the benefits of three brownfield
redevelopment cases based on their different redevelopment scenarios, the comparison
should not be generalized due to the limited number of brownfield redevelopment cases
in the study. The limitations of the study suggest that future researchers might consider
analyzing the indirect economic impacts of brownfield redevelopment using a large
number of redeveloped brownfields in various geographical locations.

Second, even though the values of R-square of the hedonic price analyses of this

study were higher than those of other hedonic studies in general, the values were still a

81

bit lower to explain the variations in housing prices. One of the reasons might be that the
modified housing transaction data still contained “non-arms length” sales even though the
study undertook several procedures to eliminate these cases. This suggests that more
accurate housing market data is necessary for improving the ability to explain variations
in values of a dependent variable.

Third, the disparity between the numbers of housing sales observations before and
after redevelopment might affect the level of statistical significance of the estimated
benefits after redevelopment of individual brownfield redevelopment projects. If time
permits, balancing the number of sales observations before and after redevelopment will
improve the level of statistical significance.

While this study is generally consistent with findings form other hedonic price
studies on the impact of amenities and disamenities, the findings of this study also need
to be compared to other non-hedonic method approaches to value one’s willingness to

pay for brownfield redevelopment projects.

82

APPENDICES

APPENDIX A Regression Analysis for PRUDDEN

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Descriptive Statistics

N Minimum Maximum Mean Std. Deviation
Price 5933 6000 500282 79975.38 38493.551
Acre 5933 .028 1 .1 30 .1 3243 .066860
Age 5933 1 1 52 80.27 20.957
Area 5933 400 5082 11 70.1 3 433.758
Base 5933 0 2307 674.61 247.452
FINBase 5933 O 1457 46.41 148.468
Garage 5933 0 1660 217.99 201 .272
Update 5933 0 1 .02 .136
Black 5933 3.9 66.7 19.085 12.6971
Hispanic 5933 4.3 33.8 13.952 5.9145
Rent 5933 13.70 86.82 38.6999 16.01126
Income 5933 10323 57768 31724.32 8641 .046
Education 5933 8.7 58.3 27.427 13.5054
Poverty 5933 4.51 53.77 22.5444 11 .06829
STY1 5933 0 1 .35 .477
STY1.25 5933 0 1 .10 .298
STY1 .5 5933 0 1 .07 .259
STY1.75 5933 0 1 .11 .311
STY2 5933 0 1 .36 .481
dumSTY 5933 0 1 .01 .088
alum_vinyl 5933 0 1 .21 .410
block_brick 5933 0 1 .01 .111
pine_lap 5933 0 1 .00 .026
wood 5933 0 1 .76 .425
dumexterior 5933 0 1 .01 .092
dum98 5933 0 1 .09 .288
dum99 5933 0 1 .1 O .300
dim00 5933 0 1 .10 .297
dum01 5933 O 1 .10 .299

 

 

 

 

 

 

00
b)

 

Descriptive Statistics (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

dum02 5933 0 1 .11 .319
dum03 5933 0 1 .11 .317
dum04 5933 0 1 .12 .330
dum05 5933 O 1 .11 .308
dum06 5933 0 1 .10 .293
dum07 5933 0 1 .06 .234
Industry 5933 .00 4983.83 1521 .8504 979.98223
Dis_PRU_B 5933 0 7920 3339.74 2970.693
Dis_PRU_D 5933 0 791 8 1227 .04 2406.490
Dis_PRU_A 5933 O 7918 827.09 2066.716
PRU 5933 291 .71 7919.65 5393.8738 1770.47550
Valid N (Iistwise) 5933
Model Summary
Adjusted R Std. Error of the

Model R R Square Square Estimate

1 .768 .589 .587 24743.476

 

 

 

 

 

a. Predictors: (Constant), Dis_PRU_A, Education, STY1.5, block_brick, pine_lap, dumSTY,
Update, dumexterior, Acre, STY1.25, dim00, STY1.75, FINBase, dum98, alum_vinyl, dum01,
dum99, Rent, Garage, dum03, Base, dum05, Black, Industry, dum07, Age, dum02, STY1,
Hispanic, Area, Income, Dis_PRU_D, Dis_PRU_B, dum06

 

 

 

 

 

ANOVA
Model Sum of Squares df Mean Square F
Regression 5.179E12 34 1 .523E11 248.786
1 Residual 3.611 E1 2 5898 6.122E8
Total 8.790E12 5932

 

 

 

 

 

 

a. Predictors: (Constant), Dis_PRU_A, Education, STY1.5, block_brick, pine_lap, dumSTY,
Update, dumexterior, Acre, STY1.25, dimOO, STY1.75, FINBase, dum98, alum_vinyl, dum01,
dum99, Rent, Garage, dum03, Base, dum05, Black, Industry, dum07, Age, dum02, STY1,
Hispanic, Area, Income, Dis_PRU_D, Dis_PRU_B, dum06

b. Dependent Variable: Price

84

 

 

 

Coefficients

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Standardized
Unstandardized Coefficients Coefficients

Model B Std. Error Beta t Sig.

1 (Constant) 45798.555 4887.076 9.371 .000
Acre 34842.593 5386.01 9 .061 6.469 .000
Age -328.242 20.537 -.179 -15.983 .000
Area 32.786 1 .227 .369 26.727 .000
Base 13.930 1 .632 .090 8.538 .000
FINBase 18.464 2.313 .071 7.983 .000
Garage 19.190 1.728 .100 11.107 .000
Update 12853.072 2419.895 .045 5.311 .000
Black -1 67.995 34.313 -.055 -4.896 .000
Hispanic -55.445 85.410 -.009 -.649 .516
Rent -249.361 36.894 -.104 -6.759 .000
Income .209 .068 .047 3.068 .002
Education 631 .608 43.199 .222 14.621 .000
STY1 -5734.537 1037.460 -.071 -5.527 .000
STY1 .25 -2948.795 1343.612 -.023 -2.195 .028
STY1.5 -132.550 1379.547 .000 -.096 .923
STY1 .75 322.825 1166.649 .003 .277 .782
dumSTY -6263.894 3775.494 -.014 -1 .659 .097
alum__vinyl -1079.134 814.832 -.012 -1.324 .185
block_brick -1606.380 2933.253 -.005 -.548 .584
pine_lap 7992.735 12401 .171 .005 .645 .519
dumexterior -6408.754 3510.007 -.015 -1.826 .068
dum98 -39254.313 2717.984 -.294 -14.442 .000
dum99 -34193.550 2727.301 -.266 —12.538 .000
dimOO -32192.123 2717.207 -.249 -11.848 .000
dum01 -21870.819 2727.976 -.170 -8.017 .000
dum02 -18041 .544 2689.644 -.150 -6.708 .000
dum03 -11094.049 2690.129 -.091 -4.124 .000
dum05 2064.431 1361.559 .017 1.516 .130
dum06 -2229.173 3457.755 -.017 -.645 .519
dum07 -13539.781 3595.244 -.082 -3.766 .000

 

85

 

Coefficients @nt’d).

 

 

 

 

 

 

 

 

 

 

 

Industry 2.142 .411 .055 5.207 .000
Dis_PRU_B 1 .816 .310 .140 5.867 .000
Dis_PRU_D 1.019 .425 .064 2.399 .016
Dis_PRU_A 1.257 .509 .068 2.487 .013

 

a. Dependent Variable: Price

Excluded Variables

 

Collinearity Statistics

 

 

 

 

 

Model Beta In t Sig. Partial Correlation Tolerance
STY2 .3 .000
wood .a .000
dum04 a .000

 

 

 

 

 

 

 

a. Predictors in the Model: (Constant), Dis_PRU_A, Education, STY1.5, block_brick, pine_lap,

dumSTY, Update, dumexterior, Acre, STY1.25, dim00, STY1.75, FINBase, dum98, alum_vinyl,
dum01, dum99, Rent, Garage, dum03, Base, dum05, Black, Industry, dum07, Age, dum02, STY1,
Hispanic, Area, Income, Dis_PRU_D, Dis_PRU_B, dum06

b. Dependent Variable: Price

86

 

 

APPENDIX B

Regression Analysis for NEOGEN

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Descriptive Statistics

N Minimum Maximum Mean Std. Deviation
Price 5345 6000 500282 77929.60 37604.1 38
Acre 5345 .028 1 .1 30 .12579 .058433
Age 5345 1 152 82.94 19.839
Area 5345 400 5082 1188.27 436.151
Base 5345 0 2307 680.21 230.625
F lNBase 5345 0 1338 41.22 134.016
Garage 5345 0 1660 203.90 198.312
Update 5345 O 1 .02 .142
Black 5345 3.9 66.7 18.962 12.2916 '
Hispanic 5345 3.8 33.8 13.708 6.0406
Rent 5345 4.57 86.82 40.7530 15.90551
Income 5345 1 0323 59904 31 160.54 8063.546
Education 5345 7.3 69.2 27.564 13.2909
Poverty 5345 1 .25 53.77 23.2125 10.53755
STY1 5345 0 1 .32 .466
STY1 .25 5345 0 1 .09 .290
STY1 .5 5345 O 1 .07 .263
STY1 .75 5345 0 1 .12 .323
STY2 5345 0 1 .39 .488
dumSTY 5345 0 1 .01 .086
alum_vinyl 5345 0 1 .20 .398
block_brick 5345 0 1 .01 .112
pine_lap 5345 0 1 .00 .027
wood 5345 O 1 .78 .414
dumexterior 5345 0 1 .01 .094
dum98 5345 0 1 .09 .291
dum99 5345 0 1 .10 .301
dimOO, 5345 0 1 .10 .293
dum01 5345 0 1 .10 .295
dum02 5345 0 1 .12 .321

 

 

 

 

 

 

oo
\]

 

Descriptive Statistics (cont'd)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

dum03 5345 0 1 .11 .318
dum04 5345 0 1 .13 .332
dum05 5345 0 1 .11 .309
dum06 5345 0 1 .09 .291
dum07 5345 0 1 .06 .232
Industry 5345 .00 4983.83 1521.9071 983.16416
Dis_N EO_B 5345 0 7917 3345.03 2920.405
Dis_N EO_D 5345 0 7918 1272 .09 2439.069
Dis_N EO_A 5345 0 7920 814.71 2037 .276
NEO 5345 96.86 7919.56 5431 .8245 1609.90282
Valid N (Iistwise) 5345
Model Summary
Adjusted R Std. Error of the

Model R R Square Square Estimate

1 .762 .581 .578 24426.885

 

 

 

 

 

a. Predictors: (Constant), Dis_NEO_A, Base, STY1.5, dumexterior, pine_lap, dumSTY, Rent,
Update, block_brick, alum_vinyl, dimOO, Hispanic, STY1.75, dum98, dum01, Acre, STY1.25,
dum99, FINBase, Garage, dum03, Industry, Black, dum05, STY1, dum07, Age, dum02, Income,
Area, Education, Dis_NEO_B, Dis_NEO_D, dum06

 

 

 

 

 

ANOVA
Model Sum of guares df Mean Square F
Regression 4.388E12 34 1.291 E11 216.320
1 Residual 3.168E12 5310 5.967E8
Total 7.557E12 5344

 

 

 

 

 

 

a. Predictors: (Constant), Dis_NEO_A, Base, STY1.5, dumexterior, pine_lap, dumSTY, Rent,
Update, block_brick, alum_vinyl, dim00, Hispanic, STY1.75, dum98, dum01, Acre, STY1.25,
dum99, FINBase, Garage, dum03, Industry, Black, dum05, STY1, dum07, Age, dum02, Income,
Area, Education, Dis_NEO_B, Dis_NEO_D, dum06

b. Dependent Variable: Price

88

 

 

 

Coefficients

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Standardized
Unstandardized Coefficients Coefficients

Model B Std. Error Beta t Sig.

1 (Constant) 45255.13? 4825.443 9.378 .000
Acre 48531 .874 6304.648 .075 7.698 .000
Age -305.497 21 .834 -.161 -13.992 .000
Area 30.381 1.281 .352 23.716 .000
Base 13.615 1.857 .083 7.333 .000
FINBase 19.150 2.678 .068 7.151 .000
Garage 17.666 1 .812 .093 9.751 .000
Update 12657.339 2420.772 .048 5.229 .000
Black -240.778 36.049 -.079 -6.679 .000
Hispanic -40.023 80.678 -.006 -.496 .620
Rent -232.876 36.760 -.098 -6.335 .000
Income .112 .071 .024 1.567 .117
Education 665.052 44.092 .235 15.083 .000
STY1 -6847.707 1066.209 -.O85 -6.422 .000
STY1.25 -2122.386 1414.632 -.016 -1.500 .134
STY1 .5 895.241 1407.488 .006 .636 .525
STY1 .75 504.693 1164.025 .004 .434 .665
dumSTY -2035.365 3996.636 -.005 -.509 .611
alum_vinyl -2021 .958 855.839 -.021 -2.363 .018
block_brick 1133.498 3041 .977 .003 .373 .709
pine_lap 12185.087 12254.827 .009 .994 .320
dumexterior -7891 .549 3571 .319 -.020 -2.210 .027
dum98 -37565.647 3091.265 -.291 -12.152 .000
dum99 -32030.141 3117.633 -.256 -10.274 .000
dimOO -30512.982 3094.900 —.238 -9.859 .000
dum01 -21274.617 3103.843 -.167 -6.854 .000
dum02 -16658.057 3062.360 -.142 -5.440 .000
dum03 -9343.866 3062.411 -.079 -3.051 .002
dum05 3298.492 1412.522 .027 2.335 .020
dum06 -1367.242 3906.718 -.011 -.350 .726
dum07 -12052.347 4072.869 -.074 -2.959 .003

 

89

 

Coefficients (cont’d).

 

 

 

 

 

 

 

 

 

 

 

Industry 2.415 .413 .063 5.853 .000
Dis_NEO_B 1.810 .320 .141 5.649 .000
Dis_NEO_D 1 .211 .460 .079 2.633 .008
Dis_NEO_A 1 .572 .556 .085 2.830 .005

 

a. Dependent Variable: Price

Excluded Variables

 

 

 

 

 

 

 

Collinearity Statistics

Model Beta In t Sig Partial Correlation Tolerance
STY2 .3 . . ' . .000
wood .a . . . .000
dum04 a .000

 

 

 

 

 

 

 

a. Predictors in the Model: (Constant), Dis_NEO_A, Base, STY1.5, dumexterior, pine_lap,
dumSTY, Rent, Update, block_brick, alum_vinyl, dimOO, Hispanic, STY1.75, dum98, dum01, Acre,
STY1.25, dum99, FINBase, Garage, dum03, Industry, Black, dum05, STY1, dum07, Age, dum02,
Income, Area, Education, Dis_NEO_B, Dis_NEO_D, dum06

b. Dependent Variable: Price

90

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

APPENDIX C Regression Analysis for BUILDERS
Descriptive Statistics

N Minimum Maximum Mean Std. Deviation
Price 6543 6000 389888 8151 5.86 34809.996
Acre 6543 .028 .997 .12415 .052845
Age 6543 1 152 81 .12 19.396
Area 6543 400 5082 1172.27 416.1 52
Base 6543 O 2307 688.94 224.227
FINBase 6543 0 1338 53.77 149.723
Garage 6543 0 1660 212.75 193.592
Update 6543 O 1 .02 .141
Black 6543 3.5 66.7 16.834 12.5167
Hispanic 6543 1.0 33.8 11.620 5.3977
Rent 6543 4.57 86.82 36.6906 18.13642
Income 6543 10323 60063 33260.09 8526.255
Education 6543 7.3 69.2 30.362 14.2917
Poverty 6543 1 .25 53.77 19.7057 10.26321
STY1 6543 0 1 .36 .480
STY1.25 6543 0 1 .11 .310
STY1 .5 6543 0 1 .08 .264
STY1.75 6543 0 1 .11 .307
STY2 6543 O 1 .35 .476
dumSTY 6543 0 1 .01 .078
alum_vinyl 6543 0 1 .30 .456
block_brick 6543 0 1 .02 .1 55
pine_lap 6543 0 1 .00 .049
wood 6543 0 1 .67 .471
dumexterior 6543 0 1 .01 .095
dum98 6543 0 1 .09 .291
dum99 6543 0 1 .10 .300
dimOO 6543 0 1 .10 .300
dum01 6543 O 1 .10 .306
dum02 6543 O 1 .12 .320

 

 

 

 

 

 

\O
H

 

Descriptive Statistics (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

dum03 6543 0 1 .11 .317
dum04 6543 0 1 .12 .327
dum05 6543 0 1 .10 .305
dum06 6543 0 1 .09 .292
dum07 6543 0 1 .05 .224
Industry 6543 .00 4279.95 1326.1126 899.77478
Dis_BUl_B 6543 .00 7915.88 3464.7455 3035.15519
Dis_BUl_D 6543 0 7915 1257.80 2481 .261
Dis_BUl_A 6543 O 7909 800.76 2064.967
BUI 6543 103.74 7915.88 5523.3038 1830.64822
Valid N (Iistwise) 6543
Model Summag

Adjusted R Std. Error of the
Model R R Square Sflare Estimate
1 .769 .592 .590 22299.002

 

 

 

 

 

a. Predictors: (Constant), Dis_BUl_A, pine_lap, dumSTY, FINBase, dumexterior, STY1.5, Industry,
Update, STY1.75, dum98, block_brick, Garage, dum99, dim00, alum_vinyl, STY1.25, dum01,
Acre, Black, Base, dum03, Hispanic, dum05, dum07, STY2, Age, dum02, Income, Area,
Education, Rent, Dis_BUl_B, Dis_BUl_D, dum06

 

 

 

 

 

 

 

ANOVA
Model Sum of Squares df Mean Square F
Regression 4.691 E12 34 1 .380E11 277.476
1 Residual 3.236E12 6508 4.972E8
Total 7.927E12 6542

 

 

 

 

 

 

a. Predictors: (Constant), Dis_BUl_A, pine_lap, dumSTY, FINBase, dumexterior, STY1.5, Industry,
Update, STY1.75, dum98, block_brick, Garage, dum99, dim00, alum_vinyl, STY1.25, dum01,
Acre, Black, Base, dum03, Hispanic, dum05, dum07, STY2, Age, dum02, Income, Area,
Education, Rent, Dis_BUl_B, Dis_BUl_D, dum06

b. Dependent Variable: Price

92

 

Coefficients

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Standardized
Unstandardized Coefficients Coefficients

Model B Std. Error Beta t Sig.

1 (Constant) 49410.215 3557.094 13.891 .000
Acre 46222.489 5710.872 .070 8.094 .000
Age -292.180 19.266 -.163 -15.165 .000
Area 30.430 1 .058 .364 28.749 .000
Base 12.790 1 .552 .082 8.243 .000
FINBase 14.252 2.036 .061 7.000 .000
Garage 19.493 1 .536 .108 12.691 .000
Update 13575.949 2011.039 .055 6.751 .000
Black -159.759 30.691 -.057 -5.205 .000
Hispanic -552.075 72.423 -.086 -7.623 .000
Rent -225.507 31.234 -.117 -7.220 .000
Income .103 .062 .025 1 .662 .097
Education 466.565 35.771 .192 13.043 .000
STY1 .25 5067.550 983.263 .045 5.154 .000
STY1.5 6468.453 1127.945 .049 5.735 .000
STY1 .75 6188.108 1050.981 .055 5.888 .000
STY2 5141 .655 864.702 .070 5.946 .000
dumSTY -5537.072 3718.662 -.012 -1 .489 .137
alum_viny| -716.206 651.757 -.009 -1.099 .272
block_brick 4235.545 1869.107 .019 2.266 .023
pine_lap -1 502.035 5602.526 -.002 -.268 .789
dumexterior -5969.166 2916.600 -.016 -2.047 .041
dum98 -30269.998 2397.321 -.253 -12.627 .000
dum99 -23951 .685 2404.249 -.206 -9.962 .000
dimOO -22248.254 2400.681 -.192 -9.267 .000
dum01 -15204.307 2393.345 -.134 -6.353 .000
dum02 -9041 .400 2367.894 -.083 -3.818 .000
dum03 -2724.464 2366.936 -.025 -1.151 .250
dum05 2263.631 1177.146 .020 1.923 .055
dum06 4002.916 2957.392 .034 1.354 .176

 

 

93

 

Coefficients (cont’d).

 

 

 

 

 

 

 

 

 

 

 

 

dum07 -5819.920 3061 .496 -037 -1 .901 .057
industry .534 .373 .014 1.431 .152
Dis_BUl_B .991 .210 .086 4.717 .000
Dis_BUl_D 1.756 .334 .125 5.255 .000
Dis_BUl_A .832 .382 .049 2.175 .030

 

a. Dependent Variable: Price

Excluded Variables

 

 

 

 

 

 

CollinearityStatistics

Model Beta In t Sig. Partial Correlation Tolerance
STY1 .3 .000
wood .a .000
dum04 a .000

 

 

 

 

 

 

 

a. Predictors in the Model: (Constant), Dis_BUl_A, pine_lap, dumSTY, FINBase, dumexterior,

STY1.5, Industry, Update, STY1.75, dum98, block_brick, Garage, dum99, dim00, alum_vinyl,
STY1.25, dum01, Acre, Black, Base, dum03, Hispanic, dum05, dum07, STY2, Age, dum02,

Income, Area, Education, Rent, Dis_BUl_B, Dis_BUl_D, dum06

b. Dependent Variable: Price

94

 

 

APPENDIX D

Regression Analysis for ALL SITES

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Descriptive Statistics

N Minimum Maximum Mean Std. Deviation
Price 8458 6000 500282 8291 2.56 37848.798
Acre 8458 .028 1 .130 .13165 .062010
Age 8458 1 152 78.10 20.204
Area 8458 400 5082 1147.67 412.148
Base 8458 0 2307 682.34 241 .1 77
FINBase 8458 0 1457 58.77 159.847
Garage 8458 0 1660 223.65 196.523
Update 8458 0 1 .02 .136
Black 8458 3.5 66.7 17.372 12.8498
Hispanic 8458 1.0 33.8 12.476 5.6853
Rent 8458 4.57 86.82 34.7491 1 7.06897
Income 8458 10323 60063 33423.02 9211 .719
Education 8458 7.3 69.2 29.219 14.8873
Poverty 8458 1.25 53.77 20.1501 11.27779
STY1 8458 0 1 .39 .488
STY1.25 8458 0 1 .12 .319
STY1 .5 8458 0 1 .08 .264
STY1.75 8458 0 1 .10 .294
STY2 8458 0 1 .31 .465
dumSTY 8458 O 1 .01 .083
alum_vinyl 8458 0 1 .30 .457
block_brick 8458 0 1 .02 .149
pine_lap 8458 0 1 .00 .043
wood 8458 0 1 .67 .471
dumexterior 8458 0 1 .01 .091
dum98 8458 0 1 .09 .287
dum99 8458 0 1 .10 .300
dimOO 8458 0 1 .10 .300
dum01 8458 0 1 .10 .306
dum02 8458 0 1 .11 .319

 

 

 

 

 

 

\O
Ur

 

Descriptive Statistics (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

dum03 8458 0 1 .11 .317
dum04 8458 0 1 .12 .328
dum05 8458 0 1 .11 .307
dum06 8458 0 1 .09 .291
dum07 8458 0 1 .06 .228
Industry 8458 .00 4983.83 1326.9808 918.43984
Dis_BFs_B 8458 .00 7919.65 3189.9175 2914.87682
Dis_BFs_D 8458 0 7918 1161.17 2329.618
Dis_BFs_A 8458 O 7918 751 .13 1957.574
BFs 8458 96.86 7919.65 5102.2176 1951 .80243
Valid N (Iistwise) 3453
Model Summary
Adjusted R Std. Error of the

Model R R Square Square Estimate

1 .789 .623 .621 23291 .119

 

 

 

 

 

a. Predictors: (Constant), Dis_BFs_A, pine_lap, Base, dumexterior, STY1.5, Black, dumSTY,
Update, dum98, STY1.75, Industry, block_brick, dum99, Acre, dim00, STY1.25, dum01,
alum_vinyl, FINBase, Garage, dum03, Hispanic, dum05, Age, dum07, STY2, dum02, Income,
Area, Education, Rent, Dis_BFs_D, Dis_BFs_B, dum06

 

 

 

 

 

ANOVA
Model Sum of Squares df Mean Square F
Regression 7.546E12 34 2.219E11 409.107
1 Residual 4.569E12 8423 5.425E8
Total 1 .211 E13 8457

 

 

 

 

 

 

a. Predictors: (Constant), Dis_BFs_A, pine_lap, Base, dumexterior, STY1.5, Black, dumSTY,
Update, dum98, STY1.75, Industry, block_brick, dum99, Acre, dim00, STY1.25, dum01,
alum_vinyl, FINBase, Garage, dum03, Hispanic, dum05, Age, dum07, STY2, dum02, Income,
Area, Education, Rent, Dis_BFs_D, Dis_BFs_B, dum06

b. Dependent Variable: Price

96

 

 

 

Coefficients

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Standardized
Unstandardized Coefficients Coefficients

Model B Std. Error Beta t Sig.

1 (Constant) 31406.776 3458.184 9.082 .000
Acre 32006.452 4523.405 .052 7.076 .000
Age -328.529 16.706 -.175 -19.665 .000
Area 32.840 .973 .358 33.752 .000
Base 14.701 1.304 .094 11.271 .000
FINBase 14.948 1 .733 .063 8.624 .000
Garage 19.354 1.403 .100 13.791 .000
Update 13644.708 1904.567 .049 7.164 .000
Black -167.270 27.512 -.057 -6.080 .000
Hispanic -79.393 67.092 -.012 -1.183 .237
Rent -141.045 29.924 -.064 -4.713 .000
Income .304 .050 .074 6.050 .000
Education 582.281 30.344 .229 19.190 .000
STY1 .25 3263.780 866.950 .028 3.765 .000
STY1 .5 5567.551 1035.629 .039 5.376 .000
STY1 .75 6312.080 995.026 .049 6.344 .000
STY2 5184.351 808.445 .064 6.413 .000
dumSTY -1852.983 3174.548 -.004 -.584 .559
alum_vinyl -1621.470 593.363 -.020 -2.733 .006
block_brick 708.062 1766.149 .003 .401 .688
pine_lap -3969.895 5846.635 -.005 -.679 .497
dumexterior -5202.279 2797.174 -.013 -1.860 .063
dum98 -35507.496 1954.686 -.269 -18.165 .000
dum99 -30224.136 1954.297 -.239 -15.465 .000
dim00 -28294.259 1 953.785 -.224 -14.482 .000
dum01 -19343.178 1950.914 -.157 -9.915 .000
dum02 -13936.576 1924.220 -.117 -7.243 .000
dum03 -7973.187 1923.234 -.067 -4. 146 .000
dum05 1619.622 1075.297 .013 1.506 .132
dum06 2164.144 2408.989 .017 .898 .369
dum07 -7838.495 2515.793 -.047 -3.116 .002

 

 

 

 

 

 

 

97

 

Coefficients (cont’d).

 

 

 

 

 

 

 

 

 

 

 

Industry 1 .261 .325 .031 3.886 .000
Dis_BFs_B 2.343 .212 .180 11 .063 .000
Dis_BFs_D 2.103 .306 .129 6.869 .000
Dis_BFs_A 1 .242 .355 .064 3.497 .000

 

a. Dependent Variable: Price

Excluded Variables

 

Collinearity Statistics

 

 

 

 

 

Model Beta In t Sig. Partial Correlation Tolerance
STY1 .3 .000
wood .a .000
dum04 a .000

 

 

 

 

 

 

 

a. Predictors in the Model: (Constant), Dis_BFs_A, pine_lap, Base, dumexterior, STY1.5, Black,

dumSTY, Update, dum98, STY1.75, Industry, block_brick, dum99, Acre, dim00, STY1.25, dum01,
alum_vinyl, FINBase, Garage, dum03, Hispanic, dum05, Age, dum07, STY2, dum02, Income.

Area, Education, Rent, Dis_BFs_D, Dis_BFs_B, dum06

b. Dependent Variable: Price

98

 

 

APPENDIX E Log-Linear Functional for ALL SITES

Model Summary

 

 

 

Adjusted R Std. Error of the
Model R R Square Square Estimate
1 .744 .554 .552 .3328298

 

 

 

 

 

a. Predictors: (Constant), Dis_BFs_A, pine_lap, Base, dumexterior, STY1.5, Black, dumSTY,
Update, dum98, STY1.75, block_brick, dim00, Acre, dum99, STY1.25, alum_viny|, dum01,
FINBase, Huspanic, Garage, dum03, STY2, dum05, dum07, Age, dum02, Income, Area,
Education, Rent, Dis_BFs_D, Dis_BFs_B, dum06

 

 

 

 

 

ANOVA
Model Sum of Squares df Mean Square F Sig.
Regression 1159.024 33 35.122 317.055 .000
1 Residual 933.174 8424 .111
Total 2092.1 98 8457

 

 

 

 

 

 

 

a. Predictors: (Constant), Dis_BFs_A, pine_lap, Base, dumexterior, STY1.5, Black, dumSTY,
Update, dum98, STY1.75, block_brick, dim00, Acre, dum99, STY1.25, alum_vinyl, dum01,
FINBase, Huspanic, Garage, dum03, STY2, dum05, dum07, Age. dum02, Income, Area,
Education, Rent, Dis_BFs_D, Dis_BFs_B, dum06

b. Dependent Variable: LnPrice

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Coefficients
Standardized
Unstandardized Coefficients Coefficients

Model B Std. Error Beta t Sig.

1 (Constant) 10.780 .049 218.498 .000
Acre .219 .065 .027 3.396 .001
Age -.004 .000 -.152 -15.707 .000
Area .000 .000 .292 25.392 .000
Base .000 .000 .108 11.967 .000
FINBase .000 .000 .046 5.802 .000
Garage .000 .000 .120 15.138 .000
Update .220 .027 .060 8.091 .000
Black -.004 .000 -.093 -9.165 .000
Huspanic -.002 .001 -.023 -2.071 .038
Rent -.002 .000 -.069 -4.774 .000

 

 

 

 

 

 

 

99

 

 

 

Coefficients (cont’d).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Income 7.261 E7 .000 .013 1.017 .309
Education .008 .000 .227 17.878 .000
STY1.25 .067 .012 .043 5.392 .000
STY1 .5 .080 .015 .042 5.409 .000
STY1.75 .110 .014 .065 7.753 .000
STY2 .106 .011 .099 9.303 .000
dumSTY .033 .045 .006 .725 .468
alum_vinyl -.025 .008 -.023 -3.009 .003
block_brick -.036 .025 -.011 -1.440 .150
pine_lap -.020 .084 -.002 -.238 .812
dumexterior -.146 .040 -.027 -3.652 .000
dum98 -.569 .028 -.328 -20.366 .000
dum99 -.503 .028 -.303 -18.017 .000
dim00 -.460 .028 -.277 -16.469 .000
dum01 -.346 .028 -.213 -12.422 .000
dum02 -.266 .027 -.170 -9.658 .000
dum03 -.188 .027 -.120 -6.856 .000
dum05 .031 .015 .019 1.997 .046
dum06 -.003 .034 -.002 -.101 .920
dum07 -.168 .036 -.077 -4.663 .000
Dis_BFs_B 4.031 E-5 .000 .236 13.358 .000
Dis_BFs_D 2.017E-5 .000 .094 4.618 .000
Dis_BFs_A 1 .856E-5 .000 .073 3.669 .000

 

 

 

 

 

 

a. Dependent Variable: LnPrice

Excluded Variables

 

Collinearity Statistics

 

 

 

 

 

Model Beta In t Sig. Partial Correlation Tolerance
STY1 .3 .ooo
wood .3 .ooo
dum04 a .000

 

 

 

 

 

 

 

a. Predictors in the Model: (Constant), Dis_BFs_A, pine_lap, Base, dumexterior, STY1.5, Black,

dumSTY, Update, dum98, STY1.75, block_brick, dimOO, Acre, dum99, STY1.25, alum_vinyl,
dum01, FINBase, Huspanic, Garage, dum03, STY2, dum05, dum07, Age, dum02, Income, Area,

Education, Rent, Dis_BFs_D, Dis_BFs_B, dum06

b. Dependent Variable: LnPrice

100

 

 

BIBLIOGRAPHY

Adler et al. 1982. K.J. Adler, R.C. Anderson, Z. Cook, R.C. Dower, A.R. Ferguson and
M.J. Vickers. The Benefits of Regulating Hazardous Waste Disposal: Land Values
as an Estimator. Washington, DC: Public Interest Economic Center

Bartsch, Charles and Barbara Wells. 2003. Financing Strategies for Brownfield Cleanup
and Redevelopment. Washington DC: Northeast-Midwest Institute.

Beck, Eckardt C. 1979. “The Love Canal Tragedy.” EPA Journal. January 1979. US.
Environmental Protection Agency.

Becker, N. and D. Lavee. 2003. “The benefits and costs of Noise Reduction.”
Environmental Management and Planning. 46(1): 97-111.

Bishop, ID. 1996. “Comparing regression and Neural Net Based Approaches to
Modelling of Scenic Beauty.” Landscape and Urban Planning 34: 125—134

Boyle, MA. and K.A. Kiel. 2001. “A Survey of House Price Hedonic Studies of the

Impact of Environmental Externalities.” Journal of Real Estate Literature
9(2): 1 1 7-144.

Brasington, D.M. 2002. “The Demand for Local Public Goods: the Case of Public School
Quality.” Public Finance Review. 30(3): 163—187.

Cheshire, Paul and Stephen Sheppard. 1995. “On the Price of Land and the Value of
Amenities.” Economica. 62(246): 247-267

Clark, D.E., L. Michelbrink, T. Allison, and WC. Metz. 1997. “Nuclear Power Plants and
Residential Housing Prices.” Growth and Change. 28(4): 496-519.

Corona, Joel Peter. 2004. Essays in Brownfield Redevelopment. Ph.D. Dissertation.
University of Connecticut.

Dale, L, J .C. Murdoch, M.A. Thayer and RA. Waddell. 1999. “Do property values
rebound from environmental stigmas? Evidence from Dallas.” Land Economics.
75(1999): 311—326.

Deaton, Brady J. Jr. 2002. Hazards and Amenities: Examining the Benefits of Hazardous
Waste Clean-up and Support for Farmland Preservation. Ph.D. Dissertation.
Michigan State University.

Deaton, B. James and John P. Hoehn. 2004. “Hedonic Analysis of Hazardous Waste Sites

in the Presence of Other Urban Disamenities.” Environmental Science & Policy.
7: 499 -508.

101

Downes, Thomas A. and J efi'rey E. Zabel. 2002. “The impact of school characteristics on
house prices: Chicago 1987-1991.” Journal of Urban Economics. 52: 1—25.

EPA. About Brownfield. Available at http://www.epa.goWbrownfields/abouthtm

EPA. 2007. Financing Brownfields: State Program Highlights. Publication No. EPA-560-
F-07-252.

EPA. 2005. Brownfields Federal Programs Guide. Publication No. EPA-560-F-05-230.

EPA. 2005. Superfund 's 25th Anniversary: Capturing the Past, Charting the Future.
Available at http://www.epa.gov/superfund/ZSanniversary/index.htm

EPA Region 5 Office of Public Affairs. 1996. Basic Brownfield Fact Sheet.

Espey, Molly and Hilary Lopez. 2000. “The Impact of Airport Noise and Proximity on
Residential Property Values.” Growth and Change 31(3): 408-419.

Farber, Stephen. 1998. “Undesirable Facilities and property values: A Summary of
Empirical Studies.” Ecological Economics. 24: 1-24

Freeman, A. Myrick III. 2003. The Measurement of Environmental and Resource Values:
Theory and Methods. 2nd Ed. Washington, DC: Resources For the Future Press

Gawande, Kishore & Hank J enkins-Smith. 2001. “Nuclear Waste Transport and
Residential Property Values: Estimating the Effects of Perceived Risks.” Journal
of Environmental Economics and Management. 42(2): 207-233

Geltman, Elizabeth Glass. 2000. Recycling Land: Understanding the legal landscape of
Brownfield Development. The University of Michigan Press: Ann Arbor, MI

Goodstein, Eban S. 1999. Economics and the Environment. 2nd Ed. Upper Saddle River,
NJ: Prentice-Hall, Inc.

Grigelis, Peter Edward. 2005. The Eflects of Environmental Contamination on
Commercial and Industrial Property Values: Do Perceptions Matter? Ph.D.
Dissertation. Georgia State University

Griliches, Z. 1971. Price Indexes and Quality Change. Cambridge, MA: Harvard
University.

Hula, Richard. 2003. Citizen Assessment of the Michigan Brownfield Initiative. IPPSR:
East Lansing, MI

102

Hula, Richard. 2003. The Michigan Brownfield Initiative and Private Market
Redevelopment: An Assessment. IPPSR: East Lansing, MI

Hwang, S. 2003. Environmental Amenities and Disamenities, and Housing Prices; Using
GIS Techniques. Ph.D. Dissertation. Texas A&M University.

Ihlanfeldt, Keith R. and Laura 0. Taylor. 2004. “Externality Effects of Small-Scale
Hazardous Waste Sites: Evidence from Urban Commercial Property Markets.”
Journal of Environmental Economics and Management 47 (1): 117-39.

Kaufman, Dennis A. and Norman R. Cloutier. 2006. “The Impact of Small Brownfields
and Greenspaces on Residential Property Values.” Journal of Real Estate Finance
and Economics. 33: 19-30.

Ketkar, Kusum. 1992. “Hazardous Waste Sites and Property Values in the State of New
Jersey.” Applied Economics. 24(6): 647-59.

Kiel, Katherine A. 1995. “Measuring the Impact of the Discovery and Cleaning of
Identified Hazardous Waste Sites on House Values.” Land Economics 71 (4): 428-
35.

Kiel, Katherine and Jeffrey Zabel. 2001. “Estimating the Economic Benefits of Cleaning
Up Superfund Sites: The Case of Wobum, Massachusetts.” Journal of Real Estate
Finance and Economics 22 (2-3): 163-84.

Kim, C., T. Phipps, and L. Anselin. 2003. “Measuring the benefits of air quality
improvement: A spatial hedonic approach.” Journal of Environmental Economics
and Management 45: 24—39.

Kohlhase, Janet E. 1991. “The Impact of Toxic Waste Sites on Housing Values.” Journal
of Urban Economics 30(1): 1-26.

Lancaster, K. J. 1966. “A New Approach to Consumer Theory.” Journal of Political
Economy. 74: 132-157.

Longo, Alberto and Anna Alberini. 2005. “What Are the Effects of Contamination Risks
on Commercial and Industrial Properties? Evidence from Baltimore, Maryland.”
FEEM Working Paper No. 111.05 Available at SSRN:
http://ssm.com/abstract=825566.

McCluskey, Jill J. and Gordon C. Rausser. 2003. “Hazardous Waste Sites and Housing

Appreciation Rates.” Journal of Environmental Economics and Management 45
(2): 166-76.

103

McCluskey, Jill J. and Gordon C. Rausser. 2003. “Stigmatized Asset Value: Is It
Temporary or Long-Term?” The Review of Economics and Statistics. 85 (2): 276-
285.

Michaels, R. Gregory and V. Kerry Smith. 1990. “Market Segmentation and Valuing
Amenities with Hedonic Models: The Case of Hazardous Waste Sites.” Journal of
Urban Economics 28 (2): 223-42.

Leggett, CG and Bockstael, NE 2000. “Evidence of the Effects of Water Quality on
Residential Land Prices.” Journal of Environmental Economics and Management.
39: 121-144.

McPherson, E.G. 1992. “Accounting for Benefits and Costs of Urban Greenspace.”
Landscape and Urban Planning. 22(1): 41-51.

MDEQ. 2008. Cleanup and Redevelopment Programs at Risk. Presented in a workshop
series entitled, “Current Status of Michigan Brownfield Redevelopment
Incentives” in Lansing, Michigan. On February 5 and 6, 2008. Available at
http://www.michigan.gov/deq/0,1607,7-135-331l_4l10_29262-185178--,00.html

MDEQ. 2007. Consolidated Report on the Environmental Protection Bond Fund,
Cleanup and Redevelopment Fund, Clean Michigan Initiative Bond Fund: Fiscal

Year 200 7. Available at http://www.michigan.gov/documents/deq/deq-rrd-
FY07ConsolidatedReport_229475_7.pdf#BRFA

Mishra, S. K. “On the Leading Approaches to Valuation of Environmental Goods and
Services.” Available at SSRN: http://ssm.com/abstract=594424

Mohamed, Rayman and Dancik, Brendan. 2007. Priming the Pump: Assessing the
Investment Impact of Michigan 29 Site Assessment Fund. East Lansing, MI: Land
Policy Institute.

Morancho, AB. 2003. “A Hedonic Valuation of Urban Green Areas.” Landscape and
Urban Planning. 66: 35—41.

Northeast-Midwest Institute. 1999. Brownfields “State of the States An End-of-Session
Review of Initiatives and Program in the 50 States. Washington DC: Northeast-
Midwest Institute.

O'Byme, P.H., Nelson, J .P. & Seneca, J .J . (1985). “Housing Values, Census Estimates,
Disequilibrium, and the Environmental Cost of Airport Noise: A Case Study of

Atlanta.” Journal of Environmental Economics and Management. 1 2: 169-178.

Patchin, Peter J. 1991. “Contaminated Properties-Stigma Revisited.” Appraisal Journal
59(2): 167-172

104

Rafson, Harold J. and Robert N. Rafson. 1999. Brownfield: Redeveloping
Environmentally Distressed Properties. McGraw-Hill: New York, NY

Riddel, M. 2001. “A Dynamic Approach to Estimating Hedonic Prices for Environmental
Goods: An Application to Open Space Purchase.” Land Economics 77: 494—5 12.

Rosen, Sherwin. 1974. “Hedonic Prices and Implicit Markets: Product Differentiation in
Pure Competition.” Journal of Political Economy. 82 (1): 34-55.

Smith, V. K. and W. H. Desvousges. 1986. Measuring Water Quality Benefits. Boston,
MA: Kluwer-Nijho.

Smith, V.K. and RB. Palmquist. 1994. “Temporal Substitution and the Recreation Value
of Coastal Amenities.” Review of Economics and Statistics. 76:119-126.

Smith, V.K., C. Poulos, and H. Kim. 2002. “Treating Open Space As an Urban Amenity,
Resource.” Energy Economics. 24: 107—129.

Tajima, K. 2003. “New Estimates of the Demand for Urban Green Space: Iimplications
for Valuing the Environmental Benefits of Boston’s Big Dig Project.” Journal of
Urban Affairs. 25 (5): 641—655.

The US. Census Bureau. 2007 American Community Survey. Available at: http://
http://factfinder.census.gov/servlet/DatasetMainPageServlet?_program=ACS&_s
ubmenuld=datasets_2&_lang=en&_ts=

The US. Census Bureau. US. Census 2000. Available at:
http://factf1nder.census.gov/servlet/DatasetMainPageServlet?_program=DEC&_s
ubmenuld=datasets_2&_lang=en&_ts=

The United States Conference of Mayors, 2006. Recycling America 19 Land: A National
Report on Brownfield Redevelopment. Vol. VI. Washington DC.

The United States Conference of Mayors, 2000. Recycling Americas Land: 3rd Annual
Report on the Status of Brownfields. Washington DC.

Tyrvainen, L. 1997. “The Amenity Value of the Urban Forest: An Application of the
Hedonic Pricing Method.” Landscape and Urban Planning. 37: 211-222.

Tyrvainen L. and A. Miettinen. 2000. “Property Prices and Urban Forest Amenities.”
Journal of Environmental Economics and Management. 39: 205-223.

Wemstedt, Kris. 2004. Overview of Existing Studies on Community Impacts of Land Use.

Working Paper #04-06. Washington DC: National Center for Environmental
Economics.

105

Whittington, Jane. 2007. “Motor Wheel Reborn.” Greater Lansing Business Monthly.
July 2007.

Wisinger, Perry (1. 2006. The Impact of Chemical Hazardous Sites on Residential Values.
Ph.D. Dissertation. Texas Tech University.

106

2 5572

93 030

l

 

'lilllllm