ArticlePDF Available

The Effect of Eminent Domain on Private and Mixed Development on Property Values

Authors:

Abstract and Figures

The objective of this study is to assess the impact of eminent domain (ED) for private and mixed development on property values in Rochester, New York, within 107 months of policy announcements and construction initiations. This study includes data on 19,707 screened house sales. By using both parametric and semiparametric models, this study concludes that the Midtown Plaza (MP) redevelopment project purely for private development generates positive policy externalities on property values across the city. However, homeowners lost property value if they lived within a one mile radius of the MP center after the policy announcement. The average citywide housing prices dropped by 8.2% after the MP demolition began, and yet, homeowners living within a one mile radius of the MP neighborhood enjoyed an 8.7% property value gain after the start of the MP demolition. There is no significant credible policy impact from the Brooks Landing (BL) project. This project for mixed development aims for both public and private revitalization. Citywide housing prices dropped by 6.8% after the start of the BL site demolition and homeowners suffered a 1.4% property value loss for each mile closer to the BL site under demolition. The semiparametric model takes spatial heterogeneities and nonlinearities into consideration; thus, due to the spatial dependence problem within the dataset, it is superior to the parametric model in this study.
Content may be subject to copyright.
The Effect of Eminent Domain on Private and Mixed
Development on Property Values
Peiyong Yu
University of Hawaii at West Oahu USA
Abstract: The objective of this study is to assess the impact of eminent domain (ED) for private
and mixed development on property values in Rochester, New York, within 107 months of
policy announcements and construction initiations. This study includes data on 19,707
screened house sales. By using both parametric and semiparametric models, this study con-
cludes that the Midtown Plaza (MP) redevelopment project purely for private development
generates positive policy externalities on property values across the city. However, homeown-
ers lost property value if they lived within a one mile radius of the MP center after the policy
announcement. The average citywide housing prices dropped by 8.2% after the MP demolition
began, and yet, homeowners living within a one mile radius of the MP neighborhood enjoyed
an 8.7% property value gain after the start of the MP demolition. There is no significant cred-
ible policy impact from the Brooks Landing (BL) project. This project for mixed development
aims for both public and private revitalization. Citywide housing prices dropped by 6.8% after
the start of the BL site demolition and homeowners suffered a 1.4% property value loss for
each mile closer to the BL site under demolition. The semiparametric model takes spatial het-
erogeneities and nonlinearities into consideration; thus, due to the spatial dependence problem
within the dataset, it is superior to the parametric model in this study.
1. Introduction
Eminent domain (ED) is popularly deemed either
an important tool for elevating the living standards in
neighborhoods and rejuvenating cities or a violation
of private property rights. Munch (1976, p. 473) de-
scribes ED as “the legal right to acquire property by
forced rather than by voluntary exchange. When a
buyer seeking to acquire a property has the power of
ED, he must attempt to negotiate a voluntary sale.
But if his highest offer is rejected, he may condemn
the property, that is, obtain a forced sale at a price de-
termined in a court of law.” Eminent domain is an
economic policy with a long history. Initially there
was with no compensation, partially because land
was abundant, but after the Fifth Amendment to the
Constitution, compensation at the market value be-
came mandatory and ED is now allowed only for
public use. Public use includes construction of roads,
parks, schools, hospitals, etc.
The use of ED is most controversial in the preva-
lent applications to a large-scale economic redevelop-
ment project: proponents of strong governmental
powers deem ED the solution to market failures and
help acquire property for purposes of redeveloping
blighted inner cities, while others are property rights
advocates who see more government failures and
deem unconstitutional such actions as transferring
property from one private party to another (Benson
and Brown, 2007). The question here is: does the use
JRAP 45(2): 173-187. © 2015 MCRSA. All rights reserved.
174 Yu
of ED for private development produce positive ex-
ternalities for the city’s property values? It becomes
more urgent to answer this question after the Kelo v.
New London lawsuit, in which the drug company
Pfizer built a new plant in 1998 in New London, Con-
necticut, aiming to bring in more jobs and govern-
ment revenues. The City of New London decided to
purchase 115 additional houses in a nearby area to
sell them to commercial developers, but 15 residents,
including Kelo, resisted, so the city used ED (545 U.S.
469, 2005). The public outrage from the Kelo case led
many states to think about the lawful applicability
and politically viable use of ED for private develop-
ment and even mixed development.
The legal justification of ED is based on the Fifth
Amendment to the Constitution, which grants the
power of ED to government only for ‘public use’ and
with ‘just compensation’ (U.S. Const. amend. V, sec.
2). The definition of public use has been broadened
to include anything that benefits the public, such as
inner city revitalization, downtown redevelopment,
and airport expansions. The U.S. Supreme Court con-
tinued to expand its definition to include aesthetic
considerations. Liston (2013) summarizes the Berman
v. Parker case in 1954 when the court ruled that the
government can transfer property from one private
party to another as part of a redevelopment plan that
serves a public purpose; however, a property owner
objected to the government’s taking a piece of prop-
erty that was not blighted. In Poletown v. City of De-
troit (1981), the Michigan Supreme Court condemned
a large tract of land to be conveyed to General Motors
Corporation, under the conjecture that a new assem-
bly plant would help revitalize the economy of the
state. Growing and extensive uses of ED during the
recent economic recession have caused widespread
concern, and these debates have been centered on
whether there is an abuse of ED for private gains.
The current concern is how the government seiz-
ing property and providing it to private developers
or individuals affects nearby property values. To an-
swer this question, I collected ED data, house sales
data, and GIS data from the City of Rochester, New
York, to explore the impact of two most recent pro-
jects that used ED on property values. The two study
cases in Rochester are the Midtown Plaza (MP) pro-
ject, which was initially announced in November
2006, and, after many false starts, the Brooks Landing
(BL) project for redevelopment, which was an-
nounced in November 2005 (the credible announce-
ment). The MP project is purely for private develop-
ment, while the BL project is a mixed development
(public and private). This study questions whether
these two different projects create the same social
benefit measured by improvements of single-family
residence values, which are determined by both the
houses’ physical and locational characteristics.
In this paper different statistical models, including
OLS (Ordinary Least Squares), fixed effects, and sem-
iparametric models, are employed to investigate the
impact of eminent domain for both private and mixed
development on property values. The general form
of fixed effects models is Y = αi + Xitβ + uit. Dealing
with the spatially referenced data, semiparametric
models have been frequently applied because they
keep the explanatory power of parametric models
and the flexibility of nonparametric models (Li and
Mei, 2013). The semiparametric model is partially lin-
ear and partially nonlinear, having general form Y =
Xβ + m(z) + ε. The function m does not have any pre-
defined functional form, and its error distribution
cannot be assumed to be of any specific type. The
commonly used parametric hedonic pricing model
assumes that there is a linear relationship between
the housing prices and locational variables. How-
ever, incorrect functional forms and omitted varia-
bles that are correlated over space produce spurious
spatial autocorrelation (Basile et al., 2014). Thus, a
semiparametric model including spatial parameters
(latitude and longitude as a vector of z in the function
m) is employed in this study to take both spatial het-
erogeneity and nonlinearities into consideration.
2. Literature review
2.1. Economic analysis of eminent domain
Research on the impact of ED on various eco-
nomic outcomes generally falls into three different
categories. The first category finds evidence that ED
promotes economic growth. Collins and Shester
(2010) used data on more than 25,000 residents in two
years (1950 and 1980) to investigate whether more in-
tensive urban renewal programs led to better eco-
nomic outcomes in 1980. This study used OLS regres-
sion model with city-level control variables and cen-
sus-division dummy variables to estimate the pro-
gram’s effects on city-level outcomes, which included
median family income, median property value, em-
ployment, and poverty rates. The pre-program con-
trol variables included housing stock, population,
and economic characteristics in 1950. The authors
found that the urban renewal programs led to higher
median incomes and higher median property values
in 1980. However, some of the census division
Effect of Eminent Domain on Property Values 175
dummy variables such as Mountain, Pacific, South At-
lantic, etc., are very sensitive to model specification
due to possible spatial dependence problems. Their
OLS model just assumes that those locational dummy
variables have a linear relationship with the three
city-level outcome variables. These potentially seri-
ous misspecifications may yield biased and/or ineffi-
cient parameter estimates (Brenner, 1977).
The second group of studies found that ED pro-
jects have mixed or no net impacts on economic out-
comes. Munch (1976) used data from the Chicago De-
partment of Urban Renewal for three large projects
during the period of 1962-1970. He also used an OLS
model to estimate the relationship between market
value, assessed value, and property characteristics
from sample data of homes sold on the open market
compared to those sold under urban renewal pro-
jects. Munch found that, under ED, high-valued par-
cels systematically received more than market value
and low-valued parcels received less than market-
value. This early literature only includes seven ex-
planatory variables, and the zoning dummy variable
is also sensitive to possible spatial dependence prob-
lems.
Carpenter and Ross (2010) examined whether lim-
iting the use of ED for private-to-private transfer of
property significantly harmed economic growth.
Their study used hierarchical linear modeling to
measure economic effects on three dependent varia-
bles: construction jobs, building permits, and prop-
erty tax revenues. They hypothesized that if efforts
to limit ED harm economic development, trends in
those three variables should have turned negative af-
ter legislation becomes effective. For building per-
mits, covariates included the number of sales of exist-
ing houses and aggregate personal income; total tax
revenues and home vacancy rates were control varia-
bles for property tax; and overall labor force minus
construction employment and building permits were
control variables for construction jobs. This study an-
alyzed each of the indicators separately by using lin-
ear models and state-level quarterly data from all 50
states from 2004 to 2007. Results indicated that there
were no negative economic consequences after the
legislative/judicial change.
The last group of studies found that ED had a neg-
ative impact on economic outcomes. Carpenter and
Ross (2009) discovered three waves of gentrification
that varied with respect to the two distinct character-
istics of state involvement and extent of gentrifica-
tion. The first wave of gentrification from 1960s to
early 1970s includes significant state involvement
mainly in the large northeastern cities in the U.S.
Meanwhile, the second wave of gentrification which
surged from the late 1970s to the early 1990s was
characterized by less state involvement but more pri-
vate market gentrification. The third wave of gentri-
fication occurred after the early 1990s with increasing
government involvement in public-private partner-
ships and many times involved entire neighbor-
hoods. Carpenter and Ross’ data included 184 areas
targeted by ED for private development. For each
project area the percentages of minority residents,
children and senior citizens, and renters and owners
along with education and poverty levels were ana-
lyzed using independent sample t-tests to study dif-
ferences between project areas and surrounding com-
munities. The study indicated that a greater percent-
age of minority residents (58%) compared with their
surrounding communities (45%) were in the areas
targeted by ED for private development. Similar re-
sults were found for education and income variables,
leading the authors to conclude that ED dispropor-
tionately hurt poor, minority, and other historically
disenfranchised community members.
Kerekes and Gulf (2011) used Dana Berliner’s data
of ED use for private benefit for all 50 U.S. states ex-
tracted from court papers and published accounts
spanning from 1998 to 2002. They used basic Poisson
models and found that ED for private benefit is uti-
lized more widely in states with higher rates of cor-
ruption, appointed Supreme Court judges, less fiscal
decentralization, and lower economic freedom.
2.2. Nonparametric hedonic models
Bin and Filho (2001) investigated 1,000 recorded
sales in the Portland-Oregon housing market be-
tween 1992 and 1994 to estimate a hedonic price func-
tion with application to additive nonparametric re-
gression modeling. They argued that the functional
form specification problem common in hedonic price
models can be conveniently addressed by modeling
the conditional mean of prices in a nonparametric en-
vironment. They compared their results to an alter-
native parametric model and found evidence of the
superiority of nonparametric model.
McMillen and Redfern (2010) used all sales of
single-family homes in 2000 that are within one mile
of Chicago’s elevated train line. They used GIS to
measure distance from Chicago’s city center and
showed how nonparametric and semiparametric
procedures assist in the specification of a hedonic
house price function. They argued that semiparamet-
ric estimation procedures can control for spatial
variation in marginal effects while also allowing for
nonlinearities.
176 Yu
Haupt, Schnurbus, and Tschernig (2010) per-
formed a replication of Parmeter et al. (2007), apply-
ing nonparametric methods for estimating hedonic
house price functions and comparing the results to
the parametric and semiparametric specifications.
They extended their analysis by using the nonpara-
metric specification test used in Hsiao and Racine
(2007) for mixed continuous categorical data and sim-
ulation-based prediction comparisons. They found
that the previously proposed parametric specification
does not have to be rejected but suggest that the non-
parametric methods still provide valuable insights
during all modeling steps.
Therefore, while there are many studies on the re-
lationship between ED and various economic out-
comes and also many studies preferring nonparamet-
ric to parametric specification, none has specifically
focused on the relationship between local property
values and ED used for both private and mixed de-
velopment. Compared to the previous similar litera-
ture mainly using simple linear models, this study
uses both the parametric and semiparametric models
to improve the measurements of ED’s impact on
property values. The parametric fixed effects model
taking into account within-neighborhood heteroge-
neities helps compare with the semiparametric
model. The semiparametric analysis in this paper is
mainly based on McMillen and Redfern (2010), in-
cluding geographic coordinates in the nonparametric
part to solve the spatial dependence problems.
3. Study area and data
The Midtown Revitalization (MP) project was
undertaken by private development company Mid-
town Tower LLC, which is a partnership of Bucking-
ham properties and Morgan Management. As shown
on Figure 1 as a square, MP is located in the center of
Rochester. It focuses on the revitalization of the city
center. This plan is attempting to transform the for-
mer Midtown Plaza into a more attractive neighbor-
hood. The announcement of the MP revitalization
project was in November 2006, and demolition
started in October 2010. The second project is Brooks
Landing (BL), which is located in southwest Roches-
ter, marked by a diamond on Figure 1. The BL project
is a public/private development project to improve
economic conditions through ED. The project an-
nouncement was in October 2005, after many false
1
The school district, library and CBD are also added on the map
using the same way to create distance variables by using the ‘near’
function of the ArcGIS in the ‘proximity’ category.
starts, and the demolition began in September 2006.
This project consists of a mixed-use development in-
cluding boat storage facility, restaurant, student
housing, parking spaces, and a credit union drive-
through operation (City of Rochester, 2012).
This study creates four dummy variables
(MPP_d, BLP_d, MPD_d and BLD_d) to split the sales
before and after the MP and BL projects policy an-
nouncements and site demolitions. Two dummy var-
iables (MPN and BLN) are created to represent the
houses located in the MP and BL neighborhoods (ar-
eas within a 0.5 mile radius of the centroid of the ED
project), and they also interact with the earlier created
four dummy variables to create four more variables
(MPP_n, MPD_n, BLP_n and BLD_n). The descrip-
tions and summary statistics of the key variables are
listed in Table 1.
There are several sources of data used to estimate
the impact of the ED projects for private and mixed
development on property values. The primary da-
taset consists of single-family residential transactions
occurring between 2000 and May 2013 in Rochester.
This study period covers the recent “Great Reces-
sion” from December 2007 to June 2009. Fourteen
yearly dummies are created to control for annual eco-
nomic fluctuations. The housing prices are trans-
formed to adjust for inflation, with the base year 2000.
Data on sale prices and property characteristics were
compiled from information provided by the Depart-
ment of Assessment & Taxation. After deleting the
multi-family, land, and commercial sales data from
the original 28,487 records, there were 19,707 single-
family residential house sales left for this study. The
single-family house data also includes the number of
bathrooms, fireplaces, bedrooms, stories in structure,
garage car spaces, per square foot, an exterior con-
struction of concrete or stucco, distance to the nearest
school, library, or CBD, etc., since they are very likely
to affect housing prices positively (Cebula, 2009).
Addresses of ED areas and GIS shapefiles of
schools, libraries, parks, recreation centers and CBD
were acquired from the Department of Neighbor-
hood & Business Development. The home addresses
are geocoded to obtain the longitude and latitude of
each observation in order to calculate the distance to
the Midtown Plaza and Brooks Landing areas
(MP_DIST & BL_DIST).
1
Brooks Landing areas in-
clude both private development and public project
areas. BL_DIST only accounts the nearest distance
Effect of Eminent Domain on Property Values 177
from each house to the area for private development.
These distance variables have to interact with the four
dummy variables (MPP_d, MPD_d, BLP_d and
BLD_d) to create four more distance variables
(MPPD, MPDD, BLPD and BLDD) to get the exact
distance impacts on property values after their policy
announcements and the start of their demolitions.
Figure 1. Study area and geocoded results.
Note: More detailed maps of the projects can be found in the appendix.
178 Yu
Table 1. Descriptions and summary statistics of key variables.
Variable
Description
St Dev
Min
Max
MPP_d
MPP_d =1 if the houses were sold after 11/2006
(when the MP policy was announced)
0.4979
0
1
MPD_d
MPD_d =1 if the houses were sold after 10/2010
(when the MP demolition started)
0.3593
0
1
MP_DIST
The nearest distance to the MP area in miles
1.0639
0.27
7.593
MPPD
MPPD = MP_DIST * MPP_d
1.4152
0
7.588
MPDD
MPDD = MP_DIST * MPD_d
0.9537
0
7.585
BLP_d
BLP_d =1 if the houses were sold after 11/2005
(when the real BL policy was announced)
0.4981
0
1
BLD_d
BLD_d =1 if the houses were sold after 09/2006
(when the BL demolition started)
0.4991
0
1
BL_DIST
The nearest distance to the BL area in miles
1.7651
0.037
9.362
BLPD
BLPD = BL_DIST * BLP_d
2.1385
0
9.360
BLDD
BLDD = BL_DIST * BLD_d
2.0813
0
9.360
MPN
MPN = 1 if houses are located within a 0.5 mile radius
of the centroid of the MP site (MP neighborhood)
0.0396
0
1
MPP_n
MPP_n = MPN * MPP_d
0.0310
0
1
MPD_n
MPD_n = MPN * MPD_d
0.0175
0
1
MP1m
MP1m =1 if houses are located within a one mile
radius of the MP neighborhood
0.1703
1
MPP1m
MPP1m = MP1m* MPP_d
0.1206
0
1
MPD1m
MPD1m= MP1m * MPD_d
0.0725
0
1
BLN
BLN =1 if houses are located within a 0.5 mile radius
of the centroid of the BL site ( BL neighborhood)
0.1595
0
1
BLP_n
BLP_n = BLN * BLP_d
0.1190
0
1
BLD_n
BLD_n = BLN * BLD_d
0.1132
0
1
BL1m
BL1m =1 if houses are located within a one mile
radius of the BL neighborhood
0.2454
0
1
BLP1m
BLP1m = BL1m * BLP_d
0.1831
0
1
BLD1m
BLD1m = BL1m * BLD_d
0.1711
0
1
BDS
The number of bedrooms
0.7841
1
32
BTH
The number of bathrooms
0.5412
1
9.5
AGE
The age of the house
25.475
0
211
LAT
The latitude of each house
0.0284
43.11
43.27
LONG
The longitude of each house
0.035
-77.68
-77.54
4. Methodology
A semiparametric model is a combination of par-
ametric and nonparametric approaches. The bench-
mark OLS model assumes a very strict functional
form in which the dependent variable is determined
by the regressors and unobserved errors based on a
fixed structure. The disadvantage of parametric
models, including the fixed effects model, is the re-
quirement that both the structure and the error distri-
bution are specified correctly. Nonparametric mod-
els, on the other hand, impose very few restrictions
on the functional form, so there is little room for mis-
specification (Powell, 1994). However, the precision
of estimators which impose only nonparametric re-
strictions is poor (Powell, 1994), and there is always a
Effect of Eminent Domain on Property Values 179
“curse of dimensionality”. A semiparametric model
combines the merits of parametric and nonparamet-
ric models.
The parametric part of the semiparametric model
is:
*
181716
15141312
111098
7654
321
*
1_1
__
__1_
1_
___*ln
it
it
mBLDnBLDmBLP
nBLPBLDDBLPDDISTBL
dBLDdBLPmMPDnMPD
mMPPnMPPMPDDMPPD
DISTMPdMPDdMPPP
(1)
The dependent variable price has a natural log form,
since the coefficient on the natural-log scale is directly
interpretable as approximate proportional differ-
ences. The procedure for the smoothing part of the
semiparametric model is LOWESS, which is a proce-
dure for fitting a regression surface to data through
multivariate smoothing: the dependent variable is
smoothed as a function of the independent variables
in a moving fashion analogous to how a moving av-
erage is computed for a time series (Cleveland and
Devlin, 1988). The smoothing degree varies, usually
falling between 0 and 1. For example, if the window
size is 0.2, it indicates that the smoothing window has
a total width of 20% of the horizontal axis variable.
Different from parametric smoothing techniques re-
quiring functional forms beforehand, this nonpara-
metric smoothing method allows the data to speak for
itself. Thus, the fitted curve drawn by lowess is gen-
erated empirically rather than through prior
specifications about any structural nature that may
exist within the data. A detailed application to hous-
ing price functions is found in McMillen and Red-
fearn (2010). The target for the nonparametric esti-
mator is a house with structural and locational char-
acteristics given by the vector X. The LOWESS esti-
mator is then derived by minimizing the following
equation with respect to α and β:
n
i
i
ii hXX
KXXLnP
1
2)())((
(2)
The kernel function K(z) determines the weight of
each house sold as an observation in estimating the
house price at target point X, with Xi X defined as
the distance between the target point and the ith
neighboring house and h a smoothing parameter
called the bandwidth. As the distance increases, the
weight declines; thus a kernel represents a decreasing
function of a distance between two objects. Though
there are various types of kernels, such as uniform,
Ephanechnikov, biweight, triweight or Gaussian, the
choice of kernel weight function usually has little ef-
fect on the results. This study uses a tri-cube kernel,
but h is more important since it determines how
many observations receive positive weight when con-
structing the estimate as well as how rapidly the
weights decline with distance. By placing more
weight on more distant observations, high values of
h (i.e., larger bandwidth) imply local regressions that
produce more smoothing than do smaller band-
widths (McMillen and Redfearn, 2010). The choice of
an optimal bandwidth is a crucial procedure in the
nonparametric part of the semiparametric analysis.
This study uses Silverman’s Rule of Thumb to deter-
mine the optimal bandwidth. Silverman proposes
the rule-of-thumb bandwidth as
)12(1
)(
v
vnkCh
(3)
where
is the sample standard deviation, v is the
order of the kernel, and Cv(k) is a constant depending
on the type of kernel used. Since this study uses the
triweight kernel, according to the Silverman Rule the
constant is 3.15 if the kernel order is 2. Since the lati-
tude and longitude of each house are estimated in the
nonparametric part, their average standard deviation
is 0.063, as given in the Table 1. Plugging this number
in the rule-of-thumb function, the optimal bandwidth
for this study is about 0.03.
5. Empirical results
The impact of ED on private property values is
largely an empirical question, since there are reasons
to expect that redevelopment using ED may either in-
crease or decrease property values in the short run.
Redevelopment acts as a form of insurance for future
neighborhood quality, raising property values with
possible positive spillovers to adjacent communities.
On the other hand, time-consuming and inefficient
redevelopment projects may reduce nearby property
values due to construction noise, congested traffic,
and lost investor/consumer confidence.
Table 2 records the parametric fixed neighbor-
hood effects and semiparametric regression results
for key variables. The regression results for the rest
of the variables are in Table A1 in the Appendix. Fig-
ure 2 records the graphical comparisons of results un-
der these two specifications: both the fixed neighbor-
hood effects and semiparametric specifications indi-
cate that the citywide housing prices increased, by
9.8% and 10.7%, respectively, after the MP policy an-
nouncement. However, they dropped by 6.7% and
180 Yu
8.2%, respectively, after the city started the MP dem-
olition. Both specifications indicate that after the pol-
icy announcement the homeowners enjoyed property
value gains, of 2.6% and 3.2%, respectively, for each
mile closer to the MP area. However, homeowners
living in the MP neighborhood suffered about 42.8%
and 43.9% loss of their property value, respectively,
under the two specifications, after the MP policy an-
nouncement.
Figure 2. Graphical parametric and semiparametric regression results for both Midtown Plaza
(MP) project and Brooks Landing (BL) project.
Effect of Eminent Domain on Property Values 181
Table 2. Regression results under fixed effects and semiparametric specifications.
Variables
Fixed Effects Model
Semiparametric Model
INTERCEPT
8.905 (0.239)***
N/A
MPP_d
0.098 (0.037)**
0.107 (0.03)***
MPD_d
-0.067 (0.031)*
-0.082 (0.026)**
MP_DIST
0.047 (0.105)
0.047 (0.114)
MPPD
-0.026 (0.011)*
-0.032 (0.005)***
MPDD
0.01 (0.009)
0.011 (0.006)
BLP_d
0.049 (0.038)
0.047 (0.03)
BLD_d
-0.067 (0.036)
-0.068 (0.026)**
BL_DIST
0.043 (0.042)
-0.001 (0.069)
BLPD
-0.003 (0.005)
-0.003 (0.005)
BLDD
0.014 (0.007)
0.014 (0.005)**
MPN
0.158 (0.087)
0.135 (0.096)
MPP_n
-0.428 (0.064)***
-0.439 (0.103)***
MPD_n
0.209 (0.035)***
0.234 (0.127)
MP1m
0.061 (0.035)
0.047 (0.027)
MPP1m
-0.071 (0.033)*
-0.073 (0.025)**
MPD1m
0.108 (0.043)*
0.087 (0.034)**
BLN
-0.031 (0.032)
-0.073 (0.036)*
BLP_n
-0.035 (0.028)
-0.033 (0.055)
BLD_n
0.112 (0.041)**
0.107 (0.055)
BL1m
-0.026 (0.03)
-0.034 (0.023)
BLP1m
-0.011 (0.031)
-0.011 (0.034)
BLD1m
0.024 (0.04)
0.02 (0.034)
Adjusted R2
0.904
AIC
0.068
House characteristics
x
x
Year control
x
x
Control groups
130
Spatial control
x
Observations
19,707
19,707
Note: standard errors in the parentheses; * 5% significance; ** 1% significance, *** 0.1% significances.
In addition, the two specification results indicate
that the homeowners living within a one mile radius
of the MP neighborhood suffered 7.1% and 7.3%
losses of their property value, respectively, after the
MP policy announcement. In contrast, homeowners
living within a one mile radius of the MP neighbor-
hood enjoyed 10.8% and 8.7% property value gain, re-
spectively, under the two specifications, after the MP
demolition began. The only difference between these
two specification results is the impact on the houses
located within the MP neighborhood after the city
started MP demolition. The fixed effects model indi-
cates that after the city began tearing down MP,
homeowners living within the MP neighborhood en-
joyed a 20.9% property value gain. However, the
semiparametric model indicates insignificance re-
lated to this variable (MPD_n).
Both models indicate very few significant impacts
on the property value from the mixed BL project. The
fixed effects model indicates that homeowners living
in the BL neighborhood enjoyed an 11.2% property
value gain after the city started the BL demolition.
182 Yu
The semiparametric model indicates that there is no
significant impact on homeowners living in the BL
neighborhood after the BL demolition began, but the
citywide housing prices dropped by 6.8% and home-
owners suffered a 1.4% loss of their property value
when they moved one mile closer to the BL site after
the city started the BL demolition. Both specifications
reach the same conclusion for this project: since there
had been many false policy announcements, the re-
cent credible policy announcement produced no sig-
nificant impact on property values. However, when
BL demolition started, it started creating real impact.
This limited impact is positive under the fixed effects
model, but negative under the semiparametric
model. The two models generate very similar results
for the MP project, which is purely for private devel-
opment, but contrasting results for the hybrid BP pro-
ject.
The AIC (Akaike Information Criterion) test result
of the fixed effects model (0.068) indicates that the
fixed effects model is already a very good specifica-
tion (0.068 < 0.1); the adjusted R2 value (0.904) also
indicates the same. However, Figure 3 below shows
that there are some violations of its parametric as-
sumptions. According to the Anderson-Darling test,
the error structure is not normally distributed (p-
value < 0.05). The residual vs. fits plot indicates that
the equal error variance assumption is also violated.
Therefore, this study confirms the appropriateness of
using a semiparametric model, which is a much more
flexible functional form than any parametric model.
Figure 3. Residual plots for the dependent variable.
Robust research has to deal with the potential mis-
specification caused by the spatial nature of the data.
Spatial heterogeneity is one obvious reason to use
nonparametric models for spatially referenced data.
When all of the variables are included in the regres-
sion model, it loses degrees of freedom, which leads
to the ‘curse of dimensionality’. The semiparametric
model is better since it captures the spatial heteroge-
neities. Testing Moran's I measure of spatial autocor-
relation results in rejecting at α = 0.05 the null hypoth-
esis that there is zero spatial autocorrelation present
in the variable Lnprice . Figure 4 indicates that there
are evident spatial heterogeneities across the plane of
the study area.
Effect of Eminent Domain on Property Values 183
Figure 4. Surface plots of Lnprice vs latitude and longitude across the years.
6. Conclusions and discussion
How housing prices vary with adjacent develop-
ment carries important policy and market implica-
tions. The impact of ED for private and mixed devel-
opment on neighboring communities is a hotly de-
bated topic. Local governments are often interested
in the process of gentrification - trying to bring new
businesses and residents into moribund city centers.
The positive aspects of proximity are related to the
revitalized economy in the city center with more
amenities added. The downsides of proximity are ad-
verse effects primarily associated with inefficient
planning, unfulfilled promises, construction noise,
blocked views, and congested traffic.
The purpose of this paper is to evaluate the impact
of ED for private and mixed development on prop-
erty values in Rochester, New York. Based on a large
and detailed sample of single-family home transac-
tions, two ED cases, the Midtown Plaza (MP) and
Brooks Landing (BL) projects, are studied. The for-
mer project is purely for private revitalization, but the
latter one is for both private and public redevelop-
ment. For a rigorous analysis, this study uses both
parametric fixed effects and semiparametric regres-
sions.
5.0
7.5
10.0
43.10 43.15 43. 20 43.25
12.5
2012
2008
2004
2000
43.25
Lnprice
Year
Latitude
Surface Plot of Lnprice vs Latitude, Year
Evidence of Spatial Heterogeneity
1
5.0
7.5
10.0
-77.65
-77.60
-77.55
12.5
2012
2008
2004
2000
-77.55
Lnprice
Year
Longitude
Surface Plot of Lnprice vs Longitude, Year
Evidence of Spatial Heterogeneity
2
184 Yu
For the MP project, there are many similar results
between the two models. Both specifications indicate
that the credible MP policy had a positive distance
spillover impact. They also indicate that the housing
prices dropped by approximately 44% for houses lo-
cated in the MP neighborhood along with approxi-
mately another 7% property value drop for houses lo-
cated within a one mile radius of the MP neighbor-
hood after the MP policy announcement. Both show
that citywide housing prices increased by approxi-
mately 10% after the MP policy announcement, but
they dropped by 6.7% and 8.2%, respectively, after
the MP demolition started. The parametric and sem-
iparametric results also show that there were 8.7%
and 10.8% property value drops, respectively, for
homeowners living within a one mile radius of the
MP neighborhood after the MP demolition began.
Both the parametric and semiparametric models indi-
cate that the BL project policy has no significant im-
pact on property values.
There are different economic incentives for differ-
ent economic participants within the city. Some local
residents claim that their city used ED to force closure
of the Midtown Plaza malls and then hand the land
over to a corporate darling and other well-connected
friends, while others are expecting downtown gentri-
fication, which helps attract high-income residents or
investors back to their neighborhood. Home buyers
might not consider residential houses a good invest-
ment in the MP area after the policy announcement
due to many uncertainties; thus, there might be less
demand for houses located in the neighborhood,
which would likely to be condemned soon. On the
supply side, some existing homeowners in the MP
area may fear that the future physical takings might
harm their property value. They might be eager to
sell their houses after the initial policy announce-
ment. This rational combination of leftward shift of
the demand curve and rightward shift of the supply
curve could cause a huge price drop, which is vali-
dated by the result of an approximately 44% property
value drop in the MP neighborhood and another 7.3%
property value drop within a one mile radius of the
MP neighborhood after its policy announcement.
On the other hand, people seem to hold positive
expectations on the revitalization of their city center.
Citywide housing prices increased by 10.7%, and
homeowners enjoyed a 3.2% property value gain for
each mile closer to the Midtown Plaza after the MP
policy announcement. Even though this positive
citywide policy shock was at the cost of a small group
of people, the previously significant negative impact
on homeowners living in the MP neighborhood
(MPP_n) dropped to insignificance (MPD_n) after the
start of the MP demolition. Compared to the 7%
property value loss for homeowners living within a
one mile radius of the MP neighborhood after its pol-
icy announcement, they enjoyed an 8.7% property
value gain after the start of MP demolition; thus,
there is a net 1.4% property value gain for homeown-
ers living in the proximity to the MP neighborhood.
Even though there was an 8.2% citywide housing
price drop after the MP demolition, there was a 10.7%
citywide housing price rise after the MP policy an-
nouncement, so there was a 2.5% net property value
gain for the whole city.
For the BL project, since it had many false starts,
the policy announcement did not produce any credi-
ble policy shock. This is validated by the results,
which indicate that there is no significance related to
the impact of the BL project policy on property val-
ues. There are negative impacts resulting from the BL
site demolition: citywide housing prices dropped by
6.8% after the city started the BL demolition, and
homeowners lost 1.4% for each mile closer to the BL
site under demolition. In sum, there are no positive
externalities for residential property values regarding
this BL project, which was used for both private and
public redevelopment. This is an example of ineffi-
cient planning.
In summary, eminent domain for private and
mixed development is not a certain tool to revitalize
the housing market in the city. It involves tradeoffs,
efficient planning, and externalities. The seemingly
small positive net property value gains from the MP
project are not likely enough to justify the huge state
funds plugged in the downtown revitalization. How-
ever, these results are only for short-term effects; if in
the long term the fully revitalized downtown brings
prosperity back and increases property tax receipts
for the city, using ED for redevelopment with a cred-
ible policy announcement might be worthwhile (effi-
cient planning). Future research could also add data
after these projects are complete.
Acknowledgements
I would like to thank Jeremy Groves for his ex-
tremely helpful guidance. I am also grateful for
Richard Cebula and Rebekka Dudensing’s valuable
comments.
Effect of Eminent Domain on Property Values 185
References
Basile, R., M. Durban, M. Minguez, J. Montero, and
J. Mur. 2014. Modeling regional economic dy-
namics: spatial dependence, spatial heterogeneity
and nonlinearities. Journal of Economic Dynamics
and Control 48: 229-245.
Brenner, M. 1977. The effect of model misspecifica-
tion on tests of the efficient market hypothesis.
Journal of Finance 32(1): 57-66.
Benson, B., and M. Brown. 2010. Eminent domain for
private use: is it justified by market failure or an
example of government failure. In Property
Rights: Eminent Domain and Regulatory Takings Re-
examined (pp. 149-170). Palgrave Macmillan.
Bin, O., and M. Filho. 2001. Estimation of hedonic
price functions via additive nonparametric re-
gression. Journal of Economics Literature 20(1): 93-
114.
Carpenter, D.M., and Ross, J. K. 2009. Testing
O’Connor and Thomas: does the use of eminent
domain target poor and minority communities?
Urban Studies 46(11): 2447-2461.
Carpenter, D.M., and J. K. Ross. 2010. Do restrictions
on eminent domain harm economic develop-
ment. Economic Development Quarterly 24(4): 337-
351.
Cebula, J.R. 2009. The hedonic pricing model ap-
plied to the housing market of the city of Savan-
nah and its Savannah Historic Landmark District.
The Review of Regional Studies 39(1): 9-22.
City of Rochester. (2012). Preliminary Findings. Bu-
reau of Planning and Zoning. Retrieved from
file:///C:/Users/pyu/Dropbox/2014/Musci/SP-014-11-
12_preliminary_findings.PDF.
Cleveland, S.W., and J.S. Devlin. 1988. Locally
weighted regression: an approach to regression
analysis by local fitting. Journal of the American
Statistical Association 83(403): 596-610.
Collins, W., and K. Shester. 2010. The Economics Ef-
fects of Slum Clearance and Urban Renewal in
the United States. Working Paper No.10-W13,
Department of Economics, Vanderbilt University.
Haupt, H., J. Schnurbus, and R. Tscherning. 2010.
On nonparametric estimation of a hedonic price
function. Journal of Applied Econometrics 25(5):
894-901.
Kerekes, B.C., and F. Gulf. 2011. Government tak-
ings: determinants of eminent domain. American
Law and Economics Review 13(1): 201-219.
Liston, D. 2013. Summary of Berman V. Parker. OLR
Research Report. Retrieved from
www.cga.ct.gov/2013/rpt/2013-R-0211.htm.
Li, T., and C. Mei. 2013. Estimation and inference for
varying coefficient partially nonlinear models.
Journal of Statistical Planning and Inference 143:
2023-2037.
Lancaster, J.K. 1966. A new approach to consumer
theory. The Economics of Leisure 74(2): 394-419.
McMillen, P.D., and L.C. Redfearn. 2010. Estimation
and hypothesis testing for nonparametric he-
donic house price functions. Journal of Regional
Science 50(3): 712-733.
Munch, P. 1976. An economic analysis of eminent
domain. Journal of Political Economy 84(3): 473-
498.
Parmeter, C., D. Henderson, and S. Kumbhakar.
2007. Nonparametric estimation of a hedonic
price function. Journal of Applied Econometrics
22(3): 695-699.
Powell, J. 1994. Estimation of semiparametric mod-
els. In R.F. Engle and D. McFadden (eds.) Hand-
book of Econometrics, Volume IV. Amsterdam:
North Holland. Chapter 41, pp. 2443-2521.
Rosen, S. 1974. Hedonic prices and implicit markets:
product differentiation in pure competition. The
Journal of Political Economy 82(1): 34-55.
186 Yu
Appendix.
Figure A1. Midtown Plaza (MP) Revitalization Project Map (source: www.cityofrochester.org).
Figure A2. Brooks Landing (BL) Revitalization Project Map (source: http://rocwiki.org/Brooks_Landing_Project).
Effect of Eminent Domain on Property Values 187
Table A1. Full results for semiparametric model with 3% bandwidth.
Variables
Estimates
Variables
Estimates
Physical characteristics
Per Square Footage
0.018 (0.0001)***
# of Rooms
0.01 (0.003)***
Additional Living Area
0.00007 (0.0002)***
# of Bedrooms
0.005 (0.005)
Fished Basement
0.0001 (0.00004)**
# of Bathrooms
0.003 (0.005)
Finished Recreation Room
0.00002 (0.00001)
Air Conditioner
0.02 (0.005)***
Basement Garage
0.006 (0.0122)
Age
-0.001 (0.0001)***
# of Stories
0.073 (0.006)***
# of Fireplaces
0.005 (0.005)
House styles
Ranch
0.056 (0.035)
Row
0.045 (0.034)
Split Level
0.105 (0.038)**
Log Cabin
0.082 (0.074)
Cape Cod
0.077 (0.034)*
Contemporary
0.087 (0.056)
Colonial
0.052 (0.034)
Duplex
-0.073 (0.052)
Old Style
-0.083 (0.07)
Mansion
0.025 (0.036)
Cottage
-0.4 (0.073)***
Township
-0.103 (0.037)**
House grades
Poor Grade
-0.647 (0.057)***
Average Grade
-0.089 (0.023)***
Poor Kitchen
-0.046 (0.009)***
Average Kitchen
-0.01 (0.006)
Average Bath
0.049 (0.008)***
Good Bath
0.062 (0.01)***
Poor Interior
-0.413 (0.063)***
Fair Interior
-0.007 (0.057)
Normal Interior
0.108 (0.057)
Good Interior
0.114 (0.056)*
Poor Exterior
-0.619 (0.087)***
Fair Exterior
-0.251 (0.058)***
Normal Exterior
-0.102 (0.058)
Good Exterior
0.096 (0.057)
House materials
Alum/Vinyl
0.018 (0.0001)***
composition
0.01 (0.003)***
Concrete
0.00007 (0.0002)***
Stucco
0.005 (0.005)
House location
School Distance
0.056 (0.035)
CBD Distance
0.045 (0.034)
Library Distance
0.105 (0.038)**
Recreation Distance
0.082 (0.074)
Annual trends
Year 2000
0.331 (0.043)***
Year 2001
0.281 (0.042)***
Year 2002
0.252 (0.042)***
Year 2003
0.234 (0.042)***
Year 2004
0.200 (0.042)***
Year 2005
0.178 (0l041)***
Year 2006
0.133 (0.034)***
Year 2007
0.085 (0.026)***
Year 2008
0.078 (0.026)**
Year 2009
0.015 (0.026)
Year 2010
0.043 (0.024)
Year 2011
0.035 (0.015)
Year 2012
-0.012 (0.015)
Year 2013
Omits
Note: standard errors in the parentheses; * 5% significance, ** 1% significance, ***0.1% significance. N =19,707.
... Public interest includes construction of roads, parks, schools, hospitals, government properties and anything which falls under the purview of social investment (Yu, 2015). Where sometimes the state revokes one's title and transfers same to another private entity, it is pertinent to note that public interest does not necessarily mean interest of the state. ...
... Kelo v. New London Yu (2015) provides the case of Kelo v. New London, in which the drug company, Pfizer, built a new plant in 1998 in New London, Connecticut. Their aim was to bring in more jobs and government revenue. ...
... The key observation from these cases is that growing and extensive use of eminent domain has caused widespread concern which has negated its intended goal (Yu, 2015). Becher (2014) argues that in Philadelphia, about 250 such formal cases exist with residents fighting back against the land-grabbers. ...
Article
Full-text available
This paper examines the role of public relations (PR) in the peaceful implementation of eminent domain in Nigeria. Eminent domain is the right of government or its agents to expropriate private property from their owners for public use, with payment of compensation. Globally, such development plans are hardly implemented peacefully due to perceived injustice, misinterpretation and the refusal of owners to let go of their property. The unrest and destruction of lives and property following such initiatives suggest that despite its prospects, eminent domain, to a large extent, may negatively impact on community life and tamper with the opportunities of minorities. Because peace amounts to harmony in personal relations, and public relations is the state of maintaining a harmonious relationship between an organization and its publics, the paper recommends that an understanding of the role of public relations in the peaceful implementation of eminent domain is crucial if the efforts being made presently in Nigeria (e.g., in Enugu State) to implement this policy will yield the desired result. Keywords: peace, public relations, eminent domain, development
... Yet, its significance cannot be overestimated, especially where such isolated cases become a norm, and culminating in a network of disgruntled clusters. Studies have alluded such disgruntlements to menaces of land conflicts and disputes, dampened land transactions, frustrated development schemes, discouraged real estate investments, litigations Hughan, 2018;Yu, 2015). Obviously, with such occurrences, pathways to the accomplishment of the acquiring authority's goals and intentions would be subject to blockades, and thus reinforces the philosophy of getting the expropriation victim on board with the acquisition process. ...
Article
Full-text available
This study applies a hedonic pricing model to the housing market of the City of Savannah, Georgia. The Savannah Historic Landmark District is located both in and adjacent to downtown Savannah. Of the 2,888 single-family homes for the period 2000-2005 for which data are available, 591 are located in the Historic District. The model of the real sales price of a single-family house in the City of Savannah environment reveals it is positively affected by the number of bathrooms, fireplaces, bedrooms, stories in structure, garage car spaces, square feet of finished living space, the presence of a deck, a private courtyard, a pool and/or hot-tub, an exterior construction of brick or stucco, the presence of an underground sprinkler system, and whether the house was new. Six spatial control variables are considered. Locations across from or adjacent to open space carry premia, as do locations on cul-de-sacs or lakes. Corner properties do not. In addition, proximity to an apartment complex is capitalized as a negative quantity, as do locations on a busy street. The real sales prices of residential properties that closed during May or July tend to be higher. In addition, houses designated as a national historical monument tend to carry with them a modest price premium, as do properties that are simply located within the Savannah Historic Landmark District.
Article
Varying coefficient partially nonlinear model Profile nonlinear least squares estimation Local linear fitting Generalized likelihood ratio test Bootstrap a b s t r a c t In this paper, we extend the varying coefficient partially linear model to the varying coefficient partially nonlinear model in which the linear part of the varying coefficient partially linear model is replaced by a nonlinear function of the covariates. A profile nonlinear least squares estimation procedure for the parameter vector and the coefficient function vector of the varying coefficient partially nonlinear model is proposed and the asymptotic properties of the resulting estimators are established. We further propose a generalized likelihood ratio (GLR) test to check whether or not the varying coefficients in the model are constant. The asymptotic null distribution of the GLR statistic is derived and a residual-based bootstrap procedure is also suggested to derive the p-value of the GLR test. Some simulations are conducted to assess the performance of the proposed estimating and testing procedures and the results show that both the procedures perform well in finite samples. Furthermore, a real data example is given to demonstrate the application of the proposed model and its estimating and testing procedures. & 2013 Elsevier B.V. All rights reserved.
Article
Spatial modeling of economic phenomena requires the adoption of complex econometric tools, which allow us to deal with important methodological issues, such as spatial dependence, spatial unobserved heterogeneity and nonlinearities. In this paper we describe some recently developed econometric approaches (i.e. Spatial Autoregressive Semiparametric Geoadditive Models), which address the three issues simultaneously. We also illustrate the relative performance of these methods with an application to the case of house prices in the Lucas County.
Article
In dissenting from the US Supreme Court’s 2005 Kelo decision upholding the use of eminent domain for private-to-private transfers of property, Justices O’Connor and Thomas asserted, based on the history of urban renewal, that eminent domain for private development would disproportionately hurt poor and minority communities. This study uses US census data and a sample of redevelopment project areas using or identified for the use of eminent domain to test the assertions of Justices O’Connor and Thomas. Results reveal that such project areas are, in fact, disproportionately populated by those who are poor, minority and less educated.
Article
The Housing Act of 1949 established a federally subsidized program that helped cities clear areas of existing buildings for redevelopment, rehabilitate deteriorating structures, complete comprehensive city plans, and enforce building codes. The program ended in 1974, but not before financing over 2,100 urban renewal projects and generating great controversy. We use an instrumental variable strategy to estimate the program’s effects on city-level outcomes. The estimates are generally positive and economically significant and are not driven by differential changes in cities’ demographic composition. We caution that the results do not imply that the program was an equitable or optimal approach to dealing with central-city problems.
Article
Locally weighted regression, or loess, is a way of estimating a regression surface through a multivariate smoothing procedure, fitting a function of the independent variables locally and in a moving fashion analogous to how a moving average is computed for a time series. With local fitting we can estimate a much wider class of regression surfaces than with the usual classes of parametric functions, such as polynomials. The goal of this article is to show, through applications, how loess can be used for three purposes: data exploration, diagnostic checking of parametric models, and providing a nonparametric regression surface. Along the way, the following methodology is introduced: (a) a multivariate smoothing procedure that is an extension of univariate locally weighted regression; (b) statistical procedures that are analogous to those used in the least-squares fitting of parametric functions; (c) several graphical methods that are useful tools for understanding loess estimates and checking the assumptions on which the estimation procedure is based; and (d) the M plot, an adaptation of Mallow's Csubp/sub procedure, which provides a graphical portrayal of the trade-off between variance and bias, and which can be used to choose the amount of smoothing.
Article
After the U.S. Supreme Court upheld in the Kelo decision the use of eminent domain for private-to-private transfer of property for economic development, public outrage was followed by attempts to restrict such use of eminent domain. Opponents of restrictions predicted dire consequences for state and local economies. This study considers whether restricting the use of eminent domain for economic development results in negative economic effects. The authors examine economic indicators before and after legislative or judicial restrictions on eminent domain across all states and between states based on the type of legislative/judicial change. Results indicate that there appear to be no negative economic consequences resulting from limiting the use of eminent domain when examining economic indicators before and after legislative/judicial change. Adopting either moderate or major eminent domain restrictions appears to create no economic ill effects when analyzing differences in trends based on the type of legislation passed or scope of judicial decision.
Article
There is a well-documented link between institutions of secure property rights and economic development. The increasing use of eminent domain to transfer property from one private owner to another for private benefit, upheld in the Supreme Court's controversial Kelo decision, undermines the security of property rights. While previous research examines the effect of eminent domain use, this paper explores which factors explain varying levels of eminent domain use for private benefit across states. The author finds that corruption, election of state Supreme Courts, federalism, and economic freedom are important determinants of eminent domain use for private benefit. Copyright 2011, Oxford University Press.
Article
In contrast to the rigid structure of standard parametric hedonic analysis, nonparametric estimators control for misspecified spatial effects while using highly flexible functional forms. Despite these advantages, nonparametric procedures are still not used extensively for spatial data analysis due to perceived difficulties associated with estimation and hypothesis testing. We demonstrate that nonparametric estimation is feasible for large datasets with many independent variables, offering statistical tests of individual covariates and tests of model specification. We show that fixed parameterization of distance to the nearest rapid transit line is a misspecification and that pricing of access to this amenity varies across neighborhoods within Chicago. Copyright (c) 2010 Wiley Periodicals, Inc.