The economic value of walkable neighborhoods

Abstract

This study investigated how the benefits of a walkable neighborhood were reflected in the American real estate market by examining the economic values of urban environmental factors supporting walking activities. Property values were used as a proxy measure for economic value and analyzed in relation to land use characteristics that have been known to correlate with walking at the neighborhood scale. Four aspects of the built environment supporting walking were included in the analyses: development density, land use mix, public open space and pedestrian infrastructure. Hedonic models were employed where the property value was regressed on the measures of the four sets of correlates of walking in a neighborhood. Models were estimated for four land use types – single-family residential, rental multi-family residential, commercial and office. The findings did not support previous arguments that increasing density weakens the quality of a neighborhood. To the contrary, the positive association of higher development density with the value of single-family residential properties detected in King County suggested that high development density might increase surrounding property values. The pedestrian infrastructure and land use mix significantly contributed to increases in rental multi-family residential property values. Higher development density with higher street and sidewalk coverage were also favored by retail service uses. In relation to land use mix, mixing retail service uses and rental multi-family residential uses helped make rental housings more attractive.

Full text

AUTHOR COPY
Original Article
The economic value of walkable neighborhoods
Dong Wook Sohn
a,
*, Anne Vernez Moudon
b
and Jeasun Lee
c
a
Department of Urban Design & Planning, Hongik University, Seoul, Korea.
E-mail: sohndw@gmail.com
b
Department of Urban Design & Planning, University of Washington, Seattle, USA.
c
Department of Urban Planning & Engineering, Yonsei University, Seoul, Korea.
*Corresponding author.
Abstract This study investigated how the benefits of a walkable neighborhood were reflected in the
American real estate market by examining the economic values of urban environmental factors supporting
walking activities. Property values were used as a proxy measure for economic value and analyzed in relation
to land use characteristics that have been known to correlate with walking at the neighborhood scale. Four
aspects of the built environment supporting walking were included in the analyses: development density, land
use mix, public open space and pedestrian infrastructure. Hedonic models were employed where the property
value was regressed on the measures of the four sets of correlates of walking in a neighborhood. Models were
estimated for four land use types – single-family residential, rental multi-family residential, commercial and
office. The findings did not support previous arguments that increasing density weakens the quality of a
neighborhood. To the contrary, the positive association of higher development density with the value of single-
family residential properties detected in King County suggested that high development density might increase
surrounding property values. The pedestrian infrastructure and land use mix significantly contributed to
increases in rental multi-family residential property values. Higher development density with higher street and
sidewalk coverage were also favored by retail service uses. In relation to land use mix, mixing retail service uses
and rental multi-family residential uses helped make rental housings more attractive.
URBAN DESIGN International (2012) 17, 115–128. doi:10.1057/udi.2012.1; published online 4 April 2012
Keywords: walking; mixed land use; neighborhood; urban design
Introduction
The concept of walkable neighborhood is at the
core of such contemporary urban theories as
Smart Growth and New Urbanism. Proponents
of these approaches argue that building walkable
communities will counteract the negative effects
of urban sprawl and alleviate traffic congestion,
air pollution, and the destruction of natural
environments in and near metropolitan areas
(Paumier, 2004). Researchers have claimed that
combining residential and commercial land uses
in walkable neighborhoods will help produce
such social benefits as affordable housing (Hess
and Lombardi, 2004; Handy, 2005), cleaner air and
water (Shapiro et al, 2002) and lower automobile
dependency (Dorn, 2004).
Other researchers have expressed doubts about
the viability of walkable neighborhoods as an
alternative to sprawl. They assert that consumers
do not care about such social benefits, simply
favoring higher standards of living such as more
room and spacious yards filled with trees and
shrubs, and that traditional auto-oriented sub-
urban developments have successfully competed
in the market place (Holcombe, 1999). They add
that citizens are so accustomed to auto-oriented
suburban space that the market for walkable
urban settings is limited.
This study addressed these issues by examining
the economic value of walking friendly urban
environments. Property values were used as a
proxy measure for economic value and analyzed
in relation to land use characteristics that have
been shown to correlate with walking at the
neighborhood scale. The approach assumed that
assessed property values were associated with
consumers’ willingness to pay, and thus reflected
r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128
www.palgrave-journals.com/udi/

AUTHOR COPY
consumers’ like or dislikes of neighborhood
features supporting walking.
Four aspects of the built environment support-
ing walking in a neighborhood were included in
the analyses: development density, land use mix,
public open space and pedestrian infrastructure.
Hedonic models
1
were employed where the
property value was regressed on the measures
of the four sets of correlates of walking in a
neighborhood. Models were estimated for four
land use types – single-family residential, rental
multi-family residential, commercial and office.
Background
A walkable neighborhood: Built environment
correlates of walking
A large body of literature in transportation and
health has documented walkable neighborhoods
as being characterized by medium- to high-
density residential development, a mixture of
land uses that are close together to reduce or
eliminate the need to drive between routine
activities. Specifically, improved accessibility to
retail stores, transit and recreational areas has
been associated with more walking (Katz, 1994;
Crane and Crepeau, 1998; Limanond and Niemeier,
2003; Morrow-Jones et al, 2004; Giles-Corti et al,
2005; Song, 2005; Moudon et al, 2007). Saelens and
Handy (2008) conducted a systematic review of
articles examining the link between built environ-
ment and walking. Their review identified several
correlates of walking with sufficient evidence.
According to the results, development density,
mixed land use and the pedestrian transportation
infrastructure were found to be associated with
walking. In contrast, few studies reported asso-
ciations between parks and open space and
walking (Clifton and Dill, 2005; Giles-Corti et al,
2005; Zlot and Schmid, 2005), whereas others
found no evidence of such a relationship (Bopp
et al, 2006; Lee and Moudon, 2006).
Among these various factors related to walking,
this study focused on the four sets of correlates of
walking shown in Table 1.
Economic value of walkable neighborhoods
Research on the economic benefits of walkable
neighborhoods has remained limited to how
selected aspects of walkability might be acknowl-
edged in the real estate market.
Residential density has been considered to
undermine the quality of a neighborhood, and
thus to decrease the values of residential proper-
ties. Most studies confirmed that the residential
real estate market did not favor urban environ-
ments with high development density (Schwanen
and Mokhtarian, 2004). On the basis of a preference
survey in which consumers were conditioned by
the respondent’s stage in the life cycle, length of
residence and socio-psychological factors, Talen
(2001) reported that Americans preferred low-
density suburban development over urban life.
Song and Knaap (2003) also noted that density –
measured by single-family residential dwelling
unit density and population density – was
negatively related to the housing values in the
Portland, OR, region, suggesting consumers’
preference for lower density single-family neigh-
borhoods.
Other research found evidence that certain
groups of people were willing to pay a premium
to live in compact environments. A comparative
study conducted by Eppli and Tu (1999) exam-
ined four New Urbanist communities and found
that there was a price premium of about 15 per
cent to live in a New Urbanist (or neo-traditional)
community over a comparable conventional sub-
urban subdivision. Lang et al (1997) claimed that
such a phenomenon represented the existence of
dual housing markets: one for conventional low-
density suburbs, and one for cities and denser
suburbs.
While the positive effects of mixing land uses
on walking trips (that is, by reducing the travel
distances to destinations and by increasing
diversity of amenities) have been demonstrated,
the findings of research on the market reaction to
land use mix have been inconclusive. For exam-
ple, Grether and Mieszkowski (1980) conducted
market experiments designed to produce mea-
sures of the effects of nonresidential land uses on
the prices of nearby dwellings, but found no
systematic relationship between nonresidential
land use and housing prices. Later, Cao and Cory
(1981) examined the relationship between non-
residential uses and residential property values.
They also noted that the effect of non-residential
activity on property value was indeterminate and
influenced by the external effects generated from
the nonresidential activities. Later, Sohn and
Moudon (2008) analyzed the effects of land use
mix on the value of office properties in King
Sohn et al
116 r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128

AUTHOR COPY
County, WA. This study found that office property
values in the office cluster decreased as the
proportion of surrounding land in retail uses
increased, which suggested that urban planning
policies seeking to mix uses in employment centers
werenotsupportedbycurrentmarkettrends.
A few studies have tried to measure the impact
of pedestrian infrastructure on the values of
nearby properties. Asabere (1990) estimated the
effects of neighborhood street patterns on housing
values using data from Halifax, Nova Scotia.
Using two categories of streets – cul-de-sac and
grid – the study found that cul-de-sacs generated
a 29 per cent price premium over the grid street
pattern, supporting the hypothesis that cul-de-
sacs attracted premium values. In contrast, Plaut
and Boarnet (2003) studied New-Urbanism-style
neighborhoods characterized by a grid street
pattern and street-oriented neighborhood design
and found that housing sales data attributed a
significant price premium to New-Urbanism-style
neighborhoods.
Regarding public open space, studies generally
confirmed that residential property values in-
creased with the nearby presence of open space.
For example, Irwin (2002) estimated the marginal
value of different open space attributes using a
hedonic pricing model with residential sales data
Table 1: The correlates of walking for the investigation
The correlates of walking Relations with walking Reference Findings
Development density In areas with higher density,
destinations can be closer
meaning that the needs for
driving decrease
Gauvin et al, 2005 Positive relation of walking to
density of destinations
Clifton and Dill, 2005 Greater walk trips with
increasing housing density
Khattak and Rodriquez (2005) Higher walking trips in
neo-traditional versus
conventional neighborhood
Land use mix By putting various amenities
in close proximity to one
another, walking becomes
viable
Lee and Moudon, 2006 Positive relations between
walking and close proximity to
a grocery store, eating places
and retails
De Bourdeaudhuij et al, 2005 Walking for transport related to
higher land use mix
Hoehner et al, 2005 Walking for transport related to
greater perceived and objective
land use mix
Public open space It supports walking by
offering opportunities for
recreation and improving
environmental quality of a
neighborhood
Giles-Corti et al, 2005 High walking more likely among
individuals with shorter
distance to highly attractive
and large public open space
Clifton and Dill, 2005 Greater walk trips with greater
park access (men only)
Zlot and Schmid, 2005 Walking for transport related to
parkland acreage
Pedestrian
infrastructure
Small street blocks shorten
distances between
activities, making walking
practical. Wide sidewalks
and safe opportunities to
cross streets obviously
support walking by
creating safe environments
Cao and Cory, 1981 Higher walking to store
frequency related to route
comfort and pedestrian
connections
Li et al, 2005 Higher walking activity among
residents in neighborhoods
with more street intersections
Fulton et al, 2005 Higher walk for transport to
school among residents of
areas with sidewalks
The economic value of walkable neighborhoods
117r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128

AUTHOR COPY
from central Maryland. He found that surround-
ing open space significantly influenced the resi-
dential sales price of houses. More recently,
researchers (Laverne and Winson-Geideman,
2003; Wolf, 2003) noted that commercial proper-
ties might benefit from attributes related to open
space such as a quality landscape and greenery.
In summary, previous studies found that
aspects of land use and the built environment
associated with walking were valued in the real
estate market. Evidence of such valuation, how-
ever, remained mixed. More comprehensive and
detailed analyses would be needed to understand
the spatial characteristics of walkable built envir-
onments that can be supported in the market.
This study sought to add to the evidence with
an individual property level analysis of the
relationships between both residential and com-
mercial property values and the primary physical
characteristics of environments that support
walking.
Methods
Research design
This research examined the economic value of
neighborhood walkability. It was conducted at the
individual level for properties located in the
urban growth area (UGA) of King County, WA.
First established in 1985, King County’s UGA has
been used to limit growth to areas with an
existing infrastructure for facilities and services.
The City of Seattle and major suburban cities
of the region such as Bellevue, Kirkland and
Redmond are located within the boundary of
UGA. More than 93 per cent of new housing in
the region has been built in the urban growth area
from 1994 to 2001 based on building permits
issued by the cities and King County.
For each individual property, the measured
characteristics of walkability included dev-
elopment density, land use mix, public open
space and pedestrian infrastructure. Economic
value was measured by the assessed property
value of individual properties in four different
land use types: single-family residential, rental
multi-family residential, retail service and office
uses.
The use of readily available data at the parcel
level
2
(the finest resolution of the built environ-
ment) with the GIS based neighborhood analysis
techniques helped bypass the modifiable areal
unit problem (MAUP)
3
derived from data aggre-
gation. The potential model bias because of the
spatial autocorrelation
4
in the property value data
was checked using the Moran’s Index test.
5
Data and sampling
The parcel level property related data sets, which
were the primary data for the analysis, came from
two primary sources: (i) a parcel level data set in
GIS obtained from Washington State Geospatial
Data Archive provided basic attributes of indivi-
dual parcels, such as parcel boundary and size.
(ii) The parcel level tax assessment data set from
the King County Department of Assessments
included the land’s and the building’s assessed
value information in addition to the detailed
description of the parcels and their development
status such as land use types, physical attributes
of buildings (for example, number of bedrooms,
number of bathrooms, fireplace, floor area ratio
(FAR), building square footage, building quality
and so on) and other miscellaneous information
on individual parcels. Both data sets were
obtained in 2004.
In addition, a variety of data sets in GIS
were used for measuring physical and socio-
demographic characteristics of a neighborhood.
Table 2 shows the list of the GIS data sets with
the brief descriptions.
The sample parcels for this study were selected
from the individual parcel records available
within the 500 square mile (1310 km
2
) Urban
Growth Boundary (UGB) area of King County,
Washington. The county’s UGB served to define
the urbanized area because it contained most of
the developed land (with 95 per cent of the
residential units). The sample parcels were ran-
domly selected using a GIS program (Arcview
GIS Arcscript – simplers.avx designed by William
Huber). Among all parcels in the King County
UGB 2289 samples of single-family residential,
837 samples of rental multi-family residential, 738
samples of retail service and 586 samples of office
parcels were randomly selected by the sampling
process. The distribution of the sampled parcels
for the four land uses is illustrated in Figure 1.
Dependent variable
Two dependent variables – (i) the logged total
assessed value of land and improvement per
parcel (for single-family residential, retail service
Sohn et al
118 r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128

AUTHOR COPY
and office models), and (ii) the logged total
assessed value of land and improvement per
residential unit (for rental multi-family residential
model) – were used as a proxy for a property’s
market value.
In Washington State, assessment for tax pur-
poses required establishing the full market value
of a parcel of land and its improvements; there-
fore, the total valuation of the properties in the
data represents 100 per cent of the fair market
values (Department of Assessments, King County,
WA). The County’s analytical method of property
assessment has been known to be reliable to
capture the market value of properties (Clapp and
Giaccotto, 1992; Janssen and Soderberg, 1999).
Measuring neighborhood walkability
Defining the spatial boundary of a neighborhood
Neighborhood boundaries needed to correspond
to the spatial range of people’s walking behaviors.
Several studies provided useful information for
determining the spatial boundary of a neighbor-
hood affecting people’s walking and transit
behavior. Ewing (1995) reported that people
walked an average of 0.3 miles for shopping
trips, 0.28 miles for accessing transit stops and
family businesses based on 1990 National Perso-
nal Transportation Survey data and an average
walking speed of 3.16 mph. A case study con-
ducted by the Federal Highway Administration
(FHWA) to build transit catchment areas for
determining walking accessibility to transit stops
used one-quarter mile as the maximum distance
that riders feel convenient to walk (FHWA, 2002).
Other studies examining pedestrian travel pat-
terns also used one-quarter mile or a 5 to 10 min
walking distance for defining the extent of a
neighborhood (Rood, 2000; Dill, 2003). Overall,
studies confirmed that walking distances ranged
between 0.25 and 0.3 miles. Given the focus of this
study on measuring correlates of walking in a
neighborhood and on assessing the impacts of
these correlates on property values, a one-quarter
mile radius airline buffer around the sampled
parcels was adopted as the spatial boundary of a
neighborhood (Figure 2).
Using a circular buffer around a sampled parcel
as the spatial unit of analysis had the advantage
of avoiding data redundancy derived from the
use of a larger predefined spatial unit such as a
census tract or transportation analysis zone as the
boundary of the spatial analysis (Sohn, 2007). For
example, given that two samples were located in
the same census tract (first figure of Figure 3),
using a census tract as the boundary of a
neighborhood produced the same value of a
neighborhood measure for the two samples as
they had the same neighborhood boundary. On
the other hand, defining a neighborhood as the
buffer area around each sample (second figure of
Figure 3) produced a unique value of a neighbor-
hood measure for each of the samples, thus
Table 2: List of data sets in GIS
Data set Data type Source Description
parks.shp Polygon shape file King County Department of
Transportation (2004)
KDelineates the locations and boundaries of
public parks in King County, along with
their names
streets.shp Line shape file King County Department of
Transportation (2004)
KThe centerlines of the street network in
King County
Kprovide detailed information on the status
of individual street segments (for example,
length, width, road class and so on).
sidewalks.shp Line shape file Puget Sound Regional County (2000) KProvides information on the location and
length of sidewalks available in the region.
busstops.shp Point shape file King County Department of
Transportation (2000)
KProvides the locations of bus stops
available in the region.
urban centers.shp Polygon shape file Puget Sound Regional County (2004) KDelineates the boundary of Seattle
downtown and urban centers designated in
PSRC Vision, 2020
Block group level
census data set
Polygon shape file US Census Bureau (2000) KThe block group level data of average total
household income, median age of
households and per cent of non-White
households
The economic value of walkable neighborhoods
119r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128

AUTHOR COPY
clearly differentiating between them based on
their neighborhood boundaries.
Measuring built environment correlates of walking
The independent variables of interest in this
study, which are development density, land use
mix, open space and pedestrian infrastructure
(Table 3), were measured in different ways by
different researchers. Various forms of density
measures were extensively used in the travel
behavior research because density was one of
the core characteristics of built environments
(Cervero, 2002). Most frequently, density was
estimated in the form of residential density
(for example, Giuliano and Small, 1993; Rajamani
et al, 2003) or employment density (Cervero, 1996;
Kockelman, 1997; Anderson and Bogart, 2001). A
major shortcoming of these two measures was
that they were only able to capture the density of
specific land uses (residential or commercial
uses). On the other hand, development density
Figure 1: Maps of the distributions of sampled parcels.
Sohn et al
120 r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128

AUTHOR COPY
was, by definition, capable of capturing a neigh-
borhood’s overall physical density level regard-
less of land use types. For this reason, the average
FAR in a neighborhood was used in the models as
a measure of development density.
Mixed land use and open space were also
important built environment correlates of walk-
ing. As they were associated with land use
patterns, the characteristics of these two correlates
could be captured by measures describing the
type and intensity of land uses. In addition, they
were also associated with proximity to potential
destinations – mixed land use (including open
space) meant that destinations were within close
proximity – proximity to destinations was known
to be the most consistent correlate of walking
(Saelens and Handy, 2008).
Most frequently used measures associated with
land use mix were heterogeneity and diversity
measures. An entropy index and a dissimilarity
index estimated the degree of uniformity of land
uses or the degree to which different land uses
came into contact with one another. The short-
coming of these measures was that they were not
able to capture the difference of specific land
use composition such as a mixed land use of
30 per cent residential and 70 per cent retail versus
a mix of 30 per cent in retail and 70 per cent in
residential uses (Hess et al, 2001; Krizek, 2003). As
an alternative way, this research estimated the
proportion of specific land use area in a neighbor-
hood. This measure enabled to conduct more
detailed analysis for the degree of land usage
for specific uses and their interactions. In addi-
tion, average Euclidean distance from sampled
1/4 miles
sampled parcels
Figure 2: Illustration of the suggested spatial unit of analysis for the neighborhood analysis.
circular buffers
around samples
a census tract
sample 1
sample 2
sample 1
sample 2
Figure 3: Comparison of the spatial units of analysis (census
tract versus a circular buffer around a sample).
The economic value of walkable neighborhoods
121r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128

AUTHOR COPY
properties to surrounding uses were estimated as
it was the simplest and the most extensively used
measure for proximity to destinations (for example,
Komanoff and Roelofs, 1993; Smith and Butcher,
1994; Handy, 1996; Talen, 2003).
The characteristics of pedestrian infrastructure
were measured in the perspective of its transit
accessibility and network connectivity. For mea-
suring accessibility to transit facilities such as bus
stops, train stations and trails, Euclidean distance
has been commonly used (Kitamura et al, 1997; Kim
and Ulfarsson, 2004; Song, 2005). This research
employed it to measure accessibility from a
sampled property to a bus stop. Although net-
work distance, which is defined as the length of
walkways from the pedestrian’s origin to a
destination, may be a more accurate measure for
pedestrian accessibility, it was not used in this
study as a large portion of the study area included
low density neighborhoods with poor sidewalk
infrastructure. When estimating pedestrian acces-
sibility in these urban settings, Euclidean distance
seemed to reflect pedestrian accessibility more
effectively than network distance, considering
pedestrians’ tendency to take shorter routes by
walking through undeveloped lands or along
streets with no sidewalks.
The characteristics of street configuration
proved to significantly affect walking and were
extensively investigated in transportation research.
Measures such as the linear length of streets,
street density, cul-de-sac density and intersection
density were developed for capturing the char-
acteristics of street configuration in the literature
(AultmanHall et al, 1997; Cervero and Kockelman,
1997; Srinivasan, 2001; Song and Knaap, 2004).
This research employed two street density mea-
sures for sidewalks and local streets (total length
of streets or sidewalks per acre). Although
including more discriminative measures describ-
ing route directness would lead to a more detailed
examination of the effects of street configurations
on property values, it called for a non-systematic
analysis (that is, visual inspection and partially
subjective decision making) of each sample.
Considering the large sample size, this was
almost impracticable and therefore these mea-
sures were not estimated.
Control variables
The effect of neighborhood walkability was
expected to be confounded by other factors
Table 3: The list of the measure for the correlates of walking
Correlates of walking Measurement Unit of
measures
Description
Development density Average FAR — Average floor area ratio of all developed
parcels in a neighborhood
Land use mix Ratio of MF area % Ratio of the area of multi-family residential
parcels in a neighborhood to the total area
of a neighborhood
Ratio of retail service area % Ratio of the area of retail service parcels in a
neighborhood to the total area of a
neighborhood
Ratio of office area % Ratio of the area of office parcels in a
neighborhood to the total area of a
neighborhood
Average distance to MF uses ft Average distance to the MF parcels in a
neighborhood
Average distance to retail
service uses
ft Average distance to the retail-service parcels
in a neighborhood
Average distance to office uses ft Average distance to the office parcels in a
neighborhood
Public open space Distance to public open space ft Distance to the closest public park in a
neighborhood
Pedestrian
infrastructure
Distance to a bus stop ft Distance to the closest bus stop in a
neighborhood
Street density ft/acre Ratio of the length of streets in ft to the acreage
of a neighborhood
Sidewalk density ft/acre Ratio of the length of sidewalks in ft to the
acreage of a neighborhood
Sohn et al
122 r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128

AUTHOR COPY
determining the economic value of a property.
Fundamental attributes of property value such as
lot size, age of building and building square footage
were considered in the model. Regional location
factors were also taken into account, and measured
by: (i) the distance to Seattle downtown, and (ii) the
distance to the closest urban center.
6
In addition,
three socio-demographic variables were included
based on US Census block group data: household
income, median household age and percentage of
non-White residents in the neighborhood. Each of
these measures was calculated by averaging the
values of the census data overlapping each prop-
erty’s designated neighborhood (that is, quarter-mile
buffer), accounting for the proportion of the block
group areas contained within the neighborhood.
V¼X
n
1
ða1v1þa2v2þ...... anvnÞ
where, V: socio-demographic measure of a neigh-
borhood; a: the ratio of the area of the census block
group to the total area of a neighborhood; v:the
value of the socio-demographic measure from a
unit of census block group; n:totalnumberof
census block groups overlapping a neighborhood
The hedonic model
Variables capturing the correlates of walking,
control variables and dependent variables for
the four sets of hedonic model (MF, RMF, retail
service and office) are listed in Table 4.
Moran’s Index test was conducted to ensure
that the spatial autocorrelation of sampled prop-
erty values was properly controlled in the model.
Model Results
The sample size ranged from 586 office properties
to 2289 single-family properties. The results of
Moran’s Index test showed that the spatial
autocorrelation of the residuals for all models
was marginal (Moran’s Indexes were less than
0.01 with z-scores less than 1.96), confirming that
the spatial autocorrelation of the sampled prop-
erty value data was sufficiently explained by the
independent variables. The test for multi-colli-
nearity showed that variance inflation factor
values below 10,
7
indicating multi-collinearity in
the model was not an issue.
The explanatory power of the four sets of
hedonic model varied. The office model had the
highest adjusted R
2
(0.824), followed by retail
service (0.724), RMF (0.574) and SF (0.357).
Control variables related to the physical attributes
of a property were significant for all models. The
effect of ratio of non-White residents and proxi-
mity to downtown were also consistent. On the
other hand, the correlates between the neighbor-
hood-scale measures of walking and property
values were noticeably different among the four
models. The detailed model results are reported
in Table 5.
Relationships between property values and
correlates of walking
The measure of development density (the average
FAR) was significantly associated with the prop-
erty values of single-family residential, retail
service and office uses, but not of multi-family
land uses. The positive direction of the relation-
ship indicated that higher development density
increased the economic value of a property.
The measures of land use mix barely showed
significant associations with single-family resi-
dential, retail service and office property values.
On the other hand, three measures of land use
mix – proximity to office use, proximity to retail
service use, and the ratio of retail service area to
the total area of a neighborhood – were found to
be significantly associated with rental multi-
family residential property values. The signs for
proximity measures showed that the values of
retail service properties increased as the proxi-
mity to office use increased (that is, the average
distance to office use decreased), while they
decreased as the proximity to retail service
increased (that is, the average distance to retail
service use decreased).
The relation between rental multi-family resi-
dential use and retail service, however, seemed
complicated. Whereas proximity to retail service
was negatively associated with assessed property
value for rental housing, the areal increase in
neighborhood retail service was positively asso-
ciated with an increase in the values of rental multi-
family residential parcels, meaning that a larger
retail service area in a neighborhood was economic-
ally beneficial to multi-family properties.
Proximity to open space was significantly and
positively associated with single-family residen-
tial property values, but did not affect property
The economic value of walkable neighborhoods
123r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128

AUTHOR COPY
Table 4: Summary of variables used in the hedonic model
Variable Name Description Unit of
measures
Dependent
Variable
Total value
a
Property values (land value þimprovement
value) logged
Log($)
Value per dwelling unit
b
Property value per unit (land value per unit þ
improvement value per unit) logged
Log($)
Independent
Variables
Measures for the
correlates of
walking
Development density Average FAR The average floor area ratio of all developed
parcels in a neighborhood (logged)
Log(FAR)
Land use mix Average distance to MF
parcels
The average distance to all MF parcels in a
neighborhood (logged)
Log(ft)
Average distance to retail
service parcels
The average distance to all retail service parcels in
a neighborhood (logged)
Log(ft)
Average distance to office
parcels
The average distance to all office parcels in a
neighborhood (logged)
Log(ft)
Ratio of MF parcel areas The ratio of MF parcel areas to the total area of a
neighborhood
%
Ratio of retail service parcel
areas
The ratio of retail service parcel areas to the total
area of a neighborhood
%
Ratio of office parcel areas The ratio of office parcel areas to the total area of a
neighborhood
%
Public open space Distance to a public park The distance to the closest public park in a
neighborhood (logged)
Log(ft)
Pedestrian
infrastructure
Distance to a bus stop The distance to the closest bus stop (logged) Log(ft)
Street length per acre The length of streets per acre of a neighborhood ft/acre
Sidewalk length per acre The length of sidewalks per acre of a
neighborhood
ft/acre
Control variables Physical attributes of
a property
Parcel size Parcel size in square feet Log(sqft)
Building square footage Building square footage Log(sqft)
Parcel size per dwelling
unit
b
Area of a parcel per unit (logged) Log(sqft)
Building square footage per
dwelling unit
b
Building square footage per unit (logged) Log(sqft)
Year built Year built Year
Socio-demographic
characteristics of a
neighborhood
Average income of
household
The average household income in a neighborhood
(logged)
Log($)
Average age of household The average age of households in a neighborhood Age
Ratio of non-White residents The ratio of non-White to White residents in a
neighborhood (logged)
Log(%)
Regional location Distance to downtown The distance to Seattle downtown (logged) Log(ft)
Distance to an urban center The distance to the closest urban center (logged) Log(ft)
a
applied only for SF, retail service and office models.
b
applied only for RMF model.
Sohn et al
124 r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128

AUTHOR COPY
Table 5: Hedonic model results
Land use types Single-family
residential
Rental multi-family
residential
Retail service Office
Dependent
variable
Property value (land value þ
improvement value) logged
Property value per unit (land
value per unit þimprovement
value per unit) logged
Property value (land value þ
improvement value) logged
Property value (land value þ
improvement value) logged
N2289 837 738 586
Adjusted R
2
0.357 0.574 0.724 0.824
Independent variable
(Standardized coefficients)
b(Significance) b(Significance) b(Significance) b(Significance)
Average FAR 0.116*** 0.057 0.157*** 0.062*
Average distance to MF parcels 0.065* 0.043 0.058 0.023
Average distance to retail-service
parcels
0.025 0.139*** 0.052 0.012
Average distance to office parcels 0.034 0.136*** 0.030 0.022
Ratio of MF parcel areas 0.001 0.026 0.043 0.038
Ratio of retail-service parcel areas 0.018 0.091** 0.015 0.008
Ratio of office parcel areas 0.018 0.002 0.031 0.002
Distance to a public park 0.030* 0.017 0.029 0.011
Distance to a bus stop 0.028 0.073*** 0.015 0.010
Street length per acre 0.018 0.102*** 0.069*** 0.086***
Sidewalk length per acre 0.013 0.080*** 0.062*** 0.026
Parcel size (or parcel size per
dwelling Unit
1
)
0.103*** 0.304*** 0.402*** 0.236***
Building square footage (or
building square footage per
dwelling Unit
2
)
0.388*** 0.480*** 0.420*** 0.593***
Year built 0.204*** 0.169*** 0.132*** 0.104***
Average income of household 0.054*** 0.041 0.007 0.045***
Average age of household 0.046** 0.118*** 0.055*** 0.024
Ratio of non-White residents 0.152*** 0.375*** 0.099*** 0.065***
Distance to downtown 0.349*** 0.479*** 0.177*** 0.115***
Distance to an urban center 0.007 0.011 0.071*** 0.054***
1 & 2: applied only for rental multi-family residential model.
* significant at 0.1 level; ** significant at 0.05 level; *** significant at 0.01 level.
The economic value of walkable neighborhoods
125r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128

AUTHOR COPY
values of rental multi-family residential, retail
service, and office uses.
Several measures of the pedestrian infrastruc-
ture were found to be positively related to
property values. First, proximity to bus stops
contributed to an increase in rental multi-family
residential property values. However, the coeffi-
cients for the other three uses were not statisti-
cally significant. Second, better sidewalk coverage
in their neighborhood was positively related to
increasing property values of rental multi-family
residential and retail service uses. Third, street
density, measured as street length per acre,
showed mixed results. It is positively related to
the property value of rental multi-family residen-
tial use, whereas negatively related to the prop-
erty values of retail service and office uses.
Control variables
Thesignificancelevelsandsignsoftherelation-
ships between property values and the physical
attributes of a property were fairly consistent and
in the expected direction. Building square footage
was the strongest correlate of property values
among all independent variables in the model.
Average household income in a neighborhood was
positively related to single-family residential and
office property values. The average household age
was a significant correlate of property values for
single-family residential, rental multi-family resi-
dential and retail service parcels. The direction of
the association, however, varied by land use types.
Whereas rental multi-family residential and retail
service uses favored a neighborhood with younger
households, the opposite was true for single-family
residential use. Racial composition mattered for all
land use types, and its effect on property values
was strong. Property values decreased as the per-
centage of non-White residents in a neighborhood
increased. Proximity to Seattle Downtown was
significant in all models. Proximity to the Urban
Center was also significantly related to property
values for retail service and office parcels; the
magnitude of its relation to property values,
however, was not as great as the proximity to
Seattle Downtown.
Conclusions
The findings of this study suggest the following
conclusions. First, although it was obvious that
most of the variation in property values was
explained by the attributes of individual proper-
ties, the neighborhood’s socio-demographic fac-
tors and regional location factors, some physical
characteristics of neighborhoods had significant
effects on individual property values. In particu-
lar, the effects of a neighborhood’s racial compo-
sition (the per cent of non-White residents) and
accessibility to the downtown (the distance from
the downtown) were substantial.
The study also demonstrated that certain land
use types were more sensitive to neighborhood
walkability than others. For example, several
measures of the pedestrian infrastructure and
land use mix significantly contributed to increases
in rental multi-family residential property values.
Retail service uses also favored higher develop-
ment density with higher street and sidewalk
coverage. In contrast, few measures of the
correlates of walking were significantly associated
with single-family residential and office property
values.
In relation to land use mix, the study showed
that for mixed-use neighborhoods, identifying
desirable land use combinations was as crucial
as formulating approaches to the spatial assign-
ment of land uses. It revealed that mixing retail
service uses and rental multi-family residential
used helped make rental housings more attrac-
tive. However, the positive interaction between
retail service and rental multi-family residential
uses could be anticipated only if these two uses
were appropriately separated. Providing suffi-
cient buffer space between rental multi-family
residential area and retail service area would
prevent the negative interaction between these
two uses, making the rental multi-family residen-
tial properties more marketable. The study also
suggests that mixing jobs with compact rental
multi-family housings could be favored in the
market as the findings showed a positive relation-
ship between the values of rental multi-family
residential properties and the proximity to office
parcels.
More importantly, in the present study, a higher
development density in a neighborhood did not
always seem to affect the marketability of
residential properties in a negative way although
increased density has been considered as one of
the main reasons for weakening the quality of a
neighborhood. To the contrary, the positive
association of higher development density with
the value of single-family residential properties
supports Eppli and Tu’s (1999) claim that market
Sohn et al
126 r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128

AUTHOR COPY
demand existed for higher density neighbor-
hoods. Further research investigating the factors
making high-density neighborhoods an attractive
living environment would help develop urban
design strategies for creating walking-friendly
urban settings that are marketable.
Acknowledgement
This work was supported by the Hongik
University new faculty research support fund.
Notes
1 Hedonic model is a regression analysis used to estimate
economic values of components that directly affect market
prices of an item. It is commonly applied to variations in
housing prices that reflect the value of local environmental
attributes.
2 GIS data consist of shape files defining the boundaries of
parcels and tables containing information on the land uses
and building attributes in the parcels.
3 The MAUP is a potential source of error that can affect
spatial studies, which utilize aggregate data sources (Unwin,
1996).
4 Spatial autocorrelation refers to the pattern in which
observations from nearby locations are more likely to have
similar magnitude than by chance alone (Legendre and
Fortin, 1989), which introduces deviation from the indepen-
dent observation assumption of classical statistics.
5 Moran’s Index is a measure of spatial autocorrelation
developed by Patrick A.P. Moran. The values can be
transformed to z-scores in which values greater than 1.96
or smaller than 1.96 indicate spatial autocorrelation
significant at 0.05 level.
6 Urban centers, designated by Puget Sound Regional Council
(PSRC) as the region’s core of current and future develop-
ment in Vision 2020, are locations that include a dense mix of
business, commercial, residential and cultural activity with-
in a compact area of up to 1.5 square miles.
7 the cutoff for potential multicollinearity (Myers, 1990).
References
Anderson, N.B. and Bogart, W.T. (2001) The structure of
sprawl: Identifying and characterizing employment centers
in polycentric metropolitan areas. American Journal of
Economics and Sociology 60(1): 147–169.
Asabere, P.K. (1990) The value of a neighborhood street with
reference to cul-de-sac. Journal of Real Estate Finance and
Economics 3(2): 185–193.
AultmanHall, L., Roorda, M. and Baetz, B. (1997) Using GIS for
evaluation of neighborhood pedestrian accessibility. Journal
of Urban Planning and Development 123(1): 10–17.
Bopp, M. et al (2006) Factors associated with physical activity
among African-American men and women. American
Journal of Preventive Medicine 30(4): 340–346.
Cao, T.V. and Cory, D.C. (1981) Mixed land uses, land-use
externalities, and residential property values: A reevalua-
tion. Annals of Regional Science 16(1): 1–24.
Cervero, R. (1996) Jobs-housing balance revisited: Trends and
impacts in the San Francisco bay area. Journal of American
Planning Association 62(4): 492–511.
Cervero, R. (2002) Built environment and mode choice: Toward
a normative framework. Transportation Research Part D 7(4):
265–284.
Cervero, R. and Kockelman, K. (1997) Travel demand and the
3Ds: Density, diversity, and design. Transportation Research
Part D 2(3): 199–219.
Clifton, K.J. and Dill, J. (2005) Women’s Travel Behavior and
Land Use: Will New Styles of Neighborhoods Lead to More
Women Walking? Research on Women’s Issues in Trans-
portation. Report of a Conference, 2, 89–99.
Clapp, J.M. and Giaccotto, C. (1992) Estimating price trends for
residential property: A comparison of repeat sales and
assessed value methods. The Journal of Real Estate Finance
and Economics 5(4): 357–374.
Crane, R. and Crepeau, R. (1998) Does neighborhood
design influence Travel? A behavioral analysis of travel
diary and GIS data. Transportation Research Part D 3(4):
225–238.
De Bourdeaudhuij, I., Teixeira, P.J., Cardon, G. and Deforche,
B. (2005) Environmental and psychosocial correlates of
physical activity in Portuguese and Belgian adults. Public
Health Nutrition 8(7): 886–895.
Dill, J. (2003) Transit use and proximity to rail – Results from
large employment sites in the San Francisco, California, Bay
Area. Transportation Research Record 1835: 19–24.
Dorn, J. (2004) Hidden in Plain Sight: Capturing the Demand for
Housing Near Transit. Oakland, CA: Center for Transit-
Oriented Development.
Eppli, M.J. and Tu, C.C. (1999) Valuing the New Urbanism: The
Impact of the New Urbanism on Prices of Single-Family Homes.
Washington DC: Urban Land Institute.
Ewing, R. (1995) Beyond density, mode choice, and single
purpose trips. Transportation Quarterly 49(4): 15–24.
FHWA. (2002) Toolbox for regional policy analysis, http://
www.fhwa.dot.gov/planning/toolbox/index.htm, accessed
16 May 2011.
Fulton, J.E., Shisler, J.L., Yore, M.M. and Caspersen, C.J. (2005)
Active transportation to school: Findings from a national
survey. Research Quarterly for Exercise and Sport 76(3):
352–357.
Gauvin, L. et al (2005) From walkability to active living
potential: An ‘ecometric’ validation study. American Journal
of Preventive Medicine 28(2): 126–133.
Giles-Corti, B. et al (2005) Increasing walking: How important
is distance to, attractiveness, and size of public open space?
American Journal of Preventive Medicine 28(2): 169–176.
Giuliano, G. and Small, K.A. (1993) Is the journey to work
explained by urban structure? Urban Studies 30(9): 1485–1500.
Grether, D.M. and Mieszkowski, P. (1980) The effects of
nonresidential land uses on the prices of adjacent housing:
Some estimates of proximity effects. Journal of Urban
Economics 8(1): 1–15.
Handy, S. (1996) Understanding the link between urban form
and non-work travel behavior. Journal of Planning Education
and Research 15(3): 183–198.
Handy, S. (2005) Smart growth and the transportation – land
use connection: What does the research tell us? International
Regional Science Review 28(2): 146–167.
The economic value of walkable neighborhoods
127r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128

AUTHOR COPY
Hess, D.B. and Lombardi, P.A. (2004) Policy support for
and barriers to transit-oriented development in the inner
city: Literature review. Transportation Research Record 1887:
26–33.
Hess, P.M., Moudon, A.V. and Logsdon, M.G. (2001) Measur-
ing land use patterns for transportation research. Transpor-
tation Research Record 1780: 17–24.
Hoehner, C.M., Brennan Ramirez, L.K., Elliott, M.B., Handy, S.L.
and Brownson, R.C. (2005) Perceived and objective
environmental measures and physical activity among
urban adults. American Journal of Preventive Medicine
28(2): 105–116.
Holcombe, R.G. (1999) In defense of urban sprawl. PERC
Reports 17(1): 3–5.
Irwin, E.G. (2002) The effects of open space on residential
property values. Land Economics 78(4): 465–480.
Janssen, C. and Soderberg, B. (1999) Estimating market prices
and assessed value for income properties. Urban Studies
36(2): 359–376.
Katz, P. (1994) The New Urbanism: Toward an Architecture of
Community. Washington DC: McGraw-Hill.
Khattak, A.J. and Rodriquez, D. (2005) Travel behavior in neo-
traditional neighborhood developments: A case study in
USA. Transport Research Part A 39(6): 481–500.
Kim, S. and Ulfarsson, G. (2004) Travel mode choice of the
elderly – Effects of personal, household, neighborhood, and
trip characteristics. Transportation Research Record 1894: 117–126.
Kitamura, R., Mokhtarian, P. and Laidet, L. (1997) A micro-
analysis of land use and travel in five neighborhoods in the
San Francisco Bay area. Transportation 24(2): 125–158.
Kockelman, K.M. (1997) Travel behavior as function of
accessibility, land use mixing, and land use balance:
Evidence from San Francisco Bay Area. Transportation
Research Record 1607: 116–125.
Komanoff, C. and Roelofs, C. (1993) The Environmental Benefits
of Bicycling and Walking. Washington DC: Federal Highway
Administration.
Krizek, K.J. (2003) Operationalizing neighborhood accessibility
for land use – travel behavior research and regional modeling.
Journal of Planning Education and Research 22(3): 270–287.
Lang, R.E., Hughes, J.W. and Danielsen, K.A. (1997) Targeting
the suburban urbanites: Marketing central-city housing.
Housing Policy Debate 8(2): 437–470.
Laverne, R.J. and Winson-Geideman, K. (2003) The influence of
trees and landscaping on rental rates at office buildings.
Journal of Arboriculture 29(5): 281–290.
Lee, C. and Moudon, A.V. (2006) Correlates of walking for
transportation or recreation purposes. Journal of Physical
Activity and Health 3(1): S77–S98.
Legendre, P. and Fortin, M.J. (1989) Spatial pattern and
ecological analysis. Vegetatio 80(2): 107–138.
Li, F., Fisher, K.J. and Brownson, R.C. (2005) A multilevel
analysis of change in neighborhood walking activity in older
adults. Journal of Aging and Physical Activity 13(2): 145–159.
Limanond, T.L. and Niemeier, D.A. (2003) Accessibility and
mode-destination choice deciusions: Exploring travel in
three neighborhoods in Puget Sound, WA. Environment and
Planning B 30(2): 219–238.
Morrow-Jones, H.A., Irwin, E.G. and Roe, B. (2004) Consumer
preference for neotraditional neighborhood characteristics.
Housing Policy Debate 15(1): 171–202.
Moudon, A.V. et al (2007) Attributes of environments supporting
walking. American Journal of Health Promotion 21(3): 448–459.
Myers, R. (1990) Classical and Modern Regression with Applica-
tions, 2nd edn. Boston: Duxbury Press.
Paumier, C. (2004) Creating a Vibrant City Center. Washington
DC: Urban Land Institute.
Plaut, P.O. and Boarnet, M.G. (2003) New urbanism and the
value of neighborhood design. Journal of Architectural and
Planning Research 20(3): 254–265.
Puget Sound Regional Council. (2004) Vision 2020. Seattle, WA:
Puget Sound Regional Council.
Rajamani, J., Bhat, C., Handy, S., Knaap, G. and Song, Y. (2003)
Assessing impact of urban form measures on non-work trip
mode choice after controlling for demographic and level-of-
service effects. Transportation Research Record 1831: 158–165.
Rood, T. (2000) Ped Sheds. San Francisco, CA: Congress of the
New Urbanism.
Saelens, B.E. and Handy, S.L. (2008) Built environment
correlates of walking: A review. Medicine and Science in
Sports and Exercise 40(7): 550–566.
Schwanen, T. and Mokhtarian, P.L. (2004) The extent and
determinants of dissonance between actual and preferred
residential neighborhood type. Environment and Planning B
31(5): 759–784.
Shapiro, R.J., Hassett, K.A. and Arnold, F.S. (2002) Conserving
Energy and Preserving the Environment: The Role of Public
Transportation. Washington DC: American Public Transpor-
tation Association.
Smith, M. and Butcher, T. (1994) Parkers as pedestrians. Urban
Land 53(6): 9–10.
Sohn, D. (2007) The effect of spatial autocorrelation in
analyzing the relationship between the characteristics of
walkable neighborhoods and multi-family residential prop-
erty values. The Korea Spatial Planning Review 54: 119–137.
Sohn, D. and Moudon, A.V. (2008) The economic value of office
clusters. Journal of Planning Education and Research 28(1): 86–99.
Song, Y. (2005) Smart growth and urban development pattern:
A comparative study. International Regional Science Review
28(2): 239–265.
Song, Y. and Knaap, G. (2003) New urbanism and housing
values: A disaggregate assessment. Urban Economics 54(2):
218–238.
Song, Y. and Knaap, G. (2004) Measuring urban form. Journal of
American Planning Association 70(2): 210–225.
Srinivasan, S. (2001) Quantifying spatial characteristics for
travel behavior models. Transportation Research Record 1777:
1–15.
Talen, E. (2001) Traditional urbanism meets residential affluence –
An analysis of the variability of suburban preference. Journal
of the American Planning Association 67(2): 199–216.
Talen, E. (2003) Neighborhoods as service providers: A
methodology for evaluating pedestrian access. Environment
and Planning B 30(2): 181–200.
Unwin, D.J. (1996) GIS, spatial analysis and spatial statistics.
Progress in Human Geography 20(4): 540–551.
Wolf, K.L. (2003) Public response to the urban forest in inner-
city business district. Journal of Arboriculture 29(3): 117–126.
Zlot, A.I. and Schmid, T.L. (2005) Relationships among
community characteristics and walking and bicycling for
transportation and recreation. American Journal of Health
Promotion 19(4): 314–317.
Sohn et al
128 r2012 Macmillan Publishers Ltd. 1357-5317 URBAN DESIGN International Vol. 17, 2, 115–128