CSIS Discussion Paper No.

67

Economic Value of Urban Landscapes

November 2005

Xiaolu Gao1 and Yasushi Asami2

1
Corresponding author. Institute of Geographical Sciences and Natural Resources Research,

Chinese Academy of Sciences. Jia 11, Datun Road, Chaoyang District, Beijing 100101,

China. Email: [email protected]

2 Center for Spatial Information Science, University of Tokyo. Kashiwanoha 5-1-5, Kashiwa,

Chiba 277-8568, Japan. Email: [email protected]

1
Economic Value of Urban Landscapes

Abstract
To support decision-makings in landscape management and related policies, an
objective methodology for evaluating urban landscapes is necessary. In particular, it is
anticipated to show the economic value of urban landscapes so that the social benefits of
regulations on landscapes can be verified. This paper developed a three-step framework
for assessing the economic value of urban landscapes, i.e. conducting standardized
landscape survey, extracting critical evaluation factors, and identifying the marginal
effect of the factors with a hedonic approach. Using Tokyo and Kitakyushu city data,
which are typical of large metropolitan areas and local cities, it was empirically
demonstrated that the compatibility of buildings and the greenery of neighborhood are
distinctively perceived factors, and in both cities, compatibility and greenery were
significant determinants of land prices. Although Tokyo and Kitakyushu city differ in
landscape features and many other ways, the economic impacts of urban landscapes
were very alike. These results empirically confirmed the usefulness of the three-step
approach for evaluating urban landscapes, and because the improvement of
compatibility and greenery require for the collaboration of residents, the results, in a
general sense, implied the importance of coordination at neighborhood level.

Keywords: urban landscape, economic value, compatibility, greenery

2
1. Introduction
In recent years, there is a growing concern for the beauty of urban landscapes.
Improving urban landscapes is taken as an effective way for enhancing the competitive
ability of cities and a way for revitalizing old dilapidated areas. A variety of public
policies have been made to fulfill this purposes, e.g. land use regulations on the height
and appearance of buildings, landscape controls on views and advertisement, and
designation of conservation areas (Nishimura et al., 2000). An example in Japan is the
implementation of ‘Landscape Laws’ in 2004. The laws provided legal foundation for
the protection of the rights of urban landscapes, and through a so-called ‘landscape
certification’ program, human judgments on the comprehensive quality of landscapes
were legitimated. It is widely accepted that the above policies have deep influences on
urban planning and landscape management. However, in practice, the evaluations of
urban landscapes are often mixtures of personal views, and there is still no formal
framework of evaluation. For these reasons, the effectiveness of the policies is
potentially limited. Therefore, it is of critical significance to establish an objective
methodology for evaluating urban landscapes.
In particular, since planning regulations on the heights, shapes, or locations of
buildings out of landscape reasons will inevitably limit the development right of land, it
is strongly anticipated to clarify the social benefits of regulations and quantitatively
show the benefits. This necessitates analyses on the economic values of urban
landscape.
So far, quite a few frameworks for landscape evaluation have been raised. For
example, Gómez-Sal et al. (2003) proposed to evaluate landscape from ecological,
productive, economic, social and cultural perspectives and defined scenarios in
comparison with which particular landscape planning or management projects can be
evaluated. Prato (2000) proposed to evaluate the sustainability of landscape
management plans by considering the biophysical and economic attributes of plans
emphasized differently by private and public decision-makers. Angileri and Toccolini

3
(1993) assessed the visual quality of rural landscapes for which they defined five
aspects in landscape perceptions, i.e. relief, vegetation, density of built-up areas, size of
cultivated fields and presence of character elements such as hedgerows or small woods.
Based on field survey, they drew landscape evaluation maps for study areas. These
works strongly suggest that empirical studies are indispensable in landscape
evaluations.
In the literature, a big variety of approaches were employed in empirical works, for
example, evaluations for the physical space and compositions of specific historical sites
(Carter and Bramley, 2002; Coeterier, 2002; Lichfield, 1988), evaluations of urban
landscapes with semantic differential technique in terms of human perceptions
(Garcia-Mira et al., 1997; Green, 1999; Imamoglu, 2000), evaluations for the quality of
design with respect to physical criteria of landscape (Fukahori and Kubota, 2003;
İpekoğlu, 2006), and so on. In addition, some addressed the economic values of
landscapes, for example, with psychological experiment (Fukahori and Kubota, 2003),
contingent valuation method (Willis and Garrod, 1993), etc. Methodologically speaking,
the above approaches have been well established.
However, although these approaches allow for evaluations from specific aspects, it
is not easy to use these methods to comprehensively evaluate urban landscapes, and the
results are sometimes not easy for understanding. On the other hand, widely used
methodologies for evaluating the economic impact of environmental goods such as
hedonic approach was rarely used in the present literature of urban landscape.
In this paper, we attempt to bridge this gap. The main concern of this study is
twofold. The first is to identify evaluating structure of urban landscape, and to
empirically investigate whether the values of urban landscape are reflected by land
market. This requires for a framework for objectively measuring the ‘beauty’ of urban
space, as well as an analytical framework for associating the perceptions for physical
environments with economic values. The second concern relates to the generalization of
the evaluation method. Since the perception of urban landscape is likely to differ among

4
cities, districts, and individuals with different preferences, to what degree a
methodology is valid and applicable is a critical issue. Through this work, we want to
establish an empirical approach for assessing urban landscape.
We selected Tokyo and Kitakyushu city for investigation, which are typical of large
metropolitan areas and local central cities, respectively. The two cities have
well-managed data on planning and urban environment, most of which are integrated
with GIS.
A sample of residential sites was chosen in each city. We conducted standardized
surveys on the landscape of these areas, i.e. in the survey subjective factors that may
lead to diverse results were controlled as much as possible. Upon analysis, it was
identified that compatibility and greenery are most distinctively perceived factors in
urban landscapes, and in addition, they are significant determinants of residential land
prices. Comparison analysis of the Tokyo and Kitakyushu samples revealed that their
evaluation structures are quite alike, though the two cities have very different
geographical and social economic conditions. This strongly suggests the objectivity and
usefulness of the developed method for assessing urban landscapes.

2. Methodology
A three-step approach was developed.
First, design a framework, by which one can qualitatively catch the physical
characteristics of urban landscapes, and then survey on sample sites and their
surrounding areas to collect objective data.
Second, extract critical factors for landscape evaluation by employing principal
component analysis (PCA).
Lastly, adopt a hedonic approach to examine if the principal characteristics of
urban landscape are critical determinants of land prices.

2.1 Sample and data

5
 In Tokyo, the sample sites were drawn from 1996-97 issues of Weekly Housing
Information, which provided information on a large amount of houses and lands for sale.
The sample was limited to transacted vacant land properties in western part of the 23
wards of Tokyo. The study area covers nine wards (Suginami, Nerima, Shinjuku,
Shibuya, Nakano, Setagaya, Meguro, Ota, and Shinagawa). We chose only a part of
Tokyo for the convenience of survey, and because transacted sites were relatively
densely distributed in this area. In addition, the sample sites were confined to land use
zones designated mainly for residential use. The size of sample with complete data is
272. Among them, 203 sites are in low-story residential zones, 37 in high-medium
residential zones and 32 in residential zones.
The sample in Kitakyushu city was drawn from an administration survey database
for 1333 properties purchased in fiscal year 2003. Similar criteria were used as that in
Tokyo, which yielded a sample of 187 vacant sites distributed across the whole city. The
proportion of samples in low-story residential zones, high-medium residential zones and
residential zones is approximately 2:1:2.
The databases for the two samples were carefully constructed. First, both include
the price of the sites. For the Tokyo sample, Weekly Housing Information provided the
final list prices of each. The administration survey data of Kitakyushu city provided the
prices given by purchasers in questionnaire. Since they are not real prices, the accuracy
is a bit low, but we assumed that they were close to real prices. Besides, the data on the
Tokyo sample from the database of Gao et al. (2005) was used. They included detailed
information on sample lots such as sizes, shapes, the width and direction of front roads,
locations and accessibility to public transportation, a variety of neighborhood
environmental attributes, as well as land use and social economic indices of chome (i.e.,
district) such as building density, gross floor-area-ratio, population density, proportion
of wooden-made structures, and so on. For Kitakyushu city, the living environment data
provided by Kitakyushu city, including the ratings on 18 aspects of living environment
of 1488 chomes, were used in addition to the administrative survey data.

6
 A list of the above data can be found in Appendix 1.

2.2 Landscape survey

Table 1 shows the landscape survey sheet. With regard to the content, we learned
from Arai (2001), which proposed to evaluate urban landscape from three aspects:
neighborhood scene, street scene and planning activities (e.g. public involvement in
local affairs and the implementation of ‘district planning’ and covenant).
In this survey, we focused more on evaluations of neighborhoods and streets, with
specific interest on the physical aspect of urban environment and for which data are
easy to be collected by observation. Along this line, we designed an 11-factor evaluation
system shown in Table 1. The factors beginning with ‘A’ are indicators of
neighborhoods and those with ‘B’ are indicators of streets. Furthermore, each factor has
several specific items, each rated with points (+1, 0, -1, etc.). The indices of the 11
factors were aggregated from the points.

Table 1 Factors for evaluating urban landscape

Factors Items Point

A1 Continuity of external walls Continuous +1
(+1, 0, -1) Average level 0
Not continuous -1
A2 Conformity in colors and materials Harmonious +1
(+1, 0, -1) Average level 0
not harmonious -1
A3 Compatibility of buildings styles Sharing common features +1
(+1, 0, -1) Average level 0
Little common features -1
A4 Beauty of skylines constructed by buildings Building height in order +1
(+3, +2, +1, 0, -1) Similar roof shape +1
beautiful rhythm with other buildings +1
Skylines in disorder -1
A5 Openness and the scale of buildings Open and relaxed +1
(+1, 0, -1, -2) Compressed street space -1
Dull without change -1
A6 Visually nice and continuous greenery forming network +1
(+2, +1, 0) Visual continuous +1
B1 Greenery of walls and trees Continuously greened walls +1
(+1, 0, -1) Average level 0
Mostly concrete blocks walls -1

7
B2 Greenery of open pedestrian spaces Having well-greened parks and playgrounds +1
(+2, +1, 0, -1) Many trees along street +1
Deserted land scattered by garbage -1
B3 Favorable pedestrian space Pleasant streetscape +1
(+1, 0, -1, -2, -3) Chaotic scenes with garbage bins or bicycles -1
Advertises in disorder -1
Illegal parking that disturb pedestrian use -1
B4 Friendly outdoor space Friendly street space +1
(+1, 0, -1) Isolated street without living atmosphere -1
B5 Decorations and street furniture Street furniture, sculpture, waterscape, etc. +1
(+2, +1, 0) Well-designed lightening, etc. +1

To keep objectivity in the survey, a detailed manual was made. The evaluations
were based on neighborhood areas within 20-25 m from the borders of each sample site.
Detailed criteria for giving points are provided, including both verbal descriptions and
pictures for reference. For an instance, the criteria for A1 (the continuous of external
walls) are shown in Appendix 2. Besides, a 1-hour training was delivered to
investigators, who came from two local investigation companies and had no
professional experience on architecture and urban design.
The site surveys were conducted for the Tokyo sample in May 2004 and for the
Kitakyushu sample in March and April of 2005. The investigators were instructed to
work on each site for 15 to 20 minutes, and finish about 10 sites every day. For each
sample site, two people implemented the evaluation. The second was asked to confirm
the points given by the first. In case that they cannot agree with each other, the scores
given by each were recorded. In addition, they were asked to take six to eight pictures
and a 15-second video for each sample area. These were later used to justify whether the
evaluation results were biased and in case of different opinions appeared, which of the
investigator’s result was better. In fact, 94% of the points given by the investigators
were unanimous and reasonable, and no significant difference was detected between the
results in two cities. A member in one of the 15 groups of Tokyo investigators tended to
rate landscape lower than his colleague. This tendency and other discrepancies in the
survey were justified with the pictures and videos.
As results, the databases of urban landscapes in Tokyo and Kitakyushu city were

8
established. Figure 1 shows the means of the indices of the 11 factors. The figures in
bracelets are t-values of the differences between the two cities.

1.2
(1.21) Tokyo
1.0
Kitakyushu
0.8
(2.89)
0.6
(-6.50) (-2.53)
(4.42)
0.4
(-0.41)
0.2 (0.41) (2.3) (-0.62)
(-0.83) (-1.5)
0.0
A1 A2 A3 A4 A5 A6 B1 B2 B3 B4 B5
-0.2

Figure 1 Comparison of the survey results in two cities

A screening of the data shows that the average levels are similar in factors such as
A1 (continuity of external walls), A2 (conformity in colors and materials), A3
(compatibility of building styles), A4 (beauty of skylines constructed by buildings), B1
(greenery of walls and trees) and B5 (decorations and street furniture but significantly
different in other factors such as A5 (openness and the scale of buildings), A6 (visually
nice and continuous greenery), B2 (greenery of open pedestrian spaces), B3 (favorable
pedestrian space), and B4 (friendly outdoor space). The mostly varied factor A5 had
reflected the fact that many samples in Kitakyushu city were located in newly
developed areas with many vacant lands. As a whole, the results reasonably suggested
the characteristics of large metropolitan areas as compared to local cities, where
building density is generally higher and residential areas are featured by less open space
and greenery.

2.3 Principal Component Analysis (PCA) of landscape data

It was found that many of the 11 variables on urban landscape were significantly
correlated. We performed a PCA in the second step to alleviate the correlation problem

9
 while to understand the evaluation structure better.
For the Tokyo data, the 11 factors were firstly classified into four categories
according to their correlations with a graphical modeling method. As a result, we have
four groups of variables, (A1, A2, A3, A4), (B1, B2, B6), (B3, B4, A5), and (B5). From
the features they represented, we could tell that they relate to the compatibility of
buildings, greenery, sense of familiarity, and the effort in blocks to preserve or create
unique characteristics, respectively. Then, PCA was conducted within each category.
This generated one principal component with eigenvalue larger than 1 for each of the
former three categories, which were named compatibility, greenery and familiarity.
They accounted for 72.8%, 66.6% and 58.2% of the classified variables,
respectively. B5 in the fourth category was omitted because eigenvalue of the factor
strongly correlated to this variable in the whole data is very small. The variables were
classified prior to PCA because the two variables signified in later discussed hedonic
modeling kept more information of original data (44.7%) than the one significant
variable did if principal components without classification were input to hedonic model
(41.4%).
In parallel, three principal components with eigenvalues larger than 1 were
generated with the Kitakyushu data. Table 2 details the results.

Table 2 Principal components of landscape factors

Tokyo Kitakyushu city

Principal Component 1 2 3 1 2 3
Eigenvalue 4.436 1.313 1.035 4.317 1.4369 1.016
Percent (%) 40.33 11.93 9.41 39.24 13.06 9.23
Cum Percent (%) 40.33 52.26 61.68 39.24 52.31 61.54
Compatib Familiarit Compatib Decoratio
Evaluation concept Greenery Greenery
ility y ility n
A1 Continuity of external walls 0.336 -0.208 -0.306 0.372 -0.234 -0.004
A2 Conformity in colors and materials 0.370 -0.298 -0.155 0.392 -0.269 -0.068

10
A3 Compatibility of buildings styles 0.381 -0.158 -0.204 0.388 -0.221 -0.028
A4 Beauty of skylines constructed by
buildings 0.383 -0.206 -0.207 0.397 -0.235 0.079
A5 Openness and the scale of buildings 0.231 -0.073 0.361 0.280 0.056 -0.002
A6 Visually nice and continuous
greenery 0.305 0.476 -0.097 0.220 0.608 -0.055
B1 Greenery of walls and trees 0.249 0.491 -0.197 0.322 0.289 0.032
B2 Greenery of open pedestrian spaces 0.275 0.487 0.037 0.226 0.548 0.144
B3 Favorable pedestrian space 0.280 -0.086 0.479 0.246 0.084 -0.203
B4 Friendly outdoor space 0.287 -0.221 0.383 0.236 -0.056 -0.079
B5 Decorations and street furniture 0.119 0.187 0.492 0.048 -0.051 0.957

In both samples, the three principal components explained for about 40%, 12%,
and 9% of the variances of the 11 variables. In total, they keep 62% of the total
information of each dataset. The eigenvectors (lower part of Table 2) revealed that the
structures of the first and two components were fairly alike in Tokyo and Kitakyushu
city. Since the first principal component strongly correlates to factors representing the
compatibility of buildings (A1, A2, A3, A4), we considered it to be a scale for
compatibility. In the same way, the second principal component was deemed to be a
scale of greenery. The third of the Tokyo sample, associated with B3, B4 and B5, was
regarded as familiarity. That of the Kitakyushu sample with strong association with B5
was named decoration.
The results suggested that compatibility and greenery are the most distinctive
features in the cognition of urban landscapes. Even in different cities, it has no much
change. In metropolitan areas, familiarity is emphasized, probably because the flavor of
people in large metropolitan areas is inversely affected by intense mixtures of industrial
use, heavy traffic, and so on. In contrast, decoration in Kitakyushu city may have
reflected the endeavors of local government and communities to preserve and create
local characteristics.

11
2.4 Hedonic analysis on landscape factors in Tokyo
In the third step, we performed hedonic analyses with the datasets of Tokyo and
Kitakyushu. In this process, the principal components of urban landscape were used as
independent variables to identify, if any, their impact on land prices.
For the Tokyo sample, we used the following linear regression functions.
UnitP = intercept + ∑ α i × X i + ∑ b j × ( X j / S ) , (1)
i j

where, UnitP is a vector of unit price derived by dividing total land prices by lot size, Xi
i=1 to m is a vector of the i-th independent variable with m being the number of
independent variables, S is a vector of lot size, and ai and bj are regression coefficients.
The dependent variable UnitP is approximately normally distributed. Terms Xi and Xi /S
were both entered because we assumed that the influence of some variables may change
with lot size.
The raw data of independent variables were transformed to suitable forms through
postulating and validating various assumptions repetitively. For example, the width of
the front road, w1, was replaced by ln(w1), thinking that as w1 increases, its marginal
influence on UnitP should decay. The width of second front road, w2 (w2=0 if second
front road does not exist), was transformed to ln(w2-1) so that the new variable is
continuous even if a second front road does not exist. After transformation, the fitting of
the model had been improved.
The specification in Equation (1) was tested against a variety of alternative functions
such as linear model regressing on Xi, log-linear regression model on Xi, and so on. The
fitting of the model in Equation (1) was satisfactory (highest R2 and lowest AIC), and
the prediction errors yielded by cross-validation test were smaller than other tested
models.
As the result of incremental stepwise regression, a model with 21 variables was
established (Table 3). The correlations of the 21 variables were weak except for that
between cul-de-sac and cul-de-sac/S. This was easy to understand because they were
integrated terms. However, statistical tests with some variables removed, and with

12
 random errors added to the correlated terms showed that the estimates of the model
were stable. This suggests that this model has not suffered much from multi-collinearity
problem.

Table 3 Regression model for unit price (in Tokyo)

Impact on
unit price Std P-valu
Variable Definition t Scope
thousand Error e
Yen/m2
Intercept 734.5 0.0432 17.01 <.0001
1 Line-Seibu* Along Seibu railway lines, 1; otherwise,0 -120.2 0.0155 -7.77 <.0001 Neighborhood
2 Line-Keio Along Keio railway lines, 1; otherwise,0 -40.2 0.0132 -3.05 0.0025 Neighborhood
3 Line-Tokyutoyoko Along Tokyutoyoko railway lines, 1;
64.2 0.0159 4.03 <.0001 Neighborhood
otherwise, 0
4 Multiple line Close to multiple railway lines, 1;
33.4 0.0120 2.78 0.0058 Neighborhood
otherwise, 0
5 UpFAR1 If effective FAR of a site is between
60-100%, 1; between 110-270%, -1; -60.3 0.0122 -4.95 <.0001 Lot
otherwise, 0
6 UpFAR2 with effective FAR of a site less than
-112.7 0.0234 -4.84 <.0001 Lot
210%, 1; beyond 220%, -1
7 UpFAR3 with effective FAR between 110-160%, 1;
18.9 0.0085 2.21 0.0279 Lot
between 170-210%, -1; otherwise, 0
8 t_station Time to the nearest train station (minute) -9.5 0.0013 -7.38 <.0001 Lot
9 Irregular shape If for an irregular shaped site, S>=70 ,
-24.3 0.0037 -6.51 <.0001 Lot
ln(S-70); otherwise, 0
10 Frontage/S Frontage sharing with main front road (in
505.5 0.2092 2.42 0.0164 Lot
meters)/S
11 ln(w1) ln(width of main front road (in meters)) 49.8 0.0168 2.96 0.0034 Lot
12 ln(w2-1) If width of the second main front road
50.0 0.0135 3.69 0.0003 Lot
w2>=2.0 m, ln(w2-1); otherwise, 0
13 Cul-de-sac/S Cul-de-sac dummy/S 4998.3 2.0419 2.45 0.0151 Neighborhood
14 Cul-de-sac With a cul-de-sac front road, 1; otherwise,
-100.2 0.0258 -3.88 0.0001 Neighborhood
0
15 Chome-elevation/S Average elevation of chome (m) 83.6 0.0357 2.34 0.02 District
16 Chome-popden Population density of chome (person/ha -0.4 0.0001 -3.03 0.0027 District
17 Chome-BCR<=40%/ with average building coverage ratio less
-4243.1 1.4081 -3.01 0.0029 District
S than 40%, 1/S; otherwise, 0
18 Chome-wooden Proportion of wooden structure buildings
-2.0 0.0009 -2.21 0.0283 District
in chome (%)
19 Unpleasant facility with unpleasant facility in neighborhood,
-119.9 0.0284 -4.23 <.0001 Neighborhood
1; otherwise, 0
20 Compatibility First principal component of landscape 7.9 0.0033 2.4 0.0171 Neighborhood
21 Greenery Second principal component of landscape 8.4 0.0040 2.1 0.0368 Neighborhood
R2 0.699
Adj. R2 0.672
N 232

13
 * Dummy variables for railway lines were based on JR Chuo line.

It was found that the determinants of land price include railway lines (line-Seibu,
line-Keio, line-Tokyutoyoko), accessibility to multiple railway lines (multiple line) and
time to the nearest train station (t-station). For example, properties along Tokyutoyoko
line, where is famous for wealthy residents and high-class houses, are 64 thousand
Yen/m2 more expensive than standard residential areas along Chuo line (based on which
the dummy variables of railway lines were generated).
Significant variables also include attributes of land such as frontage (frontage/S),
irregularity (flag) and floor-area-ratio regulated by planning controls (upFAR1, upFAR2,
upFAR3), the attributes of front roads such as width (ln(w1) and ln(w2-1)) and
cul-de-sac, as well as the attributes with regard to blocks, which involve average
elevation (chome-elevation/S), population density (chome-popden), building coverage
ratio (chome-BCR<=40%), the proportion of wooden structure buildings
(chome-wooden), and the presence of unpleasant facilities.
The estimates of the above variables as well as their signs are reasonable and the
results are of interest from many viewpoints. However, detailed explanations are
omitted to leave space for the investigation of variables on landscape.
Two principal components were significant. One is compatibility, and the other is
greenery. Land prices increase for 7.9 thousand Yen/m2 if the value of compatibility
increases for one point and for 8.4 thousand Yen/m2 if the value of greenery increases
for one point. These results have some important implications. First, they demonstrated
the positive effect of landscapes on land prices since compatibility and greenery are
positively associated with landscape factors as shown by the signs of eigenvectors in
Table 2. Secondly, since the unit land price in Tokyo is averagely 600 thousand Yen/m2
(specifically, the mean of this sample is 602.4 thousand Yen/m2), the difference of one
point in compatibility or greenery accounts for more than 1-1.5% of the total prices. We
should say that the influence by landscape amenity on land price can not be ignored.
In addition, the results suggested the importance of cooperative activities for

14
 improving landscape qualities. According to the nature of the significant variables, they
can briefly be classified to those associated with individual lots, those with
neighborhoods, and those with even broader areas. This property is shown in the
right-most column in Table 3. Because compatibility and greenery are attributes of
building groups and neighborhoods, their improvement or management, to a large
degree, depends on the efforts of whole residents rather than solely individuals.

2.5 Hedonic analysis of landscape factors in Kitakyushu city

A similar modeling procedure as above was followed with the Kitakyushu data. As
a result, we got a log-linear regression model for unit land price.
ln(UnitP)= intercept + ∑ ai × X i , (2)
i

By stepwise regression, a model with 25 variables was established (Table 4). This
model explains for 63.8% of the variance of ln(UnitP). Statistical tests showed that
there were no big multi-collinearity problems in this model and the estimates were
stable. In addition, the signs and the estimates of the variables were consistent with
expectation.
In this model, two principal components of urban landscape factors, compatibility
and greenery were significant at 0.01 and 0.1 levels, respectively. This result again
demonstrated the economic values of urban landscape. From the elastic coefficients, we
know that land price can increase for 2.9% if compatibility is one-point higher, and for
2.7% if greenery is one-point higher.

Table 4 Regression mode for ln(UnitP) in Kitakyushu city

Elastic
Impact
coefficient
No. Variable Definition on Std Error t P-value Scope
(impact on
ln(UnitP)
unit price
Intercept 11.0167 0.1414 77.91 <.0001
1 Away from main road Away from main road, 1; -0.1399 0.0396 -3.53 0.0005 0.869 Neighborhood
otherwise, 1

15
2 Away from commercial Away from commercial area, -0.1175 0.0554 -2.12 0.0355 0.889 Neighborhood
area 1;otherwise, -1
3 Road direction1 Not in east, 1; otherwise, -1 -0.0562 0.0296 -1.9 0.0595 0.945 Lot
4 Road direction2 In north, north-east or south, -0.0580 0.0194 -3 0.0032 0.944 Lot
1; in east, 0; otherwise, -1
5 Road direction3 In north, 1; in north-east or -0.0467 0.0316 -1.48 0.1413 0.954 Lot
south, -1; otherwise, 0
6 Without sidewalk Without sidewalk, 1; -0.0565 0.0206 -2.74 0.0067 0.945 Neighborhood
otherwise, 0
7 Road circulation No bad, 1; otherwise, -1 0.1478 0.0380 3.89 0.0001 1.159 Neighborhood
8 Line-not Chikuho Not along Chikuho line, 1; 0.0657 0.0315 2.08 0.0389 1.068 neighborhood
otherwise, -1
9 Line-not Hitahikosan Not along Hitahikosan line, -0.1024 0.0431 -2.38 0.0186 0.903 Neighborhood
1; otherwise, -1
10 Line-not monorail Not along monorail, 1; -0.1329 0.0295 -4.51 <.0001 0.876 Neighborhood
otherwise, -1
11 t-bus stop Time to the nearest bus stop -0.0004 0.0001 -4.42 <.0001 1.000 neighborhood
12 Solid land base Solid land base, 1; 0.1729 0.0795 2.17 0.0311 1.189 Lot
otherwise,-1
13 regulated FAR FAR restricted by land use 0.0011 0.0003 3.71 0.0003 1.001 Lot
regulations
14 Regular shape Regular shape, 1; otherwise, 0.1046 0.0214 4.89 <.0001 1.110 Lot
-1
15 Distance-shopping Within 200-500 m to the -0.0432 0.0182 -2.38 0.0187 0.958 Neighborhood
nearest shopping center, 1;
otherwise, -1
16 Chome-fire disaster In high degree of danger, 1, -0.0452 0.0171 -2.64 0.0092 0.956 District
otherwise, -1
17 Chome-fire prevention Well prepared for fire 0.0352 0.0192 1.83 0.069 1.036 District
prevention, 1; otherwise, -1
18 Chome-public Public transportation is bad, -0.1657 0.0378 -4.38 <.0001 0.847 District
transportation 1;average, 0, good, -1
19 Chome-medical facility Poor or average medical -0.0490 0.0266 -1.84 0.0672 0.952 District
facilities, 1, good, -1
20 Chome-home care Too few or too many home -0.0733 0.0212 -3.46 0.0007 0.929 District
care facilities, 1, average, -1
21 Chome-daily facility Insufficient daily facilities, -0.1485 0.0406 -3.66 0.0003 0.862 District
1;average or above, -1
22 Chome-popden Population density in chome 0.0029 0.0008 3.76 0.0002 1.003 District
(person/ha)
23 Housing density Density of houses in chome -0.0048 0.0028 -1.68 0.0947 0.995 District
(house/ha)
24 Compatibility Principal component of 0.0286 0.0084 3.41 0.0008 1.029 Neighborhood
landscape
25 Greenery Principal component of 0.0262 0.0150 1.75 0.0827 1.027 Neighborhood
landscape
R2 0.686
2
Adj R 0.638
N 187
* Dummy variables for railway lines were based on JR Kagoshima line.

3. Comparison of the evaluation in two cities

16
 The empirical results on the evaluation of urban landscapes in Tokyo and
Kitakyushu city are quite identical. Both demonstrated that compatibility and greenery
are mostly concerned attributes in landscapes, and in either case, they were significant
determinants of land prices.
Although the evaluations for 11 factors being investigated were quite different (as
shown in Fig. 1), t-tests showed that the levels of compatibility in the two cities did not
have significant difference (t= -0.25), and neither did the levels of greenery in the two
cities (t= 0.099).
In fact, the average unit price levels of the Tokyo and Kitakyushu samples are
602.4 and 73.2 thousand Yen/m2, respectively. Accordingly, the impact of compatibility
and greenery at the average unit price value in Kitakyushu city are 2.12 and 1.98
thousand Yen/m2. The estimates are comparable in scale with the results in Tokyo,
which are 7.9 and 8.4 thousand Yen/m2. With respect to absolute values, the economic
impacts of compatibility and greenery in Tokyo are higher, but with respect to the ratios
to land prices, that in Kitakyushu city are a bit higher.
Because PCA factors were linear combinations of the product of eigenvectors and
standardized landscape factors, we decomposed the estimates for compatibility and
greenery in 11 dimensions. Thereby, the marginal effects of the 11 landscape factors on
unit price were computed. Table 5 lists the results.

Table 5 Comparison of the results in Tokyo and Kitakyushu city

Tokyo Kitakyushu city

Marginal Elastic Average Elastic
effect on unit coefficient marginal effect coefficient
price (impact on on unit price (impact on
(thousand average unit (thousand
unit price
Yen/m2) price Yen/m2)
A1 Continuity of external walls (1,0,-1) +4.73 1.0079 +0.44 1.0060
A2 Conformity in colors and materials
+5.17 1.0086 +0.41 1.0057
(1,0,-1)
A3 Compatibility of buildings styles
+5.78 1.0096 +0.54 1.0074
(1,0,-1)

17
A4 Beauty of skylines constructed by
+2.76 1.0046 +0.26 1.0035
buildings (3,2,1,0,-1)
A5 Openness and the scale of buildings
-* 1.0000 +1.00 1.0137
(1,0,-1,-2)
A6 Visually nice and continuous greenery
+6.10 1.0101 +2.39 1.0326
(2,1,0)
B1 Greenery of walls and trees (1,0) +10.87 1.0180 +2.85 1.0390
B2 Greenery of open pedestrian spaces
+7.90 1.0131 +2.407 1.0327
(2,1,0,-1)
B3 Favorable pedestrian space
- 1.0000 +1.62 1.0221
(1,0,-1,-2,-3)
B4 Friendly outdoor space (1,0,-1) - 1.0000 +0.83 1.0113
B5 Decorations and street furniture (2,1,0) - 1.0000 +0.01 1.0002
* ‘-‘ means that the impact of the factor is not significant.

Table 5 confirms that the influences of landscape factors on land prices are
significant, and that the absolute effects of landscape factors are generally larger in
Tokyo while the effects with respect to unit price is somewhat higher in Kitakyushu city.
For instance, solely by 1-point increase in A1 (continuous external walls), the unit
prices can rise for 0.6-0.8%. The effects associated with greenery-related factors are
even larger.
The results provide important policy implications. One of special interests, for
example, is the results with B1 (greenery of walls and trees). In well greened areas in
Tokyo, land price is 10.87 thousand Yen/m2 higher than that of other areas. In
Kitakyushu city, the marginal effect is 2.85 thousand Yen/m2, which is also the largest
among considered factors. Actually, in many Japanese cities, local governments
encourage residents to greening their walls and fences along street by providing
subsidies or reducing tax. There are more and more examples in enlightened areas
where residents make agreements among themselves to green walls. Our analysis
strongly demonstrates that these activities were valuable.

4. Concluding remarks
As more and more people pursue for the beauty of urban landscapes, understanding
their economic values, especially their impact on land prices is valuable. It may
substantially raise the incentives of residents for preserving or creating landscape beauty.

18
It also suggests that planning policies that purposefully encourage and induce people to
do so are valuable. In addition, with the impact of landscape improvement or landscape
destroy being clarified, it makes possible to adjust benefits among residents and others,
and still, it may help public landscape management sectors to optimize their budget
plans.
The analyses confirm the usefulness of the three-step procedure for landscape
analysis, i.e., with standardized survey, PCA and hedonic analysis. Although Tokyo and
Kitakyushu city differ a lot in geographical and social economic conditions, land prices
levels, etc., their evaluation structures for urban landscape are amazingly similar. In
both cases, the compatibility of buildings and the greenery of neighborhood are
distinctly emphasized and in market, they are strongly evaluated. This implies the
importance of keeping compatibility and greenery levels in a general sense.
Throughout the analyses, it is noted that landscape aesthetics could hardly be
achieved solely by individual efforts; instead, collaborative efforts in neighborhood
level are extremely important. Therefore, activities leading to collaborative
improvement of landscapes, such as ‘district planning’ made by local government, or
‘building and landscape covenant’ made by residents should further be encouraged in
landscape management policies.

Acknowledgement
We are very grateful to the members of Evaluation of Residential Environment research
group who gave us much help on collecting data and valuable comments. This research
was sponsored by Ministry of Land, Transportation and Construction. it was also
supported by the Joint-research program of Center for Spatial Information Science,
University of Tokyo.

19
 Appendix 1 Data for two samples

Tokyo (valid sample size: 272) Kitakyushu city (valid sample size: 187)
Railway lines Railway lines
Time to the nearest station Time to the nearest station
Time to Yamanote line (railway line surrounding
Lot size
central areas)
Lot size Frontages
Frontages Shape of lot (irregular or not)
Shape of lot (irregular or not) Landform of lot
Landform of lot Number of front roads
Number of front roads Direction of front road
Direction of front road Width and lanes of front road
Prerequisites of development Right of road (public or private)
Gas Pavement of road
Width and lanes of front road Slope of road
Right of road (public or private) Cul-de-sac
Pavement of road Building setback along road
Slope of road Fence and walls
Cul-de-sac Adjacent land use (farmland, factory, parking lots, etc.)
Building setback along road Unpleasant facilities
Fence and walls Mixture of land use
Noise and vibrations Mixture of different height buildings
Adjacent land use (farmland, factory, parking lots, large Distance to public facilities (park, school, hospital,
open space, etc.) shopping center, etc.)
Unpleasant facilities (waste treatment, cemetery, etc.) Land use zone
Mixture of land use FAR and building coverage ratio designated by zoning
Mixture of different height buildings Evaluation of chome on vulnerability to fire disasters
Distance to public facilities (park, school, hospital,
Evaluation of chome on activities against fire disasters
shopping center, etc.)
Available sunshine duration (delimited by surrounding
Evaluation of chome on dangers to natural disasters
buildings)
Land use zone Evaluation of chome on hazard
FAR and building coverage ratio designated by zoning Evaluation of chome on criminal-prevention
Effective FAR Evaluation of chome on pollution and noise
Beauty area designated by planning Evaluation of chome on public transportation
Requirement for building setback Evaluation of chome on accessibility to artery roads
Economic rank of chome Evaluation of chome on welfare facilities
Planning activities (district plan) Evaluation of chome on medical facilities
High criminal occurrence area Evaluation of chome on daily facilities
Average elevation of chome Evaluation of chome on education facilities

20
Population density in chome, changing rate of population Evaluation of chome on commercial facilities
Proportion of road in chome Evaluation of chome on parks and public space
Proportion of vacant land in chome Evaluation of chome on coverage of greenery
Building coverage ratio in chome Evaluation of chome on open space
Density of wooden structure buildings Evaluation of chome on planning regulations in terms
of landscape beauty
Density of dilapidated old buildings Evaluation of chome on sustainability of environment
Evaluation of chome on balance of population

Appendix 2 Evaluations on A1 (continuity of external walls)

This factor focuses on the walls and external walls of buildings that are higher than
1.5 m (above eye-line) to see if they are well-aligned along street. Three situations are
separated.
+1 point: Most walls and buildings along street are well-aligned. They account for
more than 4/5 of the total in each side of the street. An example is shown in the
left of Fig. 2.
0 point: More than half of walls and buildings are well-aligned but no more than 4/5.
An example is shown in the middle of Fig. 2.
-1 point: There are many vacant lands, parking lots, houses without gate-walls, or
large buildings such as apartments or office buildings in the neighborhood.
Specifically, less than 1/2 of walls are aligned along street. An example is shown
in the right of Fig. 2.

+1 point 0 point -1 point

Figure 2 Pictures for reference in landscape survey

21
References
Angileri, V., Toccolini, A., 1993. The assessment of visual quality as a tool for the
conservation of rural landscape diversity, Landscape and Urban Planning, 24(1-4),
pp. 105-112.
Arai, Y. (2001) Evaluation indices of residential environment, in: Asami, Y. (ed)
Evaluation of Residential Environment. University of Tokyo Press, Tokyo, p.52
(Appendix).
Carter, R.W., Bramley, R., 2002. Defining heritage values and significance for improved
resource management: an application to Australian tourism, International Journal
of Heritage Studies, 8(3), pp. 175–199.
Coeterier, J. F., 2002. Lay people’s evaluation of historic sites, Landscape and Urban
Planning, 59(2), pp. 111-123.
Fukahori, K., Kubota, Y., 2003. The role of design elements on the cost-effectiveness of
streetscape improvement, Landscape and Urban Planning, 63(2), pp.75-91.
Gao, X, Asami, Y., Katsumata, W., 2005. Evaluating land use restrictions concerning
the floor-area-ratio of lots, Environment and Planning C: Government and Policy.
In press.
García-Mira, R., Arce, C., Sabucedo, J.M., 1997. Perceived quality of neighborhoods in
a city in northwest Spain: an individual differences scaling approach. J. Environ.
Psychol. 17, pp. 243–252.
Gómez-Sal, A., Belmontes J.A., Nicolau, J.M., 2003. Assessing landscape values: a
proposal for a multidimensional conceptual model, Ecological Modelling, 168(3),
pp. 319-341.
Green, R., 1999. Meaning and form in community perception of town character. J.
Environ. Psychol. 19, pp. 311–329.
Imamoglu, Ç., 2000. Complexity, liking and familiarity: architecture and
non-architecture Turkish students’ assessments of traditional and modern house

22
 facades. J. Environ. Psychol. 20, pp. 5–16.
İpekoğlu, B., 2006. An architectural evaluation method for conservation of traditional
dwellings, Building and Environment, 41(3), pp. 386-394 (forthcoming).
Lichfield, N., 1988. Economics in urban conservation, Cambridge University Press,
Cambridge.
Nishimura, Y., Machinami Research Group, 2000. Landscape Planning of Cities.
Gakugei Press, Tokyo.
Prato, T., 2000. Multiple attribute evaluation of landscape management, Journal of
Environmental Management, 60(4), pp. 325-337.
Willis, K.G., Garrod, G.D., 1993. Valuing landscape: a contingent valuation approach,
Journal of Environmental Management, 37(1), pp. 1-22.

23