# RECORD 844f3013b415b6de89430c28be2c24c13f35196d
# TITLE: Using GIS and Hedonic in the modelling of spatial variation of housing price in Xiamen city
# YEAR: 2013.0  VENUE: nan  DOI: 10.14246/irspsd.1.4_29
# AUTHORS: Yuan Li; Lang-Jie He; W. Xu; Hui Wang; Zi-Hao He
# PAGES: 14  CANDIDATE TABLE PAGES (1-based): [9]
# ABSTRACT: Valuing the built environment with the support of GIS and housing price model is an interesting topic because it connects the spatial thinking with economic concerns. In this paper, we chose the ordinary commercial housing in Xiamen City as our research targets, and the technology of GIS and Hedonic price model were used. Xiamen city is a typical coastal city, and from the research result, we found that the distribution of higher housing price was kept consistent with the coastal areas of Xiamen city. Besides, the average housing price of Xiamen inner-island (the old urban island area) was far higher than the outer-island. To reveal the degree of different spatial elements affecting housing price, we chose 21 variables (three types: location characteristic, structure characteristic and neighbourhood characteristic) and measured them in GIS platform. The Hedonic model showed that the main elements that influencing the price of commodity housings in Xiamen included Property Management Fee, Distance to Commercial Center, Distance to Hospital, Distance to Primary School, Near Village or Not, Near Green space or Not, Near kindergarten or Not. At the end of the paper, we provided three suggestions to balance the spatial variation of housing price in Xiamen City.


===== PAGE 1 (score 9, numeric-paren 0, table-words 0, var-words 9) =====

29 
International review for spatial planning and sustainable development, Vol.1 No.4 (2013), 29-42 
ISSN: 2187-3666 (online) 
DOI: http://dx.doi.org/10.14246/irspsd.1.4_29 
Copyright@SPSD Press from 2010, SPSD Press, Kanazawa 
Using GIS and Hedonic in the modelling of spatial 
variation of housing price in Xiamen city 
Yuan Li¹*, Lang He¹, Wangtu Xu¹, Hui Wang¹, Zizhang He2 
1 School of Architecture and Civil Engineering, Xiamen University 
2 Xiamen Urban Planning & Design Institute 
* Corresponding Author, Email: Yako79@gmail.com
Received 16 December 2012; Accepted 5 July 2013
Key words: 
GIS, Housing Price, Spatial Variation, Hedonic, Xiamen City 
Abstract: 
Valuing the built environment with the support of GIS and housing price 
model is an interesting topic because it connects the spatial thinking with 
economic concerns. In this paper, we chose the ordinary commercial housing 
in Xiamen City as our research targets, and the technology of GIS and 
Hedonic price model were used. Xiamen city is a typical coastal city, and from 
the research result, we found that the distribution of higher housing price was 
kept consistent with the coastal areas of Xiamen city. Besides, the average 
housing price of Xiamen inner-island (the old urban island area) was far higher 
than the outer-island. To reveal the degree of different spatial elements 
affecting housing price, we chose 21 variables (three types: location 
characteristic, structure characteristic and neighbourhood characteristic) and 
measured them in GIS platform. The Hedonic model showed that the main 
elements that influencing the price of commodity housings in Xiamen included 
Property Management Fee, Distance to Commercial Center, Distance to 
Hospital, Distance to Primary School, Near Village or Not, Near Green space 
or Not, Near kindergarten or Not. At the end of the paper, we provided three 
suggestions to balance the spatial variation of housing price in Xiamen City. 
1.
BACKGROUND
Housing, which is one of the most basic human needs, is a hot economic 
issue and social issues related to people's livelihood. Many factors influence 
the housing price and many researchers try to reveal the distribution rules on 
spatial variation in housing price. Some people (Xu, 1997; Jiang and Zhu, 
2005; Meng, Zhang et al., 2005; Mei and Li, 2008) used GIS spatial 
interpolation method to get spatial distribution map and contour map of 
housing price. Some people (Luo, He et al., 2002) took Lanzhou as a case 
study and constructed the SD-GIS space-time simulation model to study the 
space-time distribution changes of residential price. Some (Montero and 
Larraz, 2009) considered that space in the real estate market is important and 
official averages do not take into account the spatial correlation of housing 
price. They proposed the Kriging mean method to estimate mean housing 
price. Some (Fik, Ling et al., 2003) hold the view that absolute location must 
be viewed as interacting with other determinants of housing value. They 
presented an interactive variables approach and tested its ability to explain 
price variation in an urban residential housing market. Some (Zhang, Meng 
et al., 2004; Armstrong and Rodríguez, 2006) established the Hedonic price 
method, studied the relationships between different traffic convenient degree 
@Licensee SPSD Press 
This open access article is published under a 
Creative Commons [Attribution-
NonCommercial-NoDerivatives 4.0 
International] license. 
https://creativecommons.org/licenses/by-nc-nd/4.0/ 


===== PAGE 9 (score 27, numeric-paren 0, table-words 9, var-words 5) =====

Li, et al 
37
 
 
10%. Combined with variance inflation factor (VIF) inspection, we weeded 
out the multicollinearity factors. Finally, there were 7 variables selected in 
the model. After running the regression analysis, the results were as shown 
in the Table 3. 
Table 2. Comparison of three functions 
Function type 
R 
R2 
Adjusted R2 
Linear model 
0.901 
0.812 
0.800 
Log-linear model 
0.897 
0.805 
0.790 
Semi-logarithmic model 
0.874 
0.764 
0.754 
Table 3. Regression coefficient analysis 
Variable 
Non-standardized 
coefficient 
Standardize
d coefficient 
t 
Sig. 
VIF 
B 
Std Error Beta 
Constant 
5274.659 1446.784 
 
3.646 
.000 
 
PropertyManagementFee (X21) 4880.947 445.076 
.613 
10.967 
.000 
1.270 
NextToCC (X2) 
-.602 
.153 
-.218 
-3.939 
.000 
1.247 
NextToHospital (X6) 
1.190 
.387 
.197 
3.071 
.003 
1.675 
NextToKPSchool (X3) 
-.271 
.127 
-.123 
-2.139 
.035 
1.354 
VillageEffect (X14) 
-2501.633 968.019 
-.139 
-2.584 
.011 
1.173 
GreenFieldEffect (X11) 
2974.278 1024.754 .163 
2.902 
.005 
1.289 
KGService (X10) 
2572.500 1004.815 .137 
2.560 
.012 
1.159 
5. 
RESULTS 
The regression coefficients for 7 characteristic variables of the linear 
model could be obtained by regression analysis. We calculated the housing 
Hedonic price model of Xiamen, as shown in Formula (7): 
 
Price=5274.659-0.602X2-0.271X3+1.19X6+2572.5X10+2974.278X11-2501.63     
3 X14+4880.947 X21                                   (7) 
 
According to the results of regression analysis, we could find the main 
elements that influence the price of commodity housings in Xiamen: 
PropertyManagementFee (X21), NextToCC (X2), NextToHospital (X6), 
NextToKPSchool (X3), VillageEffect (X14), GreenFieldEffect (X11), 
KGService (X10). 
1) Proximity of the commercial centers (X2) 
Generally speaking, commercial centers were the most active areas of a 
city as they provided various urban services and consumption choices for 
urban residents, which made it very convenient for people to live. 
Theoretically, the housing price in the surrounding area was relatively 
higher. It could be seen from the Hedonic price model that the variable 
coefficient of the distance to commercial centers was -0.602, suggesting that 
the distance from housing to business centers was inversely proportional to 
the housing price. With the distance between dwellings and business centers 
expanding, the residential price would fall. More specifically, every 1 


===== PAGE 10 (score 15, numeric-paren 0, table-words 2, var-words 11) =====

38 
IRSPSD International, Vol.1 No.4 (2013), 29-42 
 
 
kilometre of the increase in distance would cause 602 RMB (Yuan) loss in 
sale price. 
2) Proximity of the hospitals (X6) 
With the improvement of living standards, people paid more attention to 
medical and health care than before. As an important kind of public service, 
medical care provided protection for the life and health of the residents. 
“Time is life”, the distances between dwellings and hospitals measured the 
residents’ medical convenience. However, the coefficient of the model 
doesn’t reflect the relationship that the houses price increases among with 
the distance increasing. The most reason is the models don’t differentiate the 
needs of health care service between different age groups. The young people 
are strong and seldom go to the hospital. They usually will never take 
hospital into account even reject live close to hospital. Meanwhile the 
elderly  people are more often go to the hospital. Considering the access of 
the data, the Hedonic model didn’t specially research the age structure of 
families then ignored the elderly’s preference close to hospital. In the 
Hedonic price model, the variable coefficient of hospitalization was 1.19, 
and it meant that the sale price would increase 1,190 RMB as the distance to 
hospitals increase 1 kilometer.  
3) Proximity of the key primary schools (X3) 
Education was the foundation of the nation, which was related to the 
quality of people and long-term development of the country. As the 
high-quality education resources were limited and unevenly distributed, 
many parents in China rent or even bought houses near key schools in order 
to offer their children with better education. As can be seen from the above 
Hedonic price model, the key primary schools had a positive effect on 
housing price and the sale price would decrease 1,190 RMB per kilometer as 
the distance to provincial-demonstrated primary schools increased. 
4) Near village or not (X14) 
With the increasing development of Chinese economy, the process of 
urbanization was becoming faster and faster. Urban construction and urban 
area expanded to the outlying regions, which made “villages surrounded by 
cities” become the most common phenomenon. The area that had not been 
urbanized could be found in the urban area, and this was called “village in 
the city”. Cities were often close to rural settlements in the outreach process. 
The rural lands possessed features such as a high building density, a low 
greening rate, narrow and hindered streets, poor sanitation and so on because 
of land property restrictions and lack of reasonable and effective planning. 
Urban infrastructure and public service facility could not reach the whole 
city, so the development of urbanized area close to villages was kind of 
limited. As can be seen from the above Hedonic price model, villages had a 
negative effect on residential lands’ externality. The average price of the 
premises near the village was 2,501.633 RMB lower than the ones far away 
from the village. 
5) Near Greenland or not (X11) 
As the living standard improves and awareness of environmental 
protection enhances, residents were more willing to purchase houses with 
comfortable environment. Being the open space of a city, public green lands 
possessed functions like landscape, airing and leisure. They were provided 
with social, economic and ecological value, and made the comfortable living 
and working environment accessible to residents. It could be concluded from 
the Hedonic price model that public green lands had a positive effect on 
urban residential lands’ externality. The average price of the housings close