ARTICLE IN PRESS
Habitat International 32 (2008) 471–484
www.elsevier.com/locate/habitatint
Urban population densities and their policy implications
in China
Minghong Tana,, Xiubin Lia, Changhe Lua, Wei Luob,
Xiangbin Kongc, Suhua Mad
a
Institute of Geographic Sciences and Natural Resources Research (IGSNRR), Chinese Academy of Sciences (CAS),
Beijing 100101, PR China
b
Research Center for Eco-Environmental Sciences, CAS, Beijing 10085, PR China
c
College of Resources and Environment, China Agricultural University, Beijing 10094, PR China
d
China Land Surveying and Planning Institute, Beijing 100035, PR China
Abstract
In China, some people believe that urban land has a large potential for absorbing more of the urban population, while
others think that urban population density is very high, and has already caused many urban problems. The population
density of 135 major cities is studied by using the city population data (Shiren kou) from the Fifth Census of China in 2000.
The data included the floating population who lived in cities for more than 6 months/year, so it could more closely reflect
the real size of the urban population. Land-use data were obtained from a digital map interpreted from remotely sensed
data collected in 2000. The results show that urban population density was fairly high in China and the average urban land
per capita of these cities was only about 76 m2 in 2000, but that the corresponding value was 106 m2 if it was calculated with
the non-agricultural population. Moreover, urban population density varied greatly between cities: from 4 103 to
22 103/km2. Regression results show the differences in urban population density were strongly related to six independent
variables, including wage per capita, city size and shape index of urban land, etc. The policies derived from the results
deserve more attention in the new land-use planning with the target year of 2020 in China.
r 2008 Elsevier Ltd. All rights reserved.
Keywords: China; Urban population density; Natural and socio-economic parameters; Policy implication
Introduction
It is not so easy to define urban population density, because of the definition of urban population is vague
and the urban area boundary is unclear. However, urban population density continues to inspire researchers
working in the fields of urban planning, geography and urban/regional economics (Kasanko et al., 2005),
because it is closely related to land use, urban traffic, environment and living quality, etc. (Camagni, Gibelli, &
Rigamonti, 2002; Li, Sato, & Zhu, 2003).
Corresponding author. Tel.: +86 10 64889450 802; fax: +86 10 64854230.
E-mail addresses: [email protected] (M. Tan), [email protected] (X. Li), [email protected] (C. Lu), [email protected] (W. Luo),
[email protected] (X. Kong), [email protected] (S. Ma).
0197-3975/$ - see front matter r 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.habitatint.2008.01.003
ARTICLE IN PRESS
472
M. Tan et al. / Habitat International 32 (2008) 471–484
In different countries/regions, there are large variations in urban population density. Generally speaking,
urban densities in American and European countries are lower compared with densely populated cities in
Asia, such as Hong Kong, Singapore, Tokyo, Seoul and Shanghai. In the Western context, high-density
development is often considered to be over-crowded and cause excessive congestion, with potentially
detrimental effects on society (Chan, Tang, & Wong, 2002; Oh et al., 2005; Williams, 1999), although more
and more researchers are noticing the negative impacts of low-density development and believe that there are
better modes of urban growth in the future, such as compact growth, growth management, etc. (Alig, Kline, &
Lichtenstein, 2004; Frenkel, 2004; Nelson, 1999). On the contrary, in Asian countries, more people believe that
dense and intense land use could be a solution used to satisfy rapidly increasing land demand for urban
activities (Chen & Jia, 2005; Lau, Giridharan, & Ganesan, 2005).
Urban population density has become one of the important concerns of policy makers and researchers (Fu,
2001), because China is a country with a huge population size of 1.3 billion and limited land resources (National
Bureau of Statistics of China (NBSC), 2005). Furthermore, as one of the most dynamic countries, China’s
sustained economic development and massive rural–urban migration have increased urban land demand. In the
1990s, built-up areas increased by 1.76 million ha, 80% of which was from arable land (Liu, Zhang, & Zhuang,
2003). At the same time, the arable land area per capita in China is only 0.094 ha, less than half of the world’s
average (Land and Resources Minister of China, 2005; Lin & Ho, 2003). Hence, the Chinese government strictly
controls the conversion of arable land to built-up land. In 1997 and 2004, the central government even stopped
approving any application for non-agricultural occupation of arable land across the country. The discrepancy
between urban land demand and arable land protection is becoming acute.
In China, more intensive urban land use is often suggested as a solution to the discrepancy. One of the
reasons is that most researchers prefer to use data on the urban non-agricultural population to calculate urban
land per capita since the data are more continuous and steady compared with other urban population data. In
the country, the urban non-agricultural population only includes the permanent residents in urban areas, who
engage in the secondary and tertiary industries and people they feed (National Bureau of Statistics of China
(NBSC), 1991–2001). These data neglect the floating population who actually work and live in urban areas.
This leads to considerable distortions of urban population in some cities with a large inflow of migrant
workers (Zhou & Yu, 2004). In most cases, urban population is under-estimated. Thus, it is easy to draw the
conclusion that urban land has a big potential for carrying an increased urban population in the country.
On the contrary, others believe that the urban population density is very high in China, which has caused many
problems. Firstly, over-development and over-concentration of urban areas results in various environmental
problems, such as air and water pollution, and water shortage (Chen & Lotspeich, 1998). Secondly, a high urban
population density leads to serious traffic congestion, especially in large cities, such as Beijing and Shanghai.
Thirdly, in part due to the high urban population density, housing prices in cities are continuously increasing. For
instance, the average housing price in China rose by 14.4% in 2004 (Liang & Ma, 2004; XinhuaEnglish, 2005).
Now, the new land-use planning (Tudiliyong Zongtiguihua Xiubian) is under way in China with the target
year of 2020. The planning will continue to place special emphasis on arable land protection through
controlling urban land expansion and intensively using existing urban land.
In this context, it is necessary to study urban population density with more reliable data, which will be discussed
in the Data section. The rest of the paper is organized as follows. In the Method section, a set of simple indicators
to interpret urban population density is introduced. Results section examines the differences of population density
for 135 selected cities across the country and the factors causing the differences. Discussion section presents several
policy implications arising from the results in the Results section. Finally, short conclusions are offered in the
Conclusions section. We hope that this study will help planners to strike a better balance between arable land
protection and urban land growth, and allow more reasonable planning for future land use in China.
Data
Urban population data
In China, there are three levels of cities, namely municipalities, prefecture-level and county-level cities. The
selected 135 cities in this paper are all above (and not including) the county level (Fig. 1). In the country, these
ARTICLE IN PRESS
M. Tan et al. / Habitat International 32 (2008) 471–484
473
Fig. 1. One hundred and thirty-five cities and their population densities (1000 persons/km2) in China in 2000. Note: The black line goes
along Huai River and Qinling Mountanin, which divide China into two regions, namely, South China and North China.
cities often administer a large area, comprising city proper, extensive exurban districts and rural counties
(Fig. 2). Population in the administration area of a city was divided into three parts in the Fifth National Census
of China in 2000: urban population, town population and rural population (National Bureau of Statistics of
China (NBSC), 2002). Fig. 2 shows the population composition of municipalities or prefecture-level cities in
China. In the Census, the city population is defined as the urban population in the city proper and exurban
districts, and does not include the town population or rural population (Shi Renkou). In addition, in the Census,
floating people were recorded at their current abode rather than at their place of registration if they lived at the
abode for more than 6 months/year. So, the city population, including some of the floating population, reflects
more closely the real size of the urban population, when compared with other population data in China (Zhou &
Yu, 2002). The city population data are used to study urban population density in this paper.
Urban land-use data
The basic urban land-use data were extracted from the 1:100,000 digital land-use map of 2000 which was
obtained from the Landsat Thematic Mapper images. These data are provided by the Resources and
Environment Data Center, CAS. The field survey and random sample check (covering a survey line of
70,000 km and 13,300 patches) testified that the average interpretation accuracy for land-use/land-cover was
better than 92% (Liu, Liu, & Zhuang, 2003). The data have been successfully employed to study land-use
changes in China (Liu, Liu et al., 2003). The original map consists of six classes of land-use types, i.e., arable
land, forest, grassland, residential areas and land for stand-alone industrial and mining sites, water, and
unused land. Here, urban land, one of the subclasses of residential areas and land for stand-alone industrial
and mining sites, is selected to study urban population density.
Urban land includes urban built-up areas in urban districts (city proper, and exurban districts) in this study.
For example, in Beijing in 2000, urban land included urban built-up areas (not including built-up areas
for rural residents) in the city proper and exurban districts of Shunyi, Changping, Fangshan and Tongzhou
(Fig. 3).
In the city proper, all population who live here for more than 6 months/year, belongs to the city population,
so it is easy to extract the urban built-up area. In an exurban district, only the members of the population, who
live in the central area where the district government is located, belong to the city population. A city often has
several exurban districts. As a result, it is possible to miss urban built-up areas in some exurban districts. In
order to ensure that no exurban districts are missed, we have examined each exurban district for all the cities
selected in this paper according to administrative maps at different levels. Thus, the spatial scope of the city
population is consistent with that of urban land.
ARTICLE IN PRESS
M. Tan et al. / Habitat International 32 (2008) 471–484
474
Rural population, town population
Counties
Exurban districts
City proper
Urban population, town population, rural population
Urban population
Fig. 2. Division of administrative areas of municipalities or prefecture-level cities in China. Note: The population in a county includes
town population and rural population, and that in exurban district includes urban population, town population and rural population. In a
city proper, there is only urban population. According to the Fifth National Census, city population includes two parts, that is, urban
population in city proper and exurban districts (National Bureau of Statistics of China (NBSC), 2002).
Fig. 3. Urban built-up areas and administrative division of Beijing in 2000. Note: Urban land in this paper refers to urban built-up areas in
the city proper and exurban districts.
In addition, socio-economic data are sourced from Urban Statistical Yearbook of China (National Bureau of
Statistics of China (NBSC), 1991–2001). Their meanings are shown in Table 1.
Method
Urban population density is influenced by various factors: natural resources endowments, economies of
scale in production and consumption, usage of cars, etc. (Alig et al., 2004; Fujita, 1989; Kasanko et al., 2005;
Oh et al., 2005). Through studying 15 European urban areas, Kasanko et al. (2005) present a more
comprehensive understanding for the differences of population density and think that geographical
surroundings, historical onset of the urbanization process, policy factors etc., lead to the differences of
urban population density among different parts of Europe, namely, compact cities in southern Europe, cities
with looser structures in eastern Europe, and central and western cities midway between the extremes.
ARTICLE IN PRESS
M. Tan et al. / Habitat International 32 (2008) 471–484
475
Table 1
The independent variables and their meanings
Variable group
Variable
Code of variable
Ncityi
Cities in North China
Geographical
surroundings
Regional
dummy
variables
Dummy
variables of city
types
PLcityi
MFcityi
Cities on the plains
Cities at the foot of mountains
Urban shape
Shape index of
urban land
Shapei
City size
Salary level
Sizei
Wageij
1000 persons
Yuan/person
Economic
structure
Investment in
fixed assets
Sgdp2i
Percent (%)
Fixij
Yuan/km2
Arable land per
capita
Arablei
ha/person
Economic factors
Social factors
Unit
Meaning
City size of city i in terms of city population
Average wage of fully employed staff and
workers of city i in year j
Share of secondary industry in GDP of city
i
Investment in fixed assets per km2 of city i
in year j
Arable land per capita in the scope of urban
districts of city i
Note: GDP means gross domestic product.
Urban population density
Geographical surroundings
Urban shape
Economic factors
Social factors
Fig. 4. Factors affecting urban population density.
However, it is unfortunate that, to date, only a few researchers have established a set of indicators to interpret
differences of urban population density.
In this study, a simply explanatory framework is established with a set of indicators in order to explain the
differences of urban population density among the selected cities (Fig. 4 and Table 1). Selection of independent
variables mainly depends on the following principles in this study. Firstly, data should be available. Secondly,
effects of an independent variable on urban population density should be easily measured. For example, some
social factors, such as historical tradition and cultural background, may affect urban population density, but the
effects are difficult to estimate. Thirdly, serious multicollinearity problems should be avoided among
independent variables. Selection of independent variables will be discussed in depth as follows.
Geographical surroundings
Geographical surroundings can largely express the characteristics of the natural conditions and socioeconomic development. This paper will consider the effects of geographical surroundings on urban population
density from two aspects: regional location and landforms around a city.
Firstly, in China, regional features vary greatly due to an area of 9.6 million km2. To express the regional
differences, mainland China is traditionally either divided into three large regions: the East, the Middle and
the West, or divided into two: the North and the South, by the geographic boundary of the Huai River
and Qinling Mountains. Since Fig. 1 clearly shows the difference in urban population density between the
ARTICLE IN PRESS
476
M. Tan et al. / Habitat International 32 (2008) 471–484
Fig. 5. Landform and spatial distribution of 135 cities in China. Note: Plain city refers to cities on the plains, and Foot and Mountain
cities are those at the foot of mountains and in mountain areas, respectively.
North and the South, the regional dummy variable of the northern cities (Ncityi) is set here. Ncityi is a binary
variable. If a city is in North China, its value equals one, and zero otherwise.
Secondly, landforms around a city may affect its population density. According to the landform
characteristics around cities, they are divided into three groups: cities on the plains (PLcityi), cities at the foot
of mountains (MFcityi) and cities in mountain areas. This division is realized by overlaying China’s landform
map and urban map (Fig. 5). Thus, two dummy variables of PLcityi and MFcityi are set and they are also
binary variables.
Urban shape
The shape index of the landscape has been used in many studies (Chust, Ducrot, & Pretus, 2004). Given that
the urban land of a city is often composed of many land patches, the shape index of city i (Indexi) is defined by
the following formula:
Pn
j¼1 Pij
q
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Indexi ¼
,
Pn
2 p j¼1 Aij
where Pij and Aij are the perimeter and area of land patch j of city i, respectively. Indexi ¼ 1 when the urban
land of a city is composed of a single circular patch. With the number of patches increasing or as the patch
shape becomes more irregular, the value of Indexi will be higher. A higher value means a longer perimeter per
unit area of urban land. A longer perimeter means a longer rural–urban interface, often characterized by low
buildings, less building form and more natural surroundings. So, a city with a higher value of Indexi may be
less compact.
Economic factors
For a city, economic factors mainly include city size, economic development level and economic structure.
Firstly, in urban areas, many firms are located close together in order to reduce transport costs and increase
the opportunities for face-to-face communication with other firms and customers (Fujita, 1989). Hence, the
urban population density may be higher because more human activities bid for scarce land resources in a
ARTICLE IN PRESS
M. Tan et al. / Habitat International 32 (2008) 471–484
477
bigger city. For example, after examining 88 US cities in 1960, Hoch (1972) found that urban density increases
with city size.
Secondly, urban economic development level can be mirrored by income per capita, GDP per capita, etc.
Because, many studies prove that land consumption per capita grows with an increase of income (Alig et al.,
2004; Camagni et al., 2002; Zhang, 2001), wage per capita is used to interpret urban population density in this
study.
However, wage per capita cannot affect urban land demand immediately. For instance, an urban family
needs some time to save up to buy a house. Hence,
the sum of average wage per capita of fully employed
P
P staff
and workers in the last 5 years (1996–2000) ( Wageij) is used as an independent variable (Table 1). Wageij
is calculated at the constant price of 2000.
Economic structure may have some effects on urban population density. Secondary industry, in particular
heavy industry, may need more urban land and less labor relative to tertiary industry, because manufacturing
processes of the former are capital intensive, and its input and output requires significant amounts of land
(Goldberg & Chinloy, 1984). Since there is an obvious collinearity relationship between shares of secondary
and tertiary industries in GDP, only the former (the variable of Sgdp2i) is employed as one of the independent
variables (Table 1).
In addition, higher investment in fixed assets per km2 (Fixij) can improve building conditions and even offset
some of the disadvantages of the natural environment, for instance, construction of transport infrastructure
and drainage. Moreover, by building some industrial and service establishments, higher investment can create
more social and economic activities, which will present more job opportunities. Hence, Fixij is selected as one
of the independent variables.
Because it is a stockP
variable, the sum of investment in fixed assets during the last
P
5 years (1996–2000) ( Fixij) is employed here. Fixij is calculated at the constant price of 2000.
Social factors
It is very difficult to estimate the effects of social factors on urban population density, because social group
covers a wide range of factors, including cultural background, historic tradition, policy, land ownership and
social fairness, etc. (Ding, 2007). Of all the social factors, the policy factor may play an important role. For
example, adopting a growth-management policy could contribute to the prevention of urban sprawl and the
preservation of open space, which can effectively control the urban de-concentration trend (Alig et al., 2004;
Frenkel, 2004; Nelson, 1999).
In China, of all the policies related to urban population density, the policy of dynamic balance of arable land
plays a central role. It emphasizes that the government at provincial level must ensure that arable land
quantity is in a state of dynamic balance in the jurisdictions (Yang & Li, 2000). This means that governments
in some regions with less arable land resources are forced to adopt a more compact urban land-use policy.
Here, arable land per capita in 1995 is used as an independent variable (Rural Survey Organization of
National Bureau of Statistics of China, 1996), for two reasons. First, for 135 cities, only the arable land data
for 1995 and 1999 can be obtained. Here, the arable land of a city only refers to that in its urban districts
(including the city proper and exurban districts) (Fig. 2). Second, impacts of arable land per capita on urban
population density cannot be reflected immediately.
Finally, as linearity is necessary in the multiple regression analysis, this paper uses logarithms of the
dependant and independent variables. Since Ncityi, PLcityi and MFcityi are binary variables, which equal one
or zero, they are not transformed into logarithms.
Results
The results show that urban land per capita of the 135 cities was about 76 m2 in 2000. However, the
corresponding value would approach 106 m2 if it was calculated with the non-agricultural population data.
The latter is about 47% higher than the former, especially in large cities with city populations more than
5 million (Fig. 6).
ARTICLE IN PRESS
M. Tan et al. / Habitat International 32 (2008) 471–484
478
Urban land per capita (m2/person)
164
144
124
104
84
64
Land 1
Land 2
44
24
4
>5 million
1-5 million
0.5-1 million
City classification
<0.5 million
Urban land per capita (m2/person)
Fig. 6. Urban lands per capita of four classifications of cities in China in 2000. Note: In this figure, 135 cities are divided into four
classifications according to city population and non-agricultural population, respectively. Lands 1 and 2 are urban lands per capita
calculated with city population and non-agricultural population, respectively. Urban land data are from the Resources and Environment
Data Center of CAS.
Land 1
120
Land 2
110
100
90
80
70
60
50
40
Cities in North
China
Cities in South
China
Average
City classification
Fig. 7. Urban lands per capita of two types of cities in China in 2000. Note: Lands 1 and 2 are urban lands per capita calculated with city
population and non-agricultural population, respectively. Urban land data are the same as Fig. 6.
Urban population density varied from 4 103/km2 in Kalamyi city in Xinjiang Autonomous Region to
more than 20 103/km2 in Fuzhou city in Fujian Province in 2000 (Fig. 1). Of all the 135 cities, 120 had urban
population densities exceeding 10 103/km2.
In North China, urban land per capita was 84 m2, much higher than 69 m2 in South China (Figs. 1 and 7).
7 of the total 8 cities with an urban population density higher than 20,000 persons/km2 were situated in the
Yangtze River Delta in South China, such as Hangzhou, and Ningbo (Fig. 1). However, it is notable that
the urban population density in South China would be lower than that in North China if it was calculated with
the non-agricultural population, and urban land areas per capita in South China and North China were 110
and 102 m2 in 2000, respectively (Fig. 7).
To shed light on the differences in urban population density, the variables in Table 1 are entered as a block
in a single step in the linear regression model of the SPSS software package. Table 2 lists the results. Overall,
our results are consistent with the discussion in the Method section above. Of all the independent variables, six
of them exhibited close relationships with the dependent variable at the 5% statistically significant level,
ARTICLE IN PRESS
M. Tan et al. / Habitat International 32 (2008) 471–484
479
Table 2
Variable coefficients in the linear regression model
Model
Unstandardized coefficients
Standardized coefficients
B
Std. error
Beta
t
Sig.
Collinearity statistics
Tolerance
VIF
Geographical
surroundings
(Constant)
Ncityi
PLcityi
MFcityi
7.735
0.298
0.097
0.050
0.848
0.044
0.045
0.044
0.489
0.143
0.077
9.116
6.852
2.166
1.153
0.000
0.000
0.032
0.251
0.643
0.748
0.737
1.554
1.337
1.356
Urban shape
Shapei
0.157
0.050
0.239
3.157
0.002
0.570
1.753
Economic
factors
Sizei
P
Wageij
P
Fixij
Sgdp2i
0.181
0.748
0.272
0.162
0.027
0.095
0.039
0.093
0.513
0.632
0.492
0.110
6.648
7.865
7.040
1.743
0.000
0.000
0.000
0.084
0.551
0.506
0.672
0.819
1.816
1.975
1.489
1.221
Social factors
Arablei
0.034
0.024
0.092
1.449
0.150
0.809
1.236
Note: Dependent variable is urban population density. And the adjusted R2 is 0.560.
P
including Salaryij, Sizei, Fixi, Ncityi, Shapei and PLcityi. Among them, the variables of Sizei and Fixij had
positive relationships with urban population density, while the remaining four variables all exhibited negative
relationships with the dependent variable (Table 2).
The relationship between Sgdp2i and urban population density was not as apparent as we expected. One of
the important reasons is that most cities in China are in the process of industrialization, and their shares of
secondary industry in the GDP were very high. The shares of 129 of 135 cities were higher than 40% in 2000,
which weakened the explanatory power of the differences of urban population density between cities.
Similarly, because of the implementation of the strict farmland protection policy across the country, the
variable of Arablei had a very weak relationship with urban population density.
MFcityi had no obvious relationship with urban population density, which meant the urban population
density of cities in mountain areas was approximately equal to that of cities at the foot of mountains (Table 2).
In the model, the adjusted R square is 0.560. Collinearity statistics results show that no obvious
multicollinearity exists among independent variables since the values of variance inflation factor (VIF) are all
lower than two (Table 2). In general, if the VIF value is more than five, there will be significant
multicollinearity between independent variables (Zhu, 2004).
Discussion
Compared with the western countries (Demograhia, 2005; Kasanko et al., 2005; Kline, 2000; Nelson, 1999),
the urban population density in China is very high. Population densities of some cities selected in this study are
even higher than those of other densely populated countries, such as Japan, Israel and India (Demograhia,
2005; Frenkel, 2004). Of the 20 urban areas with the highest population density in the world, eight are located
in mainland China (Demograhia, 2005). In addition, Hong Kong (China HKSAR) and Taipei (ChinaTaiwan) rank in first and fifth place, respectively. In the future, China’s government will continue to
implement the policy of strict farmland protection and control urban land expansion. In order to mitigate the
discrepancy between urban land growth and arable land protection, the following countermeasures are put
forward based on the results of this study.
To more exactly assess urban land demand
It is strongly necessary to assess urban land demand, in order to strike a balance between urban land
expansion and arable land protection. At present, two aspects deserve more attention.
ARTICLE IN PRESS
M. Tan et al. / Habitat International 32 (2008) 471–484
480
10000
RMB (Yuan)
8000
6000
4000
2003
2001
1999
1997
1995
1993
1991
1989
1987
0
1985
2000
Year
Fig. 8. Changes of per capita annual income of urban households, 1985–2004. Note: Data are from (National Bureau of Statistics of
China (NBSC), 2005), and income per capita is computed at constant price for 2004. RMB (Yuan) is China’s money (now about 7.5 Yuan/
US dollar).
Firstly, China is a developing country with a low urbanization rate (41.2% in 2004) (National Bureau of
Statistics of China (NBSC), 2005), so many factors causing urban agglomeration still work. As a result, the
urban population of nearly all cities is increasing, because of massive migration into cities from rural areas.
With the step-by-step cancellation of the ‘‘urban household registration system’’, more floating people will live
in the urban areas and enjoy the same urban welfare and equal job opportunities as urban residents, which can
cause urban land growth.
Secondly, urban land-use planning is often based on the assumption that urban land demand per capita is
constant over time in the country. However, the regression model shows that wage per capita had a negative
relationship with urban population density. In China, per capita annual income of urban households is
increasing rapidly. From 1985 to 2004, it increased by 339% (Fig. 8). With income increasing, land
consumption per capita may increase (Mak et al., 2007).
In China, economic growth and massive migration from the rural population are exacting an unprecedented
demand on urban land, so it is important to assess urban land demand for both maintaining urban economy
growth at a stable and relatively fast rate, and for avoiding excessive expansion of urban land and enclosure of
land.
To establish discrete criteria for quantifying urban land per capita for different types of cities
China has 660 cities, with diverse characteristics of urban development. So, it is necessary to establish
discrete criteria for quantifying the urban land per capita for different types of cities. Based on the results of
this study, several types of cities should be taken into account.
Firstly, South China has a higher urban population density than North China, due to better natural
conditions in South China, such as warmer weather, and richer water resources. But, in many studies, urban
population densities are often seriously underestimated in some cities of South China. So, in South China, we
should appropriately increase land supply to meet urban land demand in order to maintain sustainable urban
growth.
Secondly, city size had a positive relationship with urban population density (Table 2). The standardized
coefficient was 0.513. In fact, this relationship has already been mentioned in the planning criteria of urban
land per capita promulgated by the Minister of Construction of China in 1991. However, the criteria
emphasize calculation of urban population density with the non-agricultural population. This hides the real
relationship between urban population density and city size, because the non-agricultural population seriously
underestimates city size in China, especially in large cities with city populations over 5 million (Fig. 6).
ARTICLE IN PRESS
M. Tan et al. / Habitat International 32 (2008) 471–484
481
Thirdly, in the regression model, cities in plain areas were negatively related to urban population density.
This implies that urban population densities of these cities were lower than those of cities at the foot of
mountains. The main reason may be that cities in plain areas can expand in every direction with fewer natural
barriers. Owing to the importance of the plain in grain production in China, a more strict policy should be
implemented to control urban land expansion of cities in plain areas.
To efficiently use urban investment in fixed assets
P
In the regression model, the variable Fixij had a positive relationship with urban population density.
In China, urban investment in fixed assets per km2 varies greatly from region to region. From 1996 to 2000,
cities with the sum of urban investment in fixed assets more than 30,000 Yuan/km2 were mainly located in east
China, especially in the provinces of Hebei, Jiangsu, Zhejiang and Guagndong. Cities with values less than
20,000 Yuan/km2, were mostly in Northeast China and West China, including Liaoning, Shanxi, Inner
Mongolia and Guangxi (Figs. 9 and 10). Most of these belong to mining cities, such as the cities of Liaoyang,
Fushun, and Anshan in Liaoning province, Panzhihua city in Sichuan Province, and Hegang city in
Heilongjiang province. Now most of the mining cities are facing resources exhaustion, and construction
conditions are destroyed due to mining activities. Thus, these cities develop very slowly and cannot attract
more migrants. So, the cities need more urban investment for urban development.
By contrast, in some cities in developed regions, urban investment in fixed assets still increases quickly
although the amount is already very large. It is known that unscientific investment can cause serious economic
loss and even has negative effects on urban development and urban environment. For instance, in Hangzhou
in Zhejiang Province, a 22-story building on the bank of the famous West Lake was pulled down in 2007. The
building served a mere 13 years, although it cost 20 million Yuan (about US$2.7 million) and was designed to
last 100 years (Wu, 2007). Actually, in China, many high buildings and some basic infrastructures in
developed regions suffer similar fates, due to random decision making.
So, how to efficiently use urban investment is very important for China’s urban development, since high
urban investment in fixed assets cannot only improve the urban environment, but also help transform the
economic structure and create job opportunities to attract the urban population, especially for mining cities.
To properly increase population density in cities with high shape index
The regression model shows that the shape index of urban land had a negative relationship with urban
population density. In a city with a high shape index, its infrastructure cannot be efficiently used due to its
looser urban land-use structure. For instance, in Zibo city in Shandong Province, the shape index of urban
land was high and close to 6.61 in 2000. In this city, it is difficult for urban residents who live in Place A to use
Fig. 9. Urban investment in fixed assets (Yuan/km2) of 135 major cities in China from 1996 to 2000.
ARTICLE IN PRESS
482
M. Tan et al. / Habitat International 32 (2008) 471–484
Fig. 10. China’s administration map at provincial level.
Fig. 11. Urban land-use map of Zibo city in Shandong Province in 2000.
the public infrastructure in Place B (Fig. 11). Hence, in Zibo, the population density was only
7.37 103 people/km2 (135 m2 person) although it was a large city with a 1.76 million city population in 2000.
In some cities composed of several separate sub-centers, the urban population scale of each center is not
very large, which can mitigate urban problems (e.g., traffic over-crowding) to some extent. So we should
properly increase urban population density of these cities by improving the public infrastructures. In the
world, there are many cities with high shape index, which have very high urban population densities, such as
Singapore, and Hong Kong (Liang, 1973).
Conclusions
In China, the urban population density is often underestimated, because it is calculated with the nonagricultural population, which does not include the massive migration from rural areas. As a result, most
ARTICLE IN PRESS
M. Tan et al. / Habitat International 32 (2008) 471–484
483
people believe that urban land has a big potential for burdening urban population and that it is necessary to
strictly control urban land growth in order to protect arable land.
However, if taking into account the floating population, population density in most cities is very high. In
2000, the average urban land per capita was only 76 m2. Moreover, urban land per capita varied greatly among
these cities. Now, rapid economic growth and massive rural–urban migration are exacting unprecedented
demands on urban land. It is very important to assess urban land demand for different types of cities with
reliable data.
To analyze the differences in urban population density between cities, a simple explanatory framework is
established. However, a city is a very complicated system, and its formation and growth are affected by many
factors. Due to lack of data, some explanatory variables were not considered in the framework. This could be
improved in further studies, by collecting more data to enhance our understanding of the characteristics of
urban population density in China.
Acknowledgments
The authors are indebted to the Natural Science Foundation of China (Project no. 40501005), Foundation
of Knowledge Innovation Program of CAS (KZCX-YW-421) and the Foundation of National Science and
Technology Supporting Program of China (2006 BAB15B00). Professor Zhuang Dafang of the Resources and
Environment Data Center, CAS, is acknowledged for kindly providing the digital data of urban land. The
comments from two reviewers have been greatly helpful for strengthening the arguments of the paper.
References
Alig, R. J., Kline, J. D., & Lichtenstein, M. (2004). Urbanization on the US landscape: Looking ahead in the 21st century. Landscape and
Urban Planning, 69, 219–234.
Camagni, R., Gibelli, M. C., & Rigamonti, P. (2002). Urban mobility and urban form: The social and environmental costs of different
patterns of urban expansion (special section: Economics of urban sustainability). Ecological Economics, 40, 199–216.
Chan, E. H. W., Tang, B. S., & Wong, W. S. (2002). Density control and the quality of living space: A case study of private housing
development in Hong Kong. Habitat International, 26, 159–175.
Chen, A., & Lotspeich, R. (1998). Determinants of industrial residuals: Generation and treatment: Evidence from Chinese cities.
Ecological Economics, 26, 97–108.
Chen, H., & Jia, B. (2005). Centralization or decentralization: The trend of Chinese city in the rapid urbanization. City Planning Review,
30(5), 61–69.
Chust, G., Ducrot, D., & Pretus, J. L. (2004). Land cover mapping with patch-derived landscape indices. Landscape and Urban Planning,
69, 437–449.
Demograhia. (2005). World urban areas (areas of continuous development). /http://www.demographia.com/db-worldua-dens.pdfS.
Ding, C. (2007). Policy and praxis of land acquisition in China. Land Use Policy, 24, 1–13.
Frenkel, A. (2004). The potential effect of national growth-management policy on urban sprawl and the depletion of open spaces and
farmland. Land Use Policy, 21, 357–369.
Fu, Y. (2001). Site density restrictions: Measurement and empirical analysis. Journal of Urban Economics, 49, 404–423.
Fujita, M. (1989). Urban economic theory: Land use and city size. New York: Cambridge University Press.
Goldberg, M., & Chinloy, P. (1984). Urban land economics. Chichester: Wiley.
Hoch, I. (1972). Income and city size. Urban Studies, 9(3), 229–327.
Kasanko, M., et al. (2005). Are European cities becoming dispersed? A comparative analysis of 15 European urban areas. Landscape and
Urban Planning, 77(1–2), 111–130.
Kline, J. r. D. (2000). Comparing states with and without growth management analysis based on indicators with policy implications
comment. Land Use Policy, 17, 349–355.
Land and Resources Minister of China (2005). The investigation report of land use change in China. Beijing: The Dadi Press.
Lau, S. S. Y., Giridharan, R., & Ganesan, S. (2005). Multiple and intensive land use: Case studies in Hong Kong. Habitat International,
29, 527–546.
Li, L., Sato, Y., & Zhu, H. (2003). Simulating spatial urban expansion based on a physical process. Landscape and Urban Planning, 64,
67–76.
Liang, C.-s. (1973). Urban land use analysis: A case study on Hong Kong. Hong Kong: Goodyear Printing Press Ltd.
Liang, H., & Ma, N. (2004). China’s property market: Fallacies of an imminent bubble burst. Economic Flash, 1–14.
Lin, G. C. S., & Ho, S. P. S. (2003). China’s land resources and land-use change: Insights from the 1996 land survey. Land Use Policy, 20,
87–107.
ARTICLE IN PRESS
484
M. Tan et al. / Habitat International 32 (2008) 471–484
Liu, J., Liu, M., & Zhuang, D. (2003). Study on spatial pattern of land-use change in China during 1995–2000. Science in China (Series
D), 46(4), 373–384.
Liu, J., Zhang, Z., & Zhuang, D. (2003). A study on the spatial-temporal dynamics change of land-use and driving forces analyses of
China in the 1990s. Geographical Research, 22(1), 1–12.
Mak, S. W. K., et al. (2007). Privatization, housing conditions and affordability in the People’s Republic of China. Habitat International,
31, 177–192.
National Bureau of Statistics of China (NBSC) (1991–2001). Urban statistical yearbook of China. Beijing: China Statistic Press.
National Bureau of Statistics of China (NBSC) (2002). Tabulation on the 2000 population census of the People’s Republic of China. Beijing:
China Statistics Press.
National Bureau of Statistics of China (NBSC) (2005). China statistical yearbook. Beijing: China Statistic Press.
Nelson, A. C. (1999). Comparing states with and without growth management: Analysis based on indicators with policy implications.
Land Use Policy, 16, 121–127.
Oh, K., et al. (2005). Determining development density using the urban carrying capacity assessment system. Landscape and Urban
Planning, 73(1), 1–15.
Rural Survey Organization of National Bureau of Statistics of China (1996). Rural statistical yearbook of china. Beijing: China Statistic
Press.
Williams, K. (1999). Urban intensification policies in England: Problems and contradictions. Land Use Policy, 16, 167–178.
Wu, N. (2007). Bad urban planning scrutinized in cloud of dust. /http://www.china.org.cn/english/China/195765.htmS.
XinhuaEnglish. (2005). Land supply increased to help stabilize housing price. /http://english.sina.com/business/1/2005/0614/34741.htmlS.
Yang, H., & Li, X. (2000). Cultivated land and food supply in China. Land Use Policy, 17, 73–88.
Zhang, T. (2001). Community features and urban sprawl: The case study of the Chicago metropolitan region. Land Use Policy, 18,
221–232.
Zhou, Y., & Yu, H. (2002). Suggestion on revise on China’s urbanization level in standard of the fifth population census. Statistical
Research, 4, 44–47.
Zhou, Y., & Yu, H. (2004). Reconstructing city population size hierarchy of China based on the fifth population census(1). Urban
Planning, 28(6), 49–55.
Zhu, P. (2004). Modern econometrics. Shanghai: The Press of Shanghai University of Finance & Economics.