Sunteți pe pagina 1din 18

Gated community and residential segregation in urban China

Abstract
Research on gated community in Chinese cities has been growing very
fast, but empirical studies are limited, especially those on the relationship
between gated community and residential segregation. A retrospective
questionnaire survey was conducted in three gated communities in
Chongqing, China. The findings from the survey include that, after moving
into the gated communities, many homeowners contact with other people
decreases. It is also found that homeowners participation in local public
affairs decreases. These results hold even when we compare with the
reference group and control for duration of stay. They support the view
that gated community aggravates residential segregation. However, the
survey shows that many homeowners feel the income differences among
the neighbors increase and the changes of several types of external
activities dont show a consistent pattern. All these empirical findings
suggest that a complex relationship between gated community and
residential segregation exists in urban China and the removal of work unit
from the housing system affects peoples experience in gated community.
Introduction
Almost all new residential developments in Chinese cities, by name microdistrict (MD, or xiaoqu), are in the form of gated community that is
governed by the homeowners association (HOA).1 The definition of gated
community is walled and gated residential developments that restrict
public access (Atkinson and Flint 2004).2 Although most studies in the
English literature (see, for example, McKenzie 1994; Blakely and Snyder
1999) are critical of gated community, especially regarding its negative
impact on residential segregation, there are still opposing voices (Manzi
and Smith-Bowers 2005). In contrast, the views in the Chinese literature
are largely split along disciplines. Those in sociology, law and public
administration (see, for example, Chen 2009) often praise HOA for its role
in community self-governance and grassroots democracy, while those in
geography and planning (see, for example, Miu 2004) often criticize gated
community for aggravating residential segregation. It remains an open
question what is the relationship between gated community and
residential segregation in the Chinese city, especially from an empirical
perspective.
First of all, I need to point out that several terms of the segregation impact
of gated community are used in the literature, including social
segregation, residential segregation, housing segregation and
spatial segregation (Blakely and Snyder 1999; Low 2003; Manzi and
1

Smith-Bowers 2005). In this paper I will use these terms interchangeably


without paying extra attention to their minute differences although in
most of time I use residential segregation. By doing so I want to
emphasize that possible segregation effects of gated community cover
social, housing and spatial aspects, reflecting the changes of
homeowners behavior in all above dimensions.
By conducting a retrospective survey on residents in three gated
communities in Chongqing, China, we are able to obtain data about their
lives both before and after they moved into the gated communities. In this
way, we can study how their lives have changed after moving into the
gated communities and, consequently, what is the impact of gated
community on residential segregation. In the English and Chinese
literature that I can find, studies on residential segregation in China
mostly adopt two approaches. In the first approach (Huang and Yi 2009)
researchers calculate dissimilarity index or other similar indexes based on
census data, which can then be used to analyze the degree of residential
segregation. It essentially examines the changes of demographic structure
across space at a specific time. The spatial unit is often the street office.
But, this approach is difficult to apply to gated community, which is
usually smaller than the street office or even the neighborhood
committee, due to the lack of census data at this spatial scale. The second
approach (Huang 2005) relies on descriptive analysis of spatial
distribution of housing in a city. This is certainly weak in terms of
analytical depth and accuracy and relies heavily on anecdotal evidence.
My approach focuses on a particular gated community and analyzes the
changes of the homeowners perception, attitude and behavior after they
move there. In this way, I can study residential segregation based on
broader indicators such as income differences among the neighbors, their
connections outside the gated community, their participation in local
public affairs, and so on. In a sense, this is my methodological contribution
to the literature.
My findings include, firstly, that after moving into the gated community,
homeowners contact with other people decreases. Secondly, their
participation in local public affairs, especially activities organized by the
neighborhood committee, decreases after moving into the gated
community. These two findings, which hold even when compared with the
reference group (i.e. those who had lived in an MD before moving into the
current MD) or considering the duration of stay, are strong evidence for
the segregation impact of gated community, supporting mainstream
theories on this form of housing (Blakely and Snyder 1999; Low 2003;
Atkinson and Flint 2004). However, it is also found that the majority of
homeowners feel that income differences among the neighbors increase.
2

The changes of their external activities, such as shopping, dining, visiting


friends and recreation, dont show a consistent pattern. These results
suggest that the relationship between gated community and residential
segregation is very complex due to the transitional nature of the Chinese
city and history matters (Polanska 2010).
The remainder of the paper is organized as follows. The next section is
literature review, followed by brief theoretical analysis of residential
segregation
in
China. A conceptual framework is presented at the end of that section.
The fourth section is empirical analysis based on the survey. MD Y is
introduced in the fifth section as a typical case of gated community. I
analyzed the behavior of the reference group and the possible impact of
duration of stay in the destination in the sixth section. The last section
concludes the paper.
Literature review
There is a large body of English literature on private community including
gated community. Theoretical studies include both supporters (Foldvary
1994; Nelson 2009) and opponents (McKenzie 1994; Blakely and Snyder
1999) of private community. The former treats the private community as
market organization that provides local public goods. In this way, a market
of local governance can be established to enhance efficiency of resource
allocation. Critics on private community mostly focus on two aspects. First,
they argue that the private community represents the secession of the
successful from the society in a discourse of fear. It will split the society,
aggravate residential segregation, and lead to homogeneous
communities. Second, the private community will influence peoples
participation in local public affairs and, hence, weaken democracy and
civil society. It is fair to say that most critics, even including some
supporters of private community, agree that gated community
aggravates/memperburuk the problems of the private community.
Probably because it is difficult to obtain data about private community,
high-quality empirical studies are very limited. Most existing studies focus
on residential segregation, civil society and housing price. For example,
Barton and Silverman (1994) is a collection of some early studies on
private communities in the US. Some researchers (Barzel and Sass 1990;
Dilger 1992) found that private community is very efficient. Gordon (2003)
showed that private communities in California dont crowd out
participation in local politics. Pompe (2008) found that housing price in the
gated community sells at a premium of 18.6 %, which indicates positive
valuation of gated community by the market.
3

For the purpose of this paper, I am interested in the relationship between


gated community and segregation. The mainstream wisdom on this issue
is best represented by Atkinson and Flint (2004:877): gated communities
provide a refuge that is attached to social networks, leisure, schooling and
the workplace via paths which are used to avoid unwanted social contact.
My argument is that each of these spaces more or less segregates its
occupants from social contact with different social groups, leading us to
suggest that the impact of such residential division resembles a seam of
partition running spatially and temporally through cities, what we term
timespace trajectories of segregation. The problem with this strand of
thinking is that there is little empirical evidence to support this
argument (Roitman 2013:157). The contrarians include Manzi and SmithBowers (2005), who argued that gated community does not encourage
segregation since urban segregation has always existed.
As a matter of fact, there are a few empirical studies that attempt to
analyze the relationship between gated community and segregation.
Based on qualitative research in a gated community in Argentina, Roitman
(2005) found that segregation process related to gated community has
two sides, on both of which people feel segregated and discriminated
against. She (Roitman 2013) further studied social practices and
viewpoints of its residents, especially how they interact with those in nongated communities that surround them. In particular, she examined three
types of social practices: institutional communication, charity work, and
social relations and socialization settings. She found that most
neighborhood social practices contribute to segregation. The weakness of
Roitmans approach is that it relies on interviews and qualitative research
in one community. A different but methodologically similar research was
conducted in Ghana by Asiedu and Arku (2009), who found that there are
appreciable levels of interaction among the residents in three gated
communities. Their interviews with the residents reveal that there seem to
be substantial levels of interaction, albeit centered on economic
opportunities rather than social interaction, between non-gated residents
and gated residents. In addition to relying on qualitative methods, this
research only examines absolute levels of interaction without any
comparative perspective. This is definitely a shortcoming in research
design.
Another direction of research on this topic is Vesselinov (2008), who
conducted regression analyses for gating (percentage) and residential
segregation (dissimilarity index) based on American Housing Survey data.
She found that some well-established factors that affect segregation also
affect the process of gating. She concluded that gating and residential
segregation reinforce each other and, sometimes, are also alternatives to
4

each other. The problem with her approach is that it does not directly test
the causal relationship between gating and segregation.
There are a few studies on gated community in China in the English
literature. By comparing two major explanations of gated community, i.e.,
the discourse of fear and the club theory, Wu (2005) thought that the
latter is more applicable in the Chinese city. Huang (2006) argued that
gated community in the Chinese city is an extension of the cultural
tradition of collectivism. Read (2008) argued that homeowners association
can help build grassroots democracy, promote neighborhood selfgovernance and, ultimately, lead to civil society, albeit in a different way.
Given their similar historic background of planned economy, gated
communities in China should share some common characteristics with
their counterparts in Eastern Europe. Studies on the latter (see, for
example, Stoyanov and Frantz 2006; Polanska 2010) are also
accumulating. In general, most of them found that historic housing
condition in the communist era greatly affects peoples perception of and
experience in gated community. Polanska (2010) argued that gated
community is a reaction to housing conditions prevailing under
communism.
Researches on gated community in the Chinese literature have been
growing very fast. In sociology, law and public administration, most
researchers (see, for example, Zeng 2002; Chen 2009) praise HOA for its
role in grassroots democracy and community selfgovernance. In contrast,
studies in geography and planning are largely critical of gated community.
For example, Miu (2004) is an early paper on this topic that argues that
gated community is a cancer in the Chinese city. Many articles (Yang and
Min 2008; Qin et al. 2008; Yu et al. 2009; Yu and Zhang 2010; Liu and Li
2010; Song 2010; Xu and Yang 2010) simply introduce Western studies to
the Chinese audience, while some (Xu and Yang 2008; Liu and Li 2009;
Wang 2010; Dou 2010; Wei and Qin 2011) focus on the design of gated
community.
A rare empirical study is Fenget al. (2011),who did a survey in a MD in the
suburb of Guangzhou. They asked the respondents where they do
shopping, have dinner, visit friends or have recreational activities. They
found that those activities are not restricted within the walls but are
related to outside the MD. Hence, they argued that gated community
doesnt aggravate residential segregation in the context of the Chinese
city. But, there is a problem in their approach, which only examines how
the volumes or frequencies of various activities change with distance to
the MD. Residential segregation doesnt mean no relation (in the absolute
sense) to the other parts of the city but rather the decline of those
5

relations in the relative sense. This comparative perspective can be


horizontal, which means comparison between residents inside and outside
the walls, or vertical, which focuses on the change of residents behavior
after moving into the gated community.
In conclusion, both the English and the Chinese literature suggest that,
first, although many people emphasize the negative impact of gated
community on residential segregation, empirical studies are rare. On the
one hand, it is difficult to calculate quantitative indexes used to measure
segregation, such as the dissimilarity index, due to the lack of government
statistics at the spatial level of MD. On the other hand, there is no
consensus on the indicators of segregation at this scale. Second, studies
on residential segregation in the Chinese city are limited probably because
it is just an emerging phenomenon. Especially, more need to be known on
the micro mechanism of residential segregation in the Chinese city. Third,
empirical studies on the relationship between gated community and
residential segregation have so far been focusing on that between the
absolute levels of activities and gating without acomparative perspective.I
believe the latter is very important in analyzing the segregation effects of
gated community.
Residential segregation in China
Gated community is defined by its external characteristics such as walls,
gates, private guards and access control system. In our questionnaire
survey almost all (97.5 %) homeowners support installing access control
system in their MDs; 98.7 % of homeowners think it can improve security;
75.2 % believe it can increase privacy inside theMD;only41.7 %think
positively of its role in separating the MD from the rest of the city. Even
this last number is rather high considering the political sensitivity of the
question. Is this the manifestation of residential segregation in the
Chinese city? What is special about residential segregation in China? How
should we analyze the relationship between gated community and
residential segregation in urban China?
The most common reason for residential segregation is peoples prejudice
and discrimination. For example, racial discrimination is an important
reason for residential segregation in the US (Massey and Denton 1993). In
the studies on gated community, many researchers (see, for example,
Low 2003) point out that the fear of the outside world is fundamental to
its formation. This sense of fear is essentially peoples prejudice towards
those of different races or cultures. Another reason is related to the forces
described in Tiebout Model (1956). In the Tieboutian world, consumers
vote on their feet for local public goods provided in different jurisdictions,
6

matching their preferences with the bundles of local public goods, leading
to efficient inter-jurisdiction competition. In equilibrium, the sorting of
consumers will result in a spatial pattern of homogeneous communities.
Many thus regard the suburbs in American cities as a good example of
Tiebout Model. Of course, the formation of homogeneous communities is
the basis for residential segregation. Among all the factors that determine
peoples preferences for local public goods, income is obviously one of the
most important. This factor applies in probably all countries.
What is special about residential segregation in Chinese cities, in addition
to income difference? We first need to examine prejudices and
discriminations that exist in the Chinese society, among which the most
widely studied and best known is the prejudice and discrimination against
peasants. Some regard it as a fundamental reason for gated community in
the Chinese city (Pow 2007). Combined with peoples normal concern
about security, this factor can easily lead to peoples fear of the outside
world that is full of migrants from the countryside, ultimately generating
demand for gated community. It is also distinct from post-communist
counties in Eastern Europe, where urbanization and industrialization have
largely been completed. Another factor that is commonly cited in the
English literature is racial discrimination. Since race is not an important
issue in most Chinese cities, this factor is not a fundamental reason for
residential segregation.
In order to understand residential segregation that is emerging in a
market economy, we have to examine the housing system before the
economic reform started in 1979. Many authors (Stoyanov and Frantz
2006; Polanska 2010), albeit in Eastern Europe, have pointed out the
relationship between gated community and the housing system in
planned economy. The same applies in China. In the era of socialist
planned economy, housing was mostly provided by the work unit as part
of the welfare package for the employees and urban landscape was
dominated by work unit compounds. The distribution of housing within the
work unit was based on non-pecuniary criteria such as rank, title and age
(Wang and Murie 1999). Because different work units had different power
or capacity of obtaining fund or permit for building housing, inter-work
unit differences in housing condition were often bigger than that within a
work unit.3 Workers salary and welfare all vary greatly across work units.
It is fair to say that, compared to people outside the work unit compound,
residents within it is very homogeneous in spite of differences in rank and
title. After the major housing reform started in late 1990s, many
commodity housing units have been sold in the market and the majority of
them are in the form of gated communities. Homeowners in those gated
communities come from different backgrounds and have different
7

incomes. They are very likely to be more heterogeneous than those in


work unit compounds in the era of planned economy.4 Therefore, it is very
important to take into account historical background when studying
residential segregation in urban China. When compared to work unit
compounds, where many people lived before they moved into gated
communities, the latter may actually result in higher heterogeneity among
the homeowners.
Another important factor to residential segregation is culture, which
affects how people interact with each other. Chinese culture emphasizes
family and clan (Hsu 1963). Many people probably have more contact with
their relatives than their neighbors, let alone people outside the housing
community. It is generally agreed that they trust more in their relatives
than in other people (Fukuyama 1995). In other words, people are more
likely to spend time with their relatives no matter they live in gated
community or open community. In this sense, the impact of gated
community on peoples behavior may be partially offset by the culture.
Gated community per se might have limited effect on some of
homeowners contact and external activities.
In summary, residential segregation in the Chinese city is based on
income and Tieboutian forces, which are reinforced by gated communities
that tend to increase the homogeneity of homeowners and reduce their
contact with the rest of the city. These are generally the same as in the
West. However, two other factors unique to China also need to be
considered diperhatikan. One is the historical background of work unit
housing and the other is the Chinese culture. Due to the former people
may feel that their neighbors become more heterogeneous after moving
into the gated community. The latter will let people rely more on their
relatives and have less contact with their neighbors or other people no
matter where they live.
Summarizing the above analysis, Fig. 1 shows a conceptual framework for
analyzing the impact of gated community on residential segregation.
There are four indicators for residential segregation: income homogeneity,
contact with people, external activities (such as shopping and dining and
so on) and participation in local public affairs. Each indicator is affected by
gated community as well as some other factors. For example, contact
with people is affected by gated community, Chinese culture and work
unit. Gated community will tend to decrease peoples contact with those
outside the MD; in Chinese culture people are likely to spend a lot of time
with their relatives wherever they live; work-unit housing is likely to
increase peoples contact with those within the same work unit and
decrease their contact with those outside the work unit. Income
8

homogeneity in a community is affected by gated community, work unit,


Tieboutian competition and economic growth. In other words, gated
community is likely to increase income homogeneity; the experience of
work-unit housing is likely to let people feel that income differences
among neighbors increase; Tieboutian forces are likely to increase a
communitys income homogeneity; lastly, economic growth is likely to
widen income gap and result in higher income heterogeneity. The top of
Fig. 1 are those underlying reasons or factors for the four indicators of
residential segregation.
Empirical analysis
I selected three MDs, by name K, J and Y, in Nanan District in Chongqing,
China for the questionnaire survey, which was conducted from late June to
early July, 2012. These three MDs are very close to each other, belonging
to the same street office, and thus impacts from different geographic
locations or different street offices can be ignored. Besides, they have
different governance structures representing three possible combinations
of HOA and PMC (Property Management Company): MD K has an HOA but
no PMC; MD Y has a PMC but no HOA; MD J has both an HOA and a PMC.
MD Y is a middle-upper class community, MD J is a middle class
community and MD K is a low-income community. In this sense, our survey
sample is representative of residents in gated communities in urban
China, which accommodate almost all social classes. We employed a
systematic sampling method by interviewing one out of every four
households in each MD.5 The total number of housing units in the three
MDs is 2433. Total sample size is 320, within which 77 is from MD K, 117
from MD J, and 126 from MD Y. The response rate is thus about 52.6 %. For
the purpose of this section, I delete those observations who had already
lived in another MD before moving into the current MD because, in theory,
their behavior would not change much. Hence, my final sample size is
197.6
Table 1 provides brief demographic profiles of the three MDs. MD Y has
1577 households, much larger than MD J and MD K. Homeowners in MD Y
and J are better educated while only 1.3 % of people in MD K have college
or higher degree. One-fifth and one-third of people in MD Y and J,
respectively, are communist party members, a symbol of social status in
China, and that number in MD K is only 1.3 %. Car ownership in MD Y and J
reaches 42 and 33 %, respectively, while that in MD K is only 5.2 %. More
people in MD Y and J work in the government or SOEs (State-Owned
Enterprises) than in MD K. There are also more senior managers or
executives in MD Y and J than in MD K. Most important of all, average
monthly income in MD Y and J is twice that in MD K. In a word, MD Y and J
9

are middle or middle-upper income communities and MD K is a lowincome community.


The first aspect I examine is about the homeowners living condition,
which is important background information for my analysis. The average
inside floor area (tao nei mian ji) for each household increases from 83.4
to 92.4 square meters after moving into the MDs. Before moving into the
MDs, most homeowners (69.5 %) lived in housing provided by the work
units, 16.2 % in commodity housing, 13.3 % in rental housing, and few in
housing managed by the Housing Bureau. Besides, most of them regard
themselves as belonging to middle or lower class in the past. 48.8 % think
they belonged to middle class according to total household income, 30.2
% belonged to middle-lower class, 11.6 % to lower class, 8.7 % to middleupper class, and only less than 1 % to upper class. Table 2 shows the
changes in housing type, suggesting that most peoples housing condition
has improved. For example, many moved from bungalow to multi-story
building or from multi-story building to tower building. We conducted
Stuart-Maxwell Test on the change of the distribution. Stuart-Maxwell Test
checks if a square table is symmetric or, in our case, if the distributions
among multiple categories in the past and present are the same. The v2
value in Stuart-Maxwell Test is 85.6 and the P value equals 0.0000. In
other words, the improvement in housing condition is significant at 1 %
level.
Second, I investigate income differences among the neighbors and their
interaction with each other as well as with other people. Table 3 shows the
distribution of changes in peoples perception about income differences
among the neighbors. It is clear from the table that diagonal values are
the largest, indicating that many peoples perceptions havent changed at
all. However, values in the lower triangle are larger than the
corresponding ones in the upper triangle, suggesting that more people
think that income differences among the neighbors have increased. The
v2 value in Stuart-Maxwell Test is 14.9 and the P value is 0.0006,
indicating that the change is significant at 1 % level. One reason may be
that most people lived in work unit compounds, which featured high
homogeneity among the neighbors, before moving into gated
communities, which no longer group people by their work units. Hence,
neighborhood heterogeneity increases and income gap widens in peoples
perception. Another reason may be that economic growth in China has
been widening income gap wherever people live. Even if they lived in
another MD before moving into the current one, their perception of
income differences among the neighbors will still increase. These
arguments show history matters in understanding gated community,
especially in transitional economies (Polanska 2010). The same is also
10

reflected in how familiar people are with their neighbors. We can see in
Table 4 that more people agree that they are less familiar with their
current neighbors than in the past. The v2 value is 73.48 (with P value
equal to 0.0000), indicating that the change is significant at 1 % level. In
this case, an important reason may be that it takes time for people to get
familiar with their neighbors. The longer they stay in a community, the
more familiar they are with the neighbors.
Questionnaire also includes a question on how homeowners interact with
other people in their spare time. There are three categories, namely
frequent contact (1), occasional contact (2) and no contact (3). The higher
is the categorical value, the less is the frequency of contact with other
people. v2 values and P values of Stuart-Maxwell Tests are provided in
Table 5. It is clear from Table 5 that, first, with regard to the homeowners
contact with their relatives, the frequency of contact with the relatives
decreases after moving into the gated community. This change is
significant at 5 % level. Second, their contact with colleagues also
decreases, with the change significant at 10 % level. Third, their contact
with friends (outside the neighborhood) decreases. This change is
significant at 1 % level. Fourth, there is also a decline in the frequency of
contact with pepole inside the neighborhood, which is significant at 1 %
level. Fifth, their contact with people in the vicinity of the neighborhood
(about 20 min walking distance) decreases. This change is significant at 1
% level. Sixth, the mean value for contact with people in the further areas
also increases, indicating a decline in the frequency of contact. This
change is significant at 10 % level. All these evidence points to one
consistent pattern: homeowners contact with almost all types of people
decreases after moving into the gated community. This finding is
consistent with Roitmans (2013) qualitative study in Argentina, which
supports the segregation effect of gated community; it is contradictory to
Asiedu and Arkus (2009) finding in Ghana as well as Feng et al. (2011) in
Guangzhou, both of which suffer from methodological problem that only
examines the interaction among the residents in absolute rather than
relative values.
The third aspect is about the locations of homeowners external activities.
In particular, the survey asked four questions about where they do
shopping, have dinner with their families, visit friends and participate in
recreational activities. Those questions are intentionally designed to be in
the same format as in Feng et al. (2011) so that comparison can be made
between their findings and ours. The spatial location falls into one of the
three categories: within the MD, in the vicinity of the neighborhood (within
20 min walking distance), and further areas (beyond 20 min walking
distance). Categories for the frequency of the external activities include
11

seldom or never (1), sometimes (2), and often (3). Findings include, first,
that the possibility of shopping within the community doesnt change.
Table 6 shows the changes of where homeowners do shopping and the
statistics from StuartMaxwell Test. The P value for the change of the
distribution of shopping within MD is 0.862, suggesting that the change is
insignificant. So is the change of the distribution of shopping in the further
areas, with the P value equal to 0.278. In contrast, the frequency of
shopping in the vicinity of MD rises significantly.
Second, with regard to the location of dinner, the frequency for dining
within MD drops and that for dining in the vicinity of MD increases. Table 7
shows that both of those changes are significant at 1 % level. The change
of the frequency of dining in the further areas is not significant even at 10
% level.
Third, as shown in Table 8, the frequencies of visiting friends within MD
and in the vicinity of MD both decrease, significant at 1 % level. However,
the frequency of visiting friends in further areas increases, significant at 5
% level.
Forth, Table 9 shows the changes of where homeowners have recreational
activities and the statistics from Stuart-Maxwell Test. The changes for
having recreational activities within MD and in the vicinity of MD are
insignificant. Only the increase of the frequency of having recreational
activities in the further areas is significant at 5 % level.
By comparing the situations before and after moving into the gated
community, it is clear that within the community the frequencies for
shopping and recreational activities dont change much and those for
dining and visiting friends decrease. In the vicinity of the neighborhood,
the frequencies for shopping and dining increase, that for recreational
activities doesnt change much, and that for visiting friends decreases. In
the further areas, the frequency for shopping remains the same and those
for dining, visiting friends and recreational activities increase. Obviously,
the findings reported in Feng et al. (2011) and Asiedu and Arku (2009) no
longer hold once we examine the data from a comparative perspective. In
general, I can conclude that homeowners external activities within the
community decline to some extent, those in the further areas increase,
and those in the middle (i.e., in the vicinity of the neighborhood) dont
show a consistent pattern across different types of activities.
There are two possible reasons for the above findings. First, automobile
has become an important transportation mode in Chinese cities and car
ownership has been rising quickly. It is then not surprising that
homeowners external activities increase in the further areas given the
12

convenience of automobile. Second, MD is designed to be for residential


land use and commercial land use is not encouraged inside an MD. That
might be why homeowners dine less within MD. This is very different from
traditional work-unit housing that might lack integrated design and
residential and commercial land uses might be close to each other. In the
middle ground, i.e. the vicinity of MD, the above two reasons interact,
resulting in an inconsistent pattern across the four types of external
activities.
The final aspect I am interested in is homeowners participation in local
public affairs, which is addressed by three questions in the questionnaire:
Do you participate in the election of representatives of the Peoples
Congress? Do you participate in the activities organized by the
neighborhood committee? Have you ever been a volunteer? There are five
choices to each question: always (1), often (2), occasional (3), seldom (4),
and never (5). Table 10 shows the statistics from Stuart-Maxwell Test. The
responses to the first two questions dont change much before and after
moving into the gated communities. Both changes are insignificant. But,
homeowners appear to be more willing to be volunteers after moving into
MD. Albeit significant at 1 % level, this is a perplexing result.
In summary, the survey reveals that peoples living condition has
improved by moving into the gated communities. Many homeowners think
that the income differences among the neighbors have increased and
their familiarity with the neighbors has decreased. Although their contact
with other people is still dominated by that with their relatives and
colleagues, it declines for almost all types of people. Homeowners
external activities within the community somewhat decline; those in the
vicinity of the neighborhood dont show a consistent pattern of change;
the activities in the further areas increase slightly.
The case of MD Y
The analysis in the preceding section is based on the survey on the three
MDs. Although it can help to draw more general conclusions by pooling
the data from the three MDs, it might be affected by some peculiarity of a
particular MD. For example, MD K is a resettlement community in which
most residents are former peasants who lost their land in the process of
urban development. Consequently, they might exhibit behavior different
from those in ordinary MD. Therefore, I include herein a brief analysis of
the survey data from a particular MD, i.e., MD Y, as a case study that can
help us to obtain a closer picture of the gated community.
MD Y is a typical gated community for the middle-upper class, with private
guards, electric wired walls and security cameras. Most buildings inside
13

MD Y are eight-story high, Mediterranean-style buildings except five highrise tower buildings. Landscaping is beautiful with many trees and
gardens. The developer built many public facilities including swimming
pool and tennis court. There is no HOA in MD Y; a PMC is in charge of the
provision of local public goods. In a word, MD Y is a good example of gated
community that might impact residential segregation. The sample size of
the data for MD Y is 57.
Through analysis of the survey data on MD Y, I find that most conclusions
in the preceding section hold except those on homeowners participation
in local public affairs. Table 11 shows the statistics from Stuart-Maxwell
Test. It is easy to see that the changes with regard to election and
volunteers are both insignificant. Participation in the activities organized
by the neighborhood committee decreases, the change of the distributions
of which is significant at 10 % level. It is clear that homeowners
participation in neighborhood activities declines after moving into the
gated community while the likelihood of participation in the election and
that of being volunteers dont change much. I think this result is more
accurate about the impact of gated community because it is based on the
data from a typical MD and is not influenced by low-income community.
Low-income communities are more likely to participate in local public
affairs because they often rely on the neighborhood committee for
financial subsidies.
Then, how can we interpret these empirical findings? First, as a matter of
fact, most people lived in work unit housing before they move into gated
communities. Residents in work unit housing are very homogeneous in the
sense that they all work in the same work unit and their housing
conditions, salaries and welfares are all closely related. Two processes are
at work when they move into gated communities. The first is the change
from high homogeneity in the work unit to heterogeneity in commodity
housing. The second process is the impact of gated community on
residential segregation, which in theory tends to form relatively
homogeneous communities. With regard to income differences among the
neighbors, the first process is likely to increase it while the second process
is likely to decrease it. A third factor is that economic growth in China has
been widening income gaps. The net effect, as reported in our survey, is
increased income differences among the neighbors. Second, the finding
that homeowners contact with all types of people declines after moving
into gated communities is consistent with theories on residential
segregation. It indicates at least that the second process dominates
homeowners contact with other people. This is the strongest result from
my survey that is consistent with most theories (Blakely and Snyder 1999;
Atkinson and Flint 2004) as well as empirical studies (Roitman 2013;
14

Vesselinov 2008). It also casts doubt on some empirical studies (Asiedu


and Arku 2009; Feng et al. 2011) that may have problems in
methodological design. Third, some changes of homeowners external
activities are consistent with theoretical predictions of residential
segregation while some others contradict them. An important reason why
some of homeowners external activities increase in the further areas may
be that, with rising car ownership, people are more likely to travel to
further areas. This is very different from Feng et al. (2011). Given the
methodological problems in their research as discussed in the literature
review, I believe my finding is more accurate. Fourth, the case of MD Y
shows that the impact of gated community on peoples participation in
activities organized by the neighborhood committee is significant.
Because the election of the Peoples Congress representatives is not of
much real significance in Chinas political system, it is not surprising that
gated community doesnt have much impact on participation in the
election. The same can be said about being volunteers, which is tightly
controlled by the Party and often becomes an official activity. This kind of
political apathy is not surprising in the context of Chinas political system
and is not even contradictory to peoples enthusiasm for HOA that are
reported in some studies (Read 2008). In a word, my finding contradicts
Gordons (2003) finding that private community does not crowd out
peoples participation in local elections in California. This pattern of
changes of peoples participation in local public affairs supports the
argument that gated community aggravates residential segregation.
Reference group and duration
It is suggested in the previous sections that the change from work unit
housing to commodity housing (in an MD) might help explain many of my
findings. It is then interesting to compare this group of respondents
(target group) with another group who had lived in an MD before moving
into the current MD (reference group). In other words, treating the latter
group as a reference group can help us to better explore the reasons
behind my findings. If the reference groups behavior doesnt change
much after moving into an MD and the target groups behavior does
change, then we can say that the behavioral change is caused by the
difference between work unit housing and commodity housing; if both
groups behavior change after moving into an MD, then we can say that
some other factors cause the changes.
I select from the survey dataset those observations in which the
respondents had lived in an MD before moving into the current MD. The
size of this subsample is 123. Since we are interested in the behavioral
change of this reference group, the most important statistic is the v2
15

value and corresponding P value from Stuart-Maxwell Test, all of which are
included in Table 12. There are in total 23 tests on 23 different variables.
The Stuart-Maxwell test results are varied and can be categorized into the
following types:
P values are greater than 0.1 and, consequently, the changes of the
reference group are insignificant. 14 variables fall into this category,
especially those variables that stand for contact with all kinds of people. In
contrast, the number of insignificant variables for the target group is only
7. In particular, the target groups contacts with all kinds of people change
significantly while those of the reference group change insignificantly. The
only exception among the contact variables for the reference group is
contact with colleagues, which changes significantly after moving into
the current MD. But, its P value increases when compared with that for the
target group. All variables in this category support the theory that
homeowners behavior changes significantly due to the impact of gated
community.
The P value for the reference group is significant while that for the target
group is insignificant. In other words, the behavioral changes of the
reference group and target group are not consistent with the segregation
theories of gated community. There are 2 variables falling into this
category, including dining in further areas and shopping within MD. I
suspect this is due to the rise in car ownership and the design of gated
community.
The behavioral changes of the reference group and the target group are
consistent, but the P value increases for the reference group. In other
words, the change of the distribution of the variable becomes less
significant for the reference group than for the target group. This category
is still consistent with the theory that the difference between the
reference group and the target group helps to explain the behavioral
change although some other factors may also play important roles here. 6
variables are included in this category. Important ones include perception
of income differences among neighbors and familiarity with neighbors.
In the last type P values are both significant for the reference group and
the target group, but it is lower (or more significant) for the former than
for the latter. The only variable in this category is visiting friends in
further areas. I suspect this might be related to the rise of car ownership
in China. People who had lived in an MD are often richer than those who
had not and, hence, they are more likely to have a car.7 It is easier for
them to visit friends in further areas.

16

Summarizing across the above four types of behavioral changes, it is safe


to say that the difference between work-unit housing and gated
community (i.e. MD) accounts for the changes of the target group. This is
especially true for the respondents contact with all kinds of people. It is
my strongest finding for the segregation effect of gated community.
Besides, the reason why homeowners perception of income differences
among the neighbors increases for both the reference group and the
target group might be that income gaps have been increasing over time
with the economic growth. That is another factor above the change
fromwork unit housing to gated community. In a similar way, the reason
why homeowners familiarity with neighbors decreases for both the
reference group and the target group may be related to the duration of
stay in the destination, a factor that works against the segregation effect
of gated community.
Since the longer people stay in a place the more familiar they become
with their neighbors, it might be argued that the duration of stay in the
destination is an important factor to their behavioral change. Therefore,
we conduct analysis on the relationship between duration and
homeowners behavior. Table 13 lists the correlation coefficients between
duration and 23 behavioral variables. Most of them are very low, except
for familiarity with neighbors (0.29), participation in election (-0.30)
and participation in activities organized by neighborhood committee (0.34). It is evident that familiarity with neighbors increases over time.
That is the main reason why the changes of this variable are significant for
both the reference group and the target group and it eclipses the
segregation effect of gated community. Participation in election and
participation in activities organized by neighborhood committee are
second order effects of duration. Our analysis indicates that both could
not dominate the segregation effects of gated community. In summary,
duration is not an important factor to most aspects of behavioral change.
An important reason for this conclusion is that close to 85 % of
homeowners in our sample have stayed in the destination for more than 2
years, which may be long enough to diminish the effect of duration.

Conclusion
This paper empirically analyzes the relationship between gated
community and residential segregation in Chongqing, China. My
17

methodology is based on a retrospective survey that asked about


homeowners lives before and after they moved into the gated
communities. In this way, I can analyze the impact of gated community on
homeowners behavior. My major findings include, first, some evidence
supports that gated community aggravates residential segregation. For
example, both homeowners contact with all types of people and their
participation in local public affairs decrease.8 This result holds even when
we compare with the reference group or consider the duration of stay in
the destination. Put another way, the differences between work-unit
housing and gated community can best explain those behavioral changes.
Second, the changes of homeowners external activities dont show a
consistent pattern. This is different from Feng et al. (2011). Increasing car
ownership allows people to travel to further areas for some external
activities and the design of gated community reduces commercial land
use inside the MD. Third, homeowners participation in activities organized
by the neighborhood committee declines after moving into the gated
communities, implying a negative impact on participation in local public
affairs.
There are some questions remaining open for future research. First, this
paper relies on the survey in three gated communities. In the future more
communities need to be included in the survey so that more advanced
statistical methods can be employed for empirical analysis. Second, a
difficult problem is how to separate the effects of gated community from
other factors. I tried to use reference group for this purpose. More effort
should be devoted to this issue in the future research. Third, I consider
four types of activities, namely shopping, dining, visiting friends and
recreational activities, when studying homeowners external activities. It is
worthwhile to study in the future what types of activities can best
measure the effects of residential segregation. More generally, we need to
know what are the best indicators for residential segregation in the
Chinese city.

18

S-ar putea să vă placă și