Living Conditions of the Floating Population in Urban China

Housing Studies, Vol. 21, No. 5, 719 744, September 2006 Living Conditions of the Floating Population in Urban China LEIWEN JIANG Watson Institute for International Studies, Brown University, Providence, USA (Received May 2005; revised December 2005) ABSTRACT Slums and squatters are not widespread in most Chinese cities and towns even though China has been experiencing rapid urbanization. Moreover, statistics from the 2000 Census reveal that migrants do not necessarily live in poorer housing conditions than non-migrants in urban areas; some housing facilities for the floating population, who moved without changes in household registration, are even better than those of local urban residents. Using 1 per cent of the household sample of the 2000 Census dataset, this paper examines the living conditions and their determinants of the floating population in urban China. KEY WORDS: Living conditions, floating population, urban China Motivation Cities in developing countries are frequently characterized by high rates of in-migration coupled with poverty and a widespread proliferation of slum and squatter areas (Brockerhoff & Brennan, 1998; Costello, 1987; Tangri, 1968). According to the United Nations Population Division (2001), driven by increasing rural urban migration and high fertility, 95 per cent of the world s population growth in the next 30 years will be absorbed by the urban areas of less developed regions, whose population will probably rise from approximately 2 billion in 2000 to just below 3.5 billion in 2030. As the majority of the world population is migrating to urban areas, the locus of global poverty is also moving to the cities, a process now recognized as the urbanization of poverty. The combination of high population density amid poverty and limited resources makes for an environment that favours the rapid growth of slum areas. The UN-HABITAT (2003) predicts that if the current trend persists, the number of slum dwellers worldwide will rise over the next 30 years to approximately 2 billion, which accounts for about 40 per cent of the total of urban residents. To significantly improve the lives of slum dwellers is one of the 18 UN Millennium Development Goals (MDGs). Urban squatters and slum dwellers in less developed countries have been a research focus for many years (Anderson, 1928; Peattie & Aldret-Haas, 1981; Portes, 1972; Ulack, 1976, 1978). Correspondence Address: Jiang Leiwen, Watson Institute for International Studies, Brown University, Box 1970, Providence, RI 02912-1970, USA. Email: Leiwen_Jiang@Brown.EDU ISSN 0267-3037 Print/1466-1810 Online/06/050719 26 q 2006 Taylor & Francis DOI: 10.1080/02673030600807431

720 L. Jiang Some scholars observe that squatter settlements remain the predominant first destination for rural urban migrants (e.g. La Greca, 1977; Richardson, 1977), while others argue that movingintoa slumis arationalchoicefor those migrants(e.g. Portes,1972;Ulack 1976).From the negative perspective, the newcomers to the city are mostly unskilled and untrained, and have difficulty gaining steady employment and income. As a result, many of the migrants congregate in squatter communities, preserving their regional prejudices and customs (Davis, 1975). On the other hand, slums are positively regarded as communities that provide a necessary adjustment for the newcomers to the city milieu (Abu-Lughod, 1961). To the majority of the observers, slum communities are essentially an unavoidable consequence of migration toward the cities in the less developed world. Based on those studies, two assumptions can be made about the relationship between rural urban migration and the expansion of urban slums and squatters in the less developed countries: (1) slums and squatters are the consequence of rapid rural urban migration; the higher the migration rate, the faster expanding the slum and squatter population; (2) since the rural urban migrants are most likely to move into slum and squatter communities, the living conditions of the migrants are substantively worse than those of the native urban residents. These two assumptions have actually been testified by many authors, and become the premises for the study of urbanization and migration in the developing regions. However, there are a few studies that do not support these generalizations. For example, a comparative study of housing quality between the migrants and native residents in Manila shows that the migrants do not disproportionately contribute to nor suffer from the problems of slums associated with urbanization (Hendershot, 1978). Another study in a medium Philippine city, Cagayan de Oro, suggests that migrants are neither heavily settled nor segregated in the slum community (Costello, 1987). However, the findings from the Philippine cities were regarded as special cases due to either the unique characteristics of the migrants or the problem of biased samples. For instance, the housing pattern of the migrants in Cagayan de Oro City was chiefly attributed to its heavy rates of migration by young unmarried females, who mostly resided in non-slum neighbourhoods as servants, lodgers or extended relatives of the native urban residents (Costello, 1987). The conclusions of the Manila study were limited by the problems of the validity of the quantitative measures of housing quality, and the representative of the sample used (Hendershot, 1978). Nevertheless, these case studies reveal that the predominant proposition that rural urban migrants disproportionately contribute to and suffer from the housing difficulties in the slums is questionable under certain conditions. Recent studies on the housing conditions of rural urban migrants in China also give various results. While some authors indicate that housing inequality in urban China is expanding (e.g. Logan et al., 1999), and many migrants are segregated in and suffer from poor living conditions (e.g. Huang, 2003; Ma & Xiang, 1998), others argue that the informal settlements are not a viable option for the migrants since municipal authorities are intolerant of migrant congregation and squatting in China (Wu, 2002; Wu & Wang, 2002). However, all of the above-mentioned studies are based on data from sample surveys in one or several Chinese cities. Therefore, the findings of these studies are limited and cannot draw any broad conclusions for the country as a whole. Moreover, the most recent statistics from the China 2000 Census shows that the housing conditions of migrants vary significantly from region to region (Zhai & Zhang, 2003), and the floating population does not necessarily live in poorer housing conditions than the local non-migrants (Jiang & Pang, 2003).

Living Conditions of the Floating Population in Urban China 721 Therefore, this study is designed to study the housing conditions of rural urban migrants in China and to answer two main questions (1) In contrast to local urban residents, are Chinese rural urban migrants more likely to live in slums? (2) What are the major driving forces affecting the living conditions of those migrants? To answer those questions, China s social and political settings that affect Chinese people in making migration decisions and housing choices should be taken into account. A number of studies show that the housing choice in urban China represents a remarkably different pattern from those in other developing countries, where market force plays a major role (Huang, 2003). The migration decisions and housing choices of Chinese people, to a great extent, are determined by the strong regulations of urbanization policies and the transition state of urban housing reform. Under the unique household registration (or hukou) system, 1 each individual in the Chinese population is classified as having an agricultural (rural) or a non-agricultural (urban) household status. The system embodies a sharp differentiation of rights and privileges, and stringent conditions for converting from rural to urban status (Wu & Treiman, 2003). According to their hukou status, migrants in China can be categorized in permanent or temporary groups. While the permanent migrants in the urban areas holding an urban hukou are officially approved to move due to job assignment, marriage or other family reasons, the rural urban migrants are mostly defined as temporary migrants or floating population, whose changes in residence are officially regarded as temporary even if they are not in reality. While China is still trying to transform the old welfare housing allocation system into a market-oriented housing system in the urban areas, housing choices of the floating population are very limited. Given that most of the floating population in urban China consists of rural urban migrants, this paper mainly focuses on the living conditions of the floating population and their determinants, in contrast to that of the urban permanent migrants and non-migrants. The next section will briefly discuss the urbanization and housing reform of urban China in the past two decades. It is followed by a descriptive analysis of the living conditions of the floating population, compared with local urban residents and permanent migrants. A logistical regression analysis was conducted to study the major determinants of the living conditions of different groups of urban population. A brief conclusion and discussion is included in the final section. Urbanization, Housing Reform and the Floating Population in Urban China Urbanization and the Floating Population In the past two decades, the surge in the floating population in China has been closely associated with the changes of urbanization policies. At the end of the 1970s and the early 1980s, the priority of urbanization was given to the development of rural township and village industries in order to absorb rural surplus labour in situ and avoid the urban problems caused by rapid rural urban migration. In the mid-1980s, as the criteria for urban designation was adjusted in order to absorb rural surplus labour (Zhang & Zhao, 1998), a rapid proliferation of designated cities and towns and urban population was observed, while rural urban migrants increased significantly (Li & Li, 1996). Since the 1990s, urbanization has been accelerated by the urban economic reforms and the development of a market-oriented economy, leading to a flood of migrants into the urban

722 L. Jiang 75 65 55 45 35 25 15 1978 1982 1986 1990 1994 1998 2002 2006 2010 2014 2018 2022 2026 2030 2034 2038 2042 2046 2050 Figure 1. Proportion of urban population in China, 1978 2050. Source: 1978 2002 from Xinhua Data on Line. Available online at http://data.xinhuaonline.com/xhol/xh_pwd.asp; 2002 2050 from Qiu (2003). areas. This increasing spatial movement is often referred to as the floating population boom. The 2000 Census reported that more than 11 per cent of the Chinese population had left their place of household registration for more than six months. 2 Among the total floating population, 71 per cent of whom resided in the urban areas, 41.4 per cent had lived in their current place for at least five years. Among the floating population who had arrived within the past five years, about 50 per cent had moved from rural to urban areas, 31 per cent moved between urban towns, 16 per cent moved between rural areas, and only 3.5 per cent moved from urban to rural areas. In the past two decades, the proportion of urban population increased from 17.5 per cent in 1978, to 36.6 per cent in 2000, and to 39.9 per cent in 2002 (Figure 1). The government plans for this proportion to reach 45 per cent in 2010, 50 per cent in 2020 and about 70 per cent in 2050 (Qiu, 2003). The newly published China Urban Development Report (China Mayor Association, 2004) stresses the importance of mega-cities and suggests promoting the development of agglomerations. It is predicted that, as China accelerates its urbanization process, the influx of the floating population to the urban areas will continuously increase in the coming decades. As the floating population is excluded from the social welfare system designated for the local urban residents, their choices of employment, education, housing, etc., are considerably limited. Although the Chinese government is gradually reforming the hukou system and aims to eventually remove it, resistance against this reform is still strong. Therefore, the hukou system will remain active for substantial period of time and continues to affect migration, housing choices and other decisions of the Chinese. Urban Housing Reform In the early 1980s, China embarked on an ambitious urban housing reform process, through privatization and marketization, to solve the problem of severe housing shortage,

Living Conditions of the Floating Population in Urban China 723 enhance people s standard of living and prompt economic growth. There are already several review papers in English providing a good introduction to this process of urban housing reform (e.g. Tolley, 1991; Tong & Hays, 1996; Wang, 2000, 2001; Zhou & Logan, 1996). However, it is noteworthy that the urban housing reform chronologically involves several specific steps: (1) Encourage the sitting tenants to purchase their occupied public housing at subsidized prices. (2) Allow the suburban agricultural population and urban residents to construct housing in the urban periphery where land is available. (3) Promote commercial housing construction, targeting the high-income households. (4) Support affordable and comfortable housing projects for the needs of low and medium-income households. (5) Provide subsidized municipal public housing for the lowest-income urban households. (6) Lift the restrictions on the previous public housing purchased by the sitting tenant, and foster a regulated secondary housing market. Rapid economic development and housing reform has enabled China to achieve great success in housing construction. The newly completed housing space in urban areas in the period from 1980 2000 grew to more than 4.6 billion m 2. The living space per urban resident increased from less than 3.6 m 2 in 1978 to approximately 10 m 2 in 2000, while the rate of homeownership among urban residents has significantly increased from less than 20 per cent in the early 1980s to approximately 73 per cent (72 per cent for the city, and 78 per cent for the town) in 2000. However, the urban housing system reform did not take into account the welfare of the floating population. Being limited by the hukou system, the floating population theoretically has no access to public housing, affordable and comfortable housing or the land for housing construction. The only housing opportunities for the floating population are to purchase commercial housing in the market or to rent private housing. Many cities encourage the floating population to purchase commercial housing as a means of attracting investment, while mortgages are normally not available to the floating population. Members of the floating population who have bought commercial housing in the cities often obtain a blue stamp hukou status, which provides the holders with partial access to basic social welfare in the cities where they live. However, this policy only favours high-income individuals and is far beyond the reach of most of the floating population. In fact, many of the floating population regard their presence in the cities as temporary, and try to reduce their living costs and save for a return to their home villages in the future. Renting and sharing housing in the suburban rural areas where the rate is relatively low is one of their important housing choices. As a result, some suburban villages have experienced an influx of floating population, forming the so-called migrant enclaves (Ma & Xiang, 1998). Encountering the segregated employment market and limited housing access, the floating population is generally at a disadvantage in housing choices. This situation significantly affects their living condition, which is analysed in the next sections.

724 L. Jiang Data and Measurement Data Data used for this study are mainly the 0.95 per cent sample dataset of the long form of the China 2000 Census. For the first time, the 2000 Census collected information on housing conditions in family households. A random sample of 10 per cent of the total households was selected to fill out the long form, while the other 90 per cent answered the questions in the short form. In addition to the questions on general demographic information, the selected 10 per cent of households were asked to provide detailed information on their housing situations. However, the 2000 Census did not provide housing information for those who lived in collective households, and did not count those who had left their place of household registration for less than half a year as migrants. Therefore, these data make it possible to cover all types of households. As a supplement to the Census data, the dataset from the 1997 Beijing Floating Population Census was adopted, which interviewed all the floating population (of both private and collective households) who had lived in Beijing for at least three days. This provided a reference for the study of the housing of the floating population in collective households and the floating population of family households who had stayed in one place for less than six months. Measurement of Housing Conditions While housing conditions may be defined according to the nature of housing availability, affordability and qualitative aspects of the neighbourhood environment (Lawrence, 1995), choice of the indicators is often context-dependent and varies over time (Wu, 2002). Moreover, use of the housing indicators is often affected by data availability. For example, to measure the prevalence of slums in developing countries, the UN-HABITAT (2003) adopted five indicators in its statistical report: (1) access to improved water; (2) improved sanitation; (3) sufficient living space; (4) improved durable housing; and (5) secure tenure. According to UN-HABITAT, a slum is defined as a group of individuals living under the same roof lacking one or more of the five conditions. However, data for the latter three indicators are not available for most of the countries. 3 A few studies on the housing conditions of China s urban population (e.g. Logan et al., 1999; Wu, 2002) construct a qualitative index of housing quality using several variables, including availability of water, bathrooms, kitchen, fuel, etc. In the current study, four indicators are considered when measuring housing conditions: tenure, crowding, privacy and facilities.. For housing tenure, in addition to the status of owner or renter, the sources of housing are also compared (self-built, public or private housing, affordable and comfortable housing and commercial housing) and the cost of purchasing, constructing or renting.. To measure crowding, per capita rooms and building areas are considered, as well as the housing structure (bungalow, 2 6 storeys, or 7 þ storeys).. The index of privacy is constructed by summing two variables: sharing housing with other households (0 ¼ sharing; 1 ¼ none), and the function of the dwelling (0 ¼ residential as well as working or other purpose; 1 ¼ residential only). The value of the privacy index ranges potentially from 0 to 2.

Living Conditions of the Floating Population in Urban China 725. To construct the facility index, six variables are used: tap water (1 ¼ none; 5 ¼ yes), bathroom (1 ¼ none; 2 ¼ shared other type; 3 ¼ shared flushing; 4 ¼ private other type; 5 ¼ private flushing), kitchen (1 ¼ none; 2.5 ¼ shared; 5 ¼ private), bath or shower (1 ¼ none; 2 ¼ other; 3 ¼ public hot water supply; 4 ¼ private water heater), building materials (1 ¼ other; 2 ¼ grass, bamboo, or wood; 3 ¼ brick or stone; 4 ¼ concrete), cooking fuel (1 ¼ biomass and other; 2 ¼ coal; 4 ¼ electricity or gas). The first three variables represent the basic conditions of the housing, and therefore are given slightly heavier weighting than the latter three. The values of the six variables are summed to construct a comprehensive facility index that ranges potentially from 6 to 27. It was not possible to derive the other important indicator, neighbourhood environment, from the datasets due to lack of information. Therefore, the hypothesis that the floating population spatially congregates in slum areas is not directly tested in this research. However, the physical neighbourhood environment is often closely related to its housing facilities. For testing the assumption of slum incidence, the definition of slum developed by the UN-HABITAT (2003) is adopted and people living without access to tap water or bathrooms are regarded as slum dwellers. Moreover, according to UN-HABITAT, it is necessary to make the distinction between slums and shanties. While slums describe residential buildings that lack essential services or old housing that has deteriorated, shanties refer to spontaneous settlements in outskirts and undeveloped areas of the city. Given that spontaneously occupied land for housing is not tolerated by the Chinese government, shanties are not common in urban China. Comparison of Living Conditions According to hukou and migration status, the Chinese urban population in the 2000 Census can be divided into three groups: local urban residents, permanent migrants and floating population. Floating population is defined as those who have left their place of household registration for more than half a year at the time of interview (1 November 2000); a permanent migrant is an individual who has moved to a place with official change of household registration within the past five years, while a local urban resident refers to those who have stayed at their place of household registration for at least five years. Comparing the housing of the floating population and permanent migrants with that of the local residents, it would help to understand the impact of migration on living conditions; comparing the housing of the floating population with that of the local residents and permanent migrants, it could provide evidence on the impact of hukou status on living condition. Statistical analysis of the 2000 Census dataset shows that of the total urban population, 68.1 per cent were local residents, 7.3 per cent were permanent migrants and 24.6 per cent made up the floating population. Among the floating population, 54 per cent held an agricultural hukou, while the other 46 per cent had a non-agricultural one. About 40 per cent of the floating population had lived in one place for at least five years. Of the floating population arriving within the last five years, 61 per cent originated from rural areas, while 39 per cent moved in from other urban towns. Therefore, economic reform and development of market economy in the past two decades had already generated a large

726 L. Jiang proportion of the floating population in urban China. Moreover, the characteristics of the floating population had become remarkably diversified. Housing Types in Collective Households According to the 2000 Census, of the total urban population 92.3 per cent lived in family households and 7.7 per cent lived in collective households. While the proportion of collective households among urban local residents was very small (0.6 per cent), 28.8 per cent of permanent migrants and 21.3 per cent of the floating population lived in collective households. That permanent migrants had the highest propensity for living in collective households was mainly because many permanent migrants were young college students or new graduates. Ninety-four per cent of permanent migrants living in collective households were 15 to 24 years old. College students and graduates are normally accommodated in dorms by the colleges or work units, and accordingly registered as collective household members. In general, the dorms provided for the permanent migrants are adequately facilitated, although tenants of those dorms usually have to share bathrooms, eat in cafeterias and take showers at public baths. However, the housing conditions of collective households of the floating population were generally not as good as those of permanent migrants. Since there is no housing information on collective households in the 2000 Census, the dataset of the 1997 Beijing Floating Population Census is used to obtain general information on housing conditions for the floating population living in collective households. According to the Beijing 1997 survey, of the 2.3 million of the floating population interviewed, 97.7 per cent had stayed in Beijing for at least a month, 72 per cent had stayed for longer than a half-year, 53 per cent had stayed for more than a year, and 12 per cent had stayed for longer than five years (Figure 2). The differences in the definition of floating population in the Beijing survey and in the 2000 Census should be noted. All migrants who had stayed in Beijing for more than three days were counted in the Beijing Survey, while only migrants who had left their place of household registration for at least six months were regarded as floating population in the 2000 Census. Therefore, only 12 per cent of the floating population had remained in Beijing for more than five years, compared to 38.7 per cent for the whole country reflected in the 2000 Census. Moreover, 57 per cent of the floating population in Beijing stayed in collective households, which also reflects their relatively short duration of stay. Figure 2. Duration of stay in Beijing by floating population, 1997

Housing types Living Conditions of the Floating Population in Urban China 727 Table 1. Housing types for floating population in Beijing, 1997 (%) Household with migrants Migrant-only household Collective household Total Rented peasant housing 2.1 32.9 4.4 14.2 Rented urban resident housing 2.5 32.4 4.9 14.4 Rented working unit housing 3.8 14.5 12.9 12.8 Self-built 10.5 3.5 2.2 3.3 Purchased housing 3.1 1.2 0.1 0.7 Work-unit dormitory 10.5 5.4 41.2 26.4 Work place 0.7 6.5 14.0 10.3 Work plant 0.8 15.7 9.3 Employer s housing 17.1 0.8 1.7 2.6 Relative s housing 49.2 1.5 0.1 4.3 Hotel 0.1 0.3 2.2 1.4 Hospital 0.0 0.4 0.2 Other 0.4 0.1 0.1 0.1 Total 100.0 100.0 100.0 100.0 Among total floating population 7.6 35.0 57.4 100.0 The survey provides information on where the floating population lived. Statistical analysis shows that 26.4 per cent of the floating population lived in work unit dormitories, 28.4 per cent rented private housing, 12.8 per cent rented work unit housing, and about 20 per cent lived at their site of work. Compared with those living in family households, the floating population of collective households was much more likely to live in their work unit dorms (42 per cent) or just on their work sites (about 30 per cent) (see Table 1). Although no further information on the housing conditions of these living arrangements is provided, according to some studies (e.g. Wu, 2002; Jiang & Kuijsten, 2003; Wang, 2000), the housing facilities of work-unit dorms of the floating population is generally poorer than that of the permanent migrants. In particular, the on-site living constitutes the worst conditions. Therefore, it is safe to conclude that the living conditions of the floating population in collective households is usually worse than for those living in family households. The next sections of the study focus on the housing of the urban population living in family households only. Housing Tenure According to the 2000 Census, the homeownership rate in urban China had reached approximately 73 per cent; if those living in collective households are excluded, the homeownership for family households was even as high as 78 per cent (Table 2), which is higher than in most developed countries. Analysis on the source of housing shows that self-construction was the predominant method for urban residents to achieve homeownership (accounting for 41 per cent). Purchased public housing (23 per cent) was next, and then rented public housing (12 per cent). Although the government had been trying to enhance the housing market, purchasing or renting commercial housing was not common (together they account for less than 14 per cent), and purchased affordable and comfortable housing, which could be regarded as a semi-market housing type, accounts for only approximately 6 per cent. It is apparent that at this time housing reform in China was still at an early stage.

Table 2. Housing tenure of urban population in China Floating population Household registration of the householder Local resident Perm. migrant Sum Local resident Perm. migrant Floating pop. Total Source of housing Self-built 49.8 14.4 16.9 35.8 6.2 12.5 41.1 Purchased commercial housing 6.6 19.3 13.1 7.7 24.4 14.1 8.7 Purchased aff. & comf. housing 5.2 11.8 6.6 5.9 11.1 6.6 5.9 Purchased public housing 23.5 31.2 17.4 26.5 28.8 14.8 22.7 Rented public housing 10.6 15.1 17.8 17.2 19.6 18.0 12.3 Rented commercial housing 1.3 3.5 18.7 2.1 4.9 23.5 4.9 Other 3.0 4.7 9.4 4.9 5.0 10.4 4.4 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Cost of purchasing/building (Yuan) Less than 10 000 29.6 9.1 18.6 25.7 5.8 16.4 26.9 10 000 20 000 24.5 14.5 17.3 24.0 11.8 15.0 22.9 20 000 30 000 14.9 14.7 13.4 14.8 14.2 12.9 14.7 30 000 50 000 14.8 22.7 17.4 16.6 22.7 17.7 15.6 50 000 þ 15.9 37.7 32.0 18.5 45.1 36.9 19.4 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Cost of renting (Yuan) Less than 20 32.5 22.0 12.7 30.9 14.2 9.8 23.5 20 50 37.1 29.5 19.5 35.5 28.3 17.6 29.2 50 100 20.9 28.8 25.9 21.3 32.0 26.5 23.5 100 200 6.7 13.8 21.5 7.9 16.4 23.4 13.4 200 500 2.9 6.0 20.4 4.4 9.0 22.7 10.5 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 728 L. Jiang

Living Conditions of the Floating Population in Urban China 729 Compared with local residents and permanent migrants, the floating population had a lower homeownership rate, which was not unexpected. Table 1 indicates that the floating population was more likely to rent housing (36.5 per cent) than permanent residents (19.6 per cent for permanent migrants and 11.9 per cent for local residents). However, the source of housing among the floating population was much more diversified than expected. Theoretically, public housing and affordable and comfortable housing is not accessible to the floating population. Although they had a relatively high proportion of renting or purchasing commercial housing, many of the floating population also rented public housing (17.8 per cent), purchased public housing (17.4 per cent) and affordable and comfortable housing (6.6 per cent). It could be argued that the floating population themselves did not own or rent the public or affordable and comfortable housing, but might have lived with permanent residents who qualified for those types of housing. Therefore, the study further examines the housing of three groups of the floating population according to the hukou status of householders. They are the floating population living in households headed by (1) a local resident; (2) a permanent migrant; and (3) a member of the floating population. Since members of the floating population were very unlikely to head households with permanent residents, the study considers a household headed by a member of the floating population as a pure floating population household. The analyses show that the distribution of sources of housing for the pure floating population households did not differ much from that of the total floating population households (see Table 2). In practice, many of the floating population worked for and were accommodated by state-owned enterprises (SOEs). Some of the SOEs sold public housing to the sitting tenants of the floating population. In some cities or towns, the floating population rented old public housing from the municipality housing bureaus. Although the affordable and comfortable housing projects aimed to help urban permanent residents with low and middle incomes, some developers, driven by economic interests, helped the floating population purchase their properties. Moreover, many local residents moved to newly purchased commercial housing and rented out the old public housing to the floating population, even though renting out public housing by individual households had not yet been officially approved in many places. To purchase or rent housing, the floating population usually had to pay substantively higher prices than the permanent residents. For rented housing, less than one-third of the floating population in 2000 paid low rent (less than 50 yuan) (Table 2), while about twothirds of local residents and more than half of permanent migrants did so. For purchased or constructed housing, the cost was also higher among the floating population than the local residents, while permanent migrants spent the most. Although the cost of renting or purchasing housing is normally related to housing quality, the floating population usually had no access to the subsidies given to permanent residents, and almost always had to pay extra administrative fees. Housing Conditions In general, the analysis shows that although their homeownership rate was relatively higher, the quality of housing for the Chinese urban population in 2000 was still much poorer than that of their counterparts in most developed countries. For example, more than half of Chinese urban households had no shower or bath facility, 40 per cent had no access to clean cooking fuel, about 25 per cent had no bathroom, and

Table 3. Housing conditions of the urban population in China, 2000 Floating population Household registration of the householder Local resident Perm. migrant Sum Local resident Perm. migrant Floating pop. Total Crowding index 1. rooms per capita 0.81 0.80 0.71 0.68 0.68 0.71 0.79 2. building areas per capita (m 2 ) 24.1 25.1 20.0 20 21.4 19.9 23.3 3. building structures (%) bungalow 41.0 15.4 32.4 33.0 12.0 33.0 37.8 less than 7 storeys 50.8 56.9 51.3 56.1 57.3 50.1 51.3 7 þ storeys 8.2 27.1 16.0 10.9 30.7 16.8 10.8 730 L. Jiang Privacy index 1.94 1.95 1.92 1.93 1.94 1.92 1.94 1. housing function (%) residential only 98.7 98.2 93.7 98.3 98.3 92.8 97.6 residential, work, and other 1.3 1.2 6.0 1.7 1.7 7.2 2.3 2. sharing with other household (%) sharing 6.1 5.4 8.0 6.9 6.0 8.4 6.5 no sharing 93.8 94.0 91.7 93.1 94.0 91.5 93.4 Facility index 20.4 23.5 21.0 21.2 23.9 20.8 20.7 1. construction material (%) concrete 26.1 44.3 32.7 29.1 49.3 33.1 28.5 brick or stone 67.8 53.1 62.5 66.4 49.6 62.1 65.9 wood, bamboo, or grass 2.2 0.7 2.0 1.8 0.6 2.1 2.1 other 3.8 1.3 2.5 2.7 0.5 2.5 3.4 2. kitchen (%) private 88.5 91.3 78.0 86.5 92.2 75.6 86.5 shared 1.9 1.6 3.4 2.1 1.4 3.7 2.2 no kitchen 9.6 6.5 18.3 11.3 6.4 20.5 11.2 3. cooking fuel (%)

gas 53.4 78.8 67.8 63.0 84.9 68.7 57.8 electricity 1.7 2.7 2.0 1.7 1.7 2.1 1.8 coal 28.7 13.9 23.8 26.9 12.3 23.4 26.8 biomass 15.9 3.4 4.5 8.0 0.8 3.8 12.9 other 0.3 0.7 1.6 0.3 0.4 1.9 0.6 4. tap water yes 75.7 91.8 87.8 85.4 95.1 88.5 79.1 no 24.3 7.6 11.9 14.6 4.9 11.4 20.8 5. bath or shower (%) collective hot water supply 1.8 4.0 2.0 1.4 3.3 2.1 2.0 private heater 33.2 55.7 36.8 39.8 60.6 35.3 35.2 other 9.8 6.4 8.4 8.1 6.0 8.6 9.3 none 55.2 33.3 52.5 50.6 30.1 53.9 53.4 6. bathroom (%) private flushing latrine 38.5 67.4 44.6 42.9 68.3 44.4 41.4 shared flushing latrine 1.0 1.2 2.1 1.5 1.3 2.3 1.3 private other type latrine 32.7 15.4 17.2 25.3 14.4 15.3 28.6 shared other type latrine 4.0 2.7 6.9 4.6 3.0 7.6 4.5 none 23.8 12.7 28.8 25.6 13.0 30.2 24.2 Proportion of slum dwellers (%) 39.0 12.3 25.5 32.6 15.6 33.6 33.7 Living Conditions of the Floating Population in Urban China 731

732 L. Jiang more than 20 per cent had no access to tap water. Moreover, housing quality varies considerably across different groups of the urban population by hukou and migration status. Living space. Compared with local residents and permanent migrants, the floating population had less living space and fewer rooms per household member (Table 3), regardless of whether they lived by themselves or lived with permanent residents. Moreover, local residents more frequently lived in bungalows or low multiple-storey buildings (less than seven storeys), while a larger proportion of permanent migrants lived in high-rise buildings. This reflects the reality in China that many urban towns have been experiencing rapid and large-scale housing construction and population movement. Compared with older buildings, most of the newly constructed housing comprises multiple-storey buildings with larger living areas for each apartment. Many urban residents had recently left their old neighbourhoods and moved into the new buildings in different districts or towns, and therefore were identified as permanent migrants. Others, who were waiting for new construction projects that might replace their old dwellings, were categorized as local residents. Therefore, permanent migrants more often lived in multiple-storey buildings and enjoyed larger housing space than local residents. The situation for the floating population is more complicated. On the one hand, as newcomers to the urban area, the floating population were more likely to be accommodated in the newly constructed multiple-storey buildings than the local residents; on the other hand, their uncertain hukou status and unstable employment and economic situation might diminish their wish to move into new and large apartments. Moreover, to reduce living costs, the floating population would often seek cheap housing in old neighbourhoods from the tenants who moved out to the new residential communities. Privacy. Living in crowded conditions, the floating population s privacy was also compromised. Compared with local residents and permanent migrants, a larger proportion of the floating population shared housing with other households (Table 3). Some of them lived at their work place, e.g. restaurants or hospital wards. A study in Beijing shows that many in the floating population used their dwellings for business purposes (Ma & Xiang, 1998). ANOVA analysis indicates that the privacy index of housing for the floating population is significantly lower ( p, 0.001) than for local residents and permanent migrants. Housing facilities. Although they were at a disadvantage in terms of living space and privacy index, the floating population s housing facilities were not necessarily poorer than those of permanent residents (Table 3). In fact, some housing facility indicators for the floating population were better than those of local residents, although the housing of permanent migrants almost always had the best facilities. In contrast to local residents, the floating population had better access to tap water, clean cooking fuel, better bath or shower facilities and more durable housing (in terms of housing construction materials), although their kitchen and toilet facilities were poorer. This is true even after considering the hukou status of the householders of the floating population. Measured by the comprehensive housing facility index, the housing facilities were best for permanent migrants and poorest for the local residents, while the housing facilities of the floating population was in between. Moreover, the floating population living in households headed by permanent and local residents enjoyed better housing facilities than those living in households headed by the floating population.

Living Conditions of the Floating Population in Urban China 733 Slum Incidence. Adopting the methodology and measurement of the UN-HABITAT (2003), the indicators of access to tap water and bathrooms 4 are used to estimate the incidence of slum dwellers among different groups of population in urban China. The analysis shows that in 2000, about one-third of the Chinese urban population were slum dwellers, which is close to the UN-HABITAT s estimate (37.8 per cent for 2001). The slum prevalence in the floating population (25.5 per cent) was significantly lower than that among local residents (39.0 per cent), but higher than that among permanent migrants (12.3 per cent). Among the floating population, those living in pure floating population households had a higher slum incidence (33.6 per cent) than those living in households headed by permanent residents (15.3 per cent). However, it was still lower than the slum incidence of the local residents. Based on the discussion above, it may be concluded that in contrast to permanent residents, the floating population lived in more crowded situations where their privacy was compromised. However, the floating population on average possessed better housing facilities and had a lower slum incidence than local residents. Therefore, the floating population did not necessarily disproportionately suffer from severe living conditions in urban China. Given the fact that the floating population had better housing facilities than the local residents, with permanent migrants better off in all of the housing indicators, it may be argued that migration may help people to achieve better living conditions. Moreover, the fact that the living conditions of the floating population were much poorer than those of permanent migrants implies that hukou status significantly affects migrants housing choice and living conditions. It should be noted that in addition to the differences between permanent versus floating hukou status, the household registration system makes another fundamental distinction between agricultural versus non-agricultural (or rural versus urban) hukou status, which should be taken into account in the analysis. Before the introduction of economic reform in the end of the 1970s, non-agricultural (or urban) hukou holders enjoyed about 13 types of social welfare that were not available to agricultural (or rural) hukou households (Jiang & Kuijsten, 2003). Although economic reform had narrowed the income gap between agricultural and non-agricultural residents since the 1980s, a non-agricultural hukou status was still crucial for urban residents to benefit from many types of government subsidies (Li & Yue, 2004). Some local residents living in suburban areas were agricultural hukou holders whose land was encroached upon by recent urban expansion. Although they became a part of the urban population, residents of those areas did not benefit much from the process of urbanization, and their agricultural hukou status was not changed accordingly. As a result, they were not qualified to receive the social welfare benefits designated for the non-agricultural population, and could not improve their housing facilities as much as the non-agricultural urban residents. In fact, many suburban villagers rented out part of their housing to the floating population in order to earn more income. Living in only the remaining space in their houses, their housing conditions actually worsened (Ma & Xiang, 1998). Therefore, it is necessary to make the distinction between agricultural and non-agricultural hukou holders among the local urban residents Similarly, among the floating population, those with agricultural hukou status generally migrated from rural to urban areas and contributed to the growth of the urban population, while those with non-agricultural hukou status usually moved between urban towns. The non-agricultural floating population mainly consisted of three subgroups: (1) the

734 L. Jiang redundant SOE employees who sought job opportunities and conducted business away from home cities/districts; (2) the professionals and skilled workers who held better job positions and earned a higher income in other districts/cities than their places of household registration, and preferred to keep or could not change their household registrations; (3) university graduates dissatisfied with the jobs or the locations they were assigned to, and who preferred to live in the places where they could earn a better living. Therefore, the non-agricultural floating population was very different from city-ward migrants of the agricultural floating population. Considering the two fundamental elements of hukou status (permanent versus floating, non-agricultural versus agricultural), five groups are distinguished within the urban population: non-agricultural local residents, agricultural local residents, permanent migrants, non-agricultural floating population and agricultural floating population. Table 4 shows that the housing conditions between agricultural and non-agricultural hukou holders differ significantly: the living conditions of non-agricultural local residents are much better than those of agricultural local residents; the non-agricultural floating population lived far better than the agricultural floating population. Although they enjoyed more living space, the agricultural local residents had the worst housing facilities and highest proportion of slum dwellers. In contrast to the agricultural local residents, the agricultural floating population had much less living space but significantly better housing facilities. The impact of temporary versus permanent hukou status on housing conditions is not very clear, since the housing conditions of non-agricultural local residents were not as good as those of the non-agricultural floating population, while permanent migrants experienced the best housing conditions. It is noted that a proportion of the floating population with agricultural hukou status, who had moved from one district or sub-district to another within the same city/county but had not reported the change of household registration, the so called Ren Hu Fen Li phenomenon, are not really migrants in usual sense. Moreover, about 17 per cent of the total migrants moved due to demolition of old housing under the recent urban large-scale construction. Some of those households were included as floating population only because they had not yet been registered in the new districts or sub-district, and caused even more problems of Ren Hu Fen Li. The study analysed the impact of the Ren Hu Fen Li phenomenon on the living conditions of urban population by differentiating migrants by the levels of administrative boundaries they had crossed: moving within county/district, moving cross county/district of the same province, and moving from other province. Table 4 clearly shows that the shorter distance the floating population moved, the better the living condition was. Among the permanent migrants, the migration distance did not have much effect on the living conditions, while those moving within the provincial boundaries had relatively better housing. Hence, the Ren Hu Fen Li phenomenon, to an extent, might disturb the true values of housing indicators of urban population by migration status. However, the fact that permanent migrants had the best living conditions and floating population had better living facilities than local residents does not change. It shows that even the floating population moving from other provinces had a higher facility index and much lower slum prevalence than the local residents with agricultural hukou. The migrants moving due to demolition of old housing are also excluded from the analysis and the pattern was found that that floating population had a higher housing facility index and lower proportion of slum than the local residents sustains.

Living Conditions of the Floating Population in Urban China 735 Table 4. Housing conditions of urban population by Hukou and migration status in China, 2000 Per capita room Per capita building area Facility Privacy (m 2 ) % of slum index index %of total pop. Population by Hukou status Local resident with 0.77 22.4 25.5 22.3 1.94 41.7 non-agricultural hukou Permanent migrant 0.80 25.1 12.3 23.5 1.95 7.3 Floating population with 0.78 23.4 18.2 23.0 1.94 10.9 non-agricultural hukou Local resident with 0.87 26.9 60.5 17.5 1.93 26.3 agricultural hukou Floating population 0.62 16.1 30.7 18.7 1.90 12.9 with agricultural hukou Migrants cross different levels of administrative boundary Permanent mig. moving within 0.91 28.4 13.1 23.8 1.94 19.8 county/district Permanent mig. moving within 0.93 28.3 12.1 23.4 1.95 8.7 province Permanent mig. from other 0.93 27.0 15.5 22.3 1.94 2.6 province Floating pop. moving within 0.87 26.1 22.1 21.9 1.92 26.3 county/district Floating pop. moving within 0.81 21.9 25.6 20.3 1.91 21.6 province Floating pop. from other 0.75 17.0 30.2 18.2 1.88 21.0 province Total 0.79 23.3 33.7 20.7 1.94 100.0 Determinants of Housing Conditions From the above discussion, the effects of migration and hukou status on housing conditions were much more complicated than what could be learned from a simple descriptive analysis. To further understand the factors determining the living conditions of the urban population, particularly the living conditions of the floating population, a multivariate analysis was conducted by constructing a logistical regression model. Choosing the indicator of slum dwellers as the dependent variable, three logistical regression models were constructed to analyse the factors influencing the housing conditions of the floating population compared with urban permanent residents. The incidence of slum dwelling was used as the dependent variable. In the first model, the total urban population was considered; the second model considered the total floating population in urban areas; in the third model, only the floating population with agricultural hukou in the urban areas was taken into account. The independent factors included hukou type (agricultural versus non-agricultural), floating status (permanent versus floating), urban type (city versus town), duration of stay, reason for migration, occupation, education, age and household structure. The regression analysis particularly noted the effects of hukou type (agricultural versus non-agricultural) and migration status (floating versus permanent) so that the determinants of housing conditions of the rural urban