PLEASE SCROLL DOWN FOR ARTICLE

Transcription

1 This article was downloaded by:[wu, Weiping] On: 18 January 2008 Access Details: [subscription number ] Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: Registered office: Mortimer House, Mortimer Street, London W1T 3JH, UK The Professional Geographer Publication details, including instructions for authors and subscription information: Migrant Settlement and Spatial Distribution in Metropolitan Shanghai Weiping Wu a a Virginia Commonwealth University, Online Publication Date: 01 February 2008 To cite this Article: Wu, Weiping (2008) 'Migrant Settlement and Spatial Distribution in Metropolitan Shanghai ', The Professional Geographer, 60:1, To link to this article: DOI: / URL: PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

2 Migrant Settlement and Spatial Distribution in Metropolitan Shanghai* Weiping Wu Virginia Commonwealth University Given the persistence of China s migration trends since the early 1980s, migrants have begun to assert their influence on cities spatial structure. This article attempts to understand the geography of migrant residence and how it relates to the overall spatial development in metropolitan Shanghai. It explores the key geographical factors underlying migrant spatial distribution. Results are based on spatial analyses and a regression model at the subdistrict level, with data drawn primarily from the 2000 Population Census, 1996 Basic Establishment Census, and a migrant housing survey completed in The article also shows how intra-urban migrant settlement and mobility patterns in China might be distinctive from those in other developing countries, given China s unique context and institutional factors. In general, migrant distribution in metropolitan Shanghai displayed a strong degree of centrality until the late 1990s when the inner suburb became the main receptor for new arrivals. The geography of migrant residence has shifted in tandem with deconcentration of the local population and, to a lesser degree, industrial relocation. Areas with a large number of manufacturing enterprises but a smaller state sector are likely to see a high share of migrants in total population. New arrivals also are attracted to areas already concentrated with migrants. Housing availability, however, proves to be an insignificant predictor. Key Words: location behavior, migrant settlement, Shanghai, spatial distribution, urban China. Dada la persistencia de las tendencias migratorias de China desde principios de la década de los 80, los migrantes han comenzado a ejercer su influencia en la estructura espacial de las ciudades. En este artículo se intenta entender la geografía de la residencia de los migrantes y su relación con el desarrollo espacial general del área metropolitana de Shanghai. En él se exploran los factores geográficos claves que determinan la distribución espacial de los migrantes. Los resultados de basan en análisis espaciales y un modelo de regresión a nivel de subdistrito, con datos obtenidos principalmente del Censo de Población de 2000, el Censo de Establecimiento Básico de 1996 y la encuesta sobre vivienda de los migrantes que se concluyó en El artículo también muestra cómo los patrones de movilidad y asentamiento intraurbano de los migrantes en China pueden ser característicos, en comparación con los de otros países en desarrollo, dados los factores institucionales y de contexto específicos de China. En general, la distribución de migrantes en el área metropolitana de Shanghai mostró un fuerte grado de centralidad hasta finales de la década de los 90, cuando las áreas intersuburbanas se convirtieron en el principal receptor de los recién llegados. Support from the National Science Foundation (BCS ) and the United States Department of Education (P019A80016) for the field work phase of this research is deeply appreciated, as are constructive comments from multiple reviewers. The Professional Geographer, 60(1) 2008, pages C Copyright 2008 by Association of American Geographers. Initial submission, March 2006; revised submissions, December 2006, May and June 2007; final acceptance, June Published by Taylor & Francis, LLC.

3 102 Volume 60, Number 1, February 2008 La geografía de residencia de los migrantes ha cambiado conjuntamente con la desconcentración de la población local, y en menos grado con la reubicación industrial. Las áreas con un mayor número de empresas de manufactura pero con un sector estatal más pequeño probablemente verán una mayor proporción de migrantes en su población total. Los recién llegados también son atraídos hacia áreas ya concentradas con migrantes. Sin embargo, la disponibilidad de vivienda ha mostrado ser un predictor insignificante. Palabras clave: conducta locacional, asentamiento de migrantes, Shanghai, distribución espacial, China urbana. D espite China s impressive record of economic growth since the start of the reform in 1979, income disparities between urban and rural areas, as well as regional imbalances, remain large (Fan 1997; Wei 2000). Such disparities are a key driving force behind the largest tide of internal migration in the country s history. By official estimates, about 120 million rural migrants are living in urban areas ( China: Migrants, Slaves 2007). Shanghai alone has received in excess of three million migrants. The majority of migrants move to seek employment, and family migration is on the rise. The sheer magnitude of the migration is bringing significant challenges to cities and their spatial development. The persistence of China s migration trends calls for a better understanding of the spatial dynamics of migrant settlement in destination cities. This article builds on existing theories of migrant spatial distribution and analyzes patterns in Shanghai, China s largest metropolis. It addresses three research questions: (1) How is migrant residence distributed geographically within the metropolitan area? In particular, where do newly arrived migrants find housing and how have patterns evolved over time? (2) What are the main predictors of the residential distribution of migrants? (3) How is migrant spatial distribution related to patterns of local population and employment activities? The focus of the article is temporary migrants (also known as floating population) without official changes of household registration (hukou), who make up the bulk of China s internal migration. Local population refers to registered permanent residents and permanent migrants with formal changes of hukou. Adding to the complexity of understanding migrant settlement in urban China are the inevitable changes brought by the transition from a command to a market economy. In particular, urban housing and land systems have operated with a significant level of fluidity. Under reforms, housing is no longer a free public good to urban residents and commercial housing distributed through market mechanisms is increasingly the main choice in cities. Housing often is developed with private capital, giving rise to a variety of residential spaces that are replacing the prereform cellular neighborhood structure built around work units. Urban land, on the other hand, remains owned by the state and managed by municipal governments through a land leasing system. For migrants, squatting and illegal subdivisions are generally not a viable option, unlike in many developing countries, largely due to municipal authorities intolerance and tight reign over public land. A key objective of this article is to understand how migrant distribution patterns in urban China might be distinctive from trends elsewhere given the unique institutional environment. This large context, in no small way, has shaped the housing choices, mobility patterns, and living arrangements of migrants in China. Overall, they experience a housing disadvantage that is rooted in restrictions associated with the hukou system (W. Wu 2004). They can be characterized as seasonal or temporary migrants, and tend to invest little financially and socially in cities. Thus, low cost and proximity to the workplace are higher priorities in making residential decisions than tenure and amenities. Most migrants never make the transition from renters to owners, even after a lengthy residence in the cities. I also expect to show that their spatial mobility is limited in distance and intertwined with employment opportunities. Results of this article are based on spatial analyses and a regression model at the subdistrict level, which is the best resolution of available information for Shanghai (and Chinese cities in general). Data are drawn from the 2000 Population Census and 1996 Basic Establishment Census, supplemented by my own survey of 1,789 migrants completed in Because of limited data availability at the subdistrict level, I have little choice but to use

4 Migrant Settlement and Spatial Distribution in Metropolitan Shanghai 103 population and employment data from different years. The main body of the article is organized into five sections. The first is a synthesis of research on the spatial distribution of migrants in Third World cities. This also provides a conceptual framework for the subsequent analyses and places China s experience in a comparative context. The next section introduces the data sources and assesses their reliability and limitations. The third section offers a discussion of spatial development in Shanghai (particularly in terms of the local population and employment activities), as a pretext to understanding migrant residential patterns. Next, I analyze migrant housing behavior in Shanghai and metropolitan-wide geographic distribution of migrant residence. The last section is an analysis of the key geographic factors underlying the spatial distribution of migrants, primarily based on a multivariate regression model. The interpretation of the model is supplemented by the survey results and field observations conducted in select neighborhoods around Shanghai. Understanding Migrant Settlement and Residential Distribution Theoretical constructs of spatial distribution patterns of internal migrants in Third World cities have originated primarily in the context of Latin American countries. As the major migratory tides swept through these countries from the 1950s to the 1980s, most research was done around that time. Subsequent massive rural urban migration in other regions, particularly Africa and South and Southeast Asia, also has attracted the attention of researchers to study intra-urban migrant settlement patterns. These studies have evolved in a setting in which private land ownership, and housing and rental markets are functional. China s unprecedented waves of internal migration commenced in the early 1980s after the central government had relaxed population mobility control (primarily through the hukou system). Largely from rural to urban, much of the migratory flow involves circular movements of rural labor in search of work to augment agricultural income. Migrants continue to be regarded as temporary by authorities, although longer term residence is increasing (about 52 percent of surveyed migrants in Shanghai have lived in the city for five years or longer). This is in some ways similar to circular migration in African countries, where many migrants live in cities for a long time. Research shows that such semipermanent residence means that intra-urban residential movement is increasingly similar to that typical in many Latin American cities (Gilbert and Crankshaw 1999). One influential view in the literature regarding the initial location of migrant settlement is Turner s (1968) model, based on his work in Lima, Peru. It suggests a two-stage settlement process for rural urban migrants in urbanizing countries. New migrants (labeled bridgeheaders) initially seek deteriorating rental shelter, primarily in the central city but sometimes scattered across town for good access to jobs. As their income level improves, they move to build peripheral informal shanties for residential stability or ownership and then upgrade shanty dwellings over time into more substantial houses. Once this transition from rented rooms to self-help housing is made, migrants become consolidators. However, in a number of countries with continuing urbanization, inner-city areas are no longer found to be the major destination for new migrants. The expansion and redevelopment of the commercial core has led to a rapid rise in land costs. To avoid higher rents within the city, migrants are attracted to settlements in intermediate or peripheral zones, and recent arrivals tend to locate more on the outskirts (United Nations Center for Human Settlements [UNCHS] 1982; Conway 1985; Gilbert and Varley 1990; van Lindert 1991). The trend of increasing concentration of new migrants in the periphery has been observed in, for instance, Jakarta, Indonesia (Rustiadi and Panuju 2006); Karachi, Pakistan (Ahmad 1992); La Paz, Bolivia (van Lindert 1991); and Lagos, Nigeria (Afoloyan 1982 cited in Ahmad 1992). In Johannesburg, South Africa, the central city has never been the major reception area because it has been predominantly commercial and industrial from the beginning. Instead, the new townships are the main destinations for migrants (Gilbert and Crankshaw 1999). In all of these cities, the periphery location is characterized by squatter settlements or self-help housing. Such housing also serves the economic function of reducing or eliminating

5 104 Volume 60, Number 1, February 2008 housing costs for migrants (Collier 1976; Ulack 1978). For some of these and other cities, the key recipient area for new migrants also shifts as urban expansion proceeds from early to later phases. In many cities, such expansion tends to be in an outward, concentric pattern. The case of Jakarta illustrates how the shifting of migrant spatial distribution has occurred. Inmigration flows reached their peaks in the period between 1982 and Central Jakarta lost its attractiveness as migration destination earlier than other districts of the city. Most of the districts have steadily lost their attractiveness from about the beginning of the 1990s. Since then, the destination of new migrants has shifted to the suburb of Jakarta City (Rustiadi and Panuju 2006). Similarly in La Paz, the oldest part of the city was the first residence of early migrants, but recent arrivals tended to locate more outside of the central city. The intermediate zone was the largest reception area during rapid urban expansion around the 1960s. Many migrants now end up in the far periphery of the city s high plateau (van Lindert 1991). In Mexico City, the inner city was the major reception area for migrants during its early phase of urbanization (the 1940s and 1950s). Subsequently new squatter settlements began to appear outside of the inner city. By the 1960s and 1970s, the periphery was experiencing the most rapid proliferation of migrant settlement (Conway 1985). Even in Lima, on which Turner s (1968) model is based, from the 1980s onward, a growing number of new migrants settled directly in the periphery of the city, whereas central city neighborhoods are increasingly populated by second-generation migrants (Chambers 2005). In some cities, however, spatial distribution of new migrants might be less concentrated in a specific zone and instead be more dispersed. For instance, in Monterrey, Mexico, on arrival migrants scatter throughout the city (Vaughan and Feindt 1973). There is also an increasing level of heterogeneity in how major migrant receptors shift geographically with respect to the three concentric zones (central city, intermediate zone, and periphery). Research in Quito, Ecuador, shows that migrants secure affordable and available housing, be it in a central or peripheral location (Klak and Holtzclaw 1993). A more recent study of Mexico City suggests that migrant mobility, once in the city, is limited in terms of distance. Rather than moving from rental housing in the central city to peripheral settlement as suggested by Turner, most of them relocate relatively near their last places of residence (Ward 1990). This might be attributed to the sheer size of Mexico City, whereby migrants relate to only one sector of the city. Spatial distribution of migrant residence is often determined by a common set of factors. Proximity to existing or potential employment is a major factor underlying choices of site for the first residence, as well as subsequent relocation. It ranks high on the list of preferences and needs of migrants for at least two reasons. First, income generation and economic viability are a primary objective for migrants. Second, most migrants work long hours at what are almost inevitably physically exhausting jobs. Particularly for new arrivals with few acquaintances in the city, an initial residence within walking distance of jobs is essential (Conway 1985; Gilbert and Varley 1990). Others point out the importance of kinship and friendship ties, acting as social institutions (Abu-Lughod 1961; Collier 1976; UNCHS 1982; Banerjee 1983; van Lindert 1991). Migrants first place of residence in the city is largely predetermined by the location of kin or friends. New arrivals to the city might stay with members of their social networks, or rely on the information controlled by the networks to find a place to stay. The social networks that sustain migration flows also lead to spatial concentration of migrants, often in the form of satellite or daughter communities of migrants from a single village. These theoretical constructs of intra-urban distribution patterns of internal migrants will no doubt help our understanding of likely trends in China. However, Chinese migrants encounter significantly more barriers in cities. There are restrictions preventing or obstructing their settlement in destinations, ranging from labor market discrimination to the hukou system (Roberts 1997; Solinger 1999). These barriers make China s rural urban migrants more like immigrants from developing to developed counties than internal migrants within developing countries (Roberts 1997). Solinger (1999) even argues that two kinds of citizenship, urban and rural, created during the era of state socialism now form the basis of the broadest kind of social inequality in China. Whereas urban citizenship comes with provision of

6 Migrant Settlement and Spatial Distribution in Metropolitan Shanghai 105 social welfare, rural citizenship essentially entails self-responsibility in food supply, housing, employment and income. With deepening marketization, however, citizenship rights no longer neatly correlate with the rural urban dichotomy (L. Zhang 2001). There are increasing variations among migrants. A small number of successful ones have gained urban status under recent hukou reforms, particularly in smaller cities. Some have gained access to limited benefits (e.g., dormitory housing) by signing employment contracts with urban enterprises. On the other hand, some with capital and skills have resorted to self-employment and paradoxically have been living on the fringe, often literally, in the cities. The majority of migrants continue to be denied full urban benefits. Any study of migrant settlement patterns also requires some understanding of how existing urban residential areas are distributed geographically by socioeconomic status (Vaughan and Feindt 1973). Accounts of such patterns need to be adequately linked to the changing spatial patterns of the destination city and, in particular, location of employment opportunities. The formation and development of migrant settlements often reflect changes in the economic activity of the surrounding areas (Conway 1985). Market forces are increasingly the dominant force behind urban processes in China, particularly in housing and land development. There is evidence that the importance of location, which was irrelevant in socialist cities without land markets, has led to the emergence of a land rent gradient similar to that of cities in capitalist systems (Ma 2003). Some cities now show three rings of differentiated urban space: pre-1949 historic areas, a socialist planned work units ring ( ), and the new estates ring (built during the property boom years since 1985; Y. Wang and Murie 2000). Accelerated urban growth has led to increasing concentration of economic functions on the outskirts of the city proper, in the form of high-tech development zones and office and industrial parks. In addition, since the 1980s, comprehensive development or large residential development projects have replaced sector-based, project-specific development. By allowing work units to retreat from direct land development, comprehensive development reduces the traditional tie between workplace and residence in the urban space. New residential communities in turn have become much larger and are often located in the peripheral areas (Yeh and Wu 1996). As a result of such fundamental changes in urban processes, Chinese cities are experiencing socio-spatial development often seen in many other Third World cities. Housing commodification and socioeconomic differentiation brought by the planned-to-market transition is leading to the resurgence and continuation of the pre-1949 spatial division in Shanghai, detailed in a later section (F. Wu 2002b; Ma 2003). Beijing, for instance, has seen the formation of a small number of wealthy housing areas in the suburbs. Aggravated living conditions of the central city and real estate development propaganda encourage the urban affluent to move to the eastern and northern inner suburbs of the city (Hu and Kaplan 2001). Ironically these same areas also attract the largest number of migrants, although some migrants concentrate in the old central city (Gu and Shen 2003). Data Sources and Reliability The empirical analysis in this article is based on data from multiple sources, both primary and secondary. The first source is the 2000 Population Census, which is residence based. The National Bureau of Statistics estimated an undercount rate of 1.81 percent nationally in the 2000 Census, largely because of the increasing population mobility (Chan 2003). Shanghai added a new migrant form with more questions to the national standard forms, defining migrants as those without Shanghai hukou who had stayed for more than one day in the city (Shanghai Fifth Population Census Office [SF- PCO] 2002). Available data on migrants at the subdistrict level, however, include only those who had been in Shanghai for more than six months. This article hence uses this set of census data for migrants. The second source is the 1996 Basic Establishment Census. Unlike the population census, the establishment census is workplace based and surveys the local economic structure, basic economic activities, labor force, and distribution of production factors. It would be ideal if the population and employment data were from the same year. The establishment cen-

7 106 Volume 60, Number 1, February 2008 sus was conducted twice in Shanghai, in 1996 and Published information on the 2001 Establishment Census, unfortunately, is aggregated at the district level. As a result, the 1996 Establishment Census data allow for a more refined resolution that is needed in this analysis. My own migrant housing survey, the third source, was conducted between December 1998 and March 1999 in twenty-two neighborhoods of Shanghai s seventeen districts (out of a total of twenty at the time) and eleven enterprises or institutions, with 1,789 complete questionnaires from migrants. Multistage stratified cluster sampling procedures were used to select migrants in residential neighborhoods. First, a number of districts were selected in three stratified geographical zones: central city, inner suburb, and outer suburb (based on official designations). Within each selected district in the central city and outer suburb, one neighborhood was randomly picked. Because of the large size of and migrant concentration in the inner suburb, two neighborhoods were picked in each selected district there. This resulted in a total of twenty-two neighborhoods as the areal sampling units. Second, the total number of migrants in each of these neighborhoods was drawn from the 1997 Floating Population Survey, jointly conducted by the municipal security bureau and Shanghai Academy of Social Sciences. Population-proportionate-tosize procedures were then used to determine the exact number of migrants to be surveyed in each neighborhood. Selected migrants also needed to meet four criteria: aged fifteen or older, with hukou outside of Shanghai, had migrated for employment reasons, and had stayed in the city for longer than a month. Although these survey procedures did not yield a random sample, a great deal of attention was paid to increasing the quality of the data as much as possible. Geographical stratification reduced potential spatial skewedness and helped generate samples with comparable distribution patterns as in the 1997 official enumeration of migrants. To capture migrants living in informal housing and work-related compounds (e.g., construction sites and dormitories), eleven additional enterprises or institutions were selected to represent housing arrangements outside of residential neighborhoods. The definition of migrants in the survey differs from that in the 2000 Census, but this should have limited impact on the reliability of the results in this article, as data from the two sources are largely used to investigate different aspects of migrant settlement. The census data are the foundation for the analysis of local population and migrant residential distribution. The survey data, on the other hand, show migrant housing and mobility choices. Because the sample design affects the number of migrants selected from different parts of the city, the distribution of their present residence should be interpreted with limited confidence (in any case, this is mostly done to confirm trends drawn from census data). However, migrant mobility patterns are more dynamic and less subject to the effect of sampling procedures. Changing Spatial Patterns in Shanghai With a population of million (including 3.87 million migrants; see SFPCO 2002) and land area of 6,377 km 2, the Shanghai metropolitan area is governed by the Shanghai Municipal Government, equivalent to a provincial government because of Shanghai s special administrative status. Nineteen district units (before the year 2000 there were twenty), eighteen with urban designation (district, or qu) and one rural (county, or xian), are located in three geographic zones: central city, inner suburb, and outer suburb (see Figure 1). These units are then divided into subdistricts (jiedao) for urban areas or towns (zhen) for rural areas (referred to in this article as subdistricts). Between 50,000 and 100,000 residents live in each subdistrict, whose functions in the population census resemble those of census tracts in the United States. There are government agencies administering migrants at all three levels: municipality, district, and subdistrict. Each subdistrict is further divided into a number of neighborhood or village committees (referred to here as neighborhoods), which are the smallest residential administrative units. Shanghai has a history of residential differentiation, dating back to the pre-1949 period. Urban space was differentiated into upper and lower ends in the central city. Semicolonial rule (1840s 1949) delineated where the upper ends were, in the French Concession and

8 Migrant Settlement and Spatial Distribution in Metropolitan Shanghai 107 Figure 1 Metropolitan Shanghai. Source: Based on Yeh (1996). EDTZ = economic and technology development; EPZ = export processing zone. International Settlement, both located in and near the city core (see Figure 2). Shantytowns were located along the boundaries of foreign settlements and in areas designated for Chinese residents. At the time these slums were all peripheral to the city core, unlike those in twentieth century urban America (Lu 1995; F. Wu 2002a). After more than thirty years of socialist development, there were still signs of such shantytowns in several districts located immediately outside of the central downtown. However, residential differentiation markedly declined after many years of building public housing and accelerated efforts to redevelop shanty areas (W. Wu 2005). Prior to 1949, a mixed pattern of land use and industrial fragmentation characterized the central city (Hodder 1996; W. Wu 1999). The

9 108 Volume 60, Number 1, February 2008 Figure 2 Central Shanghai. Sources: Based on Gamble (2003); S. Wang and Zhang (2005). lack of a proper planning framework, when foreign settlements and Chinese districts were separately administrated, led to a situation in which the locations of factories and houses were rather random and often encircled each other. This problem was further aggravated by Mao s policy on industrial self-sufficiency and the system of administrative allocation of land before 1979 without the agency of rent or price (Hodder 1996). Even in 1990, the central city housed nearly four times more industrial (manufacturing) establishments than the inner suburb (Ning and Yan 1995). Industrial fragmentation also extends to the metropolitan level more recently, aggravated by the haphazard location of township and village enterprises across suburban districts. About 27 percent of the land is currently under industrial use, a level much higher than the average for other large Chinese cities (about 15 percent). This might be attributable to the fact that nearly 44 percent of Shanghai s industrial land use is scattered and not in concentrated forms such as industrial parks (Xiong and Luo 2000). More or less following an inverse concentric pattern, Shanghai s central city has long been the residential core, with some of the highest population densities in the world (in the range of 50,000 60,000 people per square kilometer in some neighborhoods). Deconcentration began in recent decades, albeit with uneven results. Redevelopment within and new housing construction outside the central city are two important mechanisms. Under market reforms, previously residential central areas are increasingly under pressure from redevelopment, largely for commercial and office uses. This has had the effect of pushing up housing prices in the core. Redevelopment through real estate, however, is selective. Some residential areas with extreme dilapidation and high density have been left out of recent redevelopment because of high costs associated with resettlement (F. Wu 2002a). Outside of the central city, there has been substantial housing construction in new suburban areas and satellite towns. For instance, the development of the Pudong New Area

10 Migrant Settlement and Spatial Distribution in Metropolitan Shanghai 109 accommodates close to a million residents relocated from the central city (W. Wu 1999). Many areas in the inner suburb, in particular, are experiencing rapid transition from ruraltourbanuses(oftenreferredtoasrural urban transitional areas or chengxiang jiehebu). There is an increasing juxtaposition of rural villages, resettlement housing for centralcity residents, migrant communities, and new commercial housing projects (F. Wu 2002b). The satellite town program, launched after the 1950s primarily for the purpose of industrial development, has begun to attract more population since the 1980s (accommodating about two-thirds of a million residents in 1990). Five of the seven satellite towns are in the inner suburb, including one in Baoshan, two in Jiading, and two in Minhang, whereas the sixth in Songjiang and seventh in Jinshan are in the outer suburb (see Figure 1). The functions of other small towns in the suburbs have diversified and their population size also has increased, primarily as the result of recent industrialization (see Ning and Yan 1995). Shanghai s residential patterns have changed steadily during the reform era, with the combined effect of central city redevelopment, new housing construction, and the satellite town program. Using limited subdistrict data, I have arrayed the spatial distribution of local population in 1997 and 2000 according to seven distance bands, expressed in terms of distance from the city center or the People s Square (see Table 1). About two-thirds of local residents lived within a radius of 20 km and close to half within 10 km in the year However, between 1997 and 2000, the innermost band lost nearly 5 percent of its local population. The most drastic change occurred in the band between 10 and 20 km a sharp rise of 45 percent. More specifically, the core of the central city, including Huangpu, Nanshi, Luwan, Jingan, and Hongkou districts, lost a significant amount of local residents (in the range of percent since the 1980s), but the periphery of the central city has seen some gains, particularly in Xuhui District. Population growth in the four districts of the inner suburb has been substantial, in the range of 20 to 25 percent, whereas the outer suburban population level has remained stable (Gao and Jiang 2002). As a result of such deconcentration, inner suburban subdistricts immediately outside of the central city are accommodating a large local population at a fairly high density, even though the central city remains the residential core (see Figure 3). To solve problems associated with fragmented industrial land use, Shanghai has relied on relocating factories in the central city to new districts. A number of industrial parks (often called economic and technology development zones, or ETDZs) have been created (mostly in the inner suburb), including the Jinqiao Export Processing Zone (EPZ) and Zhangjiang High-Tech Park in Pudong, Minhang ETDZ, Hongqiao ETDZ, and Caohejing High-Tech Park (see Figure 1). This process of industrial relocation, albeit slow and with mixed results, has freed up a significant amount of space in the central city and led to an industrial concentration in the inner suburb (or the km band; see Table 2). Between the central city and inner suburb, the split was close to even in 1996, compared to the situation in 1990 when the ratio was closer to 4:1 in the number of industrial establishments. By 1996, only about 24 percent of the city s total industrial establishments were still located in the central city. Industrial Table 1 Spatial distribution of Shanghai s population, 1997 and Change ( ) 2000 Distance category (km) Locals % Locals % Locals % Migrants % ,480, ,177, , ,045, ,927, ,793, , ,245, ,350, ,390, , , ,350, ,308, , , , , , , , , , , , , , , Total 12,723, ,273, , ,134, Source: SFPCO (2002).

11 110 Volume 60, Number 1, February 2008 Figure 3 Spatial distribution of population, Source: SFPCO (2002). deconcentration might have contributed, to some extent, to the spread of local residents to the inner suburb as population increases in specific districts where major industrial development is in progress (Walcott and Pannell 2006). There is now an increasing concentration of economic functions on the outskirts of the city proper. In addition to industrial relocation, at the heart of Shanghai s new development is a set of service nodes. Each of them, including Nanjing Road (Huangpu District), Huaihai Road (Luwan District), Xujiahui (Xuhui District), Yuyuan (the former Nanshi District), the Passenger Rail Station (Tianmu Road, Zhabei District), and North Sichuan Road (Hongkou

12 Migrant Settlement and Spatial Distribution in Metropolitan Shanghai 111 Table 2 Spatial distribution of industrial and service establishments, 1996 Total State-owned Industrial Service Distance category (km) Number % Number % Number % Number % , , , , , , , , , , , , , , , , , , , , , , , , Total 252, , , , Source: SBECO (1997). Note: The category of industrial establishments is equivalent to the manufacturing sector in the West. District), has become increasingly commercial in character, with shops and offices displacing residential and industrial space (see Figure 2). In addition to these municipal retail nodes, many districts have designated their own (Gaubatz 1999; S. Wang and Zhang 2005). The building of Pudong s Lujiazui central business district on the east bank of the Huangpu River adds to the city s service alignments. With these multiple new centers, Shanghai s spatial pattern might begin to resemble the initial stage of a polynucleated form (Walcott and Pannell 2006). In contrast to the distribution of industrial establishments, commercial and service establishments tend to be more concentrated in the central city. More than 60 percent of them are located within a radius of 10 km from the People s Square (see Table 2). The Geography of Migrant Residence Migrant Access to Urban Housing Despite recent efforts aimed at disassociating hukou with the provision of social welfare, a local hukou continues to be an important qualification for accessing several types of urban housing. This is particularly true for more affordable housing options. Migrants cannot acquire ownership of municipal and work unit public housing because only sitting tenants (local urban residents) can do so. New public programs of affordable housing, either for sale or rental, also target only local urban residents. Commercial housing, the only real property sector without ownership restrictions, is not affordable for most migrants. A small number of migrants (fewer than one hundred) contacted for further interviews after the housing survey reported that the housing price they would consider affordable averaged around 160,000 yuan. This is lower than what it would cost on average to buy a new commercial housing unit, about 250,000 yuan in Shanghai in 1999 (W. Wu 2004). Hence, the vast majority of new migrants rent or share accommodations on arrival. More than 60 percent of migrants in the survey are renters. Frequently migrants rent private housing in suburban areas that used to be or still are designated as agricultural within the metropolis. Because of urban expansion, many local rural residents have lost their farmland to development. However, by virtue of their rural hukou status, they have been allocated ample land to build private living quarters and tend to have much larger housing space than urban residents living in the more developed urban core. Many also expand their homes or build additional structures on their land to rent out to migrants (W. Wu 2005). Consequently, the share of migrants renting private housing is much higher in the inner suburb than in the central city (see Table 3). In more central locations, migrants can rent the so-called public housing from urban residents who have purchased ownership rights to their housing (see Table 3). On the other hand, when migrants find jobs in state and collective enterprises, they obtain the access to institutionally provided dormitory housing, a legacy of the welfare housing system. Close to 30 percent of migrants in Shanghai share such dormitory housing (see Table 3). By and large, overcrowding and lack of facilities is a feature of migrant housing. Migrants also tend to live in dwellings that are

13 112 Volume 60, Number 1, February 2008 Table 3 Migrant housing and employment distribution (based on survey results) Central city Inner suburb Outer suburb Total N % N % N % N % Housing type Renting private housing Renting public housing Dorm/workshed Staying with locals Commercial housing Other a Occupation Managers/administrators Professionals/technicians Office workers Service/sales workers Craft/building workers Factory workers Elementary occupations b Ownership type of employer State-owned Collective Private Self-employed Foreign-invested a Other housing includes self-built shed, boat, hotel or inn, living on the street or in a hallway, and staying in hospital rooms. b Elementary occupations include sales and services workers (street vendors, cleaning workers, domestic helpers, building caretakers, porter, garbage collectors, etc.), agricultural laborers, and laborers in mining, construction, or transport. Based on the International Standard Classification of Occupations (ISCO-88). temporary or makeshift structures that are used for working or storage purposes in addition to serving as residences (W. Wu 2002). It is no exaggeration to say that once in the city, migrants continue to move. With substantially higher mobility rates than local residents, they experience much more residential instability (W. Wu 2006). Close to 60 percent of surveyed migrants have made at least one move. However, such mobility is not necessarily driven by the need for tenure or even amenity. Few migrants make the transition from bridgeheaders to consolidators after years of living in the city, a trend in migrant settlement seen elsewhere in developing countries. My survey shows that only about 1 percent of migrants in Shanghai are homeowners (see Table 3). Instead, most remain trapped in the private rental sector or in dormitory housing. Home ownership has yet to become an attainable goal and, therefore, the security offered by housing tenure is less relevant as a motivation for migrants in making housing decisions. China s migrants, however, do share some behaviors with their counterparts elsewhere in that they all tend to invest little income to improve housing conditions. The critical factor often lies in the intention of migrants and their commitment to the city. Many see migration as a seasonal pursuit to augment agricultural income, and continue to maintain close ties to their places of origin. Such a temporary mentality has important implications for housing behavior. Short-term residence, for example, is likely to encourage renting over ownership. Another main explanation would lie with institutional barriers migrants encounter in the city. Specifically, the system of granting only temporary urban residence permits to migrants forces them (even those with families in tow) into more of a bridgeheader existence than they might otherwise prefer (W. Wu 2006). Although some small cities and towns recently have experienced a weakening of the hukou system, it continues to be enforced in the largest Chinese cities. The hukou system is one socialist institution that is slow to change in the course of the transition to a market economy. It remains a way for the state to exclude migrants from acquiring full access to urban citizenship and its attendant social rights (Solinger 1999). Therefore, there is a general disadvantage experienced by all migrants in cities, although those

14 Migrant Settlement and Spatial Distribution in Metropolitan Shanghai 113 few with capital and skills have fared better and gained entry into the league of homeowners (W. Wu 2002). Spatial Patterns of Migrant Residence At the turn of the 1980s when the migrant influx first began, the central, oldest part of Shanghai was the chosen residential location of new arrivals. This resembled a number of Latin American cities (e.g., Lima and La Paz) at an early stage of urbanization and with a concentric spatial structure. In the mid-1980s when Shanghai enumerated the migrant population for the first time, a larger proportion (over 40 percent) lived in the central city than in the inner suburb (W. Wang 1995). The 2000 Population Census shows that migrants remain attracted to more central locations, as more than 70 percent of them still live within a radius of 20 km of the city center (see Table 1). The survey results are even more telling when migrants zone of first settlement is arrayed by each fiveyear entry cohort. Both before 1980 and in the decade after, the majority of new migrants found their first residence in the central city (see Table 4). Such centrality of migrant distribution, before 1990, can be attributed to the city s inverse concentric spatial pattern. Preliminary results from a parallel survey in Beijing around the same time show the absence of such centrality. Across all of the same entry cohorts, more than two-thirds of migrants in Beijing consistently choose their first residence within the confines of the inner suburb. Closely resembling a multinuclear model, much of Beijing s city proper is laid out in a more sprawling urban form. However, the spatial distribution of migrants has experienced a gradual shift in Shanghai, mirroring a trend in a number of cities elsewhere in developing countries undergoing continuing urbanization. With urban expansion and downtown redevelopment, the inner suburb has become a more important receiving area for migrants since the early 1990s (S. Zhang 1998). Central city housing is becoming less attractive to migrants due to commercial redevelopment and in turn the rapid rise in costs. The largest number of migrants (close to 40 percent) now concentrates in the km band (inner suburb), based on the 2000 Census data (see Table 1). This shift is further confirmed by the survey results, which show that migrants arriving after 1995 favor the inner suburb more than the central city (see Table 4). The attraction of the outer suburb remains low, and a substantial number of new migrants still choose to settle in the old core. This shift also coincides with the deconcentration trend seen in Shanghai s local population. As discussed previously, the core of the central city has lost a significant amount of local population while growth in the inner suburb has been substantial. More specifically, the km band experienced a sharp rise in local residents (45 percent) in the late 1990s. To some extent, the emerging pattern of migrant concentration in the inner suburb lags behind the pace with which industrial establishments have been relocated out of the central city. Industries now are increasingly located on the outskirts of the metropolitan area, with more than half of them outside of the km band (see Table 2). It appears that a number of subdistricts immediately flanking the central city boundary are now residential centers for both migrants and, to some extent, the locals (see Figure 3). They are primarily in the districts of Minhang, Baoshan, and Pudong (inner suburb), as well as Putuo, Xuhui, and Yangpu (periphery of the Table 4 Spatial distribution of new migrants by entry cohort (based on survey results) Central city Inner suburb Outer suburb Total Entry cohort N % N % N % N % Before Total ,

15 114 Volume 60, Number 1, February 2008 central city). This ring area can be characterized as the urban periphery or rural urban transitional area (chengxiang jiehebu). Similar to the concept of periurban areas, urban periphery is situated at the edge of the city proper as the intermediary between agricultural use in suburban areas and urban land use in the center. Migrant concentration in the urban periphery is not unique in Shanghai, as other large cities such as Beijing and Wuhan show similar patterns (Gu and Shen 2003; Yang et al. 2003). Another useful indicator is the share of migrants in the total population (see Figure 3). The urban periphery clearly emerges as a ring of migrant concentration. More specifically, all four districts in the inner suburb have migrant shares in the range of 27 to 33 percent (see Table 5). Other districts, on the other hand, register levels less than 20 percent (with one exception). About a dozen subdistricts, primarily in the four inner suburban districts and slightly beyond, are areas where migrants make up large proportions of the total population (more than 45 percent). In some of these subdistricts, migrants even outnumber local residents. Some evidence suggests that where migrants congregate in large numbers, living environments and residential conditions seem to worsen. One of the most common problems is deteriorating sanitary conditions in such areas (S. Zhang 1998). Some dwellings are built illegally, with primitive materials and unstable structures, and are potential safety hazards. A related question is what the main intraurban trajectories of migrants are after their initial settlement in Shanghai. In Table 6, moving directions between zones of first residence and present location are shown for the various entry cohorts. Migrants who have not changed their residence since arrival (about 43 percent) are not counted. The results, surprisingly, do not resemble the outward trajectories observed in many Latin American cities. Instead, most migrants seem to remain fairly stable; more than 75 percent of movers in the survey have stayed around the same general geographic location regardless of duration of stay in the city (Table 6). This suggests that migrants tend to make Table 5 Population, employment, and housing by district (mean values) % migrants in Density of local Ratio of employment Per capita total population population 1999 to local population housing area 2000 Number of District 2000 (per km 2 ) 1996 (m 2 ) subdistricts Central city Huangpu , Nanshi , Luwan , Xuhui , Changning , Jingan , Putuo , Zhabei , Hongkou , Yangpu , Inner suburb Minhang , Baoshan , Jiading , Pudong , Outer suburb Nanhui , Fenxian Songjiang Jinshan 7.2 1, Qingpu , Chongming Central city , Inner suburb , Outer suburb City average , Sources: SBECO (1997) and SFPCO (2002). Note: In 2000, the districts of Huangpu and Nanshi were merged into one district named Huangpu.