Keywords

1 Introduction

China has experienced rapid urbanisation during the past three decades. The influx of rural-to-urban migrants into Chinese cities had reached more than 270 million persons by 2014 (NBSC, 2015). This group, however, is known as the floating population (liudong renkou) because they are not entitled to the same legal rights as permanent urban residents and are treated as ‘internal aliens’ in the cities (Zhang, 2001, p. 23). Regarding their residence, they tend to concentrate at the urban periphery, particularly in urban villages, construction sites, or industrial factory dormitories (Wu, 2008; Wang et al., 2010).

While there is a large body of literature on rural-to-urban migrants in China concerning the effects of the hukou system,Footnote 1 labour market segmentation, migration strategies, and their marginalisation in cities (Chan, 2009; Fan, 2002; Fan et al., 2011; Solinger, 1999), only recently have scholars turned to examine migrants’ housing experiences and social-spatial exclusion in cities (Wu, 2008; Wang et al., 2010; Huang & Tao, 2014). Many studies have focused on the neighbourhood effects of migrants’ enclaves and the implications for their incorporation into the host society (Ma & Xiang, 1998; Zhang, 2001; Liu et al., 2014). However, little has been done to investigate the role of migrants’ residential location in their integration into the cities. The latter issue, however, is becoming increasingly important. To facilitate land development and create more governable spaces, city governments have launched various projects to demolish and redevelop urban villages. However, the massive demolition of urban villages in the inner suburban area might push migrants farther away. Consequently, as recently noted by Wu et al. (2013), migrants in remote suburban neighbourhoods might suffer from not having access to jobs and public services, thus making it even more difficult for them to survive.

In the existing literature, understanding international migrants’ residential location is an important element in understanding their assimilation process, although there have also been critiques of research that focus on neighbourhood effects (Slater, 2013). Among other issues, there has long been widespread concern over the social implication of the suburbs being their residential location. Earlier spatial assimilation models based on international migrants’ experiences in the US considered the group’s relocation to the suburbs as an indication of their structural assimilation (Massey, 1985; Waren, 2013). However, recent evidence has indicated that international migrants’ spatial proximity to the majority group does not necessarily lead to integration, and residents of the suburban ghettoes face more barriers to much needed services than do those in central cities (Logan & Alba, 1993; Murphy, 2007; Murphy & Wallace, 2010).

Drawing on a survey conducted in Shanghai, this chapter aims to examine the place effects of suburban residence on the integration of migrants. While not denying the positive role of suburbs in the supply of low-cost housing, it is argued that peripheral locations may further reinforce migrants’ marginal position in the city in the long run.

The chapter is organised as follows. Based on urban experiences from both developed and developing countries, the next section reviews the literature in the field of migrants and suburbanisation. Then, the residential settlement patterns of internal migrants in urban China are summarised and the explanatory framework is developed. Next, the data collection and analysis methods are introduced. The empirical findings from the Shanghai survey are then discussed. Finally, the main findings are concluded, and policy recommendations are proposed.

2 Literature Review: Migrants and the Suburbs

There has been extensive research on migrants’ settlement patterns in the field of urban studies. Ecologists from the Chicago School claimed that, while newly arrived international migrants usually concentrate in neighbourhoods in the central city, later generations would relocate and disperse to the suburbs (Burgess, 1925, p. 56). These observations were later theorised by Massey (1985) as the spatial assimilation model. There is also empirical evidence that suburbanisation was a key step in the assimilation of international migrants who arrived in the US in the early twentieth century (e.g. Massey & Denton, 1987; Alba et al., 1997; Iceland & Nelson, 2008).

The spatial-assimilation model, however, has been widely challenged since the 1990s. Logan and Alba (1993) proposed an alternative model, known as the place stratification model, to depict the different impacts of suburbanisation on different races. It is argued that because the mechanisms of structural discrimination persist, minority groups might not be able to translate their individual-level characteristics such as income, education, etc., into upward residential mobility. As a result, when they move to the suburbs, they only sort themselves into suburban communities with a low status.

In the US, for example, the barrier to suburban residence was significantly reduced, and many international migrants began to settle in suburban communities immediately upon arriving (Alba et al., 1999). However, instead of dispersing into White native-born communities, international migrants living in the suburbs have continued to concentrate in ethnic neighbourhoods (Logan et al., 2004). Li (1998, 2009) proposed a new model of ethnic settlement, the ‘ethnoburb’, to capture the emerging suburban ethnic clusters of residential areas and business districts. Moreover, recent studies have indicated that minority international migrants are even more spatially segregated in their new suburban destinations (Lichter et al., 2010). Suburban residence does not necessarily lead to improvements in quality of life and could even make the international migrant groups more disadvantaged than those in the central city (Murphy, 2007; Murphy & Wallace, 2010).

The housing experiences of migrants in other countries do not follow a uniform pattern found in the US. In many developing countries, cities are experiencing the process of urbanisation. Rapid urban expansion reveals a different suburban landscape with high population densities and mixed urban and rural activities. While searching for job opportunities in the cities, urban-to-rural and stepwise migrants concentrate at the urban fringe, which often leads to massive squatter settlements (Gilbert & Crankshaw, 1999; Davis, 2006). Turner (1968) suggested that, in fact, there are two stages of settlement for these migrants. New migrants usually live in rented accommodation in central locations near their jobs. However, later, they tend to move to the urban edge and build informal shanties of their own – a relocation process indicating their permanent settlement in the destination city.

Recently, high-profile projects such as gated communities, industrial parks, and shopping malls have begun to emerge in the urban peripheral areas in these countries. However, most are built as private realms and scarcely contribute to improving the overall local economy and to softening social inequality. For instance, by examining the widespread development of gated communities in one suburban municipality in Argentina, Roitman and Phelps (2011) illustrated a typical dualistic process of suburbanisation in contemporary developing countries: poverty and exclusion in the suburbs increase when wealthy people arrive.

Therefore, in both developed and developing countries, migrants living in the suburbs might face more barriers to integration. Then how might peripheral locations influence the incorporation of newcomers into their destination city? There are several potential mechanisms that may give rise to the disadvantages experienced by suburban migrants. The most common explanation rests on the lack of good job opportunities near migrants’ residences. In the US, the decentralisation of low-skill jobs in manufacturing, construction, and services is prominent, and low-wage and often dead-end jobs dominate the labour market of fast-growing suburban nodes (Kneebone, 2009; Raphael & Stoll, 2010). This provides few routes for migrants to become upwardly mobile (Lichter et al., 2010).

The second explanation concerns migrants’ constrained mobility and accessibility on the urban edge. Empirical evidence shows that migrants’ economic opportunities and social participation are reduced due to their lack of access to public facilities and services (Foth et al., 2013; Bose, 2014). Poor suburban communities have far less access to social service resources due to a lack of adequate provision (Allard, 2004). As the private provision of public services is increasingly dominant in the suburbs, this problem can become more severe (McKenzie, 1994, 2005). Moreover, because migrants rely heavily on public transport, an inadequate public transportation infrastructure and a lack of car ownership could impose greater restrictions on them in the automobile-dependent suburbs.

Finally, spatial segregation can be worse in the suburbs. To date, no consensus has been reached on the effects of spatial segregation on migrants’ integration. Some studies have stressed the social and cultural capital migrants could acquire in their neighbourhoods (Portes & Jensen, 1987, 1989). Others, however, have found that persistent segregation delays the long-term assimilation of migrants and hence is a threat to social cohesion (Nee et al., 1994; Xie & Gough, 2011). Empirical studies have shown that limited opportunities for exposure to members of the majority ethnic groups in the neighbourhoods are associated with higher risks of poverty and unemployment (Galster et al., 1999; Feng et al., 2014). Such neighbourhoods’ poor social networks and bad reputation are also important factors leading to negative outcomes for their residents (Musterd & Andersson, 2005; Permentier et al., 2007). These problems can be more apparent in isolated and low-density suburbs as residents will be less frequently exposed to each other compared to those living in densely populated central areas.

3 The Spatial Entrapment of Internal Migrants in Chinese Cities

Internal migrants’ residential spatial patterns in China show more similarity to those found in many developing countries experiencing rapid urbanisation such as Brazil and India. Migrants tend to concentrate in the urban periphery instead of in central areas (Wu, 2008) and now constitute a major source of rapid population growth in the suburbs (Shen & Wu, 2013). Moreover, they are often spatially clustered through their place of origin in certain types of neighbourhoods, creating migrant enclaves on the urban edge (Ma & Xiang, 1998; Zhang, 2001). However, in contrast to their counterparts in self-help housing on the urban edge in many other developing countries (Turner, 1976), suburban migrants in China are mostly living in employer-provided dormitories and private rental housing (Huang & Tao, 2014).

On the supply side, in contrast to middle-class suburbia in North America, the term jiaoqu (literally, suburb) in the Chinese context refers to the periphery of the city proper or built-up area mainly in a geographical sense rather than a distinctive type of residence. Before the early 1980s, the suburbs were underdeveloped and by no means attractive residential locations, being largely dominated by rural areas and a few small-scale industrial satellite towns. After economic reform, Chinese cities immediately experienced rapid suburbanisation due to the establishment of a land market and consequent changes to land use (Zhou & Ma, 2000). However, central areas have remained prosperous and are the preferred place to live and work (Wang & Li, 2004). Due to the lack of public services and facilities, housing prices in the suburbs are relatively low when compared to those in central areas.

More importantly, there are usually large numbers of low-cost private rental houses on the city outskirts. Both formal and informal suburban developments provide low-cost rental housing for migrants. On the one hand, municipal governments expropriate farmland at a very low price and lease the land for real estate development. Such projects are carefully designed and packaged to make the suburbs seem an attractive place to live. These estates remain beyond the budget of most migrants, but a large proportion of properties in the suburbs are bought simply as investment and are then leased out via the rental market. A new type of living form for migrants known as qunzu (literally, co-renting), i.e., an apartment with two to three bedrooms that is divided and leased to several migrant families separately, has become widespread in the suburbs of many large cities.

On the other hand, any market transaction of collectively-owned farmland and villages is forbidden. Former villages have thus been engulfed by rapid urban expansion. Having no right to transfer the land for their own benefit, peasants build or extend their houses and rent them to migrants (Lin et al., 2011; Zhang, 2011). These ‘urban villages’, or chengzhongcun, on urban outskirts have become a major source of low-cost private rental housing for migrants in cities (Wang et al., 2010).

On the demand side, to a large extent, the concentration of migrants on peripheral locations reflects the disadvantage they face in the local housing market, which is rooted in both their socioeconomic status, such as income and occupation, and the unique Chinese hukou-based institutional barrier (Wu, 2004). Most migrants are usually at the bottom of the urban socioeconomic and labour market strata (Solinger, 1999; Fan, 2002). Given that the price of commodity housing has risen exorbitantly in Chinese cities, it is almost impossible for migrants to buy their own homes, let alone pay for high-price estates in central locations. Moreover, migrants without local hukou are denied welfare benefits, such as affordable housing (Tao & Xu, 2007; Xu et al., 2011). As a result, migrants have very limited housing options, living in either employer-provided housing or low-cost private rental housing.

Settling down in the suburbs is also associated with the important role of hometown-based social networks in facilitating migration. More often than not, when migrants first come to the city, they receive social support in finding jobs or accommodation from earlier migrants from the same place of origin. They usually work in the same enterprises and live in the same neighbourhoods as their relatives or friends who arrived earlier. As a result, suburban migrant enclaves such as urban villages are usually their initial residential location (Ma & Xiang, 1998; Zhang, 2001). So far, debates on the social implications of migrant enclaves are inconclusive. While policymakers usually view migrant neighbourhoods as overcrowded, chaotic, and dangerous places, many scholars highlight the positive role of these neighbourhoods in China’s urbanisation. For example, it is in urban villages that migrants can find cheap accommodation and services and adapt themselves to urban lives (Zhang et al., 2003; Liu et al., 2014).

However, a suburban residence does not indicate migrants’ spatial assimilation as found in North American countries or permanent settlements in many developing countries. An earlier survey has shown that proximity to their workplace and the availability of low-cost housing are the primary reasons why migrants chose their current houses in the suburbs (Shen & Wu, 2013). Although hometown-based social resources play a positive role in their adaptation to urban life on their arrival, their life chances may later be significantly restricted and their marginal role may be further reinforced by a constrained life in the suburbs.

First, living in migrant enclaves may limit their exposure to other groups and thus their opportunities to build inter-group social networks. Earlier research has demonstrated that those who have lived in migrant urban villages for a longer time are less likely to construct non-hometown-based or neighbourhood-based social ties (Liu et al., 2012). Second, since economic reform, Chinese cities have experienced the massive decentralisation of manufacturing jobs (Zhou & Ma, 2000). Most of these jobs provide poor pay and prospects. Third, poor mobility and accessibility to quality public services and facilities further make it difficult for suburban migrants to reach opportunities for work, education, healthcare, social activities, and so on. Subsequently, with limited social resources and restricted job-search areas, migrants may have few opportunities for upward social mobility and instead remain trapped in the suburbs. The remainder of the chapter aims to empirically examine the differences between migrants living in central and suburban areas, and how residential location affects the incorporation of migrants.

4 Data and Methods

This research is based on a questionnaire survey distributed in Shanghai. Since economic reform, the city has witnessed a massive influx of migrants. By the end of 2013, the total number of migrants in Shanghai had reached 9.535 million, accounting for approximately 40% of its total population (SSB, 2014). The spatial distribution of migrants in Shanghai has experienced rapid suburbanisation. While in the 1980s, a large proportion of migrants found their residence in central areas, the 1990s saw inner suburban areas (the zone that is 10–20 kms from the city centre) become locations where migrants would concentrate (Wu, 2008). According to the 2010 census data, high concentrations of migrants are now found in outer suburban districts. The location quotient for the percentage of migrant population shows that, on average, Minhang, Fengxian, Jiading, Qingpu, and Songjiang have a higher proportion of migrants than Shanghai, and all but Minhang are outer suburban districts (Fig. 10.1). Among others, Jiading, Qingpu, and Songjiang even have greater populations of migrants than of native citizens. In contrast, the central areas continue to be dominated by native residents, who account for 75% of their total population (Fig. 10.2).

Fig. 10.1
A distribution map of Shanghai plots the location quotient for the percentage of the migrant population ranging from 0.5 to 0.75, 0.75 to 1, 1 to 1.25, and 1.25 to 1.5. The central areas and Chongming have the lowest values, while Qingpu, Songjiang, and Jiading have the highest values.

Concentration of migrants in Shanghai by districts. (Sources: 2020 Shanghai census data)

Full size image
Fig. 10.2
A map of Shanghai and its districts has respective pie charts for migrant and native populations overlaid on it. It indicates that the central areas, Chongming, Baoshan, Pudong, Minhang, and Jinshan are dominated by native popultaion, while Qingpu, Songjiang, and Jiading are dominated by migrants.

The ratio of the migrant and native population by districts in Shanghai. (Sources: 2020 Shanghai census data)

Full size image

Spatial contexts in the central city and the suburbs are different for migrants. While the distribution of migrants is more dispersed in the central city, including traditional lilong neighbourhoodsFootnote 2 – old work-unit neighbourhoods and newly-built commodity housing neighbourhoods – in the suburbs they are much more concentrated in urban villages and factory dormitories. For example, in suburban new towns, about 70% of the migrants resided in neighbourhoods with a proportion of migrants more than 60% (Wang & Yang, 2015). Meantime, residents in the central city enjoy much better accessibility to resources than those in the suburbs. Recent studies have shown that most neighbourhoods with high accessibility to jobs, healthcare facilities, and schools locate within the inner ring of the city (Wu & Sun, 2015; Zhang, 2015; Wu et al., 2012).

The survey followed the principle of random sampling and adopted a multistage clustered sampling method. As Chongming Island remains largely dominated by rural areas and its population accounts for less than 3% of Shanghai’s total population, it was excluded from the survey. First, all sub-districts (including both jiedao and zhen), except those on Chongming Island, were divided into three strata for stratified sampling: the inner city (within the inner ring road), the inner suburb (between the inner and outer ring road), and the outer suburb (outside the outer ring road). The number of sub-districts to be sampled in each stratum was in proportion to the size of the resident population in each area based on the 2010 population census data. In each stratum, sub-districts were sorted according to their per capita income and population density; their de facto population was then adjusted according to their proportion of hukou population. This was to ensure that the structure of the sample was representative in these three terms. Then, the method of probability proportionate to size (PPS) sampling was used to select individual sub-districts. The final sample was drawn from 33 out of a total of 225 sub-districts from 12 districts in Shanghai. In each sub-district, one residential committee was randomly selected. Finally, within a chosen neighbourhood, an equal number of respondent households were selected using a simple random sampling method. They were randomly picked starting from a random house number and chosen at a fixed interval. The three-stage sampling technique ensured that the sample was representative of the inter-zone distribution of population.

The survey was undertaken from 20 July to 10 August 2013. Questionnaires were given to household heads, and most were completed immediately and collected by interviewers straight away. The survey was aided by staff from residential committees who introduced surveyors to the selected households and thus ensured a very high response ratio (95%). In total, 1420 valid questionnaires were collected, among which there were 295 migrant respondents. Notably, in this study, the term ‘migrants’ refers to those who were not born in Shanghai and had migrated to Shanghai from their place of origin. This allows for the examination of the role of hukou status in migrants’ integration. In the sample, this migrant group is somewhat underrepresented. This is partly due to migrants’ high rates of refusal and of noncontact. Nevertheless, because such levels of nonresponse can be regarded as random in this case, errors are tolerable. Table 10.1 compares migrants’ educational and occupational attainments according to the 2020 Census and the survey. In both samples, the distribution of migrants’ educational attainment is concentrated in junior and senior secondary education, while the distribution of their occupational attainment is concentrated in low-end service and industrial work. However, the survey sample has a higher proportion of migrants with technician and staff jobs. This could be because the survey includes not only rural migrants but also migrants with local hukou registration. In addition, the survey does not cover those living in factory dormitories because the lists of dormitories were not available to be included in the sampling frame. Given these two differences, the distribution of migrants’ educational and occupational attainments in the survey sample is reasonable as it reflected the compositional and distributional heterogeneity of the target group.

Table 10.1 Comparison of survey data and 2020 census statistics
Full size table

While migrants’ residential location and social exclusion might be interactive and mutually reinforcing, this study particularly focuses on the subsequent effects of living in the suburbs on migrants’ integration. This is empirically examined by comparing, first, the integration of migrants living in central and suburban locations, and second, their chances of moving into the central city. Regression analysis was employed to test for the independent effects of residential location. In the analysis, place of residence was coded as one simple dichotomous variable indicating whether a respondent lived in the suburbs (including the inner suburbs and the outer suburbs) or the central city (suburbs = 1, central city = 0). Meanwhile, other variables that have been extensively studied in research on migrants’ integration were included to control for the effects of other mechanisms potentially influencing migrants’ integration (Wu, 2012; Wang & Fan, 2012).

First, the study gauges the importance of residential locations for migrants’ integration. As the existing literature conceptualises integration as a multi-dimensional process (Gordon, 1964; Alba & Nee, 1997), two key dimensions of integration are examined here: economic and social. Economic integration is achieved when migrants’ earnings are comparable to those of local natives with similar backgrounds (Chiswick, 1978; Duleep & Regets, 1999). Because income is usually skewed, the log of household heads’ monthly income is conventionally used as a measure of economic integration. In the survey, the head of a household was identified as the primary financial supporter of the family, whose income best reflected its economic integration.

Social integration describes the process whereby migrants gradually adapt to social norms and values and participate in a broad range of social relationships in the host society (Brissette et al., 2000, p. 54). Accordingly, the measure used in this study is inter-group friendship. Migrants who are more integrated are supposed to have friends outside their own group, while those who are less integrated are likely to have friends from the same place of origin as themselves. In the survey, respondents were asked where most of their friends came from; the response categories were the same county or municipality, the same province, other provinces, Shanghai, and diverse places of origin. A dichotomous variable was then constructed to indicate whether a respondent’s social relations were locally embedded (yes = Shanghai natives and diverse place of origin = 1, no = other migrants = 0).

Specific forms of actual regressions differ depending on the nature of the dependent. For economic integration (logged monthly income as the outcome variable), the ordinary least squares (OLS) linear regression was applied. For social integration (inter-group friendship as the outcome variable), the logit regression model was adopted. As for control variables, the level of human capital was measured by whether a respondent held a university degree (yes = 1, no = 0), which is an important channel for migrants to establish their official residence in cities. Higher education is one of the criteria for obtaining local hukou in Shanghai. Occupation was coded based on Chinese Standard Classification of Occupations and collapsed into three categories: blue collar, white collar, and others. Migrants usually have limited access to higher-status white-collar jobs and tend to crowd into peripheral segments in the labour market, which could in turn reinforce their exclusion in both economic and social terms. Length of stay (number of years) may enhance migrants’ exposure and acquaintance with the host society and was thus included to indicate the potential for integration. Hukou location and hukou classification, which are usually recognised as valid measures of migrants’ integration, were used to represent the institutional factor. Finally, sociodemographic variables – i.e., age, gender, and marital status – were also included to control for the effects of individual characteristics.

Second, the study used logistic regression analysis to examine further factors influencing the likelihood of migrants’ spatial assimilation (i.e., moving to the central locations). Only respondents who reported intra-urban residential moves were included in the analysis. Their current residential location was used as the dependent variable (central city = 1; suburbs = 0). Their socioeconomic variables, economic and social integration variables, and previous residential location were all included as independent variables. If the previous residential location had independent effects on migrants’ residential moves, it would indicate migrants might be unable to translate their socioeconomic advancement into locational attainment, and vice versa.

5 Results

5.1 Comparing Migrants in Central and Suburban Locations

Table 10.2 compares the socioeconomic profiles and integration of migrants in central and suburban areas. ANOVA analysis and Pearson’s chi-square test were conducted on the continuous variables and the categorical socioeconomic variables respectively. Overall, the two groups were shown to be significantly different in many aspects, with migrants living in the central city having the better socioeconomic status. Those in the central city were much more likely to hold a university degree; nearly half of the surveyed migrants there had a university degree. They also exhibited a much higher proportion engaging in white-collar jobs, i.e., as administrative managers, technicians, and office staff, but fewer numbers of industrial workers. In terms of hukou classification, the sample included 61.4% rural migrants and 38.6% urban migrants in total. More than half of the surveyed migrants in the central city were urban migrants, while two-thirds of those in the suburbs came from rural areas. Although both groups had only a small number of individuals with Shanghai hukou, more migrants in the central city had managed to acquire one.

Table 10.2 Differences between migrants in the central city and the suburbs
Full size table

Integration of migrants in the central city was significantly better than that in the suburbs in both economic and social terms. The average income of the former group was about 1.8 times greater than that of the latter group (14,500 yuan versus 8099 yuan). Meanwhile, migrants in the central areas were more likely to have local natives as their friends than were migrants in the suburbs (37.7% versus 16.7%), although a large proportion in both groups reported that their friends came from the same province as themselves.

5.2 Economic Integration of Migrants

OLS linear regression was conducted to examine the independent effects of residential location on migrants’ monthly income (Table 10.3). Because the dependent variable was logged monthly income, we could roughly interpret the estimated coefficient as the effect, in percentage, on earnings. The results confirm that socioeconomic attributes – including gender, higher education attainment, occupation, and the length of residence – have important effects on the migrants’ earnings. Male migrants earned 20% more than female migrants. Holding a university degree reflected the human capital of migrants and had a positive effect on their income, although the effect was not significant in the full model. In terms of occupation, the income of white-collar respondents was more than 30% higher than that of blue-collar and retired and unemployed respondents. In the full model, there was also an effect of length of residence, as one extra year of residence in Shanghai increased migrants’ earnings by 13%. However, migrants’ registration status, i.e., hukou location and hukou classification, was found to be not significant for migrants’ earnings.

Table 10.3 OLS Linear Regression on logged monthly income
Full size table

After controlling for other variables in the full model, living in the suburbs was found to have a negative and significant coefficient, indicating that on average, migrants living in the suburbs earned almost 18% less than those living in the central city, even when they have similar individual characteristics such as education and occupation. The place of residence variable remained in the forward linear regression model, which presented the determinants of migrants’ earning returns, further confirming the independent effects of residential location.

5.3 Social Integration of Migrants

Table 10.4 shows the results of logistic regression on migrants’ inter-group social ties. Among socioeconomic attributes, age and holding a university degree increased the likelihood of having an inter-group social network. As older migrants were more established, they were more likely to have local social ties. Holding a university degree increased the odds of having an inter-group friendship by 3.5 times. Meanwhile, as length of stay in Shanghai enhanced the exposure to the host society, it had positive effects on migrants’ social integration: one extra year of residence doubled the odds of having friends from Shanghai and other places. Interestingly, compared to rural migrants, urban migrants from other cities were much more likely to establish friendships with local natives and people from other places. Although this effect was not significant in the full model, the hukou classification variable remained in the forward regression model and was statistically significant. This implies that it is easier for urban residents to adapt to social life in the destination city.

Table 10.4 Logistic regression on inter-group friendship
Full size table

After controlling for the effects of individual attributes in the full model, residential location was again found to affect migrants’ social integration. For the migrants in Shanghai, living in the suburbs made them two times less likely to have inter-group friendships. This variable also remained in the forward regression model, suggesting it is one of the determinants of having friends outside the migrant group.

5.4 Place Effects of Suburban Residence on Migrants’ Residential Mobility

To test the spatial entrapment of migrants in the suburbs, the residential mobility of migrants was further examined. As has been found in the existing literature, migrants in the survey showed a high level of residential mobility. About 49% of respondents reported that they had moved at least once, and about 45.7% of moves had taken place in the last three years. Among those who had never moved after coming to Shanghai, 70.5% had settled in the suburbs, confirming that the suburbs are usually migrants’ initial residential location. Moreover, only 7% of those who had previously lived in the suburbs had managed to move to the central city.

Table 10.5 shows the results of the logistic regression analysis, which tests the independent effects of previous residential location on migrants’ subsequent location choices. Besides socio-economic attributes and hukou status as used above, monthly income and inter-group friendship were also included as measures of migrants’ integration. The results suggest that, among those who move, married migrants are much more likely than single migrants to live in the central city. Local hukou also significantly increases the likelihood of moving to the central city. The level of earnings, which is an indicator of migrants’ economic integration, has a positive effect on the chance of living in the central city. After controlling for other variables, previous residential location clearly plays an important role in migrants’ subsequent residential location. Therefore, for migrants with similar socioeconomic backgrounds, those whose initial residence is located in the suburbs have much less of a chance of moving to the central city and are likely to be trapped in the suburbs.

Table 10.5 Logistic regress on migrants’ residential location after recent moves
Full size table

6 Concluding Remarks

There is a growing literature on migrants’ residential patterns in China (Wu, 2008; Wang et al., 2010; Huang & Tao, 2014). Earlier studies have found that migrants are disadvantaged in the urban housing market and usually concentrate in the suburbs. However, little is known about the place effects of suburban residence on migrants. In fact, residential location is not only a well-established indicator of people’s socio-economic status but is also a valued resource for them; it is therefore key to migrants’ integration into the host society. In the US, there is always debate about whether newcomers to the country could achieve spatial assimilation, or if structural sources of inequality always impede the improvement of international migrants’ spatial position (Massey, 1985; Logan & Alba, 1993). Based on a survey in Shanghai, this paper has empirically examined the relationship between internal migrants’ integration and their suburban residential locations. While moving to the suburbs could reflect the spatial assimilation of international migrant groups in North America and many developing countries, internal migrants may be spatially marginalised in disadvantaged suburban locations in China.

Specifically, there are three empirical findings. First, migrants living in the suburbs are earning less overall than their counterparts in the central city, even after controlling for demographic and socioeconomic characteristics. This indicates that economic returns to migrants’ human capital are spatially unequal. Second, due to the relative lack of exposure to the host society in the suburbs, suburban migrants are less likely to have inter-group friendships or a diverse social network, which is of great significance in their social integration. Finally, while the central city has advantages in terms of more and better job opportunities and living facilities, compared to migrants living in central locations, those living in the suburbs are less likely to move to the central city but instead be trapped in the suburbs. Overall, suburban residence has long-term negative effects on both the socio-economic and spatial integration of migrants.

The interpretation of these results therefore highlights structural factors as the main mechanism leading to the social-spatial exclusion of migrants in the suburbs, lending support to the place stratification model. With the establishment of land and housing markets, Chinese cities have experienced massive spatial restructuring. While both the central city and its suburbs are flourishing, the former remains dominant in both administrative and functional terms. However, because migrants are largely excluded from the formal housing market, they are squeezed out of the central locations where there are more resources to marginal suburban communities such as urban villages. Due to a relative lack of good job opportunities, poorer access to quality facilities and services and more isolated neighbourhoods in the suburbs, their life chances are in turn further constrained.

The integration of rural-to-urban migrants in the cities, which is regarded as a key factor in facilitating urbanisation and social stability, has recently become the focus of inclusive urbanisation in China. This study suggests that migrants concentrating in the relatively underdeveloped suburban locations have limited locational resources and opportunities for upward social mobility. Accordingly, beyond institutional adjustments, such as relaxing the hukou system, the improvement of the suburban infrastructure and the provision of good quality public services are also critical for migrants’ incorporation.

Another policy implication is associated with recent migrant housing policies, which focus on the demolition of urban villages and the construction of large-scale social housing communities in outer suburban areas. In fact, many of the first urban villages to be erased are in the inner suburbs of the city. They have played a positive role in providing affordable housing to rural migrants at locations closer to the urban economy. However, after these urban villages have been demolished, migrants may be displaced farther away, making it less likely for them to find a job similar to the one they had previously. Suburban large-scale social housing communities at remote locations are also at risk of being marginalised and alienated. Therefore, to reserve the locational advantages for migrants, future policies of redevelopment should consider upgrading the migrants’ current accommodation rather than opting for demolition and displacement.