Housing choices of migrant workers in China: Beyond the Hukou perspective

Habitat International 49 (2015) 474e483

Contents lists available at ScienceDirect

Habitat International journal homepage: www.elsevier.com/locate/habitatint

Housing choices of migrant workers in China: Beyond the Hukou perspective Li Tao*, Eddie C.M. Hui, Francis K.W. Wong, Tingting Chen Department of Building and Real Estate, the Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong

a r t i c l e i n f o

a b s t r a c t

Article history: Received 22 December 2014 Accepted 19 June 2015 Available online 28 June 2015

The Hukou system has been increasingly used to explain the housing choices of migrant workers in China. These workers are not as enthusiastic to transfer their Hukou to the locality as the public has expected. Moreover, the role of Hukou is declining. Only a few studies have quantitatively analyzed the important roles of the circular status and coping strategies of migrant workers in their housing choices in China. To fill such knowledge gap and to verify the role of Hukou, this paper investigates the housing tenure and housing choices of migrant workers from the perspective of household strategies. Shenzhen is selected for the case study. Interestingly, Hukou has an indirect role in migrant housing. The remittances, plan to return to their hometowns, and residential mobility plans of migrant workers significantly influence their housing choices. Income has a greater influence on housing choices than housing tenure, but the opposite result is found for social security. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Household strategies Specific housing choices Housing tenure Migrant workers Shenzhen

1. Introduction The floating population in China has dramatically increased from 6.6 million in 1982 to 236 million in 2012 because of the widening economic gap among regions and the deregulation of the floating population policy since the 1980s (Zheng & Yang, 2013). The proportion of this segment in the total population has also increased from 0.7% to 16.7% in 30 years. This floating population (referred to as migrant workers1 in this study as they typically move around to seek jobs) is generally placed at a disadvantaged position in migrant-receiving cities in terms of welfare provision. Although migrant workers substantially contribute to the locality, they are largely excluded from the local public housing system, with the exception of public rental housing in some cities or housing for so-called “talents” (i.e., highly educated and highly skilled workers). Rental housing remains the most common

* Corresponding author. E-mail addresses: [email protected] (L. Tao), [email protected] (E.C.M. Hui), [email protected] (F.K.W. Wong), tingting.chen@connect. polyu.hk (T. Chen). 1 According to the household registration status (i.e., Hukou) in the locality, migrant workers can be classified either as temporary or permanent. Temporary migrant workers do not have local Hukou and have limited access to the local welfare system. Conversely, permanent migrant workers have local Hukou, are regarded as part of the local population, and can enjoy the same social welfare benefits as the locals. http://dx.doi.org/10.1016/j.habitatint.2015.06.018 0197-3975/© 2015 Elsevier Ltd. All rights reserved.

housing choice for migrant workers (Jiang, 2006; Wu, 2002). Overcrowding in living spaces is a common quality issue among them (Li & Duda, 2010; Wang, Wang, & Wu, 2010, 2004). The Hukou2 system has become increasingly useful in explaining the disadvantages that migrant workers encounter in China (Chai & Chai, 1997; Huang & Clark, 2002; Huang & Jiang, 2009; Logan, Fang, & Zhang, 2009; Wang et al., 2010; Wu, 2004; Wu, 2012). It prevents migrant workers from settling down in a locality or obtaining homeownership because of its association with the local welfare system. On the other hand, migrant workers are not as enthusiastic to transfer their Hukou to the locality, particularly in less developed cities because they fear that they will lose their farmlands or homesteads in their hometowns (Hu, Zhu, Lin, & Wang, 2011; Lu, 2008; Zhang, 2011; Zhu, 2007). Not all migrant workers prefer to settle down in the locality (Lin & Zhu, 2010; Zhang & Chen, 2014) as they mostly consider the locality as their workplace instead of their home. Therefore, these workers continually move back and forth between the locality and their hometowns to maximize their earnings and to minimize their expenditures in the city (Zheng, Long, Fan, & Gu, 2009). The circular movement and double

2 The Hukou system is a fundamental institution in China that was introduced in the late 1950s. This system has two criteria, namely, original living place (i.e., local vs. non-local) and Hukou type (i.e., agricultural vs. non-agricultural). Every citizen is allocated a Hukou location and type, which is passed on from parents to their children.

L. Tao et al. / Habitat International 49 (2015) 474e483

residential status of these workers must be investigated. Migrant workers are enabling/active agents instead of passive recipients. Faced with disadvantages in terms of social capital, human capital, and institutional discrimination, these workers strive to cope with such disadvantages and formulate effective strategies (Liu, Wang, & Tao, 2013). The influence of coping strategies for shaping the behaviors of migrant workers in the locality must not be underestimated. Given that the housing choices of these workers vary across cities (Huang, 2004; Li, 2000; Wu, 2004), contextual factors must also be considered. “Beyond the Hukoucentered approach” is called for (Huang, Guo, & Tang, 2010; Zhu, 2007). The rest of the current paper is organized as follows. The following section provides a comprehensive review of the literature. The knowledge gaps are identified, the objectives and theoretical framework of this research are proposed, and the methods of the study are introduced. The research objects are profiled in the fifth section. Six models are established for investigating the factors that affect the housing tenure and choices of migrant workers, and the findings of this study are verified. The findings are discussed and the conclusions are presented at the end of the paper. 2. Literature review Housing is composed of various attributes, such as location, tenure, neighborhood, price, and size. Accordingly, the rationales behind housing choices are complex. Geographers and demographers are interested in the demographics, socio-economic aspects, dwelling, and characteristics of the neighborhood that affect housing choices. Wang and Li (2004) revealed that potential homebuyers in Beijing focus more on neighborhood attributes than dwellings. Li and Li (2006) examined the changes in the housing tenure among the residents of Guangzhou, China. In addition to age and education, the change in marital status can significantly affect housing tenure. The relationships among households, work units, and the state have subtle effects on the tenure choices of households, which echo the findings of Huang and Clark (2002). Several studies have investigated housing choices from other perspectives, such as lifestyles and uncertainties. ÆRØ (2006) explained why residents in Denmark opted for a particular type of dwelling by referring to their lifestyle variables and revealed that personal tradition strongly affected their dwelling choices. Jansen (2012) explored the effect of lifestyle variables on the housing choices of Dutch households with an average or high level of income. Although the lifestyle variables of individuals contribute to their housing choices, the effects of such variables are smaller than those of their socio-demographic characteristics. Zhou (2011) indicated that the uncertainties in several aspects, such as unemployment, education, and medical expenses, negatively affected the home ownership rate among families in China. The huge influx of migrant workers that is caused by industrialization and urbanization has spurred considerable research on the housing choices of these workers. Owusu (1998) reported that the duration of residence, income, family size, initial motives of migration, ties to the hometown, desire for home ownership, and intention to return all affected the home ownership of Ghanaian immigrants in Canada. Jun, Ha, and Jeong (2013) revealed that Korean Chinese residents tend to live close to their friends and relatives in Seoul. These Korean Chinese residents also tend to live in multi-family housing with low rents and are located close to workplaces and urban services. Migrants in China tend to live in collective and private rental housing than in purchased housing (Logan et al., 2009). The role of institutional factors differentiates China from other countries. The Hukou system largely restricts the housing choices of renters (Huang, 2003) and significantly

475

contributes to housing inequality (Huang & Jiang, 2009). Local and non-local divisions are the most important determinants of home ownership. Those residents without local Hukou typically live in small housing units with poor facilities. Recent migrants with rural Hukou tend to be stuck on the bottom rung of the housing ladder (Logan et al., 2009). Wu (2004) examined the institutional and socioeconomic factors underlying the housing tenure, rental sector (public or private), and housing conditions of migrant workers in Beijing and Shanghai. Sources of disadvantage are largely rooted in the Hukou system, particularly in the local-nonlocal division. Song, Zenou, and Ding (2008) examined the socio-demographic and institutional factors underlying the housing types of the residents in Shenzhen and reported that those people ascribed with nonlocal or local rural Hukou, a low level of education, young, selfemployed, low income, and short intention to stay were more likely to live in urban villages3 than in other housing types. Moreover, the housing choices of individuals vary across cities. For example, most migrants in Beijing live in dormitories, whereas migrants in Shanghai live in private rental housing (Wu, 2004). Studies on migrant housing in China largely focus on the constraints that are brought by institutional arrangements, particularly the Hukou system. However, migrant workers are “enabling agents” instead of passive recipients. They actively cope with the challenges that they are facing in the locality (Liu et al., 2013, 2014). The circular status of these migrants should also be given special attention (Fan, Sun, & Zheng, 2011; Lin & Zhu, 2010; Wu, 2002; Zheng et al., 2009). Zhu (2007) contended that the Hukou reform has limited effects on either the permanent settlement or integration of migrant workers into the locality. He suggested that policy makers should focus on the temporary nature and potential roles of migrant workers in developing their hometowns. Huang et al. (2010) indicated that the Hukou status has a declining role in the social exclusion of ruraleurban migrants. Market competition also has a very important role. Liu et al. (2013) emphasized the importance of understanding the “coping strategies” that migrant workers adopt to examine their housing experiences in urban areas. They proved that migrants who are connected to local residents rather than to more people tend to live in formal housing and enjoy better housing conditions. Wu (2002) analyzed migrant housing conditions and observed that migrant workers made their housing decisions according to the convenience to work and save in Beijing and Shanghai. From the same perspective, Hui, Zhong, and Yu (2012) examined the workeresidence matching of new immigrants in Hong Kong. Li, Duda, and An (2009) argued that the transitional economic contexts and the characteristics of individual migration could exert a greater influence on the housing providers and cost of migrants than the conventional factors in Taiyuan, China. In terms of settlement intentions, Lin and Zhu (2010) examined the diversified housing demands of ruraleurban migrants for housing security policies in Fuzhou, China and revealed that only a small proportion of these migrants intended to settle down in the locality. They also argued that housing should cater to the “floating” needs of migrant workers. Hui, Yu, and Ye (2014) contended that the type of Hukou did not affect the willingness of migrant workers from urban villages to move to public rental housing in the wake of the gradual Hukou reform in Shenzhen. Tao, Wong, and Hui (2014) reported that the type of Hukou did not affect

3 Urban villages are rooted in the dual-land system of China (i.e., urban land is state owned, whereas rural land is collectively owned). In the urban expansion process, local governments prefer to acquire only the farmland and disregard the land for homestead use to avoid the huge requisition costs that are associated with demolition, resettlement, and compensation. Therefore, villages are isolated and urban villages have emerged.

476

L. Tao et al. / Habitat International 49 (2015) 474e483

the residential satisfaction of migrant workers. Few studies have quantitatively analyzed the important role of the circular status and coping strategies of migrant workers in their housing choices. If there are any, they have only focused on the housing conditions, providers, or cost of migrant workers in several selected cities of China. The specific housing choices of these workers warrant further investigation. Previous studies have also underexamined how the ties of migrant workers to their places of origin or migrant-sending areas (in terms of remittance, home ownership in hometowns, etc.) affect their housing choices. Empirical evidence must also be obtained to verify whether the role of Hukou is declining or overstated. Based on the preceding arguments, this research aims (1) to examine the household strategies4 (including the household characteristics, ties to places of origin, and mobility characteristics) of the migrant workers; (2) to investigate the effects of household strategies on their housing tenure and specific housing choices; (3) to verify the role of Hukou in migrant housing; and (4) to propose suggestions on how to address the housing needs of these workers. Shenzhen, which mostly comprises non-local residents, is selected for the case study. 3. Theoretical framework Consumer behavior theory (Lancaster, 1966) maintains that when making housing decisions, people aim to maximize the utilities to be derived by comparing a set of alternatives under certain constraints. Migrant workers are the housing consumers in their destination cities, although their housing choices are largely restricted because of the constraints in their income, social capital, human capital, working status, and institutional discriminations. Migrant workers are enabling agents who strive to formulate various household strategies to maximize their utilities and minimize their risks in the locality, such as split households (Fan et al., 2011), continued circulation, savings orientation, remittance, and home ownership in places of origin. According to the laws of migration (Ravenstein, 1885), migrants do not reach their destination localities directly, but rather arrive in their destinations through several steps. Each migration stream tends to be accompanied by a compensating counter-stream. Lee (1966) further explored the push-and-pull factors of migration and associated them with the places of origin, destination, intervening obstacles, and personal factors of the migrants. Therefore, the migration flows, the push-and-pull factors that are associated with places of origin and destination, and the coping strategies of migrant workers must be carefully examined to explain their specific housing choices. To this end, the following hypotheses are proposed: H1 Household characteristics (e.g., type of Hukou, social security, family size in the locality, period of stay, and household expenditure) significantly affect the housing tenure and specific housing choices of migrant workers. H2 The ties of migrant workers to their places of origin, such as their remittance, residence in hometowns, and plan to return, significantly affect their housing tenure and specific housing choices. H3 Mobility characteristics, such as history of residence change within the locality, residential mobility plan, and intention to

4 Household strategies refer to “coping” or “surviving” strategies that are adopted to survive in a challenging environment by “drawing on a range of resources.” These strategies were initially applied to people in marginal positions in the society, such as peasants and immigrant entrepreneurs (Wallace, 2002).

80.0%

72.7%

70.0%

60.0% 50.0% 40.0%

37.3%

40.1%

Beijing

Shanghai

36.0%

30.0% 20.0% 10.0% 0.0% Guangzhou

Shenzhen

Fig. 1. Percentage of the non-local population in the four first-tier cities of China. Source: Statistical Yearbook 2013 of each city.

work in other cities, significantly affect the housing tenure and specific housing choices of the migrant workers. Migration is not a one-way process because such concept includes both bi-directional and multi-directional flows. Therefore, the intention of migrants to move to another city must also be considered. The revealed preference approach, which is based on actual housing choices, is used to test the aforementioned hypotheses. An indirect utility function is constructed using the observed housing choice to analyze the revealed housing preferences of the respondents (King, 1980). Those variables that cannot be directly interpreted as arguments for the utility function (e.g., socioeconomic characteristics) are included as explanatory variables (Nechyba & Strauss, 1998). 4. Research methods 4.1. Study area and data collection Shenzhen is a unique migrant city with non-local residents accounting for 72.7% of the total population in 2012 (Shenzhen Bureau of Statistics, 2014). Shenzhen, which is a coastal city in Southern China that is adjacent to Hong Kong, is a sub-provincial city of the Guangdong Province. This city has the highest proportion of non-local population among the four first-tier cities (Fig. 1) and even among all cities in China. The study area was divided into six survey districts, namely, Luohu, Futian, Nanshan, Yantian, Bao'an, and Longgang (refer to the colored parts of Fig. 2). The data collection was divided into three stages (Fig. 3). Four semi-structured interviews and a 60-person sampled questionnaire survey were conducted for the pilot study in 2009. The interviewees included four officials from the Shenzhen government and the research institutions of the region. The pilot questionnaire survey was conducted in Futian District. Face-to-face questionnaire surveys were conducted during the second stage in the entire Shenzhen from March 2010 to August 2010. The respondents comprised migrant workers who have been staying in Shenzhen for at least half a year but do not have a local Hukou. The stratified sampling method was employed, and the trades of the respondents were selected as the sampling strata (Li & Duda, 2010; Li & Zhang, 2011). Large- and medium-scale firms were randomly selected from the Chinese Enterprise Networks Yellow Pages. Small-scale firms, such as retail stores and hair salons, were randomly selected from the streets. The respondents were approached in or outside of their working places during their breaks and were given a small token for appreciation for their participation. A total of 540

L. Tao et al. / Habitat International 49 (2015) 474e483

477

Fig. 2. Survey districts of Shenzhen.

Fig. 3. Flowchart of the data collection process.

questionnaires were distributed in Shenzhen, of which 450 valid questionnaires5 were received. The sample size of each district was determined based on the proportion of its non-local population to local residents. During the third stage, two semi-structured interviews and 40 structured interviews were conducted with the Shenzhen government (in December 2011 and March 2012) and migrant workers (in May 2012), respectively, to verify the findings from the questionnaire survey.

Model 1: Household characterisƟcs Housing tenure

Model 3: Mobility characterisƟcs

Housing choices

Model 4:

4.2. Data analysis A multinomial logistic regression is employed considering that housing choice is a categorical variable and that the predictor variables are either continuous or categorical (Field, 2009). Cross tabulation analysis was used to test the independence among the nominal/categorical variables. Six models were developed to test the three hypotheses (Fig. 4).

5

Given that regression analysis requires a sample size that is at least 10 times the number of independent variables (Fellows & Liu, 2008), 450 valid responses are considered acceptable in view of the 24 independent variables (including the nominal variables) in the present study.

Model 2: Ties to places of origin

Household characterisƟcs

Specific housing choices

Model 5:

Ties to places of origin Model 6: Mobility characterisƟcs

Fig. 4. Model design.

Models 1, 2, and 3 explored the factors that affected the housing tenure of migrant workers, whereas models 4, 5, and 6 examined the factors that influenced the specific housing choices of migrant workers. To establish a more concise model, the forward stepwise

478

L. Tao et al. / Habitat International 49 (2015) 474e483

Table 1 Household characteristics. 1

Average age

29

2

Gender

3

Marital status

4

Children

5

Family members in Shenzhen

6

Education

7

From another province

8

Type of Hukou

9

Social security in Shenzhen

10

Period of stay (years)

11

Household income (RMB/year)

12

Household expenditure (RMB/year)

13

Spouse income (among married respondents)

Male Female Yes No Yes No Yes No Primary school or below Middle school College or above Yes No Rural Urban Yes No Average Medium Average Medium Average Medium Yes No

method, which combines both forward entry and backward removal methods (Wang, Zhang, & Bakhai, 2004), was employed to examine the household characteristics of the migrant workers. The forced entry method was used to investigate the ties of these workers to their places of origin and their mobility characteristics, which were of significant interest in this study. 5. Empirical results 5.1. Household characteristics of migrant workers Most of the respondents are single, young males (Table 1). The respondents have a mean age of 29 years. Males account for 68.9%. In terms of civil status, 54.2% of the respondents are unmarried, whereas 83.8% of the married respondents have children. The respondents have stayed in Shenzhen for an average of six years, and most of them (69.8%) are living alone without any family members and have completed middle-school education. Most of the respondents (65%) come from rural areas, of whom 62.5% are from another province. More than half (57.3%) of the respondents are not covered by the social security system. Their average annual household income is RMB 36,966, whereas the average household income in Shenzhen in 2010 is RMB 114,386 (Shenzhen Bureau of Statistics, 2014). The median household income of the respondents is RMB 25,000 per year, which is significantly lower than their average income and reveals a large income disparity. The respondents have conservative consumption behaviors. Their average annual household expenditure is RMB 23,160, whereas the average annual household expenditure in the city is RMB 86,063 in 2010 (Shenzhen Bureau of Statistics, 2014). Approximately 43.2% of the married respondents have spouses who do not generate any income. 5.2. Ties to places of origin and mobility characteristics of migrant workers Many migrant workers maintain close ties with their places of origin (Table 2), which may significantly affect their housing

68.9% 31.1% 45.8% 54.2% 38.4% 61.6% 30.2% 69.8% 5.6% 80.4% 14.0% 62.5% 37.5% 65.0% 35.0% 42.7% 57.3% 6 5 36,966 25,000 23,160 14,400 56.8% 43.2%

choices in the locality. Roughly 43.1% of the respondents remit part of their income to their family members in their hometowns every month, with an average remitted amount of RMB 801. Approximately 96.9% of the respondents own residences in their hometowns, whereas 38% are planning to return to their hometowns. Migrant workers maintain their mobile status after arriving in the locality, but largely in the same district. Roughly 52.9% of the respondents have changed their residence since their arrival in the region, whereas 19.6% of the respondents have changed their region of residence. Approximately 28.2% of those respondents with residential mobility have changed their regions, 20.9% intend to work in other cities, and 56.2% have a residential mobility plan in the locality. 5.3. Housing of migrant workers Dormitories and rental housing are the two most common housing choices for the respondents, which echo the findings of Ding, Qiu, and Wang (2011). Up to 44.4% of the respondents are residing in free dormitories, 24.9% are living in rental housing in urban villages, and 16.4% are residing in rental commodity housing (Fig. 5). This finding is different from that of CGSS,6 whose findings indicate that 84.6% and 6.6% of the migrant workers are living in private rental housing and dormitories, respectively. The difference in these findings can be attributed to the variance in the sampling methods and in the effects of the industry structure. Additional details are provided in the verification section. 5.4. Factors that affect the housing choices of migrant workers Six models were developed to test the three hypotheses. To avoid a considerable number of zero cells in the factor space or an inaccurate analysis, the housing types were grouped into three categories, namely, commodity housing, dormitories, and rental

6 The China General Social Survey is an annual national survey that has been conducted since 2003 by the People's University of China and the Hong Kong University of Science and Technology.

L. Tao et al. / Habitat International 49 (2015) 474e483

479

Table 2 Ties to places of origin and mobility characteristics. Ties to places of origin

Mobility characteristics

50.0% 45.0% 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0%

1

Remittance to hometown

2

Housing ownership in hometown

3

Plan to return to hometown

1

Residence change history in Shenzhen

2

Same district

3

Residential mobility plan

4

Plan to work in other cities

Yes No Yes No Yes No Not sure Yes No Yes No Yes No Yes No Not sure

43.1% 56.9% 96.9% 3.1% 38.0% 38.7% 23.3% 52.9% 47.1% 80.4% 19.6% 56.2% 43.8% 20.9% 34.4% 44.7%

44.4%

24.9% 16.4% 7.6%

3.8%

2.0%

Purchased Free Rental Rental Rental RelaƟves' commodity dormitory commodity housing in housing or friends' housing housing urban provided place villages by the work unit

0.9% others

Fig. 5. Specific housing choices of migrant workers.

housing in urban villages. Given the significant role of rental housing in urban villages in accommodating migrant workers, this type of housing was selected as the reference category. (1) Factors that affect the housing ownership of migrant workers Models 1, 2, and 3 were established to examine the housing ownership of migrant workers (Table 3). The three models were significant at the 99% confidence level. Model 1 analyzed the effects of household characteristics (i.e., age, education,7 household income,8 trade, gender, marital status, children, province, type of Hukou, income of spouse, social security, family size, period of stay, and household expenditure9) using the forward stepwise procedure. Household expenditure, family size, education, type of Hukou, social security, and period of stay were identified as significant. Migrant workers have a lower tendency to live in rental housing if they have more family members residing in the locality and a higher educational attainment. This finding is also applicable to the

7 Education was categorized into six levels, namely, primary school or below, middle school, high school, technical secondary school, college, and university. 8 Household income (RMB/year) was divided into below 10,000, 10,000 to 20,000, 20,000 to 50,000, 50,000 to 100,000, and above 100,000. Each scale nearly had the same number of the respondents. 9 Household expenditure (RMB/month) was divided into below 700, 700 to 1,000, 1000 to 1,500, 1500 to 2,000, and above 2000. Each scale nearly had the same number of respondents.

other segments of the population (Li & Li, 2006; Zhou, 2011). Each additional year of stay increases the possibility for the migrant workers to own a residence by 1.16 times (i.e., 1/0.86). The migrant workers with rural Hukou are 3.67 times more likely to reside in rental housing. The less saving-oriented migrant workers (i.e., with higher household expenditure) are also more likely to live in purchased housing. As a supplement to the literature, those migrant workers who are not covered by the social security are 6.82 times more likely to live in rental housing than those who are covered, which is not only objectively forced, but is also subjectively led by their feelings of uncertainty. Those migrant workers who are not covered by the social security typically have higher non-housing-related expenses and more uncertainties in terms of medical expenses, education fees, and employment. Based on Model 1, Model 2 considered remittances, plan to return to hometowns, and residence ownership in hometowns. Forced entry method was employed. These three variables were insignificant. Based on Models 1 and 2, Model 3 considered plan to work in other cities, history of residence change, and residential mobility plan in the locality. Residential mobility plan was significant. Those migrant workers without any residential mobility plan are 6.67 times (i.e., 1/0.15) more likely to own a residence than those who are planning to change their residence, which supports our hypothesis regarding mobility characteristics (H3). Interestingly, Hukou was not significant in Model 3. According to the survey, rural migrants account for 73.1% of the respondents

480

L. Tao et al. / Habitat International 49 (2015) 474e483

Table 3 Factors affecting the housing ownership of migrant workers. 95% confidence interval for odds ratio B(SE) Model 1: Household characteristics

Model 2: Ties to places of origin

Model 3: Mobility characteristics

Intercept 1. Expenditure scale 2. Number of family members in SZ 3. Education 4. Rural Hukou 5. No social security in SZ 6. Period of stay Intercept 1. No remittance 2. No plan to return 3. No residence in hometown 4. Expenditure scale 5. Number of family members in SZ 6. Education 7. Rural Hukou 8. No social security in SZ 9. Period of stay Intercept 1. Same residence 2. No residential mobility plan 3. No plan to work in other cities 4. Expenditure scale 5. Period of stay 6. Education 7. No social security in SZ

8.43 (1.56***) ¡0.61 (0.28*) ¡0.47 (0.19*) ¡0.51 (0.24*) 1.30 (0.59*) 1.92 (0.79*) ¡0.16 (0.05**) 9.84 (1.89***) 0.97 (0.63) 0.56 (0.68) 0.64 (1.42) ¡0.67 (0.29*) ¡0.49 (0.20*) ¡0.51 (0.25*) 1.22 (0.60*) 2.28 (0.88**) ¡0.17 (0.05**) 10.38 (1.92***) 0.83 (0.62) ¡1.90 (0.67**) 0.17 (0.82) ¡0.80 (0.29**) ¡0.22 (0.05***) ¡0.63 (0.25*) 1.52 (0.74*)

Lower

Odds ratio

Upper

0.32 0.43 0.37 1.15 1.45 0.78

0.54 0.62 0.60 3.67 6.82 0.86

0.93 0.91 0.97 11.69 32.17 0.94

0.11 0.15 0.12 0.29 0.42 0.37 1.05 1.73 0.76

0.38 0.57 1.90 0.51 0.61 0.60 3.41 9.74 0.84

1.30 2.17 30.67 0.90 0.90 0.97 11.13 54.88 0.93

0.68 0.04 0.24 0.26 0.72 0.33 1.06

2.30 0.15 1.18 0.45 0.81 0.53 4.55

7.76 0.56 5.89 0.79 0.89 0.87 19.53

Note: *p< .05, **p< .01, ***p< .001.

who have a residential mobility plan. Chi-square tests were performed to test the relationship between Hukou and residential mobility plan. The result was significant [c2 (1, N ¼ 450) ¼ 17.20, p < 0.001], which indicates that Hukou may have an indirect effect on housing tenure. (2) Factors that affect the specific housing choices of migrant workers Models 4, 5, and 6 were established to test the effects of the same three groups of explanatory variables on the specific housing choices of migrant workers. The three models were significant at the 99% confidence level. Model 4 was constructed using six variables via the forward stepwise procedure (Table 4). Compared with rental housing in urban villages, those migrant workers with a higher household expenditure, more family members in the locality, or a higher income level are less likely to live in dormitories. The migrant workers with a higher level of education are 1.39 times more likely to live in commodity housing. Each additional year of

stay in the locality increases the possibility for a migrant worker to live in commodity housing by 1.02 times. Those migrant workers who have stayed longer in the locality may have higher economic accumulation, wider social networks, and can seek more economic and emotional support in the locality than those who have just arrived. The type of Hukou was excluded from the final model, which may be attributed to the significant effect of income scale, which was insignificant in Model 1. A chi-square test was performed to determine whether migrant workers with different types of Hukou were classified under different income scales. A significant result was obtained [c2 (4, N ¼ 450) ¼ 11.16, p < 0.05], hence confirming the indirect role of Hukou in migrant housing. Based on Model 4, those variables that concerned the ties of the workers to their places of origin were entered into Model 5 (Table 5). In addition to the five significant variables in Model 4, remittance, plan to return, and gender were found significant in Model 5. Compared with rental housing in urban villages, those migrant workers who are sending remittances to their hometowns

Table 4 Factors affecting the specific housing choices of migrant workers (Model 4: Household characteristics). 95% Confidence interval for odds ratio Commodity housing B(SE) Intercept 1. Expenditure scale 2. Number of family members in SZ 3. Education 4. Trade Manufacturing Construction Transportation, warehousing and postal industry Whole sale and retail 5. Income scale 6. Period of stay Note: *p< .05, **p< .01, ***p< .001.

Lower

Dormitories Odds ratio

Upper

B(SE)

Lower

Odds ratio

Upper

0.62 0.51 0.79

0.76 0.64 0.98

0.93 0.81 1.23

0.46 0.95 0.67 0.36 0.47 0.92

0.93 2.24 1.55 0.75 0.66 0.97

1.89 5.27 3.60 1.54 0.94 1.03

¡2.04 (0.72**) 0.10 (0.13) 0.00 (0.11) 0.33 (0.13**)

0.71 0.81 1.09

0.91 1.00 1.39

1.16 1.24 1.78

3.61 (0.61***) ¡0.28 (0.11**) ¡0.44 (0.12***) 0.02 (0.11)

0.93 (0.49) 0.00 (0.53) 0.42 (0.48) 0.27 (0.42) 0.19 (0.20) 0.07 (0.03**)

0.15 0.36 0.59 0.58 0.82 1.02

0.40 1.00 1.52 1.30 1.21 1.02

1.03 2.83 3.90 2.94 1.79 1.14

0.07 (0.36) 0.81 (0.44) 0.44 (0.43) 0.29 (0.37) ¡0.41 (0.18*) 0.03 (0.03)

L. Tao et al. / Habitat International 49 (2015) 474e483

481

Table 5 Factors affecting the specific housing choices of migrant workers (Model 5:3000;Ties to the hometown). 95% Confidence interval for odds ratio Commodity housing B(SE) Intercept 1. No remittance 2. No plan to return 3. No residence in hometown 4. Expenditure scale 5. Number of family members in SZ 6. Education 7. Female 8. Trade Manufacturing Construction Transportation, warehousing and postal industry Whole sale and retail 9. Income scale 10. Period of stay

Lower

¡2.05 (0.81*) 0.15 (0.33) 0.80 (0.33*) 0.53 (0.77) 0.10 (0.13) 0.03 (0.11) 0.33 (0.13**) 0.63 (0.34)

0.45 1.17 0.13 0.70 0.83 1.08 0.28

0.95 (0.50) 0.04 (0.55) 0.44 (0.49) 0.14 (0.43) 0.14 (0.20) 0.07 (0.03*)

0.15 0.33 0.60 0.50 0.77 1.01

Dormitories Odds ratio

Upper

B(SE)

Lower

Odds ratio

Upper

0.86 2.23 0.59 0.90 1.03 1.39 0.54

1.63 4.23 2.67 1.17 1.28 1.80 1.04

4.26 (0.69***) ¡0.86 (0.28**) 0.34 (0.27) 0.85 (0.70) ¡0.33 (0.11**) ¡0.42 (0.12***) 0.00 (0.12) ¡0.59 (0.29*)

0.24 0.82 0.11 0.58 0.52 0.80 0.31

0.42 1.40 0.43 0.72 0.66 1.00 0.56

0.73 2.39 1.70 0.89 0.83 1.25 0.99

0.39 0.96 1.56 1.15 1.15 1.07

1.03 2.83 4.06 2.68 1.72 1.14

0.22 (0.38) 0.77 (0.46) 0.42 (0.44) 0.28 (0.38) ¡0.40 (0.18*) 0.04 (0.03)

0.39 0.87 0.64 0.36 0.47 0.91

0.81 2.15 1.51 0.76 0.67 0.96

1.68 5.30 3.59 1.60 0.96 1.02

Note: *p< .05, **p< .01, ***p< .001.

are 2.38 (1/0.42) times more likely to live in dormitories. Furthermore, those migrant workers who are not planning to return to their hometowns are most likely to live in commodity housing. Female migrant workers are more likely to live in rental housing in urban villages than in dormitories in Shenzhen, which can be partly attributed to the industries in which they are working. Model 6 investigated the effects of mobility characteristics (Table 6). Different from the housing tenure model, the three mobility variables were all insignificant in Model 6. Mobility characteristics have a more important role in the housing tenure of migrant workers than in their specific housing choices.

70.0%

62.5%

60.0% 50.0% 40.0% 30.0% 15.0%

20.0%

15.0%

7.5%

10.0% 0.0% Rental housing in urban villages

Dormitories

Rental commodity housing

Purchased commodity housing

6. Verification Fig. 6. Housing choices of migrant workers (verification).

6.1. Housing choices of migrant workers Two semi-structured interviews and 40 structured interviews were conducted with government officials and migrant workers, respectively, to verify the findings. According to the verification survey, migrant workers commonly choose rental housing in urban villages, which is followed by dormitories, rental commodity housing, and purchased commodity housing (Fig. 6). The differences in the housing choices with the questionnaire survey can be

attributed to the different sampling methods. The questionnaire survey employed a stratified sampling method and used trade as the sampling criterion. All five trades were evenly covered. On the other hand, the verification interview employed the simple random sampling method. Trades (Fig. 7) have an important role in the housing choices of migrant workers. Those migrant workers who are working in the construction industry have a higher tendency to live in dormitories, whereas those who are working in the

Table 6 Factors affecting the specific housing choices of migrant workers (Model 6:3000;Mobility characteristics). 95% Confidence interval for odds ratio Commodity housing B(SE) Intercept 1. Same residence 2. No residential mobility plan 3. No plan to work in other cities 4. Expenditure scale 5. Period of stay 6. Education 7. Trade Manufacturing Construction Transportation, warehousing and postal industry Whole sale and retail 8. Number of family members in SZ 9. Income scale Note: *p< .05, **p< .01, ***p< .001.

Dormitories Lower

¡2.35 (0.83**) 0.04 (0.30) 0.46 (0.31) 0.02 (0.41) 0.09 (0.13) 0.07 (0.03*) 0.35 (0.13**)

0.54 0.86 0.46 0.71 1.02 1.10

0.93 (0.49) 0.01 (0.54) 0.36 (0.48) 0.22 (0.42) 0.02 (0.11) 0.22 (0.20)

0.15 0.35 0.56 0.55 0.79 0.84

Odds ratio

Upper

B(SE)

Lower

Odds ratio

Upper

0.96 1.59 1.02 0.92 1.07 1.41

1.74 2.93 2.26 1.18 1.14 1.82

3.81 (0.69***) 0.01 (0.26) 0.03 (0.27) 0.30 (0.33) ¡0.29 (0.11**) 0.03 (0.03) 0.02 (0.11)

0.61 0.61 0.39 0.61 0.92 0.79

1.01 1.03 0.75 0.75 0.97 0.98

1.67 1.74 1.42 0.93 1.03 1.23

0.40 0.99 1.43 1.24 0.98 1.25

1.04 2.84 3.70 2.82 1.21 1.85

0.07 (0.36) 0.73 (0.45) 0.40 (0.43) 0.31 (0.37) ¡0.44 (0.12***) ¡0.40 (0.18*)

0.46 0.87 0.64 0.36 0.52 0.47

0.93 2.07 1.49 0.74 0.65 0.67

1.90 4.96 3.47 1.53 0.81 0.95

482

L. Tao et al. / Habitat International 49 (2015) 474e483

50.0% 45.0% 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0%

45.0%

30.0% 17.5% 5.0%

Wholesale and retail

AccommodaƟons, catering and other services

Manufacturing

ConstrucƟon

2.5%

TransportaƟon, warehousing and postal

Fig. 7. Trades of the migrant workers (verification).

manufacturing industry are more likely to live in urban villages (Models 4, 5, and 6). 6.2. Plan to apply for a local Hukou According to the verification survey, 70% of the 40 migrant interviewees are not planning to apply for a local Hukou. Their refusal to apply for a local Hukou is mostly attributed to the high living costs in the locality, the high housing prices in the locality, their ownership of residence in their hometowns, the owning of arable land in their hometowns, their social security, the better personal development opportunities in their hometowns, the high amount of pressure in the locality, and the difficulty of finding a satisfying job in the locality. Those migrant workers who are planning to apply for a local Hukou primarily appreciate the so-called benefits of the system, such as stability, satisfactory jobs, and favorable education opportunities for their children. 6.3. Plan to return to hometowns The respondents were asked why they were planning to return to their hometowns in the near future or why they were unsure about their return. Most respondents attribute such decision to the unaffordable daily expenses in Shenzhen, the unaffordable housing in the region, the need to look after their families in their hometowns, and the Hukou system. By contrast, some respondents are uncertain of returning to their hometowns because of the favorable working opportunities in Shenzhen, the Hukou system, the income level of the region, the availability of housing in the region, and their proximity to their family members. 7. Discussion All three hypotheses are supported.The findings on residential mobility plan, expenditure scale, social security coverage, and remittances to the hometowns largely supplement the findings of previous studies. From the perspective of split households, those migrant workers who are sending monthly remittances typically have family members residing in their hometowns. Dormitories are the most cost-effective housing option for these workers from the savings-oriented perspective. The potential housing costs can be saved as monthly remittances. Hence, having more family members who require support in hometowns may increase the likelihood of housing poverty among migrant workers (Wang et al., 2010). On the other hand, a larger household size in the locality

corresponds to a higher level of residential satisfaction (Tao et al., 2014). Four remarkable findings need to be highlighted. First, household strategies generate different effects on the specific housing choices and housing tenure of migrant workers. Remittances and plan to return to hometowns have significant effects on the specific housing choices of migrant workers, but not on their housing tenure. Residential mobility plan has a significant effect on housing tenure, but not on the specific housing choices of migrant workers. Second, income has a larger influence on the specific housing choices of migrant workers than housing tenure, which may be attributed to the low income of these workers. The majority of these workers cannot afford to purchase a housing unit. Third, social security coverage has a greater effect on housing tenure than on the specific housing choices of migrant workers. Those workers who are covered by the social security are typically engaged in relatively permanent/stable jobs. They receive guaranteed income and encounter lower non-housing-related expenses (e.g., medical expenses and educational fees). Therefore, these workers have a higher budget for their housing expenditures, tend to feel more settled, and are more willing to pursue homeownership in the locality. Fourth, Hukou has an indirect role in shaping the housing choices of migrant workers, which elaborates the effect of Hukou as well as supplements the findings of previous studies. 8. Conclusion Migrant workers in China have been largely considered as passive recipients of the disadvantages that are generated by institutional constraints, particularly, the Hukou system. These workers actually strive to formulate various strategies to cope with the challenges that they face in the locality, such as their continued circular migration, regular remittances, split households, and home ownership in their places of origin. Few studies have quantitatively analyzed the role of these strategies in the specific housing choices of migrant workers. This study aims to fill this knowledge gap by investigating the housing tenure and specific housing choices of migrant workers beyond the Hukou perspective. The effect of Hukou on the housing choices of these workers is verified in response to the declining role of Hukou and its overstated effect. Three rounds of surveys are conducted. Six models are developed to confirm the three hypotheses on household characteristics, ties of the workers to their places of origin, and mobility characteristics. Migrant workers do not regard local Hukou as important, which contradicts public expectation. The role of Hukou is primarily

L. Tao et al. / Habitat International 49 (2015) 474e483

evident in their household strategies. These strategies further shape the housing choices of migrant workers. Affordability, ties to places of origin, family issues, and working opportunities have greater effects on the decisions of migrant workers (not only in terms of housing, but also their mobility plans and local Hukou application) than Hukou. Housing providers, including the government, must focus on the characteristics of this particular population group. The findings of this study provide several implications. First, the residences and arable lands of migrant workers in their hometowns provide them with important security, hence forming the ties of these workers to their hometowns. A unified land and housing market should be established to offer more options (e.g., in terms of housing choices, mobility choices, and budget scales) for migrant workers. Second, the bi- and multi-directional migration (e.g., among migrant-sending cities, migrant-receiving cities, and other cities) among migrant workers must also be examined. Balancing regional development presents a sustainable strategy. Third, a national-level service system that integrates social security, housing provident fund, and other services must be established to allow the workers to enjoy consistent and equal services regardless of their origin and migrating regions. Fourth, the housing stocks in urban villages must be fully utilized by improving the surroundings and housing facilities of the community instead of demolishing them. Fifth, the legal system must be improved to guarantee the security, tenure, job, and wages of migrant workers. This study fills the knowledge gap on the role of household strategies in the housing choices of migrant workers in China. Empirical evidence from Shenzhen verifies the role of Hukou in the housing choices of migrants. The situation may differ across cities because of their varying industrial structures, demographic compositions, and government policies; thus, other Chinese cities must be considered in future comparative studies. Acknowledgements The authors wish to express appreciation to the anonymous reviewers for their valuable comments on this paper. This project was funded by the Hong Kong Polytechnic University's research grant (No. RPF9). References ÆRØ, T. (2006). Residential choice from a lifestyle perspective. Housing, Theory and Society, 23(2), 109e130. Chai, J. C. H., & Chai, B. K. (1997). China's floating population and its implications. International Journal of Social Economics, 24(7), 1038e1051. Ding, C., Qiu, A., & Wang, J. (2011). Migrant housing during rapid urbanization in China: typology and assessment. Urban Studies, 18(6), 49e54. Fan, C. C., Sun, M., & Zheng, S. (2011). Migration and split households: a comparison of sole, couple, and family migrants in Beijing, China. Environment and Planning A, 43, 2164e2185. Fellows, R., & Liu, A. (2008). Research methods for construction (3rd ed.). UK: WileyBlackwell. Field, A. (2009). Discovering statistics using SPSS. , London: Sage. Huang, Y. (2003). Renters' housing behaviour in transitional urban China. Housing Studies, 18(1), 103e126. Huang, Y. (2004). Housing markets, government behaviors, and housing choice: a case study of three cities in China. Environment and Planning A, 36, 45e68. Huang, Y., & Clark, W. A. V. (2002). Housing tenure choice in transitional urban China: a multilevel analysis. Urban Studies, 39(1), 7e32. Huang, Y., Guo, F., & Tang, Y. (2010). Hukou status and social exclusion of ruraleurban migrants in transitional China. Journal of Asian Public Policy, 3(2), 172e185. Huang, Y., & Jiang, L. (2009). Housing inequality in transitional Beijing. International Journal of Urban and Regional Research, 33(4), 936e956. Hui, E. C., Yu, K., & Ye, Y. (2014). Housing preferences of temporary migrants in urban China in the wake of gradual Hukou reform: a case study of Shenzhen. International Journal of Urban and Regional Research, 38(4), 1384e1398. Hui, E. C., Zhong, J., & Yu, K. (2012). Mobility and work-residence matching for new immigrants in Hong Kong. Habitat International, 36, 444e451.

483

Hu, C., Zhu, Y., Lin, L., & Wang, W. (2011). Analysis on floating population's Hukou transfer intention and its influence factors: insights from a survey in Fujian province. Population & Development, 17(3), 2e10 (in Chinese). Jansen, S. J. T. (2012). What is the worth of values in guiding residential preferences and choices? Journal of Housing and the Built Environment, 27, 273e300. Jiang, L. (2006). Living conditions of the floating population in urban China. Housing Studies, 21(5), 719e744. Jun, M. J., Ha, S. K., & Jeong, J. E. (2013). Spatial concentrations of korean Chinese and determinants of their residential location choices in Seoul. Habitat International, 40, 42e50. King, M. A. (1980). An econometric model of tenure choice and demand for housing as a joint decision. Journal of Public Economics, 14, 137e159. Lancaster, K. J. (1966). A new approach to consumer theory. Journal of Political Economy, 74(2), 132e157. Lee, E. S. (1966). A theory of migration. Demography, 3(1), 47e57. Li, S. M. (2000). Housing consumption in urban China: a comparative study of Beijing and Guangzhou. Environment and Planning A, 32, 1115e1134. Li, B., & Duda, M. (2010). Employers as landlords for rural-to-urban migrants in Chinese cities. Environment and Urbanization, 22(1), 13e31. Li, B., Duda, M., & An, X. (2009). Drivers of housing choice among rural-to-urban migrants: evidence from Taiyuan. Journal of Asian Public Policy, 2(2), 142e156. Li, S. M., & Li, L. (2006). Life course and housing tenure change in urban China: a study of Guangzhou. Housing Studies, 21(5), 653e670. Lin, L., & Zhu, Y. (2010). The diverse housing needs of rural to urban migrants and policy responses in China: insights from a survey in Fuzhou. IDS Bulletin, 41(4), 12e21. Liu, Y., Li, Z., Liu, Y., & Chen, H. (2014). Growth of rural migrant enclaves in Guangzhou, China: Agency, everyday practice and social mobility. Urban Studies, 1e20, 0042098014553752. Liu, Z., Wang, Y., & Tao, R. (2013). Social capital and migrant housing experiences in urban China: a structural equation modeling analysis. Housing Studies, 28(8), 1155e1174. Li, B., & Zhang, Y. (2011). Housing provision for rural-urban migrant workers in Chinese cities: the roles of the state, employers and the market. Social Policy & Administration, 45(6), 694e713. Logan, J. R., Fang, Y., & Zhang, Z. (2009). Access to housing in urban China. International Journal of Urban and Regional Research, 33(4), 914e935. Lu, Y. (2008). Does Hukou still matter? the household registration system and its impact on social stratification and mobility in China. Social Sciences in China, 1, 149e162 (in Chinese). Nechyba, T. J., & Strauss, R. P. (1998). Community choice and local public services: a discrete choice approach. Regional Science and Urban Economics, 28, 51e73. Owusu, T. Y. (1998). To buy or not to buy: determinants of home ownership among Ghanaian immigrants in Toronto. The Canadian Geographer, 42(1), 40e52. Ravenstein, E. G. (1885). The laws of migration. Journal of the Statistical Society of London, 48(2), 167e235. Shenzhen Bureau of Statistics. (2014). Shenzhen statistical yearbook 2013. Retrieved June 26, 2014, from http://www.sztj.gov.cn/nj2013/indexce.htm. Song, Y., Zenou, Y., & Ding, C. (2008). Let's not throw the baby out with the bath water: the role of urban villages in housing rural migrants in China. Urban Studies, 45(2), 313e330. Tao, L., Wong, F. K., & Hui, E. C. (2014). Residential satisfaction of migrant workers in China: a case study of Shenzhen. Habitat International, 42, 193e202. Wallace, C. (2002). Household strategies: their conceptual relevance and analytical scope in social research. Sociology, 36(2), 275e292. Wang, D., & Li, S. M. (2004). Housing preferences in a transitional housing system: the case of Beijing, China. Environment and Planning A, 36, 69e87. Wang, Y. P., Wang, Y., & Wu, J. (2010). Housing migrant workers in rapidly urbanizing regions: a study of the Chinese model in Shenzhen. Housing Studies, 25(1), 83e100. Wang, D., Zhang, W., & Bakhai, A. (2004). Comparison of bayesian model averaging and stepwise methods for model selection in logistic regression. Statistics in Medicine, 23, 3451e3467. Wu, W. (2002). Migrant housing in urban China: choices and constraints. Urban Affairs Review, 38(1), 90e119. Wu, W. (2004). Sources of migrant housing disadvantage in urban China. Environment and Planning A, 36, 1285e1304. Wu, F. (2012). Neighborhood attachment, social participation, and willingness to stay in China's low-income communities. Urban Affairs Review, 48(4), 547e570. Zhang, Y. (2011). Migrant workers' willing of Hukou register and policy choice of China urbanization. Chinese Journal of Population Science, 2, 14e26 (in Chinese). Zhang, Y., & Chen, J. (2014). The changing prevalence of housing overcrowding in post-reform China: the case of Shanghai, 2000-2010. Habitat International, 42, 214e223. Zheng, S., Long, F., Fan, C. C., & Gu, Y. (2009). Urban villages in China: a 2008 survey of migrant settlements in Beijing. Eurasian Geography and Economics, 50(4), 425e446. Zheng, Z., & Yang, G. (2013). Population floating and the trend in China (Zhongguo renkou liudong xianzhuang ji weilai qushi). Retrieved June 26, 2014, from The Chinese Academy of Social Sciences, Institute of Population and Labor Economics Web site http://iple.cass.cn/news/723503.htm. Zhou, J. (2011). Uncertainty and housing tenure choice by household types: evidence from China. China Economic Review, 22, 408e427. Zhu, Y. (2007). China's floating population and their settlement intention in the cities: beyond the Hukou reform. Habitat International, 31, 65e76.