Hukou Disparities and Residential Attainment among Migrant Workers in Urban

By

Renling Zhang

Abstract Extensive research has documented the pervasively less desirable residential outcomes of rural-urban migrants in urban China, but how housing outcomes vary with two main dimensions of hukou status (agricultural versus non-agricultural, local versus non-local) is unclear. Using 2008 RUMiC data, this paper examines the effects of migrants’ hukou status on home ownership, location, crowding and housing quality, by comparing local- municipality, agricultural hukou migrants with two other groups: inter-municipality migrants, and migrants experiencing rural-urban hukou transformation. Results from logistics and OLS regressions suggest three main findings. First, among migrants with agricultural hukou, gaining local hukou status is crucial in gaining homeownership, living near the urban core, enjoying more space and higher quality of housing. Second, migrants experiencing hukou transformation and becoming de jure urban residents have significantly higher housing outcomes than agricultural hukou holders. Third, consistent with prior findings is the importance of socioeconomic and migratory characteristics in determining residential outcomes.

1. Introduction

The demography of Chinese population has been transformed enormously as a result of large-scale rural-urban migration over the past three decades. Since the early 1980s, millions in the rural labor force have left their land and migrated to urban areas within or outside their hometown provinces in search of higher-income jobs, nice homes, and opportunity to improve their lives. The real life in the urban sector is, however, much different from what they were seeking—for most rural migrant workers, urban life is precarious, with low income, highly-segregated occupations, disproportional social benefits, and poor residential conditions. Despite the fact revealed by extensive research that rural migrant workers tend to remain at or near the bottom stratum of the urban society, the number of rural migrant workers attempting to enter the urban industrial and service sectors keeps increasing each year, accounting for over one fifth of the

Chinese population in 2013 (NBS, 2014). While in recent years, this population has

1 made gradual headway in gaining access to better economic opportunities and social benefits, their ultimate incorporation into the urban society in China’s rapid urbanization process is hampered by the fact that the majority of them are not institutionally recognized as urban citizens because of the long-standing hukou

(household registration) system.

Better understanding the comprehensive impacts of China’s hukou system on

Chinese rural-urban migrants is of considerable importance not only because of their sheer numerical presence, but due to their contribution to China’s fast urbanization and economic growth. Moreover, the hukou institution has been standing as the primary impediment to labor mobility and key policy factor in affecting migrants’ position in the urban social . Considerable research has explored how this institutional barrier impedes migrants’ access to job opportunity, healthcare, , children’s . Comparatively lacking is research evaluating the impacts of the hukou system on the residential attainment of migrant workers in urban China. Such research is crucial not only because of the vital role residential location and environment play in structuring migrants’ social life, but also because of the persistent and deep inequalities in residential outcomes between urban residents and rural migrant workers.

Since the outset of China’s transition, researchers have documented the strong association between an urban individual’s socioeconomic status and political privilege, and his/her access to housing of good quality and location (Logan& Bian,

1993). Accompanying the deepening of marketization is the diminishing effect of socialist welfare, and a growing weight of income in the determination of housing

2 attainment. Yet the massive influx of low-skilled rural migrant workers with agricultural and mostly non-local hukou status raises the question of how the institutional constraints imposed by the hukou system deepen the inequality in residential attainment, considering rural migrants’ socioeconomic endowment and migratory characteristics. Answering this question is crucial not only to understand the challenges that Chinese rural migrant workers may face in obtaining desirable housing and residential environment, but also because it adds to the literature regarding citizenship and housing inequality among migrant populations in transitional economies.

My goal in this study is to estimate the impacts of two essential dimensions of hukou status on Chinese rural migrant workers’ residential attainment in the urban sector. To achieve this goal, housing outcomes will be presented and compared both among migrants and between migrant and urban residents groups. In contrast to most of the previous city-specific research, I use the multi-province/city sample of household heads from the 2008 Migrant Household Survey on Rural Urban

(RUMiC) with detailed information on hukou status. In terms of residential attainment, multiple dimensions will be evaluated, including home ownership, household crowding, housing location, and housing quality.

2. Background

2.1 Housing Reform: From Welfare Distribution to Marketization

China’s housing reform started in the late 1980s as an important component of the

3 country’s transition from a planned to market-oriented economy. Before the market reform, provision of urban housing was a part of the socialist welfare system. Under the socialist regime, housing was invested and allocated by the public sectors, such as governmental agencies and state-owned enterprises. It was estimated that before the reform, more than 80 percent of the housing was publicly-owned (Chen et al., 2011).

Since the housing reform, a series of legislations and policies have been promulgated to privatize and reform the state-dominated housing system. The reform progressed slowly due to the continuing debate on the cession of state-owned land and public- sector owned housing. It was not until 1998 that the allocation of welfare housing by the government officially ended and a market-oriented urban housing market was established (Wang, 2011). The marketization of the urban housing market resulted in significant improvements in housing conditions in urban China as indicated by increasing home-ownership ratio, expanding housing space and improving housing qualities over the past two decades (Lee, 2000; Mak et al., 2007; Chen et al., 2011).

On the way to the commodification of housing, the reform underwent several critical modifications. Most studies have focused on two government-initiated approaches to buffer the shock of housing reform. One approach is the government- guided division of urban housing market, with one fully marketized and the other regulated by the government. In 1994, the Chinese government called for two distinct housing systems that suited the “socialist market economy” in the transitional period.

One is the Economic and Affordable Housing Program (EAHP), which is a government housing support program targeting low-to-medium income households in the urban

4 areas. The other is “Commodity Housing” system, which operates under market principles and involves mostly well-off, upper income urbanites, who do not qualify nor need to participate in the EAHP program (Yang and Chen, 2014).

Another important approach is the creation of the Housing Provident Fund (HPF).

Out of concern that many of the urban citizens could not afford to purchase their housing units, the Chinese government established the HPF program, which is similar to housing fund programs in other countries such as Thailand and Singapore, combines a 401(k)-like savings and account with subsidized mortgage rates and price discounts to provide a mechanism through which an employee could save for, and eventually complete, a housing purchase (Buttimer et al., 2004). The program is open to employees of government agencies, state-owned enterprises, and other public and semi-public sectors such as and hospitals. While the part and parcel of the HPF policy is stipulated by the central government, local governments have a big say in determining the specifics of HPF, thus making the qualification criteria for beneficiaries vary across provinces, cities and townships.

Accompanying the housing reform from the previous socialist welfare provision to the commodification of housing, is the change of key determinants affecting housing attainment. Early discussion of socialist housing inequality focused on the creation of privileged classes in achieving housing services. Studies of housing inequality in pre- reform socialist regime reveal that housing is a privilege for the political elite and housing attainment is closely related to positions of authority and to informal ties with the authorities (Szelenyi and Machin, 1987). With the deepening of the housing reform,

5 the importance of political position diminishes. Growing is the significance of and income in determining housing ownership and housing quality among urban residents as evidenced in extensive empirical studies (Huang 2004; Li et al. 2006; Li et al. 2007). Nevertheless, the influence of social position and informal networks to attain better housing remains important and along with new forms of socioeconomic resources, has been creating housing inequality among Chinese urban residents (Logan et al., 1999; Logan and Bian, 1993). As I shall illustrate in the following sections, socioeconomic factors such as income, education and nature of employment, together with the differentiations of citizenship, play an important role in affecting housing outcomes in today’s China, especially among the large numbers of rural-urban migrants.

2.2 Migrant Housing and Sources of Housing Disadvantage

The large numbers of inflow of laborers from rural to urban areas, from inland to coastal provinces and from agricultural into non-agricultural sectors during the last three decades is a prominent feature of China’s market transition. Despite the large number of rural-urban migrants dwelling in the urban sector and the migratory population’s profound influence on China’s urbanization process, their housing conditions have not attracted due attention either in the public policy or the academic sphere. Overcrowded workplace dormitories or highly-segregated migrant enclaves with poorly-equipped kitchen/bathroom facilities are the common depiction found in ethnographic studies of migrant housing (Zhang et al., 2003). While studies of migrant housing in China’s top- tier cities like , , Guangzhou, etc., have revealed a very low home-

6 ownership rate and undesirable housing qualities (Wu, 2002, 2004; Huang, 2004), comparatively lacking is the systematic evaluation of migrant housing attainment and literature on the factors limiting their ability to attain desirable housing.

Existing research has focused mostly on two types of explanations for the sources of disadvantage of migrant housing: stratification factors and migratory factors. The stratification perspective maintains that housing quality and housing ownership is a function of migrants’ relative socioeconomic standing (Li and Yi, 2007; Logan and Bian,

1993; Logan et al., 1999). Migrant workers with greater economic resources and social capital would have access to better housing. This explanation applies to rural-urban migrants as well as urban residents. Migrants are predominantly low-skilled laborers working in the construction, transportation and manufacturing sectors and tend to receive far below required to the level to purchase the commodity housing in the urban sector, and bank mortgages are not available to the low-income class. Their ability to purchase and houses is also hindered by the limited network connections they possess to get access to the urban housing market (Logan et al., 1999).

Further, income differences may be related to social position and bargaining power among those with different education and skill levels as they press for better housing in the housing market. Empirical findings also show that discrimination exists in the secondary housing market, as transactions in what was previously socialist welfare housing are rarely accessible to rural-migrants (Wu, 2004). The stratification perspective has been informative at explaining income and occupation difference in residential outcomes both within the rural migrant group and among urban residents,

7 especially after the housing reform period when housing was largely commodified.

While the stratification perspective has received considerable empirical support, it fails to explain the disparities of housing attainment amongst migrants and urban residents who are at relatively the same income-level. The perspective on the migratory features of migrants was formulated to explain the persistence of resident-migrant difference in housing qualities and housing ownership (Nelson, 1976; Goldstein, 1993;

Solinger, 1995). According to the migratory perspective, efforts to improve housing quality have little impact on the temporary population. This is a result of housing decisions made under different expectations of length of stay in the urban areas.

Migrants tend to make diverse housing choices, invest little of their income to improve housing conditions and demand fewer amenities and services. They prefer to stay in workplace dormitories provided by employers which reduces living expenses and increases savings for remittances. In general, only those who plan to stay in the city are willing to invest in more substantial housing. The migratory perspective is widely supported in the empirical studies of international migration, particular in the cases of undocumented immigrants who may not acquire permanent residence in the destination country (Burgers, 1998).

While these two explanations have been studied extensively in efforts to explain the differences in housing conditions between migrants and non-migrants, largely ignored are the institutional factors that prevent migrants from entering into highly-paid industries and settling down in urban areas. In the Chinese case, the hukou system remains a fundamental institution to regulate citizenship status, which in turn affects

8 individual’s socioeconomic opportunities and migrants’ decision on length of migration.

Moreover, one’s hukou status directly determines his/her eligibility to participate in the aforementioned “Economic and Affordable Housing Program”, Housing Provident

Fund and other housing supporting programs provided by the government, which play a key role in improving migrants’ residential outcomes. In the next section, I’ll review the various dimensions of the hukou system and the constraints it brings to rural-urban migrants.

2.3 Hukou System as an Institutional Barrier to Housing Attainment

In the great majority of countries around the globe, internal migration is not simply physical movement, because migrants, perhaps after a short period of stay, are also granted the right to vote and gain equal access to social welfare at the destination. What is distinctive about the internal migration in China is that the two aspects of migration— movement and citizenship—can be entirely disparate. The separation results from the hukou (household registration) system, which has been effective in controlling China’s population mobility and creating the wide rural-urban gap. A migrant can move to a new place but can be barred access to community membership-based services and welfare. People who have moved to a new place but do not possess local citizenship

(hukou) are referred to as the non-hukou population, meaning that they are not de jure residents even though they are de facto residents. Also, the other dimension of hukou, agricultural versus non-agricultural, also differentiate people’s access to rights and privileges affecting socioeconomic well-being (Treiman, 2012).

9

The hukou system was institutionalized as early as in the socialist era of the 1950s.

Although the system has been relaxed and modified several times, the core classification persists. The system divides the Chinese population into “agricultural” and “non-agricultural” sectors (often treated a little imprecisely as analogous to “rural” and “urban” sectors), with a cross-cutting distinction between “local” and “non-local” registrations. Chinese citizen’s hukou status is assigned at birth, following the mother’s status. With few exceptions, in cases such as being at tertiary or technical schooling, being service, experiencing land expropriation, or assignment to an urban job, it is very hard to change one’s hukou from agricultural to non-agricultural one.

Moreover, local status in non-agricultural (urban) sector is extremely difficult to attain for those with agricultural hukou. Even rural-urban migrants who live in a particular town or city for decades, are unlikely to acquire a non-agricultural hukou or local residence status.

Hukou status was, and to a large extent still is, a strong determinant of rights and benefits and the social privileges differentiated across the two dimensions: agricultural versus non-agricultural, and local versus non-local (four categories of hukou status are thus generated). Empirical studies show that in order to qualify for health and insurance, or , or to enroll one’s children in a public , it is crucial to have both a non-agricultural hukou and local residence status (Chan 1999;

Chan and Buckingham 2008; Wang 2004, 2005). Each dimension of the hukou status is linked to different social benefits. For the agricultural versus non-agricultural dimension, a non-agricultural hukou elevates one’s status, for no matter where this type

10 of hukou holder resides, he/she was automatically entitled to the basic benefits distributed by the government, making it highly desirable and sought after throughout the country. For the local versus non-local dimension, a local regular hukou registration defines one’s rights to pursue many activities and determines eligibility for services in a specific locality. Such services vary from city to city and province to province.

Given the important position of hukou status in accessing social benefits, hukou’s influence on rural-urban migrants housing attainment in urban China are commonly discussed without a coherent combination of both its direct and indirect effects. On the one hand, numerous research found that lacking urban (non-agricultural) and local hukou stands as the barrier for migrants to achieve housing benefits directly, such as getting enrolled in the Commodity Housing program or receiving Housing Provident

Fund which are open to qualified local urban residents only (Wu, 2004). Researchers have also identified the indirect effects of hukou in impeding migrants from achieving better residential outcomes. It is widely reflected that because of the inferior hukou status they hold, migrants are channeled to industries characterized by long hours, poor working conditions, and low pay (Wang and Zuo, 1999). Also, it has been argued that the hukou system, as an institutional barrier, hampers migrants’ integration into mainstream urban society (Yue et al., 2013). These effects of hukou limit the financial options and social connections which are indispensable for migrants to purchase housing and improve housing conditions in the post-housing-reform period. Moreover, the small chance of attaining local urban hukou also helps to explain why circular migration instead of permanent settling-down persist (Chan et al., 1999).

11

Previous studies have provided key insights in understanding the direct and indirect effects of hukou status on rural-urban migrants’ housing attainment. However, few have considered the influence of the long-standing institutions in an integrated way. Further, few have looked at the separate effect of the two dimensions of hukou status, which maybe the key for policy reform. Therefore, in my study, I consider and compare the effects of different dimensions of hukou status on rural-urban migrants’ housing attainment.

Existing residential research has focused almost exclusively on two types residential outcomes: home-ownership and neighborhood poverty level (Hall, 2013).

Both are important domains of residential life, and serve to evaluate specific aspects of housing attainment. Comparatively lacking however, is information on the detailed features of housing quality that Chinese rural-urban migrants may be lacking. In my study, in considering residential outcomes, besides home-ownership and housing location, I also measure home crowding and generate a housing quality index to measure whether various basic amenities are available, including tap water, sanitation facilities, heating fuel, and kitchen facilities.

Based on the findings and limits of previous studies, two hypotheses are postulated:

Hypothesis 1 (H1): For local-municipality hukou holders, rural-urban migrants

have lower-level residential outcomes than urban residents.

Hypothesis 2 (H2): Within the rural-urban migrant group, those who have local-

municipality residence have higher-level residential outcomes than the have-nots.

12

3. Data and Method

3.1 Data

Data sets in most previous studies on Chinese migrants have several limits. The majority of the data sets are not designed for the rural-urban group per se, thus researchers commonly use certain criteria, such as hukou status and the change of residence from last year, to select the floating sample from the total population (Treiman,

2012). In those data sets specifically targeted at the rural-urban migrant group, most are city-specific, thus restricted within certain localities (Wu, 2004). Based on my questions and hypotheses, these types of data sets are not ideal for my study.

In this study, I use data set from the Rural-urban Migration in China (RUMiC) project 1 , which consists of three parts: the Urban Household Survey, the Rural

Household Survey and the Migrant Household Survey. For the purpose of comparing migrants and urban residents as hypothesized in H1, I use the Migrant Household

Survey and Urban Household Survey. The Migrant Survey part of RUMiC data set covers over 5000 migrant households in 15 cities of 9 provinces in both inland and coastal China. It is the only representative random sampling survey of rural-urban migrants in China. In addition, the Urban Household Survey covers over 5000 urban households in the overlapping cities. The survey questionnaire requires individual members in the household to report their housing and living conditions, hukou status,

1 The Longitudinal Survey on Rural Urban Migration in China (RUMiC) consists of three parts: the Urban Household Survey, the Rural Household Survey and the Migrant Household Survey. It was initiated by a group of researchers at the Australian National , the University of Queensland and the Beijing Normal University and was supported by the Institute for the Study of Labor (IZA), which provides the Scientific Use Files. The financial support for RUMiC was obtained from the Australian Research Council, the Australian Agency for International Development (AusAID), the Ford Foundation, IZA and the Chinese Foundation of Social Sciences. 13 work and income related aspects, and basic information about migration such as length of migration, inter-province or not, and current residing city.

The RUMiC is a longitudinal survey covering two time periods: the year 2008 and

2009. Given the fact that the attrition rate is relatively high (64% 2 ) for migrant households between the first and second waves of the survey, I use the data from the first round—2008 surveys, so as to ensure sample size.

To the two hypotheses, which look at different dimensions of hukou, my final sample consists of two separate sets. For Hypothesis 1, in order to compare the effect of local and non-local hukou within the migrant groups, I use the Migrant Household

Survey and restrict my sample on the migrant household heads with agricultural hukou status. This process yields a total of 4976 migrant household heads (Sample I). To test

Hypothesis 2, which compares the disparities of agricultural and non-agricultural hukou, local hukou status should be controlled. I select 943 migrant household heads with local, agricultural hukou status from the Sample I, along with 886 urban household heads with local, non-agricultural hukou status from the Urban Household Survey, then combine the two groups of respondents into a new dataset (Sample II).

It is noteworthy that all of the 886 local urban household heads selected in Sample

II underwent change of hukou type from agricultural to non-agricultural type

(nongzhuanfei) before the interview. The advantage of selecting this group of respondents instead of the entire group of urban household heads with local hukou

2 RUMiC official website reported that the survey does not track returning migrants due to high costs. Between the first and the second wave, partly due to the high mobility and partly due to the global financial crisis, the attrition rate for the Urban Migrant Survey was 64%. In the subsequent waves the attrition rate gradually came down with the second to the third wave attrition rate being 52% and the third to the fourth wave rate being 43%. Website: http://idsc.iza.org/?page=27&id=58 14 status, which include a large proportion of urbanites born with non-agricultural hukou type, is that nongzhuanfei urbanites share more similarities with rural-urban migrants in terms of socioeconomic resources. Therefore, the effect of urban-rural hukou dichotomy on housing outcomes is better focused.

Scrutinizing the age-sex composition of the two samples, I find that both urban and migrant samples are skewed towards male heads, and migrant household heads are generally younger than urban household heads. In sample I, for the entire migrant group, the median age is 28 and 30 percent of the migrant household heads are female. In

Sample II, compared with migrant household heads, urban household heads are older

(median=42). In terms of proportion of female heads, the percentage for migrant and urban resident households are equivalent, with 34 percent and 35 percent respectively.

Based on the findings of the 2013 China Migrant Worker Report and Mallee and

Pielke’s (2014) research, rural-urban migrants tend to be young laborers, therefore explains why migrants are younger than urban household heads in my samples. Rates of female-headship are higher for migrants in my samples, compared with national level

(7% for rural residents as benchmarks for migrants and 29 % for urban residents). As migrants tend to experience spousal separation, as over 40 percent of the female-headed households are single-member households, therefore explains the higher female-head rate.

3.2 Measures

Variables of primary interest are described here. Dependent variables are

15 homeownership, household crowding, location and housing quality. Independent variables include hukou status, and control variables include items related to socioeconomic status, migration factors, and demographic characteristics.

3.2.1 Residential attainment

Home-ownership is assessed on two dimensions on which migrants report whether their housing is self-owned (homeowner=1) or not (homeowner=0). In most cases, migrants do not own the housing, and the other home-ownership scenario includes renting, and workplace dormitory provided by employers.

Household crowding is a continuous variable counting the usable/living areas (in square meters) for each person in the same housing unit. Type of housing unit greatly depends on the type of ownership. If migrants live in a workplace dormitory, then a housing unit is usually a dormitory room. In other cases, housing unit is an or an entire house.

Residential location is measured using a dummy variable. Respondents are classified into living in main districts of the city (location=1) or non-main districts

(location=0). Non-main districts of a municipality include outlying districts of the city and counties, which are under the administration of the municipal government, but are for urban land use. In general, the main district and non-main district dichotomy reflects the hierarchy of living environment to some extent. Living in main districts of a city usually reflects better access to pubic goods and services (Chan and Buckingham, 2008).

Housing quality is assessed through an index, which sums up the score of five items

16 of housing amenities. If the housing unit has tap water and heating facilities then the scores for this two items are 1 (tapwater=1, heating=1), 0 otherwise. For Sanitations, if the housing unit has both bathroom and toilets, the score is 2. If only toilet is equipped, the score is 1, 0 otherwise. For fuels, natural gas or piped gas is assigned the highest score of 2, bottled gas is 1, and 0 otherwise. Similarly for kitchen facilities, if the respondent report that the unit has a private kitchen, then 2 points are assigned, shared kitchen with other units gets 1 point and 0 if none. The housing quality index sums up the five items and the higher the index, the better the housing quality attained.

3.2.2 Hukou status

Based on the aforementioned two dimensions of Hukou status, I consider both the agricultural versus non-agricultural, local versus non-local city/county dichotomies. In the comparison between urban residents and rural-urban migrants, I only consider those with local municipality residence. Therefore the dichotomy is measured by a dummy variable (urban=1 if non-agricultural hukou; urban=0 if agricultural hukou).

According to the definition stipulated by the National Bureau of Statics of China

(2013), only those with agricultural hukou are counted as rural-urban migrants. As a result, for the comparison within the rural migrant group, I hope to see the difference in housing outcomes generated by the dichotomy of local and non-local municipality residency. The difference is measured by a dummy variable (local=1 if local municipality hukou; local=0 if non-local municipality hukou) as well.

3.2.3 Control variables

3.2.3.1 SES variables

17

Household income per capita is set to measure the total income of the household, which encompasses labor, family business, asset, and transfer income, shared by members in the same household. The total number of family income last year is reported by the household head and the variable is treated as a continuous variable. In a gradually market-oriented housing market, the effect of household income is estimated to be positively associated with housing attainment.

To assess Tenure of job, respondents are classified into contract workers

(contract=1; otherwise 0), self-employed workers (self-employed=1; otherwise 0) and precarious workers (reference group). Rural-urban migrants have a relatively low rate of signing labor contracts with employers, with only 41% reported signing either one- year labor contract or non-fixed term contract in 2013 (NBS 2014). Therefore, a contracted laborer has higher likelihood to have a securely and decently paid work.

Self-employed migrants are also more likely than non-contract precarious workers to have better earnings.

Education is measured in terms of years of schooling. Educated migrants would be more likely to have access to housing resources and information, and less likely to endure ragged living conditions.

3.2.3.2 Migratory variables

Length of migration is assessed in terms of the number of months the migrant has lived away from his/her hometown due to work or business. The effects of length of migration has been mixed in previous research. On the one hand, new-comers may possess fewer resources in the housing market and be more resilient regarding poor

18 housing quality, while those who stay in the city for a relatively long period of time generally invest more in improving housing conditions or purchasing (Huang,

2014). On the other hand, ethnographic studies on migrant enclaves in some metropolises indicate that length of migration has no direct correlation with residential attainment (Zhang et al., 2003).

Current Region is also a key factor that should be considered in housing attainment, especially for ownership. China has not only experienced large rural-urban gap in the last three decades, but also growing disparity between its inland and coastal provinces

(Kanbur, 1999). China’s coastal metropolises saw soaring housing prices and mushrooming population growth, while inland cities remain comparatively less developed, thus rendering housing more attainable for both rural and non-rural migrants

(Chen et al., 2011). Therefore I set a dummy variable to assess the different effects of living in cities in inland and coastal provinces (Coastal=1 if city is Guangzhou,

Dongguan, Shenzhen, Shanghai, Nanjing, Wuxi, Hangzhou, Ningbo; Coastal=0 if city is Zhengzhou, Luoyang, Hefei, Bengbu, Chongqing, Wuhan, Chengdu).

Intention to Move is an indicator to assess migrant individual’s expectations to stay in the current city and is measured on a five-point scale on which respondents report their likelihood to move in the next twelve months as (1) Highly likely (2) Likely

(3) Not sure (4) Unlikely (5) Highly unlikely. For the analysis, I have created two dummy variables. I categorize highly likely and likely as “likely” to move

(Intentionmove=1; 0 otherwise), and unlikely and highly unlikely as “unlikely” to move

(Intentionnot=1; 0 otherwise), with those indicated “not sure” as reference group. It is

19 worth noticing that once a migrant becomes a homeowner, he/she may decide not to move, so intention may be predicted by homeownership as much as being predictive of it. Likewise for the other dimensions of housing attainment. In my study, I acknowledge the endogeneity problem and only consider the correlation between housing attainment and intention to move.

3.2.3.3 Demographic variables

I also incorporate demographic factors of the household heads, consisting of age, age-squared, gender (female-headed vs male-headed), marital status (married vs not married) and the presence of children (having at least one child present vs none) in the current household, as control variables in my models.

3.3 Statistical Analysis

To examine migrants’ residential attainment, I estimate OLS and logistic models that regress each of the four dimensions of housing attainment on hukou status and other explanatory variables. For the two hypotheses, I use separate samples of household heads. To test Hypothesis 1, I make cross group comparisons of migrants and urban residents, with individuals in both groups holding local municipality hukou, to highlight the difference of agricultural and non-agricultural-hukou on housing attainment. To test Hypothesis 2, the sample is restrained within the rural migrant group.

For the agricultural hukou holders, I show the extent to which cross-locality migration and the corresponding disconnection of movement and citizenship would affect rural migrants’ housing attainment.

20

4. Results

4.1 Descriptive statistics

I begin my analysis by presenting the descriptive statistics of housing attainment, hukou status and other explanatory variables for rural-urban migrants and urban residents in my samples. Due to the survey design, several questions available in the housing module of Migrant Household Survey are not found in the Urban Household

Survey. As a result, for urban residents, only housing location and kitchen facilities are presented, which are items included to depict housing outcomes for migrants as well.

The two common patterns gleaned from the comparison of housing outcomes between migrants and urban residents are shown in Table 1: First, urban residents live in better equipped housing than migrants, as shown by the score for kitchen. In addition, urban residents live primarily in cities (72 percent) as compared with local hukou rural-urban migrants, among which only 41 percent resides in the main districts of cities. These findings are consistent with prior research that migrants tend to live in less-well equipped housing units and live in outlying districts or counties of a municipality.

Contributing to the lack of examination of housing outcomes within migrant group, I find that rural migrants with local hukou status have higher rates of homeownership, more housing space per person, larger percentage living in urban main districts compared with migrants without local hukou.

For the demographic characteristics of migrant household heads in my sample, most of them are young laborers, with over 60 percent in the youthful age group

21

between 16 and 34. One third are females and over half are married. Children’s presence

in the migrant household is rare, accounting for less than 6 percent. For urban residents,

the average age is higher than that of migrants and the majority of urbanites in my

sample are married.

Table 1 Descriptive Statistics

Migrant ( total ) Migrants ( non-local ) Migrants ( local ) Urban Resident (total) Mean (Std Dev) / % Home-owner 3% 2% 6% - Household crowding ( Square Meters/ person) 3.5 (2.8) 10.9 (10.8) 12.6 (11.9) - Location(city-region) 31% 29% 41% 72% Housing quality (scores) 3.7 (1.9) 3.7 (1.9) 3.6 (2.0) - Tap water 0.93 (0.2) 0.93 (0.2) 0.95 (0.2) - Heating 0.19 (0.4) 0.20 (0.4) 0.17 (0.4) - Sanitations 1.03 (0.9) 1.06 (0.9) 0.83 (0.8) - Fuels 0.91(0.7) 0.88(0.7) 0.93 (0.8) - Kitchen 0.62 (0.9) 0.62 (0.9) 0.65(0.9) 1.9 (0.2)

Hukou agricultural, non-local city/county 81% 100% -- agricultural, local city/county 19% - 100% - non-agricultural, non-local city/county - - - - non-agricultural, local city/county - - - 100%

Household income (Yuan) 2187 (1934) 2258 (1953) 1861 (1831) 59157 (44977) Household income/capita (Yuan) 1414 (1110) 1467 (1058) 1194 (1296) 27329 (37861) Tenure of job Contract laborer 50% 52% 42% 55% Self-employed 20% 20% 21% 27% Precarious 29% 27% 37% 23% Education (years of schooling) 9.2 (2.6) 9.2 (2.6) 9.1 (2.4) 10.5 (4.2)

Length of migration (number of months) 10.9 (2.3) 10.9 (2.2) 10.9 (2.4) - Coastal region 54% 64% 12% 48% Intention to move Likely to move 19% 21% 14% - Unlikely to move 50% 50% 49% -

Age 31 (10.2) 30 (10.1) 31 (10.5) 41 (9.8) Female-headed 31% 30% 35% 34% Married 54% 54% 56% 88% Children-present 6% 6% 7% -

N 4976 4033 943 886 The average length of migration for rural-urban migrants is over 10 months. They

tend to migrate to the economically dynamic coastal provinces. Specifically, local

hukou migrants are mostly staying in their original inland provinces as only 12 percent

22 of the migrants with local hukou are currently in coastal provinces. Fewer than 20 percent of them report the intention of leaving the current working site in the coming twelve months. Non-local hukou migrants are 50 percent more likely to show the intention to move than locals. In terms of socioeconomic status, migrants have far less total and per capita household income, lower education level and a slightly higher proportion engaging in precarious occupations, as compared with urban residents.

4.2 Multi-level regressions

4.2.1 “Locals vs Strangers”: within migrant group variations in housing attainment

In this section, I focus on the migrant group, looking at the residential outcomes brought by the dichotomy of local and non-local city/ county hukou. Four dimensions of housing attainment: home-ownership, household crowding, housing location and housing quality are presented in separate tables (Table 2- Table 5), with each dimension regressed in three models.

Baseline models examines overall differences of local and non-local dichotomy in residential attainment. Substantial and statistically significant local versus non-local differences in residential attainment persist in terms of all four dimensions: home ownership, location, household crowding and housing quality. Specifically, the odds of migrants with a local city/county hukou to own a housing unit are 2.86 (e1.05) times greater than have-nots. Similarly, as is shown in Table 3, migrants with local hukou have a higher odds of 1.70 (e0.53) times than those with hukou attached to a different municipality in living in the city rather than suburban areas. As is shown in Table 4, the 23 average level of usable area per person is 1.74 square meters higher for local migrants than non-locals. Finally as is shown in Table 5, those with local hukou live in higher quality housing than cross-city/county migrants, with the magnitude of decrease in housing quality index as small as 0.12 points.

Table 2 Logistic regression models of home-ownership within migrant group Home-ownership Baseline Model 2 Model 3 B SE B SE B SE Hukou-local city/county 1.05 *** 0.17 1.09 *** 0.20 1.08 *** 0.20

Household income/capita -0.22 0.15 0.04 0.16

Tenure of job (ref=precarious) Contract laborer 0.35 0.28 0.39 0.29 Self-employed 1.54 *** 0.27 1.10 *** 0.28

Education (years of schooling) 0.10 ** 0.04 0.14 *** 0.04

Length of migration 0.24 ** 0.09 0.20 * 0.09

Coastal region 0.03 0.22 -0.06 0.22

Intention to move (ref=not sure) Likely to move -0.35 0.52 -0.23 0.52 Unlikely to move 1.46 *** 0.27 1.40 * 0.27

Age (in years) -0.01 0.06

Age squared 0.00 0.00

Female-headed 0.28 0.19

Married 1.15 ** 0.35

Children present 0.95 *** 0.23

Constant -3.78 *** 0.11 -7.55 *** 1.48 -10.34 *** 1.86

Pseudo R square 0.03 0.15 0.19

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

Model 2 in Table 2-5 provides a crucial test of the stratification and migratory perspectives by adding controls for socioeconomic and migratory factors. In socioeconomic factors, both self-employed work and education exert significant positive effects on housing attainment: the odds of owning a housing unit and living in

24 cities are larger for self-employed migrants (multiplicative factors: e1.54=4.66; e1.14=3.13) and grow with each additional year of education (multiplicative factors: e0.10=1.11; e0.05=1.05). Self-employed migrants have on average 2.77 square meters of larger spaces and 1.09 scores higher in housing quality than contract and precarious migrant workers. Additional years of schooling have similar effects in improving housing quality and attenuate household crowding.

Table 3 Logistic regression models of predicting location within cities for migrants Location Baseline Model 2 Model 3 B SE B SE B SE Hukou-local city/county 0.53 *** 0.07 0.42 *** 0.09 0.40 *** 0.09

Household income/capita 0.09 0.06 0.13 * 0.07

Tenure of job (ref=precarious) Contract laborer -0.35 *** 0.08 -0.34 *** 0.08 Self-employed 1.14 *** 0.09 1.05 *** 0.10

Education (years of schooling) 0.05 *** 0.01 0.06 *** 0.01

Length of migration 0.09 *** 0.02 0.08 *** 0.02

Coastal region -0.26 ** 0.08 -0.31 *** 0.08

Intention to move (ref=not sure) Likely to move -0.30 ** 0.10 -0.25 * 0.10 Unlikely to move 0.00 0.07 -0.01 0.08

Age (in years) 0.07 ** 0.02

Age squared 0.00 ** 0.00

Female-headed 0.47 *** 0.07

Married 0.23 0.10

Children present 0.24 0.13

Constant -0.90 *** 0.03 -2.94 *** 0.48 -4.49 *** 0.60

Pseudo R Square 0.01 0.08 0.10 Notes: *p<.05, **p<.01, ***p<.001

As for migratory factors, the time length of migration for migrants play a statistically significant role in increasing migrants’ odds of home-ownership and living in cities. Housing quality improves with one additional month of migration, though the

25 magnitude of the effects is not very large. Whether the current provinces where migrants reside is coastal or not, exerts great effect in location and crowding. Living in

China’s coastal instead of inland provinces makes the odds of living in a city 0.77 (e-

0.26) times smaller. Housing space is also reduced by 3.37 square meters on average for those in coastal provinces. Lastly, intention to move or stay shows various effects in statistical significance in the three dimensions of housing attainment besides crowding.

A clear intention to move reduces the odds of living in main districts by a multiplicative factor of 0.74 (e-0.30) and by 0.39 scores in housing quality index, therefore showing a negative correlation of intention to leave and housing location and quality. Those who exhibit clear intention to stay are more likely to own a housing unit and have 0.25 more scores in housing quality than otherwise.

26

Table 4 OLS regression models of household crowding within migrant group Crowding ( Square meters per person ) Baseline Model 2 Model 3 B SE B SE B SE Hukou-local city/county 1.74 * 0.40 0.71 * 0.43 0.61 0.43

Household income/capita 3.48 *** 0.31 3.55 *** 0.32

Tenure of job (ref=precarious) Contract laborer -0.22 0.37 -0.13 0.37 Self-employed 2.77 *** 0.46 2.97 *** 0.48

Education (years of schooling) 0.45 *** 0.06 0.44 *** 0.06

Length of migration 0.08 0.07 0.08 0.07

Coastal region -3.37 *** 0.36 -3.50 *** 0.36

Intention to move (ref=not sure) Likely to move -0.98 0.45 -0.90 * 0.45 Unlikely to move 0.52 0.35 0.58 0.35

Age (in years) 0.04 0.11

Age squared 0.00 0.00

Female-headed 1.29 *** 0.34

Married -0.98 0.48

Children present 0.44 0.68

Constant 10.88 0.17 -17.29 *** 2.25 -18.85 *** 2.71

R Square 0.04 0.07 0.07 Notes: *p<.05, **p<.01, ***p<.001

Including socioeconomic and migratory factors in the regression models also result in attenuation of the local hukou coefficient in predicting crowding and housing quality.

Local hukou coefficient in Model 2 is less than half the size in Model 1 for crowding.

Also, local hukou coefficient becomes statistically nonsignificant from Model 1 to

Model 2 in measuring housing quality. Thus in these data, it appears that the lower level housing space and quality, and less chance of living in cities experienced by non-local hukou migrants, relative to that experienced by local migrants, is largely attributable to non-local hukou migrants’ socioeconomic standing and migratory characteristics. In contrast, in the model for home-ownership, the coefficient for local hukou increases, 27 indicating the still robust effect of the local versus non-local dichotomy in achieving home-ownerships. Moreover, although log odds of household location in cities decreases from 0.53 in the baseline model to 0.42 in Model 2, local hukou type is significant as in previous models.

Moreover, results in Model 3 show that controls for age, age squared, female- headed, marital status and the presence of children, coefficient for local hukou is still statistically significant in predicting migrants’ ownership and location of housing, despite the disappearance of statistical significance in predicting crowding and housing quality. A closer examination at Model 3 for these two dimensions of residential attainment shows that female-heads exerts much more outstanding effects than other demographic factors. A household headed by female would have on average 1.29 square meters larger housing space and 0.53 points higher in housing quality than those headed by males. A possible explanation is that female-headed households are more likely to have double income sources both from both spouses, therefore are more capable of purchasing or renting larger and more comfortable housing units.

28

Table 5 OLS regression models of housing quality within migrant group

Housing quality Baseline Model 2 Model 3 B SE B SE B SE Hukou-local city/county 0.12 * 0.07 0.00 0.07 -0.03 0.07

Household income/capita 0.28 *** 0.05 0.36 *** 0.05

Tenure of job (ref=precarious) Contract laborer 0.24 *** 0.06 0.26 *** 0.06 Self-employed 1.09 *** 0.08 0.97 *** 0.08

Education (years of schooling) 0.11 *** 0.01 0.11 *** 0.01

Length of migration 0.06 *** 0.01 0.05 *** 0.01

Coastal region 0.08 0.06 0.04 0.06

Intention to move (ref=not sure) Likely to move -0.39 *** 0.08 -0.35 *** 0.08 Unlikely to move 0.25 *** 0.06 0.25 *** 0.06

Age (in years) 0.01 0.02

Age squared 0.00 0.00

Female-headed 0.53 *** 0.06

Married 0.31 *** 0.08

Children present 0.59 *** 0.12

Constant 3.72 *** 0.03 -0.39 0.39 -1.21 ** 0.46

R Square 0.01 0.10 0.12

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

Overall, these results highlight the fact that, local city/county hukou migrants have a substantially higher level of housing attainment than those without this type of hukou status. While the differences in some of the dimensions of housing attainment can be explained by the two groups differentiations in socio-economic standings, migratory patterns and demographic characteristics, local hukou type has a statistically significant effect, which is not eliminated with controls fo SES, migratory and demographic factors, in determining the attainment of housing ownership and location. Thus, there is support for Hypothesis 1 regarding the effect of the dichotomy of local versus non-local hukou dimension of hukou system in affecting migrants’ residential circumstances. 29

4.2.2 Urban vs rural hukou: Cross-group variations in housing attainment

The previous models presented the variations of residential attainment explained by the dichotomy of local versus non-local hukou type within the migrant groups, and demonstrated the fact that local city/ county migrants tend to have better housing outcomes. Nevertheless, for local migrants moving from rural to urban areas within their original city/county, how the maintenance of rural hukou, which is the other important dimension of hukou status, facilitates or impedes them to achieve better housing outcomes is not explored. In this section, I examine the urban-rural hukou difference in housing, focusing on local-city/county-hukou-typed migrants and urbanites. As aforementioned, local urban residents in Sample II for this analysis were not born with urban hukou status, but experienced hukou transformation from agricultural to non-agricultural type, making the comparison more focused on the effects of the dichotomy of hukou for the two groups.

Due to the limited number of questions concerning housing attainment in the Urban

Household Survey, data on home-ownership, crowding, and housing quality are not available. Instead, information on installation of private kitchen in a housing unit and housing location is provided and can serve as important dimensions of urban housing attainment (Logan & Bian, 1993). Therefore, in this section, the dependent variables are two dummy variables: 1) whether the housing unit has a private kitchen (if Yes, kitchen=1; otherwise 0) and 2) whether the housing locates in cities (if Yes, location=1; otherwise 0). Table 6 and 7 present a series of logistic regression models designed to assess the group differences of urban-rural hukou in the likelihood of having private

30 kitchen at home and living in cities.

Table 6 Logistic regression models of private kitchen between migrant and urban resident groups Private Kitchen Baseline Model 2 Model 3 B SE B SE B SE Hukou- non-agricultural 3.22 *** 0.30 2.61 *** 0.47 2.31 *** 0.54

Log Household income 0.93 *** 0.13 0.95 *** 0.15

Tenure of job (ref=precarious) Contract laborer 0.19 0.18 0.21 0.19 Self-employed 0.97 *** 0.21 0.97 *** 0.22

Education (years of schooling) 0.10 ** 0.03 0.13 *** 0.03

Coastal Region 0.26 0.22 0.19 0.23

Age (in years) 0.04 0.04

Age squared 0.00 0.00

Female-headed 0.71 *** 0.17

Married 0.04 0.25

Constant -0.93 *** 0.07 -9.14 *** 0.99 -10.67 *** 1.27

Pseudo R Square 0.48 0.54 0.55 Notes: *p<.05, **p<.01, ***p<.001

Baseline models show the effects of the urban-rural hukou dichotomy in the likelihood of possessing private kitchen in a housing unit and living in urban areas. The statistically significant and substantially positive logit coefficients (3.22 and 1.30) in

Table 6 and 7 indicate that for respondents with hukou status matched to local city/county, those with urban hukou are much more likely to own a private kitchen and live in cities. For the urban residents, the odds of having housing units facilitated with private kitchen is 25 (e3.22) times greater than rural-hukou-typed migrants, while the odds of living in main district of the city is 3.7 (e1.30) times higher.

In Model 2, regional variable and SES characteristics consisting of household income (log transformed to control wide distribution), tenure of job and education are

31 added. Since migratory factors. Including the dummy variable of coastal provinces and factors of socioeconomic status reduces the logit coefficient for urban hukou in the model for private kitchen, indicating that some of the urban and non-urban difference in this dimension of housing can be explained by geographical variations of socioeconomic resources respondents have. Also, adding these control variables weakens the effect of hukou group difference in housing location. A closer examination of the logit coefficients of the SES and geographical variables shows that household income plays a major and positive role in determining the installation of private kitchen and housing location, which is in line with prior empirical findings on the positive association of residents’ income with housing amenities and location in urban housing market (Chen et al. 2011). Consistent with findings from the previous section, self- employed migrants and urban residents are more likely to have private kitchen and live in main districts in cities (log coefficients b=0.97 and 0.57 respectively). In addition, the odds of living in the main district of a city is 34% (odds ratio=0.66) for respondents residing in cities in coastal rather than inland provinces, which applies both for the rural-urban migrant group and the stratum of nongzhuanfei urban residents selected in my sample.

Taking respondents’ demographic characteristics into consideration in Model 3, the hukou group difference remains statistically significant in the installation of private kitchen in the housing unit, while the group difference becomes less observable in explaining housing location disparities. In accordance with my findings within the migrant group in the last section, female household heads have a higher likelihood of

32 having better household facilities, in this model private kitchen than male household heads, while female-headed households have greater odds of living in cities.

Table 7 Logistic regression models of household location within main districts of a city among migrant and urban resident groups

Location Baseline Model 2 Model 3 B SE B SE B SE Hukou- non-agricultural 1.30 *** 0.100.05 *0.30-0.560.32

Log Household income 0.46 *** 0.090.56 *** 0.10

Tenure of job (ref=precarious) Contract laborer -0.30 ** 0.12-0.100.12 Self-employed 0.57 ** 0.170.71 *** 0.18

Education (years of schooling)0.000.020.020.02

Coastal Region -0.41 ** 0.14-0.42 ** 0.14

Age (in years) 0.000.03

Age squared 0.000.00

Female-headed 0.23 *0.11

Married -0.36 *0.17

Constant-0.36 *** 0.07-3.63 *** 0.62-4.900.81

Pseudo R Square0.070.100.12

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

In all, consistent with Hypothesis 2, group difference of local city/county rural migrants and urban residents in housing attainment persist in terms amenities of private kitchen and housing location. Although both types of hukou holders are deemed local citizens of a city/county, attaining a non-agricultural hukou status increases the likelihood of improving standard of housing amenities and approaching city cores.

Besides, the hukou dichotomy, socioeconomic and regional factors also help explain the variations of housing attainment. While for the results for housing location, regional disparities and respondents’ socioeconomic resources diminish the hukou group

33 difference, housing standards regarding installation of private kitchen is still largely explained by the urban-rural hukou dichotomy for local migrants and urban residents.

These results not only highlight the significant disadvantage in housing faced by agricultural hukou holders migrating in urban areas, but also suggest that household income, self-employment, and regional disparities all play important roles in determining housing attainment in today’s urban China.

5. Conclusion and Discussion

Previous research has shown pervasive effects of hukou status on many aspects of well- being for Chinese rural-urban migrant workers, including employment in low-wage occupations, absence of social security benefits, and poor educational attainment for their children (Treiman, 2012; Chan et al., 1999). Housing and residential attainment not only are additional dimensions of well-being, they also have profound consequences for migrants’ social and economic integration (Li, 2006; Wang & Zuo,

1999) into urban society. Unlike previous research suggesting the socioeconomic disadvantage and migratory nature that affect rural migrants’ residential outcomes in urban areas (Wu, 2002; Wu, 2004), I focus on the relationship of citizenship attainment with housing and residential conditions by employing the strategy of comparing housing outcomes for migrants and urban residents with different types of hukou status.

The institutional constraints of hukou system on migrants’ housing has rarely been discussed by analyzing the separate effects of two different dimensions of hukou criteria: local versus non-local, and agricultural versus non-agricultural, I classify my samples of migrant and urban resident household heads from the 2008 Rural Urban Migration 34 in China datasets into sub-groups. Specifically, I assess the within-migrant-group differences in home ownership, household crowding, location and housing quality that result from local versus non-local hukou disparities, and evaluate the variations of private kitchen installation and housing location across groups of local city/county urban residents and rural migrants. To overcome the limitations of city or region specific studies, the data gathered from the representative random sampling survey across 15 cities over 9 provinces in China also provides me with the opportunity to examine the question from a quasi-national level.

My findings confirm the existence of persistent housing disadvantages among rural-urban migrants and the significant differences of housing attainment influenced by the hierarchy of hukou status. First and foremost, variations in several dimensions of housing attainment lie within the rural migrant group. Specifically, I find that gaining a local hukou status is critical for the rural hukou migrants to become home owners, live in the main districts in cities, enjoy more spaces in their housing units, and improve their housing quality in terms of amenities including tap water, heating, sanitation facilities, fuels, and kitchen. Certainly part of the explanations for why non-local migrants are less likely to own a housing unit is that lacking local city/county hukou restricts accessibility and eligibility to benefits of housing programs such as the

Housing Provident Fund and Economic and Affordable Housing Program, which are provided by the and targeted only at the lower-income locally registered residents. This is consistent with the argument that the hukou system as an impediment to population mobility works hand in hand with the highly segregated

35 regional administrative systems (Chan, 2010). In terms of living adjacent to the urban core, my findings are consistent with the fact that intra-city/province migrants have less access to enter urban housing market. Because in some urban housing secondary market, transactions of housing are limited to residents with local type hukou (Wu, 2004).

Moreover, lacking local hukou impedes access to local mortgage services, and makes making non-local migrants reluctant to interact with banks and other financial institutions involved in the process of buying a home, renting housing units with larger space, or improving housing qualities (Chan, 1999).

One of the other main findings of my study is that even with a local hukou type, migrants with rural type of hukou experience significant disparities in housing and residential attainment compared with those holding a non-agricultural hukou. As coded for this study, urban residents are actually prior rural hukou citizens, who experienced rural-to-urban hukou transformation before the interview and became urban hukou holders for various reasons. The relative homogeneity of characteristics other than hukou status for individuals in the two groups sharpens the effects of hukou distinction on housing attainment. I find that rural migrants live in less desirable conditions than their urban counterparts, when the likelihood of having a kitchen is considered. This is likely because lacking urban hukou status channels migrants into highly segregated industries where living in employer-provided workplace dormitories is not an option but a rule. According to nation-level statistics (NBS, 2014), the manufacturing industry attracts the highest percentage of rural-urban migrants compared with construction, transportation and service industries, while over half of the labor force in manufacturing

36 holds agricultural rather than non-agricultural hukou status. Most of the manufacturing companies locate in suburban industrial clusters and impose a strict militant factory regime on their employees. Migrants are found, in many cases, living in crowded dormitories with no sanitation amenities, and certainly without a private kitchen (Pun

& Chen, 2012; Wu, 2004). This also helps to explain why rural migrants are less likely to live in the main district of a city than urban residents, though the group difference is less pronounced. When controlling for socioeconomic factors, the effect of the rural versus urban hukou diminishes, both in terms of magnitude and statistical significance, indicating that in deciding housing location, factors other than hukou also play an important role.

My findings inform both the stratification and the migratory perspectives in explaining the housing disadvantage and inequality experienced by migrant workers.

In line with prior findings on the inequalities in access to community resources in China,

I find that both the household income and the educational level of the household heads are important indicators of housing attainment, both for migrants and urban residents.

What remains less discussed in previous literatures is the effect of self-employment on housing outcomes. In my study, I find that self-employed migrants tend to have better housing outcomes on all of the four dependent variables. A possible explanation is that

China’s market reform provides more business opportunities for those who are not eligible for contracted occupations because of their hukou status, and this groups of entrepreneurial migrants possess certain characteristics that allow them to achieve better housing outcomes than their peers. Future studies can specifically test this

37 argument. Besides socioeconomic factors, long-time migrants tend to have better housing attainment in terms of ownership, location and housing quality. Also in line with previous research is that intention to stay is highly correlated with home- ownership. Nevertheless, I acknowledge that homeownership can both be the result as well as the cause for the intention to stay in urban areas, as housing purchase is a way of hukou transformation in some regions in China (Treiman, 2012). Lastly, China’s regional disparity sharpens the variations in housing outcomes both for migrants and urban residents. Living in cities in the developed coastal regions shrinks the chance of residing in the high-priced main districts, and limits the improvement of housing quality for migrants and the stratum of urban residents in my analysis.

Thus, the research presented here not only provides a first assessment of the housing predicament rural-urban migrants confront within the urban sector, but also illuminates understandings of the relationship between citizenship and housing outcomes. Yet this research represents only a first step in developing a full understanding of how China’s long-standing hukou system maintains and reinforces housing inequality, leaving open a number of important issues for future research. For example, this paper did not explore the process of hukou transformation that occurred for the urban residents selected in Sample II. Future investigators should compare those who transformed their hukou status purely due to exogenous factors such as government land expropriation with those who transformed their status in other ways.

In this way, the effect of rural versus urban hukou dichotomy on housing attainment could be more fully understood. Also, future studies should include direct indicators of

38 accessibility of housing, such as whether respondents are receiving housing provident fund or are involved in Economic and Affordable Housing Programs supported by the government in order to show a direct link of hukou and housing benefits. More of these kinds of considerations would bolster our understanding of the institutional forces of hukou system in shaping the land scape of housing opportunities in transitional China.

39

References

Burgers, Jack. "Formal determinants of informal arrangements: Housing and undocumented immigrants in Rotterdam." Journal of ethnic and migration studies 24.2 (1998): 295-312.

Buttimer Jr, Richard J., Anthony Yanxiang Gu, and Tyler T. Yang. "The Chinese housing provident fund." International Real Estate Review 7.1 (2004): 1-30.

Chan, Kam Wing. "Fundamentals of China's urbanization and policy." China Review (2010): 63-93.

Chan, Kam Wing, Ta Liu, and Yunyan Yang. "Hukou and non-hukou migrations in China: comparisons and contrasts." International Journal of Population Geography 5.6 (1999): 425.

Chan, Kam Wing, and Will Buckingham. "Is China abolishing the hukou system?." The China Quarterly 195 (2008): 582-606.

Chen, Junhua, Fei Guo, and Ying Wu. "One decade of urban housing reform in China: urban housing price dynamics and the role of migration and urbanization, 1995–2005." Habitat International 35.1 (2011): 1-8.

Goldstein, Sidney. "The impact of temporary migration on urban places: Thailand and China as case studies." (1993): 199-219.

Hall, Matthew, and Emily Greenman. "Housing and neighborhood quality among undocumented Mexican and Central American immigrants." Social science research 42.6 (2013): 1712-1725.

He, Shenjing, et al. "Social groups and housing differentiation in China's urban villages: An institutional interpretation." Housing Studies 25.5 (2010): 671-691.

Huang, Youqing. "Housing markets, governmental behaviors, and housing choice: a case study of three cities in China." Environment and Planning 36.(2004)45-68

Huang, Xu, et al. "Residential mobility in China: home ownership among rural–urban migrants after reform of the hukou registration system." Journal of Housing and the Built Environment 29.4 (2014): 615-636.

Kanbur, Ravi, and Xiaobo Zhang. "Which regional inequality? The evolution of rural–urban and inland–coastal inequality in China from 1983 to 1995." Journal of comparative economics 27.4 (1999): 686-701.

Lee, James. "From welfare housing to home ownership: the dilemma of China's housing reform." Housing Studies 15.1 (2000): 61-76.

Li, Bingqin. "Floating population or urban citizens? Status, social provision and circumstances of

40

rural–urban migrants in China." Social policy & administration40.2 (2006): 174-195.

Li, Si-Ming, and Youqin Huang. "Urban housing in China: Market transition, housing mobility and neighbourhood change." Housing Studies 21.5 (2006): 613-623.

Li, Si-ming, and Zheng Yi. "The road to homeownership under market transition Beijing, 1980- 2001." Urban Affairs Review 42.3 (2007): 342-368.

Liang, Zai. "The age of migration in China." Population and development review 27.3 (2001): 499- 524.

Logan, John R., and Yanjie Bian. "Inequalities in access to community resources in a Chinese city." Social Forces 72.2 (1993): 555-576.

Logan, John R., Yanjie Bian, and Fuqin Bian. "Housing inequality in urban China in the 1990s." International Journal of Urban and Regional Research 23.1 (1999): 7-25.

Logan, John R., Yiping Fang, and Zhanxin Zhang. "Access to housing in urban China." International Journal of Urban and Regional Research 33.4 (2009): 914-935.

Mak, Stephen WK, Lennon HT Choy, and Winky KO Ho. "Privatization, housing conditions and affordability in the People's Republic of China." Habitat International 31.2 (2007): 177-192.

Mallee, Hein, and Frank N. Pieke. Internal and international migration: Chinese perspectives. Routledge, 2014. P.76

Massey, Douglas S. "Ethnic residential segregation: a theoretical synthesis and empirical review." Sociology and social research 69.3 (1985): 315-50.

Myers, Dowell, and Seong Woo Lee. "Immigrant trajectories into homeownership: a temporal analysis of residential assimilation." International Migration Review (1998): 593-625.

National Bureau of Statistics, National 2013 Report of Rural-urban Migrant Laborer in China http://www.stats.gov.cn/tjsj/zxfb/201405/t20140512_551585.html (2014) (Chinese)

National Bureau of Statistics, China 2015 One Percent Population Survey http://www.stats.gov.cn/tjsj/ndsj/renkou/2005/renkou.htm (2006) (Chinese)

Nelson, Joan M. "Sojourners versus new urbanites: causes and consequences of temporary versus permanent cityward migration in developing countries." Economic Development and Cultural Change (1976): 721-757.

Pun, Ngai and Jenny Chan. "Global capital, the state, and Chinese workers the Foxconn experience." Modern China 38.4 (2012): 383-410.

41

Solinger, Dorothy J. "China's Urban Transients in the Transition from Socialism and the Collapse of the Communist" Urban Public Goods Regime"." Comparative Politics (1995): 127-146.

Szelenyi, Ivan. "Housing inequalities and occupational segregation in state socialist cities: Commentary to the special issue of IJURR on East European cities." International Journal of Urban and Regional Research 11.1 (1987): 1-8.

Treiman, Donald J. "The “difference between heaven and earth”: Urban–rural disparities in well- being in China." Research in and Mobility 30.1 (2012): 33-47.

Wang, Feng, and Xuejin Zuo. "Inside China's cities: Institutional barriers and opportunities for urban migrants." American Economic Review (1999): 276-280.

Wu, Weiping. "Migrant housing in urban China choices and constraints." Urban Affairs Review 38.1 (2002): 90-119.

Wu, Weiping. "Sources of migrant housing disadvantage in urban China." ENVIRONMENT AND PLANNING A. 36 (2004): 1285-1304.

Yang, Zan, and Jie Chen. "Housing Reform and the Housing Market in Urban China." Housing Affordability and Housing Policy in Urban China. Springer Berlin Heidelberg, 2014. 15-43.

Yue, Zhongshan, et al. "The role of social networks in the integration of Chinese rural–urban migrants: A migrant–resident tie perspective." Urban Studies 50.9 (2013): 1704-1723.

Zhang, L. S. X. B., Simon XB Zhao, and J. P. Tian. "Self‐help in housing and chengzhongcun in China's urbanization." International Journal of Urban and Regional Research 27.4 (2003): 912- 937.

42