THE RELATIONSHIP BETWEEN CHINESE RURAL MIGRANTS JOB STABILITY AND THEIR HOUSING CONDITIONS IN CITIES

Transcription

1 THE RELATIONSHIP BETWEEN CHINESE RURAL MIGRANTS JOB STABILITY AND THEIR HOUSING CONDITIONS IN CITIES A Thesis submitted to the Faculty of the Graduate School of Arts and Sciences of Georgetown University in partial fulfillment of the requirements for the degree of Master of Public Policy in Public Policy By Jiemin Wu, M.G.M. Washington, DC April 19, 2013

2 Copyright 2013 by Jiemin Wu All Rights Reserved ii

3 THE RELATIONSHIP BETWEEN CHINESE RURAL MIGRANTS JOB STABILITY AND THEIR HOUSING CONDITIONS IN CITIES Jiemin Wu, M.G.M. Thesis Advisor: Thomas E. Wei, Ph.D. ABSTRACT Affordable and decent housing in urban areas is crucial for rural migrants during China s urbanization. Some rural migrants live in the dormitories provided by employers, while others self-rent at the housing market. However, due to high housing prices and other economic and institutional barriers, many self-rented rural migrants turn to low-rent housing in urban villages, where the living conditions are poor, with problems such as dirt, noise, disorder, and even crime. Many policy responses to this problem have not been successful. In this paper, I use the 2007 Survey Data of Rural Urban Migration in China (RUMiC) to study the relationship between rural migrants job stability and their housing conditions in cities. I adopt the Housing Quality Index and living space as my measurement of housing conditions. I found that having high job stability, by receiving long-term labor contracts, is associated with a higher chance of obtaining better-quality dormitories, receiving mainstream accommodation benefits, and becoming more competitive when self-renting in the regular housing market, which is the major way that the majority of rural migrants find housing in cities. Given the generally low likelihood that rural migrants have long-term labor contracts, this finding suggests that more policy concerns should be placed on working on rural migrants job stability as a way to improve their housing conditions. iii

4 ACKNOWLEGEMENT I offer my sincere gratitude to my thesis advisor, Thomas Wei, for his patient guidance throughout this process; I am also grateful to the GPPI community for their help and support. Special thanks to my family, especially my mom 罗国珍 and my dad 吴海松, for your endless love. Last but not least, thank you, Yu, for your belief in me. Many thanks, Jiemin Wu 吴洁敏 iv

5 TABLE OF CONTENTS I. Introduction... 1 II. Literature Review... 4 The Rural Migrants Housing Situation in Cities... 4 The Practices and Suggested Response to the Challenge... 5 Urban Tie... 6 III. Conceptual Model and Hypothesis... 7 IV. Data RUMiC Descriptive Statistics V. Empirical Strategy VI. Results VII. Discussion Key Discussions Caveats and Further Consideration VIII. Conclusion References v

6 Figures and Tables Figure 1 Conceptual Model Figure 2 Map of China Figure 3 Distribution of Housing Quality Index Figure 4 Distribution of Household Living Space Table 1 Descriptive Statistics of Independent Variables Table 2 Descriptive Statistics of Dependent Variables Table 3 OLS Model for Total Sample With/No Controls Table 4 OLS Model for Subgroups Dorm/SR vi

7 I. Introduction Urbanization accompanied by massive rural-to-urban migration is universally regarded as an inevitable development process for almost all economies. Associated with this development is the high demand for affordable urban housing. However, the housing conditions for rural migrants in cities are typically of lower quality. Many cities in developing countries, have witnessed a rapid growth in slums. Housing for rural migrants has become one of the key policy concerns in many developing nations (United Nations Center for Human Settlements, 1990). China is one such developing country facing policy challenges with rural migrants housing (Chan and Zhang, 1999; Wang and Murie, 2000). The pressure of rural to urban migration on China s cities ability to accommodate the extra population has increased significantly during China s rapid urbanization process, especially after implementing the policy of opening up urban labor markets to rural migrant workers in the late 1990s. On the one hand, urban authorities, particularly those in the large host cities, are complaining that the influx of rural migrants has strained the host cities capacity for housing as well as other infrastructure and administration (Seeborg et al., 2000). On the other hand, rural migrants continue to struggle to obtain decent commercial housing in the regular market and low-cost housing provided by the government. Rural migrants are legally allowed to purchase commercial housing in the regular urban housing market of small county-level cities or towns in China but not in the big cities 1

8 where rural migrants have been actually concentrated since Even when buying houses in small counties or towns, they are not eligible to obtain a mortgage. Both purchasing rights and mortgages are only granted to those individuals with local urban household registration (Hukou) status. Current high housing prices in China make it generally difficult for households to purchase housing, especially for rural migrants, who tend to have lower incomes. According to the 2007 Rural-Urban Migration in China (RUMiC) data that this paper uses, only 5% of rural migrants at that time have their self-owned houses in cities, the majority of the rest need to rent housing or live in dormitories or working sites provided by their employers. Rural migrants also have a hard time renting in the regular urban market. One main constraint is affordability. In March 2010, the average rent for a one-bedroom apartment in Beijing was 2,600 RMB/month in central districts and above 1,000 RMB/month in suburban areas, which is unaffordable to most rural migrants given their average monthly income of 1,417 RMB (Wu, 2010). Another major constraint is rural migrants high mobility as a Floating Population. As many scholars have noted (Song and Ding, 2008; Wu, 2002, etc.), many rural migrants change their jobs and locations every 3-4 months. The RUMiC data supports this observation: 40% of currently-employed migrants have either short-term labor contracts (<1 year), no contract, or are part-time workers, and 68.5% of currently-unemployed migrants last jobs were short-term contracts (<1 year), non-contracted, or part-time jobs. According to the report from the Chinese Real Estate Agents Association, generally, standard lease terms in the regular renting market are signed 2

9 for at least one year, which matches the long-term labor contract threshold. This is partially because tenants with longer-term labor contracts are typically guaranteed a more steady income to pay rent, and are more likely to stay in a given rental location for a longer period of time, thus reducing landlords risk of a tenant defaulting on rent or of needing to frequently incur search costs for new tenants. These factors decrease rural migrants accessibility to regular rental markets, even for the relatively high-income migrant groups. As a result, in China, a large number of rural migrants choose to live in urban villages where the native farmers lease their self-constructed housing for lower rents to migrants. China s separate urban-rural land policies enable the farmers to construct housing on their rural collective land for free or very low cost, which allows them to charge below-market rents in suburban villages. Farmers leasing behaviors in urban villages are also not constrained by urban authorities. They can lease to anyone for any duration. Motived by such loose leasing terms and low rents, rural migrants gradually shifted to urban villages. Nevertheless, to maximize their profits, farmers build very low-quality housing and overcrowd their housing units, which leads to a physical environment for inhabitants that is cramped, dirty, disorderly, and fraught with public safety concerns (Song and Ding, 2008; Wu, 2010; Li, 2011,etc.). In this paper, I examine the relationship between rural migrants job stability and their housing outcomes for both the rural migrant group living in self-rented housing and the group living in the dormitories (including working sites) provided by employers. Specifically, I examine how housing conditions vary with labor contract terms. Drawing on 3

10 Wu (2002) and Lin (2010) s research, I use housing quality and housing space variables to measure housing condition. I hypothesize that greater job stability for rural migrants is associated with improvements in their housing condition in cities. The results are consistent with my hypothesis, indicating that rural migrants with long-term labor contracts have a higher housing quality and larger living spaces, and the magnitudes are even more significant for the group who need to rent by themselves, as opposed to living in employer-sponsored dormitories. II. Literature Review The Rural Migrants Housing Situation in Cities There are three main strands of literature on rural migrants housing situation in Chinese cities. The first strand focuses on the impacts of the Hukou system and public housing system, specifically how they institutionalized unequal access to urban housing between urban citizens and rural migrant workers (Li 2000). Even after the housing reform during late 1990s, urban citizens continue to enjoy advantages. For example, when financing new home purchases, they can use the benefits obtained before the housing reform (i.e. sell the previously freely-provided public housing), and are also eligible for new urban housing benefits (i.e. home mortgages). These benefits did not apply to rural migrants as they were never part of urban public housing or urban Hukou systems (Wang and Murie 2000; Feng 2003). The housing inequality between urban dwellers with different Hukou statuses was maintained in the post-reform era. 4

11 The second category of research concerns the poor housing conditions of rural migrants in the informal housing sector such as urban villages (Shen and Huang 2003) and on-site housing such as construction sites (Li 2006a). Housing quality for migrants varies by occupation type, but with a generally much lower-quality level compared to urban citizens (Lin and Yu 2010; Li and Duda 2010). The third focus is on the role of different housing providers for rural migrants. Song and Ding (2008) argue that urban villages have solved a key issue in urban housing policy by providing affordable accommodations for rural migrant workers. Similarly, Li and Duda (2010) examine the patterns and conditions of housing provided by employers such as dormitories, and the economic motivation for employer provision. The Practices and Suggested Response to the Challenge In China, the most frequently used government practice to respond to rural migrants housing problems is redeveloping the urban villages where they reside (Zhang 2002; Zhang et al. 2003). Most of the implemented redevelopment programs involve demolishing villages, compensating and resettling native farmers and auctioning lands to real estate companies to convert the sites into commercial housing (Song and Ding 2008). But this approach did not take the rural migrants who were formerly residing in these areas into consideration. As a result, this approach has not really solved the problems (Li, 2011). Migrants often move to new urban villages that have not yet been redeveloped because, as described above, they often have no better alternatives (Zhang 2001). Some researchers have focused their work on approaches to addressing the housing 5

12 problem based on China s specific situation. For example, one focus is on China s institutional arrangements and ending the discriminatory Hukou system (Wu 2004b; Huang 2003). Pilot programs for removing the urban-rural Hukou division have been implemented in some municipal level cities, but haven t really solved the problem insofar that the majority of rural migrants are still unable to accumulate a sufficient amount of money to buy houses in cities. In addition, for most of them, the quality of their rented housing is still low. Other suggestions focus on governments increasing the housing supply for rural migrants. In some cities such as Beijing, Guangzhou and Chongqing, the local governments have tried to directly build public rental housing and lease the housing to migrants, or to subsidize and encourage enterprises to construct dormitory-style housing that meet minimum quality standards (Lin et al, 2010). Those flats or dormitories are mostly built in the urban fringe areas, and a symbolic amount of rent is charged (Li et al. 2007). However, this increased supply is not enough to meet the demand for housing. Those housing units are usually centered as small communities in certain places, which are not close to the scattered work places of migrant workers. In addition, the dormitory-style suites are not generally suitable for migrant families (Li 2006b). Urban Tie Studies indicate that migrants who are more likely to regard themselves as temporary members of the city often invest less of their income to improve their living conditions, and demand fewer amenities and services with respect to their housing than more permanent 6

13 residents (Goldstein, 1993; Nelson, 1976; Solinger, 1995). Evidence from Africa and Latin America suggests that, over time, when their urban ties were strengthened, temporary migrants were more inclined to settle down (Wu, 2003). And according to Lin s (2010) survey in Fuzhou, China, half of the rural migrants (44.6%) wanted to finally settle down in cities. However, their urban ties are even weaker since their major residences are always at risk of being demolished by the government during the urban sprawl process. The existing research about removing institutional discrimination and supplying more public rental housing to rural migrants suggests that these approaches have not appeared to substantially improve housing outcomes for rural migrants. In this paper, by looking directly at a nationwide sample of rural migrants, I explore a new urban tie that could influence rural migrants housing conditions in cities their job stability in cities. Given that it is strongly associated with several other factors that influence the urban ties rural migrants have, such as duration of residence in one place, employment status, income level, and family status (Conway, 1985; Gilbert and Varley, 1990; Miraftab, 1997), it may also be a strong predictor of housing conditions as well. III. Conceptual Model and Hypothesis The primary research question this paper will empirically test is, Are rural migrants job stability related to their housing condition in cities? Worker s job stability is mostly proxied by labor contract terms. According to Wu (2002) and Lin (2010), two indicators provide key measures of rural migrants housing condition: one is the usable living space, 7

14 which has long been used within China to gauge progress in housing provision; the other is a composite of qualitative aspects of housing quality. Figure 1 depicts my conceptual model. Individual characteristics of rural migrants, such as age, gender, marital status, number of dependent children, education level, health conditions, income, and remittance proportion (i.e. the ratio of income that they need to remit back to families still living in their home villages) will affect their housing outcomes. Generally, young male rural migrants are more likely to work and settle in cities. If they are also married with children and living together in cities, they will be more likely to have larger and higher-quality housing. Rural migrants with higher education levels and higher incomes are more likely to find stable jobs and settle down, and acquire higher-quality housing (Wu, 2002). However if they need to remit back a great proportion of incomes to their home villages, they may have fewer means to obtain higher-quality housing (Lin and Zhu, 2010). Rural migrants job characteristics could also influence their housing conditions. For example, according to Li (2007), rural migrant workers working in construction face the worst housing conditions, on average. Rural migrants working experience in this field is related to their working skills, and thus may partly determine rural migrants ability to obtain long-term labor contracts (Qu and Zhao, 2011). In this paper, the focus is on the importance of the labor contract term in relation to a worker s housing conditions. The workers wages and possible housing subsidies provided by employers could also impact housing conditions. 8

15 Rent and housing location (e.g., urban, inner suburbs, outer suburb and countryside) could also influence housing outcomes. In China, wealthier groups choose to live in downtown areas (except for the richest 10% of the population who live in big villas located in suburbs). Generally, holding rent equal, houses located far away from the downtown areas will be better equipped and have larger spaces. The number of residents living together in the rural migrants household also may influence the housing choice. This paper will thus compare the housing quality and living space between rural migrants with the same family size. Other macro-level factors could also contribute to determining housing outcomes. Different host cities have different policy arrangements towards employers hiring and signing long-term labor contract with rural migrants. In cities where urban authorities encourage long-term labor contracts, there may be more support for rural migrants settling down there. In more developed cities, more job opportunities are available, and the average income is higher, which could lead to higher average housing standards. Drawing on this conceptual model and previous research, I hypothesize that holding other variables constant, the housing quality is better and spaces are larger for rural migrants with long-term labor contracts. In general, employers are reluctant to give long-term labor contracts to avoid higher labor costs paying higher salaries and related insurance required by China s Labor Contract Law, such as unemployment insurance, pensions and work injury insurance for long-term employees. This is particularly true with respect to rural migrant workers, who are generally kind of disenfranchised and not 9

16 well-organized to argue for additional labor rights relative to their urban resident counterparts. Once a rural migrant gets a long-term contract, he/she is more likely to signal a greater likelihood of staying in a given city with a steady income, more social benefits (insurances) that are typically bonded to long-term contracts, and also a generally greater appeal as a tenant to landlords. IV. Data RUMiC I use the 2007 Survey Data of Rural Urban Migration in China (RUMiC). The project was initiated by a group of researchers at Australian National University, University of Queensland and Beijing Normal University and Institute for the Study of Labor (IZA). RUMiC data consists of three parts: Urban Household Survey, Rural Household Survey and Migrant Household Survey. In each part, it contains individual information of all family members covering four areas: household composition, adult education, adult employment and children; the household head answered questions about the general household such as household income, household assets, housing characteristics, and information on the rural home village. Its Migrant Household Survey is currently the largest investigation of rural migrants in China. The survey locations were selected primarily based on whether a province is one of the major receiving regions of migrants. 15 provincial capital cities or other major migrant-receiving cities were selected, including Bengbu, Chengdu, Chongqing, Dongguan, 10

17 Guangzhou, Hefei, Hangzhou, Luoyang, Nanjing, Ningbo, Shanghai, Shenzen, Wuhan, Wuxi and Zhengzhou. Figure 2 indicates the distribution of sample places in China. The distribution of the sample size across the 15 cities was loosely associated with the overall population size of the city. Within each city the sampling frame was defined on the basis of workplaces, and all businesses in randomly-selected enumeration areas within defined urban boundaries were included. During a listing process, the total number of workers and the total number of migrant workers in each workplace were recorded. The survey team then estimated the total size of the migrant worker population in each city. It was designed to be representative of that city and to provide a sampling frame for subsequent random sampling. Based on the listing data, a simple random sample of a requisite number of migrant workers for each city was selected. Once individual migrants are selected, the enumerator made an appointment with them to interview the individuals and their family. The response rate of the survey is nearly 100 percent. In total, households were interviewed in each city and a total of 5,000 household cases were obtained. Descriptive Statistics I merged the individual data on rural migrant households with their family housing and living condition data. Since the household head in the survey refers to the major contributor to the household income and the decision-maker in the family (which has been defined by the survey), I just keep the individual data for each household head and his/her corresponding housing condition data. This avoids double-counting family members living in the same household. To better understand the relationship between rural migrants labor 11

18 contract term and housing conditions, I only keep the data of currently working household heads (99.56% of the sample). Basic descriptive statistics of the key independent variables are listed in Table 1. Table 1 shows that current rural migrant workers in China are on average 30 years old with a basic 9-year compulsory education level, among which male migrants still play the dominant role as the household head in rural migrant families. They migrate from non-local rural areas to local cities to earn a much higher income than their counterparts in rural areas, as indicated by their average monthly income of yuan. By comparison, according to China National Bureau of Statistics, the average monthly income for residents in rural areas is just 300 yuan. However, apparently restricted by their education levels, almost 90% of the rural migrants are working in labor-intensive industries with comparatively lower skill requirements such as the Mining, Manufactories, Construction and Services industries. 84.9% of them self-report that their health condition is better than the average level. Rural migrants who live in the dormitories provided by their employers (hereafter as Dorm Group) share many characteristics in common with the migrants who self-rent housing (hereafter as SR group). However, significant differences still exist. The unmarried and younger males are more likely to live in the dorms, and individuals in the Dorm group have almost twice as many roommates as the SR group. Almost half of the Dorm group migrants work in the Mining, Manufactory and Construction industries, which are more likely to provide dorms in or near the working sites. Those industries also have longer working periods. For example, many construction workers need to work for 1-3 years in one construction site, 12

19 and therefore the employers may be more likely to offer long-term labor contracts to Dorm group migrant workers (50%). Nevertheless, the long-term contract assignment level of 41.4% for rural migrant workers is still much lower than urban natives rate of 79.3% (Qu and Zhu, 2011). This also has brought low attainment levels for other benefits in cities for rural migrant workers, since the availability of these benefits (such as housing subsidies) are often tied to having a long-term labor contract. Only 7.56% of the SR group workers are subsidized for their living in cities by their employers. With the previously-discussed high rental housing prices and possible perception that rural migrants are less desirable as tenants in the regular urban rental market because of job instability, it is perhaps not surprising that only 31.3% of them live in urban areas. I measure Housing Condition with two indicators housing quality and living space. I used the household inside floor areas measured in square meters to indicate their living space. To construct the Housing Quality Index (HQI), I followed the basic method that Wu (2002) adopted to build a Housing Qualitative Index. Wu s indicator is a composite of seven qualitative aspects of housing, calculated as Qualitative Index = ΣXi/7. The seven aspects included electricity, water, gas, kitchen, bathroom, structure and dwelling function whether it is for residential use, and used a single scale to calculate those dummy variables. Other housing scholars (Lawrence 1995, Wu 2003, Lin 2010, etc.) agree that housing quality encompass the variables of water, sanitation (bathroom), kitchen, heating, fuel (gas), and decoration (structure). I use six dummy quality variables: water, sanitation, kitchen, heating, fuel and decoration. I also adjusted the calculation method by first standardizing 13

20 each of the quality variables (e.g., subtract mean and divide by standard deviation), then taking a simple average of those variables, noted as HQI={ ( xi xi )}/6. Table 2 depicts δ summary statistics on housing quality and living space for rural migrants both in Dorm and SR group, and Figure 3 and Figure 4 give the basic distribution for the HQI and space variables. Water=1 if there is tap water; Sanitation=1 if there is a bathroom; Kitchen=1 if there is a kitchen; Heating=1 if there is heating system; Fuel=1 if they use gas; Decoration=1 if there is decoration with wood floors, painted walls and furniture; V. Empirical Strategy To address my hypothesis of whether the labor contract tenure (long-term or short-term) is related to rural migrants housing conditions in cities, I ran multivariate OLS regressions of housing condition (as indicated by HQI and Space) on the key independent variable of Contract as well as controlling for four major types of independent variables, listed as individual characteristics, job characteristics, macro-level variables, and housing attributes I started my estimation with two multivariate models that included all the variables described above, and ran the model on the full sample of rural migrants in my data. In addition, I adjusted for whether a rural migrant worker is in the Dorm group or the SR 14

21 group by adding a dummy variable called place, which equals 1 if the rural migrant lives in self-rented housing and 0 if the migrant lives in a dormitory. The general model is: Yi=β 0 +β 1 *contract+β 2 *Individuals+β 3 Jobs+β 4 Macro +β 5 Housing+β 6 place+ξ i Where Y i is the Housing Quality Index (HQI) or the total household living space in logarithm form. Contract is a dummy variable, =1 if he/she is a permanent worker or has a long-term labor contract (>=1 year), and equals to 0 if he/she has a short-term labor contract (<1 year) or is a non-contract, temporary or domestic worker. Individuals variables include age, age-squared, male dummy, marriage dummy (not married as the reference group), education level, health category variables (good health if self-reported better than average level, average health if self-reported around average health condition and the bad health worse than average), monthly income measured in Yuan and the Remittance Income Ratio (RIR). RIR is the percentage of a migrant s monthly income that he/she must remit back to his/her family members, who still live in a rural home village. Jobs variables include the categorical variable of industrial fields (including five fields: Mine, Manufacture & Construction, Electricity, Gas, Transportation & IT, Finance, Estate, Health & Education, Service and others); the work experience variable (wexper), which equals to the investigated year minus the year when they started working in the current occupation area; the continuous variable of monthly wages they received, and rural migrants accommodation benefits they received. I generate two dummy benefits variables: pdorm=1 if the employer provides a dorm and hsubsidy=1 if the worker receives housing subsidies. 15

22 The Macro-Level variables include the Hukou variable to indicate rural migrants registration status and city dummies to control for the different economic and social characters in each sample city. To keep the external validity of studying rural migrant workers, I drop off the sample data whose Hukou status are not rural Hukou, and so the Hukou dummy =1 if the migrant is local county rural, and Hukou, =0 if is non-local rural Hukou 1. The Housing attributes variables contain the monthly rent and location, which is expressed by a dummy variable of urban, =1 if the migrant worker lives in an urban area, including inner urban villages. I also run the general fully-controlled multivariate models separately on the Dorm group and the SR group to see if the influences of labor contracts differ by these two groups. In addition, I run my models excluding all control variables to see how much the results change on HQI and space. VI. Results Table 3 shows results from the full multivariate model with control variables and the dependent variables of Housing Quality Index (HQI) and log of Housing Space. Since HQI is a qualitative index variable, to better understand the magnitude of my interested estimated coefficient on HQI change, I convert the point coefficient into standardized effect size units, noted as Z= (βi)/(std dev).. As expected, the coefficient on the key variable 1 Except for general separating Urban and Rural residents, Hukou System in China also registers individual people s original residence or location, and divided Local and Non-local among both Urban residents and Rural residents, whom are identified as Local and Non-Local Urban Hukou and Local and Non-Local Rural Hukou. Partially due to the legacy of China s old migration policy restrictions, locals and Non-locals are separately treated by the local government. As a result, the rural migrants I discussed in this paper are all with rural Hukou and migrated to urban areas, but are differentially treated by the host cities because of their local and non-local originalities. Therefore, I also consider the effects of Local and Non-local status among rural migrants on their housing conditions. 16

23 Contract is highly statistically significant and positive in the regression, and receiving a long-term stable labor contract is associated with a 0.11 standard deviation higher housing quality index. Receiving accommodation benefits that are usually bonded with long-term labor contracts such as housing subsidies, are also associated with significantly higher HQI scores (0.19 stdev). Rural migrant families with a male household head have on average lower housing quality (-0.13 stdev). As expected education level, work experience and the earnings are positively related to housing quality. HQI score is also quite sensitive to the marginal change of RIR level. When the remittance ratio increased by 1 percentage point, it is associated with a 0.31 stdev decrease in HQI level. Among the rural migrant workers, those who live in the self-rented housing have an almost 0.70 stdev units higher HQI score on average than those living in the dormitories. Table 3 also shows the same regression as for the first HQI model but with the log(space) as the dependent variable and with additional controls for the number of residents in each household. The coefficient of the key variable Contract is also statistically significant and positive. For rural migrant households with the same family size and monthly earnings, the head of household having a long-term labor contract is associated with a 6.72% increase in their living space. The average household living space is square meters, and a 6.72% increase possibly brought 2.06 square meters space increase. Receiving accommodation benefits that are usually bonded with long-term labor contracts such as housing subsidies also are associated with a significant 7.87% increase in living space correlated with an average 2.41 square meter increase. Similar to the results for the 17

24 HQI model, rural migrant families with a male household head have lower housing quality, with a 10.35% lower amount of space as many as 3.17 square meter decrease. However, holding family size constant, married migrants actually have smaller housing space than unmarried ones with a 1.34% as many as 0.41 square meter space difference. A higher remittance ratio is also associated with having smaller living spaces in cities on average. RIR increasing by 1 percentage point is associated with a 10.09% decrease equals to 3.09 square meters in space. On average, living space for those who live in self-rented housing is almost 26.87% equals to 8.24 square meters larger than those living in the dormitories. In Table 3 I also state the results of two models including only the key independent variable of interest Contract on HQI and log(space), with no control variables. Contract in both models are insignificant and with much less magnitudes. The head of household having a long-term labor contract is only associated with a 0.02 stdev increase in HQI score and a 1.18% of decrease in their living space. Table 4 shows the results of the four fully controlled regressions I run on the dependent variables of Housing Quality Index (HQI) and log of Housing Space for the two subgroups the Dorm group and Self-Rented group. For HQI score, as hypothesized, coefficients of the key variable Contract for both Dorm and SR group are highly statistically significant and positive in the regression, and receiving long-term stable labor contract is associated with a higher effect for the Dorm group. Receiving accommodation benefits that are usually bonded with long-term labor contracts such as housing subsidies, are also associated with significantly higher HQI scores (0.13 stdev) for the SR group, but 18

25 not for the Dorm group. And living in self-rented houses, the marginal change of RIR level brings a higher decrease at their housing quality. Table 4 also shows the same regressions but now with the log(space) as the dependent variable and with additional controls for the number of residents in each household for the two groups. Similarly, the coefficients of the key variable Contract are statistically significant and positive for both groups. But receiving accommodation benefits that are usually bonded with long-term labor contracts has positive effect on SR group (6.36%)but negatively related to Dorm group(-24.94%). Holding family size constant, married migrants have smaller housing space than unmarried ones with a 4.98% difference for self-rented group; however the situation is conversed for Dorm group. But the effect of RIR is not significant and with a positive small magnitude for Dorm group which is different from full sample situation. VII. Discussion Key Discussions The key finding from these analyses is that there is a positive and significant relationship between job stability and housing condition among the total sample and two subgroups: rural migrants receiving long-term labor contracts are more likely to have better housing quality and larger living spaces in cities, even after controlling for factors such as rent, wages, occupational industry, education, and work experience. According to Li (2007), employers are more likely to sign longer labor contracts with the workers they prefer; for 19

26 instance, construction-industry employers may be more likely to hire young men in good health, even if they have low levels of education. For those employers who provide dorms to rural migrant employees, dorms with better qualities may attract more desirable migrant workers. Therefore, rural migrant workers with longer contact terms may have priority when employers are distributing access to dormitories. For those rural migrants who self-rent residences in the market, having long-term contracts usually guarantee them more stable income, more tight ties with the host cities and a greater likelihood of being perceived as more reliable tenants by the landlords and thus get housing with better conditions. Once rural migrant workers have long-term labor contracts, they are more flexible to choose housing of higher quality and with more space when they rent by themselves compared with living in the dorm. Compared to diverse housing units in the market, many dorms are uniformly designed to have relatively smaller space and to be shared by several workers, and thus are less likely to meet the diverse needs of rural migrants. Secondly, in many cases, if the rural migrants needed to remit a higher proportion of their income back to their family at home villages, it may tell a different story of their housing outcomes in cities. It may even weaken the positive effects brought by long-term contracts which guarantee a more stable and sometimes higher income to improve housing conditions for rural migrants. High RIR ratio generally indicates that, for the rural migrant, there is a stronger tie in their home village than to urban livings, such as supporting their parents or other family members, or motivating them to save or invest money in their home 20

27 villages instead of investing in cities (such as their housing in cities). Caveats and Further Consideration First, to more precisely make conclusions about my study, the data can still be improved in terms of representativeness and sample size by adding other big cities, such as Beijing, that receive large numbers of rural migrants. With urbanization further expanding to the middle and western parts of China, the sample size should also increase its coverage to those areas. Another concern is the unobservable factors. The sensitivity of the main results to the presence of control variables suggests that there are many factors that influence the type of labor contract and the type of housing. I have controlled for many of them, but there may still be some that be left out such as personal preference, individual potential motivation, and personal networks in cities. This may create some risk of omitted variable bias. For example, this survey does not differentiate between the group of rural migrants who live with their urban relatives and just pay symbolic rents and the rest of the population. Thus it is possible that some of those migrants may have better housing outcomes even without stable jobs. Similarly if the workers have personal connections known as guanxi to find a job, it is more likely they will obtain a longer-term contract. The Housing Quality Index which I have adopted in this paper as one of the major measurement of housing conditions for rural migrant workers still has some weaknesses. Due to lack of research on how to quantitatively measure housing quality, the six components of HQI are equally weighted when doing the standardization calculation. 21

28 However, for example, in the cold northern parts of China, equipping heating facilities is much more important and expensive than having water, and equally weighting access to heating and water will possibly imperfectly measure the true quality of housing in those parts of the country. In my future research, I plan to find datasets with large sample sizes and broader coverage of China s rural migrant workers, and continue to find appropriate proxies available for these unobservable factors. VIII. Conclusion As China undergoes rapid development and modernization, the country's urbanization level will increase steadily. Persistent rural migratory influx will have a significant effect on the rate of urbanization and on patterns of urban development. Experience of other developing countries, as well as the current situation in China, shows that poor housing conditions are likely to force rural migrants to eventually become the urban underclass, and only much stronger urban ties could help them improve their living conditions. In this respect, urban living experience is so crucial to rural migrants, as well as to the promotion of China s urbanization process. The findings in this paper suggest that having high job stability through the receipt of long-term labor contracts is an essential and very strong urban tie that the host cities could provide to the rural migrants. Rural migrant workers with long-term contracts have a higher chance to obtain better-quality dormitories, receive mainstream accommodation benefits, and become more competitive when self-renting in the regular housing market, which is the 22

29 major way that the majority of rural migrants find housing in cities. However, the long-term contract sign rates for rural migrants are still quite low. This suggests that more preferential policies could be considered as a way to encourage employers to sign long-term labor contract with migrant workers, as well as more strict regulatory oversight policies for the employers. 23

30 Figure 1 Conceptual Model Individual Characteristics Age Gender Marital status Children living in cities Job Natures Industrial fields Working experience Labor contract Housing Subsidies Macro Level Policy arrangements Economy Hukou Status Housing Rent Locations Type Education Health Incomes and Remittance Housing Living Quality Space Heating Kitchen, Decoration Water, Sanitation Fuel 24

31 Figure 2 Map of China Note: Provinces are underlined with red line are the sample provinces. 25

32 0.05 Density Density Figure 3 Distribution of Housing Quality Index HQI Density Distribution HQI Index kernel = epanechnikov, bandwidth = Total Self-Rented Dorm Figure 4 Distribution of Household Living Space Space Distribution Space_per_capita(M2) kernel = epanechnikov, bandwidth = Total Self-Rented Dorm 26

33 Table 1 Descriptive Statistics of Independent Variables Variable Total Dormitory Self-Rented (S-R) Observations ,325 2,631 Individual Average Age(Year) *** Male *** Married *** Average Residents No *** Education(Years) Health# Better than Avg Average Hukou# *** Average RIR# *** Job Long-Term Contract *** Industry Mine,Manuf&Const *** Elec,gas,Trans&IT *** Fn,Est,Hlth&Educ * Service *** WorkExperience(Years)# *** Income(Yuan/month) *** Housing subsidy *** Dorm Provided# *** Housing Urban# *** Rent(Yuan/month) *** Note: Tests of differences in means between the dormitory and self-rented samples were conducted. Statistically significant differences are noted with asterisks in the self-rented column (***=significant at 1% level. **=5% level. *= significant at 10% level.); Health is rural migrants health condition that self-reported by themselves compared to the average health level; Hukou =1 if the migrant is local county rural Hukou; RIR= Remittance Income Ratio = The percentage of a migrant worker s income that he/she needs to remit back monthly to his/her home village, such as to parents or other families still living at home villages; Work experience=number of years since they started working in the current occupation area; Dorm provided=1 if the employers provide optional dorms to the rural migrants; Urban=1 if the individual lives in an urban area, including inner urban villages; 27

34 Table 2 Descriptive Statistics of Dependent Variables Variable Total Dormitory Self-Rented (SR) Observations ,325 2,631 Quality Variables: Water Sanitation *** Kitchen *** Heating * Fuel *** Decoration# *** HQI# *** (1) (0.645) (1.055) Average Space# (28.467) (28.402) 32.74*** (28.368) Note: Tests of differences in means between the dormitory and self-rented samples were conducted. Statistically significant differences are noted with asterisks in the self-rented column (***=significant at 1% level. **=5% level. *= significant at 10% level.); Decoration is described with wood floors, painted walls and furniture such as sofa. HQI is reported as the mean value of Housing Quality Index for dormitory residence and self-seeking migrants, with standard deviation in the parentheses. HQI={ ( xi xi )}/6, Xi=Water, Sanitation, Kitchen, Heating, Fuel, Decoration δ Average space is reported in square meters with standard deviation in the parentheses. 28

35 Table 3 OLS Model for Total Sample With/No Controls Dependent Variable: HQI Dependent Variable: Log(Space) With Controls Without Controls With Controls Without Controls Variable CONTRACT *** *** (0.0255) (0.2888) (0.0214) (0.0233) Place *** *** (0.0468) (0.0355) Male ** *** (0.0269) (0.0215) Married * (0.0408) (0.0286) Residents NO *** *** (0.0037) (0.0043) Education *** *** (0.0055) (0.0045) Better Health Level * (0.0909) * (0.0860) RIR ** ** (0.0690) (0.0590) Mine,Manuf&C onst (0.1219) (0.0945) Elec,gas,Trans& IT *** (0.1314) (0.1034) Fn,Est,Hlth&Ed uc ** (0.1254) (0.0998) Service *** (0.1180) (0.0938) Wexper * * (0.0027) (0.0027) Income *** * ( ) ( ) Hsubsidy *** * (0.0786) (0.0460) Urban *** (0.0411) (0.0271) Rent *** *** ( ) ( ) R-Squared Observations

36 Notes: Robust standard errors in parentheses. Statistical significance is noted with asterisks (***=significant at 1% level. **=5% level. *= significant at 10% level.) Dependent variables are HQI and log(space). Controls that were added in this analysis includes place=1 if the rural migrant live in self-rented house, and age, age-squared added to capture potential nonlinearity, the number of residents in the households to compare families with same size, city dummies were also added to capture the social and economic effect, and the baseline province is Dongguan City. Income and Rent are reported in Yuan. 30

37 Variable CONTRACT Male Married Residents NO. Education Better Health Level RIR Mine,Manuf&C onst Elec,gas,Trans& IT Fn,Est,Hlth&Ed uc Service Wexper Income Hsubsidy Urban Rent Table 4 OLS Model for Subgroups Dorm/SR Dependent Variable: HQI Dependent Variable: Log(Space) Dorm SR Dorm SR *** (0.0270) ** (0.0296) (0.0403) ** (0.0038) *** (0.0059) *** (0.0805) ** (0.0703) (0.1242) (0.1436) (0.1334) (0.1244) (0.0030) *** ( ) (0.1655) ** ( ) **** (0.0444) *** (0.0427) ** (0.0611) *** (0.0194) *** (0.0087) (0.1569) *** (0.1408) (0.1965) *** (0.2083) ** (0.2084) ** (0.1934) ** (0.0043) *** ( ) * (0.0848) *** (0.0430) *** (0.0001) *** (0.0307) (0.0331) (0.0427) *** (0.0045) *** (0.0066) * (0.1134) (0.0768) (0.1583) (0.1732) (0.1648) (0.1579) (0.0034) *** ( ) ** (0.1355) (0.0001) ** (0.0299) *** (0.0275) (0.0395) *** (0.0171) *** (0.0060) * ( ) ** (0.0920) (0.1020) (0.1128) (0.1106) (0.0995) (0.0029) *** ( ) (0.0479) * (0.0279) *** ( ) R-Squared Observations Notes: Robust standard errors in parentheses. Statistical significance is noted with asterisks 31

38 (***=significant at 1% level. **=5% level. *= significant at 10% level.) Dependent variables are HQI and log(space), tested for both Dorm and Self-Rented groups. Controls were added in this analysis, including age, age2 added to capture potential nonlinearity, the number of residents in the households to compare families with same size, city dummies were also added to capture the social and economic effect, and the baseline province is Dongguan City. Income and Rent indicate the money is in Yuan 32

39 References Chan KW, Zhang L, 1999, The hukou system and rural-urban migration in China: processes and changes'', China Quarterly number 160, ; Conway D, 1985, Changing perspectives on squatter settlements, intraurban mobility, and constraints on housing choice of the Third World urban poor, Urban Geography ; Gilbert A, Varley A, 1990, Renting a home in a Third World city: choice or constraint? International Journal of Urban and Regional Research ; Goldstein S, 1993, The impact of temporary migration on urban places: Thailand and China as case studies, in Third World Cities: Problems, Policies, and Prospects Eds J D Kasarda, A M Parnell (Sage, Newbury Park, CA) pp ; Huang Y, 2003, A room of one's own: housing consumption and residential crowding in transitional urban China Environment and Planning A ; Li, B. 2006a, Floating Population or Urban Citizens? Status, Social Provision and Circumstances of Rural-Urban Migrants in China, Social Policy & Administration, 40: ; Li, B. 2007, The Transformation of Cities and Establishing Flexible Housing System in Large Cities (Chengshi Bianqian He Jianli Lingguo De Dachengshi Zhufang Tixi), International Economic Review (guoji jingji pinglun), 5: 26-30; Li, B, 2011, Life considerations and the housing of rural to urban migrants: the case of Taiyuan, Available in LSE Research Online: November 2011; Li S-M, 2000a, Housing consumption in urban China: a comparative study of Beijing and Guangzhou, Environment and Planning A ; Li S-M, 2000b, The housing market and tenure decisions in Chinese cities: a multivariate analysis of the case of Guangzhou, Housing Studies ; Lin Liyue, Zhu Yu, 2010, The Diverse Housing Needs of Rural to Urban Migrants and Policy Responses in China: Insights from a Survey in Fuzhou, IDS Bulletin Volume 41 Lin, L. and Yu, Z. 2010, The diverse housing needs of rural to urban migrants and policy responses in China: Insights from a survey in Fuzhou, IDS Bulletin, 41: 12 21; 33