Mobility and travel behavior in urban China: The role of institutional factors

Transport Policy 69 (2018) 122–131

Contents lists available at ScienceDirect

Transport Policy journal homepage: www.elsevier.com/locate/tranpol

Mobility and travel behavior in urban China: The role of institutional factors Mingzhu Yao, Donggen Wang

T

∗

Department of Geography, Hong Kong Baptist University, Hong Kong, PR China

A R T I C LE I N FO

A B S T R A C T

Keywords: Hukou Travel behavior Car ownership Structural equations model Beijing

Institutional factors, which are referred to social and economic norms and rules, define individuals' entitlement and access to opportunities and thus may play an important role in shaping individuals' mobility and travel behavior, especially in countries that are experiencing or have experienced the transition from a planned economy to market-oriented economy. Studies in Chinese cities show that danwei or type of work unit is an important institutional factor in explaining jobs-housing relationship and commuting behavior. We argue that, hukou or household registration (another institutional factor), may also explain the mobility and travel behavior of Chinese urbanites. Unlike population registration systems in many other countries, hukou, one of the most important institutional arrangements in contemporary China, determines an individual's entitlement to stateprovided benefits and opportunities and plays a crucial role in defining access to housing, jobs, car ownership/ usage, education, etc., which has far reaching implications for mobility and travel behavior. Using data collected from Beijing, we use structural equations modeling method to empirically test our hypothesis about the importance of hukou in explaining car ownership, travel time and transport mode choice for daily trips. Results show that hukou status has a significant impact on mobility and travel behavior of individuals. Specifically, local urban residents are found to have better home-work proximity and higher car ownership rate, travel more by non-motorized modes and spend less time on daily travel. This study provides insights into the complex relationships among hukou, built environment, mobility and travel behavior in urban China. The research findings can be used to assist planners and policy-makers in developing effective strategies to promote sustainable urban development.

1. Introduction 1.1. Determinants of mobility and travel behavior Mobility is a term widely used in different fields. In this research, mobility refers to transport mobility. Transport mobility is defined as the “potential” for movement, conditioned on the mobility tools one has access to, including car, transit pass, feet, etc. (Spinney et al., 2009). Mobility is inextricably linked to travel behavior, which refers to the daily life trip making behavior in terms of when, where, by what means, how long or how far trips are made. It is usually represented by common descriptive measures of travel, such as vehicle miles traveled, trip frequencies, travel time, travel distance, transport mode and so on. Existing studies have identified a wide range of determinants of mobility and travel behavior, covering socio-demographics and life circumstances (Lu and Pas, 1999; Van Acker and Witlox, 2010; Scheiner and Holz-Rau, 2013), residential built environments (Handy et al., 2005; Cao et al., 2007), new technologies like ICT (Information and Communication Technologies) and autonomous vehicles (Wang and

∗

Corresponding author. E-mail addresses: [email protected] (M. Yao), [email protected] (D. Wang).

https://doi.org/10.1016/j.tranpol.2018.05.012 Received 17 January 2018; Received in revised form 9 May 2018; Accepted 23 May 2018 0967-070X/ © 2018 Elsevier Ltd. All rights reserved.

Law, 2007; Levin and Boyles, 2015), attitudes and personalities (Gärling et al., 1998; Van Acker et al., 2010). Apart from these factors, institutional factors may also be important determinants of mobility and travel behavior, especially in countries that are experiencing or have experienced the transition from planned economy to market-oriented economy such as China. Institutional factors, which are connected with social and economic norms and rules, define individuals' entitlement and access to institutional arrangements and thus may play an important role in shaping individuals' mobility and travel behavior. Although institutional factors have direct relevance to public policies, the role of institutional factors has not yet received much research attention. Existing studies in Chinese cities have acknowledged that danwei or type of work unit—an institutional factor characterizing socialist China—plays an important role in explaining jobs-housing relationships and commuting behavior. Findings showed that commuters living in danwei compounds had shorter commuting trips than those living in houses from market sources (Wang and Chai, 2009; Zhao et al., 2011). However, hukou or household registration, another important institutional arrangement in contemporary

Transport Policy 69 (2018) 122–131

M. Yao, D. Wang

Table 1 Classification of hukou system in Beijing. Type of registration

Place of registration

Local Non-local

Reside in the same place as hukou location Reside in a different place from hukou location, holding “WRP″ Reside in a different place from hukou location, holding “TRP″

Rural (Agricultural)

Urban (Non-agricultural)

Local rural hukou Non-local rural hukou holding “WRP″

Local urban hukou Non-local urban hukou holding “WRP″

Non-local rural hukou holding “TRP″

Non-local urban hukou holding “TRP″

Notes: “rural” is the same as “agricultural” and “urban” is the same as “non-agricultural” in the hukou classification (Fan, 2008).

some extent, holders of WRP can enjoy benefits similar to holders of local Beijing hukou. But local hukou is permanent, and WRP is only effective during its validity period. According to the classification, there are 6 categories of residential status in Beijing hukou system, as shown in Table 1. The hukou system in China used to be a long-standing basis for the provision of goods and welfare, such as the basic foodstuffs, housing and jobs. Despite the moderate relaxation of rural labour mobility from 1980s, the hukou system remaining in place today still has its influence on many fundamental aspects of people's life, including housing, employment, car ownership, children's education, medical insurance, etc. (Zhao and Lu, 2010). In these aspects, differential treatment still exists, between residents with urban hukou and rural hukou, as well as between non-local and local residents. Firstly, housing inequalities in urban China is strongly affected by policies connected with hukou status, which give preferential treatment to local urban residents (Logan et al., 2009). Taking Beijing as an example, policies on indemnificatory housing, including economical purchase housing (“jingji shiyong fang” in Chinese) and low-rent housing (“lianzu fang” in Chinese) only target households with local urban hukou. Most of the low-cost commercial housing (“xianjia shangping fang” in Chinese) are built for low-income local urban households with financial difficulties and local rural households experiencing resettlement induced by land expropriation, beyond the reach of non-local households.2 For commodity housing purchase, residents holding nonlocal hukou also encounter more constraints in comparison with local Beijing residents.3 Overall, when it comes to housing supply, urban natives usually have advantages. Secondly, rural-to-urban migrants face labour market discrimination due to their hukou status, especially from the companies offering high-wage, good benefits and job security, such as state-owned enterprises (SOEs) and government and institutional organizations (GIOs) (Song, 2016). Previously, some SOEs only recruit potential employees holding local urban hukou (Beijing Labor Bureau, 1989; Fang and Chan, 2000). Although the reforms have generated a wide variety of jobproviding organizations, the state-owned work units like GIOs and SOEs still employ a large proportion of local urban workforce (Li and Liu, 2016). Besides, the hukou system exerts its influence on jobs-housing relationship in Chinese cities, particularly in relation to the floating population (also referred to as migrant workers) which do not have local urban hukou. For example, very few SOEs have provided housing assistance to temporary workers, while cheap housing at a low price was provided to employees with local urban hukou. As a result, most rural migrants are concentrated in suburban enclaves or peri-urban villages (Ma, 2004; Zhao and Lu, 2010). Finally, due to the exponential car growth and serious traffic congestion in recent years, hukou system is linked to car ownership restriction policies to control car expansion in big Chinese cities like

China, may also contribute to explaining the mobility and travel behavior of Chinese urbanites, because it determines an individual's entitlement and access to state-provided benefits and opportunities for housing and child education, employment and car ownership, which has far reaching implications for their travel behavior. Hence, this paper seeks to contribute to research on the relationship between hukou and travel behavior in China.

1.2. Hukou system in China The hukou system (household registration system) was instituted in China in the 1950s. Unlike the population registration system in many other countries, it was designed not only to provide population statistics and identify personal status, but also to achieve many other important objectives (such as population regulation) through its close connection with people's access and entitlement to government-provided benefits and opportunities. The hukou system used to be a crucial means of setting up and maintaining a block to free flows of resources (including labour) between the urban and rural areas (Chan and Zhang, 1999). But the restriction it imposed on population migration has begun to relax since the implementation of economic reforms around 1980s, in conjunction with China's economic transition from centrally planned economy system to market economy system (refer to Chan and Zhang (1999) and Chan (2009) for detailed information about the hukou system in China). In this process, there has been an increasing number of migrants, especially from rural origins, moving to cities (Guo and Iredale, 2004). To grasp the essence of the hukou system in China, it is necessary to understand its classification and conversion. The basis is a dual classification structure, namely, classification by the place of hukou registration (local versus non-local hukou) and by the type of hukou registration (agricultural versus non-agricultural hukou) (Chan and Zhang, 1999; Song, 2014). It should be noted that “rural” and “urban” here are the respective synonyms of the agricultural and non-agricultural hukou type, not referring to a person's current physical location (Song, 2014). The present hukou conversion policies in big cities (e.g. Beijing and Shanghai)—the desired destinations of the majority of migrants—are mostly in favor of the highly educated and super-rich migrants (e.g. investors and home buyer). These are the hardest places for hukou conversion and are beyond the reach of most of the rural migrants (Li et al., 2010; Song, 2014). Due to highly limited local hukou quota, measures have been taken by authority of popular cities to attract and retain talents. For example, the government in Beijing has established a residence permit system for non-local population, which includes two types of permits, one is “WRP” (Work & Residence Permit) and the other is “TRP” (Temporary Residence Permit). “TRP” is designed to regulate immigrants from outside the city without local hukou status, while “WRP” is designed to attract and retain talents, unofficially named as Beijing “Green Card”. Since 1999, government in Beijing has started to issue “WRP” to a limited group of people working in hightech field and senior managers who have investment in Beijing.1 To 1

2 Information from the Official Website of Beijing Government: http://zhengwu. beijing.gov.cn/zwzt/bjsbzxzf/t1094083.htm. 3 Information from the Official Website of Beijing Government: http://zhengce.beijing. gov.cn/library/192/34/211/898456/82041/index.html.

Information from China Today: http://www.cctv.com/lm/124/31/86441.html.

123

Transport Policy 69 (2018) 122–131

M. Yao, D. Wang

commuting behavior, but travel behavior in general. In addition, previous studies always consider hukou as an exogeneous variable when analyzing its impact on travel behavior. However, as mentioned above, hukou conversion policies in big cities like Beijing give preferential treatment to those super-rich and highly educated migrants, indicating that socio-economic backgrounds can trigger the conversion of hukou status. Thus, hukou status needs to be treated as an endogenous variable for a more comprehensive analysis, which can be realized by using structural equations model (SEM). SEM has many advantages in comparison with multiple regression model. For instance, SEM is able to test models with multiple dependent variables and differentiate direct and indirect effects, so as to capture the effect that hukou may impose indirect influence on travel behavior via its impact upon residential location, car ownership and employment, which is not examined in previous research. In view of the above discussion, this study intends to use the SEM method to achieve two main aims: (1) Comprehensively investigate the relationships among hukou, built environment, employment, car ownership and daily activity-travel behavior. (2) Empirically test our hypothesis about the importance of hukou in explaining car ownership, travel time and transport mode choice for daily trips.

Beijing, Shanghai and Guangzhou. In Beijing, local residents and workers holding WRP are entitled to apply for the vehicle ownership quota if they do not possess cars, however, additional requirements are imposed on other residents, stating that they must hold TRP and have paid social insurance and personal income tax in Beijing for at least five consecutive years.4 2. Existing studies about impacts of hukou on mobility and travel behavior So far, only limited research attention has been paid to the influence of hukou as a crucial institutional factor on travel behavior. Among them, Guo et al. (2016) explored the impacts of hukou on travel mode choice by using revealed preference data collected in Foshan City. To accommodate the unobserved heterogeneity, they tried to employ both the mixed logit and latent class methods, and the latter outperformed in terms of goodness-of-fit and prediction accuracy. Apart from travel mode choice, the influence of hukou on commuting behavior were also investigated by a few researchers. Zhao and Lu (2010) considered institutional factors including housing provision (indicated by housing source and housing ownership) as well as labour mobility management (indicated by employment sector and hukou status) and analyzed their impacts on workers' commuting time in Beijing through a regression model. They found that hukou status had a relatively significant correlation with worker's commuting time. Specifically, a worker holding a local urban hukou tended to have shorter commuting time. Similar revelation concerning the impact of hukou on commuting time was given by Zhao and Howden-Chapman (2010) when assessing the impact of hukou system on migrants' job accessibility and travel burden for work trips in the case of Beijing. Regression results showed that hukou system could lead to a decrease in possibilities of migrant family members from being employed as well as an increase in their commuting time. However, seemingly contrary findings were exposed by Li and Liu (2016; 2017) when examining the influence of hukou on commuting behavior of residents in Guangzhou. They reported that in comparison with local hukou holders, non-local hukou holders tended to have more balanced jobs-housing relationship, rely less on motorized transport, and spend less time on daily commuting. Using the same data collected in Guangzhou, they also found that local hukou workers had significantly higher job accessibility than non-local hukou workers, but the average commuting time and distance of non-local hukou workers was considerably shorter than that of local hukou workers. Qin and Wang (2018) also revealed that migrants (without Beijing hukou) tended to have shorter commute distances than residents with Beijing hukou, which is in support of Li and Liu's (2016) study showing a similar trend that migrants had more balanced jobs-housing relationship than native residents. These findings were at odds with the general expectation that spatial mismatch was more likely to be substantial for non-local hukou holders. The possible cause of the seemingly contrary findings may due to inconsistency in creation of dummy variable regarding hukou status (e.g. local urban versus other, local versus non-local) and variation across cities.

4. Beijing and the survey data Beijing is the capital and political center of China. The implementation of the reform and opening-up policy since the year 1978 has greatly facilitated the population mobility from rural areas and small cities to metropolises like Beijing. According to the official statistics published by the National Bureau of Statistics of People's Republic of China, the migrant population in Beijing has been growing dramatically, from 0.218 million (2.5% of total population) in 1978 to 8.226 million (37.9% of total population) in 2015.5 As discussed in the introduction, hukou system in Beijing has inextricable linkages with multiple aspects of people's life, including housing, employment, car ownership, travel and so on. Thus, in a metropolis consisting of mixed population holding different residential status, the hukou system is likely to play a crucial part in the life of Beijing residents. Geographically, Beijing has a total area of 7800 km2 spanning over 6 central districts (filled in purple), 6 inner suburban districts (filled in pink) and 6 outer suburban districts (filled in grey), as depicted in Fig. 1. The 6 central districts and 6 inner suburban districts form into a region with strong monocentric characteristics. Centered around the Tian'anmen Square, the 2nd, 3rd, 4th, 5th and 6th Ring Road were successively built (as shown in Fig. 1), with an average radius of 4 km, 7 km, 10 km, 15 km and 25 km respectively. To improve the linkages between urban and suburban areas, the transportation system consists of eight radial expressways and a ring-and-radial pattern of subway network mixed by loop lines and radial lines (Huang et al., 2015). According to the Beijing Statistical Yearbook (2017), 56.2% of the households live in the urban area, 31.6% of them live in the inner suburban area and the remaining 12.2% live in the outer suburban area; while 73.5% of the jobs are located in the urban area, 21.5% of the jobs are located in the inner suburban area and the remaining 5.0% are located in the outer suburban area.6 As shown by these numbers, the majority of population in Beijing reside in the urban and suburban districts, and most job opportunities are also distributed among these districts. But this doesn't imply a balanced jobs-housing relationship, because data shows that the average daily commuting time is 1.5 h and the average commuting distance is 19.3 km (Huang, 2011; Yao and Wang, 2014), indicating a significant job-housing mismatch in Beijing.

3. Research objectives As can be noted, existing studies mainly focused on the impacts of hukou on jobs-housing relations and commuting behavior. We argue that, as one of the most important institutional arrangements in contemporary China, the hukou system not only defines access and entitlement to jobs and housing, but also to car ownership, education, urban amenities and so on, which have implications far beyond

5 Information from Beijing Statistical Yearbook (2016): http://www.bjstats.gov.cn/nj/ main/2016-tjnj/zk/indexch.htm. 6 Information from Beijing Statistical Yearbook (2017): http://www.bjstats.gov.cn/nj/ qxnj/2017/zk/indexch.htm.

4 Information from the Official Website of Beijing Government: http://www.bjhjyd. gov.cn/bszn/20131231/1388482774221_1.html.

124

Transport Policy 69 (2018) 122–131

M. Yao, D. Wang

Fig. 1. The map of Beijing.

avoid the possible bias caused by interactions between household members. As a result, a final sample size of 776 is generated for modeling. Structural equations model (SEM) is adopted to test the hypothetical causal relationships and capture the impacts of exogeneous variables on endogenous variables, together with the impacts of endogenous variables upon one another (refer to Bollen (1989) and Golob (2003) for details about SEM). Based on literature, the proposed theoretical framework and data availability, seven exogenous variables and eight endogenous variables are included in the model. The specification of endogenous variables and exogeneous variables are listed in Table 3. For hukou, use local urban as one category and the rest as another category (Zhao and Lu, 2010), because local urban hukou has always been the most privileged type in Beijing. As discussed in the introduction part, residents holding local urban hukou are advantageous in terms of obtaining indemnificatory housing, purchasing commodity housing, being employed by SOEs and receiving housing assistance from SOEs, as well as applying for vehicle ownership quota. Consequently, hukou status may have both direct and indirect impacts on travel behavior, and the indirect impacts are imposed through residential location, community type, employment and car ownership. In consideration that the sampled districts in Beijing can be classified and depicted by sequential annular (Ring Road) areas, residential location is defined by the Ring Road area it belongs to, as displayed in Table 2. According to the classification of typical residential types in urban China, the community types in this study cover traditional community (traditional form of residential space, mainly formed by old neighborhoods with long history, distinctive and traditional spatial structure and architectural form, such as hutong, a traditional Chinese courtyard house) (Wang, 2002), danwei community, policy-related (social welfare) housing community and commodity housing community (refer to Wang et al. (2011) for further information about different housing units). For monthly household income, the original data only provides information about which incomes classes the respondents belong to, without the exact amount of monthly household income. To retain its nature as a continuous variable, each household is assigned the midpoint value of the 12 consecutive income classes it belongs to. No latent variables and measurement models are included because

The data used for this study was collected through in-door personto-person interviews in Beijing from September to November 2016. It covers the 6 central districts and 6 inner suburban districts of Beijing and the spatial distribution of sampled households within these districts is shown in Fig. 1. A multi-level probability-proportional-to-size (PPS) sampling method was used to sample these districts, with sampling levels conforming to the territorial administrative hierarchy. In the first level, the sample size of each district (qu) is in proportion to the total number of households in the district; likewise, in the second level, the sample size of each sub-district (jiedao) is in proportion to the total number of households in the sub-district. During the interview, household head and other members aged over 12 were required to fill in the questionnaire, which contains information about individual/ household level socio-demographic and socio-economic characteristics, residential and built environment characteristics, individual's two-day activity-travel diary (one weekday and one weekend) and so on. In total, 1884 respondents from 800 households were successfully interviewed. Table 2 lists the profile of the sampled household heads, together with their weekday activity-travel behavior. Among them, local hukou holders (combination of local urban and local rural hukou) constitute 68.7%, indicating a proportion of 31.3% for non-local hukou holders. The percentage of non-local hukou holders is slightly lower than that of National Bureau of Statistics, because the targeted households are those who have a stable residence in Beijing and mostly live together with family members. Cases like joint tenancy with non-family members are not covered, while these residents are more likely to be non-local hukou holders. 5. Multivariate analysis 5.1. Model structure and variables Fig. 2 presents the conceptual modeling framework. In consideration that hukou may have an impact on workers in terms of employment sector and jobs-housing relationship, which then exert influence on travel behavior. But these flows of influence do not apply to nonworkers, therefore, we select workers into the sample for modeling. Besides, only the representative household head will be selected to 125

Transport Policy 69 (2018) 122–131

M. Yao, D. Wang

Table 2 The profile of sampled household heads. Variable

Classification

Individual characteristics Gender Male Female Age 19–29 30–39 40–49 > 50 Educational attainment Low (lower than bachelor degree) High (bachelor degree or higher) Marital status Married Other Hukou status Local urban Local rural Non-local urban with “WRP″ Non-local rural with “WRP″ Non-local urban with “TRP″ Non-local rural with “TRP″ Employment Employed/self-employed (SOE& GIO) Employed/self-employed (non SOE&GIO) Other Driving license Have Don't have Mean time allocation on Subsistence weekday Maintenance Recreation Mean trip frequencies By car on weekday By public transit By other Mean travel time on weekday Household characteristics Presence of child (ren) under 6 Yes 182 No 618 Monthly household income < 15,000 103 15,000–24,999 356 25,000–39,999 230 > 40,000 111 Household size < =2 329 3 406 > =4 65 Household head Male head 445 Female head 355 Car ownership 0 228 1 522 >1 50 Housing tenure Owner 442 Renter and others 358 Residential location Within 2nd Ring Road 38 2nd to 3rd Ring Road 118 3rd to 4th Ring Road 205 4th to 5th Ring Road 175 5th to 6th Ring Road 205 Beyond 6th Ring Road 59 Neighbourhood type Traditional 135 Danwei 84 Commercial 531 Policy-related 47 Other 3

Cases

%

445 355 183 304 227 86 156 644 732 68 453 97 94 27 93 36 238

55.6% 44.4% 22.9% 38.0% 28.4% 10.8% 19.5% 80.5% 91.5% 8.5% 56.6% 12.1% 11.8% 3.4% 11.6% 4.5% 29.8%

538

67.3%

Fig. 2. Model structure.

776 head workers are used to calibrate the model. The baseline model has a chi-square of 2272.81 with 150 degrees of freedom, while the estimated model is 29.031 with 33 degrees of freedom. The goodnessof-fit indices suggest that the model fit the data reasonably well: the Root Mean Square Error of Approximation (RMSEA) is 0.037, smaller than 0.05; The Tucker Lewis Index (TLI) and Comparative Fit Index (CFI) are 0.924 and 0.993, both are larger than 0.9 and the latter is close to one. Table 4 shows the total, direct and indirect effects among endogenous variables, while Table 5 shows the total, direct and indirect effects of exogeneous variables upon endogenous variables. Details about the modeling results and findings will be explained in the following paragraphs.

24 3.0% 687 85.9% 113 14.1% 6.97(h) 1.64(h) 0.33(h) 1.21 0.69 0.56 1.39(h)

22.80% 77.30% 12.90% 44.50% 28.80% 13.90% 41.10% 50.80% 8.10% 55.6% 44.4% 28.50% 65.30% 6.30% 55.30% 44.80% 4.80% 14.80% 25.60% 21.90% 25.60% 7.40% 16.90% 10.50% 66.40% 5.90% 0.40%

5.2.1. Impact of hukou status The holding of local urban hukou is found to have significant and positive direct effects on employment sector, suggesting that local urban residents are more likely to work in government and institutional organizations (GIOs) and state-owned enterprises (SOEs). SOEs and GIOs usually provide higher wage, better corporate welfare and greater job security in comparison with other enterprises. This finding is in line with the employment discrimination from GIOs and SOEs on the grounds of hukou status, which was prevalent decades ago. Previously, some SOEs require their employees to have local urban hukou (Beijing Labor Bureau, 1989; Fang and Chan, 2000). Although the discrimination has been reduced in recent years, it still has certain influence in the labour market. As indicated by the total effects, the holding of local urban hukou is demonstrated to have significant impacts on community type. Specifically, local urban residents have higher probability of living in traditional communities, which is understandable, because they are born in Beijing, and can inherit traditional housing properties from last generations, or can live together with last generations in the traditional compounds such as hutongs. Another reason is that traditional compounds are mainly located within the 2nd Ring Road area (Fang and Zhang, 2003), which is the urban core of Beijing, suggesting a higher percentage of local urban residents than local rural residents in this area. Local urban hukou holders are more prone to live in danwei communities, partly due to the reason that some SOEs provide cheap housing at a low price to employees with local urban hukou, while very few of them provide housing assistance to temporary workers (Ma, 2004; Zhao and Lu, 2010). Owing to the privileges that local urban residents have over indemnificatory housing as described in the introduction, they are more likely to live in social welfare housing communities than others. These findings confirm earlier revelation that the hukou system constitutes an institutional barrier to the housing availability of non-local residents (Wu, 2004). Local urban hukou holding is proved to have significant and

multicollinearity problems among the specified variables are non-significant. 5.2. Modeling results Asymptotically distribution-free weighted least squares (ADF-WLS) estimation methods is adopted to accommodate the existence of discrete endogenous variables, which violates the assumption of normal distribution and maximum likelihood estimation method becomes unsuitable (Golob and McNally, 1997). The estimation is carried out by Mplus Version 8 that excels at handing non-normal variables. A total of 126

Transport Policy 69 (2018) 122–131

M. Yao, D. Wang

Table 3 Specification of variables. Category

Variable name

Description

Endogenous Hukou status Employment sector Community type

Hukou Employment

Residential location Jobs-housing relationship Car ownership Modal frequency

Traditional Danwei Policy Residential Ring Road Jobs-housing distance Car ownership Car frequency Public transit frequency Other frequency Travel time

1 = local urban hukou, otherwise 0 1 = work in SOEs or GIOs, otherwise 0 Commodity housing community serves as the base 1 = live in traditional community, otherwise 0 1 = live in danwei community, otherwise 0 1 = live in policy-related (social welfare) housing community, otherwise 0 Indicate which Ring Road area (annular region) the residents live Distance between residence and workplace 1 = private car owner, otherwise 0 Daily trip frequency by car Daily trip frequency by public transit Daily trip frequency by other modes, mainly non-motorized modes Daily total travel time (hour)

Child Gender Age Household income Household size Education Marriage

1 = presence of child of age 0–6 years old in the household, otherwise 0 1 = male, 0 = female Assign the midpoint of the class that a respondent belongs to as his/her age Assign the midpoint of the class that a household belongs to as the monthly household income Number of household members 1 = the respondent obtains a bachelor degree or higher, otherwise 0 1 = married, otherwise 0

Daily travel time Exogenous Child presence Gender of respondent Age of respondent Household income Household size Education level of respondent Marriage status of respondent

The significant and negative relationship between traditional community and residential location implies that traditional neighborhoods are usually located at lower Ring Roads, similar to the observations by other researchers (Fang and Zhang, 2003; Gu and Shen, 2003). For other community types, their impacts on residential distribution patterns are not significant. Residing in traditional communities are proved to have significant and negative impact on car ownership. This revelation is in consistency with the finding by Silva (2014) that car ownership levels are negatively influenced by living in a central, dense and traditional area. Owing to the location advantage of traditional communities (lower rings, closer to CBD and better access to amenities), the demand for private car ownership is less. Moreover, traditional communities are mostly built in early years when private cars were not popular and affordable to many people; therefore, planners generally overlooked parking space when designing these communities, and the inconvenience induced by parking in these traditional residential areas might be another explanation of the lower car ownership rate (Wang et al., 2011). The lower car ownership level and location advantage of traditional compounds contribute to reduced car trips and increased nonmotorized trips, as indicated by the significantly negative correlation between traditional community and car trip frequency and the significantly positive correlation between traditional community and trip frequency by other modes. For residents living in danwei compounds, they are also revealed to be more inclined to travel by non-motorized modes and less likely to travel by motorized modes, which is in agreement with the findings by Wang and Chai (2009) that dwellers in danwei compounds had higher usage of non-motorized transport mode. One possible reason is that they have better job-housing proximity, and prefer to travel by nonmotorized modes for shorter commuting distance, which is the major component of a worker's daily travel (Zhao and Lu, 2010). This is also evidenced by the significant and negative impact of job-housing distance on trip frequency by other modes. Living farther to CBD (higher Ring Roads) will have a significant positive influence on jobs-housing distance, resulted from the ring structure of the sampled region. Besides, living farther to CBD indicates longer daily travel distance as well as longer travel time. It is in line with previous research findings and the distance between home and the urban center was claimed to be an important determinant of travel distance (e.g. Naess, 1995; Hickman and Banister, 2004). An underlying reason is that urban households usually have less spatially dispersed

negative influence on jobs-housing relationship, indicating that local urban residents are likely to have better home-work proximity in comparison with other types of hukou holders, partly because they have higher chances of living in danwei compounds, which are usually located close to their workplaces (Wang and Chai, 2009; Fan et al., 2014). As expected, holding of local urban hukou is revealed to be significantly and positively correlated with car ownership, partly due to reasons that local urban residents are more financially capable of owning cars (Wang and Zuo, 1999) and policies concerning vehicle purchase impose more restrictions towards non-local residents, as mentioned in the introduction. Regarding travel behavior, as indicated by the direct effects, urban natives are found to travel less by car and public transit, but more by non-motorized modes. One possible explanation is that they usually live in the built environments with highly mixed land use and high accessibility with spatially clustered activity patterns (Khattak and Rodriguez, 2005; Tana et al., 2016). It is worthwhile to note that local urban hukou holding has significant positive mediating effects on trip frequency by car, which is channeled mainly through car ownership. This finding is reasonable because local urban residents have significantly higher car ownership level. Besides, local urban hukou holding is found to have both direct and total negative impacts on the total travel time, signifying that local urban residents spend less time on daily travel than others, which tends to be the time saving result of residing closer to CBD, better access to amenities, higher car ownership level and better home-work proximity closely linked with commuting trip (the major component of a worker's daily travel) (Zhao and Lu, 2010). 5.2.2. Impact of other endogenous variables Working in SOEs and GIOs would significantly increase the possibility of residing in danwei compounds, owing to the situation that SOEs and GIOs are more likely to provide housing assistance to their employees in comparison with other enterprises (Zhang and Rasiah, 2014). The housing assistance is usually provided in terms of danwei compounds with job-housing proximity. Consequently, workers of SOEs and GIOs tend to have shorter work-home distance. In addition, danwei compound could provide an institutional context that integrate urban productive activities and social life in the same location, through a local mixture of housing, jobs, hospitals, schools and other urban services, which further contributes to the reduced daily travel time (Ren, 2002; Qiao, 2004; Zhao and Lu, 2010). 127

Policy

Danwei

Traditional

128

Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total

– – – – – – – – – 0.171** – 0.171** – – – – 0.011 0.011 −0.050 −0.022 −0.072* 0.086 0.020 0.106 −0.109 0.035 −0.075 0.069 0.010 0.079 −0.043 0.044 0.001 −0.034 −0.021 −0.055*

– – – 0.152** – 0.152** 0.090* – 0.090* – 0.026* 0.026* 0.376*** – 0.376*** 0.007 0.000 0.007 – −0.057* −0.057* 0.331*** −0.009 0.322*** −0.248*** 0.293*** 0.045 −0.038 −0.122 −0.160*** 0.160* −0.030 0.129* −0.040 −0.094* −0.134***

– – – – – – – – – – – – – – – −0.178*** – −0.178*** 0.069 −0.085** −0.016 −0.412*** −0.027 −0.439*** 0.178** −0.391*** −0.214*** −0.012 0.094* 0.082 0.090** 0.054** 0.144*** −0.057 0.064 0.007

Traditional – – – – – – – – – – – – – – – 0.063 – 0.063 −0.157** 0.030 −0.127** 0.144 −0.001 0.143 −0.192** 0.115 −0.077 0.016 −0.075 −0.059 0.109** 0.056 0.165*** 0.139** −0.085 0.054

Danwei

Community type

– – – – – – – – – – – – – – – 0.037 – 0.037 −0.045 0.018 −0.027 0.040 −0.005 0.035 −0.017 0.033 0.016 0.007 0.040 0.047 −0.133 0.008 −0.125 0.056 −0.040 0.016

Policy – – – – – – – – – – – – – – – – – – 0.476*** – 0.476*** 0.144* 0.038 0.182*** −0.120* 0.142** 0.022 0.005 −0.017 −0.012 0.058 −0.071* −0.013 0.029 0.052* 0.082***

Residential Ring Road

– – – – – – – – – – – – – – – – – – – – – 0.081 – 0.081 −0.042 0.062 0.020 0.037 0.050 0.088** −0.140** −0.019 −0.159*** 0.211*** −0.013 0.198***

Jobs-housing distance

Note: '*': significant at 0.10 level; '**': significant at 0.05 level; '***': significant at 0.01 level. All activities are out-of-home.

Travel time

Otherfrequency

Public transit frequency

Car frequency

Car ownership

Jobs-housing distance

Residential Ring Road

Community type

Employment

Hukou

Employment

Hukou

Table 4 Total, direct and indirect effects between endogenous variables.

– – – – – – – – – – – – – – – – – – – – – – – – 0.844*** 0.166** 1.010*** −0.175 −0.260* −0.435*** – −0.289*** −0.289*** −0.253* 0.276** 0.023

Car ownership

– – – – – – – – – – – – – – – – – – – – – – – – – – – −0.388** 0.111 −0.277* −0.346* −0.005 −0.351*** 0.576*** −0.172** 0.404***

Car frequency

– – – – – – – – – – – – – – – – – – – – – – – – −0.297** −0.039 −0.337** – – – −0.140 0.095 −0.044 0.503*** −0.132 0.371***

Public transit frequency

– – – – – – – – – – – – – – – – – – – – – – – – −0.127 0.117 −0.010 −0.455 −0.033 −0.489* – – – 0.303*** −0.220 0.083

Other frequency

– – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

Travel time

M. Yao, D. Wang

Transport Policy 69 (2018) 122–131

Transport Policy 69 (2018) 122–131

M. Yao, D. Wang

Table 5 Total, direct and indirect effects of exogenous variables on endogenous variables.

Hukou

Employment

Community type

Traditional

Danwei

Policy

Residential Ring Road

Jobs-housing distance

Car ownership

Car frequency

Public transit frequency

Other frequency

Travel time

Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total

Child

Gender

Age

Household income

Household size

Education

Marriage

– – – – – – −0.178 – −0.178 0.426** – 0.426** −0.277 – −0.277 −0.033 0.048 0.015 0.143 −0.032 0.112 −0.008 0.135 0.126 0.080 −0.026 0.054 0.085 0.071 0.157 −0.238* 0.012 −0.227* 0.002 0.087 0.089

−0.316*** – −0.316*** 0.170* −0.048* 0.122 0.328*** −0.028 0.300*** −0.342*** 0.021 −0.321** 0.519*** −0.119** 0.400** 0.315*** −0.061 0.254*** 0.017 0.128 0.145 −0.094 −0.200* −0.293** 0.006 −0.115 −0.109 −0.099 0.097 −0.002 0.118 −0.084 0.034 0.062 0.027 0.089*

0.177*** – 0.177*** −0.002 0.027* 0.025 0.042 0.016 0.058 0.173** 0.004 0.177** −0.093 0.066** −0.027 −0.016 0.001 −0.015 0.062 −0.028 0.033 0.133 0.062 0.195** 0.066 0.127* 0.194*** – −0.070 −0.070 −0.083* −0.008 −0.090** −0.001 0.019 0.018

0.159*** – 0.159*** −0.004 0.024** 0.020 −0.111** 0.014 −0.097* −0.087** 0.003 −0.084* −0.257*** 0.060 −0.197*** −0.084** 0.006 −0.078** −0.068** −0.003 −0.071** 0.406*** 0.057 0.464*** −0.295*** 0.389*** 0.095** 0.040 −0.098** −0.058* −0.076** 0.014 −0.062** 0.116* −0.152*** −0.036*

0.219** – 0.219** – 0.033** 0.033** 0.048 0.020 0.067 −0.176 0.006 −0.171 −0.082 0.082 0.001 0.016 −0.021 −0.005 −0.005 0.027 0.022 0.430*** 0.024 0.454*** −0.361*** 0.401*** 0.039 – −0.112* −0.112* – 0.020 0.020 0.165** −0.176** −0.011

0.179 – 0.179 0.282** 0.027 0.309** – 0.016 0.016 – 0.053* 0.053* −0.609*** 0.067 −0.542*** −0.128 −0.018 −0.146 0.141 −0.014 0.128 0.529*** 0.054 0.583*** −0.175 0.448*** 0.273** 0.124 −0.171** −0.047 −0.023 −0.020 −0.043 0.137* −0.046 0.091

0.542*** – 0.542*** 0.116 0.082* 0.199 −0.003 0.049 0.045 −0.413** 0.034 −0.379* −0.411 0.204** −0.207 0.256 −0.036 0.220 0.042 0.187 0.229 1.005*** 0.165 1.170*** −0.333 0.989*** 0.656*** 0.084 −0.253 −0.169 −0.315** −0.154 −0.469*** 0.349** −0.185 0.164*

relative long-distance (non-walking distance) travel.

weekday activity patterns than their suburban counterparts, suggesting that individuals residing in suburban areas usually travel much farther to conduct various activities to meet personal and household needs, which induces greater need of private car ownership as well (Schönfelder and Axhausen, 2003; Páez et al., 2010; Van Acker and Witlox, 2010; Tana et al., 2016). The higher car ownership level for residents living farther to CBD may also be attributed to the traffic congestion problem, since traffic congestion in areas close to CBD is severe and has become a big headache for Beijing residents (Liu et al., 2018). To some extent, this may discourage residents living close to CBD from buying cars and take the subway instead for daily travel. A longer jobs-housing distance would lead to more time spending on daily travel as demonstrated by the significant and positive correlation between jobs-housing distance and travel time. A possible reason is that commuting trips constitute the major component of workers’ daily travel. Likewise, the trip frequency by non-motorized modes would significantly decrease for long jobs-housing distance. As expected, higher household car ownership level guarantees higher car availability; therefore, car ownership is found to have significant and positive impact on trip frequency by car, but have significant and negative impact on trip frequency by public transit and by other modes. In terms of the transport frequency by car, by public transit and by other modes, all of them tend to have significant and positive effects on travel time, which is reasonable because it is obvious that the increased trip frequency by any of these modes will lead to more travel time spending. It is noticeable that trip frequency by car, by public transit and by other modes are negatively correlated with each other. In particular, trip frequency by car is found to significantly and negatively impact the trip frequency by public transit, and vice versa, due to the strong substitutability between these two types of modes for

5.2.3. Impact of exogenous socio-demographics The casual effects of exogeneous socio-demographics on endogenous variables are shown in Table 5. It is worth to be noted that the presence of child under 6 years old tends to significantly reduce the trip frequency by non-motorized modes, possibly because heavier travel burden will be induced when travelling together with young children by non-motorized modes like walking and biking. With regard to gender impacts, male workers are more likely to be employed by SOEs and GIOs, which seems to be caused by the gender preference in these organizations, as widely studied by other researchers (Nie et al., 2002; Dong and Pandey, 2012). In comparison with female workers, male workers are found to have longer jobshousing distance and spend much more time on travel. Similar findings were reported in previous research as well (Stead, 2001; Schwanen et al., 2004; Van Acker and Witlox, 2010). Partly because women take more household maintenance responsibilities and thus are more likely to work near home. Old age people are more probable to be local Beijing residents, because migrants are mostly young people that are able to find a job and adapt to a new environment. In addition, old age people are more likely to be employees in SOEs and GIOs, and are more likely to reside in danwei communities. Before the 1980s, state-owned or collectivelyowned danwei used to be the dominant type of employment organization, and workers mostly lived in danwei communities provided by their employment organization. The market-oriented economic reform introduced since the 1980s has gradually removed the non-essential function of housing provision from danwei and changed the dominance of danwei type in employment organizations, with the emergence of 129

Transport Policy 69 (2018) 122–131

M. Yao, D. Wang

country. This research extends the literature by showing that the institutional factor hukou is an important determinant of employment, residence, mobility and travel behavior. As expected, local urban residents are proved to have a significant higher probability of being employed by government and institutional organizations (GIOs) and state-owned enterprises (SOEs), be more likely to reside in danwei communities and social welfare communities, maintain better home-work proximity, travel more by non-motorized modes and spend less time on daily travel. Apart from the influence of hukou, other interesting findings are also reported in this study. In terms of the influence of the built environment on car ownership and travel, people living in traditional communities are revealed to have lower probability of car ownership, travel significantly less by car and significantly more by non-motorized modes. For residents living in danwei compounds, they are also revealed to be more inclined to travel by non-motorized modes and less likely to travel by motorized modes. Meanwhile, longer jobs-housing distance and travel time would be induced if residents live far away from CBD. Findings in this research improve the understanding about the complex causal relationships among hukou, built environment, employment, mobility and travel behavior in urban China during the transformation of economy system, with massive rural-to-urban migration. They can serve as guidance for alteration of hukou system, design of transportation improvement plans and formulation of effective policies by addressing various needs of residents with different socio-economic characteristics, e.g. gender, income, and hukou type. These plans and policies can promote sustainable urban development through the connection with spatial distribution of households, regulation of car ownership, promotion of non-motorized travel and so on. Specifically, several aspects concerning these plans and policies can be identified. First, unlike residents holding local urban hukou, local rural and non-local hukou holders are less likely to use non-motorized transport modes, indicating more room for improvement with respect to promoting active transportation among them. Therefore, customized policy intervention measures can be designed to change their daily travel behavior towards greener mobility. For example, measures can be taken to reduce the discrimination against them regarding employment and housing, so that they could have higher chances of residing at places with better access to amenities and better home-work proximity. Second, results show that residents living in policy communities are less likely to use non-motorized modes for their daily travel, suggesting that when allocating indemnificatory housing, more considerations can be given to improve the balance between housing location and typical activity locations of dwellers. Besides, when designing the location of indemnificatory housing, proximity to employment centers is important to encourage a shift from motorized travel towards non-motorized travel. Third, local urban hukou holders are shown to have significantly higher car ownership level in comparison with other residents. This implies that car usage rationing policies are more likely to be effective towards this group of people and special attention should be paid to the characteristics of this group when formulating and implementing car usage rationing policies. For future research, some interesting extensions can be further explored regarding the linkages among hukou, mobility and travel behavior. Firstly, this study treats hukou holding as a binary variable, with local urban hukou holders as one group and the rest as another group, on the basis that local urban residents are the most privileged group. Since local rural residents, non-local residents holding “WRP” and nonlocal resident holding “TRP” may also differ in their employment, built environment, jobs-housing relationship, mobility and activity-travel behavior, it is worthwhile to treat them separately and investigate the varied impacts of these hukou statuses. Secondly, as household members in the same household may hold different types of hukou, signifying that a family member holding non-local hukou may benefit from another family member holding local hukou; therefore, analysis can be carried out across different household members to incorporate

other organization types including private enterprises, foreign enterprises, etc. (Wang and Chai, 2009). These organizations seldom offer housing assistance to their employees and the housing system reform has transferred the responsibility for housing provision from work units to the market (Ye et al., 2017). Consequently, only those old workers employed by state-owned enterprises are more probable to live in danwei communities. Because wealth accumulates with age, the elderly is more financially well-off for purchasing cars, thus they have higher percentage of car ownership than young people, and their car trip frequency is much higher due to the mediating effect of car ownership. Household income is demonstrated to be positively correlated with local urban hukou holding, partly because local urban residents are the most privileged group (Wang and Zuo, 1999). High-income households are less likely to live in traditional communities, danwei communities and social welfare communities because commodity houses are more affordable to them. Besides, they are more inclined to live adjacent to CBD (lower rings) with higher housing price but shorter jobs-housing relationship. Understandably, car ownership rate is also higher among high-income families, which is frequently evidenced by previous studies (Kockelman, 1997; Dargay, 2001; Soltani, 2005; Van Acker and Witlox, 2010). With higher car ownership, high income households travel more by cars, and less by public transit and other modes. As for household size, members in larger household size are more probable to be local urban hukou holders, partly because it is more likely for several generations of local residents to live together. Besides, households with more members tend to have higher probability of car ownership, similar to findings reported by Kim and Kim (2004). Individuals with higher education level are more likely to be employed by SOEs and GIOs, which is in support of findings by other researchers (e.g. Zhao, 2002). Consequently, the well-educated individuals have better chance of living in danwei communities because SOEs and GIOs are more likely to provide housing assistance to employees (Zhang and Rasiah, 2014). As highly educated workers have higher propensity of engaging in more specialized jobs that are generally concentrated in high-density or central business district office parks, they tend to have longer jobs-housing distance, higher trip frequency by car and longer travel time (Van Acker and Witlox, 2010). Married workers have higher probability of owning cars because car is more or less a necessity to Chinese households nowadays, with similar finding revealed by Van Acker and Witlox (2010). If people get married, they tend to travel more by auto and less by non-motorized modes, partly because getting married is always accompanied by purchasing a new house in China, which might be far away from both working places and CBD due to the skyrocketing housing price in Beijing and therefore could lead to longer travel distance as well as longer travel time. Another reason is that when people get married, the job-housing proximity tend to compromise because proximities to both spouses’ workplaces have to be considered (Wachs et al., 1993). 6. Discussion and conclusion Nowadays, urban China is undergoing the process of hukou system alteration and rural-urban migration. In this process, institutional factors like hukou that links individuals’ access and entitlement to stateprovided benefits and opportunities are likely to play a crucial role in explaining the mobility and travel behavior of Chinese inhabitants. Although existing studies have identified a wide range of determinants of mobility and travel behavior, the role of institutional factors has not yet received much research attention. To test the importance of institutional factor hukou on mobility and travel behavior, this study empirically investigated the direct and indirect effects of hukou on mobility and travel behavior using structural equations model, with indirect effects exerted through built environment and employment. The data is collected from Beijing, the capital and political center of China that has attracted a multitude of migrants from all over the 130

Transport Policy 69 (2018) 122–131

M. Yao, D. Wang

intrahousehold interactions. Thirdly, the present study focuses on the impact of hukou on daily travel behavior as a whole. For travel behavior analysis, differentiations can be made among travelling for different purposes. Another interesting perspective is to investigate the influence of mobility in residence and employment on jobs-housing relations and commuting behavior, which necessitates the collection of retrospective or panel data.

University Library arXiv preprint arXiv:1803.10686. Lu, X., Pas, E.I., 1999. Socio-demographics, activity participation and travel behavior. Transport. Res. Pol. Pract. 33 (1), 1–18. Logan, J.R., Fang, Y., Zhang, Z., 2009. Access to housing in urban China. Int. J. Urban Reg. Res. 33 (4), 914–935. Ma, L.J., 2004. Economic reforms, urban spatial restructuring, and planning in China. Prog. Plann. 61 (3), 237–260. Naess, P., 1995. Travelling distances, modal split and transportation energy in thirty residential areas in Oslo. J. Environ. Plann. Manag. 38 (3), 349–370. Nie, W., Hopkins, W.E., Hopkins, S.A., 2002. Gender-based perceptions of equity in China's state-owned enterprises. Thunderbird Int. Bus. Rev. 44 (3), 353–377. Páez, A., Gertes Mercado, R., Farber, S., Morency, C., Roorda, M., 2010. Relative accessibility deprivation indicators for urban settings: definitions and application to food deserts in Montreal. Urban Stud. 47 (7), 1415–1438. Qiao, Y., 2004. History flux of Beijing unit yard and its effect on Beijing urban space. Hua Zhong Jian Zhu (in Chinese) 5, 120–124. Qin, P., Wang, L., 2018. Job opportunities, institutions, and the jobs-housing spatial relationship: case study of Beijing. Transport Pol (in press). Ren, S., 2002. On the integration of Danwei compound urban land space. Planners (in Chinese) 11, 60–63. Scheiner, J., Holz-Rau, C., 2013. Changes in travel mode use after residential relocation: a contribution to mobility biographies. Transportation 40 (2), 431–458. Schönfelder, S., Axhausen, K.W., 2003. Activity spaces: measures of social exclusion? Transport Pol. 10 (4), 273–286. Schwanen, T., Dieleman, F.M., Dijst, M., 2004. The impact of metropolitan structure on commute behavior in The Netherlands: a multilevel approach. Growth Change 35 (3), 304–333. Silva, J., 2014. Spatial self-selection in land-use–travel behavior interactions: accounting simultaneously for attitudes and socioeconomic characteristics. J. Transp. Land Use 7 (2), 63–84. Soltani, A., 2005. Exploring the impacts of built environments on vehicle ownership. In: Proceedings of the Eastern Asia Society for Transportation Studies, vol. 5. pp. 2151–2163. Song, Y., 2014. What should economists know about the current Chinese hukou system? China Econ. Rev. 29, 200–212. Song, Y., 2016. Hukou-based labour market discrimination and ownership structure in urban China. Urban Stud. 53 (8), 1657–1673. Spinney, J.E., Scott, D.M., Newbold, K.B., 2009. Transport mobility benefits and quality of life: a time-use perspective of elderly Canadians. Transport Pol. 16 (1), 1–11. Stead, D., 2001. Relationships between land use, socioeconomic factors, and travel patterns in Britain. Environ. Plann. Plann. Des. 28 (4), 499–528. Tana, Kwan, M.P., Chai, Y., 2016. Urban form, car ownership and activity space in inner suburbs: a comparison between Beijing (China) and Chicago (United States). Urban Stud. 53 (9), 1784–1802. Van Acker, V., Van Wee, B., Witlox, F., 2010. When transport geography meets social psychology: toward a conceptual model of travel behaviour. Transport Rev. 30 (2), 219–240. Van Acker, V., Witlox, F., 2010. Car ownership as a mediating variable in car travel behaviour research using a structural equation modelling approach to identify its dual relationship. J. Transport Geogr. 18 (1), 65–74. Wachs, M., Taylor, B.D., Levine, N., Ong, P., 1993. The changing commute: a case-study of the jobs-housing relationship over time. Urban Stud. 30 (10), 1711–1729. Wang, Y., 2002. A case study on urban community in Shanghai: community classification, spatial distribution and changing tendencies. Urban Plann. Forum 6, 33–40. Wang, D., Chai, Y., 2009. The jobs–housing relationship and commuting in Beijing, China: the legacy of Danwei. J. Transport Geogr. 17 (1), 30–38. Wang, D., Chai, Y., Li, F., 2011. Built environment diversities and activity-travel behavior variations in Beijing, China. J. Transport Geogr. 19, 1173–1186. Wang, D., Law, F., 2007. Impacts of Information and Communication Technologies (ICT) on time use and travel behavior: a structural equations analysis. Transportation 34 (4), 513–527. Wang, F., Zuo, X., 1999. Inside China's cities: institutional barriers and opportunities for urban migrants. Am. Econ. Rev. 89 (2), 276–280. Wu, W., 2004. Sources of migrant housing disadvantage in urban China. Environ. Plann. 36 (7), 1285–1304. Yao, Y., Wang, S., 2014. Commuting tools and residential location of suburbanization: evidence from Beijing. Urban, Plann. Transp. Res. 2 (1), 274–288. Ye, D., Zhang, J., Xu, G., 2017. Peripherization of indemnificatory housing community under land-centered urban transformation: the case of nanjing, China. Sustainability 9 (4), 656. Zhang, M., Rasiah, R., 2014. Institutional change and state-owned enterprises in China's urban housing market. Habitat Int. 41, 58–68. Zhao, Y., 2002. Earnings differentials between state and non-state enterprises in urban China. Pac. Econ. Rev. 7 (1), 181–197. Zhao, P., Howden-Chapman, P., 2010. Social inequalities in mobility: the impact of the hukou system on migrants' job accessibility and commuting costs in Beijing. Int. Dev. Plann. Rev. 32 (3–4), 363–384. Zhao, P., Lu, B., 2010. Exploring job accessibility in the transformation context: an institutionalist approach and its application in Beijing. J. Transport Geogr. 18 (3), 393–401. Zhao, P., Lü, B., De Roo, G., 2011. Impact of the jobs-housing balance on urban commuting in Beijing in the transformation era. J. Transport Geogr. 19 (1), 59–69.

Acknowledgements This research is sponsored by two GRF grants from the Hong Kong Research Grant Council (HKBU247813 and HKBU12606215). References Beijing Labor Bureau, 1989-1998. A Collection of Selected Labor Policies and Regulations (In Chinese). Beijing Labor Bureau, Beijing. Bollen, K.A., 1989. Structural Equations with Latent Variables. Wiley, New York. Cao, X., Mokhtarian, P.L., Handy, S.L., 2007. Do changes in neighborhood characteristics lead to changes in travel behavior? A structural equations modeling approach. Transportation 34 (5), 535–556. Chan, K.W., 2009. The Chinese hukou system at 50. Eurasian Geogr. Econ. 50 (2), 197–221. Chan, K.W., Zhang, L., 1999. The hukou system and rural-urban migration in China: processes and changes. China Q. 160, 818–855. Dargay, J.M., 2001. The effect of income on car ownership: evidence of asymmetry. Transport. Res. Pol. Pract. 35 (9), 807–821. Dong, X.Y., Pandey, M., 2012. Gender and labor retrenchment in Chinese state owned enterprises: investigation using firm-level panel data. China Econ. Rev. 23 (2), 385–395. Fan, C.C., 2008. Migration, hukou, and the city. China urbanizes: Consequences, Strat. Policies 65–89. Fan, Y., Allen, R., Sun, T., 2014. Spatial mismatch in Beijing, China: implications of job accessibility for Chinese low-wage workers. Habitat Int. 44, 202–210. Fang, C., Chan, K.W., 2000. The Political Economy of Urban Protectionist Employment Policies in China. Fang, K., Zhang, Y., 2003. Plan and market mismatch: urban redevelopment in Beijing during a period of transition. Asia Pac. Viewp. 44 (2), 149–162. Gärling, T., Gillholm, R., Gärling, A., 1998. Reintroducing attitude theory in travel behavior research: the validity of an interactive interview procedure to predict car use. Transportation 25 (2), 129–146. Golob, T.F., 2003. Structural equation modeling for travel behavior research. Transp. Res. Part B Methodol. 37 (1), 1–25. Golob, T.F., McNally, M.G., 1997. A model of activity participation and travel interactions between household heads. Transp. Res. Part B Methodol. 31 (3), 177–194. Gu, C., Shen, J., 2003. Transformation of urban socio-spatial structure in socialist market economies: the case of Beijing. Habitat Int. 27 (1), 107–122. Guo, F., Iredale, R., 2004. The impact of hukou status on migrants' employment: findings from the 1997 Beijing migrant census. Int. Migrat. Rev. 38 (2), 709–731. Guo, Y., Wang, J., Peeta, S., 2016. The impacts of household registration and head of household travel behavior on travel mode choice in China. In: Transportation Research Board 95th Annual Meeting (No. 16–0693). Handy, S., Cao, X., Mokhtarian, P., 2005. Correlation or causality between the built environment and travel behavior? Evidence from Northern California. Transport. Res. Transport Environ. 10 (6), 427–444. Hickman, R., Banister, D., 2004. Reducing travel by design: urban form and the commute to work. In: Paper Presented at the AESOP Conference, Grenoble, France. Huang, X., 2011. Urban and Rural Development Report of Beijing (2010–2011). Beijing Social Science Literature Press, Beijing. Huang, D., Liu, Z., Zhao, X., 2015. Monocentric or Polycentric? The urban spatial structure of employment in beijing. Sustainability 7 (9), 11632–11656. Khattak, A.J., Rodriguez, D., 2005. Travel behavior in neo-traditional neighborhood developments: a case study in USA. Transport. Res. Part A 39 (6), 481–500. Kim, H.S., Kim, E., 2004. Effects of public transit on automobile ownership and use in households of the USA. Rev. Urban Reg. Dev. Stud. 16 (3), 245–262. Kockelman, K., 1997. Travel behavior as function of accessibility, land use mixing, and land use balance: evidence from San Francisco Bay Area. Transport. Res. Rec.: J. Transp. Res. Board 1607, 116–125. Levin, M.W., Boyles, S.D., 2015. Effects of autonomous vehicle ownership on trip, mode, and route choice. Transport. Res. Rec.: J. Transp. Res. Board 2493, 29–38. Li, L., Li, S.M., Chen, Y., 2010. Better city, better life, but for whom?: the hukou and resident card system and the consequential citizenship stratification in Shanghai. City, Cult. Soc. 1 (3), 145–154. Li, S.M., Liu, Y., 2016. The jobs-housing relationship and commuting in Guangzhou, China: hukou and dual structure. J. Transport Geogr. 54, 286–294. Li, S.M., Liu, Y., 2017. Land use, mobility and accessibility in dualistic urban China: a case study of Guangzhou. Cities 71, 59–69. Liu, X., Chow, J.Y., Li, S., 2018. Online Monitoring of Local Taxi Travel Momentum and Congestion Effects Using Projections of Taxi GPS-based Vector Fields. Cornell

131