Advertisement

Natural Hazards

, Volume 99, Issue 3, pp 1431–1453 | Cite as

Energy consumption by rural migrant workers and urban residents with a hukou in China: quality-of-life-related factors and built environment

  • Ying Jiang
  • Linghan Zhang
  • Junyi ZhangEmail author
Original Paper

Abstract

This paper compares energy consumption (electricity, gasoline, and gas) by rural migrant workers and urban residents with a hukou (a China-specific household registration system) and influential factors (including quality-of-life [QOL]-related factors, built environment, and individual and household attributes) in China. A questionnaire survey was conducted in Dalian (a coastal city) in 2014 and in Guiyang (an inland city) in 2015, respectively. A zero-inflated negative binomial (ZINB) model was applied to understand whether and how much people consume a certain type of energy. The results showed that built environment explains 8.4–21.8% and 6.3–41.4% of the total variance in energy consumption by rural migrant workers and urban residents with a hukou. The corresponding variance related to QOL-related factors was 9.1–15.8% and 4.1–22.6%, respectively. The built environment was mostly associated with electricity consumed by urban residents with a hukou, while its influences on other types of energy consumption were moderate. Mixed effects of both built environment and QOL-related factors on reducing energy consumption were observed. Thus, it is context-sensitive whether and how much compact city development and social security policy affect residents’ energy consumption.

Keywords

Energy consumption Electricity Gasoline Gas Rural migrant workers Urban residents with a hukou China Zero-inflated negative binomial model Built environment QOL-related factors 

1 Introduction

The global trend to urbanization is characterized by population explosion and massive use of various resources, including energy resources. Urbanization in China is largely due to rural–urban migration (Chen and Song 2014). China established a special household registration system called hukou to manage the flow of rural migrants into its cities (Chan and Zhang 1999). The hukou system was first introduced in cities in 1951, extended to rural areas in 1955, and then formalized as a permanent system in 1958. Even though significant modifications were made during the 1980s and 1990s, the hukou system remains essentially unchanged in nature and has had a remarkable effect on Chinese people’s lives and fates. Before economic reforms were implemented in the late 1970s, the hukou system was the most decisive factor in rural–urban migration in China. The traditional hukou system means that an urban resident’s right to use public services depends on whether he/she has an urban hukou or not. Rural migrants do not have an urban hukou; therefore, they have unequal access to public services in cities (e.g., childhood education, medical treatment, social security). Under China’s National New-type Urbanization Plan (2014–2020) (hereafter, CNP), the traditional hukou system was phased out slowly. The total number of rural migrant workers in China reached 286.52 million 3 years after the introduction of the CNP in 2014, which is an increase of 9.05 million since 2014 (National Bureau of Statistics of China 2017).

To predict future energy consumption in Chinese cities, it is important to understand the differences in energy consumption and its influential factors between urban residents with a hukou and rural migrant workers. Unfortunately, the literature does not provide easy answers. In principle, the CNP allows rural migrant workers to become a permanent resident of any city in China; however, Jiang et al. (2017) showed that only 39% of migrant workers intended to permanently reside in cities, which was partially due to the influence of quality-of-life (QOL)-related factors (e.g., unstable income, unemployment risks, medical issues). Unfortunately, the existing studies are unclear on how such QOL-related factors were associated with actual energy consumption. As Anderson et al. (2015) observed, the built environment, which is associated with the building and transportation sectors, is responsible for 62% of final energy consumption and 55% of greenhouse gas emissions globally. Thus, the role of built environment in reducing energy consumption has become increasingly important (e.g., Ward 2008; Ding et al. 2017); however, little is known about energy consumption by rural migrant workers. To reveal these above unknowns, this study compares the energy consumption by rural migrant workers and urban residents with a hukou, focusing on the roles of QOL-related factors and built environment in their energy consumption. Questionnaire data were collected from two Chinese cities (one coastal and one inland) in 2014 and 2015, respectively, and econometric models were adopted to quantify the influences of various factors on energy consumption.

The rest of this paper is structured as follows. Section 2 provides a brief literature review. Section 3 introduces the questionnaire surveys conducted in urban residents with a hukou and rural migrant workers residing in two Chinese cities in 2014 and 2015. Using the collected data, Sect. 4 clarifies the differences in energy consumption between the above two resident groups. Section 5 describes two econometric models: i.e., a zero-inflated negative binomial (ZINB) model and a Poisson model. Section 6 explains the modeling estimation results. Section 7 compares and discusses the modeling results. Finally, Sect. 8 concludes this study with a discussion of future research issues.

2 Literature review

As the largest consumer of energy worldwide, China is responsible for 23.2% of the global energy consumption and contributed 33.6% of the global energy demand growth in 2017.1 Even though the current urbanization rate in China was 58.52% in 2017,2 this rate is 37.91% when rural migrant workers in cities are excluded. Since 2014, the Chinese government has actively promoted the transformation of rural migrant workers into permanent urban residents. As a result, those migrant workers’ living standards will be improved, which may accelerate the growth of energy consumption in China.

2.1 Energy consumption of rural migrant workers

An extensive literature review found a very limited number of studies related to rural migrant workers’ energy consumption in China.

Due to the lack of statistical data for migrant workers in China, Fan et al. (2016) combined several national data sources and estimated that, in 2007, energy consumption by urban residents with a hukou was 3.06 times higher than that by rural migrant workers, which is 1.57 times higher than that by rural people. Unfortunately, the energy data was estimated, but not directly collected from rural migrant workers. Zhang et al. (2016a) analyzed annual data (from 1995 to 2011) for energy consumption published in the China Energy Statistical Yearbook and found that the gaps of energy consumption between rural and urban areas gradually became smaller each year due to the growing influence of urban lifestyles on rural lifestyles. However, Zhang et al. (2016a) did not analyze rural migrant workers’ energy consumption. Based on a questionnaire survey, Wang et al. (2017) found that rural migrant workers increased rural households’ gas and electricity consumption through the impacts of the increased income of rural migrant workers and their family characteristics, i.e., economic status, consumer preference, and regional factors.

In contrast to the above studies, Wei et al. (2014) and Jiang et al. (2017) conducted questionnaire surveys with rural migrant workers. Targeting Shanghai, Wei et al. (2014) concluded that not only household attributes (e.g., income and household size), but also residential environment and travel behavior, such as residential location and commute modes, also have a significant influence on migrant workers’ energy consumption, especially their gasoline usage behavior. Jiang et al. (2017) examined migrant workers’ future intentions to become a permanent urban resident and its influence on their future energy consumption preferences. Even though Jiang et al. (2017) did not focus on actual energy consumption, they presented a systematic review that covered migration, urbanization, and hukou in China in addition to determinants of general energy consumption behavior (e.g., sociodemographics, lifestyle, attitudes, beliefs, policies, and technologies) and studies on rural migrants’ energy consumption behavior. However, their review did not cover research on energy consumption and its associations with QOL-related factors and built environment.

No previous study compared energy consumption between rural migrant workers and urban residents with a hukou based on data collected directly from these population groups.

2.2 QOL-related factors for rural migrant workers

Xie and Jiang (2016) showed that the improvements in human capital, higher wages, and greater job stability encouraged rural migrant workers to transfer or abandon their land in rural areas. On the one hand, this land transfer or abandonment may have further driven rural migrant workers to eventually leave their rural hometowns and become permanent urban residents. On the other hand, migration from rural areas to cities left behind 61 million children (29 million living with neither parent and over 2 million living alone) in rural areas, which comprises 38% of children in rural China and 22% of all Chinese children, according to Zhao et al. (2017). Zhong et al. (2015) found that lower education, worse living conditions, poorer self-perceived physical health, migration before adulthood, infrequently calling family members, and having worked in several jobs are risk factors for lifelong major depressive disorders. Ning and Qi (2017) found that self-employed migrants (25% of the total migrant workers in 2009) were less covered by urban social securities and more discriminated against by urban hukou. As a result, they tend to lose their desire to become urban citizens. However, Cheng et al. (2014) argued that discrimination between rural and urban workers due to hukou has been declining. Cheng et al. (2014) compared rural migrant workers (i.e., rural-to-urban migrants) and urban-to-urban migrants and found that the latter were more likely to sign a labor contract than the former, the latter had higher participation rates in social insurance than the former, and having a labor contract had a greater impact than hukou status in determining whether Beijing’s floating population accesses social insurance. Lu et al. (2016) confirmed that lack of occupational protection measures and regular health check-ups and longer working hours worsened migrant workers’ general health after controlling for the following variables: age, sex, education, working hours, signing the labor contract, and buying insurance. Zhu et al. (2018) further showed that limitations on insurance coverage were strongly related to rural migrant workers’ hospitalization for schizophrenia. The above family matters, health issues, and social security issues may discourage rural migrants from staying in cities permanently. All the above-mentioned factors are called QOL-related factors in this study, because they indicate people’s quality of life, partially.

2.3 General energy research associated with built environment

Anderson et al. (2015) reviewed the literature on energy analysis of the built environment at both building and urban scales and argued the necessity of addressing the interscale interactions within the built environment. Although such interactions are not the focus of this study, both in-home and out-of-home energy consumption are examined; therefore, in this study, in-home energy mainly refers to electricity and gas and out-of-home energy mainly refers to gasoline. In the case of rural migrant workers, no study has examined how the built environment affects their energy consumption. Nevertheless, some indirect observations include Cheng and Wang’s (2013) observation that many rural migrant workers in China were concentrated in disadvantaged neighborhoods and physically and socially suffered from their poor residential environment, which affected their wages. Generally speaking, income is known to be associated with energy consumption; therefore, the above study may also imply that the built environment affects rural migrant workers’ energy consumption. Existing studies also examined the effects of built environment on choice of employment location (Liu et al. 2014), and residential segregation and commuting patterns (Zhu et al. 2017). Zhu et al. (2017) confirmed that urbanizing villages provided relatively good accessibility to job opportunities and various built environment characteristics were also related to travel behavior.

2.4 Other factors

Cui et al. (2012) revealed that wage and pension benefits as well as unobserved factors related to the migration decision affected the employment choice of rural migrant workers. Zhang et al. (2016b) showed that the income gap between rural migrant workers and urban residents with a hukou can be partially explained by differences in educational attainment, work experience, and distribution across industry, occupation, and ownership of enterprises. Even though the above factors were examined without any connection with energy consumption, they need to be incorporated into studies of rural migrant workers’ energy consumption.

All these reviews suggest that this study is the first attempt to compare energy consumption between rural migrant workers and urban residents with a hukou in China, especially from QOL-related and built environment perspectives.

3 Survey and data collection

We conducted a questionnaire survey in this comparative study. Both urban residents with a hukou and rural migrant workers were asked to report their energy consumption (monetary expenditure [RMB] and/or amount [kwh or m3] of electricity, gas, and gasoline during the four annual seasons), built environment factors measured in terms of distances to the nearest daily facilities (i.e., bus stop, railway station, supermarket, department store, park, hospital, preschool, elementary school, secondary school, high school, and workplace), QOL-related factors measured in terms of insurance purchases, residential attributes (residential location, residential duration, house type, and housing size), and individual and household attributes (e.g., ages of household members, sex, education level, marital status, annual income, and household size).

The development of coastal Chinese cities was strongly promoted nationwide due to the late 1970s reform and opening-up policy. As a result, the growth of inland cities has been slow. Therefore, we conducted our questionnaire survey in a coastal city, Dalian in Liaoning Province in November 2014 (monthly average temperature: −1 to 25 °C), and an inland city, Guiyang in Guizhou Province in June 2015 (monthly average temperature: 6 to 22 °C), respectively. The respondents were recruited randomly by controlling for the age, sex, and location distributions of the total population as much as possible. Rural migrant workers were recruited from various sectors, such as construction, housekeeping, transportation and logistics, manufacturing, food processing, information technology, wholesale and retail trade, and beauty salons. The survey was mainly based on a face-to-face interview to collect reliable data. Therefore, we employed students from the universities in the two cities as interviewers and trained them before implementing the survey. The respondents who could not be interviewed face-to-face answered the questionnaire by themselves. Each item of all returned questionnaires was checked by the employed interviewers to guarantee the quality of the survey, especially to avoid missing data as much as possible. We provided respondents with monetary incentives to encourage them to provide reliable information. As a result, we collected valid data from 667 rural migrant workers (324 in Dalian and 343 in Guiyang) and 671 urban residents with a hukou (330 in Dalian and 341 in Guiyang). The data show that the average household income of urban residents with a hukou is 1.11 times higher than that of rural migrants. According to Dreger and Zhang (2017), who conducted a survey of Chinese household income in collaboration with the National Statistical Bureau, Germany, the average wage of urban residents with a hukou is 1.28 times higher than that of rural migrants. Considering that rural migrants usually work longer than urban residents with a hukou, the above existing income gaps between the two population groups partially support the use of the data collected in this study.

4 Energy consumption of urban residents with a hukou and rural migrants

In the survey, either monetary expenditure (RMB) or amount (kwh or m3) of energy consumption was recorded. This analysis focused on the energy consumption in terms of monetary consumption. Therefore, in the case that only amount (kwh or m3) of energy consumption is available, the data are transformed into monetary expenditure based on the charging rates of the corresponding energy type(s) in the two cities. Three types of energy consumption were targeted in this study: i.e., electricity, gas, and gasoline. In relation to their usage, the survey included items of ownership of different types of end-use products, including refrigerator, washing machine, TV, microwave, air conditioner, personal computer, rice cooker, and vehicles (i.e., car, motorcycle, electric bicycle). Even though the purposes of different end uses were not investigated, electricity is mainly used by in-home electronic end users and electric bicycles, gas for cooking and heating, and gasoline for vehicles.

The aggregated results show that among the three types of energy consumption, gasoline takes the largest share: i.e., 70.1% for urban residents with a hukou and 55.7% for rural migrant workers (Table 1). In comparing the two cities, more gasoline was consumed in Guiyang than in Dalian by both types of respondents. This is partially because the income levels of both rural migrant workers and urban residents with a hukou in Guiyang (rural migrant workers: 4432 RMB/month; urban residents with a hukou: 4514 RMB/month) is higher than those in Dalian (rural migrant workers: 3454 RMB/month; urban residents with a hukou: 4173 RMB/month), even though the income gaps between two cities are relatively small.
Table 1

Differences of energy consumption between rural migrant workers and urban residents with a hukou

Type of energy

Guiyang City

Dalian City

Total

Rural migrant workers (A)

Urban residents with a hukou (B)

Ratio B/A

Rural migrant workers (A)

Urban residents with a hukou (B)

Ratio B/A

Rural migrant workers (A)

Urban residents with a hukou (B)

Ratio B/A

Cost

Percent

Cost

Percent

Cost

Percent

Cost

Percent

Cost

Percent

Cost

Percent

Electricity

79

31.84

96.5

20.10

1.2

60.2

35.33

94.9

19.15

1.6

69.9

33.22

95.7

19.62

1.4

Gasoline

156.2

62.96

349.3

72.74

2.2

76

44.60

334.2

67.45

4.4

117.2

55.70

341.8

70.08

2.9

Gas

12.9

5.20

34.4

7.16

2.7

34.2

20.07

66.4

13.40

1.9

23.3

11.07

50.2

10.29

2.2

Comparing the three types of energy, it is obvious that urban residents with a hukou consume more energy than rural migrant workers do in both cities (Guiyang: 1.2–2.7 times higher; Dalian: 1.6–4.4 times higher). In Dalian, urban residents with a hukou consume 4.4. times more gasoline than rural migrant workers do. In contrast, the corresponding ratio in Guiyang is 2.2 times more. In contrast, electricity usage shows the least difference, where the consumption ratio of urban residents with a hukou relative to rural migrant workers is 1.2 in Guiyang and 1.6 in Dalian. In comparison with rural migrant workers, gas consumption for urban residents with a hukou is 2.7 and 1.9 times higher in Guiyang and Dalian, respectively. In summary, the energy consumption of rural migrant workers and urban residents with a hukou vary with energy types.

5 Energy consumption distribution and ZINB model

Figure 1 shows the distribution histograms of energy consumption (i.e., electricity, gasoline, and gas) separately for rural migrant workers and urban residents with a hukou. Other than electricity consumption by urban residents with a hukou, not a small number of respondents reported zero consumption: i.e., 667 rural migrant workers indicated 66 zero observations (9.9%) of electricity consumption, 527 zero observations of gasoline consumption (79.0%), and 420 zero observations of gas consumption (63.0%), while 671 urban residents with a hukou indicated 358 zero observations of gasoline consumption (53.4%) and 271 zero observations of gas consumption (40.4%). As expected, there are obviously more zero observations with respect to rural migrant workers. Figure 1 also shows the energy consumption distributions excluding zero observations. Because of the distributions with zero observations, this study adopted a ZINB model (Greene 1994; Yau et al. 2003; Xu et al. 2015) to represent gasoline and gas consumption. For electricity, even though 9.9% of the rural migrant workers reported zero consumption, for the sake of comparison with urban residents with a hukou, this study adopted a classical Poisson model for electricity consumption.
Fig. 1

Distributions of energy consumption by rural migrant workers and urban residents with a hukou

The ZINB model assumes two distinct data generation processes. A Bernoulli trial is used to determine which of the two processes is used. For observation i, with probability πi, the only possible response of the first process is zero counts (Yi= 0), and with probability of 1 − πi, the response of the second process is governed by a negative binomial with mean λi and an overdispersion parameter k. Thus, the zero counts are generated from the first and second processes and the overall probability of zero counts is the combined probability of zeros from the two processes.
$$Pr\left( {Y_{i} = 0} \right) = \pi_{i} + \left( {1 - \pi_{i} } \right)(1 + k\lambda_{i} )^{ - 1/k}$$
(1)
$$Pr\left( {Y_{i} = y_{i} } \right) = \left( {1 - \pi_{i} } \right)\frac{{\varGamma (y_{i} + 1/k)}}{{\varGamma (y_{i} + 1)\varGamma (1/k)}}\frac{{(k\lambda_{i} )^{{y_{i} }} }}{{(1 + k\lambda_{i} )^{{y_{i} + 1/k}} }},\quad \, y_{i} = 1,\;2, \ldots$$
(2)
And the mean and variance of Yi are calculated as follows:
$$E\left( {Y_{i} } \right) = \left( {1 - \pi_{i} } \right)\lambda_{i}$$
(3)
$$V\left( {Y_{i} } \right) = \left( {1 - \pi_{i} } \right)\lambda_{i} \left( {1 + \lambda_{i} \left( {\pi_{i} + k} \right)} \right)$$
(4)
When the overdispersion parameter k approaches zero, the ZINB model is reduced to a zero-inflated Poisson model.
With the above probabilities, the log-likelihood function of the ZINB model is given by the following Eq. (5).
$$\begin{aligned} {\text{LL}} & = \sum\nolimits_{{Y_{i} = 0}} {\ln \left( {\pi_{i} + (1 - \pi_{i} )(1 + k\lambda_{i} )^{ - 1/k} } \right)} \\ & \quad + \sum\nolimits_{{Y_{i} = y_{i} }} {(\ln (1 - \pi_{i} ) + \ln \varGamma (y_{i} + 1/k)} \\ & \quad - \ln (\varGamma (y_{i} + 1)\varGamma (1/k)) + y_{i} \ln (k\lambda_{i} ) - (y_{i} + 1/k)\ln (1 + k\lambda_{i} )). \\ \end{aligned}$$
(5)

6 Modeling estimation

6.1 Characteristics of dependent and independent (explanatory) variables

Table 2 shows the average values of the three dependent variables (i.e., electricity, gasoline, and gas) and four categories of independent (explanatory) variables (i.e., built environment attributes, QOL-related factors, residential attributes, and individual and household attributes) for both rural migrant workers and urban residents with a hukou. Even though energy data in four seasons were collected, we use the average monthly data as dependent variables in order to mitigate the impacts of missing energy data.
Table 2

Dependent and explanatory variables for energy consumption analysis

Explanatory variables

Average values

Category

Variable name

Abbreviation

Rural migrant workers

Urban residents with a hukou

Ratio

Energy consumption

Electricity (RMB per month)

Electricity

69.87

95.72

1.37

Gasoline (RMB per month)

Gasoline

117.25

341.85

2.92

Gas (RMB per month)

Gas

23.26

50.17

2.16

Built environment attributes (distances from home to the nearest daily facilities: km)

Bus stop

D_bus_stop

0.83

0.68

0.82

Department store

D_store

0.84

0.87

1.04

Supermarket

D_super

1.98

1.82

0.92

Park

D_park

3.64

2.75

0.76

Hospital

D_hospital

2.78

2.17

0.78

Preschool

D_preschool

1.77

1.17

0.66

Elementary school

D_eleschool

2.06

1.19

0.58

Secondary school

D_secschool

2.90

1.99

0.69

High school

D_higschool

4.52

3.18

0.70

QOL-related factors: insurance purchased (1: yes, 0: no)

Pension insurance

Pension_ins

45.9%

73.0%

1.59

Medical insurance

Medical_ins

64.8%

86.4%

1.33

Unemployment insurance

Unemp_ins

23.1%

49.6%

2.15

Employment injury insurance

Injury_ins

36.7%

51.4%

1.40

Residential attributes

Residence city (1: Dalian, 0: Guiyang)

Res_city

48.6%

49.2%

1.01

Living in a rental house (1: yes, 0: no)

Rt_house

68.0%

22.0%

0.32

Housing ownership (1: yes, 0: no)

House_own

13.8%

56.6%

4.10

Housing size: 100 m2

House_size

0.58

1.16

2.00

Individual and household attributes

Number of children: persons

Child_num

0.88

0.85

1.00

Food expenditure share (%)

Food_share

28.4%

30.1%

1.06

Age (years old)

Age

35.4

37.3

1.06

Marital status

Marital

61.9%

78.4%

1.27

Household income (10,000 RMB/year)

Income

4.75

5.28

1.11

Lower education level (1: secondary school or lower, 0: others)

Low_edu

56.7%

20.0%

0.35

Urban residents with a hukou were found to consume more energy than rural migrant workers (1.37 times for electricity, 2.92 times for gasoline, and 2.16 times for gas). Much fewer rural migrant workers owned a house (13.8%) than urban residents with a hukou (56.6%); as a result, there was a much bigger number of rural migrant workers renting a house (68.0%) compared with urban residents with a hukou (22.0%). Other aggregate analyses (results are not shown in Table 2) further confirmed that on average, respondents with self-owned houses consumed more energy than those living in a rental house. For gasoline consumption, the average consumption by those with self-owned houses was 2.2 times and 3.43 times higher than those living in rental houses for both types of respondents. The housing size of urban residents with a hukou (116 m2) was almost twice larger than that of rural migrant workers (58 m2). More rural migrants (56.7%) had a lower education level than urban residents with a hukou. Similarly, the individual income of migrants on average was 47,500 RMB/year and that of residents with a hukou was 52,500 RMB/year. No remarkable differences were observed in the number of children and food expenditure. Urban residents with a hukou lived in a comparatively better built environment than migrant workers because their distances from home to the nearest daily facilities were shorter, except for the distance to the nearest department store. As for QOL-related factors, more urban residents with a hukou purchased insurance: i.e., 1.33–2.15 times higher than that for rural migrant workers. In particular, a much higher share of urban residents with a hukou purchase unemployment insurance.

6.2 Model estimation results

In the case of the ZINB model, explanatory variables are needed for both the zero consumption (i.e., zero inflation) and nonzero consumption aspects. In this study, only residential, individual, and household attributes were employed to predict the zero inflation, while the nonzero consumption aspect includes all four categories of explanatory variables shown in Table 2. The model estimation results are presented in Table 3 to compare the three types of energy consumption based on the ZINB model, except for the electricity model. Considering that electricity consumption does not include any zero observations, we tried to estimate both the negative binomial and Poisson models and found that the latter model performed better than the former model. Accordingly, in this study, we adopted the Poisson model to represent electricity consumption.
Table 3

Model estimation results for electricity, gasoline, and gas consumption by rural migrant workers and urban residents with a hukou

Explanatory variables

Models

Electricity (Poisson)

Gasoline (ZINB)

Gas (ZINB)

Rural migrant workers

Urban residents with a hukou

Rural migrant workers

Urban residents with a hukou

Rural migrant workers

Urban residents with a hukou

Coef.

Sig.

Variance share

Coef.

Sig.

Variance share

Coef.

Sig.

Variance share

Coef.

Sig.

Variance share

Coef.

Sig.

Variance share

Coef.

Sig.

Variance share

The nonzero consumption part

Built environment attributes

D bus stop

− 0.01

 

0.1%

0.01

***

1.0%

− 0.09

 

0.7%

0.20

***

3.4%

0.04

 

0.4%

0.23

**

5.8%

D_store

− 0.04

***

2.8%

0.02

***

2.1%

− 0.14

 

1.5%

− 0.07

 

0.5%

0.02

 

0.1%

0.02

 

0.0%

D_super

0.00

 

0.0%

0.01

***

3.4%

0.01

 

0.1%

0.03

 

0.2%

− 0.01

 

0.1%

0.004

 

0.0%

D_park

− 0.001

 

0.0%

− 0.01

***

5.2%

0.02

 

0.2%

− 0.04

**

1.1%

− 0.01

 

0.1%

− 0.02

 

0.4%

D_hospital

0.00

**

0.3%

0.02

***

8.7%

0.05

 

1.5%

0.01

 

0.0%

− 0.02

 

0.7%

0.08

***

3.9%

D_preschool

− 0.01

*

0.2%

0.02

***

5.0%

0.13

 

4.1%

0.07

 

0.7%

0.06

**

2.5%

0.08

**

1.2%

D_eleschool

− 0.02

***

2.1%

− 0.04

***

12.3%

− 0.03

 

0.4%

− 0.02

 

0.1%

− 0.05

 

2.2%

0.09

 

1.0%

D_secschool

− 0.02

***

3.1%

− 0.01

***

1.3%

0.12

**

7.1%

0.01

 

0.1%

0.03

 

1.6%

− 0.04

 

0.5%

D higschool

− 0.010

***

3.5%

− 0.01

***

2.4%

− 0.09

**

6.3%

0.02

 

0.2%

− 0.02

 

0.8%

− 0.004

 

0.0%

Sub-total of variance explained: A

  

12.0%

  

41.4%

  

21.8%

  

6.3%

  

8.4%

  

12.9%

QOL-related factors: Insurances purchased

Pension_ins

0.04

***

0.5%

0.11

***

7.3%

0.33

 

1.0%

0.44

**

1.6%

− 0.12

 

0.4%

− 0.32

*

1.1%

Medical_ins

0.03

***

0.2%

− 0.04

***

0.7%

0.27

 

0.6%

0.56

***

1.5%

0.36

***

2.9%

1.01

***

6.4%

Unemp_ins

− 0.18

***

5.7%

0.04

***

1.4%

1.02

**

7.0%

− 0.21

 

0.5%

− 0.47

**

4.0%

− 0.85

***

9.9%

Injury ins

− 0.10

***

2.6%

0.01

 

0.1%

0.33

 

1.0%

− 0.28

 

0.8%

0.60

***

8.4%

0.62

***

5.2%

Sub-total of variance explained: B

  

9.1%

  

9.6%

  

9.6%

  

4.4%

  

15.8%

  

22.6%

Residential attributes

Res_city

− 0.17

***

7.0%

− 0.03

***

0.7%

− 0.16

 

0.2%

0.17

 

0.3%

0.38

**

3.7%

0.07

 

0.1%

Rt_house

0.37

***

29.5%

0.03

**

0.4%

1.23

***

12.7%

1.41

***

14.2%

1.13

***

28.5%

0.24

 

0.6%

House own

0.35

***

15.0%

0.09

***

6.3%

1.58

***

11.4%

0.91

***

8.5%

0.88

***

9.4%

0.03

 

0.0%

House size

0.00

***

17.6%

0.000

***

13.9%

0.01

**

8.1%

0.001

*

2.4%

0.01

***

9.9%

0.002

**

32.5%

Sub-total of variance explained: C

  

69.1%

  

21.3%

  

32.4%

  

25.4%

  

51.5%

  

33.1%

Individual and household attributes

Child_num

0.05

***

2.1%

0.03

***

1.4%

0.05

 

0.1%

− 0.31

***

2.2%

0.21

**

3.8%

0.06

 

0.1%

Food_share

0.18

***

0.9%

0.31

***

8.3%

4.02

***

17.9%

3.48

***

13.6%

0.63

*

1.2%

3.13

***

14.4%

Age

0.00

***

0.3%

0.00

 

0.2%

0.05

***

9.3%

0.10

***

42.6%

0.04

***

13.8%

0.05

***

15.0%

Marital

0.14

***

4.8%

0.11

***

6.1%

0.37

 

1.2%

0.22

 

0.3%

0.21

 

1.1%

0.30

*

0.9%

Income

0.01

***

1.5%

0.02

***

11.7%

0.12

***

6.3%

0.10

***

4.8%

0.06

***

4.3%

0.04

**

0.8%

Low edu

0.03

***

0.3%

0.02

 

0.2%

− 0.38

 

1.4%

− 0.20

 

0.3%

− 0.07

 

0.1%

− 0.13

 

0.2%

Sub-total of variance explained: D

  

9.9%

  

27.8%

  

36.1%

  

63.9%

  

24.3%

  

31.3%

 

Total variance = A + B + C + D

  

100.0%

  

100.0%

  

100.0%

  

100.0%

  

100.0%

  

100.0%

The zero consumption part: inflation model

Residential attributes

Res_city

      

0.10

 

0.2%

− 0.17

 

0.1%

− 2.57

***

51.4%

− 1.97

***

78.4%

Rt_house

      

− 0.39

 

2.8%

0.35

 

0.5%

− 1.54

***

16.2%

0.18

 

0.4%

House_own

      

− 1.16

***

13.3%

− 0.52

**

1.4%

− 2.29

***

19.5%

− 0.87

 

15.1%

House_size

      

− 0.003

 

1.3%

− 0.01

***

83.7%

− 0.003

 

0.6%

0.00

 

1.3%

Sub-total of variance explained: E

        

17.6%

  

85.8%

  

87.6%

 

95.3%

 

Individual and household attributes

Child_num

      

0.06

 

0.3%

0.21

 

0.6%

− 0.31

*

2.6%

− 0.03

 

0.0%

Food_share

      

− 0.07

 

0.0%

− 0.51

 

0.2%

− 1.00

*

0.9%

1.19

**

3.1%

Age

      

0.004

 

0.2%

0.02

*

0.8%

− 0.005

 

0.1%

0.001

 

0.0%

Marital

      

− 1.34

***

35.7%

− 0.47

*

0.8%

− 0.19

 

0.3%

− 0.28

 

1.1%

Income

      

− 0.21

***

44.1%

− 0.21

***

11.8%

− 0.15

***

7.8%

0.01

 

0.2%

Low_edu

      

0.32

 

2.2%

− 0.23

 

0.2%

0.30

 

0.7%

0.14

 

0.3%

Sub− total of variance explained: F

        

82.4%

  

14.2%

  

12.4%

  

4.7%

Constant term

      

3.55

***

0.0%

1.82

***

0.0%

4.97

***

0.0%

0.57

 

0.0%

 

Total variance = E + F

        

100.0%

  

100.0%

  

100.0%

  

100.0%

Zero-inflation parameters

/Inalpha

      

0.51

***

 

0.29

***

 

− 0.40

***

 

− 0.08

  

Alpha

      

1.66

***

 

1.34

***

 

0.67

***

 

0.93

***

 

Likelihood ratio for testing “ramda = 0”: chibar2 (Pr ≥ chibar2)

      

1.1E + 05 (0.0000)

2.3E + 05 (0.0000)

8731.51 (0.0000)

2.8E + 04 (0.0000)

Vuong test of zinb versus standard negative binomial: z value (Pr ≥ z)

      

3.07 (0.0011)

4.42 (0.0000)

12.97 (0.0000)

8.76 (0.0000)

Number of zero observations (total sample size)

66 (667)

0 (671)

527 (667)

358 (671)

420 (667)

271 (671)

Initial log-likelihood

− 13,663.90

− 13,096.53

− 1478.45

− 3176.47

− 1857.54

− 2916 22

Converged log-likelihood

− 10,730.86

− 11,696.09

− 1362.37

− 2918.27

− 1596 40

− 2617.31

Log-likelihood ratio: chi2

5866.08

2800.88

3218.16

9773.95

5874.45

7899.36

Adjusted Rho-squared

0.21

0.11

0.06

0.07

0.12

0.09

Rho-squared

0.21

0.11

0.08

0.08

0.14

0.10

*Significant at the 10% level, **at the 5% level, ***at the 1% level

All the log-likelihood ratio tests and adjusted Rho-squared values suggest that both the ZINB and Poisson models are acceptable and the Vuong tests indicate that the ZINB model performs better than the standard negative binomial model. Furthermore, both the zero-inflation parameters and parameters of explanatory variables introduced into the inflation model support the use of ZINB models because most of the parameters are statistically significant at the 10%, 5%, or 1% levels.

To reveal the size of the effects of explanatory variables on energy consumption, the ratio of total variance explained by each explanatory variable was calculated (also shown in Table 3).

6.2.1 Electricity consumption

The model estimation results show that rural migrant workers’ electricity consumption was mostly influenced by residential attributes, which accounted for 69.1% of the total variance. Living in a rental house or not explained 29.5% of the total variance, while housing size and ownership explained 17.6% and 15.0%, respectively. In contrast, the electricity consumption of urban residents with a hukou was mostly affected by built environment factors, which accounted for 41.4% of the total variance. The most influential factor is the distance to the nearest elementary school (12.3% of the total variance), followed by that to the nearest hospital (8.7%) and to the nearest park (5.2%) and preschool (5.0%).

Regarding individual and household attributes, unique citizenship-related factors could be identified. For urban residents with a hukou, household income played the most influential role in explaining their electricity consumption, which accounted for 11.7% of the total variance. The other attributes with a larger influence were food expenditure share (8.3%) and marital status (6.1%). For rural migrant workers, however, marital status showed the largest influence with 4.8% of the total variance.

The QOL-related factors show similar influences on the electricity consumption of both urban residents with a hukou and rural migrant workers in the sense that they explain 9.1% and 9.6% of the total variance of rural migrant workers and urban residents with a hukou, respectively. In contrast, rural migrant workers’ electricity consumption was mostly associated with unemployment insurance (5.7% of the total variance), while urban residents with a hukou paid the greatest attention to pension insurance (7.3%). Only 23.1% of rural migrant workers purchased unemployment insurance, which indicated that they were concerned about their employment in the city when making decisions on electricity consumption. However, urban residents with a hukou took more serious consideration of the influence of pension insurance, which was closely related to their future QOL.

6.2.2 Gasoline

Considering the inflation model, the key determinants for rural migrant workers’ gasoline consumption were household income and marital status, which contributed to 44.1% and 35.7% of the total variance, respectively. In contrast, housing size was the most influential factor for the gasoline consumption by urban residents with a hukou, which accounts for 83.7% of the total variance. In the case of rural migrant workers, the role of housing ownership is also relatively large because it can explain 13.3% of the total variance. The parameter signs for the above influential factors indicate that for urban residents with a hukou, households with smaller housing size are more likely to have zero gasoline consumption. For rural migrant workers, households with lower income, being single, and not owning a house tend to consume zero gasoline. Similar trends can also be observed with respect to urban residents with a hukou, even though their influences are minor compared with the influence of housing size.

Concerning the nonzero consumption part, both rural migrant workers and urban residents with a hukou show that individual and household attributes played the largest role in explaining their gasoline consumption, where the corresponding influence on the gasoline consumption by urban residents with a hukou (variance share: 63.9%) was larger than that for rural migrant workers (variance share: 36.1%). Among the individual and household attributes, age showed the largest influence on the gasoline consumption by urban residents with a hukou (variance share: 42.6%, which is much larger than the second-largest influential factor, i.e., food expenditure: 13.6%). However, the most dominant factor for rural migrant workers was their food expenditure (17.9%). The second-largest influential category was residential attributes, which explained 32.4% and 25.4% of the total variance of rural migrant workers and urban residents with a hukou, respectively. For both types of residents, whether to live in a rental house or not explained the largest share of the total variance among all the individual and household attributes. For rural migrant workers, built environment attributes also played a larger role, as indicated by their variance ratio (21.8%). The distances to nearest secondary and high schools were the most influential, explaining 7.1% and 6.3% of the total variance.

Considering the purchased insurance, only unemployment insurance affected rural migrant workers’ gasoline consumption, while pension and medical insurances were influential for urban residents with a hukou. Purchasing these types of insurance stimulated greater gasoline consumption.

6.2.3 Gas

In the case of zero gas consumption, the factor “living in Dalian” played a dominant role (variance shares: 51.4% for rural migrant workers and 78.4% for urban residents with a hukou). The negative sign of this factor implied that for both rural migrant workers and urban residents with a hukou living in Guiyang, they were more likely not to use gas in their daily life. Meanwhile, owning a house encouraged people to consume gas (variance shares: 19.5% for rural migrant workers and 15.1% for urban residents with a hukou). Living in a rental house largely affected zero gas consumption by rural migrant workers (variance share: 16.2%); however, it did not impose any significant influence on zero gas consumption of urban residents with a hukou.

In the case of nonzero gas consumption, residential attributes show the largest influence on both urban residents with a hukou and rural migrant workers, accounting for 33.1% and 51.5% of the respective total variance, where living in a rental house was most influential for the consumption of rural migrant workers; however, housing size most influenced the consumption of urban residents with a hukou. The second-largest influential category was individual and household attributes, which explained 31.3% and 24.3% of the total variance of urban residents with a hukou and rural migrant workers, respectively. Age was mostly influential to the consumption of both types of residents (variance ratios: 15.0% and 13.8% for urban residents with a hukou and rural migrant workers, respectively). In the case of urban residents with a hukou, the influence of food expenditure was also large (14.4%).

Two more differences shown by electricity and gasoline consumption were: (1) purchased insurance shows a much larger influence on gas consumption; and (2) the influential power of the built environment became the smallest among all four explanatory variable categories. Comparing the two types of residents, the two largest influential built environment attributes are the distances to the nearest bus stop and hospital for urban residents with a hukou and the distances to the nearest preschool and elementary school for rural migrant workers.

7 Comparison and discussion

All models shown in Table 3 indicate that the four categories of explanatory variables play different roles in explaining the three types of energy consumption by urban residents with a hukou and rural migrant workers.

First, the most influential factors to the three types of energy consumption are compared as follows.
  1. 1.

    Electricity: residential attributes account for 69.1% of the total variance for rural migrant workers and built environment attributes account for 41.4% for urban residents with a hukou.

     
  2. 2.

    Gasoline: individual and household attributes account for 36.1% and 63.9% of the total variance for rural migrant workers and urban residents with a hukou, respectively.

     
  3. 3.

    Gas: residential attributes account for 51.5% and 33.1% of the total variance for rural migrant workers and urban residents with a hukou, respectively.

     

Thus, residential attributes mainly determine electricity and gas consumption. This finding is understandable because it is related to in-home energy consumption. In the case of rural migrant workers, whether to live in a rental house or not consistently showed the largest influence on all three types of energy consumption, among the residential attributes. In contrast, housing size was most influential on the two types of in-home energy consumption, i.e., electricity and gas, of urban residents with a hukou. Concerning the consumption of gasoline (mainly for vehicle usage), which is mostly influenced by individual and household attributes, age was the most influential factor on the consumption by urban residents with a hukou; however, the gasoline consumption by rural migrant workers was mostly affected by food expenditure. Considering the built environment, distances to nearest elementary school and hospital were most influential on the electricity consumption by urban residents with a hukou.

7.1 Built environment

The built environment affects the three types of energy consumption; however, the size of its influence is much smaller than that for residential, individual, and household attributes, except for the electricity consumption by urban residents with a hukou. These distance variables mainly affected the electricity consumption of both types of respondents, rural migrant workers’ gasoline consumption, and the gas consumption of urban residents with a hukou. The corresponding variance shares were all larger than 10.0%. Most of the distances to respondents’ daily facilities mattered in the electricity consumption by both types of respondents; however, just a few distances were influential for gasoline and gas consumption.

Compact city development (e.g., to shorten distances to various facilities) has been emphasized as a key measure to realize sustainable development. Reducing the distance to the nearest bus stop surely leads to a reduction in energy consumption by urban residents with a hukou; however, it does not affect rural migrant workers’ energy consumption. Shortening the distance to the nearest supermarket reduced electricity consumption by urban residents with a hukou, but it was not influential on other types of energy consumption. Shortening the distances to the nearest hospital also leads to less energy consumption of electricity by both types of residents, and less consumption of gas by urban residents with a hukou, but this does not matter for other types of energy consumption. Thus, locating bus stops, supermarkets, and hospitals closer to residences is effective for reducing energy consumption. Unfortunately, the distance to the nearest park works in a consistently opposite way, i.e., shortening the distance only results in more energy consumption (electricity and gasoline by urban residents with a hukou); however, this does not matter for other types of energy consumption. Shorter distances to the elementary, secondary, or high schools consistently result in greater electricity consumption. Nevertheless, the effects of other distances on energy consumption were mixed.

7.2 QOL-related factors

The influence of purchased insurance can be interpreted as a proxy variable to explain how people are concerned about their future life or their preparedness for the future. This set of variables comprised 4.4–22.6% of the total variance, where gas consumption was largely influenced by such QOL-related factors: e.g., variance ratios were 22.6 and 15.8% for urban residents with a hukou and rural migrant workers, respectively. Unexpectedly, in the cases of electricity and gas (mainly in-home energy consumption), the influences of these factors were larger for urban residents with a hukou than for rural migrant workers. As expected, gasoline consumption by rural migrant workers was influenced more by QOL-related factors (9.6 and 4.4% of the total variance in the cases of rural migrant workers and urban residents with a hukou, respectively). However, the effects of different types of purchased insurance on the three types of energy consumption were mixed. In most cases, the preparedness for the future resulted in more energy consumption because, among the 17 statistically significant insurance variables, 11 variables had a positive parameter sign and there were only six variables with a negative parameter sign. In the case of gasoline, well-prepared rural migrant workers, and urban residents with a hukou consumed more gasoline in terms of unemployment insurance for rural migrant workers and pension and medical insurances for urban residents with a hukou.

7.3 Other factors

Even though household income was positively associated with energy consumption, it only showed a relatively large influence on the electricity consumption of urban residents with a hukou. Its influence on other types of energy consumption was very limited. In contrast to our expectations, the influences of marital status and number of children, which are two important factors defining household lifecycle stages, were limited.

8 Conclusions

The rapid urbanization of China is expected to further increase energy consumption globally. Since the release of CNP in 2014, China’s urbanization has moved into a new era and increasing numbers of rural migrant workers are becoming urban residents with a hukou. Some scholars have expressed great concern about the increasing amount of marginal energy consumption and CO2 emissions (e.g., Fan et al. 2016; Shen et al. 2018); however, limited research has been conducted to investigate the differences in energy consumption between rural migrant workers and urban residents with a hukou. To fill this research gap, this study examined three types of energy consumption (i.e., electricity, gasoline, and gas) according to the two population groups. A literature review further suggested that the effects of built environment and QOL-related factors were not well examined in the literature. In this study, questionnaire surveys were conducted in Dalian in 2014 and Guiyang in 2015, respectively, and the Poisson and ZINB models were employed to describe the respondents’ electricity consumption and the other two types of energy consumption, respectively. The model estimation results showed significant differences in the electricity, gasoline, and gas consumption, and their influential factors between rural migrant workers and urban residents with a hukou. The findings can be summarized as follows.

Residential and individual/household attributes played the most influential role in explaining all three types of energy consumption by both rural migrant workers and urban residents with a hukou. The built environment was mostly associated with electricity consumed by urban residents with a hukou, while its influences on other types of energy consumption were moderate. Compact city development in terms of shortening distances from residences to various facilities showed mixed effects on reducing energy consumption: i.e., (1) shortening the distances to the nearest bus stops, supermarkets, and hospitals from residences consistently contributed to less energy consumption; (2) reducing the distances to the nearest parks from residences led to greater energy consumption of electricity and gasoline by urban residents with a hukou; and (3) locating elementary/secondary/high schools closer to residences resulted in greater consumption of electricity. Both types of residents who purchased pension insurance consumed more energy. Rural migrant workers with unemployment insurance consumed less energy. Purchasing medical insurance was associated with greater energy consumption, except for electricity consumption by urban residents with a hukou. The effects of other insurance purchases on energy consumption varied with energy types for the two types of respondents.

This study contributes to the literature in several aspects. First, this is the first study to compare the energy consumption of rural migrant workers and urban residents with a hukou from QOL-related and built environment perspectives. Second, the mixed effects of compact city development were confirmed, which suggests that compact city development may not necessarily guarantee lower energy consumption compared with sprawling cities. Third, the effects of purchased insurance indicate that future social security policies may affect energy consumption both positively and negatively.

Having summarized the above findings, however, this research study was not free of limitations. Therefore, more research is needed in the future. First, the above findings should be further verified by using more data from other Chinese cities. Second, two existing econometric models were applied in this study; however, models with behavioral decision-making mechanisms may derive different findings from the same data. Therefore, such models may be worth developing to derive more sound insights into effective energy policies. Third, retrospective panel surveys should be implemented to capture the changes in rural migrant workers’ energy consumption behaviors after the CNP was released in 2014 by comparing rural migrants who obtained an urban hukou and those who have not obtained an urban hukou. Considering the interdependencies across various life choices (Zhang 2014, 2017), rural migrant workers’ intention to obtain an urban hukou may be influenced by the various barriers encountered in their daily lives, capabilities to survive in the cities, and family matters (e.g., childhood education, caregiving for elderly members). Studies have further confirmed that energy consumption behavior was influenced by residential and travel behavior (Yu et al. 2013), ownership and usage of in-home appliances (e.g., TV, refrigerator, microwave, washing machine) (Yu 2012; Yu et al. 2012a, b). Shen et al. (2018) further showed that rural migrant workers’ migration benefits ambient air quality at the national scale because they tend to shift to cleaner residential energy use upon arrival in urban areas, although their energy use is not as clean as that of urban residents with a hukou. The panel survey should comprehensively capture these above factors and behavioral changes. Finally, similar studies should be performed in the context of other developing countries with different levels of urbanization.

Footnotes

  1. 1.
  2. 2.

    For urbanization rates in China, see http://www.gov.cn/guowuyuan/2018-02/04/content_5263778.htm (in Chinese, Accessed November 9, 2018).

Notes

Acknowledgement

This research has been financially supported by a Grants-in-Aid for Scientific Research (B), Japan Society for the Promotion of Science (JSPS) (No. 26303003) and a Grants-in-Aid for Scientific Research (A), JSPS (No. 15H02271).

References

  1. Anderson JE, Wulfhorst G, Lang W (2015) Energy analysis of the built environment—a review and outlook. Renew Sustain Energy Rev 44:149–158CrossRefGoogle Scholar
  2. Chan KW, Zhang L (1999) The hukou system and rural–urban migration in China: processes and changes. China Q 160:818–855CrossRefGoogle Scholar
  3. Chen Q, Song Z (2014) Accounting for China’s urbanization. China Econ Rev 30:485–494CrossRefGoogle Scholar
  4. Cheng Z, Wang H (2013) Do neighbourhoods have effects on wages? A study of migrant workers in urban China. Habitat Int 38:222–231CrossRefGoogle Scholar
  5. Cheng Z, Nielsen I, Smyth R (2014) Access to social insurance in urban China: a comparative study of rural–urban and urban–urban migrants in Beijing. Habitat International 41:243–252CrossRefGoogle Scholar
  6. Cui Y, Tani M, Nahm D (2012) The determinants of employment choice of rural migrant workers in China: SOEs and non-SOEs. Procedia Econ Finance 1:98–107CrossRefGoogle Scholar
  7. Ding C, Liu C, Zhang Y, Yang J, Wang Y (2017) Investigating the impacts of built environment on vehicle miles traveled and energy consumption: differences between commuting and non-commuting trips. Cities 68:25–36CrossRefGoogle Scholar
  8. Dreger C, Zhang Y (2017) The Hukou impact on the Chinese wage structure. Discussion papers 1660, DIW Berlin, German Institute for Economic ResearchGoogle Scholar
  9. Fan J-L, Liao H, Tang B-J, Pan S-Y, Yu H, Wei Y-M (2016) The impacts of migrant workers consumption on energy use and CO emissions in China. Nat Hazards 81(2):725–743CrossRefGoogle Scholar
  10. Greene WH (1994) Accounting for excess zeros and sample selection in Poisson and negative binomial regression models. Working paper EC-94-10, Department of Economics, Stern School of Business, New York University. http://ideas.repec.org/p/ste/nystbu/94-10.html. Accessed 22 July 2018
  11. Jiang Y, Zhang J, Jin X, Ando R, Chen L, Shen Z, Ying J, Fang Q, Sun Z (2017) Rural migrant workers’ intentions to permanently reside in cities and future energy consumption preference in the changing context of urban China. Transp Res Part D: Transp Environ 52(B):600–618CrossRefGoogle Scholar
  12. Liu S, Xie F, Zhang H, Guo S (2014) Influences on rural migrant workers’ selection of employment location in the mountainous and upland areas of Sichuan, China. J Rural Stud 33:71–81CrossRefGoogle Scholar
  13. Lu L, Gong X, Zeng J (2016) Health status and migration: a propensity score matching with difference-in-difference regression approach. Lancet 388(Supplement 1):S60CrossRefGoogle Scholar
  14. National Bureau of Statistics of China (2017) Statistical Communique of the People’s Republic of China on the 2017 National Economic and Social Development. http://www.stats.gov.cn/english/pressrelease/201802/t20180228_1585666.html. Accessed 22 July 2018
  15. Ning G, Qi W (2017) Can self-employment activity contribute to ascension to urban citizenship? Evidence from rural-to-urban migrant workers in China. China Econ Rev 45:219–231CrossRefGoogle Scholar
  16. Shen H, Chen Y, Russell AG, Hu Y, Shen G, Yu H, Henneman LRF, Rua M, Huang Y, Zhong Q, Chen Y, Li Y, Zou Y, Zeng EY, Fan R, Tao S (2018) Impacts of rural worker migration on ambient air quality and health in China: from the perspective of upgrading residential energy consumption. Environ Int 113:290–299CrossRefGoogle Scholar
  17. Wang P, Liu Z, Yang R (2017) An empirical study on rural household energy consumption influencing factors about migrant workers in the country. J Energy Nat Resour 6(4):52–57CrossRefGoogle Scholar
  18. Ward IC (2008) What are the energy and power consumption patterns of different types of built environment? Energy Policy 36(12):4622–4629CrossRefGoogle Scholar
  19. Wei T, Zhu Q, Glomsrød S (2014) Energy spending and household characteristics of floating population: evidence from Shanghai. Energy Sustain Dev 23:141–149CrossRefGoogle Scholar
  20. Xie Y, Jiang Q (2016) Land arrangements for rural–urban migrant workers in China: findings from Jiangsu Province. Land Use Policy 50:262–267CrossRefGoogle Scholar
  21. Xu L, Paterson AD, Turpin W, Xu W (2015) Assessment and Selection of competing models for zero-inflated microbiome data. PLoS ONE 10(7):e0129606.  https://doi.org/10.1371/journal.pone.0129606 CrossRefGoogle Scholar
  22. Yau K-K-W, Wang K, Lee AH (2003) Zero-inflated negative binomial mixed regression modeling of over-dispersed count data with extra zeros. Biomet J 45(4):437–452CrossRefGoogle Scholar
  23. Yu B (2012) Integrated analysis on household energy consumption behavior across residential and transport sectors: model development and applications. Doctoral Dissertation, Graduate School for International Development and Cooperation, Hiroshima University, JapanGoogle Scholar
  24. Yu B, Zhang J, Fujiwara A (2012a) A household time use and energy consumption model with multiple behavioral interactions and zero-consumption. Environ Plan B 40(2):330–349CrossRefGoogle Scholar
  25. Yu B, Zhang J, Fujiwara A (2012b) Analysis of the residential location choice and household energy consumption behavior by incorporating multiple self-selection effects. Energy Policy 46:319–334CrossRefGoogle Scholar
  26. Yu B, Zhang J, Fujiwara A (2013) Rebound effects caused by the improvement of vehicle energy efficiency: an analysis based on an SP-off-RP survey. Transp Res Part D 24:62–68CrossRefGoogle Scholar
  27. Zhang J (2014) Revisiting the residential self-selection issues: a life-oriented approach. J Land Use Transp 7(3):29–45CrossRefGoogle Scholar
  28. Zhang J (2017b) Life-oriented behavioral research for urban policy. Springer, TokyoCrossRefGoogle Scholar
  29. Zhang M, Song Y, Li P, Li H (2016a) Study on affecting factors of residential energy consumption in urban and rural Jiangsu. Renew Sustain Energy Rev 53:330–337CrossRefGoogle Scholar
  30. Zhang L, Sharpe RV, Li S, Darity WA (2016b) Wage differentials between urban and rural–urban migrant workers in China. China Econ Rev 41:222–233CrossRefGoogle Scholar
  31. Zhao C, Zhou X, Wang F, Jiang M, Hesketh T (2017) Care for left-behind children in rural China: a realist evaluation of a community-based intervention. Child Youth Serv Rev 82:239–245CrossRefGoogle Scholar
  32. Zhong B, Liu T, Chan SSM, Jin D, Hu C-Y, Dai J, Chiu HFK (2015) Prevalence and correlates of major depressive disorder among rural-to-urban migrant workers in Shenzhen, China. J Affect Disord 183:1–9CrossRefGoogle Scholar
  33. Zhu P, Zhao S, Wang L, Yammahi SA (2017) Residential segregation and commuting patterns of migrant workers in China. Transp Res Part D: Transp Environ 52:586–599CrossRefGoogle Scholar
  34. Zhu Y, Hu X, Yang B, Wu G, Wang Z, Xue Z, Shi J, Ouyang X, Liu Z, Rosenheck R (2018) Association between migrant worker experience, limitations on insurance coverage, and hospitalization for schizophrenia in Hunan Province, China. Schizophr Res 197:93–97CrossRefGoogle Scholar

Copyright information

© Springer Nature B.V. 2019

Authors and Affiliations

  1. 1.Graduate School of Biomedical and Health SciencesHiroshima UniversityHiroshimaJapan
  2. 2.Mobilities and Urban Policy Lab, Graduate School for International Development and CooperationHiroshima UniversityHigashi HiroshimaJapan

Personalised recommendations