Parental migration and young migrants’ wages in urban China: An exploratory analysis

Abstract

Chinese

Since the initiation of the economic reforms in 1978, generations of Chinese migrants have moved from the countryside to cities to seek job opportunities. As a result of financial constraints and institutional obstacles, many migrants leave their children at the place of origin, to be taken care of by partners, grandparents or other caregivers. Whilst previous studies primarily focus on the impacts of parental migration on children’s education and health, very few studies have examined its longer-term impacts on labour market income when children reach adulthood. Yet parental migration is likely to influence children’s human capital accumulation and skill development. Drawing on data from the 2011 Chinese Migrant Dynamics Monitoring Survey, this article fills the gap by exploring the relationship between different types of parental migration and their children’s wages when the children have grown up and migrated to work in cities. Structural models are employed to estimate both education and wage equations simultaneously to capture the direct effect of parental migration on wages, together with the mediating effect of education. The results show significantly negative relationships between parental migration and young migrants’ educational attainment and wages. Those who experienced the out-migration of both parents are most disadvantaged in the urban labour market. The study is important for policies aimed at improving migrants’ life prospects and enhancing social mobility and equality.

Keywords

China labour market left-behind children parental migration rural-to-urban migration

Introduction

China has experienced the largest rural-to-urban migration wave in human history since the initiation of the economic reforms in 1978. According to China’s National Bureau of Statistics, there were about 250 million rural migrants in 2011 (NBS, 2012). However, only 21% migrated with their entire family (Lei et al., 2017). Many migrants leave their children at the place of origin due to financial constraints and institutional obstacles resulting from the household registration (hukou) system. According to the system, migrants cannot get local hukou status automatically after they migrate, and are therefore excluded from many benefits and services at their destination, including subsidised housing, minimum living allowance and unemployment insurance. In addition, they cannot send their children to local state schools in many cities, unless they pay placement fees and/or provide all required documents such as proof of tax payments and stable housing. Even if they provide these documents, their children are not allowed to go to local high schools after the nine years’ compulsory education (Chen et al., 2013). As a result, many migrants adopt a split-family strategy by leaving their children in the care of partners, grandparents and other relatives in the countryside (Fan, 2008). The Fifth Population Census shows that there were 19.81 million rural children aged 17 or below who were left behind by one or both parents in 2000 (Duan and Zhou, 2005). According to the Sixth Population Census, the number increased dramatically to about 62 million in 2010, accounting for 38 per cent of the children in the countryside and 22 per cent of the children nationwide (All-China Women’s Federation, 2014; Duan et al., 2013). China is reported to have the largest number of left-behind children among developing countries (Lei et al., 2017).

Existing studies have focused on the impacts of parental migration on children’s education and health (e.g. Antman, 2012; Zhao et al., 2014). There is a dearth of research on the longer-term impact of parental migration on children’s labour market outcomes when they reach adulthood and migrate to work in cities. It is important to study labour market outcome because it is an important indicator of social mobility. An investigation of the relationship between parental migration and children’s wages can enhance our understanding of the consequences of labour migration for left-behind children and social mobility over migrant generations. Such knowledge is conducive to policies aimed at improving migrants’ life prospects and social equality. Western literature on inter-generational mobility focuses on the impact of family disruption, in particular the parents’ divorce or separation from a partner, on children’s labour market outcomes, through the mechanisms of reduced household income and parental engagement. However, family separation as a result of migration in China differs from single parenthood in the Western context in that migrant parents send back remittances which increase household income. Parental migration may influence children’s human capital accumulation and skill development both positively through increased investment and negatively through reduced parental care. This will affect their labour market outcomes.

This article aims to fill in the above-mentioned gap by examining the relationship between parental migration and children’s labour market outcomes after children have migrated to work in cities. Specifically, we compare the wages of young rural migrants who were left behind because of parental migration during their childhood with those of migrants who grew up with at-home parents. Young rural migrants are defined as those who were born after 1980 and brought up in the countryside before they migrated to work in cities, without local urban hukou status, consistent with the definition of ‘new-generation rural migrants’. They ‘grew up already knowing and seeing migrant work as an established way of life’ (Fan and Chen, 2014: 19), and have followed in the footsteps of their parents or other migrants by moving to cities immediately or shortly after full-time schooling. They are now the major migrant labour force in urban China. Migrant parents are defined as those who had been away from their children at the place of origin for at least one year¹ when the children were growing up.

Parental migration may influence children’s cognitive and non-cognitive development, including education, health, soft skills such as communication, attitudes, aspirations and behaviour. Since educational attainment is one of the indicators of human capital crucial for labour market outcomes, it acts as an important mediator in the relationship between parental migration and wages. Using structural models, we investigate the direct association between parental migration and labour market income, as well as the indirect one, that is, parental migration influencing children’s education, which further affects income. Four types of parental migration are distinguished: both parents being at home, maternal migration only, paternal migration only and the out-migration of both parents, to see whether heterogeneities exist among these groups.

Our data come from the 2011 Chinese Migrant Dynamics Monitoring Survey, which records information on young migrants and their parental migration and is cross-sectional in nature. Ideally we need longitudinal data which follow individuals over their childhood and adulthood, to examine the impact of family structure on labour market outcomes. However, such data do not exist in China.² Based on existing cross-sectional data, we aim to provide an exploratory analysis of the association, rather than the causation, between parental migration and labour market income, and to provide potential explanations for the association. Despite the data limitation, the importance and originality of the topic deserves a first exploration, which is conducive to suggesting directions for future research and data collection.

The rest of the article is organised as follows. The next section reviews previous studies on family disruption and children’s economic wellbeing. Studies on the impact of parental migration on children’s development in the Chinese context are also reviewed to provide context. We then introduce the research design, including survey data and methods. This is followed by the discussions of the empirical results. The final section concludes with the main findings, limitations and directions for future research.

Literature review

Family disruption and children’s economic wellbeing

Family is the most important institution influencing children’s development, including their physical and mental health and intellectual and social abilities. Family disruption is likely to influence children’s labour market outcomes when they reach adulthood. First, family disruption influences household income. According to the household production model (Becker, 1991), a decline in household income is likely to adversely affect investment in children’s education, which has a longer-term impact on children’s human capital development and productivity. Second, the absence of one or both parents may reduce the care and support provided to children. Parents may spend less time with children, supervising and monitoring their progress and behaviour. According to the social control theory (Thornton, 1991), supervision and monitoring of children’s development is crucial for children to avoid behavioural problems which might influence job performance in their adulthood. Grandparents or custodial parents are likely to exert less authority or control on children than their biological parents. Third, family disruption may bring emotional stress to children, as a result of both the disruption itself and/or lack of parental care afterwards. Such emotional stress may influence children’s skills development.

Empirical studies have been conducted in the Western context to examine the impacts of family disruption on children’s economic wellbeing, using both cross-sectional and longitudinal data. Most studies find that children living with a single parent are more disadvantaged than those with married biological parents, in terms of educational attainment and labour market performance. However, it is inconclusive whether such a relationship is causal or not, because of the endogenous nature of family structure and omitted variables. For example, Murray and Sandqvist (1990) reported a negative relationship between living with a single mother at age 13 and educational attainment at age 21 in Sweden. Using the 1984 Manitoba birth cohort, Strohschein et al. (2009) find that children who experience changes in family structure are more likely to drop out from high schools, compared with those living in stable two-parent families. Song, Benin and Glick (2012) also find a negative association between family disruption and high-school completion rate. However, once they control for economic resources before family disruption, such difference is significantly reduced. Using the 1991–1995 waves of the British Household Panel Survey, Ermisch and Francesconi (2001) find that people living with a single parent not only receive a lower level of educational attainment, but are more likely to experience economic inactivity, early child bearing, distress and smoking. Fronstin et al. (2001) agree that family disruption leads to a lower level of educational attainment for both males and females, based on data from the British National Child Development Survey. They also report that family disruption is associated with a decreased employment rate for males and with lower wages for females at age 33, even after controlling for pre-disruption family circumstances. Bjorkland et al. (2007) maintain that children who were brought up in non-intact families are more likely to obtain lower levels of education and earnings, using the 1979 cohort of the National Longitudinal Survey of Youth and the Panel Study of Income Dynamics in the US. However, they find that these negative impacts become insignificant when within-family variation is removed.

Parental migration results in a distinct form of family separation, which differs from family disruption as a result of divorce or separation from a partner in the Western context in that migrants’ remittances may lead to a significant increase in family income. This echoes with the new economic theory of labour migration, which indicates that migration is a household strategy to diversify risks by sending some family members to work elsewhere and remit money back (Stark and Bloom, 1985). Studies on international migration have shown that remittances from abroad relax household budget constraints in developing countries, which enables migrant families to invest in children’s education and health and decreases child labour (Antman, 2012; Kandel and Kao, 2001). For example, Edwards and Ureta (2003) reported that remittance from international migration has a significantly greater role in improving left-behind children’s education than local household income, in El Salvador. Yang (2008) finds that the Asian financial crisis in 1997 resulted in an increase in remittances sent by international migrants due to the devaluation of the Philippine currency. This effectively enhanced the schooling rate of left-behind children in the Philippines. Antman (2012) distinguishes between domestic and Mexico–US international migrations, and finds that fathers’ migration to the United States has a significantly positive effect on daughters’ education in Mexico, whilst fathers’ domestic migration does not. Acosta (2011) reported that remittance is particularly useful in improving girls’ schooling in El Salvador. This can be explained by an increase in women’s bargaining power in the household when men migrate, which significantly enhances resource input for girls (Antman, 2011).

Some negative consequences of parental international migration are also reported. For instance, Kandel and Kao (2001) acknowledge the positive income effects of Mexico–US international migration on left-behind children’s academic performance. But parental migration decreases children’s aspiration to obtain higher educational levels (such as university) because children have more exposure to migration to the US as an alternative route of social mobility, and their education in Mexico might not be valued in the US labour market. Lahaie et al. (2009) show that parental international migration has detrimental effects on left-behind children’s academic, behavioural and mental health issues in Mexico. Cortes (2015) compares the impacts of paternal and maternal migration on children’s outcomes, and finds that maternal migration has more detrimental impacts than paternal migration.

Parental migration and the impact on children’s development in China

Similar to international migration, many Chinese migrant parents who have moved to cities tend to send back remittances, which reduces the economic vulnerability of the household in the countryside. Traditionally, individuals rely on family members for care and support in rural areas because access to social benefits is limited. Remittance is beneficial to children’s development, as more resources are available to invest in schooling, food and nutrition. It is reported that left-behind children are more likely to live in better housing conditions and to have access to mobile phones than others (Shen et al., 2015). However, the absence of one or both parents reduces the levels of parental care, supervision and support. Living apart, migrant parents may not be able to supervise their children’s homework adequately, or communicate with teachers directly and frequently. Left-behind children may have more household chore responsibilities and therefore less time for school-related activities (Chang et al., 2010). Moreover, the absence of parents may have adverse impacts on children’s psychological development (Zhao and Yu, 2016). Finally, left-behind children may face more competing alternatives to schooling; with information about jobs in cities, they may drop out from schools earlier and follow in the footsteps of their parents to work elsewhere (Bredl, 2011).

A large number of studies have discussed both the positive and negative impacts of parental migration on children’s education in China. Some studies report insignificant effects, because remittance enhances family resources, which cancels out the negative impact of parental absence. For example, Wen et al. (2015) collected data from 864 rural young people between 10 and 17 years old in Guangxi Province in 2010, and found that parental migration does not affect children’s Chinese and maths test scores. However, other studies find significantly negative impacts. Using a survey of 7648 students in the primary schools in the rural areas of Ningxia and Qinghai in North-west China, Zhao et al. (2014) reported that parental migration leads to a reduction in a child’s maths score rank by 15.6%; in particular, a reduction of 8.37% due to paternal migration and 23.3% because of maternal migration. Zhang et al. (2014) reveal that the absence of both parents significantly reduces left-behind children’s cognitive development, a decrease of 5.4 percentile points for maths and 5.1 percentile points for Chinese, but the effects are much smaller for those children being left behind by one parent. Wang and Mesman (2015) review eight papers and find that seven of them show that left-behind children have a poorer performance on standardised tests at schools when compared with non-left-behind children.

Studies have also examined the impacts of parent migration on left-behind children’s health. Some find a negative relationship, as left-behind children are more likely to develop higher levels of stress and loneliness and lower levels of self-esteem, as a result of separation from parents and reduced parent–child interaction (e.g. Wu et al., 2015). Zhao and Yu (2016) conduct a meta-analysis of parental migration and mental health, based on 32 studies. They conclude that the left-behind experience is detrimental to the mental health of children, and especially girls, children in primary school and self-guardian children. However, using the cross-sectional survey data of 2283 rural children aged 10–18 years from Hebei Province in 2012, Shen et al. (2015) find an insignificant relationship between parental migration and depression/anxiety. The authors explain that most left-behind children in the survey lived with their mothers and other family members, who provided psychological support. This is confirmed by Ren and Treiman (2016), who show no difference in emotional wellbeing between left-behind children and those living with both parents, using data from the China Family Panel Studies.

In summary, there are both positive impacts of parental migration on left-behind children due to remittance and negative ones resulting from family separation and reduced parental care. The way the positive and negative impacts balance out depends on specific circumstances, including family split arrangements, contact with children and the rural community. Previous studies focus on the impacts of parental migration on children’s education and health. Yet, very few studies have examined the longer-term influence on labour market outcomes when children reach adulthood and migrate to work in cities. This article will use the following research design to fill this gap.

Analytical framework, data and methods

According to Becker’s (1993) human capital theory, investment in human capital increases individuals’ productivity, which in turn brings high labour market income. Human capital is defined as ‘productive wealth embodied in labor, skills and knowledge and it refers to any stock of knowledge or the innate/acquired characteristics a person has that contributes to his or her economic productivity’ (Tan, 2014: 412). Education is an important indicator of human capital which is crucial in determining wages. Previous studies have shown that parental migration influences children’s educational attainment because of remittance and reduced parental care (e.g. Zhao et al., 2014). In the meantime, parental migration may influence children’s cognitive and non-cognitive abilities, behaviour and social skills such as communication, which are also important in influencing labour market outcomes. Thus, both direct and indirect impacts of parental migration on young migrants’ labour market income may exist. Education acts as a mediator in the relationship between parental migration and labour market income.

Besides education, the wage structure in China is influenced by an individual’s gender, work experience and job-related characteristics such as occupation, industry, company ownership and geographical location, as shown in previous studies on both migrants and local residents (Knight and Song, 2005; Liu et al., 2017). There are debates on whether parental migration influences children’s job-related factors such as occupation and industry. However, there is no particular theory or robust evidence from previous studies in China supporting this hypothesis. Therefore, we focus on the direct and indirect relationship between parental migration and children’s income through the mechanism of educational attainment, and assume that children with or without migrant parents have different labour market incomes after they themselves have migrated to work in cities.

Our data come from the new-generation migrant survey of the 2011 Chinese Migrant Dynamics Monitoring Survey (CMDMS), which is organised by China’s National Population and Family Planning Commission (NPFPC). We use this survey because it focuses on new-generation migrants and records information on parental migration and the locations of primary and junior middle schools besides demographic characteristics and labour market outcomes. The survey employed a three-stage probability proportional to size (PPS) sampling method to provide a nationally representative sample of young migrants. Twenty cities in 12 provinces³ were selected purposively at the first stage, representing popular migration destinations at different administrative levels in different parts of the country. In each city, communities were randomly selected. At the final stage, migrants were recruited randomly in each community to fill in questionnaires. Altogether there are 9989 respondents who were born after 1980 and had valid labour market information. Among them, 1901 respondents hold urban hukou status and are thus removed from the study; 230 respondents who went to either primary or junior middle schools in non-hukou areas are also excluded. The survey does not record the exact time when parents migrated, but records their total years of being away from home by 2011. In order to ensure that parents migrated before their children migrated, we further remove the samples where parents’ migration duration is shorter than their children’s. The final sample size is 6580 after dropping missing values for key variables.

Structural models are employed to take into account both direct and indirect effects of parental migration. The model enables us to unravel the route through which parental migration might affect labour market income, in particular the mediating effects of education. Specifically, we estimate two models, education and income, simultaneously in a structural framework, while making their error terms correlated with each other. As years of education and income are numerical variables, the linear regression models specified below are employed:

Ed u_{i} = a_{0} + a_{1} x'_{1 i} + a_{2} x'_{2 i} + u_{i}

(1)

\begin{matrix} LnIncom e_{i} = b_{0} + b_{1} Ed u_{i} + b_{2} x'_{1 i} + b_{3} x'_{3 i} + ε_{i} \\ u_{i} ~ N (0, σ_{u}^{2}), ε_{i} ~ N (0, σ_{e}^{2}) \end{matrix}

(2)

The subscript i denotes an individual. $x_{1}$ is a vector of explanatory variables which influence both education and income, such as parental migration and gender. $a_{1}$ is the corresponding coefficients to estimate. Both the binary variable of ‘either parent being away’ and four different categories of parental migration are used in separate models. $x_{2}$ represents a vector of variables which influence education but not income, such as household circumstances at the place of origin where respondents had primary and/or secondary schools. The 2011 CMDMS survey provides information on the locations of primary and junior middle schools, on the location of original housing and on farmland allocated to the household. As shown in previous studies (Li, 2016), schools located in towns and cities have better facilities and staff resources than those in villages, reflecting the huge rural–urban divide in terms of educational resources and school quality. It is assumed that such differences may influence individuals’ educational attainment. Similarly, the location of original housing in a town or a village reflects the public services and resources the household has access to. Farmland allocated to a household may be related to the workload of family members, which might distract children from schooling activities. These variables about the place of origin are tested as to whether they are directly associated with educational attainment, but not with wages after respondents have moved to work in cities, thus fulfilling the exclusion restriction requirement. Finally, provincial dummies of places of origin are added to the education model, accounting for the geographic differences in economic development levels and educational resources; $a_{2}$ is the corresponding coefficients to estimate.

For the income model, we use the conventional Mincerian earnings equation which has been used effectively for wage analysis in China (Knight and Song, 2005; Mincer, 1974). The dependent variable is logarithmic hourly wages, calculated as monthly wages divided by the number of hours worked per month. Besides education and $x_{1}$ , also included in the model are work experience, work experience squared, marital status, and job-related characteristics such as occupation, industry and company ownership, following previous studies (e.g. Knight and Song, 2005). In addition, dummies of destination cities are added to capture the heterogeneities in different geographical contexts.

It is noted that parental migration may be endogenous, because the migration decision-making of parents may not be random, contingent on family circumstances which might influence children’s economic wellbeing. This is a common methodological challenge when examining the impact of family disruption on children’s development. It would be useful to control for family circumstances before parents migrated. The 2011 survey only provides limited information on family migration history. In particular, no information on the exact timing of parental migration and pre-migration household conditions is recorded. An alternative to reduce endogeneity bias is to use instrumental variables. However, there are four different categories of parental out-migration, which makes it extremely difficult to find suitable instrumental variables to enable identification in this study. Therefore, some of our findings may be subject to alternative explanations related to the selection problem, which we could not exclude. By focusing on the association between parental migration and labour market income using structural models, our purpose is to compare the incomes of young migrants who had or did not have migrant parents when they were growing up, and to discuss the likely mechanisms, including the direct effect and the mediating effect through education.

Empirical findings

Descriptive information

Table 1 displays the descriptive statistics of the variables used in the study. Among the 6580 respondents, 23.95% had experiences of either their mother or father working away from their places of origin when they grew up. In particular, 16.38% of the respondents experienced the out-migration of both parents; 6.84% had lone migrant fathers and 0.73% had lone migrant mothers. One explanation for the low percentage of lone migrant mothers is that females usually migrate with their partners after marriage (Fan, 2008). The respondents’ average years of education are 10.37, consistent with existing literature which states that new-generation migrants’ years of education vary between 8.9 and 10.9 years (Fan and Chen, 2014). The majority of the respondents had their original housing in villages. About 93.94% went to primary schools in the countryside, with very few educated in schools located in county towns of hukou registration. A larger proportion (32.43%) went to junior middle schools located in county towns, because no such schools exist in small villages and residents have to go to the nearest towns for secondary schools.

Table 1.

Summary of variables used in the study.

Variables	Description	Proportions (%)/means
Outcome variables
Education	Years of education	10.37
Hourly wage	Wage per hour in the previous month	10.95
LnWage	Log hourly wage in the previous month	2.27
Independent variables
If-parent-out	Either parent migrated	23.95
Type of parental migration	Both parents at hometown (base)	76.05
	Only father migrated out	6.84
	Only mother migrated out	0.73
	Both parents migrated out	16.38
Female	Female = 1	48.02
Hk_urbanprimary	Primary schools in the county town of hukou registration	6.06
Hk_urbanjunior	Junior middle schools in the county town of hukou registration	32.43
Housinghome	Housing at hometown located in a village (base)	62.64
	Housing at hometown located in a town	3.72
	Housing at hometown located in a city	1.96
	Others	31.67
Land_at_home (mu)	No land at hometown (base)	34.26
	0.01–2.99 mu land at hometown	10.27
	3–4.99 mu land at hometown	21.79
	5–6.99 mu land at hometown	16.49
	Over 7 mu land at hometown	17.19
Work experience	Work experience since first migration (years)	3.34
Married	If married = 1	39.45
Occupation	Manager or professional staff (base)	17.45
	Sales personnel	39.22
	Service personnel	2.54
	Production workers	37.96
	Others	2.83
Industry	Manufacturing (base)	41.49
	Construction	4.89
	Service	48.25
	Other	5.36
Company ownership	Privately-owned (base)	46.55
	Publicly-owned	10.11
	Foreign	17.01
	Self-employed	24.64
	Other	1.70

Note: For the variable ‘Land_at_home’, the measurement unit is ‘mu’, where 1 hectare = 15 mu.

Work experience is calculated as years since the first non-agricultural work after migration. This is justified, as most new-generation migrants moved to work in cities immediately or shortly after they completed full-time education in their hometown (Fan and Chen, 2014). According to our data, over 96% of the respondents secured a job after their first migration. For those who continued to study after migration, work experience is calculated using the formula (age – years of education − 6), according to previous studies (Knight and Song, 2005). Regarding occupation, 39.22% of the respondents worked as sales personnel, and 37.96% acted as production workers. Those in managerial or professional positions only account for 17.45%. This corresponds to previous studies which demonstrate that migrants are concentrated in low-skilled jobs, mainly in the manufacturing and service sectors (Chen, 2011). About 46.55% of the respondents worked in private companies and 24.64% were self-employed, such as street vendors and service providers like shoe repairers or housing decorators. Only 10.11% were employed in publicly-owned companies, as many jobs in those companies require local hukou status (Chen, 2011).

Table 2 presents the cross-tabulation of education and wages for young migrants with different types of parental migration. It shows that respondents with neither parent out have the highest average years of education. Those with maternal migration only have the lowest educational level. Similar patterns exist for hourly wages. Results of the two-sample T-tests⁴ and ANOVA⁵ show that significant differences exist regarding educational attainment and wages among these groups. In the next section we examine the association between parental migration and wages while controlling for demographic, work experience and job-related factors.

Table 2.

Cross-tabulation of education and wage for different types of parental migration.

Variable	Neither parent out	Either parent out	Father out only	Mother out only	Both parents out
Years of education	10.43	10.21	10.23	9.94	10.21
Hourly wage	11.13	10.06	10.18	9.73	10.02
Ln hourly wage	2.29	2.20	2.21	2.21	2.20
N	5004	1576	450	48	1078

Empirical analysis of parental migration and income

We start from the OLS regression models, with log wage as the dependent variable and education as an exogenous variable (Table 3). Parental migration is represented as a dichotomous variable in Model 1 and four categories in Model 2. The results show that parental migration has significantly negative impacts on young migrants’ wages. This may be attributed to the negative consequences of parental migration for children’s development, such as health and the development of cognitive and non-cognitive skills, demonstrated by previous studies (Wang and Mesman, 2015; Zhao et al., 2014). Most of the other variables in Model 1 have the expected impacts on wages. For example, years of education is significantly and positively related to wages; work experience has significant non-linear effects, that is, wages increase with work experience, but at a decreasing rate. Female migrants receive lower wages compared with males, all else being equal. Being married is significantly positively associated with wages. Compared with managerial or professional positions, sales personnel are paid less; production and service workers receive even lower wages. Those employed in the service and construction sectors have higher wages than workers in manufacturing factories, holding everything else constant. Compared with those in private companies, staff in publicly-owned and foreign companies enjoy a wage premium, while the self-employed are lowest paid, ceteris paribus. Self-employed migrants are usually casual workers in the informal economy, which is the least regulated by the government. When we distinguish between four different types of parental migration in Model 2, it is clear that the migration of both parents is significantly negatively associated with wages. The coefficients for paternal and maternal migration only are negative but insignificant.

Table 3.

OLS and structural models of parental migration and wage.

Variables	OLS regression		Structure model M3		Structure model M4
Variables	Wage M1	Wage M2	Education	Wage	Education	Wage
If-parent-out	−0.0508***		−0.2498***	−0.0453***
If-parent-out	(0.0117)		(0.0643)	(0.0120)
Father out only		−0.0244			−0.2704*	−0.0192
Father out only		(0.0194)			(0.1081)	(0.0196)
Mother out only		−0.0326			−0.5734	−0.0204
Mother out only		(0.0567)			(0.3176)	(0.0573)
Both parents out		−0.0631***			−0.2264**	−0.0579***
Both parents out		(0.0135)			(0.0749)	(0.0138)
Female	−0.1199***	−0.1200***	0.1290*	−0.1240***	0.1304*	−0.1240***
Female	(0.0100)	(0.0100)	(0.0545)	(0.0102)	(0.0545)	(0.0102)
Hk_urbanprimary			0.1928		0.1924
Hk_urbanprimary			(0.1225)		(0.1225)
Hk_urbanjunior			0.5206***		0.5203***
Hk_urbanjunior			(0.0614)		(0.0615)
Housinghome = housing at township			0.6325***		0.6345***
Housinghome = housing at township			(0.1451)		(0.1452)
Housinghome = housing at city			1.1573***		1.1582***
Housinghome = housing at city			(0.1959)		(0.1960)
Housinghome = others			0.3417***		0.3422***
Housinghome = others			(0.0615)		(0.0616)
Land 0.01–2.99 mu			−0.1548		−0.1536
Land 0.01–2.99 mu			(0.0964)		(0.0965)
Land 3–4.99 mu			−0.2417**		−0.2408**
Land 3–4.99 mu			(0.0751)		(0.0752)
Land 5–6.99 mu			−0.1691*		−0.1693*
Land 5–6.99 mu			(0.0818)		(0.0819)
Land 7 and over			−0.1494		−0.1488
Land 7 and over			(0.0826)		(0.0827)
Education years	0.0453***	0.0453***		0.0690***		0.0685***
Education years	(0.0023)	(0.0023)		(0.0100)		(0.0099)
Work experience	0.0372***	0.0371***		0.0369***		0.0368***
Work experience	(0.0038)	(0.0038)		(0.0038)		(0.0038)
Work experience squared	−0.1876***	−0.1867***		−0.1861***		−0.1853***
Work experience squared	(0.0297)	(0.0297)		(0.0296)		(0.0296)
Married	0.0879***	0.0877***		0.0897***		0.0896***
Married	(0.0112)	(0.0112)		(0.0112)		(0.0112)
Service personnel	−0.2183***	−0.2190***		−0.2178***		−0.2185***
Service personnel	(0.0169)	(0.0170)		(0.0169)		(0.0169)
Sales personnel	−0.0996**	−0.0993**		−0.0999**		−0.0996**
Sales personnel	(0.0341)	(0.0341)		(0.0340)		(0.0340)
Production workers	−0.1080***	−0.1085***		−0.1059***		−0.1064***
Production workers	(0.0155)	(0.0155)		(0.0155)		(0.0155)
Other occupation	−0.0897**	−0.0900**		−0.0904**		−0.0907**
Other occupation	(0.0325)	(0.0325)		(0.0324)		(0.0324)
Construction sector	0.2169***	0.2159***		0.2140***		0.2131***
Construction sector	(0.0247)	(0.0247)		(0.0247)		(0.0247)
Service sector	0.0905***	0.0903***		0.0891***		0.0890***
Service sector	(0.0169)	(0.0169)		(0.0169)		(0.0169)
Other industry	0.0898***	0.0897***		0.0897***		0.0896***
Other industry	(0.0244)	(0.0244)		(0.0243)		(0.0243)
State-owned	0.0677***	0.0675***		0.0683***		0.0681***
State-owned	(0.0171)	(0.0171)		(0.0170)		(0.0170)
Foreign	0.0639***	0.0638***		0.0632***		0.0631***
Foreign	(0.0159)	(0.0159)		(0.0159)		(0.0159)
Other ownership	0.0737	0.0738		0.0750		0.0750
Other ownership	(0.0385)	(0.0385)		(0.0383)		(0.0383)
Self-employed	−0.1229***	−0.1230***		−0.1231***		−0.1233***
Self-employed	(0.0132)	(0.0132)		(0.0132)		(0.0132)
Constant	1.9579***	1.9598***	10.0529***	1.7149***	10.0532***	1.7215***
Constant	(0.0359)	(0.0359)	(0.1280)	(0.1055)	(0.1280)	(0.1055)

Notes: * significant at 10%; ** significant at 5%; *** significant at 1%. Standard errors are in brackets. Default categories are both parents at hometown, primary school in a village, junior middle school in a village, original housing in a village, no land at hometown, manager or professional staff, manufacturing companies and privately-owned companies. Provincial and city dummies are not presented in the table.

Since educational attainment, a crucial wage determinant, is likely to be influenced by parental migration, both direct and indirect effects of parental migration on wages may exist. In Models 3 and 4, education and income are estimated simultaneously in structural models, with dichotomous and different categorical variables of parental migration, respectively. Model 3 shows that parents’ out-migration is significantly negatively associated with years of education. Although migration remittance may increase investment in children, lack of sufficient parental care, supervision and support may have detrimental impacts, leading to net negative correlation between parental migration and educational attainment. The result corresponds to previous studies demonstrating adverse impacts of parental migration on children’s education (Zhao et al., 2014). Most variables in the education model are as expected. For example, attending junior middle schools in urban areas significantly improves educational attainment, while attending primary schools in cities does not make a difference. This suggests that school quality at secondary level is crucial for individuals’ educational attainment. Those residing in towns or cities benefit from better public services and resources than rural residents. Farmland holdings between three and seven mu are negatively associated with educational attainment. Housework or agricultural work is likely to distract children from school activities. In terms of wages, parental migration exerts a direct negative effect. The absence of parents may affect children’s development of social skills and behaviour, besides educational attainment. These skills may also influence wages. Compared with the OLS results, education exhibits a larger impact on wages. With the increase of one year’s schooling, an individual receives a wage increase of 7.14%⁶ rather than 4.63% in the OLS model (Model 1). Other variables have shown similar patterns to the OLS results. An interesting finding concerns the gender effects; all else being equal, females obtain higher levels of educational attainment compared with males; however, their wages are significantly lower than those of males after controlling for all other variables. Such a gender effect in the labour market has been supported by previous studies (e.g. Knight and Song, 2005).

When different types of parental migration are added in Model 4, it shows that the out-migration of both parents is significantly negatively associated with educational attainment. Compared with those with non-migrant parents, those with lone migrant fathers receive fewer years of education. The impact of lone migrant mothers is also negative, but statistically insignificant. For the wage equation, the respondents with both migrant parents suffer from a wage loss, compared with those with non-migrant parents. The impacts of maternal or paternal migration only are insignificant.

Model 4 shows that parental migration exerts both direct effects on wages and indirect effects through educational attainment. Such direct and indirect effects are calculated in Table 4. The results confirm its negative impacts on wages. The out-migration of both parents has the largest negative effects, possibly because the absence of both parents significantly reduces the level of care, support and supervision of children, and thus adversely influences educational attainment and the development of social skills. Females are associated with higher educational attainment compared with males. However, female migrants receive significantly lower wages than males. The total effects of being female on wages are significantly negative.

Table 4.

Direct, indirect and total effects on wage.

Variables	Direct effects	p-value	Indirect effects	p-value	Total effects	p-value
Education	0.0685	0.000			0.0685	0.000
Father out only	−0.0192	0.327	−0.0185	0.018	−0.0377	0.058
Mother out only	−0.0204	0.721	−0.0393	0.080	−0.0597	0.305
Both parents out	−0.0579	0.000	−0.0155	0.006	−0.0734	0.000
Female	−0.1240	0.000	0.0089	0.025	−0.1151	0.000

As significant gender effects are observed in Table 3, we estimate the structural models for females and males separately in Table 5. For females, parental migration does not have significant impacts on educational attainment. Compared with those with non-migrant parents, the out-migration of both parents exerts negative impacts on wages. Hence, parental migration only has direct negative impacts on females’ wages. For males, the out-migration of both parents negatively influences their educational attainment as well as wages. Moreover, both paternal migrations only and maternal migrations only have significantly negative associations with education. This suggests that parental migration has a more detrimental effect on educational attainment for males than for females. One explanation is that boys may benefit more from parental supervision and monitoring than girls in terms of school activities and homework. If parents are absent, boys may experience weak social control and are more likely to develop behavioural problems than girls, such as drinking, fighting or smoking. This will influence their educational attainment adversely. Moreover, as males’ wages are higher than females’, young men may face more competitive alternatives to schooling, and are more likely to drop out of schools and work elsewhere, in their parents’ absence.

Table 5.

The relationship between parental migration and wage for females and males.

Variables	Female		Male
Father out only	−0.1874	−0.0090	−0.3642**	−0.0294
Father out only	(0.1632)	(0.0274)	(0.1440)	(0.0277)
Mother out only	−0.4612	−0.0223	−0.8998*	−0.0391
Mother out only	(0.4076)	(0.0681)	(0.5110)	(0.0975)
Both parents out	−0.0807	−0.0674***	−0.3571***	−0.0491**
Both parents out	(0.1108)	(0.0188)	(0.1015)	(0.0202)
Hk_urbanprimary	0.1869		0.2006
Hk_urbanprimary	(0.1763)		(0.1685)
Hk_urbanjunior	0.4456***		0.5927***
Hk_urbanjunior	(0.0885)		(0.0841)
Housinghome = housing at township	0.8254***		0.3955*
Housinghome = housing at township	(0.2073)		(0.2056)
Housinghome = housing at city	0.9180***		1.3806***
Housinghome = housing at city	(0.2872)		(0.2656)
Housinghome = others	0.4688***		0.2103**
Housinghome = others	(0.0868)		(0.0874)
Land 0.01–2.99 mu	−0.1623		−0.1323
Land 0.01–2.99 mu	(0.1412)		(0.1305)
Land 3–4.99 mu	−0.3755***		−0.1090
Land 3–4.99 mu	(0.1084)		(0.1030)
Land 5–6.99 mu	−0.2260*		−0.1157
Land 5–6.99 mu	(0.1208)		(0.1101)
Land 7 and over	−0.3129**		−0.0395
Land 7 and over	(0.1251)		(0.1088)
Education years		0.0614***		0.0845***
Education years		(0.0131)		(0.0148)
Work experience		0.0304***		0.0443***
Work experience		(0.0053)		(0.0054)
Work experience squared		−0.1480***		−0.2299***
Work experience squared		(0.0411)		(0.0417)
Married		−0.0090		0.1746***
Married		(0.0155)		(0.0159)
Service personnel		−0.1556***		−0.2602***
Service personnel		(0.0248)		(0.0229)
Sales personnel		−0.0744*		−0.1055**
Sales personnel		(0.0445)		(0.0515)
Production workers		−0.1288***		−0.1042***
Production workers		(0.0236)		(0.0206)
Other occupation		−0.1363***		−0.0596
Other occupation		(0.0438)		(0.0470)
Construction sector		0.0214		0.2112***
Construction sector		(0.0778)		(0.0279)
Service sector		0.0382		0.1055***
Service sector		(0.0270)		(0.0219)
Other industry		0.0864**		0.0872***
Other industry		(0.0357)		(0.0331)
State-owned		0.0303		0.0813***
State-owned		(0.0261)		(0.0224)
Foreign		0.0732***		0.0443*
Foreign		(0.0205)		(0.0245)
Other ownership		0.0731		0.0607
Other ownership		(0.0584)		(0.0504)
Self-employed		−0.1205***		−0.1294***
Self-employed		(0.0180)		(0.0188)
Constant	10.5232***	1.7588***	9.8036***	1.4862***
Constant	(0.2018)	(0.1397)	(0.1595)	(0.1587)

Table 5 also shows that the wage return to education is lower for females than for males. For one year’s increase in education, a female respondent receives a wage increase of 6.33%, while the corresponding increase for a male respondent is 8.82%. In addition, significant gender differences exist in the effects of farmland allocation and marriage. Farmland allocation significantly negatively influences educational attainment for girls only. This suggests that girls in the countryside are more likely to be influenced by housework than boys. The wage premium of marriage is only significant for males. This is probably because females are more involved in domestic work than males after marriage.

Conclusion

Millions of children have been left behind in China’s countryside by one or both parents who migrated to work in cities. Previous studies have primarily focused on the impacts of parental migration on children’s education and health. This study extends the literature by examining the relationship between parental migration and children’s labour market income when they reach adulthood and migrate to work in cities. Drawing on data on young migrants in the 2011 Chinese Migrant Dynamics Monitoring Survey, we employ structural models of education and income, and estimate the direct and indirect effects on wages of four different types of parental migration: both parents being at home; paternal migration only; maternal migration only; and the out-migration of both parents. We find not only that parental migration is associated with children’s educational attainment, which influences labour market income, but also that it has a direct effect on children’s income after they reach adulthood. Compared with respondents with at-home parents, those with lone migrant fathers obtain fewer years of education; those with two migrant parents are significantly worse off in terms of both years of education and wages. The total effects of parental migration on wages are shown to be negative, especially when both parents moved to work in cities. This may be explained by reduced parental care, supervision and support for their children. Despite the potential positive impacts of remittances, they fail to compensate for the detrimental influences of parents’ absence. Parents are responsible for discipline and supervision of their children’s development. Guardians may not provide the same levels of authority, supervision and care to children as their biological parents. Significant gender effects also exist. Parental migration exerts negative indirect effects on wages through its impact on educational attainment for males only. This suggests that discipline, supervision and monitoring from parents are more important for boys than for girls to enhance their commitment to schooling and to improve educational attainment.

Migration has resulted in significant changes to the traditional family support structure in the countryside. Parental migration is negatively associated with educational attainment and labour market outcomes, especially for those with both migrant parents. Policy initiatives are needed to support left-behind children, and to help them develop skills within and beyond school curriculums. Certain skills such as communication and teamwork might become more and more important to jobs in cities, with the development of the service sector and the upgrading of the Chinese economic structure. It is therefore important to ensure that left-behind children get access to training and support in this regard. Moreover, policies should facilitate closer supervision of children in migrant families. They should also pay attention to the system of social insurance in the countryside. In particular, rural communities can be provided with more resources to support left-behind children. An alternative is for policies to facilitate family migration, most importantly removing the barriers to the schooling of migrant children in cities so that they are entitled to the same level of education as their urban counterparts.

This exploratory study has limitations. First, it is based on a cross-sectional survey on young migrants, with limited information on family circumstances in the place of origin or the history of parents’ migration. Therefore, we are unable to control for parents’ educational attainment, time being away when children grew up and children’s age when parents migrated. These factors might influence children’s economic wellbeing. Second, we cannot entirely rule out the possibilities of endogeneity for different types of parental migration. Therefore, the association found between parental migration and wages cannot be interpreted as causal. It might be explained by some omitted confounding factors that are related to both the parents’ migration decision and their children’s economic wellbeing, such as pre-migration household conditions. Third, we compare the wages of young migrants who were left behind by their migrant parents with those migrants who grew up with at-home parents, using survey data from destination cities. Left-behind children who did not migrate to cities are therefore excluded from our analysis. Although migration has become a way of life in the countryside and many young people choose to migrate because of the sustained rural-to-urban divide, the findings of the study cannot be extended to all left-behind children. Despite these limitations, we provide an exploratory analysis of the important topic of parental migration and young migrants’ labour market income in China, focusing on both the direct effect and the mediating effect through education. We call for future research on this important topic, especially when longitudinal data with detailed information on migration experiences of family members and pre-migration household circumstances are made available. In particular, future studies should take into account the timing of parental migration, the duration of separation and reunion, the ways and frequency of contact with children and the circumstances of care-providers in the rural community.

Footnotes

Acknowledgements

We would like to thank the National Population and Family Planning Commission of the People’s Republic of China for providing the data (the 2011 Chinese Migrant Dynamics Monitoring Survey) for the study.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors are grateful for the financial support provided by the National Social Science Fund of China Youth Project (16CRK023) and the UK Economic and Social Research Council Project (ES/N007603/1).

Notes

References

Acosta

(2011) School attendance, child labour, and remittances from international migration in El Salvador. Journal of Development Studies 47(6): 913–936.

All-China Women’s Federation (2014) Shandong to improve left-behind children care system. Available at: http://www.womenofchina.cn/womenofchina/html1/news/china/16/9294–1.htm (accessed 18 July 2017).

Antman

(2011) International migration and gender discrimination among children left behind. American Economic Review 101(3): 645–649.

Antman

(2012) Gender, educational attainment, and the impact of parental migration on children left behind. Journal of Population Economics 25(4): 1187–1214.

Becker

(1991) A Treatise on the Family. Cambridge, MA: Harvard University Press.

Becker

(1993) Human Capital: A Theoretical and Empirical Analysis, with Special Reference to Education. 3rd edn. Chicago, IL: University of Chicago Press.

Bjorkland

Ginther

Sundstrom

(2007) Family structure and child outcomes in the USA and Sweden. Journal of Population Economics 20(1): 183–201.

Bredl

(2011) Migration, remittances and educational outcomes: The case of Haiti. International Journal of Educational Development 31(2): 162–168.

Chang

Dong

MacPhail

(2010) Labor migration and time use patterns of the left-behind children and elderly in rural China. World Development 39(12): 2199–2210.

10.

Chen

Adams

et al . (2013) Parental migration and children’s academic engagement: The case of China. International Review of Education 59: 693–722.

11.

Chen

(2011) Occupational attainment of migrants and local workers: Findings from a survey in Shanghai’s manufacturing sector. Urban Studies 48(1): 3–21.

12.

Cortes

(2015) The feminization of international migration and its effects on the children left behind: Evidence from the Philippines. World Development 65: 62–78.

13.

Duan

Zhou

(2005) Status of children left behind in rural China. [In Chinese].Population Research 29(1): 29–36.

14.

Duan

Zou

(2013) Major challenges for China’s floating population and policy suggestions: An analysis of the 2010 population census data. [In Chinese]. Population Research 37: 17–24.

15.

Edwards

Ureta

(2003) International migration, remittances, and schooling: Evidence from El Salvador. Journal of Development Economics 72(2): 429–461.

16.

Ermisch

Francesconi

(2001) Family structure and children’s achievements. Journal of Population Economics 14: 249–270.

17.

Fan

(2008) China on the Move: Migration, the State, and the Household. London: Routledge.

18.

Fan

Chen

(2014) The new-generation migrant workers in China. In: Wu

Zhang

Webster

(eds) Rural Migrants in Urban China: Enclaves and Transient Urbanism. London: Routledge, pp. 17–35.

19.

Fronstin

Greenberg

Robins

(2001) Parental disruption and the labour market performance of children when they reach adulthood. Journal of Population Economics 14(1): 137–172.

20.

Kandel

Kao

(2001) The impact of temporary labor migration on Mexican children’s educational aspirations and performance. International Migration Review 35(4): 1205–1231.

21.

Knight

Song

(2005) Towards a Labour Market in China. Oxford: Oxford University Press.

22.

Lahaie

Jeffrey

Tinka

et al . (2009) Work and family divided across borders: The impact of parental migration on Mexican children in transnational families. Community, Work and Family 12(3): 299–312.

23.

Lei

Liu

Hill

(2017) Labour migration and health of left-behind children in China. Journal of Development Studies 54(1): 93–110.

24.

(2016) Who decided college access in Chinese secondary education? Rural urban inequality of basic education in contemporary China. Universal Journal of Educational Research 4(10): 2380–2393.

25.

Liu

Shen

et al . (2017) Market expansion, state intervention and wage differentials between economic sectors in urban China: A multilevel analysis. Urban Studies 54(11): 2631–2651.

26.

Mincer

(1974) Schooling, Experience, and Earnings. New York: National Bureau of Economic Research.

27.

Murray

Sandqvist

(1990) Father absence and children’s achievement from age 13 to 21. Scandinavian Journal of Educational Research 29: 89–102.

28.

National Bureau of Statistics (NBS) (2012) Survey report of migrant workers in China in 2011. Available at: http://www.stats.gov.cn/ztjc/ztfx/fxbg/201204/t20120427_16154.html (accessed 18 July 2017).

29.

Ren

Treiman

(2016) The consequences of parental labor migration in China for children’s emotional wellbeing. Social Science Research 58: 46–67.

30.

Shen

Gao

Liang

et al . (2015) Parental migration patterns and risk of depression and anxiety disorder among rural children aged 10–18 years in China: A cross-sectional study. BMJ Open 5: ee07802.

31.

Song

Benin

Glick

(2012) Dropping out of high school: The effects of family structure and family transitions. Journal of Divorce and Remarriage 53: 18–33.

32.

Stark

Bloom

(1985) The new economics of labor migration. American Economic Review 75: 173–178.

33.

Strohschein

Roos

Brownell

(2009) Family structure histories and high school completion: Evidence from a population-based registry. Canadian Journal of Sociology 34: 83–103.

34.

Tan

(2014) Human capital theory: A holistic criticism. Review of Educational Research 84(3): 411–445.

35.

Thornton

(1991) Influence of the marital history of parents on the marital and cohabitational experiences of children. American Journal of Sociology 96(4): 868–894.

36.

Wang

Mesman

(2015) Child development in the face of rural-to-urban migration in China: A meta-analytic review. Perspectives on Psychological Science 10(6): 813–831.

37.

Wen

et al . (2015) Positive youth development in rural China: The role of parental migration. Social Science and Medicine 132: 261–269.

38.

Kang

(2015) Social capital and the mental health of children in rural China with different experiences of parental migration. Social Science and Medicine 132: 270–277.

39.

Yang

(2008) International migration, remittances and household investment: Evidence from Philippine migrants’ exchange rate shocks. The Economic Journal 118(528): 591–630.

40.

Zhang

Behrman

Fan

et al . (2014) Does parental absence reduce cognitive achievements? Evidence from rural China. Journal of Development Economics 111: 181–195.

41.

Zhao

(2016) Parental migration and rural left-behind children’s mental health in China: A meta-analysis based on mental health test. Journal of Child and Family Studies 25: 3462–3472.

42.

Zhao

Wang

et al . (2014) The impact of parental migration on children’s school performance in rural China. China Economic Review 31: 43–54.