Parental Migration’s Effects on the Academic and Non-Academic Performance of Left-Behind Children in Rural China
Abstract: This study investigates the impact of parental labor migration on the academic achievements and non-academic growth of left-behind children in fourth and seventh grades. Employing survey data collected from rural China in 2014, 2015, and 2016, we examine the effect of parental absence on children’s academic achievement using Propensity Score Matching (PSM) and Difference in Difference (DID) methods. The results demonstrate that left-behind children whose parents have migrated for one year have statistically significantly lower academic scores. Academic scores drop lower for fourthgrade students and students from higher-income families. There are also adverse effects on left-behind children’s confidence, teacher-student relationships, subjective well-being, and educational expectations if parents migrate for one year. Surprisingly, if parental migration lasts longer (totaling two years), these adverse effects disappear, and student’s educational expectations even improve. These results may be because, over time, the adverse effects that occur immediately after parental migration are offset by the positive effects of migration (i.e. higher income). These conclusions can inform migrant parents on ways to utilize their resources to improve the academic performance of their left-behind children.
Keywords: left-behind children, academic performance, non-academic performance, rural China, difference in difference, propensity score matching
JEL Classification Codes: J21, J24, R23
DOI:1 0.19602/j .chinaeconomist.2019.9.07
1. Introduction
With China’s rapid industrial development and urbanization since the mid-1990s, a large number of rural residents have migrated to cities in search of non-farming jobs that offer better pay (Cai, 2010). By the end of 2016, the total number of migrant workers in China had reached 169 million1. Rural labor migration has significantly contributed to China’s economic development and poverty reduction (Jia, et
al., 2016). When rural residents migrate to cities, however, they are not granted equal rights as citizens, such as access to education and healthcare, i.e. the “urban-rural divide.” This divide, plus their limited incomes, makes it hard for migrant workers to send their children to school in cities. As a result, many parents migrating to cities for jobs have left behind their children in the countryside, giving rise to the problem of numerous left-behind children (Yang, Duan, 2008). By the end of 2015, there were a total of 20.19 million rural left-behind children in the compulsory education stage across China, including 13.84
2
million primary school students, and 6.36 million junior middle school students.
The migration of rural parents will change their children’s living and study environments in many ways, and influence their human capital accumulation, future job market performance, and achievements in various respects (Chen, 2009; Li, 2013). According to the National Demographic Census of 2010, 33% of left-behind children live with their grandparents (Duan, et al., 2013). Less educated and less aware of the importance of education, the older generation may not provide proper guidance on their grandchildren’s studies, taking a toll on their academic performance (Duan, Zhou, 2005).
Rural left-behind children will constitute a key part of China’s future workforce. The education of left-behind children is of considerable significance to not only their development but to China’s human capital accumulation and economic development as well. Chinese and international academics have carried out numerous studies on the education of left-behind children. Since most left-behind children are in the stage of nine-year compulsory education (primary school plus junior middle school), existing studies focus how the migration of their parents affect their academic performance, but reached inconsistent results. Some studies find that the academic performance of left-behind children suffers significantly as a result of their parents migrating to cities for work (Li, 2013; Zhao et al., 2014; Meng & Yamauchi, 2017). Based on the Propensity Scoring Matches (PSM) method with data from the Gansu Survey of Children and Families in 2000 (GSCF), Li (2013) finds that parental migration has significant adverse impacts on children’s learning and cognitive abilities and development of relevant skills.
Other studies find that parental migration helps to improve, or at least does not reduce, the scores of left-behind children (Antman, 2012; Ambler et al., 2015; Bai et al., 2018; Zhou et al., 2015; Hou, 2015). Based on a study on 13,000 rural students, Bai et al. (2018) find that parental migration significantly improves the academic performance of left-behind children, particularly those with low scores. Based on a comparison between 141,000 left-behind children and children living with their parents, Zhou et al.
(2015) find that left-behind children perform no worse and even better than their non-left-behind peers in terms of nine indicators including education and psychological health.
A reason behind such inconsistent conclusions is that parental migration has both positive and negative effects on the academic performance of left-behind children. In the empirical studies mentioned earlier, these positive and negative effects are aggregated to arrive at the final effect. The positive effect means that parents may increase household income by migrating to cities for work, thus easing the liquidity constraint to their families and exerting a positive effect on their children’s studies by raising investment in education (Rozele et al., 1999; Antman, 2012; Ambler et al., 2015). Parental migration helps foster a positive expectation on the return to education since working in cities exposes parents to broader information about the benefits of education, such as access to better jobs. Such awareness is conducive to children’s educational gains (Batista et al., 2007). The negative effect means that the absence of parenting takes a toll on left-behind children’s study (Lahaie et al., 2009; Ye & Lu, 2011; Meng & Yamauchi, 2017). The net effect of parental migration on the academic performance of leftbehind children is subject to the sum of positive and negative effects. Conclusions of the empirical study vary with sample characteristics, methods, and variables.
Another reason for the inconsistent research conclusions is whether the positive and negative effects
occur at the same time after parents move to cities. Most studies examine whether parental migration has a positive or negative effect on left-behind children by comparing the positive and negative effects at a particular time point without revealing whether the two effects overlap in time (Hu, Li, 2009; Li, 2013). Hou (2015) finds that the negative effect closely follows parental migration, while the positive effect comes into play after a certain time lag. When parents leave for cities, their children will be left without proper care and supervision, while the positive income effect is relatively small in the short run. Parents will spend more on their children’s education only after they stay in cities for some time. Despite the use of PSM method and the control of sample selection bias, Hou’s study does not address the problem of omitted variables stemming from unobservable factors.
Existing domestic studies seldom boast highly representative samples, large sample size, and the latest data at the same time. Most studies focus on specific provinces, cities, or countries (Ye, Meng, 2010; Bai et al., 2012). Some scholars have carried out their research based on nationally representative data. Yuan and Zheng (2016) employ the 2010 China Family Panel Studies (CFPS) data. Hu and Li (2009) employ 4,967 samples collected in Beijing, Nanjing, Guangzhou, Lanzhou, and Bozhou City of Anhui Province in 2007. Data employed in these studies are out-of-date. Migrant population and their left-behind children face ever-changing socio-economic environments and public services as China’s urbanization reform moves ahead, making it necessary to carry out the research based on the latest representative data instead.
Based on the large-scale sample survey data collected from China’s countryside in 2014, 2015, and 2016, this paper employs data of 18,181 fourth- and seventh- grade students from 166 rural schools in five provinces for an analysis of parental migration’s effect on their left-behind children’s academic performance. Moreover, this paper also examines whether the positive and negative effects simultaneously come into play after parents migrate to cities for work. For this goal, this paper first describes the current status of parental migration and their left-behind children in the countryside. Then, we investigate the effects of different types of families with parents having migrated to cities (families with parents having migrated for a short period and those with parents having migrated for a more extended period) on the academic performance of left-behind children, and uncover whether the adverse effect and positive effect simultaneously come into play. Finally, we carry out a robustness test and heterogeneity analysis. The rest of this paper is arranged as follows: Part 2 introduces data and empirical model; Part 3 presents model results, including parental migration’s effects on the academic performance of left-behind children, as well as the mechanism of such effects, and carries out a robustness test and heterogeneity analysis of the results; the final section offers conclusions and policy recommendations.
2. Data and Methodology 2.1 Data Introduction
2.1.1 Data source
Data employed in this study are from the Dream Center Effectiveness Evaluation and Tracking Project - a study conducted by the Center for Chinese Agricultural Policy (CCAP) of the Chinese Academy of Sciences (CAS) and commissioned by the Cherished Dream China Education Foundation ( CDEF). The project team carried out a sample survey in the five provinces of Shanxi, Shaanxi, Hubei, Guizhou, and Fujian in 2014. The survey covered a total of 18,181 fourth- and seventh-grade students from 166 complete primary schools, junior middle schools, and nine-year education schools. Respondents include all the students of a randomly selected fourth-grade class and seven-grade class, as well as the class teacher, math teacher, and headmaster of the class. In May 2015 and May 2016, the project team carried out an intermediate tracking survey (Evaluation Phase I) and end-of-term tracking survey (Evaluation Phase 2) for all sample schools. Due to the loss of samples, only 9,606 fourth- and
seventh-grade students from 125 schools finally took part in the survey.
2.1.2 Variable explanations
Our questionnaire mainly collected the following information: Information about student academic performance and other aspects, parental migration, and student personal and family backgrounds and socio-economic conditions.
(1) Academic and non-academic performance: The principal explained variable of this paper is the academic performance of students, which is measured by standardized math scores. The project team organized a standardized math test for the students, and obtained the original math scores of students based on the test results. To make the math scores of fourth- and seventh-grade students comparable, we have standardized the original math scores between the two grades.
Apart from academic performance, this paper employs the China Adolescent Psychological Development Scale formulated by Peijing Normal University to measure the self- confidence and subjective happiness of students. The self-confidence scale contains 17 questions on three dimensions, including self-functionality, self-performance, and a sense of achievement. We have aggregated the scores of questions to arrive at an average score of all dimensions to arrive at the total average score of students’ self-confidence. The student subjective happiness scale contains student emotional index and happiness index with nine questions.
Based on the scoring rules of the China Adolescent Psychological Development Scale, the average score of student emotional index is given a weight of 1, and student happiness is given a weight of 1.1. Then, the total average score of student subjective happiness is obtained through aggregation. Aside from the above-mentioned outcome variables, the students were required to rate their relationship with their teachers in the questionnaire, and the full score was 10 points. We obtained the subjective score of teacher-student relationship based on student rating. In the questionnaire, we also asked students about the highest level of education they expected to receive, and measured their educational expectations based on their answers.
(2) Migration of parents. Currently, there is no commonly accepted definition on the left-behind children. Referencing Duan and Zhou (2005) and Ye and Murray (2005), this paper defines left-behind children as students whose one parent or both parents have migrated for work for no less than half a school year. First, this paper has excluded 2,391 students who were left-behind children in the base period to ensure that the migration status is the same for families during the base period. Second, this paper has excluded 486 families in which parents had returned to the countryside during the evaluation period, and finally retains the remaining 6,729 samples from 124 schools for research.
Based on the migration status of parents during various investigation periods, we classify the families of remaining samples into three categories: (a) families in which parents migrated for a short period, i.e. both parents stayed in their hometown during the Base Period (2014) and the Evaluation Period I (2015), and at least one parent migrated during the Evaluation Period II (2016).
( b) Families in which both parents migrated for a long time, i. e. both parents stayed in their hometown during Base Period (2014), and at least one parent migrated during the Evaluation Period I (2015) and the Evaluation Period II (2016).
(c) Families in which none of the parents migrated, i.e. none of the parents migrated during the Base Period (2014), the Evaluation Period I (2015), and the Evaluation Period II (2016).
As Table 1 shows, 8% of the samples are families in which parents migrated for a short period, 10% of the samples are families in which parents migrated for an extended period, and 82% of samples are families in which none of the parents ever migrated. Data suggest that 26% of the fourth- and seventhgrade students in the countryside were left-behind children (Table 1).
(3) Personal and family backgrounds and socio-economic conditions of students. Apart from parental migration, the student questionnaire also includes student personal information and family member
information. Personal information includes such information as gender, age, and boarding status. Family member information includes the number of siblings, household income, and education of parents. There are significant differences between families in which parents migrated and those in which parents did not migrate (Table 2). By comparing the two types of families, we find that as far as individual student characteristics are concerned, most students from families in which parents migrated boarded at school and had poor math scores. As far as family background is concerned, families in which parents migrated had more children and were poorer; families in which parents migrated had less-educated parents. In summary, there are significant differences in the characteristics of families in which parents migrated and those in which parents did not migrate at student personal and family levels. Such differences can be both the reasons for parents to migrate and causes for children’s academic performance. In the analysis of parental migration’s effects on left-behind children, therefore, we need to control for such factors and the associated endogeneity problem.
2.2 Research Methodologies
2.2.1 Combination of Propensity Score Matching (PSM) and Difference in Differences (DID) methodologies
In conducting a quantitative study on parental migration’s effects, we face many statistical problems, such as the bias of omitted variables due to unobservable factors and the self-selection bias. This paper employs a combination of PSM and DID to address the two problems. The basic idea of PSM is to find a non-left-behind child who shares similar endowment characteristics with each left-behind child through PSM, and then compare the mean values of the outcome variables of left-behind children and non-leftbehind children to estimate the effect of parental migration. The steps are as follows: (1) estimate the
migration propensity score of each household; (2) employ the non-repetitive one-on-one method to match left-behind children with non-left-behind children based on PSM’s common support; (3) conduct an equilibrium test of the matched two groups of samples.
PSM can mitigate selection bias and obtain comparable treatment group and control group samples.
However, it may only control for observable factors, and still has the problem of omitted variables for unobservable factors (Dehejia, 2005). As such, this paper combines PSM with DID methods to reduce endogeneity. Under DID method, math score after differential treatment is the explained variable, and the average treatment effect of parental migration on the math score of left-behind children is estimated based on PSM matching results.
2.2.2 Model specification
This paper specifies the following model to estimate parental migration’s effects on the academic performance of left-behind children:
Where i is student i , and s is school s. is change in the standardized math scores of student in Evaluation Period II. is families in which parents migrated for a short or long period. In investigating whether there is any significant difference in the math scores of students from families in which parents migrated for a short period and those from families in which no parent migrated, =1 denotes families in which parents migrated for a short period, and = 0 denotes families in which no parent migrated at all. In investigating whether there is any significant difference in the math scores of students from families in which parents migrated for a long period and those from families in which no parent migrated, =1 denotes families in which parents migrated for a long period, and =0 denotes families in which parents did not migrate. Parental migration status is influenced by the characteristics of individuals and families.
This study controls for some key variables that may influence children’s academic performance and whether a child is left behind. In the model, denotes control variables, which include ( 1) personal information such as gender, age, grade, and whether a student boards at school; (2) number of siblings, household income, education of parents, and other information about family background and socioeconomic status. is the standardized math score of baseline students. Through the value of coefficient and its significance level, we may assess the effects of short-term and long-term migration on the academic performance of left-behind children.
This paper carries out a robustness test from two aspects to ensure the reliability of the results: First, the outcome variables in the model are replaced with self- confidence, teacher- student relationship, and student subjective sense of happiness for analyzing parental migration’s effects on other aspects of left-behind children’s performance. Second, the outcome variables in the model are replaced with educational expectations to examine parental migration’s effects on the educational expectations for left-behind children. These two methods are employed to verify whether parental migration’s effects play out for different outcome variables at various time points rather than simultaneously.
This paper also carries out a heterogeneity test. First, we examine parental migration’s effects on the academic performance of left-behind children from families with different income levels during the base period. Based on base-period income, this paper’s total samples into low-income households (lower 50%) and high-income households (upper 50%). Based on the grouping result, we investigate parental migration’s effects on the academic performance of left-behind children from low-income households and high-income households as defined above. If the positive effect of income does exist, the income effect is relatively insignificant for high-income households, and parental migration may worsen the academic performance of left-behind children. The income effect is relatively significant for low-income households, and parental migration may improve the academic performance of left-behind children. Second, this paper respectively analyzes parental migration’s effects of the academic performance of fourth- and seventh-grade students.
3. Results and Discussions 3.1 Parental Migration’s Effects on the Academic Performance of Students
After matching the personal and family characteristics of students, the regression results in Column (1) of Table 3 indicate that after one year of parental migration, the standardized math scores of leftbehind children are significantly below those of students from families in which parents did not migrate. The regression results of Column (2) indicate that there is no significant difference in the standardized math scores of students from families in which parents migrated for two years compared with those from families whose parents did not migrate. Specifically, the math scores of left-behind children whose parents migrated for one year dropped by 0.15 standard deviations (significant at 5%), and the math
scores of left-behind children whose parents migrated for two years turned out to be higher by 0.08 standard deviations than those of non-left-behind children (statistically insignificant).
The above regression results are inconsistent with the results of most studies on parental migration’s effects on the academic performance of left-behind children. Based on the hypothesis that parental migration has positive and negative effects on the academic performance of their children as mentioned earlier, one explanation on the regression results of Table 3 is that after one year of parental migration, the negative effect comes into play immediately, i.e. parental migration has a negative effect on their left-behind children in the short term. After two years of parental migration, the cumulative positive effect outweighs negative effect, as reflected in the vanishing negative effect of parental migration on left-behind children in the longer term. Most studies did not take into account the sequence of both effects, and therefore led to conclusions contrary to this paper’s findings. Hu and Li (2009) find that the migration of a parent or both parents for over half a year has a significant adverse effect on the Chinese language or math score of the left-behind children. Although the study also uncovered that the remittances sent back home by their migrant parents could offset such a negative effect to some extent, it did not examine the timing for such an income effect to come into play.
Some studies examined the duration and size of the positive and negative effects of parental migration. Tao and Zhou (2012) find that only when both parents have migrated for a relatively long period will the academic performance of their children suffer significantly, and that parental migration’s negative impact on their children’s scores is unlikely to be offset by rising household income since the importance of parenting far outweighs the positive income effect. Results of the above studies are inconsistent with this paper’s findings for one possible reason that Tao and Zhou (2012) define students whose both parents migrated as left-behind children, while our definition is that at least one parent migrated. In this paper, the negative effect of lack of parenting is the average negative effects of migration of only one parent and migration of both parents, which may be smaller than the negative effect of the migration of both parents and therefore more likely to be offset by the positive effect. Meanwhile, Tao and Zhou (2012) only adopt the PSM approach without addressing the bias of omitted variables arising from unobservable variables. Therefore, the results of their study can be biased.
3.2 Robustness Test of the Effect of Parental Migration on the Academic Performance of Students
This paper examines parental migration’s effects on the self-confidence, teacher-student relationship, the subjective happiness of students, and their educational expectations4 to verify the hypothesis that the negative effect of parental migration precedes the positive effect. Our analysis led to similar conclusions as in Table 3 (see Table 4). Based on Column (2) of Table 4, parental migration for a year has a significant negative effect on the outcome variables of students, and Column (3) suggests that parental migration for two years will significantly raise educational expectations for students, but has no significant effect on other outcome variables. This robustness test once again verifies the hypothesis that the “negative effect immediately occurs after parental migration, while the positive effect plays out after a certain period.” Also, there is a significant positive effect on educational expectations for the left-behind children after two years of parental migration, which indicates that the positive effect of expectations on return to education will play out after parents migrate for some time.
The results of studies on parental migration’s psychological health effects on left-behind children contradict with the findings of existing studies. Currently, most studies believe that parental migration harms their children’s psychological health and increases their anxiety (Abbasi & Irfan, 1983; Wang, 2011; Wu et al., 2015). Some studies also suggest that parental migration has a positive effect on the psychological health of left-behind children (Shi et al., 2016) or does not have any psychological impact
at all (Stevens & Volebergh, 2008). Existing studies have certain drawbacks, such as the small sample size and failure to address endogeneity problem and examine changes in the left-behind children’s psychological health after parental migration for some time. For these reasons, their results can be biased.
Some studies investigated changes in the psychological health of the left-behind children after a certain period of parental migration using PSM or DID method based on large-sample data. With a threeyear cut-off point, Hou (2015) discovers that parental migration exerts a negative impact on left-behind children’s psychological health no matter such migration is short-term (within three years) or longterm (above three years). With a four-month cut-off point, Liu, et al. (2017) find that short-term parental migration of fewer than four months has no psychological impact on left-behind children’s psychological health, while long- term parental migration ( above four months) will harm left- behind children’s psychological health.
A possible reason that the above-mentioned studies arrived at inconsistent results with this paper is that these studies investigated junior middle school students who are in a critical stage of psychological development (Liao, Ruan, 2009), making it hard for the negative effect of parental migration to be offset by the cumulative net effect in the short run. In this paper, however, all respondents include
primary school and junior middle school students. While one-year parental migration may inflict an immediate negative impact on fourth- and seventh-grade students, the psychological impact for fourthgrade students is smaller. After one year, the positive effect starts to outweigh the negative effect, so that the average observed effect of parental migration on left-behind children’s psychological health is insignificant in the second year.
3.3 Heterogeneity Analysis of Parental Migration’s Effects on the Academic Performance of Left-Behind Children
There are significant differences in household’s incomes and personal characteristics among fourth- and seventh-grade students. Given this, this section will examine parental migration’s effects on the academic performance of students from families with different income levels and from different grades. Table 5 reports the results. This paper divides total samples into low-income households and high-income households by the household income level during the base period. The benefit of such classification is that all households were free from migrant parents during the base period, and the level of income has nothing to do with migration.
As can be seen from the results of Panel 1 of Table 5, parental migration had no significant impact
on the academic performance of left-behind children in the low-income group no matter the migration lasted for one year or two. For the high-income group, the academic performance of left-behind children from families in which parents migrated for a year significantly worsened, but the negative impact vanished in the second year. A possible explanation is that for low-income households, the income effect is relatively significant when parents migrate for a year, so that the positive effect of migration offsets the adverse effect. For high-income households, the positive effect of parental migration for a year is insufficient to offset the negative effect, but after two years of parental migration, the cumulative positive effect offsets the negative effect for all households.
Children of different age brackets require parenting by different degrees (Sun and Wang, 2016). This paper compares parental migration’s effects on the academic performance of the fourth- and seventhgrade students. Column (1) of Panel 2, Table 5 indicates that one year of parental migration has a significant negative impact on fourth-grade students and no significant impact on seventh-grade students. Compared with seventh-grade students, fourth-grade students require more supervision and care from their parents. After parents migrated for a year, fourth-grade left-behind children performed worse at school due to lack of oversight. Column (2) shows that after two years of parental migration, there is no significant difference in the academic performance between fourth-grade left-behind children and nonleft-behind children. Among seventh-graders, left-behind children performed better than non-left-behind children at school, although this advantage is statistically insignificant.
Results of the above heterogeneity analysis are similar to those of Table 3 and Table 4. After parental migration, the positive and negative effects will come into play at different time points. Based on the analysis of students from different grades, we find that the lack of parenting inflicts an immediate negative effect, which is more significant for fourth graders. Based on the heterogeneity analysis of households with different income levels in the base period, it can be found that the income effect takes some time to build up and is more significant for lower-income households. After a certain period of parental migration, the positive effect will offset or outweigh the negative effect for all students.
4. Conclusions and Policy Recommendations
Based on the panel data of 2014-2016, this paper employs the Difference in Differences (DID) and Propensity Score Matching (PSM) methods to estimate parental migration’s effects on their left-behind children’s academic performance. This paper uncovers that the negative effect immediately comes into play after one year of parental migration, as reflected in the worsening academic performance of leftchildren. However, the positive effect starts to outweigh the negative effect after two years of parental migration, so that parental migration no longer takes a toll on left-behind children then. This conclusion has passed the robustness test and heterogeneity analysis.
From the robustness test, we discovered that one year of parental migration has a significant adverse effect on the left-behind children’s self-confidence, subjective happiness, teacher-student relationship, and educational expectations, while the negative impact vanishes after two years of parental migration. Based on a grouped test, we find that one year of parental migration has a significant adverse effect on the left-behind children from families with higher incomes and from the fourth grade. However, the negative impact becomes offset by the positive effect after two years of parental migration. This paper verifies the assumption that the “negative effect of parental migration precedes the positive effect.”
Unlike most existing studies, we uncover that parental migration’s positive and negative effects come into play at different time points. Specifically, parental migration will leave left-behind children with less care and oversight on their study. Although left-behind children can be taken care of by other custodians, no one else can give them a sense of security as afforded by their parents. As a result, the negative effect of parental migration becomes more significant in the short run. After a certain period of migration, parents will remit cash to their rural homes to fund their children’s education and reduce
pressures for their children to work and earn their living expenses. Also, migration to cities exposes parents to more information and opportunities and raises their expectations on the return to education; migrant parents attach more importance to educating their left-behind children and use more of their income to invest in their children’s education. Based on the experience of working in cities, migrant parents will reinforce the message to their children that education makes it easier to find better jobs.
Compared with existing studies, this paper has the advantage of acquiring rural household data from a tracking survey with large-size representative samples. Nevertheless, this paper also has some drawbacks to address: Without data of changing household income and educational investment on leftbehind children during the evaluation period, we cannot directly measure the income effect. The PSM method cannot address the self-selection bias stemming from unobservable factors. In this paper, we are unable to address the mutual causality between the decision of parental migration and children’s academic performance. This paper supplements the existing body of literature, and offers a new dimension of evidence on the effects of rural worker migration in China.
References:
[1] Abbasi Nasreen & Mohammad Irfan. 1983. “Socioeconomic Effects of International Migration on Pakistani Families Left Behind.” In Arnold Fred & Nasra M. Shah (eds. ): Asian Labor Migration: Pipeline to the Middle East. Boulder, Colorado: Westview Press.
[2] Ambler Kate, Diego Aycinena & Dean Yang. 2015. “Channeling Remittances to Education: A Field Experiment among Migrants from EI Salvador.” American Economic Journal: Applied Economics, 7(2): 207-232.
[3] Antman Francisca. 2012. “Gender, Educational Attainment, and the Impact of Parental Migration on Children Left Behind.” Journal of Population Economics, 25(4): 1187-1214.
[4] Bai Qin, Zeyan Lin & Kaiming Tan. 2012. “Experimental Research on the Training Mode of Left-behind Children in Rural China—— Quantitative Analysis Based on the Changes of Mental Health Status after Field Intervention.” Management World, No.02: 62-72.
[5] Bai Yu, Linxiu Zhang, Chengfang Liu, Yaojiang Shi, Di Mo & Scott Rozelle. 2018. “Effect of Parental Migration on the Academic Performance of Left Behind Children in North Western China.” The Journal of Development Studies, 54 (7): 1154-1170.
[6] Batista Cátia, Aitor Lacuesta & Pedro C. Vicente. 2007. “Brain Drain or Brain Gain? Micro Evidence from an African Success Story.” Social Science Electronic Publishing, 86(6): 463-498.
[7] Cai Fang. 2010. “Urbanization and the Contribution of Migrant Workers——Thoughts on China’s Economic Growth Potential in the Post-Crisis Period.” China Population Science, No.01: 2-10+111.
[8] Chen Xinxin, Linxiu Zhang, Scott Rolleze & Yaojiang Shi. 2009. “The Learning Performance of Parents and Rural Left-behind Children——A Survey from Shaanxi Province and Ningxia Hui Autonomous Region.” China Population Science, No.05: 103-110+112 . [9] Dehejia Rajeev. 2005. “Practical Propensity Score Matching: A Reply to Smith and Todd.” Journal of Econometrics, 125(1-2): 355-364. [10] Duan Chengrong, Fulin Zhou. 2005. “Research on the Status of Left-behind Children in China“. Population Research, No.01: 29-36. [11] Duan Chengrong, Lidan Lu, Jing Guo, Zongping Wang. 2013. “The Basic Status of Survival and Development of Rural Left-behind
Children in China——Based on the Analysis of the Sixth Population Census Data.” Population Science, 35(03): 37-49 .
[12] Hou Yuna. 2005. “The Impact of Parents’ Migrant Workers on the Development of Rural Left-behind Children: An Empirical Analysis
Based on the Tendency Matching Method.” Education and Economy, No.01: 59-65.
[13] Hu Feng, Shantong Li. 2009. “The Influence of Parents’ Migrant Workers on the Education of Left-behind Children in Rural Areas—— An Empirical Analysis Based on the Survey of Migrant Workers in Five Cities.” Management World, No.02: 67-74.
[14] Jia Peng, Yang Du & Meiyan Wang. 2016. “Rural Labor Transfer and Poverty Reduction in China.” Labor Economy Research. 4(06): 6991.
[15] Lahaie Claudia, Jeffrey Hayes, Tinka Markham Piper & Jody Heymann. 2009. “Work and Family Divided Across Borders: The Impact of
Parental Migration on Mexican Children in Transnational Families.” Community Work & Family, 12(3): 299-312.
[16] Li Yunsen. 2013. “Self-selection, Parental Outing and Left-behind Children’s Learning Performance——An Empirical Study Based on
Investigations in Underdeveloped Areas.” Economics (Quarterly): 12(03): 1027-1050.
[17] Liao Meiling & Sijing Ruan. 2009. “A Study on the Relationship between Mental Health Level and Parental Rearing Patterns of Junior
Middle School Students in Xiamen.” Chinese Journal of Health Psychology, 17(08): 1015-1017.
[18] Liu Hongyan, Fang Chang, Yue Ai & Wang Huan. 2017. “The Influence of Parents’ Migrant Workers on the Mental Health of Left-behind
Children in Rural Areas: Based on Panel Data Research.” Peking University Education Review, 15(02): 161-174 +192.
[19] Meng Xin & Chikako Yamauchi. 2017. “Children of Migrants: The Cumulative Impact of Parental Migration on Children’s Education
and Health Outcomes in China.” Demography, 54(5): 1677-1714.
[20] Rozelle Scott, Edward Taylor & Alan deBrauw. 1999. “Migration, Remittances, and Agricultural Productivity in China.” American Eco
nomic Review, 89(2): 287-291.
[21] Shi Yaojiang, Yu Bai, Yanni Shen, Kaleigh Kenny & Scott Rozelle. 2016. “Effects of Parental Migration on Mental Health of Left-behind
Children: Evidence from Northwestern China.” China & World Economy, 24(3): 105-122.
[22] Stevens Gonneke & Wilma Vollebergh. 2008. Mental Health in Migrant Children. The Journal of Child Psychology and Psychiatry,
49(3): 276-294.
[23] Sun Wenkai & Yijie Wang. 2016. “The Influence of Parents Working Outside the Home on the Health of Left-behind Children——Re-ex
amination Based on Micro Panel Data.” Economics Quarterly, 15(03): 963-988.
[24] Tao Ran & Zhou Minhui. 2012. “Parents’ Out-of-work Jobs and the Results of Rural Left-behind Children——Based on the New Discov
ery and Policy Implications of the Empirical Analysis of Anhui and Jiangxi Provinces.” Management World, No.08: 68-77.
[25] Wang Yi. 2011.“Research on Rural Left-behind Children Education.” Journal of University of Electronic Science and Technology (Social
Sciences): no.03: 97-101
[26] Wu Qiaobing, Deping Lu & Mi Kang. 2015. Social Capital and the Mental Health of Children in Rural China with Different Experiences
of Parental Migration. Social Science & Medicine, No.132: 270-277.
[27] Yang Juhua, Chengrong Duan. 2008. A Comparative Study of Educational Opportunities for Migrant Children, Left-behind Children and
Other Children in Rural Areas. Population Research, No.01: 11-21.
[28] Ye jingzhong & James Murray. 2005. Pay Attention to Left-behind Children: the Impact of Migrant Workers from Western Rural Areas of
China on Left-behind Children. Beijing: Social Sciences Academic Press.
[29] Ye Jingzhong & Pan Lu. 2011. Differentiated Childhoods: Impacts of Rural Labor Migration on Left-behind Children in China. The Jour
nal of Peasant Studies, 38(2): 355-377.
[30] Ye Jingzhong, Xiangdan Meng. 2010. Left-behind Children from the Perspective of Migrant Parents. China Rural Economy, No.12: 68
76.
[31] Yuan Meng & Yuting Zheng. 2016. “The Influence of Parents’ Outing on the Education of Rural Children.” China Rural Survey, No.03:
53-63+96.
[32] Zhao Xue, Jian Chen, Mingchun Chen, Xiaoling Lv, Yuhong Jiang & Yehuan Sun. 2014. "Left-behind Children in Rural China Experi
ence Higher Levels of Anxiety and Poorer Living Conditions," Acta Paediatrica, 103(6): 665-70.
[33] Zhou Chengchao, Sean Sylvia, Linxiu Zhang, Renfu Luo, Hongmei Yi, Chengfang Liu, Yaojiang Shi, Prashant Loyalka, James Chu, Alexis Medina & Scott Rozelle. 2015. "China’s Left-Behind Children: Impact of Parental Migration on Health, Nutrition, And Educational Outcomes." Health Affairs, 34(11): 1964-1971.