Does Parental Absence Harm Children’s Education? Evidence from Vietnam

Abstract

This study analyzes the impacts of parental absence due to migration, death, or divorce on children’s school enrollment in Vietnam. We find children from two-parent families have a better chance of enrolling at all levels of education than those from single-parent families. Within single-family types, the negative effect on children of parental divorce is higher than that of parental death, while the effect of parental migration is the lowest. We find that children living with a single mother tend to have higher school enrollment than those living with a single father, indicating the critical role of mothers in children’s education.

JEL Classification: I1, I2, O1.

Keywords

Children’s education parental absence migration divorce Vietnam

Introduction

With rapid urbanization, there is an increasing labor force migrating from rural to urban areas in Vietnam. Data from the 2015 National Internal Migration Survey indicated that 13.6% of the Vietnamese population are internal migrants. From 2010 to 2015, 36.2% of migration was rural to urban, 31.6% urban to urban, 19.6% rural to rural, and 12.6% urban to rural (General Statistics Office, 2016). Due to the rural-to-urban labor migration, many children in rural Vietnam live in the absence of parental care. In addition to the left-behind children caused by internal migration, there has been a rapid increase in the number of children living with one or no parents due to either parental divorce or separation, or a parent’s death. The divorce rate increased from 1% in 2009 to 1.8% in 2019 (General Statistics Office, 2020).

The increase in parental absence in Vietnam due to migration, divorce, or parental death may have negative important implications on children’s education outcome. Parenting has been consistently demonstrated to positively impact children’s educational attainment. Parental involvement in education helps improve children’s educational outcomes, whether measured by test scores, school attendance, or school completion (Lara and Saracostti, 2019; Sebastian et al., 2017; Tárraga et al., 2017). A number of meta-analyses across different countries and educational levels (Castro et al., 2015; Jeynes, 2016; Ma et al., 2016) have demonstrated the positive impact of parental presence and involvement on child academic achievement, regardless of a definition of parental involvement or measure of achievement. In contrast, parental absence accompanies worse academic achievement in most studies (McKenzie and Rapoport, 2011; Mao et al., 2020; Zhang et al., 2014).

In this study, we use data from the 2014 Intercensal Population and Housing Survey (2014 IPS), a large-scale sample survey covering 5% of all Vietnamese households, to examine the effects of one-parent family types on children’s education. We find children from two-parent families have a higher chance of enrolling at all levels of education than those from single-parent families. The negative effect on children of parental divorce is higher than that of parental death (either mother or father is dead), while the effect of parental migration (either mother or father is migrating) is the lowest. We find that children living with a single mother tend to have higher school enrollment than those living with a single father, indicating the critical role of mothers in children’s education.

Our study contributes to the literature in two aspects. First, while most of the previous studies are concerned with either parental migration or parental divorce, our study examines the effects of parental absence due to different reasons, including migration, divorce, and death of mothers and fathers. The large data set allowed us to estimate the impact of different parental statuses. Different types of parental absence can have different effects on children’s education. As a result, the government should have different policies to support children’s education under different types of parental absence. Second, we examine heterogeneous effects of parental absence across parents with different educational levels and wealth levels.

The remaining paper is structured as follows. The second section reviews previous research on parental absence and children’s education. The third section describes the data set and estimation method. The next section discusses the descriptive analysis and empirical results of the impact of parental absence on children’s education. Finally, the fifth section presents our conclusions.

Literature review

Previous studies on the impacts of parental absence on children’s education outcomes fall into two types: (1) the impact of parental migration and (2) the impact of parental divorce or separation.

There is a growing literature on the impacts of parental migration on the outcomes of children left behind, mainly focusing on the children’s educational attainment. The absence of parents due to migration also has an impact on children. Theoretically, parental absence due to migration can positively or negatively affect children’s school attendance. On one hand, parental migration contributes to household income via remittances, thus alleviating household financial constraints and enhancing children’s educational opportunities. On the other hand, it can harm children’s academic outcomes because children receive less parental care and parental involvement in their schooling (Antman, 2012, 2013). Left-behind children are also more likely to spend more time working and less on schooling (Pörtner, 2016). Empirically, most studies find negative impacts of parental migration on children’s education outcomes; yet some find a positive effect. In Mexico, where parents often leave children behind to emigrate to work in the United States, Antman (2012) finds that left-behind children study fewer hours and work for more hours. McKenzie and Rapoport (2011) indicate a negative effect of parental migration on children’s education in migrant households. However, Antman (2013) finds a positive impact of migration on schooling attainment for girls, but not for boys. In the context of rural China, where one in three children below the age of 17 is living without one or both parents due to rural-to-urban migration (Zhang et al., 2014), the findings mainly indicate that parental migration reduces the educational attainment of left-behind children. Fu et al. (2017) find that left-behind children are more likely to get low scores in academic examinations. Mao et al. (2020) find that left-behind children have lower cognitive test scores and academic scores and are less likely to attend college. Zhang et al. (2014) indicate significant adverse impacts of being left behind by both parents on children’s cognitive achievements, but the effects are insignificant for children left behind by one parent.

Similarly, Zhou et al. (2014) find negative impacts of parental migration on children’s test scores only when both parents have migrated, while the migration of one parent has little effect. In contrast, Morooka and Liang (2009) find a positive impact of parental migration on left-behind children in school enrollment in Fujian, China. Also, in China, Yang and Fan (2012) state that children aged 17–18 living with only the mother due to the father migrating are more likely to enroll in high school than their peers living with both parents. Cebotari and Mazzucato (2016) studied that parental migration is likely a predictor for decreased school performance in Nigeria and Ghana but not in Angola.

The other strand of research on the impact of parental absence focuses on the effect of parental divorce or death on children’s educational attainment. Most researchers agree that living with none or just one parent, whether due to divorce, death, or widowhood, adversely impacts children’s well-being (e.g. Berg et al., 2014; Burrell et al., 2020; Haveman and Wolfe, 1995; Kailaheimo-Lönnqvist and Kotimäki, 2020; Liu et al., 2022). Most research on the causal effects of parental absence has been done in developed countries. McLanahan et al. (2013) review 33 studies on the impacts of parental absence due to death or divorce on education; only two were from developing countries (South Africa and Indonesia). They found consistent evidence of a causal effect of parental absence on educational attainment. Data limitations were the main reason that prevented the analysis of the impact of parental divorce or parental death in developing countries. The few studies in developing countries on the effects of parental death include those by Case and Ardington (2006) in South Africa, Gertler et al. (2004) and Cas et al. (2014) in Indonesia, and Beegle et al. (2006) in Tanzania. Studies on the impacts of parental divorce in developing countries include those by Chae (2016) in Malawi, Pholphirul and Teimtad (2018) in Thailand, Brand et al. (2019) as well as Cebotari and Mazzucato (2016) in African countries (Ghana, Nigeria, and Angola), and Zhang (2020) in China. These studies find negative impacts resulting from parental absence, whether due to divorce or death, on children’s education outcomes. However, an analysis using comparative reading scores from the Program for International Student Assessment for 15-year-old students in five Asian countries reports ambiguous results (Park, 2007). Compared to two-parent children, single-parent children in Hong Kong and Korea had negligible disadvantages, while those in Indonesia and Thailand outperformed the children from two-parent families; only in Japan did the negative effect of single parenthood remain significant.

The gender of the absent parent may have an effect as well. Case and Ardington (2006) find significant differences in the impact of mothers’ and fathers’ deaths on children’s education outcome. The loss of a child’s mother is strongly associated with poor schooling outcome, in terms of both school enrolment and completion. While the loss of a child’s father is significantly correlated with lower household socioeconomic status, it does not have a causal effect on children’s educational outcome. In the case of parental migration, most researchers find that maternal migration has a stronger negative effect on a child’s education outcome than paternal migration, indicating a more important role of mothers in children’s education (e.g. Arlini et al., 2019; Dunusinghe, 2021; Mao et al., 2020).

There are a large number of studies on education in Vietnam (e.g. Dang et al., 2022; Glewwe, 2004; Nguyen, 2016; Nguyen and Nguyen, 2020). In most studies, education and school enrollment are strongly correlated with parental education, income levels, household composition, geography, and ethnicity. However, there are few studies on how parental absence affects a child’s educational outcomes. An exception is Nguyen and Vu (2016), who examine the effects of parental migration for work on the time allocation of children aged 5–8 years in Vietnam, using panel data from the Young Lives surveys in 2007 and 2009. The authors find that children with parental absence tend to spend less time on home study and more on leisure and playing. Another noted study is that of De Loenzien (2016), who investigates the effect of lone motherhood on children’s school enrollment and attainment. Using logistic regression models, the author shows that school enrollment and attainment levels are lower for children of lone mothers than for children living with both parents.

Data and method

Data set

In this study, we use the 2014 Intercensal Population and Housing Survey (henceforth referred to as the 2014 IPS), which was a large-scale sample survey covering 5% of all Vietnamese households, selected from 20% of the enumeration areas throughout the whole country.¹ The survey included two types of questionnaires: about 3.4% of all households in Vietnam (equivalent to 760,200 households) were interviewed using the short questionnaire to collect information on age, sex, location of residence, and births and deaths in families, and about 1.6% of all households (equivalent to 361,650 households) were interviewed using the long questionnaire that included all the questions from the short questionnaire as well as questions on migration, education level, and births and deaths within households over the 5 years since the 2009 census. The long-questionnaire data are representative at the provincial level. The sampling units are enumeration areas, which are randomly selected according to the probability proportional to size method of the population of districts. There are 7882 enumeration areas in the long-questionnaire data sample.²

In this study, we used the sample with the long questionnaire. Because of the questionnaire design, it was impossible to include in the analysis children who did not live with at least one parent. Our sample for analysis consisted of all children whose one biological parent was the household head. Households in which grandparents were household heads were excluded because it was impossible to identify the precise relationship between parents and children in these families.³ Step-families with a stepparent and a biological parent were considered two-parent households. We limited the age of children from 7 to 22. The final sample consisted of 274,299 individuals. Our main outcome variable is the school enrolment of children and adolescents aged 7 to 22.

Estimation methods

To understand the association between parental absence and children’s education, we used regression analysis. Regression analysis provides an understanding of the effect of one explanatory variable on a dependent variable after controlling for other explanatory variables. The regression model is as follows

y_{i, j} = β_{0} + C h i l d r e n_{i, j} β_{1} + P a r e n t_{j} β_{2} + H o u s e h o l d_{j} β_{3} + u_{i, j}

(1)

where $y_{i, j}$ is a dummy variable, for example, school enrollment of individual i in household j. $C h i l d r e n_{i, j}$ is the vector of children’s characteristics, including age and gender. $P a r e n t_{j}$ represents a set of dummy variables indicating the single-parent status of children, including children living with only a father (mother migrating), children living with a mother (father migrating), children living with an unmarried mother, children living with a father (parents divorced), children living with a mother (parents divorced), children living with a father (mother died), and children living with a mother (father died). The reference group is children who are living with both parents. $H o u s e h o l d_{j}$ represents variables of households such as ethnicity of household head, household size and proportion of children and elderly in households, and wealth index. $u_{i, j}$ represents unobserved variables. The summary statistics for the variables used in our models are presented in Table 5 in Appendix 1.

We estimated the model (1) using ordinary least squares (OLS) regression. The sampling weights were used in regression, and the standard errors were clustered at the sampling units which are enumeration areas. We used specifications that differed in the selection of control variables to examine the effect of parental status on children’s education to different sets of control variables. In most models, we also controlled for enumeration area fixed effects, which are a proxy for local culture and economic levels, and ethnicity fixed effects to account for differences in terms of ethnic culture and economic factors.⁴

It should be noted that the dependent variable in equation (1) is binary, and binary models such as logit and probit are often applied. However, in this study we estimated equation (1) by OLS for the linear probability model for three reasons. First, our model includes enumeration area fixed effects, and it is a computation burden to use logit or probit to estimate a model with enumeration area fixed effects (some estimates failed to converge). OLS estimators are consistent and can be applied to a binary model (e.g. Angrist et al., 2010). Second, we examine heterogeneous effects of parental absence across parents with different educational levels and wealth levels by including interactions between these variables in regression models. However, Ai and Norton (2003) demonstrate that the sign and significance level of interaction terms in a non-linear model such as probit or logit are not the same as those of the marginal effect of these interaction terms, which can lead to an incorrect interpretation. A linear probability model allows for a straightforward interpretation of the interaction terms. Thus, we use a linear probability model instead of a logit or probit to model interaction terms. Third, we follow an approach of Lewbel (2012) to construct heteroscedasticity-based instruments for parental absence, and this method is applied for linear models. For robustness analysis, we estimated equation (1) using logit without the enumeration area fixed effects. The results are very similar to those from OLS. The results are reported in Table 7 in Appendix 1.

An important variable that affects the demand for children’s education is income. However, there were no data on income in the 2014 IPS. We constructed a welfare index to address this problem: an aggregate index derived from assets and housing variables. We used all variables available in the 2014 IPS, including living areas, housing conditions, water, sanitation, and durables. We followed the principal components approach of Filmer and Pritchett (2001) to compute a wealth index. According to this approach, a wealth index for a household j (denoted by $A_{j}$ ) was constructed as the first principal component of a vector of assets of households, including durables goods, housing characteristics, and access to utilities as follows

A_{j} = \sum_{p} a_{p} (\frac{x_{p j} - {\bar{x}}_{p}}{s_{p}})

(2)

where $x_{p}$ denotes the asset p, and $\bar{x}$ denotes a mean of households in the sample. s is a standard deviation of asset $x_{p}$ , and the p-dimensional vector of weight a is chosen to maximize the sample variance of A, subject to $\sum_{p} a_{p}^{2} = 1$ . The weight a is also called the vector of scores of asset variables, which can be estimated using principal component analysis. The wealth index is standardized to have a zero mean and a standard deviation of one. A higher value of the wealth index means a higher level of assets.

The list of variables used to construct the wealth index is presented in Table 6 in Appendix 1. In this table, we also conducted the analysis of Cronbach’s alpha to examine internal consistency of these variables (Cronbach, 1951). It shows that all variables have a positive sign, meaning all the variables have the same direction with the overall score. The alpha coefficient is estimated at 0.86, indicating reasonable consistency and correlation between the variables used to construct the wealth index (Lance et al., 2006).

Measuring the causal impact of parental absence on children’s school enrolment is a significant challenge because of the endogeneity bias due to unobserved variables. Parents’ ability or health can influence their children’s education and may affect their children through channels other than parental absence. For example, children whose parent has a health problem such as disability are more likely to have lower education (e.g. Mont and Nguyen, 2013), and at the same time their parent can have a higher likelihood of being dead than other people without a health problem. In this case, we can overestimate the effect of parental absence on children’s education.

A traditional method to address the endogeneity bias is to use instrumental variable (IV) regression. This method requires finding an IV that affects parental absence but is not correlated with the error term in the equation for children’s education. Finding such exogenous IVs is challenging. In our study, there are seven dummy variables indicating parental absence, which require seven IVs. For simplicity, we combined all these seven variables into a single dummy that indicates parental absence. Following the approach of Lewbel (2012), we construct heteroscedasticity-based instruments for parental absence. We first regress parental absence on all the control variables as equation (1) and then estimate the residuals, denoted by $ε$ . Next, we use all the control variables and construct a set of IVs for parental absence as $(X - \bar{X}) ε$ , and employ a two-stage least squares (2SLS) regression to estimate the effect of parental absence on children’s education. A condition for this approach is heteroscedasticity in errors in the regression of equation (1), that is, $c o v (X, ε^{2}) \neq 0$ , which can be tested using a Breusch–Pagan test. In this study, we present the IV estimates as the robustness analysis instead of the main finding. The reason is that with the IV method, we could only estimate the effect of a single variable of parental absence, but we were interested in the effects of different types of parental absence. In addition, our IV method used an internal IV, which is not as good as an exogenous one.

Empirical results

Descriptive analysis

Our data analysis in Table 1 indicates that 90.1% of children lived with both parents, about 8.3% with single mothers, and only 1.6% with single fathers. Compared to other developing countries, the percentage of children living with both parents in Vietnam is relatively high. For example, Pholphirul and Teimtad (2018) report that only 68% of the children in a sample of 11,000 Thai students lived with both parents. Using data from Census 2000 in China, Yang and Fan (2012) estimate that 81.8% of children aged 17–18 lived with both parents in 2000.

Table 1.

Frequency of family types by age and wealth quintile.

	Living with both parents	Living with a single father			Living with a single mother				Total
	Living with both parents	Mother migrating	Mother died	Parent divorced	Father migrating	Father died	Parent divorced	Unmarried mother	Total
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Age groups
Aged 7–15	92.2	0.3	0.5	0.6	1.6	2.9	1.6	0.4	100
Aged 16–18	88.7	0.4	0.8	0.5	1.5	5.6	2.1	0.4	100
Aged 19–22	86.2	0.4	1.1	0.4	1.6	7.8	2.0	0.4	100
Wealth index quintile
Poorest quintile	87.8	0.4	1.2	0.8	1.0	6.3	1.8	0.7	100
Near poorest quintile	88.4	0.3	0.8	0.6	1.2	6.3	1.8	0.6	100
Middle quintile	90.2	0.3	0.6	0.5	1.4	4.8	1.8	0.3	100
Near richest quintile	91.7	0.4	0.4	0.4	1.8	3.5	1.8	0.2	100
Richest quintile	92.0	0.4	0.4	0.4	2.4	2.4	1.9	0.1	100
Total	90.1	0.4	0.7	0.5	1.6	4.6	1.8	0.4	100

Father’s death was the most common reason for children to live with only their mother (4.6%), followed by parental divorce (1.8%) and father’s migration (1.6%). For children living with single fathers, maternal death was the most important reason (0.7%), followed by parental divorce (0.5%) and mother’s migration (0.4%). The notable difference in the role of parental death between living with a single father and a single mother is due to the social stigma in Confucian-influenced countries that discourage widows from remarrying.⁵ Table 1 indicates that children’s family situation varied according to their age group and wealth quintile. Overall, children in the poorest 40% of the population were more likely to live with single parents. In terms of age group, younger children were more likely to live with both parents than older children.

The Vietnamese education system is structured into five general levels: nurseries (3 years of age) and kindergartens (3–5), primary, grades 1–5; lower secondary education, grades 6–9; and upper secondary education, grades 10–12. Children are required to enter primary school at the age of 6. Students are required to sit for leaving examinations at the lower and upper secondary school level. There are a number of higher education options in Vietnam such as universities and senior colleges. In addition, students can also enroll in technical colleges or vocational or professional schools. Relative to its income level, Vietnam has achieved remarkable success in terms of its basic education outcomes. Vietnam’s school net attendance rates were 98%, 93%, and 78% at the primary, lower-secondary, and upper-secondary level in 2020 (UNICEF, 2021).

Table 2 presents the school enrollment rate of children classified by their family types. It should be noted that although children aged from 6 are eligible for primary school, we focus on children aged from 7 years in this study. In Vietnam, the school year in Vietnam begins in September. Because the 2014 IPS was conducted in April 2014, some children who were aged 6 did not attend primary school at the time of the survey. Table 2 shows that children from single-parent families generally had a lower enrollment rate at all levels than their peers in two-parent families. However, we see different results between parental migration and parental divorce or death. Children in single-parent families due to migration have similar enrollment rates to children in two-parent families and even higher enrollment rates at college degrees. In contrast, parental divorce or death is associated with much lower enrollment at all education levels, especially at the high school and the college level. Thus, it is likely that the positive income effect due to remittances in parental migration compensates for the negative impact of less parental care. Children living in a family with parental death or never-married mother showed the lowest enrollment at college level, implying they are more likely to enter the labor market to help their families.

Table 2.

Family types and school enrollment (in percent).

	Primary and secondary enrollment rate (age 7–15)	High school enrollment rate (age 16–18)	College enrollment rate (age 19–22)
	(1)	(2)	(3)
Living with both parents	93.8	60.7	23.1
Single-father families
Mother migrating	90.3	57.7	24.6
Parent divorced	87.1	48.7	19.8
Mother died	80.8	40.1	15.1
Single-mother families
Father migrating	93.6	60.9	25.1
Parent divorced	89.7	54.6	19.4
Father died	86.5	47.4	14.7
Never-married mother	89.6	50.8	12.0
Total	93.4	59.6	22.3

Comparing the single-mother and single-father families, children in the former showed higher enrollment rates than the latter at the pre-college level, indicating the relatively more important role of mothers than fathers when children are young. This impact dissipates at the college level as children from both groups showed similar enrollment rates.

The effect of parental absence on children’s education

We first run the OLS regression of parental absence on children’s education. Table 3 provides regressions of children’s education on parental absences and control variables. To examine whether the effect of parental absences is sensitive to the control variables, we estimated two model specifications: one without controlling for parental and household variables (columns 1 to 3 of Table 3) and another that included controlling for parental and household variables (columns 4 to 6 of Table 3). In the large model specifications, we also controlled enumeration area fixed effects, which address the differences in local effects between enumeration areas. Table 3 shows that the estimates of the coefficients of parental absences are quite similar in the two model specifications. We used the result from the large model specification for interpretation.

Table 3.

OLS regression of school enrollment.

Explanatory variables	Dependent variables
	Enrolled in primary or secondary (yes = 1, no = 0)	Enrolled in high schools (yes = 1, no = 0)	Enrolled in college or university (yes = 1, no = 0)	Enrolled in primary or secondary (yes = 1, no = 0)	Enrolled in high schools (yes = 1, no = 0)	Enrolled in college or university (yes = 1, no = 0)
	(1)	(2)	(3)	(4)	(5)	(6)
Gender of children (male = 1, female = 0)	–0.0117***	–0.0912***	–0.0820***	–0.0136***	–0.0913***	–0.0659***
	(0.0013)	(0.0043)	(0.0041)	(0.0013)	(0.0044)	(0.0042)
Age of children	0.1069***	2.9385***	0.2873***	0.1091***	2.7299***	0.3203***
	(0.0027)	(0.1472)	(0.0766)	(0.0027)	(0.1456)	(0.0752)
Age of children squared	–0.0058***	–0.0907***	–0.0085***	–0.0059***	–0.0844***	–0.0094***
	(0.0001)	(0.0043)	(0.0019)	(0.0001)	(0.0043)	(0.0018)
Living with both parents	Reference			Reference
Living with father, mother migrating	–0.0330**	–0.0419	–0.0067	–0.0565***	–0.1022***	–0.0357
	(0.0162)	(0.0352)	(0.0334)	(0.0150)	(0.0328)	(0.0342)
Living with mother, father migrating	–0.0116*	–0.0347*	–0.0142	–0.0361***	–0.0845***	–0.0120
	(0.0068)	(0.0204)	(0.0170)	(0.0062)	(0.0219)	(0.0169)
Living with unmarried mother	–0.0401***	–0.1026***	–0.1116***	–0.0594***	–0.1139***	–0.1011***
	(0.0129)	(0.0341)	(0.0228)	(0.0132)	(0.0356)	(0.0241)
Living with father, parent divorced	–0.0705***	–0.1437***	–0.0564**	–0.0814***	–0.1372***	–0.0656**
	(0.0115)	(0.0322)	(0.0264)	(0.0117)	(0.0326)	(0.0256)
Living with mother, parent divorced	–0.0412***	–0.0875***	–0.0644***	–0.0481***	–0.0950***	–0.0613***
	(0.0065)	(0.0168)	(0.0145)	(0.0066)	(0.0172)	(0.0149)
Living with father, mother died	–0.0955***	–0.1287***	–0.0559***	–0.0884***	–0.0892***	–0.0300**
	(0.0150)	(0.0231)	(0.0155)	(0.0145)	(0.0233)	(0.0146)
Living with mother, father died	–0.0499***	–0.1170***	–0.0722***	–0.0540***	–0.0841***	–0.0472***
	(0.0054)	(0.0094)	(0.0064)	(0.0053)	(0.0099)	(0.0067)
Parental age				0.0046***	–0.0012	0.0007
				(0.0011)	(0.0039)	(0.0030)
Parental age squared				–0.0001***	–0.0000	–0.0000
				(0.0000)	(0.0000)	(0.0000)
Less than primary school				Reference
Parents with a primary degree				0.0574***	0.1025***	0.0377***
				(0.0029)	(0.0073)	(0.0049)
Parents with a secondary degree				0.0702***	0.2025***	0.0957***
				(0.0030)	(0.0079)	(0.0064)
Parents with a high school degree				0.0703***	0.2671***	0.1694***
				(0.0030)	(0.0095)	(0.0090)
Parents with college, university				0.0614***	0.2876***	0.2760***
				(0.0033)	(0.0115)	(0.0148)
Religion (yes = 1, no = 0)				–0.0026	–0.0215**	–0.0218***
				(0.0029)	(0.0084)	(0.0071)
Migration during past 5 years (yes = 1, no = 0)				–0.0438***	–0.0827***	–0.0637***
				(0.0064)	(0.0182)	(0.0130)
Household size				–0.0146***	–0.0316***	–0.0127***
				(0.0010)	(0.0021)	(0.0016)
Proportion of children				0.0229***	0.0363**	–0.0096
				(0.0067)	(0.0179)	(0.0175)
Proportion of elderly				0.0695***	0.1232***	0.0009
				(0.0125)	(0.0327)	(0.0233)
The poorest quintile				Reference
The near poorest quintile				0.0288***	0.0723***	0.0120**
				(0.0034)	(0.0082)	(0.0057)
The middle quintile				0.0467***	0.1316***	0.0364***
				(0.0035)	(0.0092)	(0.0066)
The near richest quintile				0.0616***	0.2038***	0.0912***
				(0.0035)	(0.0097)	(0.0077)
The richest quintile				0.0701***	0.2921***	0.2062***
				(0.0036)	(0.0109)	(0.0096)
Urban areas	0.0160***	0.1460***	0.1821***	Omitted	Omitted	Omitted
	(0.0023)	(0.0074)	(0.0079)
Ethnic group fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Enumeration area fixed effects	No	No	No	Yes	Yes	Yes
Constant	0.4199***	–23.310***	–2.1635***	0.3561***	–21.433***	–2.4350***
	(0.0138)	(1.2458)	(0.7846)	(0.0256)	(1.2345)	(0.7752)
Observations	158,256	55,353	60,690	158,256	55,353	60,690
R ²	0.100	0.142	0.102	0.205	0.369	0.322

Robust standard errors in parentheses. Standard errors are clustered at the enumeration area level. OLS: ordinary least squares.

***

p < 0.01; **p < 0.05; *p < 0.1.

Table 3 indicates that children in single-parent families have lower enrollment rates than their peers in two-parent families. The effect is higher at the high school level (children aged 15–18) than primary and lower secondary school levels (aged 7–14) and college level (age 19–22). For example, living with a single mother due to divorce lowers the probability of enrollment by 9.5% in high schools, 4.8% in primary or lower secondary schools, and 6.1% in colleges in the model with control variables. This is consistent with earlier findings on the negative effect of parental divorce on children’s education in developed and developing countries (e.g. McLanahan et al., 2013; Yang and Fan, 2012).

Among the specific causes for parental absence, parental divorce is the most harmful to a child’s education, followed by never-married mothers and parental death. Parental migration lowers school enrollment at the pre-college level, yet the effect is not significant at the college level.

Among single-parent households, living with a single mother seems better for a child’s education outcome than living with a single father. The negative coefficients of living with a single mother are consistently lower than those of living with a single father, regardless of the specific cause and at most levels of education. The only exception is the effect of parental death on enrollment in higher education, in which children living with a single father are more advantaged than those living with a single mother, probably because at this level of education, the role of mothers in a child’s education diminishes compared to the effect in earlier years.

A channel in which parental absence negatively affects children’s education is by reducing the time spent caring for children. An indirect effect of parental absence is through changing income, one of the main determinants of children’s education. If income is the primary channel through which parental absence affects children’s education, controlling for income will reduce the estimated effect of parental absence. Table 3 shows that the effects in models with and without controlling for wealth index are similar. This finding suggests that income is not the main channel through which parental absence affects children’s education.

The results of the regression analysis also show several interesting findings. Boys tend to have lower enrollment rates than girls at all levels of education. Parental characteristics are strongly related to a child’s education. The age of a parent affects a child’s school attendance. Age and age-squared variables are statistically significant in models of children’s schooling, and the results suggest an inverted U-shaped relationship between parental age and children’s education. The probability of children attending school increases with parents’ age, but this peaks at a certain parental age (45 years in the regression of school attendance of children aged 7–15), after which the probability of attending school decreases with parental age.⁶ The reason for this is that the age of parents reflects income and the parents’ experience. Higher age generally correlates with higher income and more experience, which positively impact children’s school attendance, up to a certain age limit. Age and income have an inverted U-shaped relationship, that is, income increases with age, but only to a certain age, and then falls due to decreasing health and labor productivity in adults at older ages (Deaton, 1986).

Children of parents with higher education levels tend to have a higher rate of school enrollment. The impact of parental education on school enrollment is highest for youth aged 16–22. The school enrollment rate for youth in Vietnam is notably lower than that for young children. In this context, parental education is vital for the education of youth. For example, children aged 7–15 whose parents attended college or university have about a six percentage point higher enrollment rate at primary and lower secondary schools than children whose parents have less than primary education. For youth aged 16–22, the effect of having parents with college or university education is even greater, resulting in a difference of up to 28–29 percentage points in enrollment at high school and college level compared to peers with parents who have less than primary education.

There is often-cited wisdom that children in large families receive less investment from parents than those in small families (Becker and Tomes, 1976). Our analysis shows that children in large families have a lower enrollment rate than children in smaller-sized families. An additional household member reduces the enrollment rate at the primary and lower secondary levels by 1.5 percentage points, at the high school level by 3.1 percentage points, and at the college level by 1.3 percentage points. We also find that children living in migrant households are less likely to enroll in schools. Furthermore, as expected, children from wealthier quintiles have higher enrollment rates than those in the poorer quintiles. The education gap between the rich and the poor is narrow at the primary and lower secondary levels but is remarkable at high school and college levels. In particular, the gap between the richest and the poorest quintile is seven percentage points at the primary and lower secondary levels but reaches 29.2 percentage points at the high school level and 20.6 percentage points at the college level.

Our results are similar to previous research on the issue. For example, De Loenzien (2016) found that in Vietnam, the likelihood of being enrolled in school and completing primary and lower secondary level is 50% lower for children of lone mothers compared to children living with two parents. The author also found significant positive effect of parental education and wealth as well as negative effect of being boys on both enrolment and completion rate at primary and lower secondary schools.

Robustness analysis

As mentioned, the main problem in estimating the effect of parental absence is the endogeneity bias. We conducted several analyses to examine this issue. First, Table 3 shows the estimated effect of parental absence using the small and large model specifications. If the estimate of parental absence is sensitive to unobservable selection bias, it would differ significantly between the small and large model specifications. Table 3 shows very similar results from the small and large model specifications.

Second, we estimated equation (1) using a logit model. Because of the convergence problem, we cannot control for enumeration area fixed effects in the logit model. The results, which are reported in Table 7 in Appendix 1, are similar to those from the linear probability model in Table 3.

Third, we used the heteroscedasticity-based instrument approach of Lewbel (2012) to estimate the effect of parental absence. As mentioned in the estimation method section, we combined all these seven variables into a single dummy that indicates parental absence. Table 8 in Appendix 1 reports the OLS regression of children’s education on parental absence using similar model specifications as in Table 3. It shows that parental absence is negative and statistically significant at the 1% level in all the regressions. The test statistic of the Breusch-Pagan test of heteroscedasticity was over 10,000, indicating that the null of homoscedastic errors was firmly rejected. By default, we use all the control variables to construct $(X - \bar{X}) ε$ , which are the instruments for parental absence. The IV results, which are reported in Table 9 in Appendix 1, indicate a negative and significant effect of parental absence on children’s education.

Heterogeneous effects

We considered the possibility that parental absence may confound characteristics such as parental education or household wealth by running the OLS regression similar to Table 3 but with interaction terms. We ran the models with the interaction terms between parental absence and parental education and between parental absence and household wealth quintiles separately. Table 4 reports the results.⁷ Most of the interaction terms are only significant for primary and lower schooling. At this level of education, the coefficients of the interaction terms are positive, indicating that parental education and household wealth have a beneficial role in reducing the adverse impact of parental absence. In other words, the negative effect of parental absence is partly compensated if the remaining parent is better educated or if the household is relatively well-off. However, this offsetting effect is only significant for younger children from 7 to 15 years of age.

Table 4.

OLS regression of school enrollment with interactions.

Explanatory variables	Dependent variables
	Enrolled in primary or secondary (yes = 1, no = 0)	Enrolled in high schools (yes = 1, no = 0)	Enrolled in college or university (yes = 1, no = 0)	Enrolled in primary or secondary (yes = 1, no = 0)	Enrolled in high schools (yes = 1, no = 0)	Enrolled in college or university (yes = 1, no = 0)
	(1)	(2)	(3)	(4)	(5)	(6)
Living with single parent (yes = 1, no = 0)	–0.0918***	–0.0978***	–0.0344***	–0.0912***	–0.0960***	–0.0302***
	(0.0081)	(0.0125)	(0.0073)	(0.0075)	(0.0128)	(0.0077)
Parents with primary degree × Living with single parent	0.0372***	0.0050	–0.0098
	(0.0099)	(0.0187)	(0.0119)
Parents with secondary degree × Living with single parent	0.0560***	–0.0004	–0.0257**
	(0.0094)	(0.0187)	(0.0128)
Parents with high school × Living with single parent	0.0598***	0.0414*	0.0042
	(0.0096)	(0.0225)	(0.0217)
Parents with college, university × Living with single parent	0.0657***	0.0271	–0.0320
	(0.0098)	(0.0284)	(0.0373)
The near poorest quintile × Living with single parent				0.0315***	–0.0234	–0.0158
				(0.0100)	(0.0193)	(0.0111)
The middle quintile × Living with single parent				0.0384***	0.0147	–0.0129
				(0.0106)	(0.0211)	(0.0129)
The near richest quintile × Living with single parent				0.0648***	0.0204	–0.0219
				(0.0095)	(0.0220)	(0.0164)
The richest quintile × Living with single parent				0.0665***	0.0389*	–0.0261
				(0.0086)	(0.0215)	(0.0188)
Parents with less than primary	Reference
Parents with primary degree	0.0536***	0.1014***	0.0400***	0.0567***	0.1017***	0.0381***
	(0.0030)	(0.0077)	(0.0054)	(0.0029)	(0.0073)	(0.0049)
Parents with secondary degree	0.0650***	0.2018***	0.0999***	0.0696***	0.2016***	0.0961***
	(0.0030)	(0.0083)	(0.0067)	(0.0030)	(0.0079)	(0.0064)
Parents with high school degree	0.0649***	0.2624***	0.1706***	0.0697***	0.2660***	0.1700***
	(0.0031)	(0.0098)	(0.0093)	(0.0030)	(0.0095)	(0.0090)
Parents with college, university	0.0556***	0.2841***	0.2811***	0.0608***	0.2866***	0.2765***
	(0.0034)	(0.0119)	(0.0152)	(0.0033)	(0.0115)	(0.0148)
The poorest quintile	Reference
The near poorest quintile	0.0288***	0.0724***	0.0124**	0.0251***	0.0765***	0.0156**
	(0.0034)	(0.0082)	(0.0057)	(0.0034)	(0.0087)	(0.0062)
The middle quintile	0.0469***	0.1316***	0.0369***	0.0424***	0.1294***	0.0400***
	(0.0035)	(0.0092)	(0.0065)	(0.0035)	(0.0097)	(0.0069)
The near richest quintile	0.0621***	0.2039***	0.0918***	0.0558***	0.2015***	0.0960***
	(0.0035)	(0.0096)	(0.0077)	(0.0035)	(0.0100)	(0.0080)
The richest quintile	0.0710***	0.2926***	0.2069***	0.0646***	0.2886***	0.2113***
	(0.0036)	(0.0109)	(0.0096)	(0.0036)	(0.0111)	(0.0099)
Ethnic group fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Enumeration area fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Other control variables	Yes	Yes	Yes	Yes	Yes	Yes
Constant	0.3616***	–21.437***	–2.440***	0.3589***	–21.410***	–2.4257***
	(0.0256)	(1.2341)	(0.7750)	(0.0256)	(1.2338)	(0.7750)
Observations	158,256	55,353	60,690	158,256	55,353	60,690
R ²	0.205	0.369	0.322	0.205	0.369	0.322

Robust standard errors in parentheses. Standard errors are clustered at the enumeration area level. OLS: ordinary least squares.

***

p < 0.01; **p < 0.05; *p < 0.1.

Conclusions

Although more studies are needed to fully understand the relationship between parental absence and child education in Vietnam and other developing countries, this study has two main contributions. First, while most previous research focus on either parental migration or parental divorce, our study uses large-scale data from Vietnam, allowing for analysis of the effect of different types of parental absence (including migration and other family disruptions such as divorce or death) on school enrollment of children aged 7–22 years. We find that children from two-parent families have a higher enrollment rate at all levels of education than those from single-parent families. Within single-family types, the negative effect on children of parental divorce is higher than the effect of parental death, while the effect of parental migration is the lowest. Comparing the effect of single father and single-mother households, we find that children living with a single mother tend to have higher school enrollment than those living with a single father. Second, our study highlights the role that wealth and parental education can play an important role in mitigating the negative effect of parental absence on a child’s education. We find that the negative impact of parental absence in wealthier households is less pronounced than in poorer ones. The effect of parental absence is also weakened in families with better-educated parents.

Findings from our study suggest several policy implications. First, there can be educational supports for children with a single parent, especially those without parents. Currently, supports and subsidies are provided only for infant children, but not those living with a single parent. Second, it suggests that the country may consider policy interventions that strengthen the access to education of single parents, especially single mothers.

Our study has two limitations that should be noted. The first limitation is about the data set, which contains only information on school enrolment. There is no information on educational quality such as test scores in the data set. The second limitation is that we are not able to fully address the endogeneity problem in estimating the causal effect of parental absence. As a result, the interpretation of the causal effect should be taken with caution. Addressing the two limitations is beyond the scope of this study but certainly important for future studies.

Footnotes

Appendix 1

Table 9.

Heteroscedasticity-based IV regressions of children’s enrollment.

Explanatory variables	Dependent variables
	Enrolled in primary or secondary schools (yes = 1, no = 0)	Enrolled in high schools (yes = 1, no = 0)	Enrolled in college or university (yes = 1, no = 0)	Enrolled in primary or secondary schools (yes = 1, no = 0)	Enrolled in high schools (yes = 1, no = 0)	Enrolled in college or university (yes = 1, no = 0)
	(1)	(2)	(3)	(4)	(5)	(6)
Living with single parent	–0.0729***	–0.1029***	–0.1094***	–0.0717***	–0.1137***	–0.0445***
	(0.0148)	(0.0336)	(0.0266)	(0.0071)	(0.0162)	(0.0117)
Gender of children (male = 1, female = 0)	–0.0117***	–0.0913***	–0.0816***	–0.0142***	–0.0923***	–0.0704***
	(0.0013)	(0.0043)	(0.0042)	(0.0013)	(0.0040)	(0.0039)
Age of children	0.1070***	2.9469***	0.2894***	0.1091***	2.6904***	0.3478***
	(0.0027)	(0.1476)	(0.0766)	(0.0026)	(0.1359)	(0.0716)
Age of children squared	–0.0058***	–0.0910***	–0.0086***	–0.0059***	–0.0833***	–0.0101***
	(0.0001)	(0.0044)	(0.0019)	(0.0001)	(0.0040)	(0.0017)
Urban areas	0.0168***	0.1468***	0.1841***	–0.0072***	0.0148**	0.0445***
	(0.0023)	(0.0075)	(0.0079)	(0.0022)	(0.0061)	(0.0061)
Religion (yes = 1, no = 0)				–0.0027	–0.0184***	–0.0217***
				(0.0028)	(0.0066)	(0.0053)
Migration during past 5 years (yes = 1, no = 0)				–0.0463***	–0.1053***	–0.0504***
				(0.0072)	(0.0182)	(0.0116)
Household size				–0.0173***	–0.0344***	–0.0112***
				(0.0011)	(0.0023)	(0.0017)
Proportion of children				0.0393***	0.0634***	–0.0134
				(0.0075)	(0.0171)	(0.0170)
Proportion of elderly				0.0980***	0.1382***	0.0046
				(0.0130)	(0.0306)	(0.0212)
Parental age				0.0064***	0.0047	0.0037
				(0.0011)	(0.0039)	(0.0027)
Parental age squared				–0.0001***	–0.0001	–0.0000
				(0.0000)	(0.0000)	(0.0000)
Parents with an education degree				Reference
Parents with primary degree				0.0655***	0.1194***	0.0402***
				(0.0030)	(0.0068)	(0.0044)
Parents with a secondary degree				0.0804***	0.2201***	0.0921***
				(0.0030)	(0.0073)	(0.0057)
Parents with a high-school degree				0.0803***	0.2923***	0.1788***
				(0.0031)	(0.0085)	(0.0084)
Parents with college, university				0.0731***	0.3233***	0.3034***
				(0.0033)	(0.0100)	(0.0126)
The poorest quintile				Reference
The near poorest quintile				0.0304***	0.0723***	0.0066
				(0.0033)	(0.0073)	(0.0049)
The middle quintile				0.0467***	0.1274***	0.0312***
				(0.0034)	(0.0082)	(0.0056)
The near richest quintile				0.0617***	0.2046***	0.0925***
				(0.0033)	(0.0086)	(0.0067)
The richest quintile				0.0734***	0.3085***	0.2444***
				(0.0035)	(0.0096)	(0.0085)
Ethnic group fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Region fixed effects	No	No	No	Yes	Yes	Yes
Constant	0.4206***	–23.3821***	–2.1798***	0.3336***	–21.2312***	–2.8466***
	(0.0138)	(1.2491)	(0.7842)	(0.0258)	(1.1527)	(0.7385)
Observations	158,256	55,353	60,690	158,256	55,353	60,690
R ²	0.098	0.141	0.100	0.143	0.262	0.198

Robust standard errors in parentheses. Standard errors are clustered at the enumeration area level. IV: instrumental variable.

***

p < 0.01; **p < 0.05; *p < 0.1.

Acknowledgements

We would like to thank Judy McDonald, Aron Shlonsky, editor TY Wang, and two anonymous reviewers from the Journal of Asian and African Studies for their useful comments on this 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) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Cuong Viet Nguyen

Linh Hoang Vu

Availability of data and material

Data and code are available for replication.

Notes

Author biographies

Cuong Viet Nguyen holds a PhD in development economics from Wageningen University in the Netherlands. His main research fields are poverty, inequality and climate change. His research findings have been published in referee journals, such as the American Political Science Review, European Economic Review and Journal of Development Economics.

Linh Hoang Vu obtained his PhD in applied economics from the University of Minnesota, USA, specialising in the fields of poverty, agriculture and migration. His contributions to these areas have been published in referee journals, including the Journal of Agricultural and Resource Economics, Agricultural Economics, and Economics Letters.

References

Norton

(2003) Interaction terms in logit and probit models. Economics Letters 80(1): 123–129.

Angrist

Victor

Analia

(2010) Multiple experiments for the causal link between the quantity and quality of children. Journal of Labor Economics 28(4): 773–823.

Antman

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

Antman

(2013) The impact of migration on family left behind. In: Constant

Zimmermann

(eds) International Handbook on the Economics of Migration. Cheltenham: Edward Elgar, pp. 293–308.

Arlini

Yeoh

Yen

, et al. (2019) Parental migration and the educational enrolment of left-behind children: evidence from rural Ponorogo, Indonesia. Asian Population Studies 15(2): 190–208.

Becker

Tomes

(1976) Child endowments and the quantity and quality of children. Journal of Political Economy 84(4): S143–S162.

Beegle

De Weerdt

Dercon

(2006) Orphanhood and the long-run impact on children. American Journal of Agricultural Economics 88(5): 1266–1272.

Berg

Rostila

Saarela

, et al. (2014) Parental death during childhood and subsequent school performance. Pediatrics 133(4): 682–689.

Brand

Moore

Song

, et al. (2019) Parental divorce is not uniformly disruptive to children’s educational attainment. Proceedings of the National Academy of Sciences 116(15): 7266–7271.

10.

Burrell

Mehlum

Qin

(2020) Educational attainment in offspring bereaved by sudden parental death from external causes: a national cohort study from birth and throughout adulthood. Social Psychiatry and Psychiatric Epidemiology 55: 779–788.

11.

Cas

Frankenberg

Suriastini

, et al. (2014) The impact of parental death on child well-being: evidence from the Indian Ocean tsunami. Demography 51(2): 437–457.

12.

Case

Ardington

(2006) The impact of parental death on school outcomes: longitudinal evidence from South Africa. Demography 43(3): 401–420.

13.

Castro

Expósito-Casas

López-Martín

, et al. (2015) Parental involvement on student academic achievement: a meta-analysis. Educational Research Review 14: 33–46.

14.

Cebotari

Mazzucato

(2016) Educational performance of children of migrant parents in Ghana, Nigeria and Angola. Journal of Ethnic and Migration Studies 42(5): 834–856.

15.

Chae

(2016) Parental divorce and children’s schooling in rural Malawi. Demography 53(6): 1743–1770.

16.

Cronbach

(1951) Coefficient alpha and the internal structure of tests. Psychometrika 16(3): 297–334.

17.

Dang

Nguyen

OTT

, et al. (2022) Do positive income shocks benefit children’s education? Evidence from Vietnam. Applied Economics. Epub ahead of print 15 December. DOI: 10.1080/00036846.2022.2140111.

18.

De Loenzien

(2016) Lone motherhood and its educational outcomes for children in Vietnam. Marriage & Family Review 52(1–2): 162–195.

19.

Deaton

(1986) Life-cycle models of consumption: is the evidence consistent with the theory? NBER Working Papers 1910. National Bureau of Economic Research. Available at: https://www.nber.org/papers/w1910

20.

Dunusinghe

(2021) Impact of parental migration on children’s educational performance: evidence from Sri Lanka. Journal of Economics and Development Studies 9(1): 35–49.

21.

Filmer

Pritchett

(2001) Estimating wealth effects without expenditure data or tears: an application to educational enrollments in states of India. Demography 38(1): 115–132.

22.

Xue

, et al. (2017) Parental absence accompanies worse academic achievements: evidence-based upon a sample of left-behind children in rural China. Frontiers in Education 2: 1–5.

23.

General Statistics Office (2016) The 2015 national internal migration survey —major findings. Hanoi, Vietnam: Vietnam News Agency Publishing House. Available at: https://vietnam.unfpa.org/sites/default/files/pub-pdf/PD_Migration%20Booklet_ENG_printed%20in%202016.pdf

24.

General Statistics Office (2020) Completed results of the 2019 Vietnam population and housing census. Hanoi, Vietnam: Statistical Publishing House. Available at: https://www.gso.gov.vn/en/data-and-statistics/2020/11/completed-results-of-the-2019-viet-nam-population-and-housing-census/

25.

Gertler

Levine

Ames

(2004) Schooling and parental death. Review of Economics and Statistics 86(1): 211–225.

26.

Glewwe

(2004) An investigation of the determinants of school progress and academic achievement in Vietnam. In: Glewwe

Dollar

Agrawal

(eds) Economic Growth, Poverty, and Household Welfare in Vietnam. Washington, DC: The World Bank, pp. 467–501.

27.

Haveman

Wolfe

(1995) The determinants of children’s attainments: a review of methods and findings. Journal of Economic Literature 33(4): 1829–1878.

28.

Jeynes

(2016) A meta-analysis: the relationship between parental involvement and Latino student outcomes. Education and Urban Society 49(1): 4–28.

29.

Kailaheimo-Lönnqvist

Kotimäki

(2020) Cause of parental death and child’s health and education: the role of parental resources. SSM Population Health 11: 100632.

30.

Lance

Butts

Michels

(2006) The sources of four commonly reported cutoff criteria: what did they really say? Organizational Research Methods 9(2): 202–220.

31.

Lara

Saracostti

(2019) Effect of parental involvement on children’s academic achievement in Chile. Frontiers in Psychology 10: 1464.

32.

Lewbel

(2012) Using heteroscedasticity to identify and estimate mismeasured and endogenous regressor models. Journal of Business & Economic Statistics 30(1): 67–80.

33.

Liu

Grotta

Hiyoshi

, et al. (2022) School outcomes among children following death of a parent. JAMA Network Open 5(4): e223842.

34.

Shen

Krenn

, et al. (2016) A meta-analysis of the relationship between learning outcomes and parental involvement during early childhood education and early elementary education. Educational Psychology Review 28(4): 771–801.

35.

McKenzie

Rapoport

(2011) Can migration reduce educational attainment? Evidence from Mexico. Journal of Population Economics 24(4): 1331–1358.

36.

McLanahan

Tach

Schneider

(2013) The causal effects of father absence. Annual Review of Sociology 39(1): 399–427.

37.

Mao

Zang

Zhang

(2020) The effects of parental absence on children development: evidence from left-behind children in China. International Journal of Environmental Research and Public Health 17(18): 6770.

38.

Mont

Nguyen

(2013) Does parental disability matter to child education? Evidence from Vietnam. World Development 48: 88–107.

39.

Morooka

Liang

(2009) International migration, and education of left-behind children in Fujian, China. Asian Pacific Migration Journal 18(3): 345–370.

40.

Nguyen

(2016) Does parental migration really benefit left-behind children? Comparative evidence from Ethiopia, India, Peru, and Vietnam. Social Science & Medicine 153: 230–239.

41.

Nguyen

(2020) Intra-generational and intergenerational social mobility: evidence from Vietnam. Journal of Asian and African Studies 55(3): 370–397.

42.

Nguyen

(2016) Should parents work away from or close to home? the effect of parental absence on children’s time use in Vietnam. Journal of Human Development and Capabilities 17(1): 110–124.

43.

Park

(2007) Single parenthood and children’s reading performance in Asia. Journal of Marriage and Family 69(3): 863–877.

44.

Pholphirul

Teimtad

(2018) Living with parents and educational outcomes in developing countries: empirical evidence from PISA Thailand. Journal of Population Research 35: 87–105.

45.

Pörtner

(2016) Effects of parental absence on child labor and school attendance in the Philippines. Review of Economics of the Household 14(1): 103–130.

46.

Sebastian

Moon

Cunningham

(2017) The relationship of school-based parental involvement with student achievement: a comparison of principal and parent survey reports from PISA 2012. Educational Studies 43(2): 123–146.

47.

Tárraga

García

Reyes

(2017) Home-based family involvement and academic achievement: a case study in primary education. Educational Studies 44(3): 361–375.

48.

UNICEF (2021) Viet Nam SDGCW Survey 2020–2021. Education. Available at. https://www.unicef.org/vietnam/media/8686/file/Education.pdf (accessed 18 January 2023).

49.

Yang

Fan

(2012) Migration, family structure, and high school enrollment: an analysis of China’s 2000 Census. Chinese Sociological Review 44(4): 22–57.

50.

Zhang

(2020) Are children from divorced single-parent families disadvantaged? New evidence from the China family panel studies. Chinese Sociological Review 52(1): 84–114.

51.

Zhang

Behrman

Fan

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

52.

Zhou

Murphy

Tao

(2014) Effects of parents’ migration on the education of children left behind in rural China. Population and Development Review 40(2): 273–292.