Leaving A Legacy: Parental Migration and School Outcomes Among Young Children in the Philippines

Abstract

This article examines the link between parental migration and young children’s education using data from the Philippine country study of the Child Health and Migrant Parents in South-East Asia (CHAMPSEA) Project. The key research question probed here is: what difference does parental migration make to the school outcomes of young children? Specifically, it looks at factors that explain children’s school progression (school pacing) and academic performance (school achievement) using multiple regression analysis. These questions are explored using CHAMPSEA data gathered from a survey of children under 12 years of age and their households in Laguna and Batangas (n=487).

The concern that parental absence due to migration can negatively affect the school performance of children is not supported by the study. If parental migration affects school outcomes, it is associated with positive outcomes, or with outcomes which show that children in transnational households are not doing worse than children living with both parents. Positive school outcomes are best associated with a migrant-carer arrangement where fathers work abroad and mothers stay home as carers –children in these households fare very well when it comes to school pacing and school achievement. The study concludes that families and households need to provide both economic and psychological support to enhance the chances that children are at pace with their schooling and are doing well at school.

Background

We always think about the future of our kids … we hope they can finish their studies. That’s why their father is working abroad so he can support their studies. Had he stayed here, we would have nothing to support our children’s education. – Mother-carer of 11 year-old girl

… I tell them they should really study well as that will be the only legacy that their mother can give them. That’s why she works hard abroad so they will be able to finish their studies. – Aunt-carer of four year-old girl¹

The statements above indicate how supporting their children’s education is a responsibility Filipino parents try to meet. In encouraging their children to do well in school, it is customary for Filipino parents to say that education is the best legacy they can bequeath to the young ones. When economic resources constrain parents’ capacity to perform this role, parents may turn to migration to realize their dreams for their children. The aspiration to provide a better future for their children via education plays a crucial role in the decision of Filipinos to work abroad (ECMI/AOS-Manila et al., 2004; Aguilar et al., 2009; Asis et al., 2005). However, public perceptions about the impact of parental migration on children’s well-being see migration as a double-edged sword. On the one hand, remittances are seen as enhancing the children’s material well-being, but on the other hand, parental absence is perceived to deprive children of emotional support and care that are detrimental to the children’s welfare. With respect to children’s education, it is not uncommon to hear opinions that parents went abroad to further their children’s education, but their efforts were for naught because their children either stopped schooling or encountered problems in school. The culprit: parental absence (especially the absence of mothers) and the presumption that because parents are not around, the children are lacking in guidance. In general, the negative press about the presumed ill effects of migration overshadows accounts of children of migrants performing well in school. News in the media of the children of migrants garnering honors and citations during graduation exercises or awards given to outstanding children of overseas Filipinos tends to be viewed as the exception rather than the norm.²

Extant research findings, particularly studies that provide a comparison with non-migrant families, tend to be less pessimistic about the impact of parental migration on children’s education. Several studies in the Philippines have shown that supporting the education of children is one of the major uses of remittances (e.g., ECMI/AOS-Manila et al., 2004; Yang, 2006; Aguilar et al., 2009). Contrary to perceptions that remittances mostly go into the purchase of durable goods, findings from Yang’s study (2006) indicate that families of migrant workers who received increased remittances (due to the higher dollar-peso exchange rate during the 1997 financial crisis in Asia) used remittances to invest in children’s education and start small businesses. The ECMI/AOS-Manila (2004) study, a nationwide survey of 1,443 children in the ages 10-12 conducted in 2003, revealed that the children of overseas Filipino workers (OFWs) had a higher percentage attending private schools compared to the children of non-migrants. Furthermore, similar to an earlier study (Battistella and Conaco, 1998), the 2003 study indicated that the children of OFWs performed just as well, if not better, than the children of non-migrants, with the exception of the children of mother-migrants. In a study of older children 13-16 years old, children of OFWs were found to be more likely to join academic organizations and participate in extra-curricular activities; also, they were more likely to receive school awards compared with the children of non-OFWs (Edillon, 2008). Limiting the sample to children belonging to two-parent families (and thereby controlling for variations in family structures), overall, the study found that two-parent families tended to cope with the strains and stresses of migration. In turn, this contributes to the children’s adjustment to the absence of one or both parents. The fact that the care of children remains in the hands of family members suggests that the rearing, nurture and socialization of the children of migrants are likely to adhere to how the migrant parents would raise the children if they were around. In the Philippines, access to the support provided by the extended family has assuaged migrant parents’ anxieties about leaving their children (see Asis et al., 2005; ECMI/AOS-Manila et al., 2004; Aguilar et al., 2009).

In this article, we examine the links between parental migration and children’s education using data from the Philippine country study of the Child Health and Migrant Parents in South-East Asia Project (hereafter CHAMPSEA-Philippines). Undertaken between 2008 and 2010 in Indonesia, the Philippines, Thailand and Vietnam, the CHAMPSEA Project sought to examine how the migration of one or both parents affects the health and well-being of young children under 12 years old. Hypotheses concerning these links compare the children in transnational households (i.e., households with one or both parents who were working abroad) with those in usually resident households (i.e., households wherein both parents co-reside with their children) and examine the impact of migration on child health and other well-being indicators in terms of who migrates, who assumes care for the children, the characteristics of the child (particularly gender), and selected community variables (see Graham and Yeoh, in this volume).

We focus on education because of the great value placed on this aspect of child welfare in the Philippines. As mentioned earlier, existing research has suggested that remittances from migration contribute to higher likelihood of school attendance (Yang,2006; Aguilar et al., 2009; ECMI/AOS-Manila et al., 2004), especially higher attendance in private schools which is considered “quality education” in contrast to public school education (ECMI/AOS-Manila et al., 2004; Aguilar et al., 2009). This article turns to another facet of children’s schooling – school outcomes – and to examine the factors that are associated with school progression and academic performance. School outcomes here refer to two measures: school pacing, which indicates whether the child’s progress or pace in school is in keeping with expected progress by a given age, and school achievement, which indicates the academic performance of a child. The key research question probed here is: what difference does parental migration make to the school outcomes of young children? This key question is broken down into three specific research questions: (1) how do the children in transnational households compare with the children in usually resident (non-migrant) households on education-related indicators?; (2) what factors account for the school outcomes of children in transnational and usually resident households?; and (3) how does the impact of migration compare with other factors in explaining the school outcomes of children? Given the data set, these questions focus on children 9-11 years old. Between presumptions about the positive role of remittances in keeping children in school on the one hand, and alleged negative consequences of parental absence on children’s school performance on the other, the study aims to contribute to a better understanding of whether and how migration affects the school outcomes of young children in the Philippines.

Context and Conceptual Framework

School Outcomes of Filipino Children

One of the outcomes targeted by the Millennium Development Goals (MDGs) is to ensure that by 2015, “children everywhere, boys and girls alike, will be able to complete a full course of primary schooling.” Between 1990, the baseline year, and the most recently available data as of 2011, Table 1 shows that the Philippines slowly progressed in increasing the net enrollment ratio in primary education. However, the proportion of pupils who reach grade 6 and complete primary school is still markedly off the target, 74 percent as of 2011. In contrast, related indicators for promoting gender equality in the primary grades (goal 3) indicate that the Philippines has met the targets way ahead of 2015 – by 2011, the ratio of girls to boys in primary education and the ratio of girls to boys participation rates have been met. Overall, gender disparity in education slightly favors girls in terms of the ratio of girls to boys in primary education, secondary participation rates and tertiary education.

TABLE 1. MDG Indicators on Primary Education and Greater Equality: Philippines.

GOAL 2. ACHIEVE UNIVERSAL PRIMARY EDUCATION
target 2.A	Ensure that, by 2015, children everywhere, boys and girls alike, will be able to complete a full course of primary schooling
Indicator		1990	2011	2015
indicator 2.1	Net enrollment ratio in primary education	84.6	91.2	100.0
indicator 2.2	Proportion of pupils starting grade 1 who reach grade 6 (Cohort Survival Rate)	69.7	73.7	100.0
indicator 2.2a	Primary completion rate	64.2	80.0	100.0
indicator 2.3	Literacy rate of 15 to 24 years old	96.6	97.8	100.0
indicator 2.3a	Ratio of literate females to males of 15-24 year-old	1.0	1.0	1.0
GOAL 3. PROMOTE GENDER EQUALITY AND EMPOWER WOMEN
target 3.A	Eliminate gender disparity in primary and secondary education preferably by 2005 and to all levels of education no later than 2015
Indicator		1996	2011	2015
indicator 3.1a	Ratio of girls to boys in primary education	1.0	1.1	1.0
indicator 3.1a.1	Ratio of girls to boys in elementary participation rates	1.0	1.0	1.0
indicator 3.1b	Ratio of girls to boys in secondary education	1.1	1.0	1.0
indicator 3.1b.1	Ratio of girls to boys in secondary participation rates	1.2	1.2	1.0
indicator 3.1c	Ratio of girls to boys in tertiary education	1.3 (1993)	1.2	1.0
indicator 3.2	Share of women in wage employment in the non-agricultural sector	40.1 (1990)	41.4	50.0
indicator 3.3	Proportion of seats held by women in national parliament	11.3 (1992)	26.0	50.0

Excerpted from MDGWatch: Statistics at a glance of the Philippines’ Progress based on the MDG Indicators, as of September 2013, http://www.nscb.gov.ph/stats/mdg/mdg_watch.asp

Using data from the 2002 and 2004 Annual Poverty Indicators Survey, Maligalig and Albert (2008:5) found that the three most important reasons given why elementary age children do not attend school are: (1) lack of personal interest, 29 percent and 29.4 percent in 2002 and 2004, respectively; (2) others (could be too young to go to school, not admitted in school, or lack of documents such as birth certificate), 25.3 percent and 27 percent; and (3) high cost of education, 14.8 percent and 15 percent. According to them, lack of interest may actually reflect lack of financial resources, as borne out by data showing that non-attendance in school decreases as the children’s households increase in income level.⁴ Furthermore, children in poor families may forego education and work instead not only because of resource constraints but also the poor quality of education which does not encourage students to remain in school.

As Table 1 shows, gender disparities in education at all levels are either non-existent or favor girls. The net enrollment ratio in the elementary level used to favor boys until 2005, after which girls outnumbered boys. In general, girls have a higher completion rate at each level (elementary, secondary and tertiary), higher cohort survival rate and higher academic performance (Tan et al., 2011:4). Cultural factors, such as the tendency of Filipino families to be more protective of girls, may incline girls more towards school work. Also, many boys in poor families may stop schooling to work and to augment family income (Tan et al., 2011: 4). Or, in the face of limited resources, various studies have suggested that Filipino families may invest more in daughters’ education than in sons’ because the former are viewed as more committed to their families (Arguillas and Williams, 2010).

Schooling outcomes have also been explored in relation to early growth retardation. One aspect which has been documented by various studies pertains to the association of height-for-age with retention and promotion in school. Based on a longitudinal data collected from 2000 children in Cebu, Daniels and Adair (2004) examined the effect of height-for-age on age at initial enrollment into school, grade repetition, completion of primary school, and completion of high school. They found that low height-for-age is associated with late enrollment (perhaps because parents perceive tall children as more ready to learn) and the percentage of boys ever repeating a grade was more than double that of girls. Boys and girls who were taller at two years old were markedly less likely to drop out in grade school, to be behind in school, and to drop out of high school. According to the authors, the Philippine context presents contrasting profiles, characterized by high levels of nutritional deficiency among children (the country was in a recession when data were collected in the 1980s) on the one hand, and strong national support for education (mandatory primary school enrollment for all Filipino children) on the other. Although enrollment rates are impressive at elementary school level, lower completion rates reveal the impact of financial constraints that keep children out of school. Against economic odds, Filipino families strive to provide a good education for their children, and children are often reminded of the sacrifices their parents make to put them to school (ECMI/AOS-Manila et al., 2004; Asis, 2006a; Aguilar et al., 2009).

In addition to the studies cited earlier, Arguillas and Williams’ (2010) work provides insights into the impact of parental migration on the education of children in the ages 19-21. They considered three education outcomes: number of years of completed schooling, whether or not the child had completed high school, and whether or not the child had obtained some college education. A key independent variable is the family structure which took into account whether a parent had been overseas before the outcome of interest had occurred: mother had been overseas, father had been overseas, both parents had been overseas, and both biological parents were at home and had not been overseas. The multivariate analysis shows that in general, there were no statistically significant differences in the schooling outcomes of children residing with both parents and those with parents working abroad (except in cases where both parents were working abroad). Interestingly, where differences surfaced, they were sometimes positive for mother-migrant families. Children whose mothers were working overseas had completed more total years of schooling. Girls were found to fare better than their brothers on two educational measures: matriculation in college and total number of years of schooling. Where both parents had worked abroad, sons tended to complete fewer years of education than those whose parents were at home.

Key Findings in Other Contexts

The relationship between parental migration, care arrangements and school outcomes has been the subject of research in other contexts as well. The variety of methodological approaches, determinants and outcomes of interest, and age group of the children considered have yielded rich but mixed results. School outcomes have been examined in terms of school attendance, school retention, grades, completion, and progression to higher levels of education.

Parental migration has been posited as having a positive impact on household economic resources through remittances which increase household educational investment, reduce child labor, and mitigate the negative effect of parental absence. This found support in Lu and Treiman’s (2007) study of the impact of remittances sent by South African Black labor migrants to their families. Remittances also contribute to reducing gender inequalities within families and socio-economic inequities in schooling. The study noted that among migrant households, children in remittance-receiving households are better off than children in households which do not receive remittances – the latter in fact were sometimes worse off compared to children in non-migrant households. Without remittances, migrant households do not have the economic resources to compensate for the difficulties resulting from parental migration. The impact of labor migration on Mexican children’s educational aspirations and performance is mixed. Based on a random sample of 7,600 of grammar, junior high and high school level students in a Mexican migrant-sending state, high levels of US immigration were found to be associated with lower aspirations to attend a university at all levels, but a positive relation was found between US migration and grades (Kandel and Kao, 2001). According to the authors, it appears that temporary labor migration affects children in two ways: remittances allow parents to put the children to school and to enhance the children’s well-being, but on the other hand, labor migration offers an alternative pathway to mobility which reduces motivation to aspire for higher education.

In general, the impact of remittances on various school outcomes tends to be positive, although the outcomes may vary according to the gender of the migrant or the child in some cases. In rural Bangladesh, the migration of fathers and brothers has a positive impact on the schooling of both boys and girls. The migration of sisters had no effect on their siblings’ education and mothers in migration were too few to arrive at more conclusive findings (Kuhn, 2006).

On the negative side, parental absence due to migration is perceived to deprive children of important support which can result in poor school outcomes. Although it is not focused on the families of migrants, a study in the UK looking into parental engagement and student learning identified social and economic factors that prevent many parents from fully engaging with their children’s schooling. The study specified that while it is important to involve parents in school-based activities, the involvement of parents in learning at home is most likely to result in a positive difference to the learning outcomes of young students (Harris and Goodall, 2008). Studies such as this imply that when parents are absent, there is a void in the care and guidance provided to school-age children.

As cited earlier, studies in the Philippines found children in mother-migrant families tending to lag behind in academic performance compared with other children (Battistella and Conaco, 1998; ECMI /AOS-Manila et al., 2004; Asis, 2006b). This may be linked to the difficulties children and families experience when it is the mothers who leave (e.g., ECMI/AOS-Manila et al., 2004; Carandang et al., 2007). In Thailand, the long-term absence of migrant mothers had an adverse effect on school enrollment, but the migration of fathers had no such impact (Jampaklay, 2006). In Sri Lanka, 62 percent of 400 surveyed households reported changes in the behavior of children in the absence of their mothers, of which problems at school were the most frequent (26.5 percent).

Research-based knowledge on the impact of parental migration in Asia converges on the finding that overall, the children left behind are not necessarily worse off than the children living with both parents. Children and their families may not find the migration of one or both parents ideal, but in time, they learn to adjust to this reality and make the best of the situation. The availability, affordability, and immediacy offered by a variety of communication facilities have contributed to narrowing the time and space that separates migrants and their family members. In addition, the support and assistance provided by the extended family help fill the gap left by migrant parents and assure migrants that their children will be cared for while they are away (Battistella and Conaco, 1998; ECMI/AOS-Manila et al., 2004; Asis, 2006a, 2006b; Hugo and Ukwatta, 2010).

Conceptual Framework

Schooling outcomes could result from a complex web of factors. From the literature survey, we identified four sets of variables that can affect school outcomes. These include: (1) the household care resources – this cluster covers the migrant-carer status of the household, family functioning, and the index child’s or IC’s number of siblings; and family process (as indicated by the household’s migration status and caring arrangement, family functioning and the IC’s number of siblings); (2) the IC’s characteristics, including psychological well-being and nutritional status (age, sex, total difficulties and stunting); (3) household material resources (wealth index, mother’s education and father’s education); and (4) community context, as indicated by urban residence (Figure 1).

FIGURE 1. Parental Migration, Care Arrangements and School Outcomes.

As outlined in Figure 1, we consider the household as having different resources and constraints that determines its capacity to support the schooling of young children. A household’s economic status and child care arrangements can be reconfigured in the face of migration. A household’s caring and nurturing function towards its young members depends on who migrates (who will provide economic support) and who assumes caregiving. The migrant-carer status of a household is posited as a key variable in defining children’s school outcomes. This cluster also include the child’s assessment of satisfaction with family relationships which can be an indication of how well the family provides emotional support while the number of siblings present in the household can have a bearing on the competition for care (as well as material) resources within the household.

A household is also a repository of material resources – a household with ample resource can invest more in children’s education. Wealth and the human capital resources of a child’s parents can define the educational chances of young children – access to quality education, opportunities to participate in extra-curricular activities, or capacity to avail themselves of educational assistance such as tutors.

The individual characteristics of the IC are significant in school outcomes – a physical condition, such as stunting or psychological distress, can delay a child’s progress in school or could get in the way of the learning process. Boys and girls may respond to enabling factors that promote appropriate school pacing and/or academic performance. Based on data from the Cebu Longitudinal Health and Nutrition Survey, Carvalho’s (2012) findings that childhood health and nutrition, cognitive and non-cognitive abilities, and early schooling account for one-third to one-half of the intergenerational transmission of socio-economic status reinforces the importance of knowing more about school outcomes. In the context of a high migration environment, it would be interesting to explore whether the economic benefits of migration translate or contribute to better school outcomes.

The final variable is the location of households. The urban, semi-urban, or rural context of households has implications on access to schools, the scope of educational options available to households, and quality of education. In general, more educational institutions are located in urban areas, but on the other hand, the quality of education may be affected adversely by larger classes, among others factors, in urban areas. The urban disadvantage surfaced in the results of the National Achievement Test (NAT) conducted for Grade 3, Grade 6 and fourth year high school students. The NAT results for Grade 6 students show higher mean percentage scores for students in rural areas compared to those in urban areas. Rural students had mean percentage scores of 60.81 percent, 65.52 percent and 66.67 percent for 2007, 2008 and 2009, respectively while urban students scored 59.48 percent, 64.43 percent, and 64.15 percent for the same years. Education officials pointed to congestion and distractions in urban areas, such as malls, as probable reasons why urban students trail behind (Tubeza, 2009).

Data and Methods

The Philippine component of the four-country CHAMPSEA Project was conducted in the provinces of Laguna and Bulacan. Within each province, communities that are more urban in character and communities that are more rural in character were chosen. These two provinces share similar characteristics: both have had a long history of international labor migration and are among the top origin provinces of international labor migrants; both provinces are situated on Luzon island; they share a common language (Tagalog), an indicator of ethnicity; and both are among the more developed provinces in the country. The similarities “control” for variations that may otherwise be introduced by having different characteristics.

In common with other study countries in the CHAMPSEA Project, data collection was conducted in 2008. Households were selected by flexible quota sampling, with a target of 1,000 households with index children in one of two age groups (3-5 years, and 9-11 years). The absence of a suitable sample frame means that the sample is not nationally representative; rather its composition reflects the high out-migration communities selected. The data employed in this article were drawn from the survey of 1,000 households (509 from Laguna and 491 from Bulacan). The sample included both usually resident or non-migrant households and transnational or migrant households; only two-parent households were eligible to participate. For transnational households, the migrant parents had to be current migrants, i.e., they must have been working overseas for at least six months and they must not have returned home for visits during the same period. For each sample household, interviews were conducted in Tagalog with a responsible adult regarding household-related matters, the primary carer (who may be the same person as the responsible adult) of the index child, and the older child. One index child (hereafter IC) was selected per household according to quota requirements, or randomly where there was more than one qualifying child.⁵ The IC is either a young child (those aged 3-5 years old) or an older child (those in the ages 9-11). Interviews were not conducted with the young children; instead, they were invited to make a drawing of their family. The purpose of the study was explained to all eligible participants and their informed consent acquired before proceeding with the interviews.⁶ For the participating households, anthropometric measurements, including height and weight, were taken of the index children and their siblings who were between ages two and under 12 years old. The analysis of school outcomes is limited to a sub-sample of 487 elementary school-aged children. The analytical sample for the present study comprises 487 households with index children aged between 9 and 11 years, 13 cases where data were missing were dropped, which is 2.6 percent of the total sample. In terms of migration status, the sample is divided into 237 usually resident or non-migrant and 250 transnational or migrant households. In terms of child care arrangements, households were classified into: mother-carer, father-carer, or other-carer households. Information on carer arrangement was combined with the migration status of the household to create the variable migrant-carer status, a critical variable for this study. Usually resident or non-migrant households were those where children were co-resident with both parents while transnational or migrant households were those where one or both parents were working abroad. Transnational households were further categorized into the following groups: father-migrant, mother-migrant, and father and mother-migrants.⁷

Multiple regression was employed to analyze the links between parental migration and school outcomes. For school pacing, we conducted ordinary least squares regression because the dependent variable, school pacing, is an interval variable. For school achievement, we used ordered logistic regression given the ordinal nature of the dependent variable. Considering the focus of the CHAMPSEA study on parental migration and the health and well-being outcomes on the left-behind children, we started with a basic model introducing the migrant-carer status of the household as the lone independent variable to test its relationship with school pacing and school achievement. Building on this basic model, other variables were added – i.e., child’s characteristics, household wealth, family functioning, and urban residence – a process which allows a comparison of the contribution of additional variables in explaining the two measures of school outcomes. In the discussion of the results, the results of the basic model and the full model are highlighted.

Definition of Variables

Table 2 presents the variables that were considered in the multivariate analysis. The first outcome measure examined is school pacing, which indicates whether a child is behind, at pace or advanced in school level in relation to his/her age. Kuhn’s study (2006) in Bangladesh used the same variable as a dependent variable and we adopted the same procedure in the construction of school pacing. In the Philippine school system, compulsory education begins at age six.⁸ Thus, ages 9-11 would correspond to Grade 4 for nine-year olds, Grade 5 for 10-year-olds, and Grade 6 for 11-year-olds.

TABLE 2. Variables in the Multivariate Analysis.

Outcome Variables
School pacing	Continuous measure of school pacing/progression. It was created by subtracting age in completed years from years of completed schooling plus 6 to account for the average and prescribed age for starting Grade 1.
	Range: −2 to 2
School achievement	An index created from three variables: IC’s class position (1=high; 0=otherwise),9 whether IC has received a positive school report in the past six months (1=yes; 0=no), whether IC has received a negative school report in the past six months (reverse coded, 1=no; 0=yes).
	Range: 0 to 2 (where higher value is better).
Independent Variables
Migrant-carer status	Combined measure of a household’s migration status and caregiving arrangements.
	0=NM
	1=TH- mother carer (father migrant)
	2=TH- father carer (mother migrant)
	3=TH- other carer (father or mother or both parents are migrants and carer is other relative)
Total difficulties	It is the summative continuous measure of the Strengths and Difficulties Questionnaire (SDQ) subscales excluding the prosocial scale. The measure was transformed into a binary indicator using the UK cutoff points for ‘abnormal cases’ which are greater or equal to 17. (The UK cutoff point was used in the absence of a Philippine-specific cutoff point.)10
	1=scored 17 or more (denoting difficulties); 0 otherwise
Stunting	Binary indicator (stunt) created from height-for-age standardized measure (HAZ1) using <−2 SD.
	1 indicates stunting; 0 otherwise
IC’s age	Categorical measure of index child’s completed years of age where possible values are: 9, 10, 11 Reference category: 9
IC is girl	IC girl=1; 0=boy
Household socio-economic status	Categorical measure that ranks households based on level of wealth
	1= low wealth (1st and 2nd quintiles)
	2= average wealth (3rd and 4th quintiles)
	3= high wealth (5th quintile)
Mother’s education	Binary variable created from years of completed schooling where 1=primary or lower secondary 2= upper secondary or higher
Father’s education	Binary variable created from years of completed schooling where 1=primary or lower secondary 2= upper secondary or higher
Family functioning	A binary indicator created from the summed total of five standard items for family functioning based on the Family APGAR (Adaptability, Partnership, Growth, Affection and Resolve)11 score and using 12/13 as the cutoff.
	1 = good family functioning (APGAR score 13 or above)
	0 = not good family functioning (APGAR score 12 or below).
Number of siblings	0= none  1=one  2=two or more
Residence	Categorical measure of residence where 1= urban  2= semi-urban  3= rural town/village Reference category: 1

The other school outcome of interest is school achievement, which is based on an index created from three variables, all derived from the carer’s report: the class rank or position of the child, whether the child has received a positive school report in the past six months, and whether the child has received a negative report in the past six months. An index, ranging in value from 0 to 2 (higher scores are positive outcomes), was developed from these three variables.

Table 3 shows the profile of the sample of 487 children aged 9-11 and the characteristics of their households in terms of the variables of interest. The sample is about evenly divided in terms of gender composition (49.5 percent were males; 50.5 percent were females). By age, 37.4 percent are 9 years old, 39.4 percent are aged 10, and 23.2 percent are aged 11. About half (48.6 percent) live in households with both parents present, while the rest – 51.4 percent – are part of transnational households, of whom the largest group (34.7 percent) are in father-migrant households.

TABLE 3. Socio-demographic Profile of the 9-11 Year Olds: CHAMPSEA-Philippines (N=487).

Characteristic	Percent	Characteristic	Percent
Gender		Family functioning
Male	49.5	Good	57.2
Female	50.5	Not good	42.8
Age		Number of siblings
9	37.4	None	6.2
10	39.4	One	30.4
11	23.2	Two or more	63.4
% of children with psychological difficulties (based on SDQ total difficulties score)		Mother’s education
		Primary or lower secondary	24.8
	19.1	Upper secondary or higher	75.2
% stunted	15.2	Father’s education
		Primary or lower secondary	25.7
Migrant-carer status		Upper secondary or higher	74.3
NM	48.7
TH-mother carer	32.4	Residence
TH-father carer	8.4	Urban	61.8
TH-other carer	10.5	Semi-urban	20.7
		Rural town/village	17.4
Household socio-economic status (Wealth Index)
Low wealth (1st-2nd quintile)	41.3
Medium wealth (3rd-4th quintile)	40.7
High wealth (5th quintile)	18.1

The global measure of well-being (derived from responses to the question, “In general, are you happy or unhappy?”) shows that the majority (83 percent) of the children reported that they were happy or very happy.¹² This is supported by their scores in the Strengths and Difficulties Questionnaire (SDQ), an instrument developed to detect psychological difficulties. The total difficulties score sums four sub-scales: emotional symptoms, conduct problems, hyperactivity, and peer relationship problems. Based on a standard cutoff used in the UK (in the absence of a Philippine standard), 81 percent had no psychological difficulties while 19.1 percent were detected as likely to have these difficulties. Asked about what makes them happy, 46 percent cited family-related reasons (mostly, the whole family being together). The other reasons given include playing (28 percent), friends (10 percent) toys/presents (3.3 percent), good grades/praise at school (2.5 percent), and other reasons (11 percent). The things that made them unhappy were mostly things that do not go well with their families (49 percent): getting in trouble with their parents (19 percent), mother or father being away (19 percent) and quarrel with siblings (11 percent). School-related factors accounted for 16 percent of reasons provided (rejection/bullying/quarrel with classmates, 12 percent, and bad grades, four percent).

School Outcomes

Eight out of ten children reported that they enjoy school almost always or always. Based on their own assessment of their grades in school, 66 percent said their grades were about the same as their classmates’, 24 percent said their grades were better, and 10 percent considered their grades as worse than their classmates’.

The bivariate analysis of school outcomes and the migrant-carer status of the children’s households suggests some association between these two variables (Table 4). Children in TH-mother carer households seem to register the most favorable school outcomes: about a third (34.2 percent) of them were advanced in school pacing and a little more than a quarter (27.2 percent) scored the highest in school index. Contrary to earlier findings (e.g., Battistella and Conaco, 1998; ECMI/AOS-Manila et al., 2004) which reported that children left behind by mother migrants were disadvantaged in school performance and psychological measures compared to other children, the present study reveals that children in TH-father carer households were doing well in school achievement. Overall, the children in NM households fared rather poorly compared to children in TH households. As a group, they had the highest share of children who fell behind in terms of school pacing (they were ranked third among those who were advanced), and they ranked last for children who were above average in class position and in getting a positive school report compared to children in TH households. Based on the school index, the children in non-migrant households comprised the largest group among those who scored 0 and were the least likely to score the highest. The strength of the migrant-carer variable in explaining these variations is explored next in the multivariate analysis wherein the role of other variables will be considered.

TABLE 4. School Outcomes by Migrant-Carer Status: CHAMPSEA-Philippines.

	Migrant-Carer Status
School Outcomes	NM	TH-mother carer	TH-father carer	TH-other carer	Total
School pacing
Behind	21.5	10.1	19.5	15.7	17.0
On pace	54.0	55.7	61.0	51.0	54.8
Advanced	24.5 (237)	34.2 (158)	19.5 (41)	33.3 (51)	28.1 (487)
Class position
Above average	19.8	34.2	34.1	19.6	25.7
Average	75.5	62.2	63.4	74.5	70.0
Below average	4.6 (237)	3.8 (158)	2.4 (41)	5.9 (51)	4.3 (487)
Positive school report?
% yes	35.0 (237)	55.7 (158)	46.3 (41)	39.2 (51)	43.1 (487)
Negative school report?
% no	86.1 (237)	84.8 (158)	90.2 (41)	94.1 (51)	86.9 (487)
School achievement index
0	65.0	45.6	51.2	56.9	56.7
1	21.9	27.2	17.0	29.4	24.0
2	13.1 (237)	27.2 (158)	31.7 (41)	13.7 (51)	19.3 (487)

School Pacing

The results of the basic OLS model show that of the migrant-carer categories, only TH-mother carer is significantly related to school pacing (Table 5). Compared with children in non-migrant households, those in TH-mother carer households are more likely to perform well in school pacing and this difference is significant. The data suggest that children in TH-father carer and TH-other carer household are not any more positively or negatively affected by the absence of mothers or both parents.

TABLE 5. Regression Analysis of School Pacing.

	Basic Model			Full Model
Variables	Coefficients	95%	CI	Coefficients	95%	CI
Constant	−0.0127	−0.12	0.09	−0.0275 (0.1979)	−0.42	0.36
Migrant-Carer Status
(NM)
TH-mother carer	0.2658 *** (0.0850)	0.10	0.43	0.1767 * (0.0893)	0.00	0.35
TH-father carer	0.0127 (0.1400)	−0.26	0.29	−0.0706 (0.1391)	−0.34	0.20
TH-other carer	0.2087 (0.1278)	−0.04	0.46	0.0867 (0.1307)	−0.17	0.34
IC is a girl				0.0652 (0.0724)	−0.08	0.21
IC’s age
(9 years)
10 years				0.0144 (0.0838)	−0.15	0.18
11 years				−0.1778 (0.0967)	−0.37	0.01
IC has psychological difficulties				−0.2984 *** (0.0943)	−0.48	−0.11
IC is stunted				−0.2592 ** (0.1059)	−0.47	−0.05
Household socio-economic status
(Low)
Medium household wealth				0.0677 (0.0924)	−0.11	0.25
High household wealth				−0.0056 (0.1193)	−0.24	0.23
Father has high education				0.0714 (0.0989)	−0.12	0.27
Mother has high education				0.2831 ** (0.0963)	0.10	0.47
IC’s number of siblings
(No sibling)
One sibling				−0.0698 (0.1636)	−0.39	0.25
2 or more siblings				−0.1352 (0.1594) (0.0746)	−0.45	0.18
Family functioning is good				0.0598 (0.0746)	−0.09	0.21
Residence
(Urban)
Semi-urban				−0.0638 (0.0965)	−0.25	0.13
Rural town/village				−0.2236 * (0.0971)	−0.41	−0.03
R-squared	0.0361			0.1315
Adjusted R-squared	0.0261			0.1000
No. of observations	487			487

^*p < .05,

^**p < .01,

^***p < .001

Values in parentheses refer to the standard error.

Categories in parentheses refer to the omitted category.

When other variables are added to the model, TH-mother carer continues to exert a significant positive effect on school pacing (although the effect size is reduced). As in the basic model, children in TH-mother carer households are more likely to do well in school pacing compared to children in NM households. However, children in TH-father carer and TH-other carer households are not more or less advantaged in school pacing compared to children in non-migrant households. In the full model, the four other variables that are significantly affecting school pacing are the following:

• Children with psychological difficulties (based on the SDQ total difficulties score) are less likely to be at pace or advanced in school compared with children who have no psychological difficulties.

• Children who are stunted are less likely to be at pace or advanced in school compared with children who are not afflicted by stunting.

• Children residing in a rural town or village are less likely to be at pace or advanced in school than children living in urban areas.

Most of the predictors of school pacing have economic dimensions. Among the migrant-carer categories, the TH-mother carer households combine the advantages of having fathers as migrants and mothers as primary carers. With fathers likely to be employed in formal sector jobs overseas, they earn more, are more likely to send more remittances and are also less subject to difficult conditions than women migrants in less skilled occupations, particularly those employed as domestic workers. With mothers as carers, it is an arrangement which does not disrupt traditional childcare arrangements. Like most physical health indicators, stunting among young children is linked to economic disadvantages. As such, it is not surprising that it comes out as significantly related to school pacing. Having psychological difficulties is also found to affect school pacing. Thus, it is not only physical health which can be a drawback to school progression, but also the mental health of children. The other variables which reflect economic conditions are mother’s education, which contributes positively to school pacing, and residence in rural community or village, which has a negative effect on school pacing. Interestingly, household wealth or socio-economic status does not appear to have an independent effect on school pacing.

School Achievement

Turning to school achievement, the basic model establishes the positive impact of TH-mother carer households on school achievement – as shown in Table 6, children from these households are two times more likely to score highly in school achievement than children in non-migrant households (OR = 2.28, p<0.001). However, children in TH-father carer and TH-other carer households are not any more or less advantaged compared with children in non-migrant households when it comes to school achievement. The results of the full model (based on odds ratios), as presented in Table 6, show that TH-mother carer continues to have an impact on school achievement when other variables are added, although the odds are reduced (OR = 1.69, p<0.01). Nonetheless, the results suggest no significant difference between children in TH-father carer and TH-other carer households compared with children who co-reside with their parents. The other significant variables affecting school achievement are as follows:

• Children in wealthier (medium and high) households are two times more likely to have high school achievement compared with those in low-wealth households: those from medium-wealth households are twice as likely (OR = 1.93, p<0.01), and those from high-wealth households are close to three times more likely (OR = 2.79, p<0.001) to have high school achievement than their counterparts in low-wealth households. This underscores the capacity of better off families to allocate resources as needed to enhance their children’s academic performance.

• Children whose family functioning is good are more likely to have high achievement at school than children who do not perceive their family as functioning well (OR=1.48, p<.05). This suggests the importance of a supportive family environment – and children perceiving this to be so – in contributing to children’s academic performance.

TABLE 6. Ordered Logistic Regression of School Achievement.

	Basic Model			Full Model
Variables	Odds ratios	95%	CI	Odds ratios	95%	CI
Migrant-Carer Status
(NM)
TH-mother carer	2.28 *** (0.4561)	1.54	3.37 (0.3551)	1.69 **	1.12	2.55
TH-father carer	2.14 (0.7113)	1.11	4.10 (0.4186)	1.27	0.66	2.42
TH-other carer	1.3205 (0.3961)	0.73	2.38 (0.1956)	0.61	0.33	1.15
IC is a girl				1.37 (0.2388)	0.97	1.92
IC’s age
(9 years)
10 years				1.07 (0.2169)	0.72	1.59
11 years				1.00 (0.2314)	0.64	1.57
IC has psychological difficulties				0.80 0.1833	0.51	1.25
IC is stunted				0.86 (0.2288)	0.51	1.45
Household socio-economic status
(Low)
Medium household wealth				1.93 ** (0.4333)	1.24	2.99
High household wealth				2.79 *** (0.7922)	1.60	4.87
Father has high education				1.08 (0.2628)	0.67	1.74
Mother has high education				1.12 (0.2582)	0.72	1.76
IC’s number of siblings
(No sibling)
One sibling				0.83 (0.3221)	0.39	1.78
2 or more siblings				0.79 (0.3007)	0.37	1.67
Family functioning is good				1.48 * (0.2681)	1.04	2.11
Residence
(Urban)
Semi-urban				1.04 (0.2432)	0.66	1.64
Rural town/village				1.27 (0.2940)	0.81	2.00
Model chi-square	18.75 ***			57.35 ***
Pseudo R-squared	0.0196			0.0517
No. of observations	487			487

^*p < .05,

^**p < .01,

^***p < .001

Values in parentheses refer to the standard error.

Categories in parentheses refer to the omitted category.

The two dimensions of school outcomes appear to be influenced by different factors. As regards school pacing, the significant variables revolve around economic-related factors, highlighting the enabling impact of access to resources that allow children to be at pace or advanced in their schooling. Both the basic and full models indicate the migration advantage when it is fathers who migrate and mothers who assume the role of caring for the children. When mothers or both parents are the ones who migrate, children are not necessarily disadvantaged in school pacing relative to children in non-migrant households. When the focus shifts to school achievement, the migration advantage suggested by the basic model (i.e., TH-mother carer being associated with greater likelihood of higher school achievement) is sustained in the full model. As in school pacing, the combined positive effects offered by having a migrant father and caregiving provided by a mother also contribute to greater likelihood for children to attain high achievement at school. While physical health, as indicated by stunting, is not significant in school achievement, children reporting having good family functioning are around one and a half times more likely to have high school achievement than those whose family functions less well. The significance of medium and high household wealth suggests that economic factors are not only important in increasing access to education but they also figure in school achievement, most likely in terms of enabling families to muster financial resources that would contribute to enhanced learning, e.g., participation in extra-curricular activities, tutorials, or Internet access. On the whole, the cluster of variables pertaining to family resources – migrant-carer status, household wealth or socio-economic status and good family functioning – has an important to play in a child’s school achievement.

Discussion and Conclusion

Before discussing the findings in relation to the research questions, the usual caveat applies in interpreting the results of the study, particularly the limited size of the non-nationally representative sample. In addition, the delimitation of the sample to two-parent non-migrant and transnational households restricts the generalizability of the findings to other types of families. Although the results may not be representative of all households in the Philippines, nonetheless data from CHAMPSEA-Philippines identified factors that shape school outcomes in the study communities.

Concerning the first question explored in this article – how do children in transnational households compare with those in non-migrant households – the analysis uncovered findings that run contrary to popular perceptions. The concern that parental absence due to migration can negatively affect the school performance of children is not supported by the study. If parental migration affects school outcomes, it is associated with positive outcomes, or with outcomes which show that children in transnational households are not doing worse than children living with both parents. The best scenario is the TH-mother carer household where fathers work abroad and mothers stay home as carers; children in these households fare very well when it comes to school pacing and school achievement. This migrant-carer set-up combines the possibility for families and households to increase their economic resources and the continuation of the caregiving role of mothers. Although not considered as the ideal carers, the perceived disadvantage of children in father-carer or other-carer households is not supported by the study’s findings. Whereas past findings (e.g., Battistella and Conaco, 1998; ECMI/AOS-Manila et al., 2004; Asis, 2006b) found young children in father-carer households worse off than children in other households when it comes to school and some psychological indicators, this was not sustained in the CHAMPSEA study. It could be that the impact of having a migrant mother has changed over time. Investigating the factors that might have produced such change is beyond the scope of the present study and must await future research. It is also possible that the role of fathers as carers was not properly acknowledged in previous studies because of methodological issues – mothers are easier to access as research participants compared to fathers. The efforts of the CHAMPSEA Project to collect data from different types of migrant and non-migrant households provided the opportunity to involve fathers as research participants, to hear their voices, and to see their roles in a different light as well.

The factors that account for school outcomes and the specific impact of migration were the other questions explored by the study. Except for the TH-mother carer arrangement, which is positively associated with both school pacing and school achievement, the other factors that affect school pacing and school achievement can be broadly classified into economic and psychological resources. However, the specific variables that are at work vary. For school pacing, economic factors are salient, as suggested by the impact of the following variables: the negative impact of stunting and rural residence and the positive impact of mother’s high education. The negative impact of stunting on school pacing hints at the severe constraints of poor families in supporting children’s schooling; moreover, since stunting is an indicator of deprived circumstances in early childhood, it suggests the difficulty of overcoming disadvantages in school pacing faced by children in poor families. The article by Graham and Jordan (this volume) shows that children in (currently) low wealth households in the Philippine sample are three times more likely to be stunted than children in high wealth households. Although the finding is only marginally significant, nonetheless, it indicates that stunting is associated with the current, as well as the past, socio-economic status of the household. Also, children who have psychological difficulties are less likely to be at pace or advanced compared to children without this disadvantage. Broadly speaking, the same combination of economic and psychological resources significantly affect school achievement, although the specific indicators differ. Contributing to the greater likelihood of school achievement are medium and high household wealth and good family functioning. Unlike school pacing, medium and high household wealth have a more direct and independent effect on school achievement. This suggests that economically better off families are in a better position to enhance the children’s academic performance. Should children need tutoring, for example, this will not pose a problem for families with more economic resources. The disadvantaged position of children in poor households is a flashpoint for advocacy and action. However, while economic resources are important, they are not sufficient to promote positive school outcomes. Attending to the psychological health of young children and the children’s sense of positive family functioning also contributes to school outcomes. In all, the results affirm that it takes material and caregiving investments to ensure that young children are at pace with their schooling and are doing well in school.