The effects of early childhood education attendance on cognitive development evidence from vietnam

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES VIETNAM – THE NETHERLANDS PROJECT FOR M.A ON DEVELOPMENT ECONOMICS THE EFFECTS OF EARLY CHILDHOOD EDUCATION ATTENDANCE ON COGNITIVE

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

VIETNAM – THE NETHERLANDS PROJECT FOR M.A ON DEVELOPMENT ECONOMICS

THE EFFECTS OF EARLY CHILDHOOD

EDUCATION ATTENDANCE ON COGNITIVE DEVELOPMENT: EVIDENCE FROM VIETNAM

BY

TRIEU VAN VIET

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, DECEMBER 2015

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

VIETNAM – THE NETHERLANDS PROJECT FOR M.A ON DEVELOPMENT ECONOMICS

THE EFFECTS OF EARLY CHILDHOOD

EDUCATION ATTENDANCE ON COGNITIVE DEVELOPMENT: EVIDENCE FROM VIETNAM

A thesis submitted in partial fulfilment of the requirements for the degree of

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

ABSTRACT

Over the past several decades, numerous studies have documented the association between early childhood education attendance and children’s cognitive abilities This thesis aims to examine whether the positive relationship between preschool attendance and cognitive skills found in previous studies holds in Vietnam The thesis used the first two rounds of longitudinal data from the Young Lives study in Vietnam The sample is a “younger cohort” of approximately 2000 children who were born in 2001 to 2002 Ordinary least squares (OLS) and instrument variables (IV) estimation methods were applied to examine the link between preschool attendance and child cognition of 5-years-old (measured by the Peabody Picture Vocabulary Test (PPVT) and the Cognitive Development Assessment test (CDA)) In line with the literature, a positive and statistically significant relationship between preschool attendance and cognitive attainments was found for 5-years-old children in Vietnam The IV estimation produces a higher coefficient on preschool attendance than the OLS, which implies that a downward bias of OLS model In addition, children living in urban areas, children whose father and caregiver had higher education levels, ethnic Kinh children, and children

from wealthier families scored higher on cognitive tests Hence, government should

invest more in early childhood development, particularly for rural, poor and ethnic minority children to enhance their cognitive ability before entering primary education

Finally, I am very thankful to all the staff of Vietnam-The Netherlands Programme, who helped me at different stages of my thesis I would also like to thank all of my classmates

ABBREVIATIONS

ADB Asia Development Bank

CDA The quantitative of Cognitive Development Assessment test ECE Early Childhood Education

ECLS-K Early Childhood Longitudinal Study- Kindergarten

HAZ Height-for-age z-score

IV Instrument Variables

MOET Ministry of Education and Training

N Number of observations

PPVT Peabody Picture Vocabulary Test

OLS Ordinary Least Squares

OVERID Over-identification test

SD Standard Deviation

SES Socioeconomic Status

2SLS Two-stage Least Squares

UNESCO United Nations Educational, Scientific and Cultural Organization UNICEF United Nations International Children's Emergency Fund

YLs Young Lives

TABLE OF CONTENTS

CHAPTER I: INTRODUCTION 1

1.1 Introduction 1

1.2 Research objectives 2

1.3 Research question 2

1.4 Research scope and data 2

1.5 The structure of the thesis 3

CHAPTER II: LITERATURE REVIEW 4

2.1 Definitions 4

2.1.1 Early Childhood Education 4

2.1.2 Cognitive Development 5

2.2 The importance of investing in early childhood care and education 5

2.3 Empirical study review 7

2.3.1 The impact of pre-school attendance on child cognition 7

2.3.2 The impact of other factors on child cognition 10

2.3.2.1 The impact of socioeconomic status (SES) 10

2.3.2.2 The impact of number of siblings 11

2.3.2.3 The impact of child health and nutritional status 12

2.4 Previous research of early childhood education attendance in Vietnam 12

CHAPTER III: RESEARCH METHODOLOGY 14

3.1 An overview of pre-school enrolment in Viet Nam 14

3.2 Empirical strategy and econometric modeling 17

3.2.1 Empirical strategy 17

3.2.2 Econometric model 20

3.2.3 Description of key variables 23

3.3 Data source 25

CHAPTER IV: RESULTS AND DISCUSSION 27

4.1 Data descriptions 27

4.2 Regression results 33

4.2.1 Results of OLS regressions 33

4.2.1.1 The impact of preschool attendance on the CDA scores of the child 34 4.2.1.2 The impact of preschool attendance on the PPVT scores of the child35 4.2.2 Results of IV estimation with binary endogenous variable 37

4.2.2.1 The impact of preschool attendance on the CDA scores of the child 39 4.2.2.1 The impact of preschool attendance on the PPVT scores of the child41

CHAPTER V: CONCLUSION 44

5.1 Conclusion remarks 44

5.2 Policy implications 45

5.3 Limitations and recommendations for future research 46

REFERENCE 48

APPENDICES 55

Appendix 1: The construction of the wealth index 55

Appendix 2: Results of t-test on CDA score by preschool (mean-comparison tests), coded 1 for preschool attendees, 0 for non-preschool attendees 56

Appendix 3: Comparison of densities of level and natural logarithm of the CDA and PPVT scores 56

Appendix 4: Results of t test on PPVT score by ethnicity (mean-comparison tests), coded 1 for ethnic children, 0 for kinh or chinese children 56

Appendix 5: Results of t-test on CDA score by region (mean-comparison tests) 57 Appendix 6: Total bad events between CDA and PPVT model 58

Appendix 7: Results of OLS regression on CDA scores 58

Appendix 8: Results of OLS regression on PPVT score 59

Appendix 9: Results of 2SLS regression with binary endogenous variable (CDA score) 60

Appendix 10: Results of 2SLS regression with binary endogenous variable (PPVT score) 61

Appendix 11: Test for weak instruments 62

LIST OF FIGURES

Figure 1: Rates of return to human capital investment 6

Figure 2: Enrolment in pre-primary education, both sexes (number) 15

Figure 3: Gross enrolment ratio (%) in Pre-primary education 16

Figure 4: Conceptual framework for assessing early cognitive development 18

Figure 5: The distributions of CDA and PPVT scores 29

Figure 6: Kernel density estimates for PPVT and CDA scores at Round 2, by pre-school attendance 30

LIST OF TABLES Table 1: Enrolment in pre-primary education in Vietnam from 1999 to 2013 15

Table 2: Variables definitions 25

Table 3: Description of variables 27

Table 4: Children’s cognitive development by Pre-school attendance 28

Table 5: Children’s cognitive development by ethnicity 31

Table 6: Children’s cognitive development by region 31

Table 7: Results of OLS regressions on cognitive achievement 33

Table 8: Correlation between pre-school enrollment, number of siblings, 37

Table 9: Results of 2SLS regression with binary endogenous variable 39 Table 10: The OLS and IV estimates of preschool attendance on the CDA scores 40 Table 11: The OLS and IV estimates of preschool attendance on the PPVT scores 42

CHAPTER I: INTRODUCTION

1.1 Introduction

Early childhood is defined as “the period between birth and eight years of life” (UNESCO & UNICEF, 2012) Research in the field of cognition, psychology, and neuroscience has established that learning process is easier in early childhood than any other periods in the human life (Bransford, 1979; Shore, 1997) In the first three years of life, children’s brain develop rapidly, about two times more active than an adult (Shore, 1997) Additionally, economists argued that skill formation is

a life cycle and cumulative process (Carneiro & Heckman, 2003; Cunha et al., 2006; Heckman & Masterov, 2007) The skill obtained at one stage of the life cycle will enhance the skill obtainment at next stage of the life cycle, skill begets skill and that learning begets more learning Early childhood education (ECE) play a large role in building basic skills early in life such as literacy and numeracy, and these skills are critical for long-term skill development such as problem-solving abilities and capacities (European Commission, 2011) From an economic perspective, investments in early childhood development has been shown to improve cognitive and physical development, reduce crime rates, alleviate poverty, and increase economic growth (Barnett & Nores, 2012)

In Vietnam, the Education Law 2005 defines Early Childhood Education covers "the nurturing, caring, and educating children from three months to six years

of age" (The National Assembly of Vietnam, 2005) Early childhood education system in Vietnam consists of daycare centres, kindergarten, and pre-school or pre-primary education The children from three to five years old participating pre-school has increased dramatically in recent years, from 39 per cent in 1996 to 60 per cent in 2003(ADB, 2005) The number of children in ECE was 2,480,272 in schooling year 2000-2001, and this figure has increased to 2,754,094 in year 2004-

2005, an increased by 2.3 per cent (UNESCO, 2006) However, low quality education has raised the questions about what is learned in pre-school education and

how this relates to children’s cognitive skills development at the ages of 5 years in Vietnam

The effects of early childhood education on children’s cognitive outcomes have been widely studied from a variety of aspects, including the variation in methods, types of interventions, national contexts, and outcomes A number of authors have looked into the benefits of early intervention childcare and compensatory education programs (Barnett, 1995; Blau & Currie, 2006; Carneiro & Heckman, 2003; Love et al., 2005) Others have investigated the relationship between maternal employment and childcare during preschool years on cognitive development (see for literature reviews Blau & Grossberg, 1990; Waldfogel et al., 2002) There is substantial empirical evidence that ECE programs improve children’ cognitive outcomes, both in the short and in the long run However, few empirical studies have been investigated about the effect of early childhood education (pre-school) attendance on cognitive achievement in Vietnam In addition, more understanding about the relationship between schooling and cognitive development

is critical for policymakers to conduct early intervention to improve the children development This thesis will apply ordinary least squares (OLS) and instrument variables (IV) approach with data from Young Lives project in Vietnam for examining the relationship between pre-school attendance and children’s cognitive outcomes The results of analysis contribute to the existing educational literature, and the suggestions for future research are also provided

- How does preschool attendance affect the child cognitive achievement?

1.4 Research scope and data

The thesis used the first two rounds of longitudinal data from the Young Lives study in Vietnam The sample is a “younger cohort” of 2000 children who were born in 2000/1 Round 1 were surveyed in 2002, when these children were aged between 6 to 18 months old Round 2 were surveyed in 2006, when they were aged between 4 to 6 years old

1.5 The structure of the thesis

This thesis is organized into five chapters In the second chapter define the term “Early childhood education” and “Cognitive Development” and then presents empirical studies about the links between pre-school attendance and children’s cognitive development

Chapter 3: This chapter is divided into two sections In the first section describes an overview pre-school enrolment in Vietnam, and then the second section presents the empirical methodology used for econometric analysis

Chapter 4: The following section briefly describes the data used in the estimation Estimation results from OLS and IV regression on child cognitive skills are also presented

Chapter 5: The final chapter draws conclusions from analyzing the data, provides some policy implications to improve Vietnam’s pre-school education system, and guidelines for future research on this topic with Vietnam data

CHAPTER II: LITERATURE REVIEW

This chapter firstly defines the term “Early childhood education”, “Cognitive Development” The second section will explain the importance for investing in early childhood education and then presents prior studies about the links between preschool attendance and child cognition

2.1 Definitions

2.1.1 Early Childhood Education

Early childhood refers to the period from the birth to age eight years of life

It is important to note that the term early childhood education (ECE) and preschool

or pre-primary education are used interchangeably in this thesis Early childhood education is defined as all supports necessary for the growth and development of children during the early childhood period (UNESCO & UNICEF, 2012) The ECE programs typically provide direct services to the child such as education, health care and nutrition, intellectual stimulation, social and emotional care

According to Vietnam’s Education Law 2005, early childhood education is a part of the national educational system which covers “the nurturing, caring, and educating children from three months to six years of age” (The National Assembly

of Vietnam, 2005) The Education Law also stated that “the objectives of early childhood education are to help children develop physically, emotionally, intellectually and aesthetically, in order to shape the initial elements of personality

as well as to prepare children for the first grade” (The National Assembly of Vietnam, 2005) In Vietnam, ECE is basically classified into three kinds by the age

of children: (1) childcare and nursery for children from the age of three months to three years, (2) kindergarten is intended for children from the age of three years to five years, (3) pre-school or pre-primary education include the functions of the childcare, nursery and the kindergarten for children from three months up to five years of age (ADB, 2005)

2.1.2 Cognitive Development

A large body of research on the child development has distinguished between cognitive and non-cognitive skills Cognitive skills refers to the basic skills includes the use of critical thinking, literacy and numerical ability, and the ability to solve complicated problems It was measured by the standardized tests in forms of reading, writing, mathematics, and the Intelligence Quotient test (Brunello & Schlotter, 2010) Non-cognitive skills belong to personality traits and behaviors such as self-discipline, extraversion, motivation to achieve, and emotional stability (Adams et al., 2012; Brunello & Schlotter, 2010) Both cognitive and non-cognitive skills are significant in demonstrating academic achievement It is shaped in early the life cycle However, cognitive ability is mostly produced by early investments, and the economic returns from investment at early years of the life cycle are much larger than investment that come later in life (Cunha et al., 2006) That is why this study concentrates on child development in terms of cognitive outcomes

2.2 The importance of investing in early childhood care and education

Early childhood education (ECE) play a large role in building basic skills early in life such as literacy and numeracy, and these skills are critical for long-term skill development such as problem-solving abilities and capacities (European Commission, 2011) Studies in the field of cognition, psychology, and neuroscience indicated that learning process is easier in early childhood than any other periods in the human life (Bransford, 1979; Shore, 1997) In the first three years of life, children’s brain develop rapidly, about two times more active than an adult (Shore, 1997) Additionally, economists argued that skill formation is a life cycle and cumulative process (Carneiro & Heckman, 2003; Cunha et al., 2006; Heckman & Masterov, 2007) The skill obtained at one stage of the life cycle will enhance the skill obtainment at next stage of the life cycle, skill begets skill and that learning begets more learning

In recent years, a large body of literature has established the importance of investing in early childhood education From an economic perspective, the primary

rationale for investments in ECE includes internal and external benefits for children, family and society (Barnett & Nores, 2012) The potential benefits for the child involves improved health, increased cognitive development and educational attainment An example of an external benefits (social benefits) include reduced crime rate and violence, lower medical costs, promoted economic growth due to productivity enhancements In the economic literature, the rate of economic returns from early investment are highest (Figure 1), and typically for those from the vulnerable groups (Carneiro & Heckman, 2003; Cunha et al., 2006).In addition, the findings indicated that early childhood education tends to produce larger benefits for those children from socio-economically disadvantaged backgrounds (Nores & Barnett, 2010) Cost-benefit analyses have estimated the economic return to society

of the High/Scope Perry Preschool Program for disadvantaged African-American youth was $16.14 dollars per dollar invested (Belfield et al , 2006)

Figure 1: Rates of return to human capital investment

(Source: Carneiro & Heckman (2003))

In general, earlier investments that directly provide early childhood education to children are more productive and effective for improving cognition and school achievement than investments that come later in life

2.3 Empirical study review

It has been widely acknowledged that children’s cognitive development is multi-dimensional and influenced by many factors such as education, health, socioeconomic status, and other factors However, the findings indicate that “pre-school has a significant and positive impact on progress over and above important influences such as family socioeconomic status” (Sammons et al., 2002)

2.3.1 The impact of pre-school attendance on child cognition

The evidence of the effects of pre-school education participation on children’s cognitive achievement has been widely studied from a variety of aspects, including the variation in methods, types of interventions, national contexts, and outcomes Numerous researchers have focused on the benefits of early intervention childcare and compensatory education programs (Barnett, 1995; Blau & Currie, 2006; Carneiro & Heckman, 2003; Love et al., 2005) Others have investigated the relationship between maternal employment and childcare during preschool years on cognitive development (see for literature reviews Blau & Grossberg, 1990; Waldfogel et al., 2002) These studies applied ordinary least squares, instrumental variables, difference-in-difference, and propensity score matching technique Most studies have documented that early childhood education programs improved children’s cognitive outcomes, both in the short and long run Basically, these studies are classified into two main types, including experimental and quasi-experimental or observational studies, some of observational studies based on longitudinal data (Lefebvre et al., 2006)

Experimental studies are generally conducted with small-scale samples Among the experimental studies, the Perry Preschool and Abecedarian Project are the most widely reported The major conclusion of these study is that preschool programs for disadvantaged children contributed to their cognitive development both the short and long-term (Campbell et al., 2012; Schweinhart et al., 2005)

Quasi-experimental and observational studies analyze the effect of ECE from larger-scale public programs and control for other important background

characteristics that could influence children’s cognitive development Evidence of the observational studies was completed mostly in the United States Several researchers employ data from the American Early Childhood Longitudinal Study-Kindergarten (ECLS-K) to analyze the influence of pre-school education on cognitive and social development of children For instance, Loeb et al (2007) investigated the link between center-based care participation and children’s social-behavioral and cognitive skills by applying ordinary least squares, and instrumental variables method These authors found that the center program was associated with

an increase in reading and math scores However, it had a negative effect on behavioral outcomes The children enter centers between the ages of two and three have greater academic benefits than those who start at earlier and after that time The result for behavior outcomes was negative: “the negative behavioral effects are greater the earlier a child enters” (Loeb et al., 2007) The other study was carried out by Magnuson et al (2007) indicates that children from poor households exhibited the greater gains in reading and mathematics skills from early childhood interventions However, the prekindergarten program may also lead to reduce self-control and increase levels of behavior problems Overall, the effects size was about -0.14 for self-control and 0.23 for aggressive behavior

socio-Outside of the United States, the findings of other studies from large scale programs are broadly consistent with the results in the United States In United Kingdom, an analysis for children (2 years old or higher) participating pre-school also found significantly higher cognitive achievement in terms of language, pre-reading, and early number than children who do not attend a pre-school (Sammons

et al., 2002) An important finding is that the longer pre-school education attendance is associated with the higher cognitive gains (Sammons et al., 2002) For the gender factor, interestingly, girls showed significantly higher score than boys in pre-reading, non-verbal reasoning and early number concepts over the pre-school period Furthermore, children from larger families with equal or more than 3

siblings has displayed negative effects on cognitive development (Sammons et al., 2002)

Similarly, in Argentina, the study of Berlinski et al (2009) reported that the pre-primary enrollment has a positive effect on both cognitive skills and non-cognitive skills For instance, one year of pre-school increase 4.69 points for Mathematics test scores and 4.76 points for Spanish test scores In addition, the results of the regressions after control for municipality fixed effects and year by province effects illustrated that the program has positive influence on student’s self-control in terms of discipline, class participation

In another major study in Ethiopia, Woldehanna et al., (2012) employed ordinary least square, fixed effects, instrumental variable estimation and propensity score matching techniques to analyze the impact of early childhood interventions on cognitive achievement outcomes The results indicated that the pre-school attendance was also found to be positively related to educational attainment for 5-years-old children Additionally, the econometric results of the instrumental variable approach, ordinary least squares, and fixed effects estimation displayed larger standard errors than the propensity score estimation

Nevertheless, the negative effects of ECE program were found in some studies from Canada Lefebvre et al (2006) utilized both a double difference and a triple difference estimation technique to investigate the potential impact of the subsidized daycare policy of the province of Québec in Canada on children’s preschool achievement The econometric results provided an evidence that the subsidized daycare policy is associated with poorer cognitive development (PPVT score) for children of age 5, but it does not influence on children of age 4 Baker et

al (2005) also found strong effects on maternal labor supply, but negative effects of new childcare program for child development

In summary, studies of the effects of early childhood education attendance on children’s cognitive outcomes are mixed results However, most studies have provided evidence that children from different context and countries receive

substantial cognitive benefits from early childhood interventions, both in the short and long run (Nores & Barnett, 2010)

2.3.2 The impact of other factors on child cognition

2.3.2.1 The impact of socioeconomic status (SES)

Beside pre-school education attendance factor, numerous cross-sectional and longitudinal studies have emphasized the relation between household’s socioeconomic status and cognitive attainment of children in early childhood (Grantham-McGregor et al., 2007) There are several ways to measure socioeconomic status (SES), but most comprise some characteristic of maternal education, paternal education, parental income, and parental occupation (Bradley & Corwyn, 2002) In recent years, several investigators have used home resources as

an indicator of household SES background (Entwislea & Astone, 1994; McLoyd, 1998; Sirin, 2005) This indicator includes computers, reading and learning materials, as well as room study However, home resources are not used as commonly as income, education, and occupation indicators (Sirin, 2005)

Over the past several decades findings on the associations between SES and children's cognitive abilities has accumulated Numerous studies have indicated that children belong to families with a high SES such as family income and parental education are associated with higher levels of school achievement than children from families with a low SES (Guo & Harris, 2000; Taylor et al, 2004; Waldfogel

et al, 2002; Duncan et al 1994) In developing countries, the links between socioeconomic status and student’s academic outcomes are also well documented in several studies For instance, a positive relationships have been found between household wealth, mother’s and father’s schooling at age one on children’s cognitive skills at age five in Ethiopia, Peru, India, and Vietnam (Benjamin T Crookston et al, 2014) Among the traditional measures of SES indicators (income, parental education, occupation), White (1982) used the first meta-analytic study and argued that household income accounted for the biggest amount of variance in academic achievement

A number of mechanisms have been suggested to explain the less cognitive development of low SES children, with most including differences in access to learning materials and resources Indeed, children from poor family (low SES) tend

to have less access to different learning materials and cultural resources during childhood period, and therefore limit their cognitive development (Bradley & Corwyn, 2002; Brooks-Gunn et al, 1995) For example, poor children are less likely

to have regular school attendance, take part in cultural events, or visit a museum As

a result, these factors are associated with poorer cognitive development (Bradley & Corwyn, 2002; Brooks-Gunn et al., 1995; Ready, 2010) Moreover, socioeconomically disadvantaged children are considerably more likely to have less opportunity to learn, resulting in these children gain fewer academic skills than wealthy children (Alexander, Entwisle, & Olson, 2001)

2.3.2.2 The impact of number of siblings

Sibling status has also been studied as an important determinant related to children’s educational outcomes Several investigators have proposed in many instances that children with many siblings obtain lower score of cognitive ability tests than children with few siblings (Brody, 2004; Downey, 2001) The theory of

“resource dilution” offers a basic explanation for the inverse relation between the

number of siblings and cognitive development (Blake, 1981) This theory argued that parental resource which include economic and interactive for children is limited It means additional siblings may reduce the availability of parental time, and financial resources to existing children (Blake, 1981) On the other hand, children with three or more sibling are less likely to be enrolled in pre-school than children with one sibling which affect children cognitive development (Chiswick & DebBurman, 2006) Nonetheless, there are some researchers supporting the hypothesis that the effect of sibship size on the quality of the children is spurious (Guo & VanWey, 1999; Rodgers, 2001)

2.3.2.3 The impact of child health and nutritional status

Nutritional status is a reflection of overall health and it is a critical factor in determining subsequent academic achievement of children during pre-school age In general, there are three nutritional status indicators, including weight-for-age, height-for-age, and weight-for-height These indicators are described by standard deviation units (z-scores) from the median of the reference population (WHO, 2007) Among of these indicators, height-for-age z-score (HAZ) is a measure of chronic malnutrition that relate to mental function (Boo, 2009) In developing countries, HAZ indicator is typically used as a measurement for children’s nutritional status and it is widely used as an input of cognitive achievement production function of children (Aurino & Burchi, 2014; Boo, 2009)

Most cross-sectional and longitudinal studies have found significant the relation between height-for-age and children’s school achievement in less developed as well as developed economies (Grantham-McGregor et al, 1999) These studies indicate that HAZ is associated with higher levels of cognitive skills

in childhood in the India, Philippines, Jamaica, and Kenya (Boo, 2009; Glewwe et

al, 2001; Grantham-McGregor et al, 1997; Sigman et al, 1989) For instance, Boo (2009) used novel panel data from two groups of children in India and found that raising HAZ at the age of one by one standard deviation will increase cognitive test scores by 24 percent of a standard deviation at age 5 However, some researchers have found that children’s cognitive development was not significantly related to height-for-age (Bogin & MacVean, 1983; Church & Katigbak, 1991)

In summary, children’s cognitive skill is affected by a variety of factors, including pre-school attendance, socioeconomic status, child characteristics, child health and nutritional status

2.4 Previous research of early childhood education attendance in Vietnam

Studies have found the positive effects of pre-school participation on student’s cognitive outcomes in Vietnam (Rolleston & James, 2012; Watanabe et al., 2005) Children who attended in pre-school program had improved language

and literacy skills, and gains in the understanding of basic number concepts compared to students who did not attend to these classes (UNESCO, 2004)

For instance, the study was carried by Watanabe et al (2005) using a sample

of 313 children aged 6.5– 8.5 year in 2 communes of Thanh Hoa province to determine whether an early childhood development project improve the cognitive development outcomes through a nutrition intervention The authors have used generalized linear model to analyze effects of interventions on Raven’s tests scores.

The evidence suggested that the program interventions for preschool-age children raise the potential for cognitive skills, particularly for malnourished children

Other study was carried by Rolleston & James (2012) employ data from Young Lives, to examine the development of cognitive ability, basic literacy and numeracy during pre-school education participation in four less developed countries: Ethiopia, India, Peru and Vietnam The study documented evidence that

“early advantage at the household is a key determinant of skill acquisition in all four countries” and early enrollment in school is associated with higher child’s cognitive score, basic literacy and numeracy in disadvantaged contexts However, relativelyfew empirical studies have been investigated about the effect of early childhood education attendance on cognitive outcomes in Vietnam In addition, more understanding about the relationship between schooling and cognitive development

is critical for policymakers to conduct early intervention to improve the children development This study employs data from Young Lives in Vietnam to examine whether preschool attendance help to improve cognitive development of 2000 index children who were born in 2000-2001 Two estimation approaches are utilized to explore the influence of the pre-school attendance in Viet Nam: ordinary least squares and instrument variables method

In conclusion, a large body of research suggested that attending preschool had considerable positive effects on cognitive development both in the short and over the long-term Socioeconomic and nutritional status and child’s characteristics are also associated with cognitive outcomes of children

CHAPTER III: RESEARCH METHODOLOGY

This chapter is divided into two sections The first section describes an overview of pre-school enrollment in Viet Nam In the second section presents a conceptual framework for modeling the relationship between preschool attendance and cognitive skills and then describes the methodology used to estimate this association

3.1 An overview of pre-school enrollment in Vietnam

Since the Vietnam economic reform program “Doi Moi” in 1986, the Ministry of Education and Training (MOET) is responsible for the management, supervision, developing programmes, and inspection of the application of early childhood education related regulations In general, early childhood education is classified into three main kinds by the age of children, including day-care centres and nursery, kindergarten school, and pre-school or pre-primary school With respect to objectives for ECE, the Education Law states that ECE helps children promotes their physical, emotional, aesthetic and intellectual development, which prepare for children entering Grade 1 of primary education (The National Assembly

of Vietnam, 2005) There are four main types of providers for early childhood education: state, semi-state, community and private provider Early childhood education is not compulsory in Vietnam Vietnamese children are relatively strong pre-primary enrollment when compare to other countries with similar economic development

With regard to pre-primary education, in recent years, the total number of children attended pre-primary education has increased from 2,124,142 in school year 1999-2000 to 3,551,082 in school year 2012-2013 (see Table 1) Data from the Ministry of Education and Training report that the percentage of early childhood teachers meeting qualification standards was 93.22% of teachers in nurseries and 97.57% of teachers in kindergartens in school year 2012-2013 Most importantly,

the proportion of nursery teachers meet national standards has increased dramatically from 59.73% in 2005 to 93.22% in 2013

Table 1: Enrollment in pre-primary education in Vietnam from 1999 to 2013

Year Total

number Female Male Minority Public Non Public

Gross enrollment ratio(%)

1999-2000

2,124,142

1,023,646

1,100,496

238,330

1,039,162

1,084,980 40.13

2000-2001

2,113,594

1,039,633

1,073,961

240,757

1,064,904

1,134,124 41.77

2001-2002

2,120,345

1,033,753

1,086,592

114,341

851,055

1,269,290 45.15

2002-2003

2,143,881

1,023,447

1,120,434

301,373

866,932

1,276,949 49.96

2003-2004

2,175,053

1,054,398

1,120,655

305,500

906,290

1,268,763 55.33

2004-2005

2,332,658

1,092,598

1,240,060

337,640

1,044,403

1,288,255 60.59

2005-2006

2,511,239

1,092,780

1,418,459

357,141

1,138,001

1,373,238 66.13

2006-2007

2,617,167

1,149,677

1,467,490

383,636

1,212,083

1,405,084 67.46

2007-2008

2,687,037

1,308,022

1,379,015

398,572

1,270,618

1,416,419 66.50

2008-2009

2,810,625

1,341,342

1,469,283

417,608

1,457,940

1,352,685 66.65

2009-2010

2,901,633

1,374,341

1,527,292

452,539

1,609,634

1,291,999 68.67

2010-2011

3,070,794

1,420,183

1,650,611

489,968

2,062,500

1,008,294 72.18

2011-2012

3,320,328

1,549,499

1,770,829

545,037

2,628,513

691,815 77.18

2012-2013

3,551,082

1,627,390

1,923,692

594,603

3,047,328

503,754 -

Source: the Ministry of Education and Training (MOET) and UNESCO

Figure 2: Enrollment in pre-primary education, both sexes (number)

Source: the Ministry of Education and Training (MOET) and UNESCO

500.000

Figure 3 below shows that the gross enrolment ratio in pre-primary education has increased rapidly in recent years, which was 40.13% in school year 1999-2000, had reached 77.18% by 2011-2012 This growth was accomplished by increasing the numbers of school and teachers For instance, the number of pre-primary school

at the end of school year 2012-2013 increased by 5.851 to 10.146, compared to the 1999-2000 school year

Figure 3: Gross enrollment ratio (%) in Pre-primary education

Source: the Ministry of Education and Training (MOET) and UNESCO

Despite the increasing trend of the enrolment rate at pre-school, there are still several challenges in terms of the quality and access to ECE programs in Vietnam (Pham & Nicola, n.d.) Firstly, the number of preschool in the year 2011 was 13.172 with 138.843 classrooms, but only 48% of classrooms meet standards for children and 25% of the classrooms have acceptable equipment (Minh Nguyen, 2012) There was a shortage of over 26.886 classrooms for early childhood education in the 2011-2012 school year, while the number of children at pre-school age is increasing every year As a result, many children in remote areas cannot go to school due to the quantity of kindergarten does not meet the growing needs of these areas (Minh Nguyen, 2012) In many schools, the class size does not meet the standards regulation On the other hand, the lack of investment for developing

teaching methods and learning environment has affected children’s cognitive development (Pham & Nicola, n.d.) Secondly, the poor and ethnic minority children from the rural areas face challenging barriers to accession to basic education (UNESCO, 2006) Rural children with access to early childhood education were about 35 per cent in 2004 Research from Young Lives in Viet Nam found that after controlling for household resources, the ethnic minority children was significantly less likely to access to education (Le et al., 2008) The reason is that parents cannot afford the cost for their children participating in pre-school, language barriers, or distance to go to school

Overall, Vietnam’s early childhood education system has got many achievements over the last decade The gross enrolment ratio has increased dramatically, reached 77.18% by the end of school year 2011-2012 Five year-old preschool children receive public assistance under the provisions of Decision No

239 of the government in period 2010-2015 Despite the rapid growth in the economy and increasing the trend of enrolment rate at pre-school, there is still a wide difference in access to education for the poor and ethnic minority children Low pre-school enrolment rates and quality of teaching remain a concern in disadvantage areas

3.2 Empirical strategy and econometric modeling

3.2.1 Empirical strategy

Figure 4 presents the analytical framework which illustrates the hypothesis that early childhood education attendance affects child cognition at age 5 in Vietnam Other determinants of cognitive outcomes represented in the framework, which have been identified in the chapter 2

The dependent variables: two measure of children’s cognitive

development

Measurements of child cognitive ability employed in this analysis are based

on Peabody Picture Vocabulary Test (PPVT) and Cognitive Development Assessment (CDA) test The PPVT is a kind of test which measures the receptive

vocabulary of children Several empirical studies used PPVT test score as a measure for cognitive achievement (Baydar & Brooks-Gunn, 1991; Blau & Grossberg, 1990; Le, 2009; McCulloch & Joshi, 2002; Woldehanna Tassew, 2011) The CDA refers to the test which reflects children’s quantitative ability at preschool age This test was used to measure the cognitive development of children in numerous studies in developing countries, such as Kenya, Peru, Vietnam and India (Cueto et al, 2009) Therefore, children’ score on the PPVT and CDA test are considered as a proxy measure of cognitive outcome of preschool children in the analysis

proxy-Figure 4: Conceptual framework for assessing early cognitive development

- Child age

- Child sex

- Ethnic group

Child cognition

at age 5

- Number of siblings

- Shocks (Total bad events)

and parental education are associated with higher levels of school achievement than children from families with a low SES (Guo & Harris, 2000; Taylor et al, 2004; Waldfogel et al, 2002; Duncan et al 1994) However, the evidence suggests “pre-school has a significant and positive impact on progress over and above important influences such as family socioeconomic status” (Sammons et al., 2002) In developing countries, a positive relationships have been found between wealth index, mother’s and father’s schooling on child cognition in Ethiopia, Peru, India, and Vietnam (Benjamin T Crookston et al, 2014) In line with previous literature, this study use wealth index, father’s and caregiver’s education as exogenous variables

In Vietnam, each region has different cultures, infrastructure conditions, and economic development which have been effect on human capital development In term of the early child development index score, Central Highlands region has the lowest score with 68.2 per cent, following by Mekong River Delta with 79.8 per cent, and then the Northern Midland and Mountain areas with 81.8 per cent (GSO, 2011) Moreover, community a child is born into is also a contributing factor to child cognitive development (Le et al., 2008) Ideally, the study also controls for family characteristics, including urban region, rural in North-East (Lao Cai), South Central Coast (Phu Yen), and Mekong River Delta region (Ben Tre)

Nutritional status of early childhood is also an important factor of child cognitive outcomes Glewwe et al (2001) argued that better early childhood nutrition display increasingly positive children’s academic achievement In developing countries, height-for-age z-score indicator is typically used as a measurement for nutritional status, and it is widely used as an input of cognitive achievement production function (Aurino & Burchi, 2014; Boo, 2009) In line with previous literature, this thesis use long-term health problem and height-for-age z-score indicator as measurements for nutritional status of five years old children

In addition, minority status is an important determinant in cognitive achievement in Viet Nam Recently, Glewwe et al (2012) have looked at the gap in

learning between Kinh and ethnic minority children in Vietnam They used the first two rounds of Young Lives data and reported lower on mathematics and reading tests for ethnic minority children In addition, recent surveys carried out by Vietnam’s GSO in 2010-2011 indicated that ethnic minority children age 36-59 months lagged behind Kinh or Hoa children in terms of literacy-numeracy and learning (GSO, 2011) Therefore, child sex, child age, and minority status are also considered as exogenous variables in the study

In general, the model includes pre-school attendance and other factors which expected to influence children’s cognitive development: height-for-age z-score, wealth index, age, gender, minority status, long-term health problem indicator, and community characteristics

3.2.2 Econometric model

This thesis will apply Ordinary Least Squares (OLS) and Instrumental Variables (IV) estimation to explore the extent to which preschool attendance affects cognitive achievement of children in pre-school age in Vietnam The study estimate several different models for each dependent variable

Hypothesis: There is relationship between pre-school attendance and

children’s cognitive outcomes

The study starts by specifying cognitive achievement to be a function of the children’s preschool participation and other factors To estimate the econometric model, this study refers to a general specification presented by Woldehanna & Gebremedhin (2012), the basic regression equation is:

Equation (1): ln (COij) = β +β1P R Ei j + β2Xi j + Ui j

Uij = Qij+ εij

Where (Qij) is the effect of unobserved variables and εij is white noise COijdenotes child i’s cognitive development outcomes, living in community j, measured

by PPVT and CDA score; PREij is an indicator of preschool attendance at age three

or higher; Xij is a set of household and child characteristics: height-for-age z-score

in round 1, father’s and caregiver’s education, wealth index in round 1, months old

of the child, gender (dummy for male), community characteristics; Uij is measurement error and unobserved factors

The dependent variables have a skewed distribution It suggests a linear model can provide “poor predictions because it restricts the effects of regressors to

be addictive” (Cameron & Trivedi, 2010) The solution uses in this study is to transform the dependent variable (PPVT and CDA scores) by taking the natural logarithm The purpose of taking the natural logarithm for dependent variables is to obtain residuals that are more normally distributed and to improve the model fit

Endogeneity of preschool attendance

The equation (1) is likely suffers from the problem of endogeneity due to the omitted variables Traditional OLS estimates are biased if there are any unobserved factors which affects cognitive outcomes through preschool attendance For instance, the number of siblings in households are observed for preschool attendance (Chiswick & DebBurman, 2006) but it does not include into the model Furthermore, participation in the preschool is probably associated with socioeconomic and environmental shocks, such as a natural disaster or victim of crime, which are also likely to negatively impact upon child cognitive achievement

As a result, it leads to underestimation of the effect of preschool attendance, and the result can be biased due to unobserved factors Therefore, several studies have used instrument variables (IV) approach to examine children’s cognitive abilities-effects

of preschool attendance The instrument must satisfy two conditions to be valid The first is that the instrument must be strongly correlated with pre-school attendance but not be a direct cause of the child cognition The second is that the instrument set is uncorrelated with unobserved error term For example, the study

by Magnuson et al (2007) used the level of state spending on prekindergarten programs as an instrument for prekindergarten participation in Ecuador The recent work by Woldehanna & Gebremedhin (2012), used community dummies and

Shocks as instruments for preschool attendance in Ethiopia Evidence from Young

Lives research team in Vietnam reported that number of siblings is significantly related to preschool enrolment (Le et al., 2008)

Ideally, this thesis also presents results for the IV approach, where number of siblings and total bad events (socioeconomic and environmental shocks) are used as

instrument variables for the enrolment of children in pre-school in Vietnam Total bad events are constructed from a set of eight dummy variables, including a natural

disaster (1), decrease or change in food availability (2), livestock died (3), crops failed (4), Job loss/source of income/family enterprises (5), severe illness or injury (6), victim of crime (7), divorced or separated (8) These events are considered to decrease the economic welfare of the household and expected to have an influence

on children pre-school attendance A dummy variable equal to one if any of big events or changes happened to household since the mother became pregnant with the index child Total bad events are calculated as the sum of eight shocks dummies

The IV estimation is accomplished by using two-stage least squares (2SLS) approach with a probit first stage (Wooldridge, 2010) In the first stage pre-school attendance (endogenous dummy variable) is regressed on number of siblings, total bad events and other exogenous to get predicted value The second stage use predicted value of pre-school attendance from the first regression as the explanatory variable in the structural equation, and then estimate using OLS The procedure is

carried by using the “treatreg” command in Stata This command estimates two

regressions simultaneously (probit regression in first stage and OLS in the second stage) The first stage equation for this analysis is:

Equation (2): PREij= β1SIBLINGij + β2 BADEVTij + β3Xij + α + εij

Where SIBLINGij denotes the number of siblings in a family (excluding the index children); BADEVTij is the total number of bad events Xij is a set of household and child characteristics The model assumes that fathers’ and caregivers’ education does not change across Rounds of the survey

3.2.3 Description of key variables

The Peabody Picture Vocabulary Test (PPVT) was firstly designed by Dunn and Dunn in 1959 The PPVT is a test of receptive vocabulary for 2.5 years old children to adulthood and it has been widely used as a universal measure of cognitive ability The PPVT consists of 204 questions grouped into 17 sets with 12 items each and are ranked by the difficulty of the questions (Dunn & Dunn, 1997) The task of the children is to identify the picture which best represents the meaning

of word reads out by the interviewer The PPVT score is calculated as the sum of the correct answers In Vietnam, the PPVT test was employed to evaluate the cognitive skills for children 5 years old in Round 2 of the Young Lives project

The Cognitive Development Assessment (CDA) was designed to evaluate children’s quantitative ability at preschool age Generally, the CDA has three subtests, including the time, spatial relations, and quantity However, only the Quantity subtest was applied in the Young Lives study The CDA quantity subtest contains 15 items Each item has three or four images and the task for the children is

to choose a picture that best reflected the question posed by the interviewer For instance, “Look at the bowls of eggs Point to the bowl that has the most eggs”.The CDA score is calculated as the sum of all the items (1 for correct, 0 for incorrect) in the test for the children The CDA test was used to evaluate the children’s cognitive ability in developing countries, such as Kenya, Peru, Vietnam and India (Cueto et

al, 2009) Similarly, the CDA test was used to measure cognitive development for 5 year old children in Round 2 of the Young Lives survey in Vietnam

Pre-school attendance refers to an indicator for whether the children attend pre-school at age three or higher A dummy variable takes value of 1 if children participated in pre-school at the second round survey, otherwise is 0 Minority status is measured by a dummy variable It has value of 1 if children belong to an ethnic minority

Nutritional status is measured by height-for-age z-score (HAZ) at Round 1 of data collection when children were aged between 6 and 18 months Height-for-age

is expressed in Z-scores The term Z-score refers to “the deviation of an individual’s value from the median value of a reference population, divided by the standard deviation of the reference population” (WHO, 1997) Z-Score allows for the comparison to be made across age groups and indicators Young Lives employed the WHO standards reference to calculate the height-for-age z-score HAZ is defined as the number of standard deviation (SD) of a child’s actual height to the median height of the children of their age When the HAZ is a negative value, it means that child's actual height is less than the median height Children were considered as stunted if they had a z-score of height-for-age below -2 standard deviations (WHO, 1997)

The Young Lives wealth index is constructed by three indices, including housing quality, consumer durable, and services index (see Appendix 1) The housing quality index is a composite of the number of rooms per person and reflects the floor, roof, and wall materials This index has value between 0 and 1 The second index is the consumer durables index In Vietnam, consumer durables index

is measured by 9 items, including radio, motorbike/scooter, car, mobile phone, refrigerator, bicycle, television, landline telephone, and fan A dummy variable takes value of 1if household owned any of these items The consumer durables index is calculated as the total of 9 dummy variables and then divided by 9 to give a value between 0 and 1.It should be to note that productive assets are not included in this measurement Finally, services index is the simple mean of 4 dummy variables, which reflect a family’s access to safe drinking water, electricity, toilet, and cooking fuel (see more in Table 2) The wealth index is constructed to evaluate the socioeconomic status of a household and provide a ranking of families through information on their wealth The wealth index is a continuous score from 0 to 1

Father’s education refers to father’s years of schooling If the level of

education is equivalent to vocational technology school or university, the variable is

14 or 16 correspondingly Caregiver’s education is defined similarly

Table 2: Variables definitions

Minority status Child is ethnic minority

Coded 1 for ethnic minorities, 0 for Kinh or Chinese

indicator Dummy variable

Dummy = 1 if yes, 0 otherwise Negative

Urban Whether the household is located

in a urban

Coded 1 for urban, 0 for otherwise Positive Northern Upland Rural household in Northern

Upland

Take a value of 1 if yes, 0 for otherwise Negative Mekong River Delta Rural household in Mekong River

Delta

Dummy = 1 if yes, 0 otherwise Negative Central Coastal Rural household in Central

Coastal

Dummy = 1 if yes, 0 otherwise Negative Siblings Number of children born to the child’s mother Discrete

Total bad events

The number of economic shocks have happened since the mother became pregnant

Discrete

Community characteristics were constructed by dummy variables coded 1 for urban residence and 1 for rural residence in North-East region (Lao Cai), South

Central Coast (Phu Yen), Mekong River Delta (Ben Tre) In order to avoid multicollinearity problem, this study does not include all of the community characteristics into the model

Long-term health problem indicator is measured by a caregiver report A dummy variable equals to 1 if the children has long-term health problem Number

of siblings refers to the number of children born to the child’s mother (born alive),

excluding the index children

3.3 Data source

Data comes from Young Lives (YLs) YLs is an international organization study about childhood poverty in India, Ethiopia, Peru and Vietnam In each of the countries, 3000 children were enrolled and tracked every three year over a 15-year period It is divided in two cohorts: a younger cohort consists of 2000 children who were born in 2000-2001, an older cohort include 1000 children who were born in 1994-1995

This thesis employs the first two rounds of data from the Young Lives Longitudinal Survey in Vietnam, which were collected in 2002 (the first round) and

2006 (the second round) The sample was a group of approximately 2000 children (a younger cohort) in 20 communities of five provinces, covering different geographical areas: Lao Cai, Hung Yen, Phu Yen, Da Nang and Ben Tre province For further information on the selection of sample can be found in the Young Lives website (www.younglives.org.uk)

Sampling: In each community (a sentinel site), 100 index children for younger group were selected by an equivalent of random sampling The survey was conducted in two periods of time and reflect a wide range of variables on characteristics of children and their families The first period (Round 1) of survey was started in 2002, when the children were 6 to 18 months old The second round

of data collection was carried out from 2006 to 2007 in the same children, when the children were about 4 to 6 years old

CHAPTER IV: RESULTS AND DISCUSSION

This chapter has two main sections In the first section presents descriptive information on data available from the Young Lives survey in Vietnam (which was collected in 2002 and 2007) The second section discusses the results of the OLS and IV estimates that attempt to examine the effects of preschool attendance on children’s cognitive outcomes

4.1 Data descriptions

Descriptive statistics are reported inTable 3

Table 3: Description of variables

Child is ethnic minority 0.144 0.351 0 1 Exogenous

Height-for- age z-score -1.128 1.310 -7.150 8 Exogenous

Long term health