Ethnic gaps in child education outcomes in Vietnam: an investigation using Young Lives data

ABSTRACT

There are large gaps in child education outcomes between the Kinh majority and non-Kinh minorities in Vietnam. This paper seeks to understand the reasons for these ethnic gaps. The examination employs Probit and multilevel regression models, and associated decomposition techniques. The results show that Vietnam’s ethnic gap in school enrolment is mostly attributable to household characteristics such as household expenditure and father’s education. Gaps in schooling progress and performance are explained by a broader set of variables such as child, household, commune, school, and peer characteristics.

KEYWORDS:

• Ethnic gap
• education outcomes
• decomposition
• Vietnam

JEL:

• I24
• O15

Acknowledgements

I thank Blane D. Lewis and Raghbendra Jha for their valuable comments. I also thank the reviewers for their constructive comments. The data used in this publication come from Young Lives, a 15-year study of the changing nature of childhood poverty in Ethiopia, India, Peru and Vietnam (www.younglives.org.uk). Young Lives is funded by UK aid from the Department for International Development (DFID), with co-funding from Irish Aid. The views expressed here are those of the author(s). They are not necessarily those of Young Lives, the University of Oxford, DFID or other funders. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 Using the asset index derived from the simple count method to represent household economic condition was found to yield consistent results as controlling for household expenditure (Montgomery et al. 2000), or using the index derived from the principal components analysis (Bollen, Glanville, and Stecklov 2002; Paxson and Schady 2007) as well as from various other methods (Filmer and Scott 2012).

2 The estimation results of the models with the same set of regressors are provided in Table A2 and Table A3. Most results are consistent with those from the models with the full set of regressors.

3 An alternative method to estimate models of enrolment and SAGE by using household data is multi-level models, with child and commune level data. However, the nonlinear relationship between covariates and the depended variable leads to a difficulty in decomposition technique. Table A4 provides the estimation results of enrolment and SAGE from multilevel mixed effect Probit model. Because the estimated values of coefficients in the multilevel mixed effect Probit model are quite similar to those in Probit model (Table 4), I expect that the decomposition results derived from the two methods, if available, should be similar, too.

4 Although the absence of school characteristics might cause bias in estimated school travel time, this bias is believed to have minor effects on decomposition results, which are the main interest of this study. As shown in the decomposition results, school travel time is not an important contributor to the ethnic gap in schooling progress. Moreover, the estimations of maths and Vietnamese test score equations (Tables 7 and 8) show that school travel time is insignificant when school characteristics are controlled for.

5 In the data, 87% and 94% of the household heads are male in Kinh and non-Kinh groups, respectively. The proportion of younger cohort children seeing their mothers daily is 93% compared to 85% of them who see their fathers daily. Most of the children, 96%, have their mother as their primary caregiver.

6 The decomposition for the Tobit model of SAGE shows that the total explained part accounts for 102% of the gap.

7 To check whether this finding is driven by a high correlation between mother’s education and father’s education as suggested by Becker (1973) as regards assortative mating between men and women, education of the father and education of mother is in turn excluded from estimations (see Table A7 and Table A8). The results are still consistent with those when education of both parents are included: father’s years of schooling are the dominant factor explaining the enrolment gap and a minor contributor to the SAGE gap; mother’s education is insignificant in the enrolment gap but significantly contribute to the SAGE gap.

8 The positive sign of the dummy variable for parents’ education that is unknown, despite being insignificant in some cases, shows that parents’ education unknown by children has a stronger influence on test scores than that known by children. A further investigation shows that Kinh and older children, and children speaking Vietnamese at home, are less likely to know their parents’ education (Table A1). Hence, the stronger impact of unknown parents’ education might be partly due to a positive association between test scores and Kinh ethnicity, age in month, and speaking Vietnamese at home.