Population policy, family size and child malnutrition in Vietnam – Testing the trade-off between child quantity and quality from a child nutrition perspective

Highlights

• Young Lives-Vietnam data were analyzed.

• The sex of the first-born is used as an instrumental variable for sibship size.

• Having one more sibling reduces 5-year-olds’ height-for-age by 0.30 SDs.

• Having one more sibling reduces 5-year-olds’ weight-for-age by 0.26 SDs.

Abstract

This paper estimates the quantity-quality (QQ) trade-off of children from a child nutrition perspective, using data on 2000 children born in 2001–2002 from the Young Lives project in Vietnam. Using the sex of the first-born in the family as an instrumental variable for family size, we found that having an additional sibling lowers a 5-year-old Vietnamese child’s height-for-age and weight-for-age Z scores by 0.3 and 0.26 standard deviations (SDs), respectively. This finding is robust to a series of checks, providing strong evidence of the quantity-quality trade-off of children. In more policy-relevant terms, our estimates suggest that the violation of Vietnam’s Two-Child policy (i.e., having more than two children) may have caused reductions in 5-year-old Vietnamese children’s height-for-age and weight-for-age by, respectively, 0.49 and 0.57 SDs.

Introduction

Demographers, economists, and other social scientists have long been interested in the role family plays in shaping children’s future. Among the most commonly observed family characteristics, family size (i.e., “the number of children in the family”, or “sibship size”, used interchangeably below) has attracted considerable attention, mainly because of the empirical regularity that children from larger families on average have poorer human capital outcomes than those from smaller families (e.g., Blake, 1981,1989; Schultz, 2008). Such an empirical regularity, supported by the theory of the quantity-quality (QQ) trade-off of children (Becker and Lewis, 1973; Becker, 1981),¹ has important implications for population policy in developing countries that are characterized as having both a high fertility rate and a low human capital stock. Many countries in Asia, such as China, Indonesia, Singapore, and Vietnam, have adopted family planning policies to control family size for decades, in the belief that population control is an effective means of enhancing human capital formation for future generations.

Yet it may be premature to infer causation from larger family size to poorer human capital outcomes of children, and thus to advocate population control policies, solely based on the afore-mentioned empirical regularity. If, for example, parents’ fertility and child human capital investment decisions are jointly made, then the commonly-observed negative association between family size and child outcomes may reflect reverse causality from the latter to the former or influences of unobserved factors that affect both (e.g., genetic fitness). Fortunately, with the identification strategy devised by Angrist and Evans (1998), which exploits parental preference over child sex-mix to create instrumental variables (IV) for family size, considerable progress has been made in obtaining credible causal estimates of family-size effects in various settings (e.g., Angrist et al., 2010; Black et al., 2005; Conley and Glauber, 2006). Important modifications of this strategy have also been made to fit the particular form of fertility preference in Asian countries, which is dominated by Asian parents’ strong desire for sons stemming from Confucian thought (UNFPA, 2012). More specifically, the sex of the first-born child was used as an IV for family size to circumvent sex-selective abortions that occur in higher-order birth parities in a number of recent studies conducted in East Asia (e.g., Lee, 2008; Kang, 2011; Chen, 2017).

However, applications of this child sex-mix identification strategy have been limited to studies of family-size effects on children’s educational outcomes, such as years of schooling (Black et al., 2005), grade repetition (Conley and Glauber, 2006), test scores (Angrist et al., 2010), as well as parental monetary investment in child education (Lee, 2008). Less attention has been paid to (family-size effects on) children’s nutritional outcomes. This lack of attention is unfortunate, because children’s nutritional status is not only a measure of their current standard of living, but also a key determinant of their socioeconomic attainment in adulthood. Numerous studies have shown that malnutrition can undermine children’s physical growth (Case and Paxson, 2008; Currie, 2009), cognitive development (Glewwe and Miguel, 2008; Grantham-McGregor et al., 2007; Duc, 2011; Walker et al., 2011), mental well-beings (O’Neil et al., 2014; Liu et al., 2019), and social skills (Liu and Raine, 2017), whereby limiting their productivity and earning potentials in adulthood. Worse still, some aspects of child malnutrition, stunting (low height-for-age) in particular, cannot be reversed by extra nutrition intakes or better healthcare received later in life. It follows that the most effective way to avoid the afore-mentioned detrimental consequences of child malnutrition is to prevent child malnutrition from occurring as early as possible. If large family size is indeed a significant cause of child malnutrition, then implementing population control policies may be an effective means of mitigating child malnutrition. It is, therefore, of interest to investigate how family size (and the related population policies) may affect children’s nutritional outcomes, especially in developing countries where child malnutrition is prevalent. The child sex-mix identification strategy offers an opportunity for such an investigation.

Vietnam provides an interesting case to study. Vietnam enacted its Two-Child policy (sometimes referred to as the “One-or-Two-Child” policy) in 1988,² two years after the initiation of its official program of economic reforms (Doi Moi) that brought about rapid economic growth in subsequent decades. Aiming to stem its rapid population growth in the 1980s, as well as to “build families with fewer and healthier children”, the Two-Child policy allowed for no more than two children in each Vietnamese family.³ The decades that followed witnessed a rapid decline in the total fertility rate in Vietnam, from 3.8 children per woman in to 2.1 in 2006 (General Statistics Office of Vietnam, 2008). Substantial improvements in children’s nutritional status, at least measured by height-for-age, were also observed during the same period (Glewwe et al., 2004; Humphries et al., 2017).⁴ These observations naturally raise a question: did the Two-Child policy contribute to the rapid improvement in Vietnamese children’s nutritional status? Given the recent discussion on whether the Two-Child policy should be relaxed or tightened (Pham et al., 2013), the answer to this question can certainly help inform that discussion.

To shed some light on this question, the present study attempts to estimate the causal effect of family size, as well as that of parents’ decision to have a third child (i.e., in breach of the Two-Child policy), on Vietnamese children’s nutritional status, using the Young Lives (https://www.younglives.org.uk/) data from Vietnam (YLVN) on 2000 children around age 5 in 2006. To address potential endogeneity of parents’ fertility decisions, we follow the recent literature and exploit Vietnamese parents’ prenatal son-preference to create IVs for family size. Similar to their counterparts in other Asian countries, Vietnamese couples have a strong desire for sons (Haughton and Haughton, 1995; Croll, 2000; Guilmoto, 2012). Under the Two-Child policy, the strong son-preference implies that the number of sons born in the first two birth parities in a family will have a strong predictive power for the likelihood that this family will violate the policy to have a third child, as well as the total number of children this family will eventually have. To further circumvent sex-selective abortions in the second parity (due to parents’ hope to have at least one son within the two-child quota), we follow Lee (2008) and Chen (2017) and use the sex of the first-born as the IV for family size. Consistent with the prediction of the QQ trade-off of children, our IV estimates reveal strong negative effects of family size on children’s nutritional status: other things being equal, having one more sibling lowers a 5-year-old child’s height-for-age and weight-for-age scores by, respectively, 0.30 and 0.27 standard deviations (SDs). Violation of the Two-Child policy (i.e., having more than two children in the family) is found to have lowered a child’s height-for-age and weight-for-age by 0.57 and 0.49 SDs, respectively. These findings suggest that reductions in family size due to the Two-Child policy were likely a key driver of the rapid improvement in Vietnamese children’s nutritional status since the 1990s.

Our study makes two contributions to the literature. First, while this study is neither the first to examine the impact of family size in Vietnam (see e.g., Dang and Rogers, 2016) nor the first that exploits the sex of the first-born to identify family-size effects (see e.g., Millimet and Wang, 2011), it is the first that bridges this Asia-specific instrument variable and the context of Vietnam. More specifically, this is the first study to exploit exogenous variations in child sex-mix to identify the family size-child nutrition relationship in Vietnam Second, the YLVN data contain valuable information on parents’ fertility preference, in particular, the desired numbers of sons and daughters, which was not available in previous studies. Not only can this information help us control for parents’ fertility preference in the analysis, but it also provides a number of ways to help assess the robustness of our estimation results, whereby strengthening our findings.

The remainder of this paper is structured as follows. The next section describes the data and the analytical sample. Section 3 develops the framework underlying our empirical analysis. Section 4 presents and discusses our findings. The final section draws conclusions and points out a number of directions for future research.