Principal investigator: Viviane Pires Ribeiro
Paper title: CHILD SURVIVAL, HEIGHT FOR AGE AND HOUSEHOLD CHARACTERISTICS IN BRAZIL
Article authors: Duncan Thomas, John Strauss and Maria-Helena Henriques
Location of the intervention: Brazil
Sample size: 41.233 households
Main theme: Health Insurance
Primary variable of interest: Height and Child Survival Rate
Type of intervention: Study by region of household variables that influence child height and survival.
Assessment method: Experimental Evaluation (RCT)
Evaluation Context
The National Household Expenditure Survey (ENDEF) is a large-scale household expenditure survey in Brazil. This survey was useful to the authors due to the extent of the dataset, which covers a considerable number of families, and the breadth of questions asked to these families. Questions such as consumption and expenditure, income, height and weight of all members, and fertility of each woman were included. Although the data are tabulated at the regional level, the authors found no other work published at the time that explored the richness of the data at the household or individual level.
Intervention Details
Brazil was divided into the same seven regions used in the National Household Sample Surveys (PNADs). The authors used information from 41.233 households, seeking to determine how household characteristics are related to the height of children under eight years of age. Since a child's height varies according to sex and age, Thomas et al. (1990) observed that within each of the four age ranges, the average child height is greater in the urban south, and in each region, height is greater in urban areas. They also identified important regional differences in child nutrition outcomes that are independent of income; on average, income is higher in the urban Northeast compared to the rural South.
For the analysis of infant survival rates, the authors distributed the mothers' ages into four age groups. The differences between urban and rural South are small, although few mothers lost a child in the urban sector. In the Northeast, mortality rates are higher and the difference increases with age; furthermore, there is a significant difference between urban and rural sectors.
Methodology Details
Thomas et al. (1990) adopted Becker's (1965) model of household behavior, assuming that families exercise choice over consumption, including leisure. Thus, the number of births, infant survival rates, and child height can be thought of as outputs of a health production function. Some of these inputs are under parental control (such as food consumption, breastfeeding time, allocation of parental time, and use of community services); others are predetermined (such as parental height, education, or the incidence of disease in the community).
Result
The authors found a positive and significant impact of maternal education on child height and survival. Both outcomes are affected by paternal education, although in the case of survival, the effect of maternal education is stronger.
The mother's and father's education are significantly associated with a child's height. In urban Northeast Brazil, for example, compared to having an illiterate mother, a child with a literate mother will be about 1,6% taller, 2,5% taller if the mother completed primary school, and 4,2% taller if she completed secondary school. The marginal effects of maternal education are slightly smaller in the South. The same pattern is identified in rural areas, although the magnitudes are considerably smaller.
The height of both mother and father are important determinants of a child's height, having a greater impact in rural areas. Maternal height also has a significant effect on child survival. The effect of income on survival rates is positive and significant, except in the urban South.
The impacts of household-level variables differ by region and urban or rural residence, particularly in terms of child survival outcomes. Maternal education has a greater impact on child survival in the Northeast compared to the South; a similar pattern is identified for family income in relation to height and child survival. The impact of maternal height on child survival is greater in rural areas and in the Northeast. These differences suggest that regional factors play an independent and important role, as well as modifying the impact of household-level variables.
Lessons in Public Policy
Prior studies to that of Thomas et al. (1990) focused primarily on income inequality in Brazil and paid little attention to other indicators of well-being, such as survival and child height, which are characterized by substantial inter-regional and inter-regional heterogeneity. Given this gap, the authors sought to determine the impact of household characteristics on child survival and height, conditional on age. The results indicate that parental education has a positive and significant effect on both outcomes. Although part of this effect is through income, most of the educational effect is independent of family income. Thus, other studies that focused on the role of education solely as a means of increasing income underestimated its importance in improving well-being. The authors argue that education can be considered a proxy The socioeconomic characteristics of the family demonstrate that the educational effects remain positive and significant, even after controlling for height and family income.
Parental heights are important determinants of child height and survival. In both cases, the influence of maternal height is greater than that of paternal height, possibly reflecting the mother's greater role in raising the child. The results found by the authors support the view that height is a useful predictor of human well-being, even when controlled for income and education. Therefore, policies aimed at increasing the long-term nutritional status of a population are likely to have complementary effects on long-term child survival.
Reference
THOMAS, Duncan; STRAUSS, John; HENRIQUES, Maria-Helena. Child survival, height for age and household characteristics in Brazil. Journal of Development Economics, v. 33, no. 2, p. 197-234, 1990.
