Family Planning, Gender Preference, and Child Health: The Case of Bangladesh

Juhua Yang, Brown University

One of the rationales for family planning is to promote women and children¡¯s health. In recent decades, family planning has been widely adopted in developing countries. Meanwhile, child health in these countries, as indicated by mortality and nutrition, although still less than optimal, has substantially improved. Are family planning and child health associated with each other? What is the nature of their relations, if they are related? What is the place of gender ideology and the value of children in the relationship between family planning and child health? These broad questions motivate my paper. Specifically, my paper investigates the relationship between family planning and child nutrition in Bangladesh. I am particularly interested in how family planning, as a macro level institution, may interact with factors at micro and household level, thereby affecting child health. Using 1996-1997 Bangladesh Demographic and Health Survey, I examine the determinants of nutrition as indicated by child anthropometrical measures. Conceptually, family planning could affect child health in multiple ways. I identify four possible pathways for family planning to affect child health. First, it could affect ideology, specifically ideas about the value of children and gender preference. To motivate couples to accept the small family norms and practice family planning, gender equality has been promoted since the 1950s in Bangladesh (NIPORT 2001). This would presumably erode traditional son preference, change parental views on the value of children, thereby producing fewer children but rearing them in a better and more equal way. While all children¡¯s health is expected to improve, female children may particularly benefit from the family planning campaign. Second, the commonly held view about the mechanism for the effect of family planning centers around family configuration of sibship. Specifically, family planning improves child nutrition by widening birth interval, reducing higher parity births, and thus decreasing family size, which in turn increases the share of household resources for each child thereby improving their nutrition. Thirdly, family planning affects child health and nutrition through improving the availability and use of health infrastructure. In Bangladesh, family planning and health programmes work together. Trained family planning and health staff was deployed across the country to deliver family planning and health services down to the village level (Mahmud and Mahmud 2000). Health and family planning clinics were also established at communities (NIPORT 2001). These clinics, together with the extensive propaganda of family planning programmes, not only provide better access to health infrastructure, but also increase knowledge of health, nutrition and childcare, the last pathway in family planning-health linkage. This in turn motivates people to make a better use of health resources, thereby improving child health and nutrition. DATA AND METHOD Sample I use 1996-1997 Bangladesh Demographic and Health Survey to test my hypotheses. A children¡¯s file is extracted from the women¡¯s survey, household survey and the service provision assessment survey. The total sample size is 5073, including children 0-59 months old. Variables Table 1 presents variables that will be used in this analysis, their definitions, means and standard deviations. Three different anthropometric measures will be used as dependent variables. Height-for-age and weight-for-height are numerical measures; malnutrition, composed of height-for-age and weight-for-height, is a dichotomous measure indicating if a child is malnourished. A child two standard deviations below the WHO/NCHS median on either measure is coded malnourished. Method After identifying and defining the dependent and independent variables, the next step of this research is to perform a multivariate and multilevel analysis, including individual, household and community levels. Corresponding to the different measures of the dependent variable, two statistical methods will be adopted. For height-for-age and weight-for-height, I use ordinary least regression model. For the dichotomous measure, malnutrition, logistic regression technique is used, with its equation being specified as: Log [Pijk/(1-Pijk)] = aIijk + bHjk + cCk + dCfp + ¥å The numerical dependent variables, htage (height-for-age) and wtht (weight-for-height), are interpreted as a linear function of individual, household, and community factors. The dichotomous measure is interpreted as the log odds that individual child i in household j and community k is mal nourished (Pijk) in comparison with those not (1-Pijk). Iijk, Hjk, and Ck represent respectively the vectors of individual, household and contextual characteristics while Cfp is the vector of family planning or related attributes. Parameter estimates, a, b, c and d, indicate the effects of covariates at each levels. Finally, ¥å represents variations not accounted for by variables specified in this research. I will first run a model with only family planning and intervening variables to see their gross effect on the outcome variable. Then a second model with all control variables will be run, and results from this model allow me to examine the net effect of family planning on child nutrition. A third model containing interaction terms between gender and family configurations (birth order, spacing and siblings) is employed, which will enable me to compare if family planning programs affect girls and boys differently. Each of the 3 models will be run three times, using different dependent variables. The aim of this project is to assess whether family planning brings about a relative shift in child nutrition in general, and female child nutrition in particular, by examining the triangular relationship among family planning, child nutrition, and gender preference in Bangladesh. Preliminary analysis indicates that family planning is associated with child nutrition through the functioning of family configurations. Interaction terms between child health and sibship variables yield mixed outcomes for boys and girls across measures. Reference Mahmud, Simeen, and Wahiduddin Mahmud. 2000. Bangladesh. In Promoting Reproductive Health: Investing in Health for Development, edited by Forman, Shepard, and Romita Ghosh, pp.19-48. Boulder: Lynne Rienner Publishers. National Institute of Population Research and Training (NIPORT). 2001. Bangladesh Demographic and Health Survey, 1999-2000. Calverton, Md.: Macro International.