Determinants of Nutritional Outcomes of Children in India: A Quantile Regression Approach

In this paper, we use quantile regressions on data from the 2005-06 wave of the Indian National Family Health Survey to study the determinants of child body-mass-index, height-for-age, and hemoglobin at different points of the conditional distribution. Our results show that only considering the conditional mean of the entire distribution can yield misleading results. In light of compelling evidence on sex-selective abortion and infanticide, we use a Heckman correction for our quantile regression to control for the “underreporting” of female births documented by Rose (1999). We find that household maternal health and education have larger effects at the lower end of the distribution than on the upper end, for all three child nutritional indicators. Results show that iron supplements are less effective at increasing hemoglobin levels in the worst-off children. We argue that policy interventions must account for socioeconomic diversity or have little hope of meeting their target.