Nighttime light intensity and child health outcomes in Bangladesh

This paper explores the effect of urbanization on child health outcomes in Bangladesh. We use nighttime light intensity as a measure of urbanization and find that the higher intensity of nighttime light, the higher the degree of urbanization, which positively affects child health outcomes. We apply a novel methodology that combines the non-parametric and panel regression approach with the Gradient Boosting Machine that originates from machine learning algorithms. Our estimates suggest that one standard deviation increase in nighttime light intensity is associated with a 1.52 rise in Z-score of weight for age. The maximum increase of weight for height and height for age score range from 5.35 to 7.18 units. We perform several robustness tests, including a wide-ranging set of controls in generalized additive models, and find robust positive relationship holds. Our findings make several contributions: first, we rationalize our empirical findings in a utility and hybrid production function showing that urbanization's marginal effect on children's nutritional outcomes is strictly positive. Second, the relationship is nonlinear and U-shaped, where low and higher light intensity leads to poorer child health outcomes, with increases being observed to be positive at all times and along the trend line. Finally, our findings are closely linked to the effective policies in reducing children's malnutrition in low- and middle-income countries. We suggest that policies favoring small towns are more effective in improving child health outcomes than expanding megacities.