Using the National Longitudinal Survey of Youth in the Us to Study the Birth Process : A Bayesian Approach

This paper employs the National Longitudinal Survey of Youth in the US to study the birth process. We develop a simultaneous equations model with seven endogenous variables: four birth inputs (maternal smoking, maternal drinking, first trimester prenatal care, and maternal weight gain), three birth outputs (gestational age, birth length, and birth weight), and twenty-four exogenous variables. The estimation is Bayesian. Separate analyses are performed on five different groups: Main Whites, Supplemental Whites, Blacks, Hispanics, and Native Americans. In all groups, we find sizeable correlation between the disturbances in the four input and three output equations and among output disturbances. For gestation, the effect of maternal weight is positive and substantial, while the effect of maternal age is consistently negative and substantial for Main Whites, Blacks, and Hispanics. The effects of smoking, drinking, prenatal care, and weight gain vary in sign and magnitude across the groups. For birth length, male infants are on average longer. The effect of maternal height is noticeable but small in magnitude, and the effect of maternal weight is noticeable only for Main Whites. The effect of smoking is consistently negative, and substantial for Main and Supplemental Whites. The effects of drinking and prenatal care vary across the groups. Both weight gain and gestation have consistently positive effects. For birth weight, male infants are on average heavier except Hispanics. The effect of maternal height is noticeable for Main Whites and Hispanics. The effect of maternal weight is noticeable and consistent across the groups. The effect of smoking is consistently negative, and substantial for Main and Supplemental Whites. The effects of drinking and prenatal care are small and vary across the groups. Weight gain has a small positive effect except Supplemental Whites. The effect of gestation is positive and fairly comparable across the groups