Bayesian analysis of an econometric model of birth inputs and outputs

This study offers a simultaneous equations model of the birth process with seven endogenous variables: four birth inputs (maternal smoking, maternal drinking, first trimester prenatal care, and maternal weight gain) and three birth outputs (gestational age, birth length, and birth weight). The data are taken from the National Longitudinal Survey of Youth. Our analysis conditions on twenty-nine exogenous variables including four racial dummies to account for the widely cited racial differences in birth outputs. We find that there is sizeable correlation between the disturbances in the four input and three output equations and among output disturbances, and that results from our simultaneous equations model are substantially different from those using the single-equation approach. It appears that the High/Low Risk Birth Weight Puzzle remains unresolved under our modeling framework.All correspondence to Kai Li. The authors gratefully acknowledge the research support of the Social Sciences and Humanities Research Council of Canada, and helpful comments on an earlier draft by the editor (Junsen Zhang), two anonymous referees, Sid Chib, Kei Hirano, John Geweke, Gary Koop, seminar participants at the University of Alberta, the University of California at Irvine, University of Kansas, University of Minnesota, Queen Mary College of University of London, Washington University at St. Louis, and participants of the 2001 Winter Meetings of the Econometric Society in New Orleans. We are solely responsible for any errors contained herein. Responsible editor: Junsen Zhang.