Estimating Mortality by State of Birth and Race Using Census and Vital Statistics Data

Dan Black, University of Chicago
Yu-Chieh Hsu, NORC at the University of Chicago
Lowell Taylor, Carnegie Mellon University

Demographers often estimate population parameters by combining data from separate sources. We propose a generalized method of moments (GMM) estimator for such cases, and to illustrate this statistical approach we estimate mortality rates using data from the U.S. Census and from the Vital Statistics. Specifically, we estimate mortality rates by race, gender, birth cohort, and State of birth for cohorts born in the 1930s. On a substantive level, resulting estimates are interesting: In mid-life (ages 40 through 60), State-of-birth effects are quite large for men, but not for women. Among both black and white men mortality is generally higher for individuals born in relatively low-income States---typically Southern States.