How Do Mortality Differences across the U.S. Matter?

Yuan Peng, Huazhong University of Science and Technology
Megan Cassidy Chiang, Stanford University
Wenyun Zuo, Stanford University
Zhen Guo, Huazhong University of Science and Technology
Shripad Tuljapurkar, Stanford University

BACKGROUND: There are large mortality gaps among U.S. states measured by life expectancy, lifespan variance, and crude death rate. Yet these measures tell different stories, since they are affected differently by mortality change across space and time. OBJECTIVE: Identify the contributions of age-specific mortality rate differences to gaps in the three aggregate measures. METHODS: We make novel use of decompositions to analyze how age-specific mortality differences determine each of the three mortality measures. For life expectancy and lifespan variance, we separate the size of age-specific mortality differences from the sensitivity to those differences. For crude death rate, we separate exactly the age-specific contributions of population age structure, and mortality rates. CONCLUSIONS: Differences in under-5 mortality do not significantly affect differences in any measure. Mortality differences at middle and old ages play a significant role in differences in life expectancy. In contrast, differences in lifespan variance are more affected by mortality differences at the middle ages than at old ages. The significance of mortality differences differs in the two sexes. Crude death rates are largely driven by the elderly, with cross-state differences primarily explained by the share of the older population, while the role of mortality differences has grown over time. CONTRIBUTION: We use a sensitivity-based decomposition approach and the Kitagawa decomposition to analyze gaps in three aggregate mortality indicators. Our approach is essential because an age group's contribution to a gap in a given aggregate indicator can result from large mortality rate differences, high sensitivity of the indicator, or shift in population composition.

Keywords: Mortality and Longevity, Decomposition analysis, Mathematical demography , Data and Methods

See paper.

  Presented in Session 72. Subnational and Spatial Variation in Mortality