Old Age Mortality Measurement and Modeling

Natalia S. Gavrilova, University of Chicago
Leonid A. Gavrilov, University of Chicago

Mortality measurement at old ages poses serious challenges to researchers: age reporting at extreme ages may be inaccurate; the small number of persons at advanced ages often leads to pooling together data for people belonging to different birth cohorts (which results in data heterogeneity); and finally, standard assumptions used for annual mortality estimates may be questionable when risk of death is particularly high. This study uses data from the US Social Security Administration Death Master File (DMF) to make old age mortality estimates for more homogeneous single-year birth cohorts, and mortality rates measured at more narrow (monthly) age intervals. Analysis of extinct birth cohorts (born in 1887-1894) found that the extent of old age mortality deceleration is surprisingly small even at extreme ages (100-105 years), and is decreasing further with improving data quality. The study was supported by NIA (grant AG028620).