Life tables under current risk composition based on observed, fixed characteristics

Abstract

We examine how differences in risk composition between the overall synthetic life table cohort versus one that would be living under current mortality conditions influence life table mortality statistics. We propose formulas (i) for adjusting the force of mortality and life table statistics, eliminating heterogeneity introduced by the lagged risk composition of the synthetic cohort based on observed, fixed characteristics; and (ii) for decomposing the difference between two statistics under current risk composition, life expectancy (LE) and average lifespan deprivation (Dep) into the effects of differences in composition and of the summary mortality statistics of the subgroups. We use data on educational attainment in Denmark in 1991–1995 and 2011–2015 to illustrate these methods. The empirical example shows that the difference between the “standard” LE and the one under current risk composition is noteworthy, but the gap for Dep is negligible. Additionally, both the increase in the LEs of the educational groups and the shift in the composition of the population due to educational expansion contributed to the increase in the LE under current educational composition of both sexes. The observed decrease in Dep over the study years for both sexes resulted predominantly from the decrease in Dep of educational groups.