Decompression of period old-age mortality: When adjusted for bias, the variance in the ages at death shows compression

Formal derivations and empirical evidence, in the framework of Fries-Kannisto's hypothesis, show that the indicators of mortality compression based on age-at-death distribution, left-censored at a fixed old age, may be subject to a bias toward showing mortality decompression in the case of a mortality decline. The previously reported increasing variance in ages at death above fixed old ages in developed countries was mainly the effect of a mortality shift, not decompression. When adjusted for this bias, the indicators of variance in ages at death show a compression of period mortality