The morphology of the life table

As living standards improve, mortality declines, and the age-pattern of human mortality changes. Yet, there is no one pattern of mortality decline., and different populations may have a different age structuring of mortality, even though they have a similar average level of mortality.
In the present paper we consider the dimensionality of mortality using a random sample of 109 population tables (male and female) drawn from the Human Mortality Database. Factor analysis of log( p(x))
values shows a three factor solution to be adequate, the three factors being:
1. Mortality to age 50 for men, to age 80 for women
2. Mortality for ages 50 to 80 for men
3. Mortality above age 80 for men and for women.
We compare these factor analytic measures with measures proposed in the mortality literature as measures of the shape of the mortality / survivorship curves and find that most of these reproduce necessary changes in the shape of the life-table functions as mortality declines, but do not distinguish between the different shapes of curves in life tables with similar overall levels of mortality. Following the logic of the factor analytic model, we thus suggest that the different elements of the life table are better expressed by age-specific survival probabilities, focussing on younger ages, middle ages (for men) and old ages