Mortality synthetic indicators. Theoretical aspects and an application to Italy

Michela Camilla Pellicani, Università di Bari
Alessandro Polli, Università di Roma "La Sapienza"

Our aim is the estimation of a non weighted indicator of mortality in the Italian population, and its temporal comparison with a standard indicator based on weighted sums of mortality rates by age derived by mortality tables. The construction of a non weighted (or synthetic) indicator of mortality has not been matter of detailed discussion in the scientific literature and, as a matter of fact, there’s only a limited number of publications on this subject. The rationale of a non weighted approach, in the analysis of fertility and migration, is the immediate interpretability of the measurements obtained in terms of number of events expected for an individual. The same interpretation, on a conceptual basis, has no sense in the case of non renewable event, such as deaths, where the number of events is by definition equal to one, no more, no less. However, in case of non recurring events, this measure can always be interpreted as an indicator of the general level of mortality in a population (in terms of timing) historically or geographically determined. Moreover, by this measure, it is always possible to determine the distribution of age at death, summarized by the median age, in order to estimate an alternative indicator to the life expectancy one. We would like focus therefore on non weighted indicators for their interesting features. First, they are additive. Moreover, non weighted indicators are sensitive to the level of disaggregation by age, unlike the weighted ones. The synthetic indicator of mortality has a key advantage over similar indicators based on a weighted sum, because in space/temporal comparison, it can neutralize the bias resulting from dynamic factors, like age−specific lifestyles resulting in different mortality rates (Sardon 1998). The adoption of this procedure can neutralize the homogenization effect, a central feature of all weighted sum procedures.

  See extended abstract

Presented in Session 71: Mortality measures and models

´