Abhatoo - Centre Nationale de Documentation

This paper discusses the question: which characteristics are appropriate for a measure of period mortality and how are these characteristics met in conventional and tempoadjusted life expectancy? According to our perspective, a period mortality measure should include exclusively the current mortality and should enable comparison of periodspecific mortality conditions of two populations or the changes between two periods without depending on past or future trends. By using a simple population model, we show that the conventional period life expectancy does not meet these demands since it includes specific assumptions regarding future mortality, which differ between different populations and at the end can lead to paradoxes disturbing its practical purpose. Tempo-adjusted life expectancy, however, is free of these compositional effects and thus enables the analysis and comparison of pure period-specific mortality conditions. From these considerations we also derive an interpretable definition for tempo-adjusted life expectancy. We suspect that this lack of definition could be a major reason for the general rejection of mortality tempo-adjustment. Finally, we present estimates for tempoadjusted life expectancy for the period 2001-2005 for 41 countries showing that tempo effects and their adjustment are not only a technical issue but can have significant impacts on the interpretation of period mortality.