Future mortality in the Netherlands: application of a new methodology

Leo van Wissen, University of Groningen
Fanny Janssen, Population Research Centre, Groningen
Anton Kunst, University of Amsterdam

Objective of this paper is to obtain future levels of life expectancy (e0, e65, e80) for the Netherlands up to 2050 by separately projecting smoking-related and non-smoking related mortality, and taken into account the mortality experiences of the opposite sex and in 10 other European countries. Data on all-cause mortality, lung cancer mortality and population by age, sex, and year for the Netherlands was obtained through Statistics Netherlands for the period 1970-2006. For the other European countries these data were obtained through the Human Mortality Database and WHOSIS. Non-smoking-related mortality is estimated using an adapted version of the indirect Peto-Lopez method. Both all-cause mortality and non-smoking-related mortality were projected into the future by means of the Lee-Carter and Lee-Li methodology. Smoking-related mortality is projected into the future by means of assumptions on the future course of the proportion of mortality caused by smoking. Based on the proposed methodology, the life expectancy at birth in 2050 will raise to 88.1 years for Dutch women, and 83.8 years for Dutch men. These values are higher than the most recent forecast of Statistics Netherlands, especially for females. The separate prediction of smoking-related mortality and non-smoking-related mortality led to lower life expectancy values in the short run and higher life expectancy values in the long run, especially for females, not so much for males, and especially at the higher ages. The effect of including the experience of other populations, when already separately projecting non-smoking and smoking-related mortality, is not big, and leads to a higher life expectancy in the long run. Clearly, different projection methods and choices lead to different projection outcomes. The use of coherent projections and separate projections for non-smoking-related mortality and smoking-related mortality gives rise to more stable results and for the Netherlands higher levels of life expectancy in 2050.

  See extended abstract

Presented in Session 71: Mortality measures and models