Mortality forecasting in the context of non-linear past mortality trends: an evaluation

Having accurate and high-quality mortality forecasts has become increasingly important due to the general increase in life expectancy and the social consequences of this.The aim of this dissertation was to evaluate mortality forecasting in the context of non-linear past mortality trends. In this way, it contributes to the debate on the degree of subjectivity in mortality forecasting, but also to the evaluation, validation, and further improvement of the mortality forecasts of Statistics Netherlands (CBS).When making mortality forecasts, different approaches and assumptions are used, in which additional, often subjective, information is included to deal with non-linear mortality trends. A careful examination of past trends, and a careful assessment of the pros and cons of including additional information, is therefore important. The mortality forecast in the context of non-linear past mortality trends may be improved by making explicit adjustments for the distorting effects of smoking on mortality trends and the use of mortality developments in other countries. The specific choices that are explicitly stated in a method proved more important than the choice of the forecasting approach.The methods used by CBS to include mortality trends from other countries in the forecast and to project smoking-attributable mortality, proved valid. The matching of the forecast to recent observations has been adjusted.Both the mortality developments themselves and the underlying factors, our knowledge hereof and the mortality forecasting methodology are constantly changing. Flexibility in both making mortality forecasts and interpreting his outcomes is therefore required