Hide/Show Apps

Modelling and forecasting age-segmented mortality: evaluation of lLe-Carter method and its extensions

Download
2019
Alşan Kılıç, Raife Sıda
Mortality dynamics deal with the human mortality from birth to death giving insights for the population in different aspects; age, year, gender and enable to ascertain crucial mortality trends as well. Over the last decades, ample new methods have been developed and the mortality modelling has been evolved into more effective ways. Among these models, the pioneering and the most seminal one is Lee-Carter model (Lee & Carter, 1992). From its development, Lee-Carter model has been intensively studied and its variants with different structures have been proposed. While the performance of mortality models has been examined under various issues so far, the modelling of mortality over age patterns for different stages of human life has yet to be studied. Therefore, in this study, our purpose is to utilise a new approach differing from literature by investigating the performance of piecewise mortality models via structural breakpoints for age dimension. Accordingly, the mortality data of three different countries for the eighty-one years averagely; Australia, Japan and Portugal are examined by Lee-Carter model and its extensions with regards to the age structure of corresponding countries. The results reveal that each country has different age structures for female and male subpopulations and the age intervals within these structures have different characteristics regarding significance of accuracy of these methods. The findings imply the need for a use of different age breaks for female and male subpopulations and the selection of best suited methods for each corresponding age intervals regarding accuracy of mortality forecasts.