Mortality Rate Forecasting Project

In recent years, the rapid aging of the population has been a growing concern for governments and societies. Concerns in many developed counties are concentrated on the substantiality of pensions, health and aged care systems, given the fall of mortality rates. This results in a surge of interest in modeling and forecasting age-specific mortality rates accurately for government policy and planning. Any improvements in mortality forecasting have an immediate impact in guiding policy decisions regarding of the allocation of current and future resources. In particular, future mortality rates are of great interest to the insurance and pensions industries.

Mortality Rate Forecasting Project investigates a novel functional time series method, namely weighted functional principal component regression, to model and forecast mortality rates. This project will comparatively examine the point forecast and distributional forecast performance of this newly proposed approach with nine existing methods used in demographics. Age- and sex-specific populations from 18 developed countries (in total 36 data sets) will be used for this study.

Within this project, there are ten algorithms to implement, and each can be time consuming. A rough estimation reveals that the computing time for 10 algorithms can take up to 20 hours, even if only one data set is used. With the computational power of Monash's high performance computing (HPC) facility, the researchers are able to process such large amounts of data sets simultaneously, with a significantly increased computing speed.