An evaluation of the forecasting performance of the Poisson Common Factor Model (Institute of Actuaries of Australia Grant)
Project description: Most current mortality forecasting methods treat males and females either independently or in aggregate; relatively little attention has been given to forecasting female and male
mortality separately but on a coherent basis. Independent forecasting of sex-specific mortality may produce unrealistic forecast sex differentials, with implications for applications in financial services and public policy. This project will conduct out-of-sample tests of forecast performance of the newly-proposed Poisson common factor model, using data for several countries. The outcome will be a better understanding of the advantages that common factor models offer over current best methods for forecasting mortality for separate but related population subgroups.
Project team: Leonie Tickle, AFAS