Mortality models: comparison and application in old-age populations of selected countries

Abstract

This research examined which of the five well-known chosen extrapolative mortality models best captured the trends in old-age population mortality for different age groupings in four different countries. Mortality rates from the Human Mortality Database for the United Kingdom, Poland, Japan and Taiwan were used, encompassing males and females in the 65-89 age group. This allowed assessments to be made across developed and emerging economies, and across Europe and Asia. Comparisons were made across models to understand why some work better for some age groupings in some countries. The research considered the goodness-of-fit of these well-known mortality models to historical population mortality rates, assessed the range of projected future mortality rates, and evaluated the financial impact of mortality uncertainty on annuity prices across the subject populations. Some of the findings which emerged were that the Booth-Maindonald-Smith model tended to work best for most of the selected populations, particularly for female or Asian populations. Perhaps surprisingly, retiring females in the emerging economies can be expected to possibly outlive males in the developed economies selected. In a low yield environment, uncertainty around mortality has a noticeable impact on the range of pricing of annuities. The extent of mortality uncertainty is expected to be less for developed than in emerging economies, and less for females than males.