Modelling and projecting mortality improvement rates using a cohort perspective

- In this study, we consider how to model and project mortality rates.
- We take a different view from the literature.
- We consider the mortality improvement rates for cohorts and so focus on the underlying trend.
- This is an alternative to the conventional approach which models directly the mortality rates over time.
- We set up the modelling framework and then compare results from the different approaches using a case study.

We investigate the feasibility of defining, modelling and projecting of (scaled) mortality improvement rates along cohort years-of-birth, that is, using a cohort perspective. This is in contrast to the approach in the literature which has considered mortality improvement rates that are defined by reference to changes in mortality rates over successive calendar years, that is, using a period perspective. In this paper, we offer a comparison of the 2 parallel approaches to modelling and forecasting using mortality improvement rates. Comparisons of simulated life expectancy and annuity value predictions (mainly by the cohort method) using the England & Wales population mortality experiences for males and females under a variety of controlled data trimming exercises are presented and comparisons are also made between the parallel cohort and period based approaches.