Forecasting Changes in Mortality: A Search for a Law of Causes and Effects

In this article, we express a concern that certain commonly accepted methods of predicting mortality will likely prove to be inadequate in the future. Specifically, the Lee-Carter method, which overall has been empirically successful, uses auto-regressive moving average (ARMA) technology and contains no structural mortality equation. This structure means that no information other than previous history can be introduced. We argue that rapid advances in medical science are taking place and that failure to reflect this information in our projection methodology will make resulting projections unsuitable.