Forecasting Life Expectancy in the Presence of Structural Breaks

Frederik Peters, University of Rostock

Since 1950, most developed countries have exhibited structural changes in mortality decline. This complicates extrapolative forecasts that require the presence of a steady long-term trend. This study tests whether the impact of the tobacco epidemic explains the structural changes in mortality decline, as is presumed in earlier studies. For this purpose, sex-specific mortality trends were investigated in 20 developed countries between 1950 and 2009 for possible structural changes. It was found that removing the impact of smoking from mortality trends took away about a third of the trend breaks. For the remaining trend breaks, adjusting for smoking attenuated the degree of change in mortality decline. Therefore, taking the tobacco epidemic into account should become standard procedure in mortality forecasts to avoid a misleading extrapolation of trends. Nevertheless, more research is needed to identify additional factors, such as healthcare policies and innovations in medical treatment, to explain the remaining structural changes.