Forecasting educational differences in life expectancy

Abstract

Forecasts of life expectancy (LE) have fueled debates about the sustainability and solidarity of pension and health care systems. However, within populations there are large and growing inequalities in life expectancy between high and low educated groups that are relevant for these debates. In this paper, we present an approach to forecast LE for different educational groups within a population. As a basic framework we will use the Li-Lee model which has been developed as to coherently forecast mortality for different groups. We adapted the Li-Lee model to distinguish between overall, gender specific and education specific trends in mortality and extrapolated the time-trends in a flexible manner. We will illustrate our method for the population above age 65 in the Netherlands. Several data sources spanning different time windows were used to construct time series of mortality by gender, age and education. Our extended Li-Lee model was used to forecast mortality rates and translate them into estimates of LE. Results suggest that LE is likely to increase for all educational groups for both men and women in the Netherlands but that differences in LE between educational classes widen. Several sensitivity analyses illustrate advantages of our proposed methodology.