Bayesian Probabilistic Projections of Life Expectancy for All Countries

Abstract

We propose a Bayesian hierarchical model for producing probabilistic forecasts of male period life expectancy at birth for all the countries of the world from the present to 2100. Such forecasts would be an input to the production of probabilistic population projections for all countries, which is currently being considered by the United Nations. To evaluate the method, we did an out-of-sample cross-validation experiment, fitting the model to the data from 1950–1995, and using the estimated model to forecast for the subsequent ten years. The ten-year predictions had a mean absolute error of about 1 year, about 40% less than the current UN methodology. The probabilistic forecasts were calibrated, in the sense that (for example) the 80% prediction intervals contained the truth about 80% of the time. We illustrate our method with results from Madagascar (a typical country with steadily improving life expectancy), Latvia (a country that has had a mortality crisis), and Japan (a leading country). We also show aggregated results for South Asia, a region with eight countries. Free publicly available software packages called and are available to implement the method.

Future inequalities in life expectancy in England and Wales – Authors' reply

“…For this reason, rigorous validation of forecasting models should be based on holding back data and measuring the bias and error of the forecasts relative to the withheld data, as we and others have done in forecasting studies. 6 Supplementary figure 1 and tables 2-4 of our Article show that the chosen model best predicts data over the entire withheld period at the national as well as district levels, with virtually no bias.…”

Bayesian probabilistic population projections for all countries

“…We generated probabilistic projections of the period female life expectancy at birth using the Bayesian hierarchical model of Raftery et al (26). The model is described in more detail in the SI Text, where we also specify the prior parameter settings we used, which are given in Table S1.…”

Integrating uncertainty in time series population forecasts: An illustration using a simple projection model

“…These extensions, which we are currently investigating, are likely to reduce the uncertainty of population forecasts in comparison to those presented in this paper. Here, the recent work on modelling fertility (Tuljapurkar 1999;Heilig et al 2010;Alkema et al 2011Alkema et al , 2012, mortality (Pedroza 2006;Girosi and King 2008;Raftery et al 2013) and migration (Brierley et al 2008;Bijak 2010; Raymer, J., Wiśniowski, A., Forster, J.J., Smith, P.W.F. and Bijak, J.…”