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

In this paper, we illustrate the application of Bayesian time series models to obtain future population estimates with uncertainty. The focus is on a simple projection model with the historical data representing population change in England and Wales from 1841 to 2007. The Bayesian forecasts to the year 2033 are compared to those produced using classical methods. The former are also extended to include stochastic volatility and model averaging. We show that the Bayesian approach is adept in capturing multiple sources of uncertainty in population projections, including model model uncertainty. The inclusion of stochastic volatility improves the fit of the models and reduces the predictive uncertainty levels. The paper ends with a discussion and ideas for further research.