Estimating the Uncertainty in Population Projections by Resampling Methods

This paper proposes a new approach to introducing quantitatively measured uncertainty into population projections. It is to a lesser degree based on past time-series than other approaches, since it uses random walk models for migration, mortality and fertility, for which upper and lower bounds are defined. No parametric distribution is fitted to the observations, but the random walk is resampled from the past data. By putting bounds on the level that fertility can reach in the future, further substantive information is introduced that transcends the information derived from the observed time series. By sampling 10.000 path of the random walks in fertility, mortality and migration, the distributions of population size and structure up to 2050 for Austria, Mauritius and USA are estimated.

VIEW CONTENT