Assessing the uncertainty in populationprojections : a test based on the12th coordinated population projection for Germany

In this paper I use variance based sensitivity analysis to investigate the uncertainty in projections of future population arising from the assumptions taken. Instead of for a limited number of scenarios the uncertainty is explored within the range of assumptions for the model parameters through sampling. To this purpose, the cohort component method and the assumptions taken in 12th coordinated population projection for Germany have been reengineered in a sensitivity analysis setting. This allows attributing the uncertainty in the outputs of the demographic model (e.g. size of the population) to the uncertainty in the model inputs (e.g. total fertility rates).