Evaluating the Accuracy of Population Forecasts

Abstract

In this paper the accuracy of population forecasts is discussed. Various papers on errors of population forecasting are reviewed and summarized. The results are stated in six theses. The main findings show that no clear dominance of any one forecasting method can be determined, that the logarithmic forecast errors are more or less independent of the length of the forecast horizon, and that saturation models underestimate the population development in the long term, whereas geometric and polynomial trend models overestimate it. Finally, the accuracy of population forecasts is compared with the accuracy of short-term and long-term economic forecasts. It is found that the error of population forecasts is smaller than that of economic forecasts. However, the logarithmic error decreases with the length of the forecast period for most economic variables.