The forecast accuracy of local government area population projections. A case study of Queensland

Abstract

Users of population projections tend to assume that they are provided accurate predictions of future demographic trends. However, there is limited evidence to either support or refute this assumption because almost no research has evaluated the accuracy of past projections in Australia. This paper assesses the forecast accuracy of seven previous rounds of local government area population projections for Queensland. The analysis reveals errors to be quite large in absolute terms and in relation to State forecast errors, but respectable compared to those reported in other studies. Relative to simple extrapolative forecasts, the official Queensland projections are shown to have performed quite well. Fractional response models are employed to determine the extent to which forecast errors can be predicted on the basis of local area characteristics. The concluding section suggests ways in which forecast error might be reduced and how users can be informed about the possible magnitude of error in current projections.

Francisco Rowe

Professor of Population Data Science

My research interests include human mobility and migration; economic geography and spatial inequality; geographic data science.