Over the past 25 years, significant improvement has been made in demographic forecasting. This study reviews major developments in population forecasting since 1990, with focus on time-series methods. Using the Human Mortality Database and Human Fertility Database, this study also compares point forecast accuracy of forecasting methods. The results indicate that the performance of forecasting methods depends on many factors. It appears that it is not constructive to pursue a single best demographic forecasting method. Complementary combinations of different approaches, such as statistical models and expert knowledge on future demographic change, would be recommended to improve the utility of population forecasting.
Further, this study also indicates the importance of understanding and measuring uncertainty surrounding forecasting.