A Comparative Analysis of the Accuracy of the United Nations' Population Projections for Six Southeast Asian Countries

Khan, H.T.A. (2003). A Comparative Analysis of the Accuracy of the United Nations' Population Projections for Six Southeast Asian Countries. IIASA Interim Report. IIASA, Laxenburg, Austria: IR-03-015

Preview

Text IR-03-015.pdf Download (158kB) | Preview

Abstract

This report analyzes the accuracy of the United Nations' population forecasts in the past, based on six Southeast Asian countries: Indonesia, Malaysia, Singapore, Philippines, Thailand, and Vietnam. The study uses available projected and estimated age-structured data published by the UN from 1950 onwards. An exploratory data analysis has been carried out to examine the accuracy of the UN forecasts. The study reveals that there is heterogeneity in the accuracy of the UN projections for different countries and the errors are age specific. For example, large errors in forecasts of age structures have been found for both the younger (0-4 years) and the older (70 years and over) cohorts for each country. However, the magnitude of errors becomes narrow with a shorter projection horizon. The analysis also shows that gradual improvement in the accuracy of projections occurs over time. The heterogeneity in error is due to the wrong assumptions made in various past projections; thus, the decomposition of the total errors provides us with interesting scenarios about the base (population) error as well as the change error (or unknown error). It has been found that, generally, the base error and the total error have consistently been decreasing over time. On the other hand, the change error does not follow any particular increasing or decreasing path. Until recently, much was unknown about the causes of the change error in forecasting; the determinants are very important to demographers in order to improve the overall accuracy of population forecasting. In short, the main findings are: i) age-specific errors are inconsistent in sign; ii) there has been a gradual improvement in the accuracy of forecasts; and iii) this increase in accuracy is due to improvements in jump-off errors, not to the forecasts of change. Finally, the present study identifies some reasonable causes of errors and makes policy suggestions.

Actions (login required)

View Item