Frontiers of Population Forecasting

Most users of demographic data in government and industry require population forecasts-soundly based information on future population trends at the local, national, or global level. In the eyes of the public this is a principal justification for the discipline of demography And increasingly, sophisticated users expect not a single "best guess" of a population trajectory - perhaps straddled by "high" and "low" variants - but a range of demographic futures with associated indications of uncertainty, or even formal confidence intervals. Or they may want full-fledged scenarios of the development of the demographic system in different economic or environmental circumstances. Many population agencies, on the other hand, continue to generate population "projections" by methods that have been virtually unchanged for decades, with no assessment of probability and often claiming no ambition to predict. It is time to reexamine the procedures of population forecasting and to respond to these emerging demands by users. The nine chapters in this volume take on this task. The key issues they address include: What population characteristics beyond the standard variables of age and sex should routinely enter population forecasts? When should forecasts take account of economic or environmental feedbacks? How is forecasting accuracy to be assessed and what is the past record? What is the state of the art of stochastic time series modeling of population change? How can users cope with probability distributions? What scope is there for application of methods to incorporate expert opinion into population forecasting? Recent years have seen substantial advances in forecasting methods. These are beginning to be applied in population. For end users of forecasts who are familiar mainly with UN or similar population projection series and for the many professional demographers whose knowledge of projections has not progressed much since graduate, school, this volume opens a window on significant new developments in this field