Population forecasts, public policy, and risk

Many aspects of public policy require assumptions about the course of future population. As examples, social security and pensions must meet the needs of future retirees, private and public health care must provide services to a population whose age composition and medical needs change over time. For such matters, an essential role in analysing alternative policies is played by forecasts of future population composition and numbers. In principle, accurate prediction follows from accurate scientific understanding of the dynamics of population. But the science of population and other complex systems such as economy and environment – subjects at the core of policy – is constrained by scale and complexity, and is certainly inadequate for accurate long-term prediction. Policy makers, policy analysts, and forecasters cannot and do not wait for scientific progress. In practice, forecasters rely on a combination of scientific knowledge, historical analysis, and expert judgement. In these circumstances, it is hardly surprising that point forecasts of population over a time horizon of a few decades or longer have often been wrong. Conventional high-medium-low scenarios are intended to do better by indicating a range of outcomes, but even these have limited value because they do not include an explicit assignment of the relative likelihood of the predicted alternatives