Ex-ante assessments of the accuracy of forecasts (cont.)
The third approach to quantifying the precision of forecasts is to use expert opinion to try and narrow the horn of uncertainty. The process involved is this:
- An individual or expert group is asked to come up with central, maximum and minimum plausible forecasts of vital rates. The range between the upper and lower bounds that is obtained is usually narrower than that obtained from the statistical extrapolation of past trends. This can be done by the Delphi method, in which a panel of experts makes successive rounds of forecasts anonymously until they converge on an agreed answer.
- These judgement-based 'confidence intervals' for the various input parameters of the projection can then be used to produce confidence intervals for the future population. This is done by repeated projection of the population using either varying combinations of assumptions or random draws from the uncertainty distributions for the various inputs of the type used in the purely statistical approach.
- In essence, this approach attempts to use expert knowledge to improve forecasts without formally incorporating that understanding of the causal processes involved into the forecasting model.
- The main concern about the approach is that expert opinion may be hopelessly divided or overly conservative. For example, no scientific consensus exists as to whether Western countries are near the upper limit of life expectancy for the human species. Moreover, many demographers doubt that our understanding of fertility in these countries is such that we can confidently predict, in the medium term, that total fertility could neither plunge to below one child per woman nor rise again to well above replacement level.