Abstract:
The idea is developed of using in computation for input-output models of a range of unit characteristics averaged over sectors rather than their unique, most probable values, as done in conventional forecasting methods. In choosing the range use is made not only of statistic data on the behavior of the process in the past but also of the forecast values of aggregated indices obtained by expert evaluation. The relations of the resultant model over the entire forecast span are to satisfy simultaneity requirements and the scatter of extreme values of the performance criterion or other forecast indices must be minimal.