Recent years have seen an increasing cross-fertilization between the fields of decision analysis and forecasting. Decision-analytic models often require forecasts as inputs, and aspects of the Bayesian decision-theoretic framework underlying decision analysis have proved useful to forecasting, particularly in contexts where subjective judgemental inputs are required. This paper describes the use of decision tree analysis for forecasting and illustrates its use for corporate divisional forecasting and planning. A specialized decision-analytic technique, acts as events, is also described and illustrated to forecast a new product's earnings. Conclusions are drawn about the applicability of decision analysis for forecasting.
KI;Y WOKIX Decision analysis Forecasting Corporate planning
This paper presents and illustrates an adaptation of decision-analytic techniques for forecasting. The adaptation adopts the basic perspective of conventional decision analysis as reflected, for example, in Raiffa (1968). In particular, it calls for a forecast to be presented in personal probability terms (rather than, say, as a maximum likelihood estimator and a classical confidence interval), and can incorporate judgemental as well as observational inputs. However, the approach also makes a significant departure from decision analytic tradition. In particular, an alternative to the practice of using 'rollback' to model subsequent acts in decision trees is proposed.
TRADITIONAL DECISION TREES
Decision tree analysis is the oldest and most widely used form of decision analysis. Managers have used it in making business decisions in uncertain conditions since the late 1950% and its techniques are familiar (see, for example, Brown et al., 1974). Over the past several years, however, the manner in which people conduct decision tree analysis has expanded. Today's analyst has at his or her disposal not only an array of computer supports that make quick turnaround possible but also the accumulated wisdom of analysts over the last twenty years.
One particularly important expansion is the use of the probabilistic modelling aspects of decision trees to develop forecasts of, for instance, future sales and profits, which in turn can be