Requisite decision modelling: a case study

This case study in decision analysis concerns a company that had to decide between continuing to manufacture an old product that might in the near future by banned by the government or introducing an improved but conventional product that would beat the ban but might lose market share to competing products using microchip technology. A decision tree with three attributes describing the consequences over a ten-year horizon modelled the problem. Implementation on a micro computer facilitated extensive sensitivity analyses, the final round of which were conducted by the Board of Directors. More and more pessimistic assumptions were made until the decision switched from the new to the old product; at that point no Director believed all the assumptions. Thus, agreement was reached about the decision even though the Directors disagreed about the uncertainties. The case illustrates `requisite' rather than optimal decision modelling and shows the essential roles of problem structure and sensitivity analysis.