Choice is a value statement. On inferring optimal multiple attribute portfolios from non-optimal nominations

Abstract

Sutton and Green have previously introduced an approach to the analysis of multiple attribute problems based on a data envelopment analysis-like formulation. This required that options should be both discrete and explicitly defined. Here, the concept is extended to cover portfolios, where it is frequently impracticable to pre-specify and describe all available options explicitly. At the same time, the idea of using decision makers' choices as indicators of their in-built values is explored. The approach is illustrated using a multiple attribute share selection example. The proposed method allows somewhat greater use to be made of decision makers' preference information revealed in the selection of well-considered favoured choices, as well as being able to take analytic account of other preference elicitation mechanics. It can be also used for other decision structures where it is not possible to describe candidate choices individually in advance.

Keywords:

• decision analysis
• multiple attribute
• value function
• elicitation of value
• portfolios
• data envelopment analysis