The Use of Decision Analysis in the Public Sector

Abstract

Criticisms of the use of normative models in the analysis of decision making in government and in complex organizations, have usually stressed the practical difficulties of obtaining the required data and subjective inputs. The present paper calls attention to an intrinsic limitation of the normative approach, stemming from the impossibility of deriving the conditions for a second-best policy from those specifying a first-best, but incompletely constrained, solution. The ‘second-best theorem’ of welfare economics is only one particular manifestation of a very general phenomenon.

Constraints represent compact summaries of policy-relevant knowledge, and it is suggested that even when a normative analysis is impossible or appears fruitless, the decision analyst can perform a most important function by organizing and expressing the available knowledge as a set of constraints defining the boundary of the region of feasible policies. Such an approach has indeed been used to forecast the range of likely outcomes for water pollution policies and, more recently, in the field of health policies.