Over the last ten years the concept of generalised cost has been gaining an increasing hold in both the theory and practice of transport planning, particularly in Britain and Australia; and in some circles it is now even beginning to be treated with awe as a major break-through with far reaching implications over a much broader area of economic analysis and public sector decision making. This article challenges the current orthodoxy, and suggests that the approach suffers from a number of major theoretical and practical weaknesses: principally that it gives an inaccurate and misleading representation of the interaction between variables relevant to a decision problem, providing little confidence that it can be of much use as a predictive tool; that it has no justification in theories of consumer behaviour, contrary to what is often supposed; and that there are considerable confusions set up by the British Government's recommended approach to evaluation and to the choice of measuring units.
Most of these faults stem from falling into "index number traps" which involve reading into indices extra properties which they were not given when constructed, and which are often inconsistent with the basic principles on which they were designed. Five of these traps are discussed: the generalised cost index itself, together with the associated indices of elasticity, utility, time value and consumers' surplus.
Suggestions are made not only about the kind of procedures that should replace the inappropriate generalised cost methodology, but also about how the current British practice should in any event be modified so as to eliminate as many as possible of its deficiencies.
The article gives only the author's personal views. He is indebted to Phil Goodwin for the several invaluable discussions that stimulated its preparation and to him, David Lewis, David Quarmby and Elyot Turner for helpful comments on earlier drafts.