ABSTRACT
More formally, however, the cost-benefit criterion to be adopted can be expressed in simple notation form as Vi > 0 , where V1,V2, . . . ,Vn are the net valuations of each of the n persons affected by the project, where a positive V valuation indicates a net benefit, and a negative V valuation a net loss to the person. Clearly, if the aggregate valuations sum to a positive figure, the aggregate of benefits exceeds the losses, and a potential Pareto improvement is realized. (More precise measures of such valuations in the form of compensating variations will be introduced later.) The above criterion is better regarded as necessary though perhaps not sufficient,
inasmuch as it may have to meet some additional political requirement, say, that V exceed a certain figure or else exceed a given benefit-cost ratio. Another reason why our V > 0 above may be deemed insufficient is that, as
it stands, it makes no provision for the distributional impact of the project. Since a number of ways have been proposed for attaching distributional or other weights to the valuations, none of which, however, we find acceptable, we defer these proposals, and our objections to them, to Chapter 3. In the meantime, although the criterion we have adopted (simply that the sum
of all valuations be positive) is straightforward enough,3 our difficulties begin once we start to trace all the repercussions and bring them into the calculations. These difficulties, which require extended treatment, are to be found chiefly in the concepts and measurement of consumer surplus and rent, in the distinctions between shadow prices and transfer payments, in evaluating a range of spillover effects, in the choice of investment criteria, and in proposals for dealingwith future uncertainty. They are dealt with in that order in the parts that follow.
