Michael Nuwer wrote:
As for cost/benefit analysis, interpersonal comparisons of utility are not necessary if the analyst equates marginal benefits with marginal costs, where benefits and costs are measured in terms of shadow prices. Then they have determined how many, say, satellites to build, and they can deduce with their theory that that quantity represents the best of all possible worlds, without measuring the total utility.
I would also add that the assumptions which make this practice possible are not very realistic. But again, I thought most people on this list would already know this. A demand curve shows how much of a commodity someone is willing to consume at different prices. One way to constructed this curve is as a locus of points derived from consumer equilibriums that correspond to different prices. (This is done in all of the textbooks with indifference curves deriving demand curves.) In this interpretation, there is a demand curve and a marginal utility curve which are the same, but they are not identical. That is, one curve lies exactly on top of the other curve. Therefore, in the mainstream story, we can use any information rung out of the demand curve (like market price) to infer information about the marginal utility curve (like marginal value). But this works only if the consumer is "rational," i.e., is a utility maximizer. A negatively sloped demand curve can always be constructed whenever there is a constraint separating what is possible from what is not possible. So, the demand curve of non-rational consumers will slope downward, but this curve can no longer be interpreted as a locus of points representing consumer equilibriums. The demand curve is no longer the same as the marginal value curve and thus the demand curve no longer represents the consumer's willingness to pay for a commodity.