Discounting of future costs and benefits is used ubiquitously in evaluation of both private investments and public projects. Though highly technical in practice, the technique would seem (at least a first glance) to be based in a simple, commonplace, and almost undeniable empirical observation. Given a choice between being handed a hundred dollars today and hundred dollars a year or two from now, wouldn't you rather just take the hundred right now? After all, even if you don't have any immediate use for the money, you can always invest it, so that after a year to two it will be worth at historic rates of return $110 or even $120.
But when applied to very long time frames and large complicated situations with many unknowns or hard-to-knows, the subject of discounting tends to trip up and befuddle the greatest minds. Take climate policy, where the proper approach to discounting has become immensely controversial. In a recent New York Review of Books article, Freeman Dyson approvingly discusses recent work by Yale economist William Nordhaus while criticizing Sir Nicholas Stern, lead author of the British government's monumental 2007 review of climate policy.
Rather inexplicably, Dyson says incorrectly that Stern "rejects the idea of discounting future costs and benefits when they are compared with present costs and benefits." Actually Stern stands accused not of that but of employing an excessively low discount rate, so that future benefits accruing from costly efforts to prevent climate damage appear bigger in present-day terms than they really should. (The higher the discount rate--the closer it gets to the normal long-term rate of return expected on investments--the smaller future benefits will be relative to current expenditures.)
In another recent article on this thorny subject, Oxford University ethicist John Broome correctly juxtaposes Stern's preferred 1.4 percent discount rate with Nordhaus's 6 percent, nicely graphing the implications. But having done that, Broome seems to provide a philosophically incomplete account of discounting, and takes the reader into a thicket of ethical complications and conundrums where we'd really prefer to have a path cleared.
According to an authoritative treatment, the idea of discounting goes back two hundred years, to economists writing soon after Adam Smith. As it was elaborated in the following century and a half, mainly by economists in a noted "Austrian school," it came to have two main components: "time preference" (that is, our predisposition to take our pleasure now and put off pain), and diminishing marginal utility (since we'll be richer in the future, added goods will have lesser proportional value). In 1937, the young Paul Samuelson--the economist who grounded the whole field in advanced mathematics, turning it into a quantitative science--published a paper in which he gathered the aspects of choosing between present and future values into one rather simple formula, which carried the day. Though Samuelson had reservations about his procedure, which our source says further research would validate, his modus operandi was so simple and elegant it was irresistible, so that it became the standard for almost all cost-benefit analysis.
Broome, writing in the June issue of Scientific American, for some reason leaves pure time preference out of his account of discounting--perhaps he reasons that personal feelings about present and future happiness have no philosophic standing? Instead he focuses strictly on issues of marginal utility, arguing in essence that the well-being of future generations should not be highly discounted relative to our well-being. That attitude is consistent with Stern's but leads, Broome himself concedes in a sidebar, to some bizarre considerations:
"If humanity become extinct or the human population collapses [as a result of climate change], vast number of people who would otherwise have existed will not in fact exist. The absence of so much potential humanity seems an overwhelmingly bad thing. But that is puzzling. If nonexistence is a harm, it is a harm suffered by nobody, since there is nobody who does not exist."
Do we really need to go there? Perhaps not. Last year, in a respectful but critical review of the Stern report, Harvard economist Martin L. Weitzman suggested that some of its key conclusions--its calls for rather aggressive and expensive actions to constrain greenhouse gas emissions now--could be better justified in terms of an insurance argument rather than the usual Samuelsonian analysis in which all projected costs and benefits are calculated to reveal an optimal consumption path.
Now, in an even more technical followup paper, Weitzman argues (if I'm following him correctly) that in situations involving very improbable but distinctly possible catastrophes, standard cost-benefit analysis is crippled not only by disagreements about discounting but also deep uncertainties in the assessment of consequences as such. That is, if we can't really know how much damage could result from atmospheric greenhouse concentrations being (say) more than twice what they were before the industrial revolution began, how can we even begin to assess the net present value of that damage?
Weitzman's work appears to be a mathematical elaboration of arguments put forth several years ago by the conservative Chicago jurist Richard Posner, who argued (1) that it would be worth taking expensive action to reduce the probability of even a very improbable catastrophe, if the catastrophe is big and bad enough; and (2) that uncertainties about the catastrophe's likelihood and effects, rather than undermining the case for action, support it. That is, the higher the probability that climate change might be less severe than generally expected, by symmetrical statistical reasoning, the higher the probability it might also be even worse.