Back to the Drawing Board : Economics, Discounting

Cost-Benefit Analysis in Light of the Non-Identity Problem

Donny with an A — Tue, 19 Feb 2008 17:48:00 GMT

[Cross-posted on the parent blog]

So earlier I wrote about the role played by discounting in doing cost-benefit analyses on the impacts of climate change. I concluded that discounting of future damage is unethical because it treats future people as if their interests matter less than present people's. But recently, I've also been discussing the implications of the Non-Identity Problem, and it should be clear that cost-benefit analysis needs to explain its relevance in light of this problem.

For those who haven't been paying attention (or have only recently begun seeing my blog at its spiffy new alternative location), I explained the relevance of the Non-Identity Problem like this:

If we were to act to prevent or mitigate climate change, we would bring it about that people would spend their money on different things, travel to different places, meet different people, get different jobs, and most importantly, have different children (just think how tiny are the chances of a particular spermatozoon fertilizing a particular egg!). In 100 years, it's likely (if not certain) that the world would be populated by an entirely different set of people.

As a consequence of this "fact" (I will accept it as one), we are pretty much forced to say that the people who inherit a world affected by climate change are no worse off than they could have been, because if we had caused less climate change, they wouldn't have existed. Accordingly, it seems difficult to see how we could say that climate change "harms" anyone; if we did anything differently "to" them, they'd simply not exist.

So if the people who would face climate change will be different people than the ones who would have existed if we didn't cause climate change, how can we reasonably talk about costs being incurred as a result of climate change? It seems like when we talk about costs, we do rely on some sort of counterfactual, based on what would have happened if the event in question hadn't happened. For example, let's say I'm talking about a cost imposed on me by a car accident. What I have in mind is that there is a difference between what actually happened to me and what would have happened to me if the accident hadn't happened.

And when we talk about costs imposed by climate change, it seems like we're using the same sort of thinking: the costs imposed by climate change represent the difference between what happens to people in a climate change scenario, and what would have happened to them in the absence of climate change. But as I've said, what would happen to them in the absence of climate change is that they wouldn't exist. So how can we say that a cost has been imposed?

It's my view that this is actually not a problem for cost-benefit analyses at all. When we talk about what would have happened if a particular event had not occurred, I don't think it's necessary that it would actually have been possible for the event not to have occurred. I might say, "What costs and benefits did I incur as a result of being born male instead of female?" I couldn't have been born female; if my parents had a female child, it wouldn't have been me. But I still think we can ask such a question without speaking utter gibberish.

Some might be quick to point out that doing so would involve a lot of serious difficulties, because we'd have to hypothesize exactly what kind of life "I" would have lived, and we'd need to somehow compare that life to the one I already have. In the same way, it's extremely difficult to establish what someone's life would have been like if climate change hadn't affected them, and probably harder still to compare that hypothetical life to the one that actually happens. But it's important to see that this problem isn't confined to situations characterized by the Non-Identity Problem. The same kind of difficulties seem to be present when we ask, "What costs and benefits did I incur as a result of majoring in philosophy?" And it seems to me that any cost-benefit analysis is going to have to face these problems.

So back to the real question: does the Non-Identity Problem create any new problems for cost-benefit analysis? It does if we think of costs as representing harmful deviations from alternative possibilities. As I pointed out earlier, the concept of harm seems to include the idea of being moved away from a baseline, and the sort of baseline we'd need to refer to here is one where the individual couldn't possibly be on the baseline. If you couldn't exist if certain things didn't happen, then it's hard to see why we would say that you're harmed by their happening. But costs don't need to be thought of as harmful to people. As I alluded to earlier, I wouldn't want to say that I was harmed by being born a male instead of a female. My being male seems to be a necessary condition for my existence. But I can still try to determine what costs being a male has imposed on me.

So the fact that we can't consider the costs involved in future cost-benefit calculations to be harmful doesn't prevent us from being able to conduct the cost-benefit analysis. But one thing we have to keep in mind is whether the costs that we'd be measuring have any ethical significance. I want to think more about that, so I'll stop here.

Does the Fact that Individuals Discount Entail the Existence of a Social Discount Rate?

Donny with an A — Fri, 15 Feb 2008 22:09:00 GMT

[Cross-posted on the parent blog]

In my last post, I discussed the idea of discounting as it relates to cost-benefit analysis. I reached the conclusion that discounting treats future people's interests as if they were less significant than our own, and that if cost-benefit analysis aims to make people the best off, then this seems like a bad practice. I received a reply from a fellow with the handle of TokyoTom, which said the following:

Donny, I don't think that you've at all demonstrated that we don't discount - viz., that we try to make decisions on the basis that the preferences of people who do not exist today should weigh as much as our own.

I would disagree with that conclusion myself. Clearly individuals act on the basis of their own preferences, which preferences may take into consideration the supposed preferences of others, including future generations. These others simply don't have a vote on what my preferences are - and is the collective actions of billions of individuals alive today that similarly make decisions that bring about tomorrow.

Tom (at least I assume his name is Tom) is absolutely right to say that individuals clearly act as though value in the future is worth less than the equivalent value today. If I were trying to argue that people actually do make decisions as if future people matter just as much as they do, I would be easily refuted. In fact, I would be hard pressed to believe even that people behave as though future people matter very much at all, never mind as though their interests were equal to their own.

But I never argued that individuals don't discount (in fact, I specifically acknowledged that they do), or that individuals consider future individuals to be just as important as themselves. Rather, I argued that discounting future damage in cost-benefit analysis is unjust. What's the difference? I'll try to illustrate with a series of examples.

Let's say that we're trying to decide whether to put a garbage dump in a neighborhood populated exclusively by an ethnic minority (say, Hmong folks). We perform a cost-benefit analysis to see what we should do. In the first scenario, let's say the Hmong folks in the neighborhood would prefer not to have the garbage dump in their neighborhood, and the folks who live outside of the neighborhood would prefer to have it there (not because of any malice, but rather because they would gain use from it). If (once we equalize for different valuation of money and all that) the cost-benefit analysis shows that the outsiders would be willing to pay more to have the dump than the Hmong folks would to not have it, then we'd say that there's a net benefit to putting the dump in; it's worth doing. And as far as we ignore all the problems with cost-benefit analysis (that is, we don't care what we do to the Hmong people as long as it represents a net gain, and we're okay with treating a single metric as properly representing the wellbeing of these people), then that's all there is to it. The cost-benefit analysis has worked exactly as advertised.

But now let's say that the outsiders didn't want the garbage dump because they would benefit from it, but rather because they're evil hillbillies and they despise the Hmong people. The benefit to them is not a self-interested benefit, but rather a benefit derived from the cost to others. Perhaps if we give this kind of benefit equal standing, the garbage dump goes in. But that seems like the wrong conclusion. We might say the same if the garbage dump doesn't go in because the Hmong people don't want the outsiders to get any benefit, even though they wouldn't really mind the dump being there. That's why most people who advocate cost-benefit analysis try really hard to ensure that the costs and benefits they're measuring reflect only the costs and benefits to the individuals they're surveying.

Accordingly, we wouldn't want to say that the importance of future individuals' wellbeing can be accounted for in cost-benefit analysis by seeing how present people value their wellbeing. What matters is how much they value their wellbeing. Once we recognize this, then it becomes clear what we do when we discount their costs and benefits compared to current people's costs and benefits. What we do is to say that their costs and benefits are less significant than those of present people. And it is this practice which I claim to be unjust.

Cost-Benefit Analysis, Discounting, and Climate Change

Donny with an A — Fri, 15 Feb 2008 04:29:00 GMT

[Cross-posted on the parent blog]

I wrote a paper last semester on the notion of discounting future damage (I'll explain what this means below), and I wanted to revisit the issue now that I've done a little more research, to see if I still agree with what I wrote then. Basically, my paper examined how our views of the proper role of discounting are dependent on our views about what social policy is trying to achieve, and what kind of problem climate change poses. Rather than putting my whole paper online and critiquing it, I'm going to split it up into pieces and post each separately. In my paper I examined four paradigms: (1) The goal of social policy should be to allocate resources to their most efficient uses, and climate change represents a challenge to accomplish this task in a changing world; (2) The goal of social policy should be to maximize the overall good, and climate change represents an obstacle in the way of achieving this goal; (3) Climate change represents an externality, and the goal of a climate policy should be to internalize the externalized costs; (4) Climate change represents an overenclosure of the commons, and the goal of a climate policy should be to remedy this injustice. In this post, I will first go over what I mean by "discounting future damage," and then I will address the first paradigm listed above.

So how does discounting play into discussions about climate change? The most significant impacts of climate change will not occur for a significant amount of time: we're talking decades or even centuries. The issue is how important that damage is compared to the equivalent amount of damage today. In his essay, "Global Climate Change: A Challenge to Policy," Kenneth Arrow wrote that the dispute "...surrounds the appropriate value for the social rate of time preference. This...allows for discounting the future simply because it is the future, even if future generations were no better off than we are. The Stern Review [a report released by economist Nicholas Stern discussing the effects of global climate change on the world economy] follows a considerable tradition among British economists and many philosophers against discounting for pure futurity. Most economists take pure time preference as obvious." So when we talk about discounting future damage, what we're concerned with is whether or not it's acceptable to treat future damage as being less important, just because it's going to occur in the future.

So with that in mind, let's look at the paradigm of cost-benefit analysis: policy should allocate social resources in the most efficient manner, and climate change just represents a challenge for doing that. In its most rudimentary form, cost-benefit analysis is a tool which allows decision makers to allocate resources in the way that best matches some relevant set of preferences. For social decision makers, the relevant set of preferences would clearly be those of society as a whole. Since groups are composed of individuals, advocates of the cost-benefit approach feel that it is reasonable to extrapolate society's preferences from the preferences of individuals. This view is implicit in the position taken by economist Jerry Taylor, who favors discounting future damage at a rate of 5% per year, because it "...matches the return on Treasury bills - or, put another way, [it is] the figure people apply themselves when considering the value of money today versus the value of money tomorrow."

Because the simple cost-benefit perspective considers society as if it were a single decision maker, needing only to allocate its own resources according to its preferences, it is immediately clear why discounting would seem obvious. The existence of a preference for value sooner rather than later is a basic economic assumption which is rooted in cold empirical fact. From this mindset, the question is not whether to use a discount rate, rather what discount rate to use. Some, like Jerry Taylor, use the discounting practices of the current marketplace. Others, like economists Richard Newell and William Pizer, try to predict how market discounting practices will vary over the discounting period, suggesting a plausible range of 2-7%. But to debate the validity of using discounting practices at all would be like asking a banker whether she thought she should charge interest on a loan, or asking an investor whether he cared about getting a return on his money.

So if we accept the view sketched above, it's clear that discounting is not only acceptable, but almost obvious. But what should we think of this view? I want to offer a few objections. First, cost-benefit analysis doesn't properly account for the individuality of its subjects, and does not take into consideration the idea that individuals should not be sacrificed for the sake of others. Second, cost-benefit analysis supposes that all harms can be quantified according to a single metric, which doesn't seem right. Third, even if we ignore the first two problems, it seems like discounting is problematic when you consider the goals of cost-benefit analysis. Let me flesh these out a little.

The first objection is basically taken from Anarchy, State, and Utopia, where Nozick writes, "...there is no social entity with a good that undergoes a sacrifice for its own good. There are only individual people, with their own individual lives. Using one for the the benefit of others, uses him and benefits the others. Nothing more. What happens is something is done to him for the sake of others. Talk of an overall social good covers this up...To use a person in this way does not sufficiently respect and take account of the fact that he is a separate person, that his is the only life he has. He does not get some overbalancing good for his sacrifice, and no one is entitled to force this upon him..." I think Nozick is absolutely right here; we can't weigh future people's interests and current people's interests as if they were all held by the same person. Some notion of proper respect for each group as ends in themselves seems necessary, and the paradigm discussed here clearly lacks that.

The second objection, that a single metric is a suspicious way to evaluate wellbeing, is taken from an essay, "Values in the Economics of Climate Change," where Michael Toman wrote, "One other critique of climate change economics as a guide to policy involves the use of a single-dimension new benefit measure for evaluating different outcomes. This reflects the standard assumption in economics that all costs and benefits are commensurable and interchangeable once expressed in a common metric (a monetary metric as a representation of unobservable utility). There may be serious measurement problems in implementing such a reductionist metric, but as a concept the notion of full tradeoffs and thus full potential compensability of losses from climate change is ubiquitous in the economic model. This view differs from alternatives that see different kinds of values as less commensurable, e.g., some losses of natural beauty or function simply cannot be compensated by other welfare gains." Personally, I tend to think that these latter kinds of views are probably closer to being right. For example, if the Hindus of India are forced to abandon the Ganges as a result of climate change, what kind of compensation could we reasonably expect them to be satisfied with?

But even if we ignored the fact that the cost-benefit model is ethically suspect, and that comparing every harm according to the same metric is methodologically suspect (never mind the fact that we could probably never conduct the kind of calculation necessary), there would still be another problem. The third objection arises from the fact that calculations of "costs and benefits" are supposed to reflect utility, and therefore social preferences. The problem is that, as we discussed earlier, the cost-benefit model is perfectly comfortable with the idea of discounting. In his essay, "Environmental Risk, Uncertainty and Intergenerational Ethics," Kristian Skagen Ekeli pointed out that "To discount the future implies that current interests and preferences count for more than those of future generations." When we say that future damage should be discounted, what we're basically saying is that "society," which is supposedly neutral between its individual members, prefers current people to be happy over future people, simply because they live earlier. How this makes sense is beyond me. It seems that if we were trying to allocate resources to impartially reflect their most efficient uses, we would need to weigh people's interests as being equally significant.

So hopefully those objections demonstrate two things. The first is that cost-benefit analysis is a really crappy way to deal with the issue of climate change. But if we use it anyway (which I suspect people will do, because that's how economics is done nowadays), then we shouldn't discount future damage. To do so would treat future people as if they mattered less than present people, and that seems obviously unacceptable. I am, of course, conspicuously ignoring the Non-Identity Problem completely, and I want to deal with that issue, but I guess I'll leave that for later.