Ben Trachtenberg writes:
This Article presents two new arguments against “discounting” future human lives during cost-benefit analysis, arguing that even absent ethical objections to the disparate treatment of present and future humanity, the economic calculations of cost-benefit analysis itself – if properly calculated – counsel against discounting lives at anything close to current rates. In other words, even if society sets aside all concerns with the discounting of future generations in principle, current discounting of future human lives cannot be justified even on the discounters’ own terms. First, because cost-benefit analysis has thus far ignored evidence of rising health care expenditures, it underestimates the “willingness to pay” for health and safety that future citizens will likely exhibit, thereby undervaluing their lives. Second, cost-benefit analysis ignores the trend of improved material conditions in developed countries. As time advances, residents of rich countries tend to live better and spend more, meaning that a strict economic monetization of future persons values the lives of our expected descendents above those of present citizens. These two factors justify “inflation” of future lives that would offset, perhaps completely, the discount rate used for human life. Until regulators correct their method of discounting the benefits of saving human lives in the future, the United States will continue to suffer the fatal costs of underregulation, and agencies will remain in violation of legal requirements to maximize net benefits.
I don’t agree with the “underregulation” argument (growth matters too, especially with a low discount rate), but there is much to be said for Ben’s argument. Do note that there are numerous complications in any correct analysis of this problem.
This is related to the issue of whether the future experienced utility of a given individual is properly discounted. Gary Becker says yes. Daniel Kahneman says no. The case for discounting is that “we” just don’t care as much about our future utility. But should we? In one sense, there is no “should” here. From the perspective of economics alone, there is not much to say once the consequences of discounting or not are made clear.
As to the issue of cost-benefit analysis and other lives (or the experience utility of future people), perhaps we ought to care as much about them as they care about us. But seriously, do we really want to treat not-yet existing people better than we treat ourselves? These are ethical and not economic issues.
What should be the precise discount rate, if we agree to discounting? You will never settle this one.
Perhaps it’s just me, but I don’t believe in the moral claims of non-existent entities….
Anyway, with an effectively infinite number of future persons, wouldn’t a utiliitarian with no-discounting ram the optimum savings rate to close to 100% and effectively stuff each current generation to labour solely for the future?
Re the savings rate point, it’s true that this presents something of a dilemma for proponents of the zero rate of pure time preference. Practically, you can ameliorate these issues by appealing to:
(a) intergenerational equity concerns (if the future is going to be richer, egalitarians/prioritarians will be less inclined to redistribute wealth towards them, and be more likely to require them to shoulder parts of the burden for future generations beyond theirs); and
(b) uncertainty that future generations will continue to exist (why sacrifice for potentially non-existent futures.
This isn’t entirely satisfactory, but I don’t think that is related to discounting so much as it is to generic problems with combining consequentialist maximands with infinities.
There is no issue for which the discount rate of future lives matters more than that of abortion. By this argument we are clearly massively under-regulating abortion.
There is no issue for which the discount rate of future lives matters more than that of [my pet issue]. By this argument, we are clearly massively [under/over]-regulating [my pet issue].
Bollocks. We’ve discussed this before, here. (Although half the comments on that post appear to have been lost in the site migration, so this wayback link is probably better if you actually want to follow the discussion. See also, this Crooked Timber thread).
So what exactly is the current discounting rate for human life? And who are the regulators explicitly using such a rate in their analyses. Is there any major regulation that could have gone another way had a different rate been used; i.e. is this just an academic construct or does the human discounting rate actually influence “real” policy decisions?
No. In my experience, cost-benefit analysis influences nothing, except perhaps occasionally regulations for which an agency finds it difficult to quantify any benefits at all. And even then, not always.
The EPA uses the discount rate on human lives in their cost-benefit analysis. It caused a stir when during the Bush Administration, the EPA altered the discount rate such that fewer environmental projects were cost effective. I’m not arguing for or against the decision by Bush’s EPA – just pointing out that it is used and it does have consequences.
The appropriate discount rate is always situation dependent and usually requires some judgment beyond objective factors. Just like a hypothesis test, you can achieve almost any result you want in a cost-benefit analysis by tweeking the assumptions.
The discount rate is crucial in legal judgments when determining the amount of damages for losses of life or abilities. Typically, though, you’re discounting the stream of expected future income rather than the inherent value of human life.
Actuaries for private insurance must determine a discount rate of human life for properly setting premiums. In theory, Social Security and Medicare should be doing the same. Is it worth $2 million to prolong the life of an 80-year old for approximately four more months when the government is paying that $2 million?
Federal agencies commonly use discount rates of 3 percent and 7 percent in their regulatory impact analyses. EPA has used 10 percent for major regulations. See the linked article at page 1324 (and footnote 40).
As for the choice of rate making a real difference, consider the “Benzene case,” discussed at page 1329 (and footnote 54), in which the Supreme Court struck down a regulation (later recognized as totally sensible and reenacted years later) because the Court thought the science wasn’t sufficiently impressive (meaning the benefits were too speculative). It’s not an exact parallel, but it demonstrates that the measure of benefits affects whether regulations become law.
I think the effect is more pronounced (but harder to measure) in regulations that never get published. If an agency’s high discount rate makes a proposed regulation seem to have costs that outweigh the benefits, the public will never see it.
As future humans (hopefully) vastly outnumber current humans, does the lack of a discount rate imply we should divert (say) 50% of GDP to improve their lot by 1%?
Pure utilitarianism is silly.
I’ve always thought about discounting future lives not in terms values but rather information. The reason to discount is that future lives are worth less to us but rather that the farther into the future one looks, the less one can guess what future lives will be like and, therefore, what our descendants would ask of us. What should the people of the 19th, 18th, 17th century have done for us had they been valuing our 21st century lives a highly as their own?
What should the people of the 19th, 18th, 17th century have done for us had they been valuing our 21st century lives a highly as their own?
Do we know empirically if they did or did not?
Well, if they did, there were cheap ways to help us. For example, as Benjamin Franklin did (https://secure.wikimedia.org/wikipedia/en/wiki/Benjamin_Franklin#Death_and_legacy) when he left 1000 pounds to gain interest for a few centuries and turn into >$7 million and pay for schools.
Sparing 1000 pounds in Ben Franklin’s time wasn’t very cheap.
It was a heck of a lot cheaper than sparing >$7m now! http://www.measuringworth.com/ppoweruk/ says that 1000 pounds at his death is about 96k now, or around 153k USD, or *1/46th* of what the sum grew to. If you are a utilitarian who doesn’t temporally discount, by adopting a Franklin strategy your money could be 46 times larger and presumably do 46 times as much good (in addition to the good done over the intervening centuries by not consuming your capital and lending it out).
It wouldn’t even be very hard for ordinary middle-class persons to adopt this strategy: middle-class people routinely have much larger life insurance policies without undergoing any financial hardship.
“What should the people of the 19th, 18th, 17th century have done for us had they been valuing our 21st century lives a highly as their own?”
Getting as many children as they we are able to – and they did. That is why there are some many of us now. That we are having so few children shows that we value 22nd etc century lives less than they did.
This is the critical point. Discounting the future is first and foremost an act of humility. It accepts that our vision is unclear. That the author is apparently decrying a high discount rate as an argument in favor of more regulation is an amusing confirmation.
“We are doing things which might possibly drive humanity extinct in the future. But we should discount this because there is a possibility that it will in fact not drive humanity extinct. Since we don’t know for certain, it isn’t worth worrying about. Just go ahead.”
It’s a peculiar form of humility.
Do me a favor: specify the each of these things which might possibly drive humanity to extinction. Specify the probability for each of these things actually causing our extinction for each of the next twenty decades. Be careful to include error boundaries.
Right Wing Nut, I can’t specify the probability that anything in particular will cause humanity to go extinct.
For example, consider the possibility that nuclear war or reactor accidents might bring the radioactivity level too high for humans to survive. How can anybody estimate the probability of nuclear war over the next 200 years? We don’t even have great estimates about how bad radiation is for you. It’s possible that human beings are designed so that by the time we get too many germ-line mutations for the population to sustain, we go sterile and only people with acceptable genes can have children at all. If that were the case, we could handle higher levels of background radiation than it looks like. If we had some problem that killed or sterilized 80% of the world population one generation, and 70% the next generation, and 40% the generation after that, the long-run effects would be small. Is it that way? We do not know.
Similarly, there are a collection of people who are sure that our background radiation levels are too low. They point to unpublished statistical studies which indicate that in a ring around Hiroshima and around Chernobyl etc, people are healthier than usual. They say low-level radiation up to maybe 10 time the current background or maybe 50 times is actually a significant benefit. If they are right, then a nuclear war would be good for the survivors, assuming it was a small enough war that there would be survivors. Are they right? I do not know. It’s possible they are right that they cannot publish their results because reputable journals censor anything whose conclusions do not fit their prejudices. I tend to believe that people who say that scientific journals censor them are usually unpublishable because their methods are flawed, but they could still be right. Even if their research is wrong, they could be right by accident.
And yet I still believe that humanity would probably be better off without nuclear war and without great big reactor accidents. If we could give up on nuclear technology completely, a serious threat to humanity’s long-run survival would be gone. While using nukes is a long-run risk. Possibly some other risk might arise that would be even worse than nukes, that nukes can help us with. Maybe without nukes the flying-saucer people would show up and slaughter us all for food. I think the risk of nukes is bigger than the risk of flying-saucer people, but it would be hard for me to quantify both of them.
We can discount future people on the assumption that there will be a whole lot of them and they’ll all be very rich, so they don’t need anything from us. But if we drive them extinct, then beyond the time they go extinct their marginal benefit is zero. All those rich multitudes we have assumed will have been killed off, an unknown but presumably large number of generations of them. Maybe a million generations of humans at 20 billion per generation? That’s a whole lot of people to kill off, even if you discount them pretty heavily.
So I say it’s a peculiar form of humility to make big important decisions like for example about nuclear war, on the assumption that future generations don’t matter.
It’s one thing to figure that extinction of humanity is important, to we should be very very careful about likely risks until we have a clear idea how big those risks are.
It’s something else to say we just don’t know so we should go ahead and do drastic things without knowing.
First, I never stated or implied that future generations don’t matter. Indeed they do, but just as I don’t value my own self in thirty years time the same way as I value it today, I should value others fifty or a hundred years out even less.
Nukes are a great example. First, I never suggested that nuclear war would be a good thing. In fact, I’ll specify that it would be a horrible thing with negative consequences in all time horizons. (Start with half a dozen air bursts to wipe out electrical transmission & generation in most of the West.)
But I would suggest that by deterring (up until now) a general war for the last sixty-five years that the threat of nuclear war has been a tremendous good. (16.5 million killed in WW I, 60 million killed in WWII–get the idea?) Surely Desert Storm and Operation Iraqi Freedom indicated that our capacity for industrialized death is growing, not the reverse.
If you meant to imply that the West should undertake whatever measures are necessary to prevent Iran from creating nuclear weapons, I heartily agree. Those people are not going to be deterred as the Communists were. But you seem to be suggesting something else entirely.
How on earth does one “do away with nuclear technology”? There isn’t a developed country on earth that could not produce a device in less than a year. (Including the SF operation to obtain the uranium.) This suggestion goes entirely beyond hubris–it is childish fantasy.
As for nuclear power, it beats anything else out there. Per megawatt, it is cleaner and safer than anything else we’ve got. (As an added bonus, it doesn’t fund crazies.) If people were willing to treat it as any other form of power generation, we would have more R&D, making it even safer and cleaner.
Right Wing Nut, it certainly makes sense to value yourself 30 years later less than you do now. After all, in 30 years you’ll be 30 years older and getting decrepit. You won’t be able to enjoy stuff as much, and you might prefer to just go ahead and die. You are going to get old and decrepit and die eventually, so the older you get the less you matter to anybody.
But that isn’t true of humanity as a whole. Many species last only 15 million years or so, but some survive longer and some evolve into new species that continue on. Every living thing today has an unbroken line of descent more than 200 million years. The idea that we might go extinct so we shouldn’t worry about doing things this generation which might drive them extinct, is circular.
You never suggested that nuclear war would be a good thing. I suggested that nuclear war might be a good thing in the long run. Of course it’s bad for the immediate victims, which is true of all war. I say that we don’t know enough about the long-term effects of nuclear war to tell whether it would be good or bad for us in the long run. However, there is an obvious chance that it could drive us extinct. It’s hard to quantify that chance. But the more nuclear wars we have, the more likely extinction becomes. Even if it turns out that one nuclear war has good results, there is definitely a point of diminishing returns and beyond megaton N or war M the results get worse.
I can’t tell you the chance that a nuclear war will kill everybody, or how long it will take to kill everybody after it happens. I say, don’t have a nuclear war until you’re pretty sure about that. If you kill everybody in the world, even if it takes 30 years or 50 years or 200 years to do it, you have done a very bad thing.
But I would suggest that by deterring (up until now) a general war for the last sixty-five years that the threat of nuclear war has been a tremendous good.</em.
You can't deter a general war with nukes unless you make a credible threat of using them. Every time you make a credible threat, you risk actually getting into a nuclear war.
This is a form of gambler's ruin. Every time you play, you have a large chance to win a small reward, and a small chance to lose a giant penalty. You don't know how small the chance is, or how giant the penalty will be. You do not know whether this is a good deal or not. But you might think you know because you have 65 years of experience winning. Likely if one time you lose that bet we'll both be dead and I won't get to say I Told You So.
Playing this game is not an act of humility.
As for nuclear power, it beats anything else out there.
We won’t know that until we have enough experience with nuclear power to get a baseline on the cost of major accidents. We completely lack that baseline now, because we have never yet had a major accident. So we don’t know how bad they will be.
Gambler’s ruin.
Again, this is not humility.
I value myself less because I don’t know if I’ll be alive then. BTW, if you’ve taken out a mortgage, you’ve made the exact same decision.
I could attempt to clarify the details of my position wrt nuclear war, but you’ve ignored my primary point: you’re demanding that the genie go back in the bottle. Its childish wishful thinking, nothing more.
I value myself less because I don’t know if I’ll be alive then.
Yes. And further, the chance that you will live to be 150 is extremely small. But we have no reason to think humanity will go extinct in the next million years except by accident or by malicious action.
You can do things which might get you killed, serene in the knowledge that you don’t have all that long to live anyway. Doing that for the human race is stupid.
I could attempt to clarify the details of my position wrt nuclear war, but you’ve ignored my primary point: you’re demanding that the genie go back in the bottle. Its childish wishful thinking, nothing more.
I think that there’s a strong chance that we will mostly stop building nukes and completely stop admitting to building nukes, after the next nuclear war. People will look at what happens to the losers of that war, and what happens to the winners, and they will decide that this is not an experiment that’s worth repeating. Democracies will decide that this is not something they will tolerate. Authoritarian governments will decide that this is not something they will authorize. They might build some bombs but will not threaten anybody with them, because the cost of admitting they have them is more than the value of the threat.
It would be nice if we could do that without having the war first, but I agree with you that is extremely unlikely.
I think this is an interesting variation on or maybe addition to the old argument about whether future consumption, like future money, should be discounted. Frank Knight argued that it shouldn’t. It has a lot of applications to tax policy, such as whether a consumption tax is better than an income tax and (same question, really) whether an income tax should allow a deduction for all interest (a position advocated by Henry Simons, following Knight).
The lower the discount rate, the more justification for regulation to prevent future harms, but also the more justification for policies which favor future growth over current spending, such as regulation to prevent future harms.
The discount rate for humans is a big deal in the global warming debate, because the costs of mitigation are immediate, but the harm is very far down the road – 50 years at least. If you use traditional measures for discount rates, you find that it’s quite difficult to find a scenario under which it makes sense to spend huge amounts of money on mitigation today. That’s one reason why people are trying to justify extremely low discount rates for humans.
Of course, if one accepts a low discount rate for humans, it would be even harder to justify current levels of deficit spending.
I discuss the relevance of discount rates to climate change policy starting at page 1343 of the linked article. You are certainly correct that the higher the discount rate for future persons, the less beneficial current regulations aimed at reducing future climate change will appear.
I’m not sure I agree about your last point, however. It would depend on how well we invest the money we borrow. If we spend all the money on beer, future persons might have reason to complain. But it we spend it on the education of folks who will save future persons from cancer (or perhaps who will invent a machine capable of reversing the climate change we fail to prevent), I imagine they’d appreciate our foresight.
Look at this as an inter-generational distribution issue: Future generations will be filthy rich compared to us. Redistribution ,however, is supposed to go from the rich to the poor. Therefore, we can let the little critters fend for themselves. Besides, what have future generations ever done for me?
Future generations will be filthy rich compared to us.
They will if we leave them enough stuff to build on. Which we won’t if we discount them too much.
Chicken and egg….
The chicken and egg problem goes away if you distinguish between different rationales for discounting (pure rate of time preference vs. intertemporal elasticity of substitution).
Conchis, I have the feeling that you said something important and I missed it. Would you mind explaining at somewhat greater length?
My thought, taken to extremes, was that if we decide future generations will be rich so we don’t need to invest in things that will help them but we consume instead, then future generations will not be so rich. Because their wealth is built on our investment. If we don’t leave them anything then they will have to start from scratch, minus the nonrenewable resources we consumed and minus the species we drove extinct.
Since our choices about what to leave to future generations have a profound effect on how rich they will be, we can’t use how rich they will be as a criterion for how much to leave them — unless we take into account the effects of our choice.
So for example, we will not leave future generations fossil fuels — unless the people who think that unlimited amounts of methane are constantly being created deep in the earth, which will be available forever, are right. If future generations have cheap safe nuclear fusion, they will be rich. But if nuclear fusion reactors are still 40 years in the future for them, they will not be.
I actually think I misinterpreted the nature of the problem your were alluding to. My apologies.
“we can’t use how rich they will be as a criterion for how much to leave them — unless we take into account the effects of our choice.”
For most policy decisions at the margins we’re likely to be talking about, the effects of our choices will be small enough that they don’t matter much. However, for policy decisions where they do matter (see e.g. the Stern report) the solution is exactly as you suggest: we take into account the effects of our choices on future consumption. What we make an assumption about is the intertemporal elasticity of substitution, which determines how fast the marginal value of consumption declines as people consume more. For any given growth rate, this will imply a discount rate on future consumption, but the resulting discount rate isn’t a primitive of the model, and so you don’t end up with an irresolvable circularity.
(Over and above this discount rate we will also assume a social rate of time preference that reflects (a) how much less we care about the utility of future persons just because they’re in the future; and (b) the probability that future generations won’t be around to enjoy any utility at all. Depending on your particular theory of population ethics, you may actually be concerned about some level of circularity in point (b) as well – at least if you’re considering policy options that will significantly affect the probability of an extinction event. However, at that point we start getting into some very muddy philosophical waters.)
FWIW, I seem to recall John Quiggin having a good primer on this stuff somewhere in his commentary on the Stern report, but can’t seem to find it now.
I do not understand the first argument, or why it would imply a different discounting factor. Sure, future saved lives will (most likely) be valued higher due to higher income and a higher willingness to pay: but this implies that the benefit side of the CBA should be affected, i.e. a saved live in 15 years should be valued at the current VSL multiplied by expected growth times the income elasticity of VSL. Then this VSL (higher than the current for a life saved this year) should be discounted as usual.
I agree that discounting is too high if benefit parameters are not adjusted accordingly. But if this is not done, it is a problem of the valuation of benefits, not of “too high” discounting, no?
Thank you for your comment. You are right that if future lives are corrected calculated (i.e., by “inflating” the current VSL at a proper rate), then I have no problem with applying existing discount rates to such benefits (or at least not much of a problem with it). My argument is that, as a practical matter, the effective discount rate for future lives will be lower if CBA calculations account for health inflation and wealth inflation.
In other words, imagine that an agency uses the common discount rate of 7 percent as part of its RIA. If health inflation and wealth inflation combine to justify an annual inflation factor of 4 percent for future lives, then the effective annual discount rate for a life saved in the future will be less than 7 percent (it will be about 3 percent).
And if the agency is using the other common discount rate of 3 percent, then an annual inflation factor in the neighborhood of 4 percent would completely counteract the effect of discounting for future human life, meaning a discount rate of zero would be appropriate.
So when I argue that current CBA practices “overdiscount” future human lives, it’s equivalent to stating that they “undervalue and then properly discount them.”
Could I suggest that future generations have utility value only insofar as they are valued by us, the living? And the utility being weighed is OUR utility, in thinking thoughts of a happy, prosperous future humanity. NOT that of the (non-existent) people themselves. The discount rate should reflect that; our present preferences for our descendents, and nothing more.
If moral duties to non-existent entities exist, then no-end of absurdities can be promulagated… I might start by demanding massive increase in AI funding to maximise the number of future sentient machine intelligences….
“If moral duties to non-existent entities exist, then no-end of absurdities can be promulagated [sic]…”
This doesn’t necessarily follow: you can place (discounted) value on the lives of future persons conditional-on-their-existing, without creating a duty to bring them into existence in the first place. The conditional claim that “if I expect to have a child I should care for it” does not entail “I should have a child”.
There have been no cases of benzene-induced leukemia in the U.S. in which the causative exposures occurred after the U.S. Supreme Court struck down (1980) the Emergency Standard (1977). Indeed, the SMR for leukemia in the cohort on which both the emergency and eventual 1ppm standard was based has been falling for years and is rapidly approaching background. That’s because all of the injurious exposures occurred during the war years when exposures were commonly an order of magnitude higher than the standard the U.S. Supreme Court kept in place. Thus, neither the 1977 nor the 1987 regulation did anything other than drive up costs (as reflected in the overall leukemia rates recorded in SEER which show no effect of changing standards on leukemia incidence).
The linear no-threshold dogma, the precautionary principle and this “don’t apply positive discount rates to human lives” business share the same purpose – driving perceived risk/loss to infinity so that regulations of chemicals/cell phones/whatever can be driven to infinity. The usual advocate for the position ever since Ind Union Dept, AFL-CIO v. API http://scholar.google.com/scholar_case?case=968580795714943926 has been Carl Cranor http://www.youtube.com/watch?v=tSoRWPiaJ-c who’s an old school Silent Spring – “man is responsible for his suffering” environmentalist (the movement is nowadays called “environmental justice”) sort still stuck in 1970 with the belief that industrial toxins are either the primary or at least a significant source of human illness in the U.S.
“The linear no-threshold dogma, the precautionary principle and this “don’t apply positive discount rates to human lives” business share the same purpose”
Perhaps they have the same effect, but that’s not the same thing. They actually have quite different rationales (which is reflected in the fact that I generally buy the ‘zero rate of pure time preference’ arguments, but not the others).
Comments on this entry are closed.