Abstract: Traditionally, the discounting debate has been dominated by those who advocate equality between generational interests and those who think future generations’ interests should be discounted at some positive rate. This paper argues for a novel view: future generations’ interests should be negatively discounted. First I defend the claim that we have greater reason to promote the well-being of those who are more morally deserving. Then I explain why we should expect future people to be more morally deserving than us. Throughout this argument, scepticism about moral desert looms large. Nevertheless, given the context of moral uncertainty under which the discounting decision must be made, a negative discount rate remains a live option even in the face of a moderate degree of scepticism about moral desert.

Keywords: future generations, discounting, desert, moral uncertainty, moral luck.

Practical ethicists and policymakers alike must grapple with the problem of how to weigh the interests of future people against those of contemporary people. This question is most often raised in discussions about our responsibility to abate climate change (Greaves, 2017), but it is also pertinent to the mitigation of other existential risks, disposal of nuclear waste, and investment in long-term scientific enterprise. To date, the debate has been between those who defend the practice of discounting future generations’ well-being at some positive rate and those who argue that the only morally defensible discount rate is zero. This paper presents an argument for a negative discount rate:

P1. There is reason to prioritise the well-being of those who are more morally deserving.

P2. There is reason to expect that future people will be more morally deserving than us.

C. Under moral uncertainty, there is reason to prioritise the well-being of future people over our own. 

Before I defend the above argument, two clarifications are in order. First, this paper is concerned with the appropriate rate of pure time preference; other factors which affect thediscount rate, such as exogenous extinction risk, are therefore ignored. Second, some claimsin this paper presuppose that there are unchanging moral facts; I believe that a negative discount rate could be defended without this metaethical commitment, but that task is not undertaken here.

The plausibility of desert-based moral theories

Everything else being equal, it is better that good things accrue to those who deserve them more rather than those who deserve them less. The credibility of P1 turns on whether there is any philosophical weight behind the superficial appeal of this claim. A simple thought experiment suggests that there is. Suppose you encounter two strangers who are suffering to indistinguishable extents from extreme pain. You discover that one of the pair, Aggressor, viciously attacked the other, Victim, thereby causing Victim’s pain. By a strange fluke, Aggressor’s attack backfired and also injured Aggressor. As it happens, you are in possession of a single dose of a potent analgesic. It seems you have more reason to relieve Victim’s pain than Aggressor’s, even though the only differences you are aware of between them are those which bear on their respective moral desert. The implication of this case is that when scarce resources are to be distributed, the moral desert of the recipients is relevant to our choice of allocation. Endorsing this conclusion entails accepting a moral theory according to which we have greater reason to promote the well-being of those who are more morally deserving.

An objection levied against some desert-based moral theories is that they are inconsistent with the welfarist principle that it is always good to alleviate suffering (Crisp, 2021). This principle can be accommodated within a desert-based moral theory, however. In axiological terms, the upshot of the above thought experiment is not that pain relief for Aggressor is of neutral or negative value, but merely that it is less valuable than pain relief for Victim. It may be that pain always has disvalue, but the magnitude of its disvalue depends on moral desert (Kagan, 2005). Likewise, in deontic terms, the point is not that there is no pro tanto reason to alleviate Aggressor’s suffering, but rather that any such reason is defeated by a stronger reason to alleviate Victim’s suffering. Perhaps the weight of our reason to alleviate suffering – although there always is one – depends on the deservingness of the sufferer. This objection is therefore problematic for at most some desert-based moral theories, but not every such theory, so it does not threaten P1. 

A more generic worry about the plausibility of desert-based moral theories is that it may be difficult to supply a philosophically compelling account of moral desert. Moral desert may be defined with reference to character, motivations, intents, or actions. This paper is not concerned with comparing different analyses of the concept, and for the remainder it is assumed that an agent’s moral desert is at least correlated with – although not necessarily constituted by– the degree to which they are disposed to act morally. P1 may appear suspect, however, if there is reason to doubt that any real characteristic of human agents is denoted by ‘moral desert’. One basis for doubt is the possibility that human agents lack free will and therefore the capacity for moral responsibility necessary to ground desert claims. I will not argue against this possibility, except by pointing out that it is inconsistent with certain robust intuitions, for instance that Aggressor and Victim differ in their moral desert. Below, I suggest that the central argument of this paper is not undermined by a moderate degree of free will scepticism as long as we continue to have an appreciable level of credence in at least one desert-based moral theory.

Evidence for moral progress

The argument for the premise that future people will be more morally deserving is by induction. The long-term trajectory of human development is towards the widespread acceptance of more tolerant, considered, and compassionate moral attitudes. The sphere of moral concern of the average person is more inclusive than ever before, having expanded from the tribe to embrace more of humanity and now other sentient creatures (Singer, 1981). This process has been punctuated by the abolition of slavery, the invention of human rights institutions, and the establishment of the animal advocacy movement. Should this trend continue, the moral attitudes of future generations will be systematically superior to our own. An obvious objection to this claim is that recent centuries have also involved some of the most grievous moral catastrophes ever witnessed, including war on an unprecedented scale, genocide, and deplorable inequality. This objection takes too narrow a view of human history, however. In the context of the broader trend described above, such cases have the status of outliers, not counterexamples. P2 does not turn on whether we are more morally deserving than our grandparents, just as its implication is not that our grandchildren are certain to be more morally deserving than us. Rather, the point is that when we take actions with consequences that will potentially reverberate through thousands of generations, it is appropriate to recognise that the long-term trend of human development is characterised by moral progress.

To secure P2, some supplementary assumptions are required. First, the fact that moral attitudes are improving relative to presently-accepted moral standards does not suffice to show that they are improving tout court. This requires the additional claim that moral standards are converging on objective moral facts. Although it is impossible to non-circularly prove the superiority of contemporary moral standards, they do trump past standards by at least one quasi-objective metric. Namely, contemporary moral standards enjoy a greater degree of internal coherence. For instance, the paradoxically racist and sexist ‘egalitarianism’ of the Enlightenment has been supplanted by a general acceptance of the moral equality of all human beings, which is more internally coherent since it does not ground claims about relative moral worth in arbitrary distinctions between persons. Second, since P2 is about individuals rather than society at large, it is necessary to stipulate that changes in societal moral standards will coincide with changes in the moral attitudes of the average person. This is plausible: there is a close relation – at least causal, possibly constitutive – between the moral attitudes accepted by individuals and those endorsed by the society which comprises them. Finally, an agent’s moral desert is indicated by their dispositions to act morally, not merely their moral attitudes. To bridge the gap between improved attitudes and improved acts, it must be the case that future people are disposed to act in accordance with their moral attitudes to at least the same degree as we are disposed to act in accordance with ours. If these conditions are satisfied, it follows that future people will be more disposed to act morally than contemporary people.

As it stands, the argument for P2 faces a serious challenge. There may be circumstances in which a more reliable disposition to act morally does not translate into greater moral desert, in particular if circumstantial luck is involved. Circumstantial moral luck arises whenever an agent who acts rightly would not have done so had environmental factors outside of their control been different (Nagel, 1979). It may seem that any systematic difference between generations in the disposition to act morally would be due to circumstantial luck. For instance, suppose it is right to donate a sizeable proportion of one’s income to charity, and that more people will do so in the future than at present because people in the future will be better off and so find it less onerous to be charitable. In this case, the relative beneficence of future people cannot ground a desert claim because it is due to environmental factors beyond their control. After all, nobody chooses the material conditions of their historical context. The objection can now be clearly stated: if circumstantial luck undermines desert claims, and the superior moral conduct of future people is the product of circumstantial luck, then the superior moral conduct of future people cannot ground a desert claim.

Fortunately, this challenge can be met, even granting that circumstantial luck undermines desert claims. This is because the hypothesised superior moral conduct of future people is not – or at least not primarily – due to circumstantial luck. Perhaps changes in environmental factors will make it easier for future people to act morally as the above example suggests, but the argument for P2 does not rely on that. Rather, the argument relies on a systematic improvement in moral attitudes – construed broadly as judgements concerning which actions are right and which are wrong. Unlike the conditions of their environment, moral attitudes are – in a morally significant way – endogenous to the agent who holds them. A person’s sensitivity to the morally relevant features of a situation and their motivation to respond appropriately are partly constitutive of who they are as a moral agent. These considerations are apt to ground a desert claim if anything is.

Another objection lurks nearby, however. There is a sense in which even agents’ moral attitudes may be considered a product of luck: constitutive moral luck arises when factors outside an agent’s control determine what kind of person the agent is (Nagel, 1979). Since we cannot always choose what to value and what to deplore – just as we cannot always choose what to believe or what character traits to manifest – constitutive luck may play a role in the determination of our moral attitudes. To the extent that we can choose our moral attitudes, the degree of self-control we are able to exercise may also be a matter of constitutive luck. Nevertheless, it is plausible that constitutive luck does not undermine desert claims even though circumstantial luck does. If someone enters a burning building to save people within, they might deserve credit for being (constitutively) lucky enough to be a heroic person without deserving any more or less credit based on whether they are (circumstantially) lucky enough to succeed. If constitutive luck does undermines desert claims, however, then P2 will be false. In fact, given the ubiquity of constitutive luck, it is plausible that if constitutive luck undermines desert claims, then no desert claim is well-grounded (Levy, 2011). It should now be clear that this objection is a variation on the general scepticism about moral desert discussed above. Although I will not argue against this scepticism, I will suggest below how it may be sidestepped.

Intergenerational discounting under moral uncertainty

The above discussion has demonstrated that there are strong reasons to accept P1 and P2, but we may have lingering doubts. For instance, we have seen that both premises presuppose that a real quality is denoted by ‘moral desert’, and that this presupposition depends on substantive assumptions about free will, moral responsibility, and luck. Nevertheless, we must choose how to weigh the interests of future people against our own. This is a morally significant decision which cannot be postponed until all normative debates are settled, making it a prime example of a choice which must be made under conditions of moral uncertainty. Given this context, if we are committed to maximising the expected moral value of our decision, it is not necessary that we have absolute confidence in P1 and P2 for there to be a pro tanto reason to apply a negative discount rate (MacAskill et al, 2020). Rather, it is sufficient that we have an appreciable degree of credence in at least one desert-based moral theory and we believe it is likelier that future generations will be more morally deserving than us as opposed to less so. The magnitude of the negative discount rate applied will depend on the specifics of the desert-based moral theory, the extent of anticipated moral progress, and our levels of confidence in P1 and P2. 

Of course, decisions taken under moral uncertainty must also account for ‘interaction effects’ between competing moral theories(MacAskill, 2019). To take an illustrative example, imagine that a policymaker is reasonably confident that there is reason to prioritise the well-being both of those who are more morally deserving and of those whose well-being is lower ex ante. Suppose also that they believe people in the future will be both more morally deserving and better off than we are now. The policymaker would then have reason to consider a marginal improvement in the well-being of future people more worthy of promotion than a comparable improvement in well-being today, and reason to consider it less so. Whether it is appropriate for them to apply a positive, negative, or zero discount rate depends on the details of their moral judgements and nonmoral beliefs, and the complex nexus of their interaction.

The significance of the argument presented in this paper is not that it settles the question of how to weigh the interests of different generations against one another, but that it provides a pro tanto reason to prioritise the interests of future people while the question remains unsettled. This reason must be balanced against reasons to afford less weight to future generations’ interests than our own. If this novel reason were ignored, a treatment of intergenerational discounting under moral uncertainty would likely recommend a relatively low positive discount rate, based on a trade-off between reasons which favour a relatively high positive rate and countervailing reasons for a rate of zero. In light of this paper’s argument, however, such an analysis need not recommend a positive discount rate; the optimal discount rate may be zero, or even negative.

In sum, I have argued that we have an unacknowledged reason to go beyond regarding future generations as entitled to equal treatment and instead prioritise their interests over our own: this is what they deserve.

Acknowledgements: For valuable feedback on an earlier draft, I am grateful to Amber Dellar, Roger Crisp, an anonymous reviewer for the Journal of Practical Ethics, and an audience at the Oxford Uehiro Centre.

References

Crisp, R. (2021) ‘Well-Being’ in Zalta, E. N. (ed) Stanford Encyclopaedia of Philosophy.

Greaves, H. (2017) ‘Discounting for Public Policy: A Survey’, Economics and Philosophy (33): pp. 391-439.

Kagan, S. (2005) The Geometry of Desert (Oxford: OUP)

Levy, N. (2011) Hard Luck: How Luck Undermines Free Will and Moral Responsibility (Oxford: Oxford University Press).

MacAskill, W. (2019) ‘Practical Ethics Given Moral Uncertainty’, Utilitas 31(3): pp. 231-45.

MacAskill, W., Bykvist, K. and Ord, T. (2020) Moral Uncertainty (Oxford: Oxford University Press).

Nagel, T. (1979) Mortal Questions (New York: Cambridge University Press).

Singer, P. (2011) The Expanding Circle: Ethics, Evolution, and Moral Progress (Princeton: Princeton University Press).

Forthcoming