2. The original position and what it represents

What kind of problem is the savings problem in the original position? Is it a reflection of a general obstacle to our thinking about intergenerational justice, or merely a queer consequence of Rawls’s theoretical construct? How might we begin to understand the relevance of the highly abstracted features of this philosophical problem to the concrete questions policy-makers face in thinking about future-oriented actions? I suggest that we begin by clarifying the role and function of the original position before turning to the problem and the various revisions that have been proposed. This will help us not lose sight of the relevant practical problems – such as if, and to what extent, we are obliged to consider the interests of future people in contemporary decision-making – when treading through some rather abstract theoretical intricacies. It may also help us see why it is so hard to visualise a fair bargaining situation for negotiating principles of justice while keeping in mind the idea of society as a fair system of cooperation over time.

There is no need to reiterate all of its features, but we do need to recall the underlying idea of the original position, as this seems so easily lost in discussions. The first thing to note, then, is that Rawls intends it to be ‘an expository device’ that helps us, that is, you and me, to organise our intuitions relevant to justice. ‘The idea’, he writes, ‘is simply to make vivid to ourselves the restrictions that it seems reasonable to impose on arguments for principles of justice, and therefore on these principles themselves’ (Rawls Reference Rawls1971, 18). The original position models what we consider, on due reflection, reasonable conditions for reasoning about such matters or, as Rawls also puts it, the reasoning of citizens of a well-ordered society, that is, a society effectively regulated by principles of justice that all citizens publicly accept.Footnote ⁶

We are, however, not citizens of such a society, nor do we always apply criteria of reasonableness in normative discussions. We do not only fall short of what we would consider good standards of reasoning – for instance, by being biased or making inconsistent judgements – but also, just as often, are blind to our own failures. In order to mitigate such shortcomings, which we must if we are to organise our society by justified principles of justice, we could try to simulate the process of reasonable choice. Thus, Rawls writes: ‘[a]t any time we can enter the original position, so to speak, simply by following a certain procedure, namely, by arguing for principles of justice in accordance with these restrictions’ (Rawls Reference Rawls1971, 19).

This understanding of the original position brings out just how demanding Rawls’s approach is, both for the theorist and for, let us say, the practitioner: it requests of the former that it gather and assemble what, on due reflection, can be considered criteria of reasonable or fair argumentation; and of the latter that it be competent enough to simulate reasoning in keeping with those criteria. This should humble us somewhat as theorists. We may be mistaken about what the criteria for a fair choice situation are, especially so when factoring in the novel features of the intergenerational setting, as we must for future-oriented decisions. We should also bear in mind that a desideratum for an interpretation of the original position thus is that it is feasible for people to simulate reasoning in accordance with it.

Describing the original position as a device of representation that people can adopt to simulate a target system is, however, not satisfactory. The two desiderata encapsulated in that – that it be descriptively adequate and facilitate the simulation of the target system – are at any rate insufficiently specified. This can be gathered by reflecting on the methodology, called constructivism, Rawls develops in his later writings (Reference Rawls1980; Reference Rawls2005).Footnote ⁷ I will soon explain what I take constructivism to mean, but consider first how it influences the presentation of the original positions. Rawls writes: ‘[the] role [of the original position] is to establish the connection between the model-conception of a moral person and the principles of justice that characterise the relations of citizens in the model-conception of a well-ordered society’ (Rawls Reference Rawls1980, 520). We can see here how the envisaged role for the original position now embodies a further function. It functions as a ‘mediating conception’ between the conception of the person and the normative principles and so establishes a connection between the two (cf. Rawls Reference Rawls1980, 516). Rawls intends the original position to carry out this function by being a model, not only of an idealised choice situation, but also of an idealised chooser. The moral conception of the person as free and equal, for instance, is part of the model conception of justice as fairness (cf. Rawls Reference Rawls, Sen and Williams1982, 182).

Although there is no unison and undisputed meaning of constructivism, some of its core ideas can elucidate the revised role of the original position.Footnote ⁸ Constructivism, Rawls (Reference Rawls1980, 516) explains, ‘specifies a particular conception of the person as an element in a reasonable procedure of construction, the outcome of which determines the content of the first principles of justice’. This is not only a paraphrase of his earlier method (that is, the method of reflective equilibrium), but also a substantial addition to it, namely: the emphasis put on the conception of the person for justifying principles of justice. This methodological recommendation thus underlines the role played by the person in the making of fair decisions. In other words, to model a fair decision situation involves modelling what persons would have to be like in order to make such fair decisions. Constructivism accordingly qualifies the first desideratum of the original position: for it to be descriptively adequate of its target system, it needs a richer description that includes also a conception of the person.

There is also a second qualification. To get to that, let us first note some things about what constructivism is not. First, it is not a method for discovering normative ideals; rather, it assumes some such ideal, such as the need for justice. What it does is to pose two preliminary methodological questions: (1) What conditions must obtain for a normative justification of the considered ideal to be acceptable; and (2) How must persons be constituted for that to happen? Constructivism thus complicates the justification of any normative ideal: it does not accept any conditions as conducive to normative justification, and not any person as able to reason accordingly. The theorist’s job is to assess not the acceptability of individual principles, but rather the conditions for any acceptable justification and what ideal person such conditions assume. The second thing to note is that constructivism is not, or at least not directly, a method for normative justification. The ultimate justification of any normative ideal is if it can be accepted in reflective equilibrium, possible only in a well-ordered society (see Rawls Reference Rawls1980, 534; cf. Rawls Reference Rawls2005, 28). Furthermore, not only principles of justice but also criteria of fairness are ultimately justified by practitioners rather than theorists. It is not the case, Rawls argues, that criteria of reasonable or fair decisions can be taken out of thin air, nor simply be invented by the theorist.

That said, the theorist still has a crucial role to play, which takes us back to the revised function of the original position:

Rawls gives the theorist the role of assembling the most essential convictions of what makes for a fair decision and to connect these in ways conducive to the justificatory process (cf. Rawls Reference Rawls2005, 108). Here, we get to the second qualification, which regards the second desideratum of the original position. Any revision of it must resonate with conceptions of society and person found in the public culture of a democratic society. This is important both to get any simulation off the ground and for simulators to learn anything from reasoning in accordance with the model conception. Although the original position cannot fully justify principles of justice, it can approximate such a justification, rightly characterised.

Before we move on to assess proposed solutions to the savings problem in the original position, let us add one other clarification. If the original position is a representation of an ideally fair choice situation, the well-ordered society, then the savings problem in the original position too has a counterpart. That the parties in the original position would not adopt a just savings principle represents a possible way in which a society could fall short of being well ordered. That is, the possibility that a purportedly fair reasoning process leads to the undesirable conclusion that society should be organised such that nothing is saved to future generations. Let us call this the target problem. This, of course, takes for granted a judgement about that possibility, namely that the decision to neglect the interests of future people is unfair and so wrong. Although also this judgement could be challenged, I will proceed under the assumption that it is valid, or as Rawls (Reference Rawls2005, 123) would put it, a fact about ‘the possibilities of construction’.

However, granted the assumption, against whom would it be unfair? There is, to wit, an ambiguity in the representational relation between the savings problem in the original position and the target problem: either we understand the original position as modelling the conditions for a fair agreement among contemporaries of a well-ordered society, or the conditions for a fair agreement among its contemporary but also future members (or still more inclusively, all affected). This ambiguity comes from the subjects to the allegedly unfair agreement being unspecified: it is unclear whom the theorist should take into account in conjecturing fair conditions. Thus, there are at least two versions of the target problem: (1) unfairness among contemporary citizens and (2) unfairness among non-contemporary citizens. Deciding which one is relevant is challenging because although accounting for the problem in the second sense is more intuitive – a no savings decision will, after all, primarily affect future people – there are good countervailing reasons for sticking to the first. It is, for instance, natural to think that the original position represents the activity of collective reasoning under certain conditions, and such reasoning can, of course, only comprise existing persons. Relatedly, if the explanation for why a policy or proposal is unfair is that someone subjected to it have reasons to not accept it, then policies and proposals cannot be unfair against future people, as they can neither accept nor reject things in their current non-existing state.

There is, however, an important, though subtle, difference in conceptualising decision-making of relevance to future generations as either taking into account the likely impact on future people’s interests or, as it were, striking a fair deal with future people. In other words, between acknowledging a moral obligation regarding future people and a moral obligation owed to future people (cf. Mazor Reference Mazor2010, 381; Brandstedt Reference Brandstedt2015). Thus, even though future people are not subjects to contemporary agreements, they will in many cases be affected. In other words, they are third parties (cf. Rawls Reference Rawls1971, 128) to future-oriented decision-making of today, and accordingly their interests are made relevant. To the extent that present people represent the interests of future people, their interests are turned into proper claims that need to be balanced against competing ones. As I will argue in section four, there are surprisingly many ways in which future people’s interests are represented in projects and practices of today. If that argument is made convincing, it shows that there are reasons for us – here and now – to reject indifference to the interests of future people, even if those reasons ultimately derive from a justificatory process only involving contemporaries.

3. Assessing solutions on a model and target level

Now, the question is not if there is a way in which the original position can be revised so that it does not produce the counterintuitive result. There are indeed several such possible revisions, as we shall soon see. The relevant question is if a revision can be justified in a non-ad hoc way.

A first possible revision is mentioned by Rawls (Reference Rawls1971, 139–140) already when he presents the savings problem in the original position, and although he quickly dismisses it, it has been picked up by some others as promising (e.g. Barry Reference Barry1989, 505–510; Gardiner Reference Gardiner, Gosseries and Meyer2009, 114–115). It is to drop the assumption, which we have been working under but not yet spelled out, namely that the parties in the original position all are contemporary: the so-called ‘present-time-of-entry interpretation’. To do so would transform the original position into a ‘general assembly’ composed of representatives of both the actual and merely possible generations, negotiating the terms of a truly intergenerational contract. Rawls (Reference Rawls1971, 139) dismisses this idea for the good reason that it would ‘stretch fantasy too far; the conception would cease to be a natural guide to intuition’.Footnote ⁹ If the original position is an expository device, then it is essential that we can adopt it, so that it can facilitate the organisation of our intuitions. But it is difficult if not impossible to imagine oneself in a negotiation, which has on the table principles threatening the existence and identity of those represented. Furthermore, how could any principle be accepted in such a situation, when each decision results in the non-existence of some represented possible people?Footnote ¹⁰

A second revision is to make a ‘motivational assumption’ of the parties in the original position. In the first configuration, the only requirements placed on their conception of the good are ones of rationality (e.g. that they take effective means to their ends). The idea of a motivational assumption is to introduce a further, and different kind of, constraint on their conception of the good. If we think of the original position as a kind of dynastical model, where the parties as ‘heads of families’ represent not only their own conception of the good but also that of their closest descendants, then we might avoid the unwanted conclusion (Rawls Reference Rawls1971, 128–129). In this reconfigured model, the parties have reason to save for future generations because it is instrumental to the furthering of their rational self-interests in this newly specified role.Footnote ¹¹

I will not here repeat all of the criticism that have been raised against this proposal, but just mention one salient point, in a slightly new take. Assuming that the motivational assumption solves the savings problem in the original position, it is far from clear that it solves the target problem. If it does not, maybe that justifies Jane English’s (Reference English1977) judgement that the solution is ad hoc. On the target level, the motivational assumption seems to suggest that the unfairness comes from lack of care. Must we, accordingly, not only be reasonable to our fellow citizens, but also empathic towards our closest relatives when we justify the conditions of a just society? Think of the, seemingly well-ordered, society organised fully negligent of the interests of future people, and of how its citizens relate to their descendants. The citizens certainly seem to exhibit carelessness and lack of empathy, even to the extent that it is hard to understand their behaviour, but is it unfair that they do so, or just puzzling?Footnote ¹² Conversely, most real people, of course, care greatly about their offspring and are highly motivated to further their well-being, yet we have climate change and a whole set of intergenerational problems. The target problem is unlikely due to lack of care, and so a motivational assumption is unlikely to solve the savings problem in the original position.Footnote ¹³

Even if the above reasoning is an elaboration thereof, it is in full agreement with Rawls’s argumentation. There are indeed good reasons for dismissing the general-assembly interpretation and for not opting for the motivational assumption. We now come to a point, however, where there are reasons to deviate from his course of reasoning. Rawls presents his considered response as a simple solution to the savings problem in the original position, whereas it is unclear to me both if it solves the problem and if simplicity should be the arbiter of that. The proposal is of another kind of revision of the original position. If the motivational assumption is an addition to the parties’ (thin) conception of the good, this is proposing to revise the choice situation itself, that is the set of conditions – such as that principles must have general form (see Rawls Reference Rawls1971, 130–142) – it is composed of.Footnote ¹⁴ The proposal is the following:

I will refer to this as the additional constraint. Rawls’s description of it as a simple solution presumably comes from thinking of it as preserving both the idea of society as a fair system of cooperation between generations and the present-time-of-entry interpretation of the original position. However, if we fully take into account the idea of the considered model presented above, this solution begins to look less satisfactory.

I should first clarify that additional constraint is not a proposal to shift from, what Rawls calls, ‘nonideal theory’ to ‘ideal theory’, that is to introduce a condition of general (or strict) compliance, as some have thought (e.g. Attas Reference Attas, Gosseries and Meyer2009, 201–203; Heyd Reference Heyd, Gosseries and Meyer2009, 178–181). It is not because that assumption is already in place: this is ideal theory. Rawls has all along assumed that the principles of justice are worked out under the assumption that all subjected comply – and this has not prevented the savings problem in the original position. This is, of course, premised on the first version of the target problem. According to this, when negotiating a conception of justice, it is irrelevant whether or not past generations have complied and similarly whether or not future generations will comply – except if such things can be suspected to affect the stability of the conception. Apart from that the fairness of the choice is decided by considering only relations between contemporaries. Thus, the reason for why it seems rational, though unfair, for the present generation to not save for future generations is not because of non-compliance, either on part of some members of the present (that has been assumed away) or on part of some other generation (that is not relevant). Strict compliance is not the solution.

Consider now how additional constraint seemingly solves the savings problem in the original position. In the revised model, there are reasons for the parties to adopt a just savings principle. Any decision to not save would defy additional constraint, as it would not be compatible with wanting previous generations to have followed it. In other words, it is inconsistent for the parties to uphold simultaneously the following two desires: to want previous generations to have saved and to want to not save for future generations. Consider now how this could be reconstructed on the target level. Basically, the idea is that what is missing, and what accounts for the unfairness of the considered problem, is the application of something like an universalisability test to proposed savings policies. What accounts for the shortfall of fairness in the less than well-ordered society accordingly is that the savings decision (i.e. of zero savings or even dissaving) contradicts what its citizens implicitly want previous generations to have done for them. I write ‘implicitly’ because the citizens presumably need not make the desire about past savings explicit: mere appreciation and consent implicit in accepting the institutional order of society could be enough to contradict a desire for no future savings. On yet another formulation of this proposal, fairness demands that all desires relevant to intergenerational savings have the form of a law, such that it could be held by any, and so all, generations.

There are, as I see it, at least three reasons for being somewhat sceptical of additional constraint as a solution.Footnote ¹⁵ The first is that it effectively requires a particular desire, namely: to want previous generations to have saved at a certain rate. The proposal thus is not just that fairness demands of citizens that they iron out any rational inconsistencies in their sets of desires before allowing them to influence the fabric of society. That is, another requirement already assumed, namely: ordering or completeness (Rawls Reference Rawls1971, 133–134). As noted, the proposal does amount to rationally require a particular desire. That is, however, a controversial point to make, for instance, in standard decision theory. The second reason for scepticism comes from additional constraint being more specific, that is restricted, than any of the other constraints of the original position. Whereas the others are formulated in general terms and all are plausible candidate conditions for an explication of the concept of right, additional constraint seems to have only one function, that is, to solve the savings problem. The worry thus is that it is ad hoc, which would be mitigated only by independently justifying it as being in concordance with what we would, on due reflection, consider fair. Until such justification is supplied, there is something wanting about additional constraint.

Finally, the third reason for scepticism relates to the previously discussed desiderata of the original position as an expository device: How does additional constraint stand up to an evaluation of the model in terms of its function? To repeat, Rawls thinks additional constraint is justified because it is a reconfiguration of the original position that produces the intuitively desirable result while preserving the present-time-of-entry interpretation. Taking for granted that it should also be descriptively adequate, the seemingly only other desideratum of a solution, we can infer, is that it maintains the present-time-of-entry interpretation so that it can be adopted. But as was argued in section two, this is to under-appreciate the role the original position could play. It is not just a workable device of representation, but more specifically of a particular choice situation with citizens suitably related to a certain self-understanding. So, although we can enter the original position with additional constraint and reconstruct the reasoning to the intuitive result of a just savings principle, to be content at that would be to short-change the potential of the model.

4. Sustainability of values – Another primary good

On the basis of the discussion in section two, one aspect of Rawls’s theory stands out as especially interesting in responding to the savings problem in the original position and its target problem. It has not yet been discussed, although it plays a crucial role in motivating the choice of the parties in the original position. It is Rawls’s account of ‘primary goods’.

Rawls first defines primary goods as ‘things that every rational man is presumed to want’ (Rawls Reference Rawls1971, 62), but later revises it to mean ‘persons’ needs as citizens’ (cf. Rawls Reference Rawls, Sen and Williams1982; Rawls Reference Rawls2005, 179). These are either social or natural, but since it is only goods of the former kind that are under direct control, it is those he focuses on. Rawls (Reference Rawls2005, 181) writes: ‘The basic list of primary goods (to which we may add should it prove necessary) has five headings as follows:

1. (1) basic rights and liberties […];

2. (2) freedom of movement and free choice of occupation against a background of diverse opportunities;

3. (3) powers and prerogatives of offices and positions of responsibility in the political and economic institutions of the basic structure;

4. (4) income and wealth; and finally,

5. (5) the social bases of self-respect’.

These are specified by ‘identifying a partial similarity in the structure of citizens’ permissible conceptions of the good’ (Rawls Reference Rawls2005, 180).Footnote ¹⁶ No matter what they desire or value, these are things citizens need to advance their conception of the good. It is thus ‘a shared idea of rational advantage’, or a standard of legitimacy that allows for interpersonal comparisons in a situation of opposed and incommensurable interests, such as that of a contemporary democratic society (cf. Rawls Reference Rawls, Sen and Williams1982, 161; Rawls Reference Rawls2005, 180). The underlying problem of finding a public understanding of what are appropriate claims to make in a pluralist society is solved by the account of primary goods. There is also a counterpart problem on the level of the original position. The veil of ignorance deprives the parties of the possibility of unfair bargaining, and also of the possibility of a rational agreement because of the scant information allowed. The idea of primary goods solves this too. If we assume that the parties prefer as large shares as possible of primary goods, then it is rational for them to agree on principles of justice, even without having to assume anything about the determinate conceptions of the good of those represented.

I now want to defend the claim that the savings problem in the original position, that is, the legitimising of no savings to future generations, can be dealt with by an addition to the primary goods. As a sixth category to the existing five, I propose: sustainability of values. Sustainability of values is a primary good in the sense that it gives values broadly conceived – underlying preferences, commitments, projects and traditions – point and meaning by guaranteeing their continuation over time. I will soon motivate this proposal further, but consider first how it solves the problem, on both model and target levels. With sustainability of values as a primary good, it is rational for the parties to agree on a just savings principle because that advances this primary good irrespective of what projects, traditions and preferences the citizens represented have once the veil of ignorance is lifted. As to the target problem, it can be solved in the following way: the group of people treat one another unfairly by choosing not to save anything for future generations, as doing so jeopardises the preferences, commitments, projects and traditions whose value and meaningfulness are conditioned on a future continuation and maintenance of society. Put in another way, to not save is unfair because it treats persons as severely limited temporal beings, like mayflies if you will.

The primary good sustainability of values can be motivated by drawing attention to what Samuel Scheffler (Reference Scheffler and Kolodny2013) recently has put forward as ‘the afterlife conjecture’.Footnote ¹⁷ Scheffler uses ‘afterlife’ in an unconventional sense, as the continuation of the material and social world after our personal death. His idea is that by reflecting on scenarios where it is common knowledge that there will be no such afterlife, we learn something about the concept of valuing and what it presupposes. The main point is presented as a conjecture: ‘people would lose confidence in the value of many sorts of activities, would cease to see reason to engage in many familiar sorts of pursuits, and would become emotionally detached from many of those activities and pursuits’ (Scheffler Reference Scheffler and Kolodny2013, 44). The reason for this, Scheffler argues, is that the afterlife is a condition for many, if not all, of the things that matter to us to continue to matter, be important and valuable.

The afterlife conjecture thus reveals hidden dimensions of valuing, for example, that it is to some extent conservative: ‘to value X’, Scheffler (Reference Scheffler and Kolodny2013, 60) explains, ‘is normally to see reasons for trying to preserve or extend X over time’. There are some examples in which this is especially evident. On a general level, these examples are of institutions, practices, activities, ways of life, traditions and group-dependent projects that have temporally distant goals and diffused benefits. More concretely, think of, for instance, cancer research, social activism and large infrastructure projects. These are projects and practices whose point and purpose directly depend on the afterlife. Less obviously, participation in cultures, religions and traditions, such as taking on a national identity and carrying on a particular cultural celebration, also have such dependence. And most far-reaching, one may follow Scheffler’s proposal to the conclusion that the afterlife is a condition for anything at all to matter.

Setting aside that strongest version of the thesis,Footnote ¹⁸ the gist of the afterlife conjecture can enrich our understanding of agency in the intergenerational setting. The dependence on the continuation and preservation of things of a transboundary character, expressed by the afterlife conjecture, suggests that present-day individual interests are temporally interlaced with the interests of future people. That there will be a future, in which people continue on and maintain contemporary projects and practices, is a condition for many things mattering today. Thus, measures taken to increase the chances of such maintenance are purposeful irrespective of what specific transgenerational projects and practices we invest in.

The introduction of sustainability of values in the original position is a revision or, if you will, an update of the general knowledge allowed to the parties, but leaves intact their motivation. It is, in other words, compatible with the (thin) conception of the good Rawls calls ‘goodness as rationality’. It does not assume that the parties, or those represented, have any particular determinate conception of the good, other than that it is a reasonable one, subordinated to the fairness norms of the choice situation. Thus, although the proposal seemingly is similar to the motivational assumption dismissed above, it clearly differs in this respect. The motivational assumption ascribes a determinate, if only partial, conception of the good, which includes care for ones descendants. Even though such an ascription would make the model descriptively more adequate of actually existing people – most of us certainly care about our descendants – it would distort the underlying target problem. The original position should be descriptively adequate of how fair choices are made in the well-ordered society, not of how actual choices are made in our actual society. It is, in this sense, a normative model. To repeat, the theorist’s job is to make explicit common sense normative judgements and connect them in ways which facilitate the justificatory process. That cannot always, or even very often, be carried out by surface-level descriptions of what people actually care about and prefer. A better approach is the constructivist one that informs the development of the account of primary goods. That is, to think about what is in a type situation a salient conception of the person and about what that involves.

Given the intimate connection between normative ideals and conceptions of the person, as stressed by constructivism, we can work from two directions. The savings problem in the original position could be understood as the result of assuming a too sparse conception of the person; and the other way around, reconsidering the conception of the person could facilitate the elaboration of novel normative requirements. With this we get a kind of openness lacking in additional constraint. In addition, we avoid the ad hoc charge, as sustainability of values is independently justified rather than just assumed to solve the modelling problem. The conception of the person that goes with it is not created out of nothing.

The proposal can be elaborated on somewhat by briefly discussing three worries one might have in relation to it. First, the way in which sustainability of values was presented can suggest that it is a claim about all conceptions of the good. That is, a proposal that whatever is valuable to people, they have reason to maintain it over time. But is this really true of all values and of all conceptions of the good? Can we not imagine, say, a hedonist, who only values net pleasure satisfaction, and that such a person could rejoice without being the least worried about or dependent on a future continuation of the world? Although, as noted above, Scheffler also gives us reasons to doubt that hedonistic values are correctly described in this way, this is not my response (see fn. 20). We might well imagine a rational ‘presentist hedonist’ – like the Ancient philosopher Aristippus (Wolfsdorf Reference Wolfsdorf2013, 22) – as one conception of the good in a well-ordered society. But in order for rational hedonists to be reasonable in the Rawlsian sense, they must blind themselves to that particular conception of the good in negotiating principles of justice. Although their own conception of the good may not mandate future savings, they do not know this. But they do know that sustainability of values is a primary good. Thus, the presentist hedonist is not a challenge to the discussed solution to the savings problem in the original position.

The second worry, or rather set of worries, concerns how sustainability of values seems to bring Rawls’s theory closer to communitarian theories (e.g. de-Shalit Reference de-Shalit1995). One might, for instance, be concerned that it amounts to a thick and controversial conception of the good, which would make the theory parochial, or at least less than universally applicable. Or, alternatively, that it makes the theory conservative. The response to such worries can only be sketchy here as there is insufficient space to develop the more concrete discussion of the just savings principle a full answer would require. But note, as a tentative response, that the relevant sense of value neutrality is not jettisoned by the proposal, as was hopefully made clear in the reasoning above. Note also that the Rawlsian theory developed here is on a higher level of abstraction than communitarian theories.Footnote ¹⁹ Sustainability of values summarises reasons to preserve values underlying various practices and projects, but these are not necessarily reasons to preserve any particular culture or tradition that may instantiate them. Because of that the resultant theory cannot be said to be improperly conservative.Footnote ²⁰

The third worry relates to the previous two. Perhaps the arduousness of finding a solution to the savings problem in the original position has been in vain. Joseph Heath (Reference Heath2013) argues that the non-reciprocity problem is not a genuine problem for contractualist theories. We see this once we drop the abstraction of non-overlapping generations and work out the implications of the relevant sense of (indirect) reciprocity, he argues. The easiest way to demonstrate his point – without the game theoretical language in which it is expressed – is using the example he uses consistently, namely that of a pay-as-you-go pension scheme. Such an arrangement is unfunded in the sense that those who currently benefit, the old generations, are paid not by their previous savings but by direct payments from the young generations. The question thus arises for the young generations: Why should they dispense a fraction of their salary to the old generations’ pensions when they cannot expect anything in return from the elderly? The answer comes from understanding the structure of the scheme, how benefits flow ‘upstream’ from younger to older generations. Granted the continuing existence of the system, the young can expect that just as they benefit the old generation, one day they will be the beneficiaries of their future successors. On the assumption that the benefits of the system outweigh the costs, and that it is expected to continue indefinitely, it is rational for each participant to contribute to its maintenance, in spite of the lack of direct reciprocity.

A conclusion from Heath’s argument, one may think, is ‘if it ain’t broke, don’t fix it’. But that would be a jump. Heath’s argument is convincing with regard to existing intergenerational projects, such as pay-as-you-go pension schemes, but not more generally. Once such systems are set up, each participant stands to benefit from the preservation of the system and the lack of direct reciprocity between payment and benefit is no obstacle to that. This is so even on the standard assumption of a narrow conception of self-interests, such as revealed preference, something for which the existence of such systems is proof of. However, it is much less clear how it can account for a more general sense of intergenerational unfairness, which covers also the unfairness of, say, unmitigated greenhouse gas emissions. As there is yet no effective mechanism for abating climate change, it is not in anyone’s narrow self-interest to pay for climate change mitigation. The challenge with respect to many of the most pressing intergenerational problems is to justify why the present generation is morally required to reserve resources, save or invest in the absence of anything like a mechanism for indirect reciprocity. This can be met with what has been proposed above. Sustainability of values is an explication of our self-interest, from which we could learn something that cannot be read straight off of revealed preference, namely that the future matters greatly to us.

Funding

This work was supported by the Swedish Research Council under [grant number 437-2014-6692].