1. Introduction
Among other things, egalitarian theories assess when and why distributive inequalities are objectionable. How should egalitarians assess inequalities between generations? This important question in intergenerational ethics is underexplored by egalitarians.Footnote 1 This article helps to fill this gap.
I defend the following claims.
First, distributive egalitarianism is the view that, other things being equal, equal distributions in some good are intrinsically better than unequal distributions (see section 2). Proponents of distributive egalitarianism typically focus on distributions between contemporaries. I argue that, when applied across generations, distributive egalitarianism significantly loses intuitive plausibility (section 3).
Second, this result undercuts the intuitive case for distributive egalitarianism. Attempts to discount or exclude intergenerational inequalities from the theory's scope are unsuccessful (sections 4 and 5). And because it must always include all past and future people, distributive egalitarianism can no longer invoke the intragenerational judgements that made it attractive in the first place (section 6).
Finally, I suggest we should shed distributive egalitarianism and stick with instrumental and relational reasons against distributive inequality instead: such reasons likely suffice to capture egalitarian intuitions, including in intergenerational cases, yet have none of the problems produced by distributive egalitarianism (section 7).
2. Distributive egalitarianism
2.1. What is distributive egalitarianism?
Distributive egalitarianism holds that, other things equal, equal distributions of some good are better than unequal distributions.
Telic distributive egalitarianism holds that – beyond any potential instrumental disvalue – unequal distributions are intrinsically bad, other things being equal (Parfit Reference Parfit1997).Footnote 2
Telic distributive egalitarianism can be non-institutional or institutional. Non-institutional views hold that inequality between two persons is intrinsically bad even if they are not connected through any institution or, indeed, any relationship whatsoever. Institutional views, in contrast, hold that distributive inequality is only intrinsically bad between persons who share in a common practice or institution. For example, non-institutional views would hold that when people live in completely separated parts of the world – so-called Divided World Cases – any distributive inequality between them is still intrinsically bad. Institutional views would deny this.
In what follows, I use ‘distributive egalitarianism’ as a shorthand for telic, non-institutional, distributive egalitarianism. I only briefly return to institutional views below.
In my discussion, I make several simplifying assumptions.
First, I do not discuss the currency of distributive egalitarianism.Footnote 3 It could be about distributions of goods that contribute towards wellbeing, such as resources, capabilities or opportunities, or about distributions of wellbeing directly. In sections 2–5, the numbers used in examples represent wellbeing or something similarly substantive like quality of life.
Second, I set aside issues involving choice, responsibility and luck. Although not discussed directly, my arguments still matter for luck egalitarianism, a species of distributive egalitarianism that judges inequalities as intrinsically bad only when they are due to luck but not choice.Footnote 4 My examples assume people are neither responsible for being worse off nor independently deserve to be.
Third, while distributive egalitarianism is about the goodness or badness of outcomes, I do assume that axiology also generates normative reasons for actions.Footnote 5
Finally, in my examples, I exclude risk and uncertainty and assume populations are fixed across outcomes.Footnote 6 These assumptions are unrealistic, of course. But discussions around risk, uncertainty and population ethics involve separate thorny philosophical puzzles. Isolating separate problems should help us make philosophical progress.
2.2. The intuitive case
I here focus mostly on the intuitive case for distributive egalitarianism which I take to be:Footnote 7
The Intuitive Case: distributive egalitarianism is a necessary part of any plausible egalitarian theory because it best captures egalitarian intuitions that would otherwise go unaccounted for.
We can make the intuitive case in the abstract. Consider two contemporaneous populations and two outcomes, O 1 and O 2. There is an equal number of people in each population and no inequality within populations (Table 1).
There are two outcomes and two contemporaneous populations. Distributive egalitarianism implies that O 2 is better than O 1.
Distributive egalitarianism would prefer O 1 over O 2, which seems intuitively plausible.
Less abstract intuitions also support the Intuitive Case:
Nagel's Move: a couple must decide whether to move to the city or the suburbs. The suburb would benefit their child without disabilities more than the city would benefit their child with a disability. Distributive egalitarianism seems to capture that parents ought to give greater weight to the interests of their disabled child (Nagel Reference Nagel1979: 123–24; also see Parfit Reference Parfit1997).
Disability: luck egalitarianism seems to capture that, intuitively, it is unjust when persons with disabilities for which they are not responsible are distributively worse off than people without disabilities (Arneson Reference Arneson1989; Cohen Reference Cohen1989). It is unjust that some persons have, through no fault of their own, fewer opportunities by birth.
Health Inequality: when we distribute scarce medical resources, many people think we should not just maximise the unweighted sum of aggregate health benefits (e.g. maximise the sum of quality-adjusted life years) but also consider health inequalities and how well off or badly off patients are relative to each other. Distributive egalitarianism seems to capture these concerns (Eyal Reference Eyal2018).
3. Intergenerational distributive egalitarianism
In some intergenerational cases, distributive egalitarian judgements seem far less intuitive than those surveyed above. Parfit early on asked whether we should really be moved by 13th century Inca peasants being much worse off than we are today (Parfit Reference Parfit1991: 7). Consider here a similar worry applied to three non-overlapping generations:
Distributive egalitarianism would judge O 2 to be pro tanto better than O 1, because wellbeing is more equal across generations. Of course, distributive egalitarians tend to be pluralists: distributive equality is not the only property of distributions we care about. Importantly, we should also care about how much aggregate or average wellbeing there is. So, in Table 2, distributive egalitarians would say that while we have egalitarian reason to prefer O 2, we should likely prefer O 1 all things considered, given the greater wellbeing it contains. (This explains the ‘other things equal’ rider in my definition of distributive egalitarianism.)Footnote 8
There are two outcomes with three generations. Distributive egalitarianism judges that we have pro tanto egalitarian reason to prefer O 2 over O 1.
But, still, a reader might even find this pro tanto judgement odd.
First, imagine you are in G 3 and could bring about O 2 rather than O 1 by making yourself worse off. It seems intuitively surprising that you should have any reason to make yourself worse off without benefiting anyone, particularly so when you cannot retroactively make deceased generations better off.
Second, imagine alternatively that you are in G 2. It seems intuitively surprising that you should have any reason to make future generations worse off just so their level is closer to yours yet no one benefits from this reduction. As Simon Caney writes:
… could we really insist on equality if we could leave future people better off at no extra cost to those alive now? […] Must we – in the name of intergenerational equality – stem the flow of benefits? That seems very unappealing. (Caney Reference Caney, Kanbur and Shue2018: 167)
Finally, distributive egalitarianism is both forward-looking and backward-looking. In its ranking, inequalities between us and future people matter just as much as inequalities between us and past people. This symmetry between backward- and forward-looking comparisons might seem intuitively surprising.
Again, some distributive egalitarians respond that O 1 is indeed better than O 2, but only pro tanto. Shlomi Segall, for example, argues that intertemporal levelling down is no worse than contemporaneous levelling down (Segall Reference Segall2016). After all, he maintains, there is something regrettable about Inca peasants being so much worse off than we are.
But consider now a variation of the levelling down objection to which the pluralist response is less powerful. In this variation, we split our current generation into two equally large groups, G 3.1 and G 3.2. Within these groups, everyone has equal wellbeing.
Imagine you are in G 3.2 and could change the outcome from O 1 to O 2 and thereby reduce G 3.1's wellbeing from 70 to 50. Now, pluralist distributive egalitarianism might advise you to stay in O 1 to preserve greater aggregate wellbeing (or prevent a Pareto-inferior outcome). But the pluralist response is now less powerful than in Table 2. Because now distributive egalitarianism would tell you that, as far as your egalitarian reasons are concerned, you should bring about O 2, as that would reduce the inequality between generations G 1, G 2 and G 3. And given that previous generations were so much worse off, intergenerational equality should take priority over intragenerational equality. This judgement seems counterintuitive as an egalitarian judgement. Intuitively, it does not seem very egalitarian to recommend levelling down the worst-off people alive today to reduce inequalities with poor dead people.
Distributive egalitarians can respond that current inequalities matter too. The egalitarian prescription is not obvious, because intergenerational inequality must be balanced with intragenerational inequality. But here we could fill in the story a little more. Instead of just two previous generations, we can add many more generations before G 1 with low wellbeing levels until the numbers have intergenerational inequalities outweigh intragenerational inequality. The sheer number of deceased poor people could swamp our egalitarian concern with the contemporary poor (at least in my example). Again, such a judgement does not seem intuitive as a specifically egalitarian judgement. Admittedly, the argument works against most but not all (conceivable) measures of inequality. We could imagine measures of equality that would avoid it (although they might yield counterintuitive judgements elsewhere).Footnote 9
In my next example, all intergenerational equality measures agree. In this example, I split each generation into two equally sized groups. The groups are different each time. For simplicity, also assume they are not related to the previous generation to exclude concerns around social mobility and such.
If we treat intergenerational and intragenerational equality alike, we could rank O 1 and O 2 by assorting all wellbeing levels of each outcome into one distribution and then calculate the inequality therein. Doing so, a pluralist intergenerational egalitarian would be indifferent between O 1 and O 2, as they are identical and thus also identical in terms of wellbeing and inequality. But if we are in the business of ‘capturing egalitarian intuitions’, O 2 intuitively seems better as an egalitarian outcome.
Objections like Table 2 have been raised before. And distributive egalitarians have given (brief) responses to them. But to date there has been no critical discussion of how plausible such responses are. A central contribution of this article is to show that the intergenerational problem is in fact more damning than realised. To do so, I show that neither of the following two responses given on behalf of distributive egalitarianism works.
First, some distributive egalitarians would endorse Supplementation. Distributive egalitarianism is not a comprehensive axiological let alone moral theory. Once we supplement it with other considerations, we would typically have good reason against sacrificing intragenerational for intergenerational equality.Footnote 10 For we often have instrumental and relational reasons to care more about inequalities between contemporaries and, even more so, between people who relate to one another (see section 7 for such reasons). But in sections 6 and 7 I argue that Supplementation fails to save distributive egalitarianism: adding it to relational and instrumental reasons against distributive inequality is unnecessary and imports counterintuitive implications.
Second, instead of Supplementation, other distributive egalitarians would respond by modifying distributive egalitarianism. As Temkin writes:
[M]any people have a ‘bias’ toward the present and future. … For an egalitarian, [such a bias] may involve either rejecting or ‘discounting’ the value of equality within past generations, or between past generations and other generations (be they past, present, or future). Though rarely recognized or explicitly acknowledged, I suspect many egalitarians would incline toward such a view. Surely, one might think, it is not morally objectionable that most of those in our society are much better off than the vast majority of those who lived in previous centuries, at least not to the same extent that it is objectionable that some members of our society are much better off than others… (Temkin Reference Temkin1995: 99)
Let us try to spell out Temkin's ideas.Footnote 11 I first take up the idea of discounted distributive egalitarianism. Afterwards, I discuss distributive egalitarianism restricted to intragenerational inequality or forward-looking inequalities.
4. Discounted distributive egalitarianism
Discounted distributive egalitarianism applies a temporal discount factor on how strongly we weigh egalitarian concerns. Distributive inequalities matter most between people living at the same time and they matter more between generations the closer in time they are to each other. Discounted distributive egalitarianism includes a concern with intergenerational inequalities whilst also placing greater weight on contemporaneous inequalities. Accordingly, in Table 3, it might rank O 1 over O 2 on egalitarian grounds to preserve intragenerational equality. In Table 4, it would rank O 1 over O 2, because O 1 contains intragenerational inequality. If our discount factor is high enough, the badness of intragenerational inequality would overpower that of intergenerational inequality.
There are two outcomes with three generations with the third generation split into equally sized groups. Most versions of distributive egalitarianism judge O 2 to be the more egalitarian outcome.
There are two outcomes with four generations with each generation split into equally sized groups. Distributive egalitarianism treats intergenerational and intragenerational inequalities alike and thus remains indifferent between O 1 and O 2.
Note two features of such egalitarian discounting.
First, I assume the exponential discounting framework standardly used in economics. One implication is that, for some far-future generation, the egalitarian discount rate will weigh in so heavily that any inequality between us and them is not intrinsically bad anymore.
Second, I only apply a discount factor to the egalitarian considerations but not to other considerations like wellbeing. Most philosophers argue that applying a ‘rate of pure time preference’ on wellbeing is implausible.Footnote 12
However, this combination meets with two challenges.
First, our reasons against a pure time preference might also speak against discounting egalitarian concerns. And some egalitarians do invoke arguments against pure discounting in support of undiscounted intergenerational egalitarianism (Segall Reference Segall2016; Vrousalis Reference Vrousalis, González-Ricoy and Gosseries2017).
Second, discounted distributive egalitarianism treats time and space as structurally different. In intragenerational comparisons, distance is seen as an arbitrary factor that does not annul egalitarian reasons (Caney Reference Caney2006). For intergenerational comparisons, in contrast, time gradually reduces egalitarian concerns. How can we justify this structural difference?
Maybe there is a good response to both challenges. But, in any case, discounted distributive egalitarianism also yields implausible judgements. Compare three generations G 1 (our generation), G 2 (the next non-overlapping generation), and a generation GX so far into the future that the egalitarian discount rate would have removed any egalitarian reasons (Table 5).
There are two outcomes and very many generations. Discounted distributive egalitarianism judges that O 2 is the more egalitarian outcome, as inequalities between G 1 and G 2 matter but comparisons with G x do not.
For discounted distributive egalitarianism, the inequality between G 1 and G 2 matters. As they are temporally contiguous, the egalitarian concern will only be mildly discounted. But the inequality between G 1 and GX does not matter at all (because of the exponential nature of discounting). We thus have to prefer O 2, because there is less discounted inequality overall, and average wellbeing is equal in O 1 and O 2.Footnote 13 However, it seems implausible that, once you think intergenerational inequality matters intrinsically, it should matter strongly between G 1 and G 2 but then not at all between G 1 and GX, particularly considering GX is so much worse off than G 1.Footnote 14 Implausibly, discounted distributive egalitarianism tells us to prefer O 2 over O 1 on egalitarian grounds.Footnote 15
We should reject discounted distributive egalitarianism.
5. Time-restricted and forward-looking distributive egalitarianism
Temkin's other suggestion was that some egalitarians might simply hold that distributive inequality is only bad between contemporaries but not across time (Temkin Reference Temkin1995: 74). Call this time-restricted distributive egalitarianism. If such a time restriction is plausible, distributive egalitarianism could keep plausible intragenerational intuitions, like Nagel's Move and Health Inequality, yet avoid intergenerational judgements that seem less intuitive.
However, for several reasons, time-restricted distributive egalitarianism is not plausible.
First, since Sidgwick, several philosophers have argued that favouring a distribution of value along one dimension – such as space, time, people, or states of the world – gives you reason to favour such a distribution across other dimensions too. For example, Sidgwick argues that we ought to be impartial about distributions of wellbeing across our lifetime and that this – with some debatable bridge premises – gives us reason to be impartial about interpersonal distributions (Sidgwick Reference Sidgwick1907: III. xiii; IV. ii).Footnote 16 Similarly, if you think intragenerational inequalities are intrinsically bad, you at least have prima facie reason to think intertemporal inequalities are intrinsically bad too. Of course, such Sidgwickian arguments are not uncontroversial, so I do not just assume they work. But I think they create an explanatory burden: we require a good reason as to why we should treat one dimension so differently from another.
Second, distributive egalitarianism boasts a robust commitment to equality. It holds that inequality is intrinsically bad even outside of relationships, institutions, and so on. We cannot easily sidestep such a commitment through what might, upon closer inspection, turn out to be status quo bias or vested interest (and ‘ideology’ in the Marxist sense). For example, distributive egalitarianism would urge us to consider inequalities at the global level, because geographical location and place of birth are morally arbitrary (Caney Reference Caney2006). If distributive egalitarianism indeed has such a broad scope, inequality should intuitively also matter between those who live at different times. For when one is born, and not just where, is morally arbitrary too (Schuppert Reference Schuppert2011: 305; Segall Reference Segall2016).
Finally, a more abstract – and somewhat less central – argument puts metaphysical pressure on simultaneity. Special relativity implies there is no absolute simultaneity. This is a problem for time-restricted egalitarianism which relies on absolute simultaneity. Imagine two populations P 1 and P 2, with P 1 living on earth and P 2 living on a superfast spaceship. P 2's speed is much faster relative to P 1's. We could conceive of a situation in which the two populations overlap only a little bit and from only one frame of reference. So, from one population's reference frame, the two populations will be contemporaneous. For the other population, they will not. What would be a just distribution between P 1 and P 2? According to the time-restricted view, this question admits of an answer from one reference frame but not the other.
Distributive egalitarians could try to resist some of the above arguments.
First, distributive egalitarians might become (even) more pluralist. They could say we should adopt an egalitarian principle for intragenerational distributions yet a different principle for intergenerational distribution. For example, they might add a prioritarian or sufficientarian principle for intergenerational cases.
While feasible, this answer is insufficient to defend time-restricted distributive egalitarianism.
First, Occam's Razor prima facie speaks against it. Other things equal, a theory of distributive ethics is preferable, the broader its plausible scope. For example, questions of distributive ethics come up for both intra- and intergenerational questions – both in academic and non-academic discourse. Both questions fall within the plausible scope of a theory in distributive ethics.
But the more damaging problem with this pluralist response, besides Occam's Razor, is that it would still require a justification for its time restriction. Why is it that equality matters intrinsically so much between contemporaries but becomes irrelevant between generations? Without a principled answer that goes beyond intuition-fitting – and that directly addresses the above objections – my above argument against time-restricted distributive egalitarianism stands.
Someone suggested to me that time-restricted egalitarians might draw on the Non-Identity Problem (Parfit Reference Parfit1984: 16). Actions that change intergenerational distributions will typically also impact who will come into existence. Given the well-known complications in population ethics, distributive egalitarians might want one principle for contemporaneous comparisons and another for variable, non-identical populations. Distributive egalitarianism would thus impose a person-affecting scope restriction.
However, this response fails.
First, the person-affecting restriction would not neatly map onto the time restriction, as they only overlap contingently. Imagine intergenerational cases where changes in intergenerational payoffs do not change the composition of future generations (as I have done in my previous discussion). The connection between intergenerational distributions and variable populations is merely contingent and thus unsuitable as a principled ground for the time restriction.
Second, the person-affecting restriction would still allow comparisons with past generations. And it would be a weird intergenerational egalitarian view indeed that considered all past inequalities but ignored those with future generations.Footnote 17
Overall then, we should reject time-restricted distributive egalitarianism.
The final option for distributive egalitarians – also mentioned by Temkin – is what I call forward-looking distributive egalitarianism: we include intragenerational and (undiscounted) comparisons with future generations but exclude comparisons with past generations. Prima facie, this view does not seem arbitrary: when concerned with egalitarian distributions, it seemed counterintuitive to compare ourselves with past people; at the same time, it might make sense to focus on people whose wellbeing can still change.Footnote 18
However, forward-looking distributive egalitarianism also fails.
First, it fails to provide an answer to my earlier examples. For example, remember Table 2:
According to forward-looking distributive egalitarianism, generation G 3 has no egalitarian reason to level down. However, imagine again that G 2 has the power to level down G 3's wellbeing: forward-looking egalitarianism still gives pro tanto reason to level down future wellbeing. Or consider again Table 4:
No generation except for the last has a reason to prefer intragenerational over intergenerational equality.
Second, forward-looking distributive egalitarianism leads to temporal inconsistencies. Consider (Table 6):
There are two outcomes and three generations with the third generation in O 2 split into two equally sized groups. Each generation can switch outcomes. Forward-looking distributive egalitarianism results in temporally inconsistent judgments.
Assume each generation can choose either the top or bottom row wellbeing, independently of what the previous generation has chosen. The time at which we assess distributions will impact which one is judged as most egalitarian. For example,
At t 1: [G 1 at 1,000; G 2 at 1,000; G 3 at 1,000.06]
At t 2: [G 2 at 1,000.1; G 3 at 1,000.06]
At t 3: [G 3 at 3,000]
Judged at t 1 – and from an ‘atemporal’ perspective that aggregates forward-looking inequality across all generation – [1,000; 1,000; 1,000.06] is the most egalitarian distribution. Yet when the time comes to choose the outcomes that make up this distribution, forward-looking egalitarianism would prefer other choices. For example, forward-looking distributive egalitarianism would tell G 2 that 1,000.1 is the most egalitarian choice and tell G 3 that 3,000 is the most egalitarian choice.Footnote 19
Forward-looking distributive egalitarianism is not a plausible view.
6. Supplementation
So, if distributive egalitarians can neither discount nor exclude inequalities across generations, it needs to treat inequalities across temporal locations as equally intrinsically bad.
Now, as I mentioned earlier, distributive egalitarians might accept this conclusion but then provide a second response, Supplementation: when supplementing their theory with instrumental and relational reasons against distributive inequality, they avoid the counterintuitive implications explored in section 3.Footnote 20 I say more about what such reasons are below, but here are some quick examples: economic inequalities between contemporaries might come with problems like hierarchy and domination, worse political institutions and envy. Those problems do not apply to intergenerational inequalities. Therefore, distributive egalitarians can explain why contemporary inequalities – at least in things like income, power and so on – are worse than intergenerational inequalities.
Combining distributive egalitarianism with instrumental and relational egalitarian reasons is entirely possible. However, in this and the following section, I argue that Supplementation ultimately fails to save distributive egalitarianism, because the combination is not plausible. Specifically, I argue that (i) adding distributive egalitarianism to the mix – alongside relational and instrumental reasons – is unnecessary: distributive egalitarianism does not help capture any egalitarian concerns that would otherwise go unaccounted for. Moreover, (ii) adding distributive egalitarianism to the mix is counterproductive: relational and instrumental reasons are more plausible on their own.
The argument for (ii) is straightforward: we have seen that distributive egalitarianism must always make egalitarian comparisons by looking at distributions for all people that have ever lived and will ever live. Such comparisons produce counterintuitive intergenerational implications. For example, distributive egalitarianism might imply that we should sometimes sacrifice some degree of aggregate wellbeing and intragenerational equality for the sake of inequalities between us and people thousands of years before or after us. Moreover, contrary to what some distributive egalitarians have assumed, they can neither ‘discount’ nor ignore intergenerational comparisons. So, if my arguments are correct, adding distributive egalitarianism means adding counterintuitive implications. Other things equal, a combination with distributive egalitarianism is thus less plausible than one without.
The remainder of this article defends (i): adding distributive egalitarianism is not only counterproductive but also unnecessary: what is plausible about it can be captured by relational and instrumental reasons alone.
To do so, consider the following response to my claim above: distributive egalitarians can respond that – besides its idiosyncratic judgements about cosmic inequalities – distributive egalitarianism also adds many intuitive judgements that would otherwise not be captured. For example, remember the three cases from earlier. In Nagel's Move, distributive egalitarianism seems to capture that parents should give greater weight to the interests of their child with disabilities. Similarly, luck egalitarians hold that it is unjust that some persons have, through no fault of their own, fewer opportunities when they are born with a disability. Finally, distributive egalitarianism might capture egalitarian intuitions in public policy. For example, we should not just maximise the unweighted sum of aggregate health benefits but also consider health inequalities (Eyal Reference Eyal2018). So, the response is that unless we add distributive egalitarianism to the combination, egalitarianism fails to capture these intuitive intragenerational judgements.
However, I think this response fails. What I call The Infectiousness Objection holds that once we appreciate its intergenerational implications, distributive egalitarianism actually fails to capture intragenerational intuitions. When describing intragenerational cases, distributive egalitarians suggest that when we make a decision that affects persons A and B, say the children in Nagel's Move, we must pay close attention to the comparative outcomes between A and B. However, once you accept that all people matter – past and future, near or far, dead or alive – and that they all matter equally, the comparison between A and B becomes unimportant. Billions of people lived in the past, billions live elsewhere in the world, and billions or more are still to be born. Moreover, the difference in wellbeing levels between them and A and B might be much larger than the difference between A and B. So, if the parents in Nagel's case wanted to apply distributive egalitarianism to their decision, they must focus on how their decision will affect cosmic intertemporal distributions instead of the comparative standing of their children. Therefore, the intuitive story distributive egalitarians tell about such cases is unavailable to them. Cosmic intergenerational distributions swamp any comparative concern between the two children.
Or consider health inequalities within a country: imagine you distribute Japanese health care resources and think the distribution should, in part, reduce inequalities in life expectancy (e.g. between social groups). Distributive egalitarianism would not capture why such differences matter. Previous centuries contained far more people with far lower life expectancy. Moreover, more people are still to come in Japan and in all other countries and, if the optimists turn out right, their life expectancy might be even higher. So, ultimately, distributive egalitarianism gives us no good reason to be particularly worried about differences in life expectancy within Japan.
The intergenerational case is thus more damaging for distributive egalitarianism than first meets the eye: it also undercuts the general intuitive case in its favour.Footnote 21 Adding distributive egalitarianism does not help us capture egalitarian intuitions.
A distributive egalitarian reader suggested the following response to the Infectiousness Objection: while true that distributive egalitarianism cannot capture our intuitions in real-world conditions, it captures our intuitions in thought experiments. In such thought experiments, we focus only on limited comparative assessments and exclude everyone else. For example, in Nagel's Move we simply stipulate a model without any other people. There are only four people that have and will ever live: the parents plus the child with and the child without the disability. For Health Inequality, we stipulate that there are no past and future people and no people outside of Japan. For those models, distributive egalitarianism gives us the answer that we should prioritise the child with the disability and those currently worse off in Japan.
However, this move fails: while distributive egalitarianism can make plausible judgements in unrealistic thought experiments, these judgements have next to no bearing on any of the real-life cases our intuitions are actually about.
First off, these stylised thought experiments are not models of real-world situations. To make them relevant to real-world cases, we would have to reintroduce all past, present, and future people, which would again swamp whatever judgement distributive egalitarianism produced in the stylised model. Accordingly, intuitively correct judgements in such stylised models are largely irrelevant for real-life cases.
Second, because they lack ‘external validity’, stylised models no longer capture our intuitions, and thus no longer make for strong arguments. What made distributive egalitarianism attractive was that it seemed to capture real-world egalitarian concerns. But such an argument is weakened if distributive egalitarianism only provides a justification for prioritising the disabled person in practically irrelevant thought experiments but not in the real-world cases of health and disability we are interested in and to which our intuitions apply.Footnote 22
Another distributive egalitarian responded to me admitting that, yes, distributive egalitarianism cannot capture intragenerational intuitions but does capture other intuitions: it is intuitive to think that it is bad (or unjust) that Inca peasants were worse off than we are or that some far-future people might be better off than we are (or relatedly that aliens in other galaxies might be better or worse off than we are). If nothing else, distributive egalitarianism captures these intuitions.
Of course, it's hard to argue with someone's intuitions, particularly if you do not share them. But, as a matter of methodology, when distributive egalitarians report having such intuitions, I do not think it does that much to support the Intuitive Claim. Intuitions about cosmic injustice between distant generations are not widely shared – even various distributive egalitarians do not share them (like Temkin and Caney above). They seem more like the intuitions one adopts after having been a distributive egalitarian for a while. Compare them with intragenerational intuitions like Nagel's Move, Health Inequality, and Disability. Such intragenerational intuitions seem like credible ‘gateway intuitions’ into distributive egalitarianism, as they are more widely shared. If they held up, they would significantly add to the case for distributive egalitarianism, including to people not yet convinced of its truth. None of this seems true of the purely cosmic intuitions. Therefore, the intuitive case for distributive egalitarianism is seriously weakened if solely built on cosmic inequality intuitions instead of more credible intragenerational ones.Footnote 23
7. What should egalitarians believe (about intergenerational inequalities)?
I have argued that when we take intergenerational inequalities seriously, distributive egalitarianism loses much of its intuitive plausibility: its extension to intergenerational inequalities generated unappealing implications and threatened its intuitive case built on intragenerational intuitions. Adding distributive egalitarianism to an ‘egalitarian package’ is (i) unnecessary and (ii) counterproductive. In this section, I further defend (i) and suggest that an alternative egalitarian view captures our intuitions around intergenerational inequalities without any of the counterintuitive implications produced by distributive egalitarianism.
Before doing so, note two alternative views I do not discuss.
First, in response to my discussion, we could reject egalitarianism across the board and look for non-egalitarian alternatives, such as prioritarianism, maximin or sufficientarianism.Footnote 24
Second, we could opt for institutional instead of non-institutional distributive egalitarianism. On this view, distributive inequalities are intrinsically bad but only between persons who share in a practice or institution (Tan Reference Tan2008).Footnote 25 One challenge for such views is to explain why inequality is intrinsically disvaluable within such institutions but not outside of them. If inequality is what one is worried about, what explains this ‘moral discontinuity’ as Liam Murphy calls it (Murphy Reference Murphy1998)? Moreover, when we encounter serious inequalities outside our institutions, should we not sometimes create better institutions to tackle such inequalities? Does the reach of existing institutions not itself become a question of egalitarian justice? Institutional distributive egalitarianism is a live option and simply asking challenging questions does not amount to a proper critique. However, a thorough discussion, let alone critique, of it is beyond my scope.Footnote 26
Instead, I now continue on with the third option: egalitarians can do without distributive egalitarianism entirely, including in intergenerational cases, if they attend to the relational and instrumental reasons against inequality.
7.1. Non-intrinsic egalitarianism
In his article What Should Egalitarians Believe?, Martin O'Neill defends what he calls ‘non-intrinsic egalitarianism’ (O'Neill Reference O'Neill2008). Roughly, he rejects distributive egalitarianism but holds that various ‘non-intrinsic’ reasons against distributive inequality sufficiently capture egalitarian concerns. Moreover, he briefly discusses inequalities across borders and argues that the value question (‘why value equality’) also provides a suitable answer to the scope question (‘between whom does inequality matter?’).
Much recent theorising gives reasons against distributive inequality that chime well with O'Neill's non-intrinsic egalitarianism. Here are some central arguments.
First, increasingly, egalitarians connect the instrumental disvalue of distributive inequality with the rich empirical work on inequality in psychology and the social sciences (Scanlon Reference Scanlon2018; Schmidt and Juijn Reference Schmidt and Juijn2024; Venkatesh Reference Venkatesh2023: 7). Stark distributive inequality will bring about worse outcomes, even when outcomes are judged from non-egalitarian standpoints. First of all, economic income and wealth have decreasing marginal value for individuals. For example, reported life satisfaction goes up with increasing income, but the increase is logarithmic. In expectation, doubling any one person's income leads to the same increase in life satisfaction (around 0.3 points on the life satisfaction scale, according to Stevenson and Wolfers (Reference Stevenson and Wolfers2013: 14)). For example, we could increase Peter's annual income from $100,000 to $200,000, and expect his life satisfaction to go up roughly as much as doubling Sarah's income from $5,000 to $10,000. So, if we had an additional $100,000 to distribute per year, giving it to people with an income of $5,000 each, we would, in expectation, add twenty times more life satisfaction than giving it all to Peter.
Note that one can make such instrumental arguments against inequality without being a utilitarian. Even non-consequentialists mostly hold that we have a pro tanto duty to prevent the bad and promote the good (Barrett Reference Barrett2022). Moreover, it is not just aggregate wellbeing that might matter in a non-comparative sense. For example, distributing economic goods more equally, we might help more people meet their basic needs (O'Neill Reference O'Neill2008).
The above reasons against inequality hold even between people who do not share any institutions or relations. Other instrumental reasons, in contrast, work through relationships and institutions. For example, research suggests egalitarian societies tend to exhibit better somatic and mental health, higher levels of trust, less crime, higher life satisfaction and other good things. Moreover, simply having less than others likely has negative psychological effects and can lead to social exclusion.Footnote 27 Finally, stark distributive inequality might lower the quality of political institutions (Acemoglu and others Reference Acemoglu, Naidu, Restrepo and Robinson2015; Bartels Reference Bartels2018). Much research in social science debates whether correlation here indicates causality or something else. Without settling individual disputes, I think that aggregating our credences across different positive effects still justifies a strong pro tanto case for reducing inequality.
Finally, some relationship-mediated reasons against inequality are inherently egalitarian (O'Neill Reference O'Neill2008). Relational egalitarianism has recently emerged as the main competitor to distributive egalitarianism.Footnote 28 Relational egalitarianism holds that institutions should be such that individuals relate to one another as equals. The point of equality is not to establish equality in some good but to prevent undesirable inegalitarian relationships and, more positively, help build a society in which individuals can relate to one another as equals (Anderson Reference Anderson1999). Iris Marion Young, for example, urges us to eradicate domination and the ‘five faces of oppression’: exploitation, marginalisation, powerlessness, cultural imperialism and violence (Young Reference Young1990: 2). With renewed interest in republicanism, domination in particular has received much attention (McCammon Reference McCammon and Zalta2018).
Now, while relational egalitarianism does not consider distributive inequalities intrinsically bad, it still supplies good reasons against them. Stark distributive inequalities can stand in the way of relational egalitarian goods (Elford Reference Elford2017; Moles and Parr Reference Moles and Parr2019; Schemmel Reference Schemmel2011). Sometimes such a connection can be direct and straightforward. Tim Scanlon, for example, argues that when public institutions simply have no reason to distribute a good unequally, only an equal distribution of some good would preserve that citizens relate to one another as equals and are treated as such (Scanlon Reference Scanlon2018: 2). For example, race, income or gender should typically not determine the quality of policing services one receives. In other cases, economic inequalities translate into relational inequalities more indirectly. For example, stark economic inequality can often bring with it unequal political influence and thereby stand in the way of citizens relating as political equals (Bartels Reference Bartels2018; Scanlon Reference Scanlon2018: 6).
The above non-intrinsic arguments taken together provide a strong pro tanto case against distributive inequality: in standard empirical conditions, we should typically prefer less distributive inequality (other things being equal).
Let us now return to intergenerational distributions.
7.2. Suggestions for intergenerational inequalities
The non-intrinsic egalitarian case does not imply that inequalities in wellbeing, as described in the matrices in section 4, are intrinsically bad. Still, when we move on to consider distributive inequalities in other goods, such as economic income and wealth, non-intrinsic egalitarianism is likely enough to capture central egalitarian concerns and intuitions about distributive inequalities, including in intergenerational cases.
-
(i) Levelling down
I suggested that egalitarians should not want to level down existing people's wellbeing to reduce inequalities with past generations. Non-intrinsic egalitarianism gives us no reason for doing so, as such inequalities are not intrinsically bad. Moreover, levelling down goods like income do not generate any instrumental value for past generations. We cannot, for example, transfer our resources to Inca peasants. For the most part, non-intrinsic egalitarianism will ignore backward-looking distributive comparisons.
-
(ii) Intragenerational inequalities matter in ways that inequalities between non-overlapping generations do not
Distributive egalitarianism somewhat counterintuitively judged that, as egalitarians, we must sometimes prioritise inequalities with past generations over inequalities between existing people (Table 3, e.g.).
Non-intrinsic egalitarian arguments, in contrast, include goods with a contemporaneous bias. For example, consider relational goods like non-domination, the absence of status hierarchies and democratic equality. Those typically work through our contemporaneous relations or are most important to us when they do. Or consider that status anxiety and social comparisons can reduce our wellbeing. Such comparisons are typically strongest within contemporaries (and even more so within groups that relate geographically, culturally and so on). Accordingly, non-intrinsic egalitarianism captures that intragenerational equality intuitively matters in a way that intergenerational inequalities do not.
-
(iii) Non-intrinsic egalitarianism includes a plausible concern for the future
However, does non-intrinsic egalitarianism conversely care enough about future people? Here are some suggestions that it will.
First, one reason to worry about unequal distributions was purely instrumental: distributive inequality might lead to deprivation and bring about outcomes that are bad in non-egalitarian terms (e.g. because there is less wellbeing or because basic needs are not met). This argument can apply across generations too. Therefore, equality's instrumental value exerts egalitarian pressure on intergenerational distributions.Footnote 29
Second, non-intrinsic egalitarianism shows why we should care about distributive inequalities within generations and groups. By extension, we should also care about future intragenerational inequalities. For example, most models to assess climate change policies typically focus on average (or aggregate) consumption in future generations. Non-intrinsic egalitarianism implies this is a mistake. We should also consider how a policy would affect how consumption is distributed within future generations. For example, maybe allowing too much climate change would disproportionately harm the global poor and exacerbate within-country inequalities. Non-intrinsic egalitarianism urges us to take those into account too, for both instrumental and relational reasons. And recent work suggests such egalitarian concerns would significantly affect policy recommendations, for example, on how to calculate the social cost of carbon (Dennig and others Reference Dennig, Budolfson, Fleurbaey, Siebert and Socolow2015).
Finally, our relations towards future generations can change depending on how close we are to them and on how far generations overlap. According to non-intrinsic egalitarianism, such relations can affect the extent to which distributive inequalities matter between generations. For example, we might not really stand in cooperative relations with far-future generations but we clearly do with people in overlapping generations. Moreover, such relationships can affect us psychologically and affect how we conceptualise our relations to others. For example, a common narrative around Germany's pay-as-you-go retirement system is that of a ‘generational contract’ (Generationenvertrag). While partly rhetorical, such language does suggest that relational equality has some purchase for cooperative undertakings across overlapping generations. Again, for egalitarians this is a plausible verdict. We intuitively care more about distributive inequalities across overlapping generations – and to some extent across close but not overlapping generations – in ways that do not apply to inequalities between generations thousands of years apart from each other.Footnote 30
Overall, non-intrinsic egalitarianism gives us good reasons against distributive inequality while likely also producing plausible judgements in intergenerational cases. Of course, I have only gestured at rather than spelled out such intergenerational arguments. Moreover, relational egalitarianism goes beyond the distributive concerns discussed here. Andreas Bengtson, for example, argues that relational egalitarianism implies we must not treat future generations paternalistically (Bengtson Reference Bengtson2019). Still, the above makes an initial case that invoking the recent wealth of non-intrinsic arguments against distributive inequality might be enough to capture plausible egalitarian judgements, including in intergenerational cases. Accordingly, shedding distributive egalitarianism is not a serious cost.Footnote 31
8. Conclusions
In this article, I have enquired how egalitarians should think about inequalities across generations. I have defended the following claims:
First, distributive egalitarianism loses significant intuitive plausibility when applied across generations.
Second, attempts to discount or exclude intergenerational inequalities are unsuccessful: distributive egalitarians must weigh all comparisons equally.
Third, combining distributive egalitarianism with relational and instrumental reasons does not salvage it. Such a combination is possible but not plausible. Adding distributive egalitarianism means adding counterintuitive intergenerational judgements. It is also unnecessary. Because intergenerational comparisons tend to swamp intragenerational ones, distributive egalitarianism can no longer capture intragenerational intuitions. Finally, relational and instrumental arguments against inequality likely suffice to capture intergenerational egalitarian concerns without encountering the problems produced by distributive egalitarianism.
Overall, it seems egalitarians are better off without distributive egalitarianism. For a plausible egalitarianism across generations, my money is on relational and instrumental concerns.
Acknowledgements
I would like to thank Jacob Barrett, Andreas Mogensen, Kasper Lippert-Rasmussen, Andreas Bengtson, Marc Fleurbaey, and two anonymous referees for helpful written comments on earlier versions of this paper. I would also like to thank audiences at the Centre for Ethics at the University of Zürich, my colleagues at the Philosophy Faculty during a Griph colloquium talk, and participants at the Normative Theory Group workshop at the University of Hamburg for helpful comments and discussions.
Competing interests
None.
Andreas T. Schmidt is a Professor of Moral and Political Philosophy at the Faculty of Philosophy and Centre for PPE at the University of Groningen.
Open access
