2. Competing Claims Egalitarianism and Prioritarianism

According to Voorhoeve and Fleurbaey (VF) (Reference Voorhoeve and Fleurbaey2012: 397):

Clause (iib) underscores the normative significance of relative levels of well-being. According to CCE, the importance of relative levels derives from their role in distributive justification when there are competing interests. VF (Reference Voorhoeve and Fleurbaey2012: 397) note that inequality lends force to the claims of the worse off when evaluating the justifiability of alternatives. And Otsuka and Voorhoeve (OV) (Reference Otsuka and Voorhoeve2009: 183–4) write:

OVF maintain that one of the virtues of CCE is that it delivers plausible verdicts in contexts involving risk. Other accounts of priority to the worse off, they contend, do not. Specifically, prioritarian views, which take absolute levels of well-being to be the ground of priority to the worse off, are said to face a pair of objections if they apply to both intrapersonal and interpersonal cases involving risk. Versions of prioritarianism that regulate both intrapersonal and interpersonal conflict are called unrestricted, whereas restricted versions of prioritarianism apply only to interpersonal conflict and avoid these objections altogether.Footnote ¹

OVF argue that, with respect to intrapersonal conflict, we can justify, irrespective of absolute levels, imposing the risk of a loss on someone by appealing to the prospect of a greater gain because it is the same individual who faces the risk of loss and prospect of gain. In balancing the prospect of a gain against the risk of a loss for this individual, they argue, we are looking after this person’s interests and gambling on his or her behalf. It is the unity of the individual that makes such balancing appropriate and justifies maximizing expected utility within a life. In contrast, there is no such unity among a group of individuals. Therefore, we should not balance possible benefits and possible burdens across lives in the same way that we do within a life. Because unrestricted prioritarianism sometimes obstructs expected utility maximization in one-person gambles by giving priority to avoiding lower absolute levels in possible outcomes, it is said to ignore the unity of life. And if unrestricted prioritarianism does not ascribe any morally important difference to whether a possible benefit and a possible burden are at stake for a single life or separate lives, then it is not adequately sensitive to the separateness of persons.

These objections are illustrated by a pair of cases modelled after those presented by OVF.

ONE PERSON CASE

The numbers in the boxes indicate levels of cardinal utility. Smith will be exposed to an equiprobable gamble unless a stranger intervenes. Nonintervention maximizes Smith’s expected utility. Per OVF, agents have an undefeated reason to choose nonintervention on Smith’s behalf.

TWO PERSON CASE

Smith and Jones will be exposed to an equiprobable gamble unless a stranger intervenes. In light of the fact that the possible gain and possible loss in the ONE PERSON CASE and TWO PERSON CASE are close in size, OVF argue that these cases should be treated differently. The difference, they claim, is explained by observing that the gamble is justified in the ONE PERSON CASE by the unity of life: we impose a risk of loss on Smith so that Smith might benefit. In the TWO PERSON CASE, however, we impose a risk of loss on Smith not for Smith’s sake but so that Jones might benefit. This tradeoff is not justified by any unity between Smith and Jones, and it ignores the separateness of persons. Smith and Jones have distinct claims that oppose one another regarding exposure to the gamble. According to OVF, Smith’s claim to intervention is stronger than Jones’s claim to nonintervention because Smith would be worse off than Jones no matter what the outcome of the gamble. In contrast, unrestricted prioritarianism, OVF originally charged, delivers the same verdict in both cases. It thereby fails to adequately appreciate the difference between the unity of life and the separateness of persons.

Derek Parfit (Reference Parfit2012) has countered that there are versions of unrestricted prioritarianism that can show sensitivity to prudential justifications in risky contexts and account for the shift in the moral weight of possible benefits and possible burdens when moving from intrapersonal to interpersonal conflict, thereby delivering different verdicts in the twin cases. One such example is Parfit’s hybrid version of unrestricted prioritarianism, which applies prioritarian weighting to both final well-being and to expected well-being (Smith’s expected well-being differs in the two cases). OVF concede Parfit’s points, yet they argue that unrestricted hybrid prioritarianism still fails to adequately respect the unity of life and the separateness of persons: an occasional refusal to do what is in a person’s expected best interests in one-person gambles persists, and there remains an inadequate sensitivity to the existence of competing claims.Footnote ²

Whether OVF are correct about unrestricted prioritarianism is not the subject of this paper.Footnote ³ Instead, this paper presents a largely internal critique of both CCE and HE, in part, by appealing to some of the very norms unrestricted prioritarianism is said to contravene.

3. Critique of Competing Claims Egalitarianism

CASE 1

CASE 1 is like the TWO PERSON CASE in that B has a claim to be exposed to the gamble and C will be worse off than B no matter what the outcome of the gamble. When evaluating cases like CASE 1, we should distinguish the following questions. (1) Is there a pro tanto reason counting against nonintervention? (2) If so, what should an agent do all things considered, given that there is a pro tanto reason counting in favour of nonintervention? My focus here is on (1).

With regard to C, clause (i) of CCE appears to be satisfied. C’s interests are at stake, as his life can go better or worse if he is exposed to the gamble, which suffices to grant C a claim according to CCE. Given that clause (i) is satisfied, advocates of CCE might be tempted to think that there are competing claims in CASE 1 and that inequality between B and C is thus morally significant. To the contrary, I will argue that CCE cannot coherently establish the presence of conflicting interests, competing claims and inequality’s importance in cases like CASE 1.

If B and C have competing claims, then we must be able to identify them either ex ante or ex post. Consider first the ex ante perspective. If there are ex ante competing claims, then one should expect to find an ex ante conflict of interests according to CCE. We can better appreciate C’s ex ante perspective by considering isolated C:

CASE 2

Regarding their preferred measure of utility, VF (Reference Voorhoeve and Fleurbaey2012: 385 n.10) write:

From the perspective of C’s self-interest in CASE 2, the value of C’s chance of a benefit perfectly balances the value of C’s chance of a loss, so the risky alternative and the secure alternative have equal expected utility for C. C could rationally choose nonintervention, and so could someone acting on C’s behalf.

Returning back to CASE 1, C’s prospective gain, prospective loss and baseline utility level are the same as in CASE 2. Thus, what is in C’s self-interest is the same in both cases. The same prudential justification for nonintervention that can be offered to C in CASE 2 can be offered to C in CASE 1. Thus, there is no ex ante conflict of interests between B and C.

One possible response is suggested by the following statement from VF (Reference Voorhoeve and Fleurbaey2012: 397): “On the Competing Claims View, in both our one-person and our two-person intrapersonal case, there is only a claim on Albert’s behalf. It is natural to take this to be a claim for him to be exposed to a sufficiently attractive gamble.” In light of this statement, it might be thought that individuals have an interest in avoiding gambles that are insufficiently attractive, and thus claims against being exposed to such gambles. Because the gamble is not favourable to C ex ante – the possible gain is not larger than the possible loss – it might be argued that the gamble cannot be deemed sufficiently attractive to C.Footnote ⁴ If so, then C’s ex ante interests ground a competing claim against exposure to the gamble in CASE 1.

However, this approach leads to an odd result given OVF’s measure of utility, namely, that we can promote C’s ex ante self-interest – it would be expectedly better for C – if we choose an alternative that has an expected utility identical to that of the only other alternative available. Furthermore, recall CASE 2, in which orthodox decision theory informs us that C, or someone acting on his behalf, should be indifferent between intervention and nonintervention, as the expected utility of each option is equal. Accordingly, someone acting on C’s behalf in CASE 2 would be permitted to expose C to the gamble, per the above remarks from Voorhoeve and Fleurbaey. But if C has a claim against exposure to the gamble in CASE 2 because it is insufficiently attractive, then agents are forbidden from doing that which orthodox decision theory permits in one-person gambles. If OVF endorsed such a claim, then this would render their conception of the unity of life obscure, for equally sized equiprobable gains and losses in one-person gambles mysteriously do not perfectly balance one another. It would also undermine OVF’s ability to wield orthodox decision theory against unrestricted prioritarianism’s treatment of one-person cases, since OVF would be guilty of (sometimes) violating orthodox decision theory’s verdicts for one-person gambles. These difficulties can be avoided if OVF simply deny that C has any claim regarding the alternatives in CASE 2, given that the gamble is neither favourable nor unfavourable to C. But if that is so, then it is sometimes not the case that a claim can be made on a person’s behalf if that person’s interests are at stake, thereby undermining clause (i) of CCE.

It is difficult to see how C’s ex ante self-interest can ground a competing claim in CASE 1 in a manner that coheres with OVF’s other commitments.Footnote ⁵

Let us now consider the possibility that C’s ex post interests are the basis of C’s claim to intervention in CASE 1. Suppose that CCE ascribes to C a competing claim in the possible outcomes in which he is made worse off than B as a result of exposure to the gamble. This ascription does not sit well with OVF’s other remarks. According to OVF, the possible outcomes of a gamble for someone represent possible lives or possible futures of a single, unified person. It is this unity, the individual who faces possible exposure to a gamble, to whom a claim belongs. Assigning claims to an individual’s possible lives appears to fragment the individual and deny his or her unity. It is the former – but not the latter – conception of claims in risky contexts that allows OVF to rebut an argument advanced in support of unrestricted prioritarianism that we treat the possible outcomes in a one-person gamble as if they each contain a claim-bearer.Footnote ⁶ These considerations appear to speak against the attribution to C of a different claim in each possible outcome of the gamble in CASE 1.

Setting this concern aside for now, how might proponents of CCE explicate the notion that B and C have competing claims in the possible outcomes of CASE 1? Recall that in their initial presentation of CCE, OV (Reference Otsuka and Voorhoeve2009: 183) described distributive claims as “claims to a given increment of improvement”. The advantage of characterizing claims in this manner is that it allows CCE to avoid the levelling down objection, unlike telic egalitarianism (OV allege), which assigns distributive claims on the basis of the intrinsic badness of inequality, irrespective of a person’s interests being at stake.Footnote ⁷

It might be objected that OV’s description of distributive claims as claims to a given increment of improvement does not commit them to the view that this description is exhaustive of what a claim can involve.Footnote ⁸ However, the cases OVF use to illustrate CCE all involve claims to increments of improvement. Moreover, it is clear from VF (Reference Voorhoeve and Fleurbaey2012: 397) that clause (iia) of CCE – promoting a person’s interests – refers to individual, noncomparative benefits or harms. Similarly, Otsuka (Reference Otsuka2012: 371) notes that “the competing claims approach has purchase only when there is a choice between alternatives which will either benefit one person or benefit another person”. All in all, it appears that the view of claims as claims to increments of improvement best represents what OVF had in mind when discussing CCE.

Let us return to the TWO PERSON CASE. Are there ex post competing claims in this case?

TWO PERSON CASE

Suppose that State 1 of the gamble obtains. Jones has a claim to this increment of improvement, or put another way, a claim against the alternative choice of intervention, which would have deprived Jones of this benefit. In contrast, Smith does not have a claim to an increment of improvement in this outcome: Smith remains at his baseline level and his level of well-being would be the same if the alternative choice of intervention were made. Hence, there are no competing claims in State 1. Similar reasoning applies if State 2 obtains, although in this outcome, Smith would have a claim to an increment of improvement – a claim in favour of the alternative choice of intervention that would have prevented his loss – while Jones would not have such a claim since he remains at his baseline level and would not have benefited if intervention were chosen instead. Hence, there are no competing claims in State 2. If this is right, then neither state of the world contains ex post competing claims in the TWO PERSON CASE.

A similar result can be derived for CASE 1. For example, State 2 of the gamble in CASE 1 is identical to State 2 of the gamble in the TWO PERSON CASE. And if nonintervention in CASE 1 results in State 1, both B and C – according to the above model of ex post evaluation – have claims against intervention, which would have deprived each of them of a benefit. If the claims of B and C are aligned in State 1,Footnote ⁹ and if State 2 does not contain competing claims, how, then, can there be ex post competing claims in CASE 1 according to CCE?

A closer inspection of clause (iib) of CCE presents further difficulties. According to this clause, “(ii) his claim to have a given alternative chosen is stronger … (iib) the worse off he is relative to others with whom his interests conflict”. We have already established that there is no ex ante conflict of interests in CASE 1 that coheres with OVF’s broader theory, unlike the ex ante conflict of interests in the TWO PERSON CASE. Nor is there an ex post conflict of interests in CASE 1. In State 1 of this gamble, both B and C receive benefits, and the size of C’s benefit is not determined by the size of B’s benefit. This jointly beneficial outcome does not exhibit a conflict of interests. In State 2, C suffers a loss while B remains at his baseline level; specifically, B neither experiences a gain nor avoids a loss in State 2. Relative to intervention, B’s interests are not promoted at C’s expense if State 2 results from nonintervention.

By way of contrast, consider OV’s (Reference Otsuka, Voorhoeve and Olsaretti2018: 78):

ANTI-CORRELATED CASE

Though there is no ex ante conflict of interests between Frank and Gwyneth, there is an ex post conflict of interests between them regarding the choice of alternatives (moving to the city or suburb). Whoever turns out to be disabled, that individual would be better off in the city, while whoever turns out to be nondisabled, that individual would be better off in the suburb. Thus, one person’s interests are promoted at another’s expense in every state of the world in the ANTI-CORRELATED CASE. In CASE 1, however, there is no state of the world in which one person’s interests are promoted at another’s expense. If a gamble results in an ex-post conflict of interests, then we should expect to observe that one person’s interests are promoted at another’s expense in at least one possible outcome of the gamble.Footnote ¹⁰

We have entertained the possibility that C might have different ex post claims in CASE 1 depending on which outcome of the gamble materializes, one claim grounded in an actual benefit to C and the other claim grounded in an actual loss to C. These ex post claims appear to be equal in strength due to the equal size of the gain and loss, so how can consideration of C’s ex post claims determine that an agent should choose intervention rather than nonintervention on C’s behalf? The problem can be framed another way. According to VF (Reference Voorhoeve and Fleurbaey2012: 392), what an individual would have achieved in another state of the world is relevant to the assessment of actual outcomes and the justifiability of choice. In CASE 1, how can what C would have gained in State 1 of the gamble fail to justify his loss in State 2?

Here is one possible explanation. Unlike C’s loss in CASE 2, if C loses in CASE 1, C also becomes worse off than B. A loss that creates or worsens inequality between affected individuals counts for more than a loss of the same magnitude that does not do so, advocates of CCE might be tempted to say. This explains why C’s possible gain – which also occurs in the setting of inequality – fails to justify his loss in CASE 1. Ex ante, C might now be said to have a competing claim against nonintervention. Ex post, C’s claim in State 2 is stronger than the claim C would have had if State 1 materialized.

This proposal, however, conflicts with OVF’s framing of CCE. According to CCE, a conflict of interests and competing claims exists independently of any inequality-weighting, as exemplified in the TWO PERSON CASE and as clause (iib) presupposes. Moreover, the conflict itself renders inequality important according to CCE. From the ex ante perspective, however, the above proposal proclaims that C’s competing claim arises from the fact that C’s possible loss is inequality-weighted and thereby outweighs a possible benefit of equal size to C. Ex post, the proposal holds that inequality renders C’s claim in State 2 stronger than C’s claim in State 1, despite the absence of conflicting interests or competing claims in either state. This proposal presupposes that inequality’s importance is independent of the existence of competing claims, and thus inequality’s importance would not be explained by CCE.

It appears, then, that proponents of CCE cannot coherently show that C’s possible gain fails to justify C’s possible loss or C’s actual loss in CASE 1. That is, they cannot show this without either offending against their conception of the unity of life, partly undermining their critique of unrestricted prioritarianism, or presupposing inequality’s importance independently of conflicting interests. In virtue of clause (i) of CCE, proponents of CCE appear to be committed to ascribing a claim to C in CASE 1. On an ordinary understanding of claims, claims require others to do specific things.Footnote ¹¹ But the claim CCE confers to C is oddly practically inert, for it does not demand the choice of a particular alternative on C’s behalf by agents deciding what to do.

It might be objected that, because each alternative has equal expected utility for C, C’s interests are not at stake and clause (i) of CCE is therefore not satisfied.Footnote ¹² I do not think OVF would endorse this strategy, as it appears to ignore the ex post perspective, a perspective OVF clearly regard as very important, as exemplified by their discussion of the ANTI-CORRELATED CASE. Surely our choice matters for C’s well-being. C’s life will go better or worse unless we intervene. If C loses the gamble in CASE 1 and becomes both intrapersonally and interpersonally worse off, it would be absurd to tell C that his interests were never at stake.

By virtue of the absence of conflicting interests and competing claims, CCE deems inequality in CASE 1 to be morally irrelevant. But consider now:

CASE 3

As in CASE 1, C’s possible gain perfectly balances his possible loss from the perspective of C’s self-interest in CASE 3. There are no competing claims and inequality is irrelevant in CASE 3 according to CCE. Moreover, in CASE 1 and CASE 3, nonintervention offers B the same possible increment of improvement – a 50% chance of gaining 10 units of utility. If inequality is irrelevant in both cases and cannot modulate the strength of claims, B’s unopposed claim appears to be equally strong in both cases. Thus, CCE implies that we would have just as much reason, just as stringent a duty, to expose each person to the gamble in CASE 3 as we would in CASE 1, even if in CASE 3, B is initially well off and C is initially badly off.

Yet some, perhaps many egalitarians will deny that these cases are morally equivalent because they perceive that inequality is morally worse in CASE 3 than in CASE 1. Like CASE 1, CASE 3 involves inequality of risk: C faces a risk of loss while B does not. But unlike CASE 1, CASE 3 involves large baseline inequality and large outcome inequality. There appears to be a weaker pro tanto reason to expose each person to the gamble in CASE 3 than in CASE 1, and a stronger pro tanto reason counting in favour of intervention in CASE 3 than in CASE 1. But CCE denies all of this because it determines that there is separability in both cases: the value of changes in each person’s utility is completely independent of the other person. CCE demands that we ignore comparative considerations in CASE 1 and CASE 3 despite the fact that, in contrast to a one-person case, the interests of separate persons are at stake and their potential futures or fates are linked, that is, despite the fact that what we choose for one person will affect the other person. No argument has been given by egalitarians preoccupied with competing claims as to why such linkage cannot itself make comparative considerations morally relevant.Footnote ¹³ Since CCE is not capable of registering a moral difference in an agent’s reasons for action between CASE 1 and CASE 3, and insofar as this difference appears to be explained by differences in inequality, then CCE does not adequately explain the moral importance of inequality.Footnote ¹⁴ Proponents of CCE simply might insist that inequality is irrelevant in these cases because they believe that the presence of conflicting interests is a necessary condition for inequality’s importance. This latter assertion is not only tendentious, but it is also no longer endorsed by OVF.

4. Critique of Hybrid Egalitarianism

In their later work, OVF embrace another form of egalitarianism, which regards all unchosen, undeserved inequality as intrinsically bad. VF (Reference Voorhoeve and Fleurbaey2016: 943–944) argue that egalitarian concern should encompass unaffected individuals. And OV (Reference Otsuka, Voorhoeve and Olsaretti2018: 68) write: “To us, it is clear that there is unfairness when some are better off than others and that this unfairness is absent when people are equally well off, so that leveling down is in one respect good.” Inequality in final utilities and inequality in expected utilities each has intrinsic disvalue according to OVF’s version of hybrid egalitarianism (HE).Footnote ¹⁵ In what follows, I compare HE with CCE and examine HE’s treatment of risky fixed-identity casesFootnote ¹⁶ in addition to its coherence with the ideals OVF have promulgated.

According to HE, nonintervention in CASE 1 is morally bad in one respect, as it engenders inequality in expected utilities and inequality in final utilities. And per HE, nonintervention in CASE 3 is morally worse than nonintervention in CASE 1 since the former yields greater inequality in expected utilities and final utilities than the latter.

It is worth making explicit some of the other ways in which HE differs from CCE. First, HE deems as intrinsically bad all undeserved, unchosen inequalities obtaining between past, present and future persons.Footnote ¹⁷ When deciding how to distribute aid between two individuals, it appears that HE deems as morally relevant the levels of numerous other individuals, all of whom are unaffected by the distributive problem at hand. Unless there is a principled rationale for excluding these other relations from consideration, then it appears that HE, unlike CCE, encumbers agents with exceedingly onerous information requirements and inordinately complex computation problems whenever distributive decisions must be made. Second, while CCE requires maximizing expected utility in one-person gambles (if there are claims to be exposed to sufficiently attractive gambles), HE implies that agents have strong reasons not to deliberate as if they are dealing with one-person cases, for acting as if there are one-person cases ignores the intrinsic badness of unchosen, undeserved inequalities.Footnote ¹⁸ These inequalities abound in our world and its history. Inequality of this kind is even relevant for decisions an agent might make about how to advance his or her own interests.Footnote ¹⁹ This is not to say that other-regarding egalitarian considerations always trump self-regarding values on this view. The point is that the former is always eligible to be weighed against the latter when deciding what to do. Agents are not entitled to deliberate in purely prudential terms with no regard for the importance of inequality, even if an agent is the last human survivor. CCE rejects this result. Regarding CCE, VF (Reference Voorhoeve and Fleurbaey2012: 397) write: “It is therefore not the case that, on this view, inequality (whether of prospects or of outcomes) is intrinsically bad.”

The point of highlighting these differences between CCE and HE is not to pick a side. Rather, one aim of this section is to underscore the foundational conflict between CCE and HE: the former denies while the latter affirms the intrinsic badness of inequality. If my criticisms of CCE are valid, then that does not itself vindicate HE. But insofar as egalitarians reject my criticisms of CCE and believe that CCE is vindicated, they should also reject HE. In fact, if CCE is affirmed, then HE is susceptible to some of the very criticisms OVF have levied against unrestricted prioritarianism, as I will now show.

OVF note that there are different versions of HE. One subfamily of HE, which OVF favour, employs “expected equally distributed equivalents” or EEDEs to decide between alternatives. The EEDE of each possible outcome of a prospect is determined as follows. In the absence of inequality, the EEDE of an outcome is the sum of its average expected utility at the moment of decision and its average final utility. If, however, there is either inequality in expected utilities or inequality in final utilities or both, then the EEDE of an outcome is the sum of some value that is less than the average expected utility at the moment of decision, if there is inequality in expected utilities, plus some value that is less than the outcome’s average final utility, if there is inequality in final utilities. (OVF do not specify how much less.) The probability weighted sum of these outcome values is computed for each prospect, and agents are instructed to choose the prospect with the highest EEDE.Footnote ²⁰

Consider now the following cases:

CASE 4

CASE 5 (non-equiprobable)

The EEDE of intervention in both cases is 100. The EEDE of nonintervention in CASE 5 is < 99.8, while the EEDE of nonintervention in CASE 4 is < 99. According to OVF’s preferred version of HE, intervention is preferable in both cases, but our reason to eschew nonintervention in CASE 5 is weaker than our reason to eschew nonintervention in CASE 4 – the least choice-worthy option.

However, the gamble in CASE 5 is unfavourable to C. C is significantly more likely than not to become both intrapersonally and interpersonally worse off. C’s risk of loss is neither outweighed nor perfectly balanced by B’s possible gain or C’s possible gain. In contrast, C is unaffected in CASE 4. If it is morally significant when evaluating inequality that a person’s probable loss in a multi-person gamble is neither interpersonally nor intrapersonally outweighed or balanced – as one might have thought based on OVF’s discussion of the TWO PERSON CASE – then HE does not capture this significant difference between CASE 4 and CASE 5.

Moreover, in CASE 5, there are (ex ante) conflicting interests that give rise to competing claims between B and C. But the presence of competing interests/claims seems not to make a moral difference on HE: they play no role in the determination of EEDEs. Yet one would have thought, in light of OVF’s critique of unrestricted prioritarianism, that the existence of competing claims is supposed to make a profound moral difference to the justifiability of choice, one whose acknowledgment is necessary to adequately respect the separateness of persons. If the presence of such competing interests and claims has independent moral significance, then HE may not be adequately sensitive to their presence.

Now, it might be said that HE employs a different conception of distributive claims than CCE. According to this conception, unchosen, undeserved inequality is intrinsically bad because it is unfair. Unequal chances and unequal outcomes are what determine the unfairness of particular alternatives and an individual’s claim regarding those alternatives, independently of welfare being at stake.Footnote ²¹ In CASE 4, for example, C’s claim to fairness competes with B’s claim to beneficence, which is grounded solely in a prospective improvement in B’s well-being. Thus, it might be argued that it is not the case that CASE 4 and CASE 5 differ in virtue of the presence of competing claims.

But according to HE, these competing claims do not explain inequality’s importance, for inequality is intrinsically bad, bad independently of welfare being at stake. HE treats beneficence as a distinct value that often competes with the value of equality. Contrary to CCE, HE does not ascribe any role to claims to beneficence in determining why it matters that some are (or would be) worse off than others. So, it is unclear how, according to HE, the presence of competing claims can render any alternative morally worse or less justifiable than another from the standpoint of inequality and unfairness.

Let us consider further the model of claims attributed to HE, where one kind of distributive claim is completely independent of a claim-bearer’s interests being at stake while another kind of claim is grounded in beneficence. This model implies that in CASE 5, but not in CASE 4, C has two claims against nonintervention: a claim grounded in fairness in light of inequality in chances and inequality of outcome levels, and a separate claim grounded in the fact that his well-being is significantly more likely than not to be negatively impacted from exposure to the gamble. If C has two such claims in CASE 5, then C apparently has more objections to his exposure to the gamble in CASE 5 than C has to B’s exposure to the gamble in CASE 4. Given C’s greater number of objections in CASE 5, it might be thought that nonintervention in CASE 5 should be harder to justify than nonintervention in CASE 4. HE yields the opposite conclusion: nonintervention in CASE 4 is the least justifiable option because it has the lowest EEDE. HE appears to be inadequately sensitive to competing claims, which by OVF’s own standards, suggests that HE fails to adequately respect the separateness of persons.

5. An Alternative Egalitarian View

Taking stock of the argument thus far, we have seen that CCE limits the importance of inequality solely to contexts in which there are conflicting interests, whereas HE has significantly broader scope – it treats all unchosen, undeserved inequality as intrinsically bad. But CCE does not adequately explain why relative levels are morally significant when the interests of separate persons are at stake, and thus CCE’s process of justification is deficient. And HE fails to appreciate the significance of a person’s interests being at stake for adequately respecting the separateness of persons. In light of the shortcomings of these two views, how, then, should egalitarians proceed?

It might be argued that OVF never intended for CCE or HE to stand alone. Instead, CCE and HE should be regarded as complementary components of a broader egalitarian theory or modular principle.Footnote ²² The thought here is that each view can compensate for the other’s deficiencies when combined. One problem with this combined view concerns how to understand the underlying basis of the moral importance of inequality. For one of the components undermines the other. CCE denies that inequality is intrinsically bad, but the intrinsic badness of inequality is a cornerstone of HE. CCE holds that inequality matters only when there are conflicting interests, whereas HE deems inequality to be morally important even when it involves individuals whose interests are not at stake. We don’t seem to have a coherent picture here of why it matters that some are worse off than others.

Perhaps CCE and HE can be integrated in a more cohesive way in the following modular principle: we first dismiss all cases that do not involve competing claims to beneficence, and for those cases that do, only then do we count inequality in expected utilities and inequality in final utilities.Footnote ²³ This modular principle integrates CCE with a version of HE that significantly differs from the view articulated in VF (Reference Voorhoeve and Fleurbaey2016), for it abandons the very broad egalitarian concern VF (Reference Voorhoeve and Fleurbaey2016: 943–944) indicate should include individuals whose interests are not at stake. It renders HE’s application and inequality’s importance conditional on the competing claims framework. But even if OVF were to adopt this modular principle, this still would not address some of the concerns I raised earlier. Since there are no competing claims to beneficence in CASE 1 and CASE 3, this modular principle also determines that inequality is irrelevant in these cases, and thus fails to explain why these cases appear to be morally different.

CCE, HE and the above modular principle do not exhaust the ways in which relative levels of well-being can be regarded as morally important. Egalitarians might combine facets of CCE and features of HE without endorsing either view. This can be done in a way that captures inequality’s importance in CASE 1 and CASE 3, as I will now explain.

Egalitarians, I believe, should reject the idea that unchosen, undeserved inequality is intrinsically morally important. Instead, the significance of lives going equally well is grounded in how we ought to value and respond to persons. Part of this response involves regarding each person as mattering and as mattering equally.Footnote ²⁴ It also involves recognizing that the importance of a person’s interests is rooted in the more fundamental importance of the person who bears those interests.Footnote ²⁵ A person’s interests matter because the person matters, that is, because of a prior value each person is considered to have, and have equally. We have reason to care that one person’s life goes as well as any other because we have a prior reason to equally care – from a moral point of view – about the persons living those lives. Ascribing moral importance to relative levels expresses our equal concern for persons and their well-being.Footnote ²⁶ However, the concern that the lives of persons go equally well should be circumscribed. Egalitarians should endorse the idea that distributive justification is owed to someone, if and only if, that person’s well-being depends on the distributor’s choice.Footnote ²⁷ For those individuals whose well-being we cannot affect, it may be appropriate to feel sympathy for them, but this does not entail that our choices must be justifiable to them. One has no claim that one’s position be taken into account in the distribution of well-being when one has no skin in the game. Combining these ideas suggests the following view. When the well-being of multiple persons depends on what we do, the distributive justification owed to them should count undeserved, unchosen inequality as a relevant factor. Ignoring relative levels in such contexts is inconsistent with a concern that each person’s life goes as well as any other, a concern rooted in the more basic equal concern for persons that we ought to have. What is owed, in part, to individuals is that distributive choices reflect the requisite concern that their lives go equally well when each person’s interests are at stake,Footnote ²⁸ a duty that is discharged by giving appropriate weight to relative levels. None of this implies that the importance of relative levels stems from competing claims to increments of improvement, either ex ante or ex post. That is, egalitarians should embrace a person-affecting version of egalitarianism, i.e. reject HE, and they should also reject CCE as a basis for egalitarian concern.

Though there may be different ways of developing these ideas, one possible egalitarian view counts – perhaps among other things – inequality in expected utilities and inequality in final utilities only when the well-being of individuals depends on what we do.Footnote ²⁹ According to this person-affecting hybrid egalitarianism – PAHE – the moral importance of inequality is not conditioned by the competing claims framework, contra the above modular principle, so PAHE regards inequality in CASE 1 as morally significant. And PAHE deems nonintervention in CASE 3 as morally more objectionable than nonintervention in CASE 1 with regard to inequality, as there is greater inequality in expected utilities and final utilities in CASE 3.Footnote ³⁰ (I am still only discussing the significance of inequality rather than what should be done all things considered.)

Yet PAHE can morally distinguish CASE 1 and CASE 3 from CASE 4, despite the fact that there are unequal prospects and unequal outcomes in all of these cases. PAHE implies that inequality in CASE 4 is morally irrelevant since only one person’s well-being depends on what we do. PAHE ties the moral importance of relative levels of well-being to the demand for distributive justification when each person’s well-being is at stake. Ignoring someone does not amount to maltreatment when we cannot affect this person’s life.

How does PAHE assess inequality in cases in which there are competing claims to beneficence? There are two possibilities here. First, proponents of PAHE might think that competing claims to beneficence do not matter to the assessment of inequality per se; instead, they are pertinent to the all-things-considered justification about what to do. Second, although PAHE does not regard conflicting interests as the ground of inequality’s importance, PAHE nonetheless might treat such conflicts as an aggravating factor, rendering inequality morally more objectionable. PAHE does not preclude this second possibility, as each person’s interests being at stake is essential to inequality’s importance. The manner in which these interests are at stake, e.g. whether they give rise to competing claims to beneficence, might also matter for the assessment of inequality. I leave unresolved here what form PAHE should take.

PAHE has several other advantages over HE. Because PAHE does not count inequalities involving unaffected persons, it does not encumber agents with exceedingly burdensome information requirements and inordinately complex computation problems resulting from the relative levels of countless unaffected individuals whenever distributive decisions must be made. And PAHE preserves as a genuine possibility prudential reasoning unfettered by a concern for inequality. Agents should not count inequality in their deliberations when they can affect only one person, for unchosen, undeserved inequality does not, in itself, provide reasons for action.

Despite these advantages over HE, it might be objected that PAHE is not truly a person-affecting theory. PAHE, in deeming inequality to be morally relevant in CASE 3, for instance, offers at least a pro tanto reason for intervention, which avoids worsening inequality. Intervention in CASE 3, however, appears to amount to a form of levelling down, i.e. doing what is expectedly worse for B and expectedly better for no one. If intervention in CASE 3 were still deemed to be better in some respect, then this violates the principle of personal good for prospects (PPGP): one prospect cannot be better than another if there is no one for whom it is expectedly better. The basis for intervention seems to be non-person-affecting.Footnote ³¹

Even if PAHE recognizes a pro tanto reason for acting in a way that is expectedly better for no one, it does not follow that the grounds for so acting are in no sense person-affecting. A proponent of PAHE might argue that the pro tanto reason for intervention in CASE 3 is grounded in fairness. Fairness is a relational matter, and unfairness in CASE 3 is reflected in B and C’s exposure to unequal risk and C’s becoming even further worse off than B in either outcome of the gamble as a result of changes to both of their well-being. But contrary to the conception of fairness operative in HE, the conception of fairness operative in PAHE requires that each person’s welfare is at stake in the distributive problem at hand. Because C is unaffected by our choice in CASE 4, C is not treated unfairly when we expose B to the gamble. In contrast, the unfairness of nonintervention in CASE 3 stems from how the choice of nonintervention would affect C relative to how it would affect B.Footnote ³²

For those attracted to the language of claims, the concern for fairness in CASE 3 might be articulated in another way: C might be said to have a claim against nonintervention grounded in fairness. If C has a claim to fair treatment here, then this claim would be unlike the claims posited by CCE, for clause (ii) of CCE is not satisfied. A claim to fair treatment counting in favour of intervention is not contingent on demonstrating an ex ante or ex post conflict of interests.

If C has a claim to fair treatment in CASE 1 and CASE 3, then this claim also would not be explained by Broome’s (Reference Broome1990, Reference Broome1991) theory of fairness. Broome (Reference Broome1991: 174) has argued that the importance of (in)equality can be grounded in his theory of fairness. As Piller (Reference Piller2017) helpfully explains about Broome’s theory, all claims are noncomparative and fairness simply mediates between competing noncomparative claims. However, if C has a claim to fair treatment, then C’s claim would be a comparative claim. Tomlin (Reference Tomlin2012) has argued that given Broome’s view of claims as all noncomparative, fairness – which is relational matter – cannot be owed to someone on Broome’s theory. But we can owe it to others to be fair if there are claims to fair treatment.Footnote ³³

Regarding the cases I have presented, PAHE does better than CCE, HE or their combination in accounting for the moral importance of relative levels of well-being, respecting the separateness of persons, and satisfying the ideal of justifiability to each person.Footnote ³⁴ But whatever advantage exists here – someone might object – is only for those already committed to the importance of relative levels and comparative considerations in general. Restricted prioritarians reject such commitments, while still arguing that restricted claim prioritarianism (RP) respects the unity of life, the separateness of persons and justifiability to each person. And prioritarians such as Adler (Reference Adler2012: 314–337) have even argued that fairness is best understood in terms of RP. Egalitarians cannot simply reassert the importance of relative levels against those who see no value in them, so further argument is required to adjudicate between PAHE and RP.Footnote ³⁵

6. Critique of Restricted Prioritarianism

CASE 1

Suppose that all of the most prominent versions of RP – Williams (Reference Williams2012), Adler (Reference Adler2012) and Nebel (Reference Nebel2017) – would ascribe a claim to B in CASE 1. According to these versions of RP, C’s exposure to the gamble in CASE 1 is morally no more concerning in any respect than C’s exposure to the gamble in CASE 2, and C lacks grounds for complaint in either case.

CASE 2

Restricted prioritarians would deny that C has any claim regarding the alternatives in CASE 2: from the perspective of C’s self-interest, agents should be indifferent with respect to whether or not C is exposed to the gamble, according to orthodox decision theory, which Williams, Adler and Nebel all endorse. It is difficult to see how, on a prioritarian view, C could acquire a claim to intervention in CASE 1 simply because of the addition of B, for whether an individual possesses a claim regarding exposure to a gamble depends only on facts about that individual’s well-being, according to prioritarians.Footnote ³⁶ If there were a new reason to oppose exposing C to the gamble in CASE 1 that was not present in CASE 2, then this appears to be due to C’s relation to B, which violates a core prioritarian tenet of separability.Footnote ³⁷

Consider next the following remark from Adler and Holtug (Reference Adler and Holtug2019: 21): “Morality is a framework for resolving interpersonal conflicts … In cases of interpersonal conflict … the separateness of persons comes into play and motivates prioritarian weighting.” If Adler and Holtug mean that prioritarian weights only apply in cases of interpersonal conflict, then this implies that prioritarian weights do not apply in CASE 1. Similarly, Williams and Nebel argue that interpersonal conflict is a necessary condition of prioritarian weighting, which would commit them to making the same judgement about CASE 1. But recall:

CASE 3

If prioritarian weights do not apply in CASE 1 because there is no interpersonal conflict of interests, then prioritarian weights also would not apply in CASE 3 for the same reason. In the absence of prioritarian weighting, one would have the same decisive reason to expose B and C to the gamble in CASE 3 as one would have in CASE 1, as B’s unopposed claim in each case is grounded in the same unweighted possible benefit – a 50% chance of gaining 10 units of utility. Given that B’s claim is equally strong in both cases, RP implies that the stringency of the corresponding duty of nonintervention would be the same in CASE 1 and CASE 3, even though in CASE 3, B is initially well off and C is initially badly off. This is an odd form of prioritarianism, for within interpersonal gambles, RP regards providing a possible benefit to someone who is very well off – B in CASE 3 – to be just as important as providing the same possible benefit to someone who is significantly less well off – B in CASE 1. And it regards an actual loss to a badly off person – C in CASE 3 – to be morally no worse than an actual loss of the same size to a significantly better off person – C in CASE 1. If the separateness of persons is so crucial for the justification of prioritarian weighting according to proponents of RP, then it is unclear why there should be no concern for how well-off people are when the interests of separate persons are at stake and their fates depend on what we do.Footnote ³⁸

If prioritarian weighting is triggered only when there is an interpersonal conflict of interests, then “the strength of our reasons to benefit a person can depend on whether and how our act would affect other people”.Footnote ³⁹ In CASE 3, for example, the strength of B’s claim depends on how our act would affect C. For if the gamble were slightly unfavorable to C in a modified CASE 3, other things being equal, then B’s claim would count for less than in the original CASE 3, in virtue of a relation of conflict between B and C. It appears, then, that all of the versions of RP considered here violate a core prioritarian tenet of separability.Footnote ⁴⁰

Unrestricted prioritarians undoubtedly will regard RP’s denial of a moral difference between CASE 1 and CASE 3, as well as RP’s violation of separability, as objectionable.Footnote ⁴¹ Restricted claim prioritarians simply might bite the bullet like their claim egalitarian counterparts and deny that differences between CASE 1 and CASE 3 have any bearing on an agent’s reasons for action. They might insist on the necessity of there being a conflict of interests in order for prioritarian weighting to obtain. It is not, however, clear why such a condition should be deemed necessary for a purportedly nonrelational distributive theory. Resolving interpersonal conflict is an important function of morality, but to my knowledge, no argument has been given by competing claims prioritarians or competing claims egalitarians as to why it should be morality’s only function.Footnote ⁴²

The demand that distributive weights or values only apply in the presence of conflicting interests seems less plausible an idea in the case of RP than in the case of CCE. RP in particular, and the competing claims framework in general, fail to exhaust the significance of fairness. Nonintervention in CASE 3 involves greater unfairness than nonintervention in CASE 1, and this difference is explained by PAHE, not RP.