Equality versus Priority

Equality versus Priority 1 Michael Otsuka and Alex Voorhoeve Here we discuss two leading theories of distributive justice: egalitarianism and prioritarianism. We argue that while each has particular merits and shortcomings, egalitarian views more fully satisfy a key requirement of distributive justice: respect for both the unity of the individual and the separateness of persons. The argument proceeds as follows. In Sections 1 and 2, we introduce egalitarianism and prioritarianism, respectively, and apply these views to cases in which we are certain how things would turn out if we choose one way or another. We then introduce cases of risks which deprive us of such certainty. Sections 3 through 5 analyse various forms of prioritarianism where decisions are made under conditions of risk and raise objections to each of them. Section 6 develops the most plausible version of egalitarianism under risk and explains why it is less vulnerable to these objections than prioritarian views. 1. Egalitarianism under Certainty There are many forms of egalitarianism. Social and political egalitarianism holds that material and social inequalities are bad when and because they undermine individuals’ ability to live as equal citizens who are willing to offer and abide by fair terms of social cooperation (Norman, 1998; Anderson, 1999; O’Neill, 2008). It follows that social and political egalitarians regard inequalities as problematic when they lead to domination. One example is the political domination that arises when large inequalities in wealth lead to the control of media by an elite; another example is the power asymmetry in the workplace that occurs when the worst off are dependent on those who are better off for a minimally decent existence. Social and political egalitarians also object to stigmatizing differences in status, such as exist or have existed between men and women, aristocrat and commoner, or Brahmin and Dalit. Finally, on this view, inequalities are bad when and because they give rise to particular morally problematic attitudes. These include servility, envy, and a lack of self-respect among the worst off and arrogance and a jealous guarding of relative advantage 1 We are grateful to Thomas Hurka, Benjamin Lange, Serena Olsaretti, and Asbjørn Schmidt for comments and to Jerod Coker for research assistance. 1 among the better off. Besides often being objectionable in themselves, these attitudes undermine citizens’ attachment to an ideal of social cooperation among persons who owe each other a justification for their common institutions (Tawney, 1964: 37-38). In this chapter, we shall set aside these objections to inequality, important though they are. For they are not the topic of dispute between egalitarians and prioritarians. Instead, this dispute revolves around a further question: Is inequality objectionable beyond the ways in which it threatens the aforementioned egalitarian social and political ideals? We shall begin our attempt to answer this question with a case that is closely modelled on one that Thomas Nagel presents in his famous essay “Equality” and which has played a key role in subsequent discussions of equality and priority. Two-Child Case with Certainty. Imagine that you are the parent of two young teenagers, Ann and Ben. Ben has recently been diagnosed with a condition that will soon give rise to a serious physical disability, but Ann has been given a clean bill of health. You have recently lost your job, but have offers in two places. You must choose whether to take a city job and move your family to cramped urban accommodation in an unpleasant neighbourhood with mediocre schools or to take a job on the outskirts of town and move your family to the open spaces of a suburb with excellent schools. Either option would be equally good insofar as your own wellbeing is concerned. But they would not be equally good insofar as the well-being of each of your children is concerned. If you move to the city, able-bodied Ann, who loves nature and sports, will be hemmed in and have only a mediocre education, but soon-to-be-disabled Ben will have access to special medical facilities that will somewhat, but far from wholly, alleviate the effects of his disability. If you move to the suburb, Ann will flourish, but Ben will not receive this treatment. Moreover, if you move to the suburb, the boost in well-being to Ann will be greater than the boost in well-being to Ben if you move to the city. Table 1 represents the relevant levels of lifetime well-being for your two children. 2 2 These numbers are to be interpreted as follows: an incremental improvement in well-being of a given size always does as much good for the person (has as much prudential value for the person) no matter the level of well-being from which this increment arises. Moreover, for every individual, 0 is a quality of life such that, from the perspective of the good of the person living that life, it is a matter of indifference that she lives that life or 2 Table 1. Final utilities for the Two-Child Case with Certainty Ann Ben City 60 39 Suburb 70 30 Considering only the well-being of your two children (that is, bracketing the level of well-being of others, yourself included), what should you choose? Utilitarianism calls for the maximization of the sum total of well-being (utility). 3 So a utilitarian would opt for the suburb in this case. However, it strikes many, including Nagel, that one should move to the city for Ben’ sake. Nagel would maintain that one has egalitarian reason to move to the city, which overrides the utilitarian grounds one has to move to the suburb in this case. As he writes: “If one chose to move to the city, it would be an egalitarian decision. It is more urgent to benefit the second child. … This urgency is not necessarily decisive. It may be outweighed by other considerations, for equality is not the only value. But it is a factor, and it depends on the worse off position of the second child. An improvement in his situation is more important than an equal or somewhat greater improvement in the situation of the first child” (1979: 124). According to the principle of equality to which Nagel alludes here, it is in itself bad if some are worse off than others (through no choice or fault of theirs). This badness is above and beyond the bad effects of inequality. The badness resides in the inequality itself, so to speak. The badness is the unfairness of some being less well off than others through no fault or choice of theirs (Cohen, 1989; Arneson, 1997; Temkin, 2001). 4 (We shall assume throughout that no one is responsible for her level of well-being.) never existed at all. (In this, we follow Adler, 2012: 219-220.) We also stipulate that 100 is an uncommonly good life. We say more on the measure of well-being when we introduce risk in Section 3. 3 4 In this chapter, we shall treat 'well-being' and 'utility' as synonyms. It is instructive here to draw a contrast with the badness of having less than enough to survive. The badness of having less than enough to survive often accompanies severe inequality. But this badness is in no way 3 The principle of equality doesn’t explain why it is better for all to be equally well off rather than equally badly off. It therefore requires supplementation. One such supplementation involves an appeal to a principle according to which it is in itself better if people are better off. In other words, an egalitarian should be a pluralist who combines a belief in equality with a belief in the importance of improving people’s utility. (We shall discuss one such form of pluralist egalitarianism in Section 6.) In Nagel’s two-child case, it would appear that equality overrides total utility, since the gain in equality of a move to the city seems more important than the loss in total utility of such a move. One gains only one unit of total utility by moving to the suburb, yet one very substantially reduces the inequality between the two children. Notwithstanding his concern for improving well-being, a pluralist egalitarian is subject to what has come to be known as the levelling down objection (McKerlie, 1984: 232; Parfit, 1995: 17-18). Suppose that Ann would, due to natural causes, develop the same severe disability as Ben. The pluralist egalitarian’s concern for Ann’s well-being might lead him to regard such levelling down as bad all things considered. Nonetheless, he is committed to the claim that it is in one way better, because fairer, if Ann and Ben are equally badly off because equally impaired. The levelling down objection holds that there is nothing good at all in such levelling down, and that egalitarianism is therefore false. Some egalitarians regard this objection as without merit (see, for example, Temkin, 2003: 67-68). We share this view. 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 levelling down is in one respect good. Others, however, regard it as a powerful objection. (This is true even of some once attracted to egalitarianism; see, for example, Arneson, 1999: 232-233.) Those who are so moved may be on the lookout for an alternative to egalitarianism to account for the conviction that you should move to the city in Nagel’s case. Prioritarianism is one such alternative. 2. Prioritarianism under Certainty constituted by the relation of inequality. It consists of the non-comparative fact of not having enough to live on in absolute terms. 4 In fact, as Derek Parfit points out, such an alternative to egalitarianism might be found in the very quotation of Nagel’s with which we introduced egalitarianism. In his re-interpretation of Nagel, Parfit writes that this passage “contains the idea that equality has value. But it gives more prominence to another idea. Nagel believes it is more important to benefit the child who is worse off. That idea can lead us to a quite different view” (1995: 19). On this different view, helping Ben is more important, not because one can improve his well-being by a greater increment (one can’t, in this case), but because an increment in well-being is morally more valuable the lower (in absolute terms) the level of well-being from which this increment arises. The idea that an improvement in a person’s well-being that arises from a lower absolute level is morally more weighty (and should therefore have priority over an equally large improvement in a person’s well-being that arises from a higher absolute level of well-being) is at the heart of the Priority View. Prioritarianism is not subject to the levelling down objection. On the Priority View, it is in no way better if Ann were to become impaired, since the only change in value, on this view, would be that (morally weighted) welfare is lost. Nothing is gained. Prioritarians are therefore insensitive to the elimination of the unfairness of some being less well off than others through no fault of theirs. This insensitivity to unfairness highlights the fact that the Priority View is not an egalitarian view. As Parfit explains: “Egalitarians are concerned with relativities: with how each person’s level compares with the level of other people. On the Priority View, we are concerned only with people’s absolute levels” (1995: 23). By contrast: “...we [prioritarians] do not believe in equality. We do not think it in itself bad, or unjust, that some people are worse off than others. That is what makes this a distinctive view” (1995: 22). Parfit offers the following illuminating analogy, to illustrate the non-comparative nature of prioritarianism: 5 “People at higher altitudes find it harder to breathe. Is this because they are higher up than other people? In one sense, yes. But they would find it just as hard to breathe even if there were no other people who were lower down. In the same way, on the Priority View, benefits to the worse off matter more, but that is only because these people are at a lower absolute level. It is irrelevant that these people are worse off than others. Benefits to them would matter just as much even if there were no others who were better off” (1995: 23). In sum, one can characterize prioritarianism as the conjunction of the following three claims: (Diminishing Marginal Value): Each person’s utility has diminishing marginal moral value; (Separability): The moral value of (an increment in) a person’s utility does not depend on how anyone else fares; (Maximization): We ought to maximize the sum-total of the moral value 5 of the utility of persons (which is their priority-weighted utility) (McKerlie, 1984; Parfit, 1995; Adler, 2012; Broome, 2015). 6 5 Here and elsewhere, when we speak of 'moral value' and its maximization, we do not mean to imply that prioritarianism is necessarily a form of consequentialism, according to which the fact that it would maximize moral goodness or value provides the explanation of the rightness of an act. Maximization is consistent with the affirmation of a non-consequentialist (e.g., contractualist) explanation of the rightness of the maximization of moral value. It is also consistent with the view that the magnitude of moral value is simply a representation of the strength of our reasons for action. 6 While Parfit’s 1995 Lindley Lecture is the locus classicus for the Priority View, a view with these characteristics has a prior history in both welfare economics and moral philosophy, though it took some time for it to be recognized as distinct from egalitarianism. In economics, Serge-Christophe Kolm (1969) and Anthony Atkinson (1970) independently formulated a social welfare function according to which each individual’s income has diminishing marginal moral value and which sums this value across individuals (thereby respecting Separability). Their view is distinct from utilitarianism because they appeal not to the diminishing marginal individual utility of income but rather to its diminishing marginal social (or moral) value. However, while the Kolm-Atkinson social welfare function can therefore be regarded as prioritarian, they refer to it as a form of pluralist egalitarianism. Amartya Sen (1973: 39) argued that in order to clarify the Kolm-Atkinson function’s special concern for those who are badly off, one should take individual utility (rather than income) as the object that has diminishing marginal moral value. He also argued that egalitarians need not accept Separability, because they may be concerned with how some fare relative to others. John Broome (1989) went further, arguing that egalitarians should reject Separability precisely because it implies a lack of concern for how some fare compared to others. In moral philosophy, Paul Weirich (1983) was the first to propose a view which respects Diminishing Marginal Value, Separability, and Maximization. (He did so apparently unaware of the work by Kolm, Atkinson and Sen on social welfare functions of this kind.) Weirich described his view as “a compromise between 6 Parfit argues that many people who may take themselves to be egalitarians because they are especially concerned with those who are worse off than others are really prioritarians. Nagel, he suggests, is one such person. Nagel says that it is “more urgent” to benefit the disabled child in the two-child case with certainty. Parfit asks: “Would it be just as urgent to benefit the handicapped child, even if he had no sibling who was better off? I suspect that, on Nagel’s view, it would. Nagel would then, though using the language of equality, really be appealing to the Priority View” (1995: 26). Parfit’s comment raises the following general questions: When considering the fate of one person in isolation from the fate of others, what sort of benefit should we provide this person? And how, if at all, do situations in which the fate of only one person is at issue differ morally from cases in which the fates of more than one person are at issue, and in which the interests of these people may conflict? In order to answer these questions, we turn to cases involving risk. equality and utility” but failed to note that Separability is in tension with standard egalitarianism. Dennis McKerlie (1984), however, did distinguish two “different forms of egalitarianism”, one of which cares about reducing inequality (and therefore rejects Separability) and the other which accepts Separability along with Diminishing Marginal Value and Maximization. McKerlie argued that while the latter avoided the levelling down objection, it was vulnerable to a different objection: in holding that utility has diminishing marginal moral value even in one-person cases, it implausibly requires individuals not to accept expectedly advantageous gambles (1984: 235). McKerlie also noted that in making the moral value of a person’s utility independent of whether the distributional choice one faces involves trade-offs within a life or across different lives, the view does not “attach special importance to the difference between [separate person’s] lives” (1984: 233). (These observations lie at the heart of the criticism of the Priority View we formulate in Section 3 above.) It appears that Larry Temkin (1983: 232-234) was the first to argue unreservedly that a view which respects Diminishing Marginal Value, Separability, and Maximization (which he refers to as “extended humanitarianism”) is not an egalitarian view, given its lack of concern for how some fare relative to others. Nonetheless, he argued that it might be worth endorsing such a view in addition to egalitarianism. Parfit’s contribution was to sharpen the contrast between egalitarianism and prioritarianism and to argue that prioritarianism is superior to egalitarianism. (We are grateful to Jerod Coker for research assistance on the origins of prioritarianism in economics and philosophy.) 7 3. Final-Utility Prioritarianism Consider the following one-person case, which is a transformation of Nagel’s two-child case into a case of a single child who has an equal chance, so to speak, of turning out as either of Nagel’s two children (Otsuka and Voorhoeve, 2009: 188; Otsuka, 2015: sec. I). One-Child Case: You have a single early-teenage child, Cathy, who is now healthy but who has a 50% chance of developing the aforementioned disability and a 50% chance of remaining healthy. Before you know how her health will develop, you must now decide whether to take the job which necessitates a move to the city, which would enable her to receive specialist treatment if she were to develop the disability, or to take the job in a suburb, which will lead her to thrive if she is able-bodied. Due to economic circumstances, these moves are permanent—if, say, you have taken the job in the suburb, you will not be able to later move to the city in the event that Cathy develops the disability. Table 2: Final and expected utility in the One-Child Case. Final utility Expected utility p = 0.5 Cathy able p = 0.5 Cathy disabled City 60 39 49.5 Suburb 70 30 50 In considering this case, it is necessary to make further assumptions about the measure of utility we are employing. We shall assume a measure of utility on which a prospect has higher expected utility for a person just in case it would be preferred for that person’s sake after rational and calm deliberation with all pertinent information while attending to her self-interest only. (A person’s expected utility is just the probabilityweighted sum of her utility in each possible state of the world.) One prospect has the same 8 expected utility as another for a person just in case such deliberation would yield indifference between the two prospects. 7 Moreover, in order to establish what the Priority View requires in this case, we must determine how it deals with risk. One prioritarian approach is to maximize expected priority-weighted final utility (Rabinowicz, 2002; Adler, 2012). That is, one maximizes the probability-weighted sum of priority-weighted utilities in each possible state of the world. We shall refer to this as “final-utility prioritarianism.” On this view, you must choose the city in the One-Child Case, for the same reason you must do so in the Two-Child Case with Certainty. In the latter, final-utility prioritarianism justifies the move to the city on the grounds that improving a person’s well-being from 30 to 39 is more morally valuable than improving a person’s well-being from 60 to 70. The greater moral value of the improvement from 30 to 39 implies that moving to the city will have greater expected moral value in the One-Child Case. Two features of final-utility prioritarianism stand out. First, because of the way in which it distinguishes moral value from personal (or prudential) value, final-utility prioritarianism holds that one must sometimes choose an option that is contrary to the expected interests of the only person whose well-being is at issue (McKerlie, 1984; Rabinowicz, 2002; McCarthy, 2008; Otsuka and Voorhoeve, 2009). Second, it treats some risky intrapersonal trade-offs in which only one person’s interests are at stake as involving the very same moral calculus as interpersonal trade-offs in which the interests of different people conflict (McKerlie, 1984; Otsuka and Voorhoeve, 2009). Both aspects are problematic. First, in the One-Child Case, moving to the suburb uniquely maximizes Cathy’s expected utility. Given our assumptions, someone solely concerned with Cathy’s interests would therefore prefer it on her behalf. This means that no 7 In other words, we assume that the measure of utility is derived from idealized preferences satisfying the Von Neumann-Morgenstern axioms. (For discussion of how assuming this rather than another measure of utility determines the plausibility of prioritarianism, see Greaves, 2015 and Otsuka, 2015.) Note that this measure does not presuppose any particular view on what utility is. One might believe that two options have the same expected utility for a person just in case she would, if ideally rational and self-interested, be indifferent between these options without also believing that utility consists of preference satisfaction. One might maintain that utility consists of something other than preference satisfaction and hold that the specified idealized preferences fully track the magnitude of this other thing (Otsuka and Voorhoeve, 2009: 172-173, n. 3). More generally, throughout, we assume that orthodox decision theory applies, according to which under risk, a decision-maker ought to maximize the expectation of what he takes to be the relevant value (so that a utilitarian ought to maximize the sum-total of expected utility, a final-utility prioritarian the sum-total of expected priority-weighted utility, etc.). 9 matter how things turn out, if you were to opt for the suburb, you would be able to offer her the following prudential justification: I had to balance a 50 percent chance of you thriving (rather than leading a cramped existence and facing mediocre schooling) if you were healthy against a 50 percent chance of your having access to specialist medical care (rather than not having such access) in case you developed a disability. I balanced these two potential effects on your well-being from the perspective of your self-interest alone. From this perspective, the expected value of the benefits of the suburb outweighed the expected value of the benefits of the city. I therefore chose the former. In so doing, I did the best I could for you, given the information I had at the time. We believe this gives you strong reason to choose the suburb. Moreover, when you consider Cathy's prospects in isolation from how well off anyone else is, you have, in our view, no countervailing reason to move to the city. We conclude that when one so brackets others’ well-being, it is at least permissible to move to the suburb. More generally, contrary to final-utility prioritarianism, in risky intrapersonal trade-offs in which inequality is not an issue, it is reasonable to accord equal moral weight to equally large increases in a person’s utility, independently of the baseline from which these increases take place—i.e., to maximize her expected utility rather than to give priority to her fate if she turns out to be worse off. 8 Indeed, when only one person’s well-being is at issue, it is curious to decide what to do for this person, not on the basis of the personal value of increments in her wellbeing, but instead on the basis of their priority-weighted value, where this value is understood to be wholly impersonal in nature. Such cases highlight that a prioritarian who is committed to maximizing impersonal value is committed to choosing contrary to the prudential interests of everyone concerned, even when inequality is not at issue. 9 8 For an extended argument for this conclusion, see Otsuka (2015). This conclusion has been much debated. For concurring opinions, see McCarthy (2008); Otsuka and Voorhoeve (2009); Williams (2012); and Rendall (2013). For dissenting views, see Rabinowicz (2002); Crisp (2011); O’Neill (2012); Parfit (2012); Porter (2012); Segall (2014); and Bovens (2015). For a review of the debate, see Weber (2014). 9 The observation that the Priority View is committed to choosing an alternative that is expectably better for no one in such cases has led some to argue that its supporters cannot wield the levelling down objection 10 Second, it is objectionable to apply the same moral calculus in risky intrapersonal trade-offs and interpersonal trade-offs. Cathy’s two potential futures are unified because they are both potential futures of hers. This unity makes it appropriate to balance the potential advantages and drawbacks of each alternative to her from the perspective of her interests. 10 That is why, in making risky decisions that affect a person alone, we naturally speak of ‘choosing for her sake’ and why, if you move to the suburb, you can offer Cathy a prudential justification for this move, no matter how things turn out. By contrast, if you move your two children to the suburb, no comparable prudential justification can be provided to the disabled child. This is because moving to the suburb was never in his interest, as only a different person (his healthy sister) could ever possibly have benefitted from that. In other words, in the Two-Child Case with Certainty, the separateness of persons makes it harder to justify forgoing the gain to Ben, who will be worse off in any case, in order to provide a slightly greater gain to Ann. In applying the same priority weights to possible increments in utility in both cases, final-utility prioritarianism therefore fails to respect the difference between the unity of the individual and the separateness of persons. 11 We emphasize that this second criticism of final-utility prioritarianism is distinct from the first. That is, one can disagree with our judgment that it is permissible to maximize Cathy’s expected utility but still object to final-utility prioritarianism’s failure to track whether a trade-off takes place within two possible futures of the same individual or across the boundaries between people. By way of illustration, consider the following case (Voorhoeve and Fleurbaey, 2012). against egalitarians (Segall, 2014). For further discussion of whether prioritarianism is vulnerable to some form of the levelling down objection, see Persson (2001), Brown (2003), and Porter (2011). For further discussion of the apparently impersonal nature of the Priority View, see Persson (2001). For a statement of the implausibility of a purely impersonal case for prioritarianism, see Otsuka (2015: sec. VI). 10 We assume that the children in question are in their early teens and that the disability in question is merely physical in order to ensure that Cathy can safely be regarded as the same person in both futures. 11 Empirical studies which compare people’s attitudes towards intra- and interpersonal trade-offs generally find a substantial difference between the two, with subjects giving a great deal more weight to improvements in utility for the worst-off in interpersonal trade-offs than in intra-personal trade-offs (see, e.g. Ubel et al., 1996; Nord et al., 1999; however, see Bognar, 2012 for criticism of this claim about the empirical literature). These studies therefore reveal that many people’s judgments respect the difference between the unity of the individual and the separateness of persons. 11 Intra- versus Interpersonal Case: You have two children, Denise and Edmund, who are both in their early teens. You must choose to move to the city or to a suburb. Both children will fare moderately well in the city. By contrast, the impact of moving to the suburb on your children is less clear, because there is uncertainty about the quality of the local school’s offerings in sports and art: it either, with 50% probability, has excellent sports and decent arts teaching, or, with 50% probability, has decent sports and poor arts education. In the former case, the suburb will prove especially beneficial to one ‘sporty’ child, but leave the other child as well off as (s)he would be in the city. In the latter case, the suburb will prove somewhat burdensome to one ‘arty’ child, but leave the other as well off as in the city. The possible benefit of the suburb to the sporty child exceeds the possible burden of the suburb to the arty child by a given amount of utility, d. In order to complete the description of this case, we need to fill in which child may be benefited and which child may be burdened by a move to the suburb. Consider the following two scenarios. Intrapersonal Scenario: Denise has special interests in both sports and arts. If you moved to the suburb, she would therefore face both the chance of the benefit and the risk of the lesser burden. Edmund is unaffected by your choice. Interpersonal Scenario: Denise has special interests in sports, Edmund in arts. If the suburb would prove beneficial, Denise would reap the benefit and if it would prove burdensome, Edmund would bear the burden. Table 3 represents this case (d > 0). 12 Table 3. Final utilities for the Intra- versus Interpersonal Case. One benefited in suburb (p = 0.5) One burdened in suburb (p = 0.5) Denise Edmund Denise Edmund 60 60 60 60 Suburb in intrapersonal scenario 70 + d 60 50 60 Suburb in interpersonal scenario 70 + d 60 60 50 City In the intrapersonal scenario, it is the same child, Denise, who will either thrive if the suburb offers excellent training in sports and decent arts teaching or be thwarted in her development if it offers poor arts education. Choosing the suburb therefore exposes Denise to a risk of doing less well than she might for the sake of giving her a shot at doing better than she otherwise would. (Edmund’s welfare is unaffected either way, so there is no conflict between your children’s expected interests.) By contrast, in the interpersonal scenario, the child who might thrive because of the suburb’s opportunities for sports (Denise) is different from the child whose interests might be thwarted there because of the lack of arts education (Edmund). In the latter scenario, there is therefore a conflict of interest between Denise and Edmund. Choosing the suburb would involve imposing a risk of a burden on Edmund, which, if it materialized, would make him worse off than Denise, in order to give Denise a chance at a gain that, if it materialized, would make her better off than Edmund. We submit that these differences make it much more difficult to justify a move to the suburb in the latter scenario. Suppose the difference d between the larger possible benefit and the smaller possible burden of the suburb is just large enough for it to be permissible to move to the suburb in the intrapersonal scenario—the balance between the possible benefit and possible burden is such that the former outweighs the latter even after one has given any extra weight one believes is required to Denise’s situation if she is worse off, and one has compensated for the badness of inequality (if any) in this case. (We 13 therefore do not assume that it is always permissible to maximize Denise’s expected utility in this scenario.) We submit that for this d, it is impermissible to move to the suburb in the interpersonal scenario, because it would not be justified to expose Edmund to a chance of being worse off than he might be for the sake of giving Denise this shot, thereby ensuring that she will be better off than Edmund. However, because it applies the same priority weights to increments in utility in intra- and interpersonal trade-offs, final-utility prioritarianism cannot account for this difference in justifiability. It holds that for every d for which the suburb is permissible in the intrapersonal scenario, it is also permissible in the interpersonal scenario. It is interesting to compare these observations about final-utility prioritarianism with the well-known criticism of utilitarianism by David Gauthier (1963: 121-127) and John Rawls (1999). As Rawls put it: “The striking feature of the utilitarian view of justice is that it does not matter ... how [the] sum of satisfactions is distributed among individuals any more than it matters ... how one man distributes his satisfactions over time... For just as it is rational for one man to maximize the fulfilment of his system of desires, it is right [according to the utilitarian] for a society to maximize the net balance of satisfaction taken over all of its members. The most natural way, then, of arriving at utilitarianism ... is to adopt for society as a whole the principle of rational choice for one man … Utilitarianism does not take seriously the distinction between persons” (1999: 23-24). Unlike utilitarianism, final-utility prioritarianism is sensitive to how lives go: it gives more weight to benefitting a person if his life will go badly than it does to benefitting a person if his life will go well. So, on this view, the boundaries between persons matter in the following sense: if one person is badly off and another well off, one has stronger moral reason to confer a benefit of a given size on the former (Otsuka, 2012: 365-366). But final-utility prioritarianism fails to take seriously boundaries between persons in the following respect: it is insensitive to whether the life that goes well and the life that goes badly are possible lives of the same person or rather the lives of different people (McKerlie, 1984: 233; Otsuka, 2012: 368). While utilitarianism can be said to ignore the separateness of persons, final-utility prioritarianism can be said to ignore the unity of the individual. The 14 Rawlsian objection to utilitarianism is that it “adopt[s] for society as a whole the principle of rational choice” that is appropriate only for a single person. Final-utility prioritarianism has the opposite problem: It adopts for a single person a principle that is appropriate only for a society consisting of many separate persons, whose interests clash with one another (Porter, 2012). Indeed, when we compare the One-Child Case with the Two-Child Case under Certainty, we find two key differences.12 In this two-child case, in addition to the absence of a prudential justification to the less-well-off child for a move to the suburb, there is the presence of a competing-claims-based justification for a move to the city (Otsuka, 2012: 371). To see the force of the latter, consider that one could pose the following rhetorical question about a proposed move to the suburb in the Two-Child Case under Certainty: “How ... can [you] justify providing a benefit of a given size to someone who is already better off in order to make him better off still, when [you] could instead provide ... [nearly as] large [a] benefit to someone else who is worse off?” (Otsuka and Voorhoeve, 2009: 183-184). The competing claims approach has purchase only when there is a choice between alternatives which will either benefit one person or benefit another person. The strength of an individual’s claim is a function of how much utility he stands to gain and from what baseline level of utility, relative to another who might benefit instead (Otsuka, 2012: 371). It is again instructive to compare this observation with a distinct form of the “separateness of persons” objection first formulated by Nagel against utilitarianism. Nagel argues that utilitarianism fails to recognize the significance of competing claims: “[Utilitarianism] depends on an application to interpersonal conflicts of the same principles which are used to settle conflicts between reasons arising from the interests of a single person. The conditions of choice corresponding to this principle are that the chooser should treat the competing claims arising from distinct individuals as though they all arose from the interests of a single individual, himself. 12 These differences are also apparent in the Intra- versus Interpersonal Case. 15 He is to choose on the assumption that all these lives are to be amalgamated into one life, his own. ... But this ... completely distorts the nature of the competing claims, for it ignores the distinction between persons...” (1970: 138). There are therefore at least two respects in which final-utility prioritarianism fails to take seriously both the unity of the individual and the separateness of persons: (i) It is insensitive to the availability of prudential justifications; (ii) It is insensitive to the existence of competing claims. 4. Expected-Utility Prioritarianism There is, however, a version of prioritarianism against which these objections have less force. On “expected-utility prioritarianism,” expected utility is the currency of distributive justice. On this form of prioritarianism, each person’s expected utility has diminishing marginal moral value. Moreover, the moral value of (increments in) a person’s expected utility does not depend on anyone else’s expected (or final) utility. We ought to choose the option with the greatest sum of moral value (i.e. the greatest sum of priority-weighted expected utility). Expected-utility prioritarianism avoids the discrepancy between prudential and moral evaluation in single-person cases that plagues final-utility prioritarianism. As a consequence, it treats intra- and interpersonal trade-offs differently: in intrapersonal tradeoffs, it requires maximizing expected utility; in interpersonal trade-offs, it gives priority to whoever has lower expected utility. In the Two-Child Case with Certainty, it requires choosing the city (since Ben has worse prospects), but in the One-Child Case, it mandates choosing the suburb. In the Intra- versus Interpersonal Case, it holds that you ought to choose the suburb for every d > 0 in the intrapersonal scenario, but also that, for some such d, you ought not to choose the suburb in our interpersonal scenario. It therefore recognizes the difference in their justifiability. Clearly, then, expected-utility prioritarianism recognizes the force of prudential justifications. It also recognizes the force of some competing claims. However, the following case demonstrates that there is one respect in which it does not recognize the force of competing claims. 16 Anticorrelated Case: You have two children, Frank and Gwyneth, each of whom has a 50-50 chance of being disabled. You know that one of them will turn out disabled, and the other healthy. You can either move them to the suburb or the city. The outcomes associated with being disabled in the city and the suburb are as in the Two-Child Case with Certainty; the same goes for being able-bodied (see Table 4). Table 4. Final utilities for the Anticorrelated Case. Frank disabled (p = 0.5) Gwyneth disabled (p = 0.5) Frank Gwyneth Frank Gwyneth City 39 60 60 39 Suburb 30 70 70 30 This Anticorrelated Case is like the One-Child Case, and unlike the Two-Child Case with Certainty, in this respect: there is a prudential justification to both Frank and Gwyneth for the move to the suburb. But this Anticorrelated Case is like the Two-Child Case with Certainty, and unlike the One-Child Case, in this respect: If we come to know what the outcome will be, then it would be transformed into the Two-Child Case with Certainty (Otsuka, 2012: 374). These similarities and differences can be seen in terms of our overarching theme of the difference between intra- and interpersonal trade-offs. In pure intrapersonal trade-off cases such as the One-Child Case, there are no conflicts of interest between any individuals, neither in terms of their expected utilities nor in terms of final utilities. The unity of the individual then gives us reason to maximize expected utility. In pure interpersonal trade-off cases such as the Two-Child Case with Certainty, there are conflicts of interest between individuals both in terms of their expected utilities and their final utilities. In such cases, the separateness of persons provides a reason to give extra weight to improvements to those who are worse off. The Anticorrelated Case is a mixed case, in which there is no conflict between individuals’ interests in expected utilities, but there are conflicts between individuals’ final-utility interests. If Gwyneth turns out to be the child with a disability, then the city is in her final-utility interest while the suburb is in Frank’s. If Frank turns out to be 17 the child with the disability, then the opposite is true. This conflict of interest gives rise to competing claims on behalf of whoever turns out to be able-bodied and whoever turns out to be disabled. In our view, in this mixed case, you have decisive reason to respond to this conflict of interest by helping the worse off child and moving to the city. For how can one justify providing a benefit of a given size to someone who will already be better off in order to make this person better off still, when one could instead provide nearly as large a benefit to someone else who will be worse off, and who would not even reach the (unimproved) level of the better off person if (s)he (the worse off person) were benefitted? “For he that hath, to him more shall be given, at the expense of he who hath not” is not a sound moral principle (Otsuka, 2012: 376). 13 However, expected-utility prioritarianism ignores this conflict of interest. It responds only to the fact that the suburb is in each child’s interest ex ante and therefore mandates moving to the suburb. 5. A Hybrid Priority View In response to some of these objections, Parfit (2012) proposes a view that combines elements of expected utility and final-utility prioritarianism. On this hybrid Priority View, it is true both that an increase in expected utility is more valuable, the lower the level of expected utility from which it takes place, and that an increase in final utility is more valuable the lower the level of final utility from which it takes place. 14 This hybrid view avoids some of the objections raised to the other versions of prioritarianism. The expectedutility-prioritarian element in the view recognizes the force of prudential justifications. In the Intra- versus Intrapersonal Case, it will therefore judge it easier to justify the suburb in the intrapersonal scenario than in the interpersonal scenario. Meanwhile, the final-utilityprioritarian element in the view ensures that one ought to look after whoever will turn out disabled in the Anticorrelated Case. Because it recognizes some differences between intraand interpersonal trade-offs and gets these cases right, we consider this hybrid view more plausible than either of the other two prioritarian views surveyed here. 13 For further discussion, see Fleurbaey and Voorhoeve (2013) and Frick (2013). Parfit leaves open how to balance the importance of increments in expected utility against increments in final utility, when these conflict. For an analysis of ways of doing so, see Bovens (2015) and Voorhoeve and Fleurbaey (n.d.). 14 18 Nonetheless, while this view boasts some of the strengths of each of its component elements, it also inherits some of their weaknesses. The final-utility-prioritarian component ensures that in risky intrapersonal trade-off cases such as the One-Child Case, it will sometimes recommend the alternative that is contrary to the person’s interests. It therefore remains inadequately sensitive to the presence of prudential justifications (Otsuka, 2015). Moreover, it remains inadequately sensitive to the presence of competing claims. By way of illustration of the latter point, consider the following case. Correlated Case: You have two children, Helen and Isaac, each of whom has a 50-50 chance of being disabled. Their fates are bound together. Either both of them will turn out disabled or both will turn out healthy. You can either move them to the suburb or to the city. The outcomes associated with being disabled in the city and the suburb are as in the Two-Child Case with Certainty; the same goes for being ablebodied (see Table 5). Table 5. Final utilities for the Correlated Case. Helen and Isaac able (p = 0.5) Helen and Isaac disabled (p = 0.5) Helen Isaac Helen Isaac City 60 60 39 39 Suburb 70 70 30 30 In this Correlated Case, the expected utility of each of the children for each alternative is just as it is in the Anticorrelated Case. Moreover, the expected priorityweighted final utility of each option is the same in both cases. On the hybrid view, you therefore have just as much reason to choose the city in the Anticorrelated Case as in the Correlated Case. We submit that this is mistaken. As we argued above, in the Anticorrelated Case, there is a conflict of interest in each state of the world. This gives rise to competing claims between the child who ends up able-bodied and the child who will end up disabled, the latter of which, we argued, is stronger and ought to be satisfied. By contrast, in the 19 Correlated Case, since the children’s fates are bound together, there is no conflict of interest between them. There are therefore no such competing claims. The presence of competing claims in the Anticorrelated Case and their absence in the Correlated Case entails that you have stronger reason to choose the city in the former. Because it fails to recognize this, hybrid prioritarianism fails to adequately respond to the presence or absence of competing claims. A further difference between these cases forms the basis of a final objection to all the prioritarian views we have discussed (Broome, 1989; 2015). When one child’s good fortune is the flipside of the other’s ill fortune, as in the Anticorrelated Case, moving to the suburb increases inequality between them. By contrast, when the children experience good or ill fortune together, as in the Correlated Case, neither opting for the city nor choosing the suburb generates inequality. If such inequality is unfair and therefore bad, then, contrary to all prioritarian views, we have a reason against moving to the suburb in the Anticorrelated Case which we do not have in the Correlated Case. In sum, the hybrid prioritarian view proposed by Parfit (2012), while superior to both final-utility prioritarianism and expected-utility prioritarianism, is subject to the following objections: (i) A failure to adequately respond to the presence of prudential justifications; (ii) A failure to adequately respond to the presence of competing claims; (iii) A failure to respond to unfair inequality. We shall now examine to what extent egalitarian views can avoid these objections. 6. Egalitarianism under Risk As mentioned, egalitarians are pluralists, because all sensible egalitarians care about improving well-being as well as about reducing inequality. Recent work has identified one family of pluralist egalitarianism as especially attractive, because it handles the 20 aforementioned cases well (Fleurbaey, 2010). This family evaluates each distribution by its “equally-distributed equivalent,” or EDE.15 To illustrate this idea, consider first a case of certainty, such as moving to the suburb in our original Two-Child Case under Certainty, where Ann’s utility is 70 and Ben’s is 30. Suppose that you should be indifferent between moving to the suburb and a move to a hypothetical town where they each have 45 utils. Then 45 is the equally-distributed equivalent of moving to the suburb. Next suppose that you should be indifferent between moving to the city, where Ann’s utility is 60 and Ben’s is 39, and moving to a hypothetical locale where each has 47 utils. Then 47 is the EDE of moving to the city. Since the latter is higher than the EDE of the suburb, you should move to the city. More generally, under certainty, when an alternative leads to inequality and one ought to be averse to this inequality, then the EDE of this alternative is less than the average utility generated by that alternative. When an alternative contains no inequality, then the EDE is, naturally, the average utility generated by that alternative. In risky cases, insofar as we are concerned with final utilities, this approach tells us to evaluate each alternative as follows. First establish the EDE for each possible distribution of final utility that may result from the alternative. Then take the probability-weighted sum of these values. A risky alternative that invariably yields the same anonymized pattern of inequality in final utilities, such as moving to the suburb in the Anticorrelated Case (with one child at 70 and one at 30), is then evaluated by the value of the EDE of this distribution, which will be less than the average expected utility. But a risky alternative that yields equality in each possible outcome, such as moving to the suburb in the Correlated Case, is evaluated at the expected utility of each person in the population. In the absence of inequality, the view is therefore utilitarian; in the presence of inequality, it gives weight to both reducing inequality and increasing utility. Such a final-utility egalitarian view gets several of our cases right. In our One-Child Case, it favours moving to the suburb, because that is prudentially most valuable for the child. In the Two-Child Case with Certainty, it mandates moving to the city, because this 15 The EDE was introduced in welfare economics by Kolm (1969) and Atkinson (1970). However, as mentioned in fn 6, the social welfare functions they proposed were prioritarian, not egalitarian. The distinctively egalitarian credentials of an EDE social welfare function become apparent when one considers cases under risk. 21 reduces inequality at modest cost. In the Correlated Case, it requires the move to the suburb, because it is to the expected advantage of each and does not generate any inequality. By contrast, in the Anticorrelated Case, it mandates the city, because this reduces inequality. It therefore marks the difference between intrapersonal trade-offs that do not generate inequality on the one hand and interpersonal trade-offs on the other. However, our Intra- versus Interpersonal Case shows that an exclusive concern with the distribution of final utilities is not enough to fully respect the difference between the unity of the individual and the separateness of persons (Voorhoeve and Fleurbaey, 2012). Final-utility egalitarianism alone cannot draw a distinction between the suburb in the intrapersonal scenario and the interpersonal scenario. The reason is that in both scenarios, the suburb leads to the same inequality in final utilities. This issue can be dealt with by extending egalitarianism to cover not merely the distribution of final utility, but also the distribution of expected utility. 16 Such a concern for inequalities in chances meshes well with the egalitarian concern for fairness. A given inequality in final utilities is less unfair when each person has a chance to end up better off than when the worse off have no such chance (Broome, 1990; Arneson, 1997; Temkin, 2001). In the Intra- versus Interpersonal Case, there is greater inequality in expected utility in the inter- than in the intrapersonal scenario. Since such inequality in expected utilities partly determines the (un)fairness of the alternatives, choosing the suburb in the intrapersonal scenario is easier to justify because it is less unfair to Edmund. More generally, this hybrid egalitarianism, which favours equality in both expected utility and final utility, deals successfully with all cases we have considered. In these cases, it is therefore duly sensitive to (i) the availability of prudential justifications; (ii) the presence of competing claims; and (iii) the unfairness of inequality. In so doing, it fully respects the difference between the unity of the individual (which gives us reason to choose in a person’s 16 For details on how to develop the EDE for hybrid egalitarianism, see Voorhoeve and Fleurbaey (n.d.). For an alternative egalitarian response to the Intra- versus Interpersonal Case, see Hyams (2015). Vallentyne (2002) proposes a view which is exclusively concerned with inequalities in expected utility. Such a view is implausible, in our view, for the same reason that expected utility prioritarianism was shown to be implausible in Section 4 above: It does not regard choosing the suburb in the Anticorrelated Case as less justifiable than choosing the suburb in the Correlated Case. It therefore fails to respond to the presence of competing claims and of unfair outcome inequality in the Anticorrelated Case. 22 expected interest) and the separateness of persons (which gives us reason to give extra weight to the claims of those who are less well off than others with whom their claims compete and, we submit, to reduce unfair disadvantage). We therefore regard it as superior, in these respects, to prioritarian views. 17 It matters that some are worse off in comparison with others, both in prospect and in final outcome. References Adler, M. (2012). Well-Being and Fair Distribution. Oxford: Oxford University Press. Anderson, E. (1999). “What Is the Point of Equality?” Ethics 109: 287–337. Arneson, R. (1997). “Postscript to ‘Equality and Equal Opportunity for Welfare’.” In L. Pojman and R. Westmoreland (eds.) Equality: Selected Readings. Oxford: Oxford University Press, 238–241. Arneson, R. (1999). “Egalitarianism and Responsibility,” The Journal of Ethics 3: 225–247. Atkinson, A. (1970). “On the Measurement of Inequality,” Journal of Economic Theory 2: 244–263.Bognar, G. (2012). “Empirical and Armchair Ethics,” Utilitas 24: 467–482. Bovens, L. (2015). “Concerns for the Poorly Off in Evaluating Risky Prospects,” Economics and Philosophy, in press. Broome, J. (1989). “What Is the Good of Equality?” In J. Hey (ed.) Current Issues in Microeconomics. Macmillan, 236–262. Broome, J. (1990). “Fairness,” Proceedings of the Aristotelian Society 91: 87–101. 17 More precisely, we regard it as superior to any prioritarian view that respects Separability, which is generally regarded as a core tenet of such a view. We have not here considered versions of prioritarianism that violate Separability because they pluralistically combine a commitment to prioritarian value with a sensitivity to comparative considerations. One such version has been proposed by Andrew Williams (2012). On this view, prioritarian weights are applied to final utility, but such weighting is triggered only in the presence of competing claims; when there are no competing claims, a person simply has a claim to whatever maximizes his expected utility. Since the moral importance of increments in a person’s utility depends on whether there are others whose claims compete with his, this view violates Separability. Moreover, on our conception of the conditions under which there are competing claims, this view is not embarrassed by any of the cases so far considered. In the One-Child Case and the Correlated Case, it favours the suburb, since this is in the expected interest of each and there are no competing claims. In the Two-Child Case under Certainty and the Anticorrelated Case, it favours the city, since these cases involve competing claims and we therefore ought to have special concern for those who are worse off. Given its success in dealing with the cases we consider in this chapter, such a pluralistic versions of prioritarianism holds promise as capturing, in a different manner from egalitarianism, the moral significance of the fact that some are worse off in comparison with others. See Weber (2014) for further discussion. 23 Broome, J. (2015). “Equality versus Priority: A Useful Distinction.” Economics and Philosophy, in press. Brown, C. (2003). “Giving Up on Levelling Down,” Economics and Philosophy 19: 111–134. Cohen, G.A. (1989). “On the Currency of Egalitarian Justice,” Ethics 99: 906–944. Crisp, R. (2011). “In Defence of the Priority View: A Response to Otsuka and Voorhoeve,” Utilitas 23: 105–108. Fleurbaey, M. (2010). “Assessing Risky Social Situations,” Journal of Political Economy 118: 649–680. Fleurbaey, M. and Voorhoeve, A. (2013). “Decide as You Would with Full Information! An Argument against ex ante Pareto. ” In N. Eyal, S. Hurst, O. Norheim, and D. Wikler (eds.), Inequalities in Health: Concepts, Measures, and Ethics. Oxford: Oxford University Press, 113–128. Frick, J. (2013). “Uncertainty and Justifiability to Each Person: A Response to Fleurbaey and Voorhoeve.” In N. Eyal, S. Hurst, O. Norheim, and D. Wikler (eds.), Inequalities in Health: Concepts, Measures, and Ethics. Oxford: Oxford University Press, 129–146. Gauthier, D. (1963). Practical Reasoning. Oxford: Clarendon Press. Greaves, H. (2015). “Antiprioritarianism,” Utilitas 27: 1–42. Hyams, K. (2015). “Hypothetical Choice, Egalitarianism, and the Separateness of Persons,” Utilitas 27: 217–239. Kolm, S.-C. (1969). “The Optimal Production of Social Justice.” In J. Margolis and H. Guitton, (eds.), Public Economics. London: Macmillan, 145–200. McCarthy, D. (2008). “Utilitarianism and Prioritarianism II,” Economics and Philosophy 24: 1– 33. McKerlie, D. (1984). “Egalitarianism,” Dialogue 23: 223–237. Nagel, T. (1970). The Possibility of Altruism. Princeton: Princeton University Press. Nagel, T. (1979). “Equality,” in Mortal Questions. Cambridge: Cambridge University Press, 106–127. Nord, E. J.-L. Pinto Prades, J. Richardson, P. Menzel and P. Ubel (1999). “Incorporating Societal Concerns for Fairness in Numerical Valuations of Health Programmes,” Health Economics 8: 25–39. Norman, R. (1998). “The Social Basis of Equality.” In A. Mason (ed.) Ideals of Equality, Oxford: Basil Blackwell. 24 O’Neill, M. (2008). “What Should Egalitarians Believe?” Philosophy & Public Affairs 36: 119– 156. O’Neill, M. (2012). “Priority, Preference, and Value,” Utilitas 24: 332–348. Otsuka, M. (2012). “Prioritarianism and the Separateness of Persons,” Utilitas 24: 365–380. Otsuka, M. (2015). “Prioritarianism and the Measure of Utility,” Journal of Political Philosophy 23: 1–22. Otsuka, M. and A. Voorhoeve (2009). “Why It Matters that Some Are Worse Off than Others: An Argument against the Priority View,” Philosophy and Public Affairs 37: 171–199. Parfit, D. (1995). “Equality or Priority?” The Lindley Lecture. The University of Kansas. Parfit, D. (2012). “Another Defence of the Priority View,” Utilitas 24: 399–440. Persson, I. (2001). “Equality, Priority, and Person-Affecting Value,” Ethical Theory and Moral Practice 4: 23–39. Porter, T. (2011). “Prioritarianism and the Levelling Down Objection,” Ethical Theory and Moral Practice 14: 197–206. Porter, T. (2012). “In Defence of the Priority View,” Utilitas 24: 349–364. Rabinowicz, W. (2002). “Prioritarianism for Prospects,” Utilitas 14: 2–21. Rawls, J. (1999). A Theory of Justice, rev. ed. Cambridge, Mass.: Harvard University Press. Rendall, M. (2013). “Priority and Desert,” Ethical Theory and Moral Practice 16: 939–951. Segall, S. (2014). “In Defense of Priority (and Equality).” Philosophy, Politics and Economics (online first). Sen, A. (1973). On Economic Inequality. Oxford: Oxford University Press. Tawney, R. (1964) Equality (5th edition). London: Unwin. Temkin, L. (1983). Inequality. Ph.D. Thesis, Princeton University. Temkin, L. (2001). “Inequality: A Complex, Individualistic, and Comparative Notion,” Philosophical Issues 11: 327–353. Temkin, L. (2003). “Equality, Priority, or What?” Economics and Philosophy 19: 61–88. Ubel, P.A., G. Loewenstein, D. Scanlon, and M. Kamlet (1996). “Individual Utilities Are Inconsistent With Rationing Choices: A Partial Explanation of Why Oregon’s CostEffectiveness List Failed,” Medical Decision-Making 16: 108–116. Vallentyne, P. (2002). “Brute Luck, Option Luck, and Equality of Initial Opportunities,” Ethics 112: 529–557. 25 Voorhoeve, A. and M. Fleurbaey (2012). “Egalitarianism and the Separateness of Persons,” Utilitas 24: 381–398. Voorhoeve, A. and M. Fleurbaey (n.d.). “Priority or Equality for Possible People?” Manuscript. Weber, M. (2014). “Prioritarianism,” Philosophy Compass 9/11 756–768. Weirich, P. (1983). “Utility Tempered with Equality,” Noûs 17: 423–439. Williams, A. (2012). “The Priority View Bites the Dust?” Utilitas 24: 315–331. 26