I.
Amongst various versions of luck egalitarianism, we can distinguish two broad approaches: the Dworkinian approach, and the Cohenian approach. [1] The Dworkinian approach emphasises the importance of hypothetical choice alongside actual choice. Inequalities are just, claims Dworkin, when they arise either from chosen risks or from risks against which agents would not have insured. [2] Dworkin writes, for example, that, if it were possible to know how much insurance an individual would have bought against the risk of developing a handicap, then we should „tailor compensation to what particular individual in fact would have bought.” [3] And in the absence of information about individual hypothetical choices, we should „say that but for (uninsurable) brute luck that has altered these equal odds, the average person would purchase insurance at that level, and compensate those who do develop handicaps accordingly.” [4] Dworkin also introduces other distinctions meant to map onto his distinction between ‘ambition sensitivity’ and ‘endowment insensitivity,’ such as a distinction between tastes which an agent identifies with and those which they do not identify with. However, I do not concentrate on these other difference in the present paper, because the paper does not aim to defend Dworkin per se against Cohen. Rather, the aim of the present paper is to defend one particular feature of the Dworkinian approach, namely its focus on hypothetical choice, against the Cohenian and contemporary alternative focus on actual choice.
The Cohenian approach, on the other hand, places the emphasis on the inequality-justifying force of actual choice rather than hypothetical choice. Inequalities are just, claims Cohen, to the extent that they eventuate from actual choices, including actual choices to take risks. For example, Cohen writes that his view „favors compensation for all deficits in (‘inoffensive’) welfare which do not reflect the subject's choice.” [5] Among other prominent formulations of luck egalitarianism such as those described by Arneson and Roemer, the actual-choice approach has come to dominate. [6] Arneson, for example, writes that,
When persons enjoy equal opportunity for welfare in the extended sense, any actual inequality of welfare in the positions they reach is due to factors that lie within each individual's control. Thus, any such inequality will be nonproblematic from the standpoint of distributive equality. [7]
Several intuitive counter-examples have been posed against the claim that the inequalities resulting from chosen risks are just. [8] But luck egalitarians have come to view such counterexamples as challenging them to clarify, within the actual-choice approach, the precise conditions under which risks must be chosen, in order that the inequalities arising from the choice be just. Luck egalitarians have not tended to view such counter-examples as requiring a return to the Dworkinian focus on hypothetical choice alongside actual choice.
The present paper sheds new light on the nature of just inequalities. I shall argue that the move away from the Dworkinian focus on hypothetical choice alongside actual choice, to the Cohenian focus on actual choice alone, was a mistake. In fact, an account of luck egalitarian fairness that appeals to the inequality-justifying properties of both actual choice and hypothetical choice does a much better job of capturing luck egalitarian intuitions about justice than can any account that relies exclusively on actual choice. I shall show why the various attempts to tidy up an actual-choice approach to justified inequalities cannot succeed in capturing our intuitions about the matter. In particular, I shall reject the requirement that a choice be made in the presence of a reasonable option, in order for the choice to qualify as inequality-justifying. Only when we allow a role both for actual choice and for hypothetical choice, I shall argue, can we give an accurate account of justified inequalities.
II.
What types of event can justify subsequent inequality arising from that event? Or, as I shall put it, what types of event qualify as inequality-justifying? At the core of luck egalitarianism’s self-understanding is the thought that inequality-justifying events are certain types of uncoerced (actual or hypothetical) choice. I shall not seek to justify or explain that core intuition in the present paper. The aim of the discussion that follows is, rather, to clear up confusion about the types of uncoerced choice that qualify as inequality-justifying events. As I noted in the introduction to this paper, recent attempts to clarify the nature of inequality-justifying events have tended to proceed within a Cohenian framework, aiming to describe the conditions under which actual choices must be made in order to justify inequality. In what follows, I shall examine several such attempts, and in so doing argue that we ought to adopt a modified version of Dworkin’s approach, referring not only to actual choice, but also to hypothetical choice.
First, what claims do Dworkin and Cohen make about inequality-justifying events? Dworkin claims that two types of event qualify as inequality-justifying events: chosen risks, and unchosen risks against which an agent would have insured. Cohen, for his part, claims that only choices and chosen risks are inequality-justifying events. Although Cohen claims to be sympathetic toward Dworkin’s hypothetical-insurance approach, [9] it is nevertheless actual choice rather than hypothetical choice that gets foregrounded in Cohen’s discussion of inequality-justifying risk. This is also so in the work of his commentators and in other prominent formulations of luck egalitarianism, such as those described by Arneson and Roemer. Neither Dworkin nor Cohen state clearly what distinguishes a chosen risk from an unchosen risk, but the basic idea seems to be that a chosen risk is a risk that the agent knew that he was exposing himself to (or, we might add, was negligently ignorant of) at the time he opted to pursue a course of action that exposed him to the risk. [10]
I shall argue that Dworkin and Cohen were right about choices being inequality-justifying events, and that Dworkin was nearly right about unchosen risks being inequality-justifying events when agents would have insured against them. But both Dworkin and Cohen, I shall argue, were wrong about chosen risks. The problem is not only that Dworkin and Cohen initially claimed that all chosen risks were inequality-justifying – a claim against which intuitive counter-examples are easy to muster. Rather, the problem is that Dworkin and Cohen framed subsequent debate about chosen risks in a manner that rendered impossible an accurate account of inequality-justifying chosen risks. The problem with both Dworkin’s and Cohen’s accounts of chosen risks, and with subsequent literature on the topic, is that the accounts take place in the context of what we might think of as a one-stage model. The model is one-stage in the sense that it requires us to ask, in a single step, whether or not both the choice to take a risk and the eventuation of the chosen risk – call such combinations choice-and-chosen-risk units – ought to be treated jointly as inequality-justifying events. For example, the model requires us to ask, in a single step, whether or not both the decision to walk on the ice and thereby risk falling through, and the subsequent falling through the ice into the pond, should be treated jointly as inequality-justifying events. The one-stage model does not permit luck egalitarians to ask separately whether the choice is inequality-justifying, and whether the eventuation of the chosen risk is inequality-justifying. As such, it requires luck egalitarians to affirm the following claim, which I shall argue is false:
C: The eventuation of a chosen risk qualifies as an inequality-justifying event if and only if the choice to take the risk qualifies as an inequality-justifying event.
In order not to assume the truth of C, I shall adopt a two-stage model in place of the one-stage model. The two-stage model asks, in two separate steps, first, whether or not the choice to take a risk ought to be treated as an inequality-justifying event, and second, separately, whether or not the eventuation of the chosen risk ought to be treated as an inequality-justifying event. The two-stage model requires us to disaggregate the effect of the choice from the effect of the eventuation of the chosen risk. Such disaggregation is made possible by disaggregating the effect of the choice from the effect of the chosen risk, by looking separately at the change in opportunity brought about by each event. The two-stage model does not assume that C is false, but it does create the conceptual space to argue, as I shall eventually do, that C is false.
Figure 1 illustrates the difference between the treatment of a choice-and-chosen-risk unit by the one-stage model, and its treatment by the alternative two-stage model. The one-stage model asks in one step whether the change in opportunity from t 1 to t 3 (the choice-and-chosen-risk unit) in the diagram is an inequality-justifying event (e.g. a loss of 50 for an agent who chooses to walk on the ice and falls through). The two-stage model, by contrast, asks separately whether, first, the change in opportunity from t 1 to t 2 (the choice) is an inequality-justifying event (e.g. a loss of 20 for an agent who chooses to walk on the ice); and, second, whether the change in opportunity from t 2 to t 3 (the chosen risk) is inequality-justifying event (e.g. a loss of 30 for an agent who falls through the ice). The two-stage approach therefore has the advantage over the one-stage approach that the former, unlike the latter, leaves open the possibility that an agent who chooses to walk on the ice might be held responsible for the consequence of his decision to walk on the ice in terms of its effect on expected outcome (a loss of 20) but not for the consequences of his falling through the ice (a further loss of 30).
Luck egalitarians have attempted to capture their intuitions about the nature of inequality-justifying events by describing criteria for inequality-justifying events. In what follows, I shall examine some of these criteria. In the manner permitted by the two-stage model, I shall ask separately, first, which choices qualify as inequality-justifying events and, second, which risks qualify as inequality-justifying events.
III.
Under what conditions should we treat choices as inequality-justifying events? Under the present heading, I include both choices (but not the risk) that are part of a choice-and-chosen-risk-unit and choices with no risk attached.
Perhaps the most widely endorsed requirement for choices to qualify as inequality-justifying events is the requirement that an agent must have had a reasonable alternative. Several examples seem to provide an intuitive case in favour of this requirement, or something like it. Vallentyne, for example, describes the case of a man who is struck by lightning, and points out that we would not want to treat his being struck by lightning as an inequality-justifying event simply because he had the unreasonable option to lie down and thereby avoid the danger of being hit by lightning. [11] Likewise Eyal gives the example of Acceptable, whose choice to live in an earthquake zone does not seem to justify the inequality resulting from an earthquake that destroys his home, since it would have been unreasonable for him to move away from the earthquake zone. [12]
It sometimes seems odd, however, to insist that, when an agent has only one reasonable option, and when he chooses that option, his choice is not inequality justifying. Vallentyne gives the example of an agent who has the choice either to pursue an education and live a wonderful life, or to become a drug addict and pursue a terrible life. [13] Against the reasonable alternative requirement, Vallentyne argues that the agent’s choice to pursue an education and live a wonderful life was might be inequality-justifying even in the absence of a reasonable alternative. But the reasonability requirement is easily altered to avoid this objection. In its altered form, the requirement insists that, in order for a choice to qualify as inequality-justifying, the option set from which the agent chose must have included at least one reasonable option, which may have been the chosen option. [14]
It seems to me a mistake to insist on the requirement that the agent have a reasonable option in order for his choice to qualify as an inequality-justifying event. In order to demonstrate the fallacious nature of the reasonability requirement, we should begin by noting that an option might be considered unreasonable either in respect of its expected outcome, or in respect of the level of risk attached to it (including its possible prognoses), or some combination of these. Since for the moment we are only considering when the choice part of a choice-and-chosen-risk unit is inequality-justifying, and not when the chosen risk part of a choice-and-chosen-risk unit is also inequality-justifying, we can set aside, for consideration in the following section, worries about the level of risk attached to an agent’s options. We can postpone risk-related concerns until we come to look at the question, which risks qualify as inequality-justifying? The question of present interest is, rather, whether we should accept some version of the requirement that an agent must have an option that is reasonable in respect of the expected outcome that it affords, in order for his choice to qualify as an inequality-justifying event. In order to answer the stated question, we need to distinguish three different interpretations of reasonable. On all three interpretations, I shall argue, we should reject the reasonability requirement.
The first sense in which an option might be considered reasonable in respect of its expected outcome is that the expected outcome of the option compares sufficiently favourably with that of the agent’s other options. However, this reading of reasonability, by inviting a comparison with the expected outcome of other options, is plausible only when the reasonability requirement is couched as a demand for a reasonable alternative to the option chosen. But as I argued above, we should not, for the reason described by Vallentyne, couch the reasonability requirement as a demand for a reasonable alternative to the option chosen. When the reasonability requirement is couched instead as a demand for at least one reasonable option, which may be the option chosen, then is no reason at all to invite a comparison with the agent’s other options. Rather, when the reasonability requirement is so couched, we seem to have in mind one of the following two interpretations of ‘reasonable.’
The second sense in which an option might be reasonable in respect of its expected outcome is in the sense that the expected outcome of the option meets some minimum absolute threshold. [15] Perhaps we might think that the option to become a drug addict and pursue a terrible life is unreasonable in this sense. But it seems to me that we should not be concerned about whether an agent has a reasonable option in the sense of having an option with a sufficiently high expected outcome. Provided that the luck egalitarian redistributive scheme has thus far been implemented, such that any agent whose opportunities are adversely affected by a non-inequality-justifying event has had this change remedied, there are two possible reasons why an agent would not have an option that is reasonable in the sufficiently good sense. Neither of these reasons seems to motivate a case for treating the choice as an inequality-justifying event.
One reason why an agent might not have an option that is reasonable in the sufficiently good sense is that past inequality-justifying events caused his opportunities to become insufficiently good. He chose, for example, to become a drug addict and pursue a terrible life, forgoing the opportunity to pursue an education and lead a wonderful life. On the present interpretation of reasonable, the choice to become a drug addict would qualify as an inequality-justifying event, because the option set from which the agent chose did include at least one reasonable option. Having chosen to become a drug addict, the agent now faces a further choice between becoming a heroin addict and becoming a cocaine addict, neither of which are reasonable options in the sufficiently good sense. Hard-line luck egalitarians will have no sympathy for this agent’s situation, and will see no reason to worry about the poverty of the options that the agent now faces. But even more forgiving luck egalitarians ought not to say that we should treat the agent’s choice between heroin and cocaine as a non-inequality-justifying event, on the ground that the choice was made in the absence of an option that was reasonable in the sufficiently good sense. To so treat it would only be to restore the agent’s opportunity for aggregate outcome to what it was before the choice between heroin and cocaine, but not before the choice to take drugs, which is still to leave him with insufficiently good opportunities. By treating the agent’s choice between heroin and cocaine as a non-inequality-justifying event, forgiving luck egalitarians will not achieve the result that they want to achieve, which is to limit the impact on the agent’s opportunities of his decision to become a drug addict rather than pursue an education. So neither an unforgiving nor a forgiving luck egalitarian have reason to claim that an agent’s choice must have been made in the presence of an option that is reasonable in the absolute sense, in order to qualify as an inequality-justifying event.
The other reason that an agent might not have a reasonable option is that, even if all his past choices were optimal, there might simply not be enough to go around. That is, conditions of existence might be such that no agents, regardless of their choices, are able to enjoy options that are reasonable, in the sense of having an expected outcome that meets some minimum threshold for sufficiency. A worldwide climate-induced famine, for example, might induce such conditions. I do not want to insist (or to deny) that under such circumstances we should still abide by the demands of luck egalitarianism. One might think, for example, that under such circumstances we should not abide by the demands of luck egalitarianism, on the grounds that it is better for some to have enough and some to have nowhere-near-enough than for all to have not-quite-enough. But even if we should continue to abide by the demands of luck egalitarianism under such circumstances, there seems to be no reason at all to suppose that we would want to treat choices made in the absence of a sufficiently good option – which would under such circumstances include all choices – as non-inequality-justifying events. To treat all choices as non-inequality-justifying events would be to require redistribution of opportunities after all choices, and there seems to be no reason to do that. In fact, quite the opposite seems to be true: under conditions of extreme scarcity, there is more reason than ever to insist that agents internalize the costs of their suboptimal choices.
Finally, the third sense in which an option might be considered reasonable in respect of its expected outcome is in the sense that it compares sufficiently favourably with the options faced by other agents. A demanding version of this interpretation insists that agents must have faced options with the same expected outcome as the options faced by other agents. [16] But it is not a problem if the expected outcome of the agent’s options is not be the same as the expected outcomes of other agents’ option sets, provided that this state of affairs is the result of prior inequality-justifying events. Whenever an agent faces a choice between different options, he cannot properly complain that the opportunity value of his option set as a whole is unreasonably low by comparison to that of other agents. It could not be higher, provided that redistribution has so far taken place along the required lines, without his having opportunities that are problematically higher than it should be. It is therefore a mistake to claim that when agents do not face options the expected outcomes of which are insufficiently great compared to those of other agents, their choices are not inequality justifying.
We need compare the expected outcome of an agent’s options neither with each other, nor with some absolute standard, nor with the expected outcome of other agents’ options, in order to distinguish choices that do qualify as inequality-justifying events from choices that do not qualify as inequality-justifying events. In fact, it seems to me, we need not make any comparisons in order to identify choices that qualify as inequality-justifying events, because there is no need to impose any structural conditions on the types of choice that qualify as inequality-justifying events. (I say ‘structural conditions’ in order to exclude coerced choices and also to remain neutral about whether choices motivated by certain types of altruistic reasons qualify as inequality-justifying events. [17] The arguments that follow do not preclude such exclusion.) In order to motivate this claim, recall the reason that it has in the past been thought necessary to insist that choices qualify as inequality-justifying only when they satisfy certain structural conditions. It has been thought necessary to insist on such conditions because numerous intuitive examples, such as Vallentyne’s lightning case and Eyal’s earthquake case above, attest to the fact that we should not treat all choice-and-chosen-risk units as inequality justifying, even when they are motivated by purely self-interested considerations. Now in the context of the one-stage model, in which existing accounts of luck egalitarianism are framed, any attempt to explain the intuition that not all choice-and-chosen-risk units qualify as inequality-justifying events will necessarily require that we treat both certain choices and certain chosen risks as non-inequality-justifying events. This is because the one-stage model requires us to endorse C (the claim that the eventuation of a chosen risk qualifies as an inequality-justifying event if and only if the choice to take the risk qualifies as an inequality-justifying event), the claim that the eventuation of a chosen risk qualifies as an inequality-justifying event if and only if the choice to take the risk qualifies as an inequality-justifying event. The one-stage model does not therefore permit us to say that the choice component of a choice-and-chosen risk unit is an inequality-justifying event, whereas the risk component is not.
Once we switch from the one-stage model to the two-stage model, however, we can see that there is no need to impose any structural conditions on the types of choice that qualify as inequality-justifying, and therefore no need to worry about the supposedly elusive nature of the distinction between inequality-justifying choices and non-inequality-justifying choices. For, in the context of the two-stage model, we need not try to explain why not all choice-and-chosen-risk units intuitively seem to be inequality justifying by trying to describe the conditions under which both choices and chosen risks qualify as inequality-justifying events. In the context of the two-stage model, we are free to reject C, and to insist that the reason that not all choice-and-chosen-risk units intuitively seem to be inequality justifying is because not all chosen risks are inequality-justifying events, even though all choices are inequality-justifying events.
IV.
I have argued that choices always qualify as inequality-justifying events. Let us look now at risks. Under what circumstances do chosen risks qualify as inequality-justifying events?
Dworkin and Cohen initially claimed that all choice-and-chosen-risk units qualify as inequality justifying, and therefore that all chosen risks qualify as inequality-justifying events. But as noted above, it is not difficult to describe intuitive cases in which choice-and-chosen-risk units do not seem to be inequality justifying (e.g. Eyal’s earthquake case, Vallentyne’s lightening case). I argued that the failure of such choice-and-chosen-risk units to justify inequalities is not due to the failure of the choice part of the unit to qualify as an inequality-justify event. It is, rather, due to the failure of the chosen risk part of the unit to qualify as an inequality-justifying event. We are therefore left with the task of describing a criterion to distinguish those chosen risks that do qualify as inequality-justifying events from those chosen risks that do not qualify as inequality-justifying events. I shall examine two conditions thought to distinguish inequality-justifying chosen risks from non-inequality-justifying chosen risks. The first of these conditions claims that we should identify inequality-justifying chosen risks by comparing the risk faced by an agent with the risk faced by other agents. The second of these conditions claims that we should identify inequality-justifying chosen risks by looking at the risk associated with the other options of the agent who faces the risk. Once again, I shall show how, when we measure the payoffs of different risks in terms of changes in opportunity. By identifying the deficiencies in these two putative conditions, I shall describe a new condition that accurately captures the intuitive distinction between inequality-justifying chosen risks and non-inequality-justifying chosen risks.
Lippert-Rasmussen discusses the view that we should, among other conditions, distinguish between inequality-justifying chosen risks and non-inequality-justifying chosen risks by asking how the risk faced by the agent compared with the risk faced by other agents. He writes that, subject to certain qualifications, the inequality that obtains between two agents as the result of the eventuation of a chosen risk is justified if their gambles “involved identical outcomes with identical probabilities.” [18] Lippert-Rasmussen does not think that this condition is a necessary condition for a chosen risk to qualify as inequality-justifying, because he allows that, where the matching gambles condition is not met, unequal gambles can still be inequality-justifying if the gambles „were equally good from the perspective of both gambles.” But Lippert-Rasmussen does nevertheless claim that matching gambles is part of a sufficient condition for a risk to qualify as an inequality-justifying event.
Lippert-Rasmussen is right to think that matching gambles is not a necessary condition for a risk to qualify as an inequality-justifying event. If agents are equally prefer their gambles then it should not be an obstacle to their chosen risks to qualifying as inequality-justifying events that they did not face matching gambles. But Lippert-Rasmussen is wrong to think that matching gambles is part of a sufficient condition for a chosen risk to qualify as an inequality-justifying event. For one thing, for the reasons given above, it ought not to matter if the payoffs of the various prognoses are not identical, since agents may unproblematically face option sets with different opportunity values as a result of prior inequality-justifying events.
We might, then, modify Lippert-Rasmussen’s proposal, so that it insists only that it partly suffices for a chosen risk to qualify as an inequality-justifying event that agents face equally risky options, even though the payoffs attached to the risks may differ. It would not prevent the chosen risk qualifying as an inequality-justifying event on the matching gambles criterion if Albert had faced a 50:50 gamble between 10 and 20, whereas Bertie had faced a 50:50 gamble between 30 and 40. But it would, however, prevent the chosen risk from qualifying as an inequality-justifying event on the matching gambles criterion if Albert had faced a 50:50 gamble between 10 and 20 whereas Bertie had faced a 50:50 gamble between 25 and 45, or a 25:75 gamble between 0 and 60. It would do so, not because of the different opportunity values of the agents’ gambles, but rather, because of the different levels of risk attached to the agents’ gambles. Even if we make this modification, it still seems that we should reject the modified version of Lippert-Rasmussen’s proposal, because what is important is not what risks different agents faced, but only the extent to which they preferred those risks. Suppose, for example, that two agents face identical risks, but that one agent is more risk loving than the other. As such, one agent would have chosen the risk that he actually faces over any other risk (with the same expected outcome) that he might have faced. The other agent, on the other hand, would have chosen a less risky option over the risk that he actually faces, if he had had the opportunity to do so. In this case, despite the identical risks faced, we should not treat the eventuation of both risks as inequality-justifying events. [19]
Whereas Lippert-Rasmussen’s formulation distinguished between inequality-justifying chosen risks and non-inequality-justifying chosen risks by comparing the risk faced by an agent to the risk faced by other agents, Otsuka claims that we should instead identify inequality-justifying chosen risks by looking at the risk associated with the other options of the agent who faces the risk. He writes that ”inequalities traceable to luck are not unfair if and only if they are the result of option luck arising from gambles when there is a reasonable alternative to gambling whose outcome is certain (or at least nearly so).” [20] In the terminology of inequality-justifying events, Otsuka’s claim should be interpreted as requiring that, in order for a risk to qualify as an inequality-justifying event, the agent must have had a safe and reasonable alternative to facing the risk. Recall that I distinguished above between three senses of reasonable. Otsuka makes clear that by reasonable he has in mind a combination of the first and second senses. He treats a safe alternative as reasonable if it is both at least as good as the chosen risky option, and if it is sufficiency good to secure a decent life.
Consider the following objection to Otsuka’s view. Because it is framed in the context of the one-stage model, the formulation is intended to decide both whether a choice is an inequality-justifying event and whether a chosen risk is one. But we ought, rather, to use the formulation to determine only whether the chosen risk is an inequality-justifying event, and not also whether the choice is an inequality-justifying event. If an agent chooses a risky option in the absence of a reasonable and safe alternative, then Otsuka’s formulation would require that we treat both the choice and the chosen risk as non-inequality-justifying events. But, contra Otsuka’s formulation, we should in fact treat only the chosen risk as a non-inequality-justifying event, but treat the choice as an inequality-justifying event.
The second respect in which Otsuka’s formulation is problematic is that it focuses on actual options rather than hypothetical options. What seems important in deciding whether a chosen risk qualifies as an inequality-justifying event is not whether the agent actually had the option of a reasonable and safe alternative, but rather, whether he would have chosen a reasonable and safe alternative, if he had had the option to do so. [21] The difference between Otsuka’s formulation and the hypothetical choice formulation emerges in the case of risk-loving agents who do not face safe alternatives. Suppose that an agent faces a risk to which he does not have a reasonable and safe alternative. Being risk loving toward this risk, the agent would decline a reasonable and safe alternative in favour of facing the risk, even if he had such an alternative. Otsuka’s formulation would have to say that the eventuation of this risk is a non-inequality-justifying event, because the agent does not have a reasonable and safe alternative. A hypothetical choice formulation, on the other hand, would claim that the eventuation of the risk is an inequality-justifying event, because the agent would have chosen to face the risk over a safe and reasonable alternative if he had been able to do so. It seems that, intuitively, the latter gives a more accurate account of our intuitions on this matter. For example, suppose that an agent has no choice but to gamble in a casino. As it happens, the agent loves gambling and would have chosen to gamble in a casino even if he had had a safe and reasonable alternative. Luck egalitarians should not treat the gamble as a non-inequality-justifying event and thereby compensate the agent for his losses in the casino, or take away his gains, even though he had no safe and reasonable alternative to gambling.
I have argued that, if any chosen risks at all qualify as inequality-justifying events, then we should identify them neither by comparing the risk associated with the agent’s chosen option with the risk associated with other agents’ options, nor by comparing the risk associated with the agent’s chosen option with the risk associated with the agent’s other options. Rather, the intuitions described above ought to lead us to the conclusion that we should identify inequality-justifying risks by asking whether the agent would have chosen an alternative that is both safe and reasonable, in the sense that its expected outcome is the same as that of the risky option, if he had been offered such an option. To the extent that an agent would have chosen a safer alternative with the same expected outcome as the risk, rather than face the risk, we should treat the eventuation of the chosen risk as a non-inequality-justifying event and redistribute accordingly. The examples above suggest that this formulation best captures our intuitions about inequality-justifying chosen risks.
V.
This paper set out to give an account of justified inequalities. Its conclusions are as follows. First, I argued existing accounts of luck egalitarianism do not accurately distinguish inequality-justifying events from non-inequality-justifying events. Second, I argued that luck egalitarians should impose no structural conditions on the types of choice that qualify as inequality-justifying events. Third, I argued that luck egalitarians should treat chosen risks as non-inequality-justifying events to the extent that the agent facing the risk would have declined the risk in favour of a safe alternative with the same expected outcome as the risk.