1. INTRODUCTION
Imagine a group of people who are completely equal: they have equal resources, or they are at an equally high level of welfare, or they have equal opportunities for welfare, or whatever your favourite view of the currency of distributive equality is. There are no relevant differences in their circumstances, talents, or their efforts in pursuing their life plans. But suppose one day they all make the choice to take up smoking. Each of them smokes twenty cigarettes a day. After some time, some – but not all – of these people develop lung cancer. They need costly medical care. Does justice demand that their treatment is covered from the taxes they have all paid, including those who have remained healthy?
On a popular egalitarian view, it is unjust if some people are worse off than others through no choice or fault of their own. This is the central idea of the view known as luck egalitarianism.Footnote 1 As a corollary, luck egalitarians typically hold that it is not unjust if some people are worse off than others due to the choices they have made. For instance, Richard Arneson writes:
distributive justice does not recommend any intervention by society to correct inequalities that arise through the voluntary choice and fault of those who end up with less, so long as it is proper to hold the individuals responsible for the voluntary choice or faulty behaviour that gives rise to the inequalities. (Arneson Reference Arneson1990: 176)
Thus, most luck egalitarians will point out that the people in the example took a gamble when they started smoking. That some of them developed lung cancer is not a mere accident. It is true that those who did were unlucky. But their bad luck is not merely bad ‘brute’ luck; it is a consequence of a chance they took, knowing all the risks. It is a matter of ‘option’ luck. There is no injustice if some are worse off than others due to their own informed choices. Justice is not concerned with inequalities due to bad option luck.
I said most luck egalitarians would make this response because there are of course significant differences between the views of different luck egalitarians. For one thing, some luck egalitarian views concern desert, rather than responsibility. Perhaps on these views, the unlucky smokers do not deserve their bad outcome, even if they are responsible for it. Therefore, it is more precise to talk about luck egalitarianism as a family of views. In this paper, I set desert aside.
It is also customary to distinguish between ‘thin’ and ‘thick’ conceptions of luck. On the thin conception, brute luck is simply what one is not responsible for, and what one is not responsible for is a matter of brute luck. Luck is the inverse of responsibility. The thick conception, in contrast, employs some substantive, moralized notion of responsibility. In what follows, I focus on thin responsibility luck.Footnote 2
Returning to our smokers: things might be a bit more complicated than the standard luck egalitarian response suggests. For you can imagine the people suffering from lung cancer respond the following way:
It is not true that we are worse off through our own choice or fault. We made the same choices as others, and we are all equally at fault for ignoring the health risks of tobacco. Our illness is due to our individual susceptibility, which is beyond our control. How could we be worse off through our own choice or fault, rather than mere bad luck, when we acted no differently than anyone else?
In effect, these people are saying that their lung cancer is a genetic disease. It is due to bad brute luck.Footnote 3
The example is meant to highlight, in an intuitive way, that the distinction between brute luck and option luck may not be as straightforward as it initially seems. Thus, insofar as it is supposed to be used as the basis for attributing responsibility, luck egalitarians need to sharpen it. They need to explain more carefully when an outcome can be attributed to brute luck or to option luck. They need to explain when it is justified to hold people responsible for bad outcomes.
Of course, luck egalitarians are not unaware of this point. Arneson adds the caveat that assigning fault is appropriate only ‘so long as it is proper to hold the individuals responsible’. Yet few luck egalitarians have much to say about the conditions for attributing responsibility. Typically, they rely on compelling examples and rough intuitive judgements instead. This raises the worry that in the end, luck egalitarianism attributes ‘choice and fault’ in arbitrary ways, and this makes the view vulnerable to what is known as the neutrality objection: the claim that rather than holding people responsible for all outcomes that are due to choice, the view ends up selectively picking only some outcomes. Thus, the view isn’t grounded in luck, but on implicit value judgements about the worth of choices and activities. The luck in luck egalitarianism isn’t thin after all.
In this paper, I argue that this isn’t just the consequence of a mere oversight. The problem inevitably arises because the distinction between brute luck and option luck cannot be made in a value-neutral way. There is no such thing as thin responsibility luck. That this has not been adequately realized is because the distinction has mostly been just taken for granted. I argue that the neutrality objection poses a much more serious problem for luck egalitarianism than usually acknowledged. Unless luck egalitarians provide a defensible account of attributing responsibility, their view remains incomplete.
If my argument is correct, it has far-reaching consequences. Luck egalitarians are typically liberal egalitarians: they believe that the state should remain neutral between different conceptions of the good. They are anti-perfectionists.Footnote 4 But if luck egalitarianism cannot meet the neutrality objection, then it seems it has to abandon its commitment to anti-perfectionism. Working out the details of this implication, however, is beyond the scope of this paper. Here I will focus on the neutrality objection and thin responsibility luck.
2. BRUTE LUCK AND OPTION LUCK
In my example, all the people took a chance when they started smoking. The bad outcome for those who developed lung cancer seems to be a matter of option luck. When they complain that they had acted no differently than those who did not get sick, their complaint is that the outcome was a matter of brute luck instead.
The distinction between brute luck and option luck is fundamental for luck egalitarianism. Luck egalitarians often write as if you could use it for determining which outcomes people should be held responsible for. For instance, Ronald Dworkin asks: ‘When and how far is it right that individuals bear the disadvantages and misfortunes of their own situations themselves, and when is it right, on the contrary that others . . . relieve them from or mitigate the consequences of these disadvantages?’ His answer: ‘individuals should be relieved of consequential responsibility for those unfortunate features of their situation that are brute bad luck, but not from those that should be seen as flowing from their choices’ (Dworkin Reference Dworkin2000: 287).
The distinction between the two kinds of luck was introduced by Dworkin in the course of developing his view on the equality of resources. On his way of drawing the distinction, ‘option luck is a matter of how deliberate and calculated gambles turn out – whether someone gains or loses through accepting an isolated risk he or she should have anticipated and might have declined.’ Brute luck is luck that isn’t a matter of such gambles. If you are hit by a meteorite, you have bad brute luck; if you win on the stock market, you have good option luck.Footnote 5
But it’s a bit more difficult to pin down where precisely to draw the line between brute luck and option luck than Dworkin’s examples suggest. Take the case of the meteorite. Being hit by a falling meteorite is plainly a matter of brute luck. But it’s not a matter of what you do; it’s just something that happens to you. Surely, Dworkin does not want to suggest that only things that happen to you can be matters of brute luck. It should be possible that the outcomes of some gambles are matters of brute luck. And the same should be true of option luck too: when the people in the example develop lung cancer, that is something that happens to them, yet luck egalitarians surely want to say it is not a matter of brute luck.
There is anyway the condition that option luck is a matter of only deliberate and calculated gambles. It follows then that the outcomes of gambles that are not considered with some care are not matters of option luck. But this is a bit odd. The favourite examples of luck egalitarians involve reckless motorcyclists, daredevil mountaineers, smokers and drunkards. These examples hardly suggest cool-headed, prudent decision makers. This way of drawing the distinction seems to make some of the standard examples of luck egalitarians irrelevant.
Dworkin is not unaware of these complications. He suggests that the distinction between brute luck and option luck can be a matter of degree (Reference Dworkin2000: 73). But a degree of what? If the idea is that more deliberation and calculation makes an outcome more of a matter of option luck, the problem does not disappear at all. If the less you deliberate about a choice, the more its bad outcome is a matter of brute luck, then the less responsible you are for it. This seems to get things the wrong way around.
The claim that option luck has to do with isolated risks is similarly puzzling. What makes a risk isolated? One way to understand the idea might be that a risk is isolated when the outcomes have no long-term consequences. If you stay out partying one night, you may ruin your next day, but the choice makes no difference in the long run. But again, this way of understanding the idea would disqualify the standard examples of luck egalitarians. The choice to drive recklessly may have consequences that are not isolated. Is the accident you cause a matter of brute luck then? Surely not.
Finally, consider the idea that the decision maker ‘should anticipate’ the outcomes of a gamble in order for it to count as an instance of option luck. How should we understand this? Perhaps the point is that an outcome can be a matter of option luck even if the gamble was not deliberate and calculated, as long as the decision maker ought to have deliberated the right way – obtained all the relevant information, calculated the odds correctly, and so on. Hence the examples of reckless drivers and careless mountaineers, smokers and drunkards are saved. But the distinction between brute luck and option luck becomes normative.
Notice how far things have got from the original, simple way of drawing the distinction – a distinction between the part of our fate that has to do with choice and that which is ‘the work not of people but of nature or brute luck’ (Dworkin Reference Dworkin2000: 287). In particular, we have moved from the idea that the distinction has to do with the choice an individual faces to the idea that it has to do with the way the individual ought to reason about the choice. What initially looked like an external, factual feature of the world turned into an internal, normative feature of the decision maker.Footnote 6
3. CHOICE, RISK AND CONTENT
It is worth trying to draw the distinction in a different way. The arguments of luck egalitarians sometimes suggest that the distinction between brute luck and option luck can be made at a high level of abstraction. That would have the advantage that responsibility for choices can be assigned formally, regardless of the content of choice.
To illustrate the idea, consider the simple choice in
Figure 1. The sequence of events unfolds from the top. Thus, a decision maker, A, has to decide whether to move left, with the result that she gets 1 for certain. This value can represent whatever you think is important for distributive justice: welfare, resources, exercising a capability, and so on. If the decision maker moves right, she takes a gamble, represented by a move by ‘Nature’ (denoted by N). Depending on the outcome of this random move, she might end up with 0 or 2. Nature’s move represents the person’s luck. In all of my examples, I will assume that the probabilities that Nature moves left or right are equal, and that A knows this.
Figure 1
Suppose that A moves right and ends up with 0. She takes a gamble and gets unlucky. Is the outcome a matter of brute luck or option luck? Considering the figure only from node N, the outcome could be considered an instance of brute luck: A has no control over Nature’s move at this point.
But if you consider the whole figure, Nature’s move at N appears to be a matter of option luck. After all, A could have moved left, securing the certain outcome of 1. Since she voluntarily took a risk with full knowledge of the possible outcomes and their probabilities, she is responsible for her predicament. So she has no claim to any compensation.
I take it that luck egalitarians would agree. They would recommend moving left, given that A wants to avoid ending up in a bad situation without compensation. The decision maker might of course have other reasons to consider when she deliberates over her choice. But, on luck egalitarianism, one reason to consider is that moving right might lead to a bad outcome for which, as a matter of justice, the decision maker has no claim to be compensated.
But this leads to a worry. It appears now that luck egalitarianism merely countenances risk aversion. It recommends choosing a certain prospect over an uncertain one with the same expected value: in the example, it recommends choosing 1 rather than the gamble between 0 and 2 with equal odds. Sometimes luck egalitarians do seem to have something like this in mind. Shlomi Segall, for instance, writes:
A person who suffers bad option luck was, we could say, ‘in the business’ of risk-taking. It is the consequence of that type of business that she may reasonably be expected to bear. . . . And what we owe each other (as a matter of distributive justice), according to standard luck egalitarians, extends to neutralizing luck only in cases where it has not been tempted.Footnote 7
But, if this were what luck egalitarians really believed, theirs would be an absurd view. We are all ‘in the business’ of taking risks every day of our lives. A life with no risk is impossible – and it would not be recognized as a human life. Plainly, many risks, both everyday and extraordinary, are absolutely worth taking. (Think of the risks you took to pursue philosophy.) Distributive justice cannot require, as a general rule, to avoid risks.
If that is correct, then it follows that you can’t ‘read off’ merely from the bare bones representation of a choice whether it is proper to hold people responsible for the outcome of that choice. You cannot formulate the distinction between brute luck and option luck at a high level of generality. As a result, you can’t avoid considering information on the content of the choice. As I shall put it, the distinction between brute luck and option luck cannot be made without substantive considerations. Being responsible for an outcome cannot merely be a function of having chosen the act that led to that outcome.
Dworkin’s formulation is vulnerable to the same problem. To see this, consider
Figure 2. In this choice, if the decision maker moves left, she may end up with 0 or 1, depending on her luck at N 1. If she moves right, she may end up with 0 or 2, depending on Nature’s move at N 2. Now suppose that moving left requires no action on the part of the decision maker: it is just where the chips may fall if things go on unimpeded and A lets, so to speak, Nature take its course. Moving right, in contrast, requires some action on A’s part – perhaps she has to pull a lever or press a button or move away from the place where meteorites are likely to fall. So, you can say that the outcome following N 1 is something that just happens to A; the outcome following N 2 is something she has chosen.
Figure 2
Now there is a sense in which the outcomes of moving left are the product of brute luck. After all, if no choice is involved, ending up with 0 or 1 is something that merely happens to the decision maker (given Dworkin’s way of making the distinction). So if she ends up with 0 (she is hit by a meteorite), she has a claim to compensation. Plausibly, she should be brought up to the level at which she would be if her luck turned out to be better. In the example, that means she should be brought up to 1.
Notice that if this is the case, then the choice depicted in Figure 2 simply collapses into the choice depicted in Figure 1. For if A can expect compensation in case her luck runs out at N 1, then moving left becomes an alternative with the certain outcome of receiving 1. This, in turn, leads back to the objection from risk aversion that I raised for the previous example.
The point can be generalized. If a system of luck egalitarian social policies determines ex ante the list of risks for which people can expect to be compensated, it transforms uncertain prospects into certain ones. But no system of social policy can remove all risks. Hence luck egalitarians must select those risks for which they don’t hold people responsible, and so those that give rise to claims to compensation. However, there is no formal way for selecting these risks. From an abstract representation, you cannot determine which outcomes are a matter of brute luck or option luck. At this level of generality, the distinction between brute luck and option luck cannot be the basis of attributing responsibility. To decide when it is proper to hold people responsible, luck egalitarians cannot avoid making substantive judgements. They cannot avoid evaluating the content of choices.Footnote 8
4. LUCK AND RATIONALITY
On Figure 2, choosing to move to the left at the initial node is clearly irrational. A can end up no worse off by moving right rather than moving left, and she has an even chance of ending up better off. Moving right weakly dominates moving left. By compensating only those who go with the default option – by assumption, moving left in the example – luck egalitarianism introduces perverse incentives. It countenances an option that is irrational.
Let’s now drop the assumption that moving left is the default option. Luck egalitarians may insist that the correct way to interpret the example is that the decision maker has to choose between two gambles. I assume they agree that choosing one of the gambles (moving left) is irrational when the other (moving right) is available.
What should they say about the demands of justice in this case? They may say that whatever happens, A has no claim to compensation, since whichever alternative she chooses, she takes a deliberate gamble whose outcome is a matter of option luck. But that seems unduly harsh. After all, A has no riskless prospect: it is not her fault that whatever she chooses, she is faced with risk. So luck egalitarians may, instead, say that A is not at fault if she chooses the weakly dominant prospect and moves right. Therefore, she has a claim to compensation if she ends up with 0 when she could have ended up with 2. But she has no claim to compensation if she ends up with 0 when she could have ended up only with 1 because of her choice.
Consider now another decision maker, B, who chooses to move left at the initial node. He ends up with 0 rather than 1. When it comes to compensation, luck egalitarians tell him that it is due to his own choice and fault that he ended up with the worse outcome. Meanwhile, A, who moved right at the initial node but also ended up with 0, collects her compensation. It seems now that B has a complaint. He might put it this way:
It is unfair that A has a claim to compensation and I do not. Whatever happened to A happened to me too. If she has a claim because she ended up with the worse outcome due to bad brute luck, then I, too, ended up with the worse outcome due to bad brute luck. If I ended up with the worse outcome due to my bad option luck, then she, too, did the same. If avoiding the bad outcome was beyond her control, then my bad outcome was just as much beyond my control. In fact, control and luck have nothing to do with where we both ended up. The only difference is that I made an irrational choice. But luck egalitarianism is not supposed to be about rewarding people who make rational choices and punishing those who do not – it is supposed to be about neutralizing the effects of bad brute luck and protecting people against bad outcomes over which they have no control.
It seems to me that B’s complaint is essentially correct. Whatever the difference between A’s and B’s predicaments, it has nothing to do with luck.Footnote 9
5. INITIAL EQUALITY OF OPPORTUNITY
So far, following Dworkin, I have taken a ‘bottom-up’ approach to identifying instances of brute luck and option luck. I have looked at choices made at particular decision nodes. There is an alternative, ‘top-down’ approach that looks only at the initial decisions rather than subsequent choices and random moves. The views that take this approach posit that justice demands that the entire decision trees of different individuals be equivalent. This family of views emphasizes initial equality of opportunity.
Arneson writes:
Equal opportunity for welfare obtains among persons when all of them face equivalent decision trees – the expected value of each person’s best (= most prudent) choice of options, second-best, . . . nth-best is the same.Footnote 10
There are two important points to note here. The first is that decision trees should be equivalent, rather than equal: they don’t need to have the same structure as long as their expected values from the initial decision node are equal. Second, Arneson formulates the view in terms of expected value rather than expected utility: as long as the trees have the same expected value, it does not matter how the decision makers evaluate them. If two decision makers face the choice depicted in Figure 1, it does not matter if one of them is risk-averse (preferring the certain outcome) and the other is risk-seeking (preferring the risky prospect). Even though the decision makers evaluate their prospects differently, their situation is equivalent from the perspective of initial equality of opportunity.
To be sure, Arneson’s identification of the value of options with prudence complicates the interpretation of this view.Footnote 11 Roughly, prudence has to do with how you allocate goods between your present and future (how much you save for retirement, say) and how you manage risks. A prudent person tries to avoid unnecessary and unreasonable risks, but assumes reasonable and worthwhile ones. So Arneson must have in mind a more complex concept than expected value. Notice that prudence is a substantive notion: prudent choices cannot be identified without considering the content of choices.
But let that pass. Is it plausible to maintain that there are no demands of justice when initial decision trees are equivalent with respect to their expected value? Consider
Figure 3to see why not.
Figure 3
The figure represents the decision trees that are faced by two individuals, A and B. The left-hand side of the figure with label (A) represents A’s situation, the right-hand side B’s. They have to make a choice which may then be followed by a random move by Nature; once again I am assuming that all outcomes following a move by Nature are equiprobable. Although these examples are greatly simplified, they are just like the sort of decision trees that defenders of initial equality of opportunity views have in mind. Their idea is to identify an initial situation from which subsequent branches – sets of choices and chance events – can be evaluated. Even if the details are more complicated, the basic idea is the same.
The first decision maker, A, can either choose a certain outcome by moving left or a risky prospect by moving right. If she moves left, she will get 3. This is the expected value of this alternative. If she moves right, she ends up with 2 or 4 with equal chances. The expected value of this choice is also 3 (0.5 × 2 + 0.5 × 4). The second decision maker, B, has no risk-free option. If he moves left, he ends up with either 0 or 6 with equal chances; if he moves right, he ends up with 1 or 5, also with equal chances. The expected value of both of these choices is 3. Therefore, A and B, in their initial choice situations, face equivalent decision trees. Consequently, initial equality of opportunity is satisfied. Whichever outcome each person ends up with raises no concern of justice. Wherever the chips may fall, they have no grounds for complaint.
But this does not seem right. In terms of risk, the two decision trees are very different. It is not unreasonable for B to complain that while he is compelled to take great risks, A has to choose between a safe option and one that, speaking loosely, has much less risk. He might reasonably prefer to be in A’s situation, given his preferences, values and tolerance of risk. That he is not is beyond his control – a matter of brute luck. There is a difference between his situation and A’s that initial equality of opportunity cannot account for. In order to properly evaluate their decision trees, A and B will have to make substantive judgements about the risks they face down the line. Thus, equivalence of decision trees is not a sufficient basis for determining the demands of justice.
6. REASONABLE RISKS AND THE NEUTRALITY OBJECTION
Some of these problems are recognized by some luck egalitarians. They have pursued different strategies to maintain the distinction between brute luck and option luck. One is to draw it in terms of reasonable avoidability. This modifies the luck egalitarian view to a certain extent. Martin Sandbu, for instance, writes:
Justice does not require us to hold people responsible for the risks they can only avoid by choosing unreasonable alternatives. But it does require us to hold people responsible for the risks they can avoid by choosing reasonable alternatives.Footnote 12
What matters, on this kind of view, is not merely whether an outcome is ‘the work not of people but of nature or brute luck’, but also whether you could have reasonably avoided it. Sandbu spells his proposal out the following way:
deliberate and calculated gambles do not give rise (only) to option luck when the safe option is sufficiently bad, that is, when risks can only be avoided by going below a minimally entitled level of resources
and
deliberate and calculated gambles do not give rise (only) to option luck when the safe option is sufficiently worse . . . than the risky choice, that is, when risks can be avoided only by incurring a high expected opportunity cost of safety.Footnote 13
An illustration of this view is given in
Figure 4. Assume that the ‘minimally entitled level of resources’ is 1. The decision maker, A, can choose the ‘safe option’ that leaves her below the minimum level, or take a gamble that can result in 2 or 4 with equal odds. (Alternatively, assume there is no minimum level, but 0 is sufficiently worse than the expected value of the risky prospect.) Thus, on this view, choosing the gamble is not an instance of option luck. If A ends up with 2, it is a matter of bad brute luck, and she has a claim to compensation.
Figure 4
One immediately apparent problem with this view is that the person choosing the gamble will have a claim to compensation regardless of the value of the outcomes of the gamble as long as the outcome of the certain prospect is below the minimum level. Thus, A would have a claim to compensation if the outcomes of the gamble were 200 and 400 with equal odds, rather than 2 and 4. Millionaires who take risks to become billionaires could complain of bad brute luck if they only become super-rich rather than mega-rich. If there are others like them who did become billionaires, they would have a claim to compensation on this version of luck egalitarianism.
Another problem is that the proposal has nothing to say in response to the problems I raised in Section 3. One way to understand the proposal is that a gamble is reasonable only when its outcomes dominate the (sufficiently bad) outcome of the certain prospect (that is, all of them are sufficiently better than the certain outcome). In that case, the problem illustrated by Figure 1 remains, since the worst outcome of the gamble is worse than the certain outcome. Thus, the gamble remains an instance of option luck. But that just leads back to the original problem. Sandbu’s proposal does not solve it.
There are other authors who attempt to capture the distinction between brute luck and option luck in terms of avoidability. Peter Vallentyne suggests (but does not endorse) the following view:
the occurrence of an event is due to brute luck for an agent if and only if the agent could not have (reasonably) avoided the possibility of its occurrence.Footnote 14
Vallentyne recognizes that mere avoidability won’t do: as I have argued, that you could have avoided a risk by choosing a certain outcome cannot be what luck egalitarians have in mind. Hence Vallentyne adds a reasonability condition: an outcome can be a matter of brute luck even if the person could have avoided it by making a different choice, but it would have been unreasonable for her to do so. An outcome is a matter of option luck when the person could have avoided it by making a different choice, and it would not have been unreasonable for her to do so.
This gets us to the heart of the matter: what makes a choice reasonable? Vallentyne suggests that a reasonable choice is that which is in the individual’s best interest – or at least adequately in her interests (Vallentyne Reference Vallentyne2002: 533). Of course, it can be seen immediately that this answer does not help with the situations illustrated in Figures 1 and 3: in those cases, it is not clear which choice is in the decision maker’s interest in the absence of a view on which risks are reasonable. For instance, what is in the person’s best interest in the example of Figure 1 is getting 2; but that’s not the correct question to ask. The correct question to ask is what is in the person’s best interest, given the risk she faces. So you can’t use the person’s interests to determine which choices are reasonable, and hence not matters of option luck, because your view on what is reasonable is supposed to determine what is in the person’s interests. Consequently, the account is circular.
Here is where my arguments go. Suppose that the distinction between brute luck and option luck indeed depends on reasonability. What determines whether the outcome of a choice is a matter of brute luck or option luck is whether the risk the choice includes is reasonable (or reasonably avoidable). However, whether a risk is reasonable is a substantive issue: it cannot be determined without judgements about the value of the opportunities the choice offers (that is, without judgements about its content, and, possibly, about the value of having the choice itself).Footnote 15 Hence there is no neutral, content-free way to determine which outcomes are matters of brute luck and option luck. Luck egalitarians either have failed to recognize this problem, or they have, for the most part, just ignored it.
One reason this problem can’t be ignored is that it underlies one of the standard objections to luck egalitarianism. Opponents of the view have long worried that luck egalitarianism singles out certain types of behaviours for which people are held responsible while it ignores others. Luck egalitarians have not exactly dispelled this impression by tending to rely on rough intuitive judgements about individual responsibility. Their examples usually feature smokers, drunkards, obese gluttons, reckless drivers or mountaineers, and adrenaline junkies. They sometimes suggest that morbidly obese people should be held responsible for needing knee replacements; they seldom suggest that former marathon runners should be held responsible for the same thing. It is often suggested that women who get lung cancer as a result of smoking should be held responsible for their health need; it is never suggested that women who get breast cancer as a result of postponing childbearing should be equally responsible.Footnote 16 Thus, there is a worry that luck egalitarianism collapses into hectoring moralism in practice.
This is known as the neutrality objection to luck egalitarianism. However, it is commonly considered merely as a practical objection to luck egalitarian views. That is, it is thought that as well as choosing better examples, luck egalitarians can meet it by building the appropriate safeguards into their theory.Footnote 17
But my arguments show that the neutrality objection isn’t merely a practical problem. It concerns the core of the theory. And luck egalitarians have not shown that there is a neutral way of drawing the brute luck–option luck distinction.
Luck egalitarians hold that it is unjust if some people are worse off than others through no choice or fault of their own, but it is not unjust if some people are worse off than others due to the choices they have made. When asked which choices can be attributed to people in the relevant sense, many luck egalitarians point to the distinction between brute luck and option luck: those, they say, which are a matter of option luck. This suggests that you just have to identify the choices that are matters of option luck in order to determine the demands of distributive justice. You can stay neutral with respect to the content of those choices. But it turns out that the brute luck–option luck distinction itself conceptually depends on judgements about the content of choices. Luck egalitarians cannot formulate the demands of justice in a way that remains neutral between preferences and values.
7. LUCK EGALITARIANISM IN HEALTH CARE
I am going to briefly illustrate my argument with an example from health care. This is an area to which luck egalitarianism has often been applied, partly because it is thought to have striking implications. These implications are more complex than is often recognized; nevertheless, I’m going to ignore these complications.Footnote 18
Undoubtedly, the most striking implication of luck egalitarianism for health care might be holding people responsible for the bad outcomes of their unhealthy behaviours and lifestyle choices. It seems that illness due to smoking, alcohol abuse, unhealthy diets or chosen pregnancies are matters of option luck and therefore provide no grounds for claims on the rest of society. In one way or another, luck egalitarian views punish those who have risked their health and ended up with a health need. Many critics of luck egalitarianism have argued that this implication is too harsh.Footnote 19
In his application of luck egalitarianism to health care, Shlomi Segall attempts to avoid the most implausible implications of the view by reformulating the distinction between brute luck and option luck. He defines brute luck as
the outcome of actions (including omissions) that it would have been unreasonable to expect the agent to avoid (or not to avoid, in the case of omissions). . . . [T]his unreasonableness criterion shifts the focus of attention from the individual to the society. It asks not whether the individual has acted in a reasonable way, but rather whether it is unreasonable for society to expect the individual to avoid a certain course of action.Footnote 20
As Segall points out, this formulation can avoid the most glaring problems. After all, it would be unreasonable for society to expect women to forgo childbearing, or to refuse treatment to women who have postponed childbearing for career reasons and thereby increased their risk of breast cancer. Perhaps it would also be unreasonable to hold people responsible for unhealthy lifestyle choices, like smoking or excessive alcohol consumption, when they have few other opportunities for recreation or pleasure.
But shifting the focus to society from the individual does not solve the neutrality problem. For illustration, consider a striking study. In this study, participants were asked to hypothetically allocate heart transplants between patients with and without a history of intravenous drug use, with and without a history of smoking, and with and without a history of eating an unhealthy high-fat diet against their doctor’s recommendation. A strong relationship was found in the responses between patients’ past behaviour and the willingness of participants to provide them with heart transplants. Participants recommended lower priority to patients with a history of eating a high-fat diet, even lower priority to patients with a history of smoking, and the lowest priority to patients with a history of intravenous drug use, as compared in each case to patients without such histories.
However, participants continued to recommend low priority to patients with a history of intravenous drug use even when it was explained to them that these patients had better survival rates than others after transplantation (ostensibly because of the effects of drug use on their immune system). The results were similar when smoking and unhealthy diet were used as examples. And there was no effect on the priorities determined by participants in any of the cases when they were told that past behaviour was not the cause of heart failure – when the participants apparently knew that the health need arose purely as a matter of bad brute luck. The researchers concluded that people’s judgements did not reflect beliefs about responsibility; rather, they reflected disapproval of certain behaviours, regardless of their contribution to health needs.Footnote 21
Suppose, for a moment, that the preferences of the participants of this study represent what ‘society’ can reasonably expect of individuals. After all, if reasonable avoidability is determined by societal expectations, the most straightforward way to determine their content is to survey people. The beliefs revealed in this study do reflect people’s expectations of how others should behave. But they do not reflect any consideration of what these hypothetical patients could have avoided – what outcomes were within their control or due to their own choice or what risks were reasonable. Instead, they seem to reflect mere prejudice.Footnote 22
Of course, this is just an illustration. Luck egalitarians will surely reply that running surveys (or relying on political institutions) is not the way the idea of reasonable avoidability should be given content. But what is the alternative? On this question, luck egalitarians who adopt the idea of reasonable avoidability remain, for the most part, silent. One exception is Segall, who, in later work, suggests employing the concept of prudence:
what I take prudence, and conversely option luck, to consist of [is this:] An action should be understood as prudent, on my understanding, when it would be unreasonable to expect the agent to avoid it (or undertake it, in the case of omission).Footnote 23
The circularity of this definition should immediately be apparent: prudence is defined in terms of what is unreasonable to expect an agent to avoid, and what is unreasonable to expect an agent to avoid is defined in terms of prudence. And option luck is defined as a matter of what is reasonable to expect the agent to avoid. Set up this way, the concept of prudence sheds no light on the distinction between brute luck and option luck at all. We have just introduced yet another epicycle in the search for a way of drawing the distinction.
Prudence in any case couldn’t be used to draw the distinction in a way that avoids the neutrality objection. What is prudent, just as what is reasonably avoidable, is a matter of substantive judgements: if you want to save up for your retirement, you can, say, put your money in a savings account or you can invest it in a more risky retirement fund that may increase or decrease the amount available to you when you retire. In the absence of further details, it’s not clear that the less risky option is more prudent. That claim would need to be defended on the basis of substantive considerations.
In sum, we are still left with the neutrality problem: what is prudent, avoidable, or reasonable to expect from people is a matter of substantive judgements. The shift to societal expectations just covers up this fact. Luck egalitarians still have not provided an account that can distinguish between brute luck and option luck in a neutral way. To be sure, perhaps they can swallow the bullet: they can abandon relying on the brute luck–option luck distinction, and accept that responsibility cannot be assigned in a value-neutral way. In other words, they can abandon thin responsibility luck – even if, for many luck egalitarians, it is the cornerstone of the theory. But these luck egalitarians have a price to pay: their view remains vulnerable to the charge of moralism. Ultimately, I suggest (though have not been able to argue for in detail), it forces them to give up their anti-perfectionism.
8. ECONOMICS AND PHILOSOPHY
Luck egalitarianism interests philosophers and economists alike. But they also tend to interpret the view differently. In this paper, I have focused on formulations that may be more familiar to philosophers. Hence it is worth addressing whether my arguments make any difference to the way economists think about luck egalitarianism.Footnote 24
One reason to doubt they do is that economists tend not to talk much about luck in their theories of equality of opportunity (their preferred name for luck egalitarianism). They focus, instead, on the distinction between choice (or effort) and circumstance. Whether any inequality in outcomes raises a concern of justice depends on whether it reflects differences in the circumstances of individuals or differences in their choices (the ‘effort’ they have made). While differences in circumstances are beyond individual responsibility, people can be held responsible for their choices. The fundamental distinction is between circumstance and choice, rather than brute luck and option luck.
But it should immediately be clear that this approach does not eliminate the problem. It just transforms the issue of distinguishing brute luck and option luck into the issue of distinguishing circumstance and choice. Take the example of the smokers I began with. Is it a matter of circumstance or choice that all these people smoke twenty cigarettes a day? On the one hand, they all made the choice to take up smoking; but, on the other hand, everyone smokes just as much in this society, and the unequal outcome reflects a difference in circumstance (people’s genetic propensity to develop lung cancer). Therefore, the kind of problems that I raised for luck can be raised for the distinction between choice and circumstance. As far as I can see, many of my arguments could be straightforwardly reformulated for this distinction.
To be sure, this claim would need a detailed demonstration that is beyond the scope of this paper. Nonetheless, a simple example might be helpful. Suppose that in the smokers example each person knows their own genetic propensity to develop lung cancer as a result of a history of smoking. Sure enough, those with the propensity do develop lung cancer. Those who do not are now asked to contribute to the treatment of those who ended up worse off. In response, however, they argue that they did not make the same choice as the others did. For them, the choice was whether to take up smoking knowing that they don’t have the genetic propensity to develop lung cancer. For the others, the choice was whether to take up smoking knowing that they do have the genetic propensity to develop lung cancer. Given this knowledge, one’s genetic propensity is not a matter of circumstance. It’s part of the choice one makes. Or so it could be argued.
Let’s vary the example a bit more. Suppose the people with the genetic propensity to develop lung cancer come from a background where smoking is prevalent. Everyone in their families and social environment smoked. Can’t they argue that for them smoking was part of the background conditions? When they took up smoking, they didn’t ‘really’ make a choice, as opposed to those for whom smoking was not part of the background conditions – and who therefore had to make a real ‘effort’ to take it up, even though all other relevant things between the two groups were equal. Then, for the first group of people, smoking might be considered a circumstance; for the rest, it might be seen as a matter of choice.Footnote 25
An objection that one could make to my analysis is this.Footnote 26 My examples seem puzzling only because I don’t take into account the difference between ex ante and ex post egalitarianism. Roughly, an ex post egalitarian holds that the inequalities that raise a concern of justice are to be found in the final outcomes between those who made the same choice, once the uncertainty has been resolved. An ex ante egalitarian, in contrast, holds that the inequalities that raise a concern of justice are those that occur before people make their choice, to be found in the prospects they face.Footnote 27 If you take the ex post perspective, the puzzles disappear. In particular, in all of the examples I have discussed, the inequalities in outcomes will be matters of brute luck.
There are a couple of problems with this objection. For one thing, it does not address the difficulty of distinguishing between circumstance and choice. For another, remember that many luck egalitarian philosophers are concerned with the question whether a choice is reasonably avoidable. In my initial example, perhaps they would say that the choice to take up smoking was reasonably avoidable. Hence, they might argue, the resulting inequality is not objectionable, and hence not a concern of justice, even though people made the same choice from identical circumstances.
This has nothing to do with the distinction between ex ante and ex post egalitarianism. Consider a similar choice about taking up an office job. People’s circumstances are the same, except that some of them will develop chronic lower back pain from working in an office. The choice of taking up the office job (let’s assume) is not reasonably avoidable (people have to work). So the resulting inequality, presumably, will be objectionable.
In these two examples, there is no objectionable inequality from the ex ante perspective. In both examples, however, there is an objectionable inequality from the ex post perspective. Now consider the examples from the perspective of reasonable avoidability. In the first one, presumably, there is no objectionable inequality, since the choice to take up smoking is reasonably avoidable. In the second one, however, presumably there is an objectionable inequality, since the choice to work in an office is not reasonably avoidable. For luck egalitarian philosophers, there is an important difference between the two cases, unrelated to the ex ante/ex post distinction.
9. CONCLUSION
The distinction between brute luck and option luck is fundamental for luck egalitarianism. Indeed, many luck egalitarians write as if it can be used to specify which outcomes people should be held responsible for. In this paper, I have argued that this is impossible to do. Luck egalitarians tend to rely on rough intuitive judgements about individual responsibility instead, which makes their view vulnerable to the neutrality objection. I have shown that attempts to avoid this objection are unsuccessful. The neutrality problem is not merely a practical problem for luck egalitarians; it stems from the core of their theory. Perhaps luck egalitarians will be tempted to say that they don’t need an account of the distinction between brute luck and option luck; perhaps they can leave it to society to determine what choices are reasonably avoidable.Footnote 28 But, as I have argued, the neutrality objection shows that their theory is hardly intelligible in the absence of such an account. As long as luck egalitarians don’t provide a defensible way to distinguish brute luck and option luck – or some other defensible account of attributing responsibility – their view remains incomplete.
ACKNOWLEDGEMENTS
I would like to thank participants at a workshop on Addressing Ethical Obstacles to Person Centred Care at Gothenburg University, a workshop on the Role of Personal Responsibility in Healthcare at the University of Oslo, and a seminar at the Chaire Hoover d’Éthique Économique et Sociale at the University of Louvain, Louvain-la-Neuve. I am also grateful to Andreas Albertsen, Shlomi Segall and Bruno Verbeek for discussion, as well as the editor of this journal for numerous suggestions for improvement. Financial support from the Swedish Research Council and from the Swedish Research Council for Health, Working Life and Welfare is gratefully acknowledged (2014–4024).
