Carl Knight has written an interesting and penetrating book about the strand of distributive justice commonly referred to as luck egalitarianism. While luck egalitarianism appears in different guises, the book is concerned specifically with the view that ‘. . . inequalities are justified only insofar as they reflect differential exercises of responsibility, and equalities only insofar as they reflect equivalent such exercises’ (230). In short, luck egalitarianism is the happy union of two plausible and forceful ideas; egalitarianism and responsibilitarianism.

Knight addresses many issues in the book. First of all, he takes on the intractable question of what exactly it is that is worth equalizing. Is it resources, welfare or advantage (to name some of the most well-known alternatives)? In the course of the two opening chapters, both equality of resources and equality of advantage are discussed and rejected. Equality of (opportunity for) welfare looks more promising, but Knight finds that its success depends ‘. . . on the availability of a robust conception of welfare’ (2). Thus, he sets out to refine welfarism, making it immune to some of the most important objections that have been pressed against it. The result is an affective-state theory. On this account, a person has a higher level of welfare the better her present mood is (71).

While it might well be the case that the present mood account avoids some familiar objections, it seems open to lines of criticisms that Knight does not address. For instance, assume that there are two mutually exclusive ways of benefiting a group of badly off people. Either we can shift resources from the better off to the badly off, thus enabling the latter to lead good lives, or we could provide the badly off with a cheap mood-enhancing drug. Even if the drug was efficient and free of side-effects, my view is that we should shift resources rather than handing out prescriptions, even if the latter would provide better moods than the former. My view could of course be wrong, but it seems nonetheless that Knight's present mood account could do with a more extensive defence.

In chapter 3, Knight attempts to define what an egalitarian theory really is. Three relevant criteria are presented. First, an egalitarian theory ‘. . . considers the bare fact that a person is in certain circumstances to be a conclusive reason for placing another relevantly identically positioned person in the same circumstances, except where this conflicts with other similarly conclusive reasons arising from the circumstances of other persons, in which case a compromise must be reached’ (92). Secondly, ‘. . . the theory . . . can be stated as ‘equality of x for all persons’, making no explicit or implicit exclusion of persons or individuals and showing no greater concern and respect for some rather than others’ (98). Lastly, ‘. . . the theory pursues equality in a dimension that is valuable to egalitarians’ (101).

The main question, in the context of the book, is whether or not luck egalitarianism can count as a version of egalitarianism proper. The question arises because, first, many theories ordinarily thought to fall (far) beyond the realm of egalitarianism nevertheless mandates equality of something. Libertarianism holds that every person should have an equal set of libertarian rights, while utilitarianism holds that each person's welfare or utility is equally important. Secondly, luck egalitarianism is a theory that, despite its name, potentially allows for quite substantive inequalities. The conclusion, however, is that neither libertarianism, nor utilitarianism (or prioritarianism) can reasonably be seen as egalitarian theories. Luck egalitarianism, on the other hand, meets all three criteria (117, 161).

Chapter 4 addresses the substantial literature criticizing luck egalitarianism, and Knight argues, quite persuasively, that the criticism is to a large extent overstated, at least insofar as the theory is given a welfarist interpretation. He concedes, however, two points. First, that luck egalitarianism in some circumstances might be too harsh on the irresponsibly badly off. If a person through her own choices winds up in a state of misery, the theory does not preclude the view that she should be left to her own devices. On the other hand, luck egalitarianism does not preclude the view that she should be helped either, for instance through mandatory universal health schemes. Moreover, most luck egalitarian theorists make sure that some safeguard is in place for the negligently badly off (140–141). Secondly, Knight allows that luck egalitarianism faces a challenge with regards to dependent caretakers. Some persons may suffer shortfalls in welfare because they choose to pursue valuable and worthy tasks. Knight argues that this problem can be overcome by regulating markets partly on the basis of the social value of different pursuits (123).

In chapter 5 it is argued that responsibilitarianism so near as entails luck egalitarianism, because it is implausible to hold people responsible for their welfare levels in such a way that two equally responsible persons justifiably enjoy divergent welfare levels (170). This welfare gap, since it is not a result of varying displays of responsibility, must stem from factors over which the individuals have no control, and it is thus unjustified, at least if responsibilitarianism is to be taken seriously (176–178).

The chapter also includes a more metaphysical discussion. Since luck egalitarianism makes distribution sensitive to responsibility and choice, it seemingly presupposes a libertarian or compatibilist view of free will. Therefore it could be argued that the theory is incompatible with hard determinism, and that it rests on insecure metaphysical ground. Knight, however, argues that if determinism is true, then no display of (real) responsibility is possible, and if so, luck egalitarianism holds that no inequalities are justified. Pure outcome egalitarianism, on the other hand, is less plausible, or so the author argues, because that theory is not flexible in the way luck egalitarianism is (181). It is wholly unresponsive to whether or not genuinely free choices are possible. This puts outcome egalitarianism in a worse position than luck egalitarianism.

It seems to me that Knight is completely right in arguing that luck egalitarianism is compatible with different views on free will. But it is less convincing that the theory is in this regard superior to outcome egalitarianism. This would be true if outcome egalitarians relied on the truth of hard determinism as a justification for their responsibility-insensitivity. If that was the case, however, outcome egalitarians would really just be luck egalitarians with a particular metaphysical standpoint. Alternatively, and more likely, outcome egalitarians hold the position that outcomes should be equal regardless of whether or not determinism is true. That may be implausible, but if so, it is implausible for normative reasons. It seems that Knight tacitly assumes that outcome egalitarians ought to take responsibility into consideration if it exists. But that is just to assume that outcome egalitarians ought to be luck egalitarians. It is not to show that outcome egalitarians are in a metaphysical tight spot.

In any case, luck egalitarianism is so far looking good, but in the last chapter, the theory is really put to the test. Here Knight discusses a range of criticisms that all relate to the well-known levelling down objection to (standard) egalitarianism. This objection exploits the fact that inequalities can be eradicated by making all individuals worse off. Since it seems that a situation in which A and B enjoy welfare levels of 9 and 10, respectively, is better (in all respects) than a situation in which they both have welfare levels of 2, say, many have argued that egalitarianism is implausible.

The levelling down objection is a challenge to luck egalitarianism as well. If A and B are equally responsible, the objection applies straightforwardly. If, on the other hand, A and B both enjoy welfare levels of 10, but A has behaved more responsibly, then a distribution in which A has a welfare level of 2, and B a welfare level of 1, might be better according to luck egalitarianism (depending, of course, on the extent to which A's and B's responsibility levels diverge). In the former (10, 10) scenario A and B have not had equal opportunity for welfare, and hence the equality is unjustified, but in the latter (1, 2), they have indeed had equal opportunity for welfare (or so we can stipulate), and the inequality is justified. As noted, further related objections are similarly problematic for the theory. These all show that luck egalitarianism has trouble with inefficiency – ‘. . . the use of resources in ways which benefit no one, or produce much smaller overall benefits for society than alternative uses . . .’ (203). In response, Knight proposes that a version of responsibility-sensitive prioritarianism might be a better theory of distributive justice.

This proposal is heavily influenced by Arneson's original responsibility-catering prioritarianism, which consists of three elements: ‘The moral value of a gain is (1) greater, the greater the welfare or utility it affords the person, and (2) greater, the lower the person's lifetime utility or welfare prior to receipt of this gain.’ In addition,

Knight accepts (1) and (2), the combination of which yields a form of prioritarianism. He wants, however to adjust (3). First, he suggests that the priorities in (3) should be lexical. Further, he proposes a series of refinements of (3) in order to make it more clearly responsibilitarian. ‘The basic idea . . . is that persons should receive what they are responsible for, except where giving them something they are not responsible for disadvantages nobody’ (224, emphasis in original).

Some questions arise in relation to Knight's revised responsibility-catering prioritarianism. First, very little is said about how we are to weigh the different elements of the theory against each other. Consider the following scenario. A is very badly off. While her display of responsibility is such that she is only entitled to a very low level of welfare, her actual welfare level is a tad lower even than that. Suppose that A can gain one unit of welfare if we take one unit of welfare away from a million people who are just slightly better off than A, and who all have a tad more welfare than their responsibility levels entitles them to. (I assume here that it is extremely costly to raise A's welfare by one unit). What would the all-things-considered view be like? The reason the question arises, of course, is that both (2) and (3) mandate that the transfer should be made. A is worse off than the million other individuals, and the resulting distribution would bring a million and one individual's welfare levels more in line with their respective levels of responsibility. But this runs directly in the face of the utilitarian element in (1). How much does a massive welfare loss count compared to the pull of (2) and (3)?

Surely, any theory of this kind will have to face some sort of weighing between different principles, but if what Knight envisages is a form of radical pluralism of the three ideals, this should be explicitly mentioned, and if not, some suggestion as to how to resolve conflicts should be offered.

The scenario described above (A and the million others) also raises a different question. Knight suggests that (2) is a version of leximin (217). Given that (3) is also given a lexical interpretation, both (2) and (3) imply that A should receive the benefit, even at the cost of taking a unit of welfare away from a million people. This suggests that responsibility-catering prioritarianism too is threatened by inefficiency, in the form of expensive resource transfers, even though the extent of this inefficiency of course depends on the way in which the trade-offs should be made. However, even when tempered by the utilitarian element in (1), the theory at least seems to mandate that resources should sometimes be used in a way that produces ‘. . . much smaller overall benefits for society than alternative uses . . .’ (203).

Now, whether one finds this implausible or not, will depend on one's views on inefficiency, and perhaps also on the absolute welfare levels of the disadvantaged. In any case, Knight's abandonment of luck egalitarianism in the first place was caused by inefficiency problems. The inefficiency related to his version of responsibility-sensitive prioritarianism is clearly less pervasive than the levelling-down problems haunting egalitarianism, not least because of the utilitarian part of the theory. It might be the case that responsibility-catering prioritarianism is suited to strike a balance that accords with our considered judgements (209), but it is nevertheless surprising that Knight takes no explicit note of this potential problem.

Further, responsibility-catering prioritarianism will sometimes call for increased inequalities. Assume that A has a low level of welfare, and that this level is in line with her display of responsibility. Assume further that B and C have higher welfare levels than A, but that their displays of responsibility are such that they are entitled to an even higher welfare level. Now, if by removing one unit of welfare from A we could increase the welfare levels of both B and C by one unit (it is cheaper to increase their welfare), so that their levels of responsibility now are in line with their welfare levels, it seems that we, on balance, should do so. A is worse off, and undeservedly so, but more welfare obtains, and there are now more people who have welfare levels that accord with their levels of responsibility. And if the leximin element in (2) denies this particular distribution, it may not do so if we add more people on the beneficiary side. (On a related note, see Otsuka and Voorhoeve (Reference Otsuka and Voorhoeve2009) for a detailed argument to the effect that prioritarianism violates the separateness of persons.)

Perhaps this is the price (former) luck egalitarians have to pay for avoiding the levelling down-objection, and perhaps they should accept this cost. But I do not think that Knight provides sufficient reasons for concluding one way or the other. More generally, it seems that while luck egalitarianism is thoroughly scrutinized and consequently found indefensible, responsibility-catering prioritarianism is not given the same treatment. It appears plausible at the outset that prioritarianism avoids the objections found fatal to luck egalitarianism, but it is less clear that prioritarianism is the better theory all things considered.

Nevertheless, this is a book well worth reading for all those interested in contemporary distributive justice. The exposition and assessment of luck egalitarianism is enlightening, thought-provoking and useful. The same can be said about the development of responsibility-catering prioritarianism, even though, as indicated, a more critical assessment of the latter would have further strengthened the book.