[Rescuing Justice and Equality. By G.A. Cohen. Harvard University Press, 2008. Xvii + 430 pages.]
The title of G.A. Cohen’s remarkable book suggests an obvious question. Cohen wishes to rescue justice and equality; but from whom or what are these in danger? Cohen’s target will strike many readers as surprising: it is the false views of John Rawls that he wishes to combat. Rawls’s famous difference principle (more exactly, the first part of it), as Rawls thinks it will apply, allows inequalities that justice properly excludes. Further, Rawls holds wrong positions in metaethics as well: he wrongly makes moral principles depend on matters of fact and he confuses the nature of justice, a question about the essence of a concept, with “rules of regulation,” properly a matter for decision.
It does not follow from this massive dissent from Rawls that Cohen thinks badly of A Theory of Justice. Quite the contrary, he regards the book as one of the greatest works of political philosophy ever written: “I [Cohen] believe that at most two books in the history of Western political philosophy have a claim to be regarded as greater than A Theory of Justice: Plato’s Republic and Hobbes’s Leviathan“ (p. 11). 1
Cohen’s attitude toward Rawls to a large extent parallels my own reaction to Rescuing Justice and Equality. It is a major work, the product of a philosophical intelligence of great depth and power. Few if any contemporary thinkers can match Cohen in his ability to grasp what is at stake in an argument and to raise devastating objections. (Those who wish a quick demonstration of Cohen’s skill should read his painstakingly minute dissection of Andrew Williams on the “publicity argument,” pp. 344 ff.) He possesses, in addition, the ability to extract every drop of meaning from a long and complex text. Even more important, the questions he discusses are of the highest significance and his answers to them often profound. Cohen does philosophy the way it should be done.
That said, I often cannot accept Cohen’s arguments. One might think, “Of course: how can a libertarian who rejects Rawls as overly egalitarian fail to disagree with someone who criticizes him from the left?” But I do not propose in what follows to evaluate Cohen from a libertarian standpoint. Rather, setting aside to the extent I am able my own beliefs, I shall endeavor to consider the book’s arguments on their own terms.
In order to understand Cohen’s criticism of the difference principle, it is essential to bear in mind his own conception of justice. He is a “luck egalitarian”: people with greater-than-average talents and abilities should not in justice receive more wealth and income than others, even if their work is more productive and valuable than their less-fortunately-endowed coworkers. People do not deserve the abilities by which they surpass others, and
my own animating conviction … [is] that an unequal distribution whose inequality cannot be vindicated by some choice or fault or desert on the part of (some of) the relevant affected agents is unfair, and therefore, pro tanto, unjust, and that nothing can remove that particular injustice. (p. 7)
(Cohen speaks here of “distribution” in general, but I think the chief issue between him and Rawls concerns the distribution of wealth and income in particular. At any rate, my discussion will be confined to this.) 2
Cohen thinks that this view of equality is intuitively plausible, and he suggests that one difference between the Oxford-style philosophy that he practices and Rawls’s Harvard approach is that the latter relies more on theory: “Oxford people of my vintage do not think that philosophy can move as far away as Harvard people think it can from pertinent prephilosophical judgment” (p. 3). I wonder whether the contrast to which Cohen calls attention is quite as sharp as he thinks: does not Rawls’s “reflective equilibrium” rely to a great extent on our moral intuitions? (Cohen does mention reflective equilibrium a few times but says little about it.) But Cohen’s stress on moral intuitions does help explain the clash between libertarians and their opponents. Those of libertarian inclinations tend not to hold it unfair for those with superior talents to benefit from them. Given this difference, the intractable disputes between egalitarians and libertarians are readily understandable. 3
“Those of libertarian inclinations tend not to hold it unfair for those with superior talents to benefit from them.”
If, roughly speaking, justice is equality, then Rawls’s difference principle must be rejected. Rawls allows inequalities to the extent that they benefit the least-well-off class in a society. 4 In what way can such inequalities benefit the worst off? Primarily, Rawls has in mind incentives: people who receive more money will respond by working more. Also, incentives may direct people into jobs that increase benefits to the poor. 5 The results of these incentives may increase the gains to the worst off over their share in a regime of strict equality.
Cohen is not content to counterpose his principle to Rawls’s. He deploys a formidable battery of arguments designed to show that the difference principle undermines itself. In a case where, because of the effect of incentives, the worst off benefit from inequalities, can we not readily construct a situation in which they will benefit still more from equality? We have only to imagine that the talented do not require incentives in order to work harder or to shift to jobs that secure greater benefits to the poor. In that case, the poor could gain still more, since some or all of the incentive payments would go to them.
If the talented are committed to egalitarian justice, must they not try to act in exactly this way? It will not do to answer that they are unable to do so, that it is just a psychological fact about many people that they respond positively to incentives. They might be able to change, if they tried; and if they could not, is it not likely that different social institutions, prolonged over time, could alter people’s attitudes?
But why should people try to act in this way? Here Cohen appeals to what he takes to be a fundamental principle: the personal is the political. We cannot insulate our fundamental commitments to justice in the institutions of the “basic structure” but must act justly in our personal lives.
The difference principle can be used to justify paying incentives that induce inequalities only when the attitude of talented people runs counter to the spirit of the difference principle itself; they would not need special incentives if they were unambivalently committed to the principle. (p. 32) 6
(Given this understanding, Cohen can say that he “accepts” the difference principle, since if talented people do not require monetary incentives, the principle as it pertains to these is vacuously true, and there may be nonmonetary incentives that would not trouble an egalitarian; but taken as Rawls thinks it will apply, in which talented people do respond to monetary incentives, he rejects it.)
I think that Cohen is right that the principles of justice apply to our daily lives as well as to the basic structure, but this point leaves a Rawlsian position intact. Let us imagine ourselves to be egalitarians: how would we wish to distribute income and wealth? Equal shares for everyone obviously suggests itself. Suppose, then, that everyone earns the same hourly rate of pay, regardless of his productivity or type of employment. At this rate, each person decides how many hours to work and in what job. Then, relative to the preferences for work and jobs that each person has, we have an equal distribution of income. Suppose now that incentives are allowed, provided that the shares of the worst off are better than they are under equal distribution.
Does this not remain a coherent conception of equality? Each person, to reiterate, is guaranteed at least as much as he would get under absolute equality, given the preferences for work and jobs of people in that situation. Further, luck egalitarianism has some effect, though it is not controlling, since the talented gain greater shares only if they benefit the worst off. True enough, the worst off do not fare as well as they would under absolute equality, if talented people alter their self-interested preferences to benefit the worst off as much as possible. But all this shows is that there is another, and more demanding, conception of equality besides the one I sketched out. The “spirit” of the first conception does not lead to the second; rather, they are two different positions. My first conception is of course roughly Rawlsian. If I am right that it is coherent, Cohen has failed to overthrow it. He has merely contrasted it with his own position.
Cohen maintains that an acceptable principle of justice must preserve a sense of community among those subject to it. To do so, it must pass what he calls a “comprehensive justification,” in which not only a policy, but the behavior of those who act under the policy is justified.
Now, a policy argument provides a comprehensive justification only if it passes what I shall call the interpersonal test…The test asks whether the argument could serve as a justification of a mooted policy when uttered by any member of society to any other member. (p. 42, emphasis in original)
The difference principle, taken to justify large inequalities, in his opinion fails this test: a talented rich person could not in good conscience present it to a poor person. But I fail to see why this is so. Suppose he says, “When I decide whether to work additional time or whether to shift jobs from what I most prefer to what would help you more, I take account of your interests. I’m willing to have part of what I earn go to you. But I rate my own interests higher than yours: I will respond positively to monetary incentives that allow me to retain a greater share of what I produce.” Why is this unacceptable? Further, the talented person who says this is not refusing tout court to give additional funds to the poor beyond what the difference principle mandates: he is simply denying that he is, on grounds of justice, required to do so.
Cohen seems to have in mind the fact that the rich person’s own decisions make what he says true: if he acted more altruistically, he would not respond to these incentives. But surely this feature by itself does not suffice to render an interpersonal justification improper, even in cases where my decisions lead to an outcome the person I address would prefer otherwise. Suppose that I say to someone about to steal my computer, “You can’t take that! It’s mine.” The truth of “it’s mine” depends on my behavior, since I could, if so inclined, donate the computer to the thief. But this hardly shows a moral defect in what I say.
The case that the difference principle passes the test becomes even stronger when one adds that Cohen himself sometimes countenances a limit to equality because of “personal prerogative.” (The term comes from Samuel Scheffler.) Cohen does not say what may legitimately be included in the personal prerogative, and he is not committed to allowing a prerogative at all; but if it includes the freedom to choose one’s occupation and one’s choice between work and leisure, incentives may be the cost of inducing the talented to shift jobs or to work harder in ways that benefit the worst off. It is evident that Cohen does not think the prerogative extends this far; but I fail to see why the view that it does cannot pass the internal justification test; and if it does pass it, then even luck egalitarianism does not rule out incentives.
Cohen, in response to David Estlund, acknowledges that the prerogative argument has some force but suggests it will not much help Rawls. 7 Does not Rawls endorse inequality, regardless of the motives of the better off, so long as this helps the worst off? I do not understand how “motives” enter the picture, unless the talented aim to accumulate money just in order to have more than others. That attitude seems inconsistent with a commitment to egalitarianism. But it is not inconsistent with egalitarianism to value material goods highly or to demand a high price to work more or to shift jobs All that is relevant is that, absent the monetary incentives, the talented would prefer working less hard or in other jobs. Perhaps Cohen is thinking of a case where someone receives more money that is needed to get him to work in a way that best helps the worst off, but Rawls’s difference principle would not be satisfied in that situation: the excess could be redistributed to the poor without loss. (If the talented misstate their own preferences for strategic reasons, this is Cohen’s “bluffing case” [pp. 57 ff].)
Cohen now responds to a related argument in defense of the difference principle. Suppose that one begins from equality. If inequalities are allowed, everyone will be better off: the distribution with the inequality is strongly Pareto superior to an equal distribution. As in the previous argument, incentives lead the talented to produce more. Do we not have good reason to endorse the inequality? Is it unjust to do so?
Cohen’s answer is typically ingenious. If everyone gains, then can we not construct another Pareto-superior distribution by reallocating the unequal gains, in that way preserving equality? Would not an egalitarian have to say that this distribution is better than the Pareto-superior distribution that fails to preserve equality?
But what if, under these conditions, the talented do not want to work enough, or in the right occupations, to generate the increased income for everyone? Would not the new distribution then require a “Stalinist” policy in which the state ordered people to assume certain jobs? Further, if people’s preferences for occupations were counted in determining whether they are better off, the new distribution would not be Pareto superior. (This “freedom” objection actually comes up later in the book, but I think it is best addressed here.) Cohen responds that the talented could freely decide to work in the way required for the gains, even if doing so meant that they would have to accept jobs they would otherwise reject.
He thinks that people committed to equality should be willing to accept such conditions. Talented people usually have jobs that are highly fulfilling; and this is likely to remain so even if they do not work in their highest-valued activity. A doctor who prefers to be a gardener should not abandon medicine, if work in that profession better promotes a higher equal standard of income for all, so long as work as a doctor is satisfying. Cohen does not think that people are morally required to accept work they find onerous; but they may not insist on working in their most-preferred field, if they acknowledge the demands of egalitarian justice. He does not discuss extensively the preference for leisure, but I suspect he would adopt a similar line: people cannot insist on as much leisure as they like, so long as they have a reasonable amount compared to others. 8
It is now clear why the appeal to personal prerogative does not much move Cohen. He has a demanding notion of what egalitarian justice requires. But once more, I do not think that he has succeeded in generating any internal pressure on the difference principle. Why cannot a Rawlsian simply deny that people are morally required to shift their preferences in the way that Cohen wants? If Cohen then says that if they do not, they depart from the demands of equality, that is true only if one accepts his own conception of equality — or one similarly demanding.
In his assault on the Pareto argument for inequality, Cohen claims that its proponents fall into a fallacy committed by Nozick’s Wilt Chamberlain argument. Nozick, it is claimed, argued that if one begins from a patterned principle like equality, then one would have to move to accepting inequalities. People have the right to spend their equal shares as they wish; but what if their spending results in Wilt Chamberlain’s having a great deal of money? Would not egalitarians have to acknowledge that this situation preserves justice? But, Cohen says, a common criticism is that Nozick
fails to see that the principles that enjoin D1 [the initial situation of equality] also prohibit the move to D2 [the unequal outcome]: Nozick takes D1 as established, and he succeeds in disestablishing it only because he ignores what established it. (p. 170) 9
In like fashion, says Cohen, you “cannot begin with equality because all inequalities are morally arbitrary in origin, and therefore unjust, and then treat an unequalizing Pareto-improvement as lacking all stain of injustice” (p. 170). This does not correctly represent the Rawlsian starting point. The claim is not that all inequalities are unjust, but rather that since the possession of superior talents is arbitrary from the moral point of view, the extra productivity generated by these talents does not by itself generate a claim to greater shares. It does not follow from this claim that any deviation from equality is unjust. Equality for the supporter of Rawls is only a starting point, not an absolute requirement of justice. The Pareto argument is then not vulnerable to the objection alleged against Nozick.
Cohen next addresses the “basic-structure” objection to his criticism of the difference principle. This objection denies that people are required in their personal lives to act in accord with the demands of egalitarian justice. So long as basic institutions of society, e.g., the tax system, are in order, individuals are free to act from whatever motives they wish. (They must, of course, obey laws that require redistributive taxation.) Cohen, in this view, is wrong to claim that the personal is the political. I think that Cohen successfully rebuts this objection: he shows that there is no clear way to separate the basic structure from more local institutions, including the family, and personal behavior. But the main problem I have so far found with Cohen’s discussion is not a version of the basic-structure objection. I do not claim that accepting the difference principle has no consequences for individual conduct. Cohen and Ronald Dworkin may well be right, e.g., that people must cultivate an “egalitarian ethos.” My difficulty involves the nature of those consequences. Cohen has without adequate warrant seen these consequences in the light of his own luck egalitarianism.
Cohen makes clear the sort of response he would offer to my argument in chapter 4, “The Difference Principle.” I claim that Rawls is not a luck egalitarian; Cohen answers that to a large extent he is. He acknowledges that there is some evidence to the contrary. The “canonical” interpretation of the difference principle allows inequalities that do not benefit the worst off so long as they do not harm them. Nevertheless, Cohen thinks that at heart, Rawls is fundamentally sympathetic to luck egalitarianism. He quotes a passage from A Theory of Justice in which Rawls criticizes the “liberal interpretation of the two principles of justice” (i.e., equal opportunity without the first part of the difference principle). This “still permits the distribution of wealth and income to be determined by the natural distribution of abilities and talents” (p. 166, quoting TJ, pp. 73–4, emphasis added by Cohen). Concerning this passage, Cohen says,
This implies that the income distribution should not be determined by the talent distribution, and not merely that the fact that an income distribution reflects the talent distribution does not justify it. (p. 166)
Is there not, though, an ambiguity in “determine”? For Cohen’s reading to hold, one must take this to mean “in any way affect” rather than “totally bring about.” Suppose, e.g., that someone claims that tenure decisions should not be determined by the number of publications in top-ranking journals. He would more plausibly be taken to be claiming that these publications should not be the sole consideration, not that they should have no influence at all. In like fashion, I suggest, Rawls is better read as saying the superior talents should not by themselves generate the income distribution, not that they cannot properly in any way affect it.
Admittedly, this is merely a different reading from Cohen’s, and the point is doubtful. Much more important is another passage of A Theory of Justice, in which Rawls seems directly to address luck egalitarianism, or something very much like it, when dealing with equality of opportunity:
First we may observe that the difference principle gives some weight to the consideration singled out by the principle of redress. This is the principle that undeserved inequalities call for redress; and since inequalities of birth and natural endowment are undeserved, these inequalities are somehow to be compensated for… [The principle of redress] is plausible as most such principles are only as a prima facie principle, one that is to be weighed in balance with others.… Now the difference principle is not of course the principle of redress.” (TJ, pp. 100–01, see also the corresponding passage of the 1999 edition, p. 86)
Cohen might respond that matters change when the distribution of wealth and income, rather than equality of opportunity, is on the agenda. But it seems to me much more plausible to think that Rawls’s attitude does not change. Here too, the desire to counteract the results of undeserved inequalities has considerable force, but this consideration is not alone decisive.
I think that a principal use of the moral arbitrariness claim for Rawls is defensive: it is his reply to the supporters of the “system of natural liberty” who claim those who produce more deserve more income and wealth. Taken this way, incidentally, Rawls has an answer to Nozick’s famous objection: even if you do not deserve to benefit from your superior talents, it does not follow that you are not entitled to do so. Rawls can say, “On my theory, you aren’t entitled to do so, unless this satisfies the difference principle.” 10
As we shall later see, Cohen has one more powerful argument to defend his alignment of Rawls with luck egalitarianism; but to understand it, we must first consider what he says about metaethics. He argues that
a [moral] principle can reflect or respond to a fact only because it is also a response to a principle that is not a fact. To put the same point differently, principles that reflect facts must, in order to reflect facts, reflect principles that don’t reflect facts. (p. 232, emphasis removed)
Judged by this standard, Rawls is weighed in the balance and found wanting: for him, moral principles do depend on facts.
In defending this view, Cohen presents a brilliant discussion of whether one can derive an “ought” from an “is.” I at first thought that his principle of factual insensitivity begged the question against the view that “ought” follows from “is”: does not claiming this make the derived “ought” dependent on facts? Further, before reading Cohen, I thought that the anti-Humean position fell before an objection. If, e.g., one claimed that from “Human beings have X as their natural end” it follows that “Human beings ought to pursue X,” is one not in fact assuming that “One ought to pursue one’s natural end”? But then, I thought, there is no derivation of an “ought” from an “is”; rather, one is applying a premise about an “ought.”
Cohen solves both problems at once. The proponent of the derivation thinks that, to take my example, “Human beings have X as their natural end” semantically entails “Human beings ought to pursue X.” But if he holds this, he must also accept, “If human beings have X as their natural end, they ought to pursue X.” This thesis does not depend on the existence of any facts, so the defender of the derivation does not violate the factual insensitivity thesis. Further, it does not at all contradict the claim that an “ought” follows from an “is” that one accepts an “if, then” general premise. Rather, precisely the point of the anti-Humean thesis is that if one understands the meaning of the factual claim, one cannot rationally refuse to accept the general principle. (Cohen takes no position here on whether the anti-Humean thesis is true.)
I venture to suggest one slight modification of Cohen’s thesis. He says,
Suppose that proposition F states a factual claim and that, in the light of, on the basis of, her belief that F, a person affirms principle P. We may then ask her why she treats F as a reason for affirming P. And if she is able to answer that question, then her answer, so I believe, will feature or imply an affirmation of a more ultimate principle (call it P1) … a principle moreover, that holds whether or not F is true and that explains why F is a reason for affirming P. (pp. 233–34, emphasis in original)
Cohen wants to say that the truth of P1 does not depend on the truth of F, but I suggest that it is not right to express this by saying that P1 “holds whether or not F is true.” This overlooks the possibility that F is metaphysically necessary. In that case, P1 is not dependent on the truth of F, but it is not metaphysically possible that F be false.
Cohen might respond that if the truth of F is metaphysically necessary, then “If F were not true, P1 would still be true” is a true counterfactual with a necessarily false antecedent, but I doubt that Cohen wishes to go in this direction; and, in any case, the semantics of such counterfactuals are controversial. In would be better, I think, to state the factual insensitivity thesis in a way that avoids this problem. Cohen may think that my problem could not arise because there are no metaphysically necessary facts; but if he thinks this, he needs to say so.
Cohen makes another very useful metaethical distinction, and this distinction returns us to luck egalitarianism and the difference principle. He distinguishes between fundamental ethical principles, such as principles of justice, and rules of regulation. The former are matters for investigation, not decision. If luck egalitarianism is correct, e.g., this is not because people have decided to accept it: it either is correct or not. By contrast, we do decide which rules of regulation to adopt. These are general rules that determine how a society is organized. To arrive at them, we consider the principles of justice, along with other virtues, as well as facts about society. On the basis of all these things, we decide what to do.
Thus, contrary to what readers might so far have surmised, a society run as Cohen wishes might well allow for inegalitarian incentives. The difference from Rawls is that there would be no claim that such incentives were consistent with justice. Rather, they violate justice but, in the existing circumstances, allowing them is judged best to do when all the virtues are weighed against each other.
Rawls fails to grasp this distinction. He makes the principles of justice matters for decision. People in the original position decide what principles will best advance their interests; and in doing so, they take account of general facts such as the susceptibility of people to monetary incentives. To view justice in this way is to confuse principles of justice with rules of regulation.
It is because Rawls misses this that he does not grasp his real commitment to luck egalitarianism. He officially accords luck egalitarianism some weight but balances it against other considerations, such as incentives. If he had made the proper conceptual separation, the true strength of his commitment to luck egalitarianism would have been evident.
I do not think that Cohen’s argument succeeds. Suppose, as seems reasonable, that one agrees that the principles of justice are not to be equated with the rules of regulation for a society; only the latter, not the former, are matters for decision. It does not follow from this that it is false that the way to discover the principles of justice is to ask what rules for society self-interested people in the original position would choose. (These rules are not, strictly speaking, rules of regulation for the original position, since they are based on self-interest, not a balancing of the various virtues.) Rawls, at least in A Theory of Justice, does not think that it is a matter for decision that the principles of justice are to be arrived at in this way: this is rather something he holds to be correct. Cohen may respond that unless Rawls did confuse principles of justice with rules of regulation, he would not think it is plausible that the principles of justice are to be discovered in this way. I do not know whether this is true; my claim is that Cohen has not shown that, because of the way Rawls thinks the principles of justice are derived, he is guilty of the confusion in question. I conclude that no more than before has Cohen shown that Rawls is “really” a luck egalitarian.
Although I disagree with many of its arguments, Rescuing Justice and Equality is a great book. Every reader will gain much by studying it. Libertarian readers will see how a first-rate mind views justice in a very different perspective from our own.
- 1Cohen concentrates entirely on A Theory of Justiceand does not discuss defenses of the difference principle that rely on the “public reason” of Political LiberalismIt is clear, though, that he is thoroughly familiar with that book.
- 2For an important criticism of luck egalitarianism by a nonlibertarian, see S.L. Hurley, Justice, Luck, and Knowledge (Harvard, 2003) and my review in The Mises Review, Summer 2003. Cohen does not accept Hurley’s criticisms: see his “Luck and Equality: A Reply to Hurley,” Philosophy and Phenomenological Research 72<.em> (2006), pp. 439–46.
- 3For a view that accepts some, but not all, of the luck egalitarian intuition, see Thomas Sowell, The Quest for Cosmic Justice (Free Press, 1999) and my review in The Mises Review Spring 2000.
- 4Actually, there is a distinction between a strict and lax interpretation of this principle. According to the former, “inequalities are forbidden unless they render the worst off better off” (p.157); but the lax interpretation, also known as the canonical or lexical account, ranks a distribution that increases the gains of only the better off higher than one that does not, so long as distributing the increase to the worst off would not improve their overall position. This complication will surface at one point later.
- 5Cohen cites Philippe van Parijs on other ways inequalities can benefit the worst off, e.g., concentrating profits in the hands of good investors, but these are not his primary focus (pp. 320–21). Another benefit of inequality is that goods initially profitable only as luxuries that only the wealthy can purchase often later spread to the masses when cheaper production techniques are devised. See on this Friedrich Hayek The Constitution of Liberty (Chicago, 1960), pp. 42ff and Bertrand de Jouvenel, The Ethics of Redistribution (Liberty Press, 1990).
- 6Cohen mentions Marx and the feminists as sources for connecting personal behavior with institutions, but in an odd sense, Nozick also anticipates his discussion. Nozick maintains that because people do not follow the difference principle on the “micro” level, this gives us some reason to reject the principle as part of the basic structure. See Anarchy, State, and Utopia (Basic Books, 1974), pp. 204ff. See also the criticism by Thomas Nagel in “Libertarianism Without Foundations” in Jeffrey Paul, ed., Reading Nozick (Rowman and Littlefield, 1981).
- 7 I don’t mean to suggest that Estlund includes unlimited freedom of choice of occupation within the prerogative.
- 8What happens if someone shifts to a less-preferred job that leaves him fairly satisfied, but less satisfied than some of those who are not required to shift jobs? I wish that Cohen had discussed this case.
- 9 I think that this objection is based on a misunderstanding of Nozick’s argument. Nozick’s point was that preserving a pattern often requires very strong restrictions on what people can do, not that supporters of a pattern are logically required to accept changes in it. I am indebted here to conversations with Nozick.
- 10One libertarian response to Rawls is to claim that people do deserve, in certain circumstances, to benefit from their superior talents. See on this David Schmidtz, Elements of Justice (Cambridge, 2006) and my review in The Mises Review Spring 2006.
