The Idea of Justice

by Amartya Sen

  305

  t h e i d e a o f j u s t i c e that enslaved people remain slaves even if their choices never conflict with the will of their master.

  There can be little doubt that the republican concept of freedom is important and captures one aspect of our intuitions about the claims of freedom. Where I would disagree is in the claim that the republican idea of freedom can replace the perspective of freedom as capability.

  There is room for both ideas, which need not be a source of tension at all, unless we insist on a single-focus idea of freedom, against which I have already argued.

  Consider three alternative cases related to a disabled person A who cannot do certain things by herself, without help.*

  Case 1: Person A is not helped by others, and she is thus unable to go out of her house.

  Case 2: Person A is always helped by helpers arranged either by a social security system in operation in her locality (or, alternatively, by volunteers with goodwill), and she is, as a result, fully able to go out of her house whenever she wants and to move around freely.

  Case 3: Person A has well-remunerated servants who obey – and have to obey – her command, and she is fully able to go out of her house whenever she wants and to move around freely.

  In terms of ‘capability’, as defined in the capability approach, Cases 2 and 3 are largely similar as far as the disabled person is concerned (this refers to the freedom of the disabled person only, and not to that of the servants, which would raise other issues), and they both contrast in the same way with Case 1, in which she does lack the capability in question. There is clearly something of substance in this contrast between being able to do something and not being able to do it, since it does matter what a person is actually capable of doing.

  The republican approach would, however, see the disabled person as being unfree in both Cases 1 and 2: in Case 1 because she cannot do what she wants to do (that is, come out of the house), and in

  * This illustration is adapted from my ‘Reply’ to Pettit’s essay, along with two other interesting and important contributions respectively from Elizabeth Anderson and Thomas Scanlon in Economics and Philosophy, 17 (2001).

  306

  e q u a l i t y a n d l i b e r t y Case 2 because her ability to do what she wants to do (in this case, to go out of the house) is ‘context-dependent’, depending as it does on the existence of a particular social security system, and it may even be ‘favour-dependent’ on the goodwill and generosity of others (to invoke distinctions invoked by Pettit). It can certainly be said that A is free in a way in Case 3 that she is not in Case 2. The republican approach captures this difference, and has a particular discriminating power that the capability approach lacks.

  However, all this does not remove the importance of the distinction on which the capability approach focuses: can the person actually do these things or not? There is an extremely important contrast between Case 1, on the one hand, and Cases 2 and 3, on the other. In the former case, A lacks the capability to come out of her house and is unfree in this respect, whereas in Cases 2 and 3 she has the capability and freedom to go out of her house whenever she wants. It is this distinction that the capability approach tries to capture, and it is a momentous distinction to acknowledge in general and to be recognized in the making of public policy in particular. Placing Cases 1 and 2 in the same box of non-freedom, without further distinction, would steer us towards the view that instituting social security provisions, or having a supportive society, cannot make any difference to anyone’s freedom, when dealing with disabilities or handicaps. For a theory of justice that would be a huge lacuna.

  Indeed, there are many exercises in which it is particularly important to know whether a person is really able to do the things that she would choose to do and has reason to choose to do. For example, individual parents may not be able to set up their own school for their children, and may be dependent on public policy, which may be determined by a variety of influences, such as national or local politics.

  And yet the establishment of a school in that region can be sensibly seen as increasing the freedom of the children to be educated. To deny this would seem to miss out an important way of thinking about freedom that has both reason and practice behind it. This case contrasts sharply with a case in which there are no schools in the region and no freedom to receive school education. The distinction between the two cases is important enough and on this the capability approach concentrates, even though in neither case can the person bring about 307

  t h e i d e a o f j u s t i c e her own schooling independently of the support of the state or support from others. We live in a world in which being completely independent of the help and goodwill of others may be particularly difficult to achieve, and sometimes may not even be the most important thing to achieve.

  The tension between capability and republicanism as approaches to freedom arises if and only if we have room for ‘at most one idea’.

  It arises when looking for a single-focus understanding of freedom, despite the fact that freedom as an idea has irreducibly multiple elements.* The republican view of freedom, I would argue, adds to the capability-based perspective, rather than demolishing the relevance of that perspective as an approach to freedom.

  The plurality does not, however, end there. There is also a distinction that concentrates on whether a person’s failure of capability is due to the interference of others – an issue that was already raised.

  We concentrate here not on the power to intervene effectively whether or not that power is exercised – that would be a republican concern

  – but the actual use of such interference. The distinction between potential and actual interference is significant, and one that evidently did engage that pioneer of modern political thought – Thomas Hobbes. Even though Hobbes might have had some sympathy for the

  ‘republican’ or ‘neo-Roman’ point of view in his early thought (an approach that was quite current in the political thinking in Britain at that time), Quentin Skinner shows convincingly that Hobbes’s understanding of freedom crystallized on a non-republican view, focusing on whether there is actual interference or not.† The focus on

  * Philip Pettit is clearly tempted by the unifocal view – what he sees as a comprehensive understanding of freedom: ‘the position defended here will help to bear out the case for thinking about freedom comprehensively, not just in a compartmentalized way’

  ( A Theory of Freedom, 2001, p. 179). Pettit is talking here about a different kind of duality, involving such issues as free will, but his motivational remark would seem to apply also to the particular internal contrast – what he may see as ‘compartmentaliz-ation’ – under discussion, involving the republican and the capability-based approaches to freedom.

  † See Quentin Skinner, Hobbes and Republican Liberty (Cambridge: Cambridge University Press, 2008). Even in his early work, Elements of Law (1640), Hobbes showed some hostility to the thesis that there would be some violation of liberty even in the absence of actual interference, but he did not develop an alternative theory in that 308

  e q u a l i t y a n d l i b e r t y the interference of others as the central feature of the negation of liberty is, thus, a Hobbesian idea.

  There is no embarrassment in accommodating several distinct features within the idea of freedom, focusing respectively on capability, lack of dependence and lack of interference.12 Those who want one canonical understanding of the ‘true’ nature of freedom may underestimate the very different ways in which ideas of freedom and non-freedom can enter our perception, assessment and evaluation. As William Cowper puts it, ‘Freedom has a thousand charms to show/

  That slaves, howe’er contented, never know.’ When it comes to distinct concepts, a thousand may be difficult to manage, but there should be no great difficulty in being able to see several different aspects of freedom as being complementary rather than competitive. A theory of justice can pay attention to each. Indeed, the approach to justice prese
nted in this work makes room for pervasive plurality as a constituent feature of the assessment of justice. The plurality of aspects of freedom fits right into that capacious framework.

  t h e i m p o s s i b i l i t y o f t h e

  p a r e t i a n l i b e r a l

  The capability to influence an outcome in the direction one wants can be, as has already been argued, an important part of freedom. The understanding of an outcome can, when relevant, take substantive note of the process through which a final state – a culmination outcome – comes about (the process-inclusive view of an outcome is called a ‘comprehensive’ outcome). In social choice theory, which is concerned with social states (as discussed in Chapter 4), the result-oriented view of freedom has particularly received attention. And particular book. But his rejection of that republican perspective came through loud and clear by the time he wrote Leviathan (1651), which is armed with an alternative approach as well, in which actual interference is the central issue. Indeed, as Skinner argues, ‘Hobbes is the most formidable enemy of the republican theory of liberty, and his attempts to discredit it constitute an epoch-making moment in the history of Anglophone political thought’ ( Hobbes and Republican Liberty, p. xiv).

  309

  t h e i d e a o f j u s t i c e many of the issues of liberty and freedom that have been discussed in social choice theory have been within this framework.

  A result that has generated something of a literature of its own is a rather simple theorem called ‘the impossibility of the Paretian liberal’.

  This takes the form of showing that if people can have any preferences they like, then the formal demands of Pareto optimality may conflict with some minimal demands of personal liberty.13 I shall not try to show how this impossibility theorem works, but illustrate it instead with an example that has been much discussed. There is an allegedly pornographic book and two possible readers.* The person called Prude hates the book, will not like to read it, but would suffer even more from its being read by the other person – called Lewd – who loves the book (Prude is particularly bothered that Lewd may be chuckling over the book). Lewd, on the other hand, would love to read the book, but would prefer even more that Prude reads it (agonizingly, Lewd hopes).

  So, ‘what to do?’ as we say in the subcontinent. There is here no liberty-supported case for no one reading the book, since Lewd clearly wants to read it, and it is none of Prude’s business to interfere in that decision. Nor is there a liberty-based case for Prude reading the book, since he clearly does not want to read the book, and it is none of Lewd’s business to weigh in in that choice in which he is not directly involved. The only remaining alternative is for Lewd to read the book, which would of course be exactly what would happen if each person were to decide what to read (or not read). However, in their preferences, as described, both Prude and Lewd prefer Prude’s reading the book over Lewd’s reading it, so that self-chosen alternative seems to go against the Pareto principle judged in terms of what the two individuals like, since both like Lewd’s reading it less than Prude’s reading it. But the other two alternatives violated minimal demands of liberty, so nothing can be chosen that satisfies the specified demands of social choice, since each available alternative is worse than some

  * In the early, innocent days of the 1960s I fear I was naive enough to choose as an example D. H. Lawrence’s Lady Chatterley’s Lover. I was influenced by the fact that Penguin Books had just before that time fought and won a case in the British courts to be allowed to publish precisely this book.

  310

  e q u a l i t y a n d l i b e r t y other alternative. Hence the impossibility of simultaneously satisfying both the principles.

  This impossibility result, like other impossibility theorems in social choice theory, is meant to be the beginning of a discussion about how the choice problem is to be tackled – not the end of any possible argument. And it certainly has served that purpose. Some have used the impossibility result to argue that, for liberty to be effective, people should respect other people’s liberty to make their own choices rather than paying more attention to the choices of other people in their personal lives than they do to their own personal lives (as is the case with both Prude and Lewd here).14 Others have used the mathematical result to argue that even the Pareto principle, allegedly sacred in traditional welfare economics, may have to be violated sometimes.15

  The case for this lies in the fact that the individual preferences here are narrowly other-regarding, and their status is compromised by the recognition that, as John Stuart Mill put it, ‘there is no parity between the feeling of a person for his opinion, and the feeling of another who is offended at his holding it’.* Still others have argued for making the right to liberty to be conditional on the person respecting the liberty of others in his own personal preferences.16

  There have been other proposed solutions as well. One that has been much discussed can be called ‘solution by collusion’. This is the suggestion that the problem is resolved if the parties involved have a Pareto-improving contract, whereby Prude reads the book to prevent Lewd from reading it.† How much of a solution is this?17

  There is, first, a very general methodological point. A Pareto-improving contract is always a possibility in any Pareto- inefficient situation. To say that does nothing to undermine the problem faced in a world in which individual choices take one to a Pareto-inefficient outcome. Note also a general problem with this way of seeking a

  * Questioning the unconditional acceptance of the priority of the Pareto principle was, I must confess, my principal motivation in presenting this result. See also Jonathan Barnes, ‘Freedom, Rationality and Paradox’, Canadian Journal of Philosophy, 10

  (1980); Peter Bernholz, ‘A General Social Dilemma: Profitable Exchange and Intransitive Group Preferences’, Zeitschrift fu¨r Nationalo¨konomie, 40 (1980).

  † A great many commentators have sought this way ‘out’. One of the most recent is G. A. Cohen, Rescuing Justice and Equality (2008), pp. 187–188.

  311

  t h e i d e a o f j u s t i c e

  ‘solution’. A Pareto-improving contract may not be viable, since the incentive for breaking it can be strong.18 This may not be the principal argument against seeing a solution to the problem through collusion (the main argument against this alleged resolution may relate to the reasoning behind the two parties offering and accepting such a contract), but it is one argument to be considered before taking up the more serious issues. We have to consider the credibility of such a contract, and the difficulty of ensuring its compliance (i.e. how to make sure that Prude actually reads the book rather than just pretending to).

  This is no mean problem but, perhaps more importantly, attempts at enforcing such contracts (for example, the policeman ensuring that Prude is actually engaged in reading the book and is not just turning the pages) in the name of liberty can powerfully – and chillingly –

  endanger liberty itself. Those who seek a liberal solution that would demand such intrusive policing in personal lives must have a rather odd idea of what a liberal society should be like.

  Of course, such enforcement would not be necessary if people were to conform voluntarily to the agreement. If individual preference is taken to determine choice (no variations at all on any other grounds

  – those discussed in Chapter 8), then this possibility is not open, since Prude will not read the book, given that choice (that is, in the absence of intrusive policing). If, on the other hand, preferences are taken to represent people’s desires (not necessarily their choices), which is perhaps more sensible in this case, then it is, of course, possible to argue that even though Prude and Lewd both desire to act in a way contrary to the contract, they need not actually act that way, since they have signed a contract and thus have reason to resist being slaves to their desires. But if that question is raised, and actions that go contrary to felt desires are permitted, then we have to ask a prior –

  and more basic – question for
this ‘solution by collusion’: why should we assume that Prude and Lewd would choose to have such a contract in the first place (even though they may desire the corresponding outcome – seen simply as a ‘culmination outcome’)?

  It is not at all obvious why Prude and Lewd must go for a peculiarly

  ‘other-regarding’ social contract by which Prude agrees to read the book he hates in order to make eager-to-read Lewd refrain from reading it, and Lewd in turn agrees to forgo reading a book he would 312

  e q u a l i t y a n d l i b e r t y love to read in order to make reluctant Prude read it instead. If people attach some importance to minding their own business rather than just following their desires, then that odd contract need not in fact materialize (cf. ‘I think Ann would be much happier if she divorced Jack, and I wish she would do this – and so let me jump in and tell her to do this’). The good liberal practice of reading what one likes and letting others read what they like can perhaps survive the apparent temptations of having this remarkable contract. It is hard to see solution by collusion as a solution at all.

  For some inexplicable reason, some authors seem to believe that the issue in question is whether rights are ‘alienable’ (in the sense of people being permitted to trade away particular rights) and whether the persons involved should be allowed to have such a contract.19 I see no reason why rights of this kind should not in general be taken to be open to contracting and exchange through mutual agreement.

  Indeed, there can be little doubt that people do not, in general, need anyone else’s (or ‘the society’s’) permission to have such a contract.

  But they do need a reason, which is where the rub is. To offer as a reason, as some have done, the fact that such a contract would be the only way of getting – and sustaining – a Pareto-efficient outcome is to beg the question, since one of the motivations for discussing the impossibility result is precisely to question and assess the priority of Pareto efficiency.