Equality a moral realistic view
c 1998 Copyright S. F. Hartkamp Lutkenieuwstraatje 34-36 9712 AZ Groningen email:
[email protected]
All rights reserved. Except for the quotation of short passages for the purpose of criticism and review, no part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of the author.
CIP-GEGEVENS KONINKLIJKE BIBLIOTHEEK, DEN HAAG Hartkamp, Steven Frederik Equality; a realistic view. Towards a simple measure of inequality. Thesis Groningen. - Includes bibliography - Includes index - Includes Dutch summary ISBN 90-9012254-0 NUGI 611 Cover: Rum, Scotland, 1996 Photography by G.C. Verduyn Printed by Krips b.v., Meppel
RIJKSUNIVERSITEIT GRONINGEN Equality; a moral realistic view
(Towards a simple measure of inequality ) Proefschrift ter verkrijging van het doctoraat in de Wijsbegeerte aan de Rijksuniversiteit Groningen op gezag van de Rector Magni cus, dr. D.F.J. Bosscher, in het openbaar te verdedigen op donderdag 15 april 1999 om 14.15 uur door: Steven Frederik Hartkamp geboren op 4 april 1957 te Meppel
Promotores: Prof. dr. J.W. de Beus Prof. dr. G.E. Lock
Contents Acknowledgements List of symbols
: : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : :
1 Introduction
1.1 Purpose of the study 1.2 The traditional background assumptions of the modern political ideal of equality. 1.3 Overview
: : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : :
I A new framework for equality
1
1
3 11
15
2 Moral universalism
2.1 Introduction 2.2 Why moral universalism? 2.2.1 The use of moral universalism 2.2.2 Arguments for moral universalism, and moral particularism 2.3 Moral realism 2.3.1 Davidson's radical interpretation 2.3.2 The relation between moral beliefs and actions. 2.3.3 Why follow particular moral reasons rather than prudential reasons? 2.3.4 Justi cation of moral beliefs 2.4 Moral realism, its particularism and equality 2.5 Summary 2.6 Appendix 1
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
v
ix xi
17
17 20 20 31 36 36 38
42 44 53 56 59
3 Volitional individualism
3.1 Introduction 3.2 The problem of interpersonal comparisons 3.2.1 Necessity of interpersonal comparisons 3.2.2 Impossibility of interpersonal comparisons 3.3 Necessity challenged 3.3.1 External e ects as reasons for redistribution 3.3.2 Non-envy analysis 3.4 Impossibility rejected; realistic individualism 3.5 Summary 3.6 Appendix 1 3.7 Appendix 2 3.8 Appendix 3 3.9 Appendix 4
63
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
4 Moral value monism
63 65 65 67 70 70 74 100 105 108 110 111 112
115
4.1 Introduction 115 4.2 Monism versus pluralism 116 4.2.1 Reasons for value monism 116 4.2.2 Reasons for moral pluralism 126 4.3 Equalisanda instead of one equalisandum 135 4.3.1 Subjective welfare as equalisandum 136 4.3.2 O ensive and expensive tastes excluded by social ideals137 4.3.3 Objective welfare as equalisandum 139 4.3.4 Resources as equalisandum 141 4.3.5 Access to advantages as equalisandum 143 4.3.6 Equalisanda 146 4.4 Summary 151 : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : :
5 The new framework and its criticisms 5.1 5.2 5.3 5.4 5.5 5.6
Introduction Conservatism Imperialism Quasi-realism Disagreement Summary
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : :
153
153 156 163 167 173 179
II Equality
181
6 The meaning of the ideal of equality 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8
Introduction Arguments an ideal of equality has to answer The levelling down objection Person a ectingness The meaning of equality The reference The other views revisited Summary
183
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : :
7 Measures of inequality
7.1 Introduction 7.2 Lorenz dominance 7.3 Incompleteness 7.3.1 Extensions of Lorenz dominance 7.3.2 Acceptance of incompleteness and its explanation 7.4 Properties of Lorenz dominance revisited 7.5 Properties of a measure of inequality 7.6 Summary 7.7 Appendix 1
213
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : :
: : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
8 A simple measure of inequality
8.1 Introduction 8.2 The construction of a simple measure 8.2.1 The measure of equalisanda 8.2.2 Independence 8.2.3 Euclidean distance as basis 8.3 The arguments for complexity revisited 8.4 Speci city and aggregation 8.5 Summary 8.6 Appendix 1
183 185 188 192 197 201 210 211 213 215 220 221 228 233 237 239 242
243
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
243 245 245 247 254 257 261 263 266
III A practical consequence
269
9 Equality of health
271
9.1 Introduction 9.2 Macro level decisions 9.2.1 Socio-economic related di erences in health 9.2.2 Choices in health care policies 9.3 Micro level decisions 9.3.1 Maximising qalys, hyes and lyars 9.3.2 Are qalys, hyes and lyars proper equalisanda? 9.4 Equality of health in medical practice 9.4.1 What if there is no cure? 9.4.2 The general practitioner in an egalitarian perspective 9.5 Summary : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : :
: : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : :
Bibliography Index Samenvatting
: : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
271 272 273 279 280 282 285 290 292 296 299 303 319 I
Acknowledgements Writing a thesis seems to be possible while accepting solipsism, however the fact is that it owes a lot to others and it shows that solipsism is false. This thesis is the result of a research project which was initiated by the Department of Political Science of the Catholic University of Nijmegen on the measurement of inequality. I thank the members of the Department of Political Science for their hospitality. I am indebted to all who made this thesis possible. In particular I am indebted to Trudy van Asperen, by whom my interest in moral philosophy was awakened by her way of teaching moral philosophy. It is sad that although I could show her the outline of the thesis, I could not show her the result because she unfortunately died in 1993. I owe much to my two promotores, Grahame Lock who introduced me to the project and who let me work in freedom and Jos de Beus who helped me to look critically at my own ideas in a way that stimulated me to explain them in a more clear way. They did their best, the remaining mistakes in reasoning which will be discovered, are mine. Also I owe much to Wim Fievez with whom I followed the rst courses in philosophy, he convinced me that Davidson's approach to language and interpretation could not be circumvented. In writing I was assisted by Frans Wiersma, who also accompanied my rst courses in philosophy, he explained to me which rules for clear writing I have shown to have forgotten. In writing this thesis I realised more and more that it was inspired by the people of Schiermonnikoog who trusted me to share their experiences on important aspects of their life, while I was working on their island as a general practitioner. Without them this thesis would have been di erent. Finally, I thank my family and friends, especially my mother who let her son follow the courses in philosophy and supported it, and Ineke who bore the underestimated costs of enduring my involvation in the project and the restless and desperate attempts to deal with the unpredictable results of a laptop with a full harddisc with system failures. Groningen, April 1998 ix
x
ACKNOWLEDGEMENTS
List of symbols The following list contains the most important symbols which are used in this study. a, ak ,b: goods, distribuenda aki : a distribuendum of type ak belonging to person i ~a, ~ak : an allocation (a1; a2; : : :; ai; : : :; an), or (ak1 ; ak2 ; : : :; aki ; : : :; akn), of the distribuendum a, or ak , among persons 1; : : :; n A, B : allocations of goods ak ; al etc., or bk ; bl etc., among persons 1; : : :; n P A: summation of the elements of the rows of the matrix matrix Pni=1 aki P A P B: all the values of the summation of the elements of the rows of A are equal or greater than the values of the summation of the elements of the rows of B : descriptions of actions C : conditions of situations i; j : persons I(): index representing the seriousness of inequality l: the amount of leisure minf ; g: the minimum of the set f ; g M(): monotone increasing transformation function ~p: price vector, or probability vector1 1
It will be clear from the context what is meant.
xi
xii
LIST OF SYMBOLS
ri : the amount of time spent by person i to work Ri (aj ): the amount of time needed for i to produce aj : permutation of persons ~u: (1; 1; : : :; 1) wi (): well-being of person i dw (a) : derivative of wi with respect to a, sometimes represented by w () da W(): social welfare function x, y: goods, equalisanda2 ~x: an allocation (x1 ; x2; : : :; xi; : : :; xn) of equalisandum x among persons 1; : : :; n xref : the allocation in which all persons are equally well o with respect to the equalisandum x due to a (re)distribution of the available distribuenda : better `all things considered' I : worse regarding inequality LD : better regarding the Lorenzcurve, `the Lorenzcurve lies above' 8: for all 9: there exists ^: and _: or/and j=: is satis ed in those models in which the premises are satis ed : material equivalence i
0
2 sometimes x and y represent a variable in logical formulas, for example in the appendix 1 of chapter 2. The context will make clear what is meant.
Chapter 1 Introduction
1.1 Purpose of the study The purpose of this study is the articulation of a political ideal of equality within a framework of moral realism, moral particularism, realistic individualism and moral value pluralism. Traditionally equality as a political ideal is formulated in a framework that is characterised by moral universalism, volitional individualism and moral value monism. The proposals of for example the economists Tinbergen, Varian, Roemer, and the political philosophers Rawls and Dworkin, aiming at a general `overall' evaluation of allocations regarding inequality, were developed within this traditional framework. The ideal of equality did not receive much attention within moral realism, which is the view according to which moral values are considered to belong to the objective external world, beyond the desires of individuals, that is inspired by the work of Wittgenstein, Davidson, McDowell, Hurley, and Dancy. It can even be said that a proper formulation of equality considered as a moral and political ideal that is valuable apart from its political and economic consequences is lacking.1 This lack is remarkable because, as is explained below, the traditional framework blocks a proper development of an ideal of equality. This study is an attempt to correct for this lack. The aim of this study is restricted to the articulation of an ideal of equality apart from its consequences or causes. The related questions: `Who is responsible for it?' and `How to arrive at a more egalitarian distribution?' are not discussed in this study. To be clear it is not suggested that these are 1 Hurley argues in favour of equality because political decision making is less biased if there is equality [Hurley, 1989] [Hurley, 1993].
1
2
CHAPTER 1. INTRODUCTION
not important, they are.2 But he goal in this thesis is to provide an analysis of the ideal of equality and to make clear what we mean when we invoke the ideal of equality in the evaluation of political decisions. It concerns the basic issues of moral and political arguments as they are discussed in for example, Nagel's Equality and Partiality rather than those issues that are discussed in Barry's Justice as Impartiality, in which the practical political problems concerning the disagreement on political and moral issues, receive most attention and in which these basic issues are taken for granted. A political ideal of equality has to indicate how distributions should be evaluated regarding this ideal. In the debate on the ideal of equality, it is argued that the ideal is complex because it is not always clear how two distributions should be evaluated and compared to each other regarding inequality [Sen, 1973] [Temkin, 1993]. It cannot always be determined which distribution is more or less unequal, or in other words, which distribution is more or less bad regarding equality. The ordering regarding equality is held to be incomplete. This incompleteness makes the articulation of an ideal of equality rather dicult because it shows that the ideal is not appropriate for all situations. This demands an explanation that comes from the ideal of equality itself. But whatever the restrictions on the applicability of the ideal of equality, a proper articulation of the ideal of equality will have to indicate an answer to the question: `What is the ordering of a distribution regarding the ideal of equality?', in other words `What is the measure of inequality?' A measure of inequality is meant to re ect the moral badness regarding inequality. Consequently, simply turning to a conventional or a stipulative measure, for example the deviation from the mean, is not sucient. A measure concerns the moral badness and not just a mathematical value. The reason why inequality is morally bad has to be looked at. Therefore, the question: `Why is equality desirable?' has to be answered. An answer to the question concerning the desirability of equality supposes that another question has been dealt with, namely the question concerning the aspect with respect to which a distribution should be equal, i.e. the question on the equalisandum. Without a determinate equalisandum the content of the ideal of equality remains obscure. Hence the question: `What is the equalisandum?' has to be answered too. So, in this thesis, in which the goal is the development of a formulation of a political ideal of equality, I have to answer the following questions: 1. What is the proper equalisandum? 2 For the time being, the author sympathises with the view of Carens on the issue of how to reach and enhance equality [Carens, 1981].
1.2. THREE TRADITIONAL BACKGROUND ASSUMPTIONS
3
2. Why is equality desirable? 3. What is the ordering according to the ideal of equality? within a moral realistic framework, with moral particularism, realistic individualism and moral value pluralism. In the next section of this introduction, the traditional background assumptions of the modern political ideal of equality are presented. Subsequently, I show that the traditional framework that is characterised by moral universalism, volitional individualism and moral value monism leads to serious diculties in the development of an ideal of equality. Abandoning this traditional framework and turning to moral realism with its moral particularism, realistic individualism and moral value pluralism, can open up the possibility of the development of an ideal of equality with a simple measure that represents a complete ordering of the moral badness of inequality in a distribution problem. This introduction ends with a short overview of the chapters in this thesis.
1.2 The traditional background assumptions of the modern political ideal of equality. The modern ideal of equality emerged as an element of the famous triad in France in 1789: `Liberte, Egalite, Fraternite'. Although supposing the ideals of the triad are congenial to each other might be an idiocy as Hare suggests[Hare, 1989, p. 170], these ideals were nevertheless developed as a triad and not just separately. The ideas that led to liberty were also the ones that led to equality and brotherhood. The ideals have the same background. This common background can help us in understanding what is meant by equality as a political ideal. It reminds us that equality is one ideal among others and that, although equality di ers from fraternity and liberty, it is linked up with them. Equality and liberty just as fraternity arose as ideals in an era in which the idea of a strict hierarchical society in which everything had its own divinely ordered place in one common world was declining. Individualism developed in early Christian thought and led to the idea of a contract as the basis of society. In Christian thought, the individual soul receives its eternal value from its lial relation with God and this relation is the basis for the brotherhood of men; Christians come together in Christ of whom they are its members [Dumont, 1983, p. 43 .]. `There is no such thing as Jew and Greek, slave and freeman, male and female, for you are all one person in Christ Jesus'
4
CHAPTER 1. INTRODUCTION
(1 Corinthians 12: 13). This idea of spiritual individualism transformed into the triad via a process of change from religious a airs belonging to a region external to this world into a airs of a region enclosing this world. Before Constantine in the fourth century AD became a Christian, the Church was essentially looking away from this evil world in which Christians were persecuted. Their orientation was directed to eternal a airs. But with a Christian emperor the Church had to review her relation with worldly a airs and she did. In about 500 AD. Pope Gelasius I proposed his theory of the two swords. The spiritual interests and eternal salvation, which were held to belong to the most important matters, fell within the domain of the Church and the less important temporal a airs fell under the jurisdiction of the king, Christian emperors need bishops for the sake of eternal life, and bishops make use of imperial regulations to the course of temporal a airs. [Sabine & Thorson, 1973, Gelasius Tractatus IV,11 cited on p. 189] This dual hierarchy in which the Church accepted the king as authority in the less important a airs changed in 754 AD, when Pope Etienne II traversed the Alps to meet Pepine to get his protection against the Lombards and to o er him the legitimisation of his power as a king, which he apparently needed. This resulted in 800 AD in the coronation of Charlemagne as emperor by Pope Leon III. Now, the Church had supremacy in worldly and extra worldly a airs. As a consequence, the individualism until now only seen in a spiritual way was injected into worldly a airs under the supremacy of the Church. Thereafter the Church had to withdraw from the political arena but individualism remained as is seen in for example the nominalism of William of Ockham with the consequential idea that society is not existent beyond the individuals belonging to it. The reformation armed individualism in this world. This arising modern individualism resulted in the idea of a contract as the basis of society [Rosanvallon, 1989]. In this era, contract theories of society and the state developed together with the idea of the market. In a contract individuals meet and settle something. It presumes individuals who can have di erent wants, desires and aims, and who can come to an agreement, which is expressed in a contract. The individuals are di erent in their desires but similar in being a contractor and similar in having interests of their own. Of course, the contract was not a totally new concept. It was used in the classical world too, but its extension was new and its meaning changed. A contract was now seen as a general universal method of regulating human relations and that was new. It became a way of regulating the passions
1.2. THREE TRADITIONAL BACKGROUND ASSUMPTIONS
5
of individuals. Before, contracts were merely applicable in particular situations for particular persons or institutions, and they lacked the universal constitutive function of regulating human relations.3 The very idea of a contract as the basis of society and the state presumes moral universalism, by which is meant that rules that specify what kind of actions or situations speci ed in publicly observable terms are right or wrong, good or bad, are necessary for moral judgements and that those rules form the basis for moral judgements. In contracts it is stated what kind of actions should be done or should not be done. It should be publicly possible to recognise whether the contract is broken or not. So, according a contract, some kind of actions described in publicly observable terms are wrong or right. Actions with such and such properties are judged as right or as wrong because of the rules speci ed in the contract. The universality of a basic contract and the universality of the market is accompanied by a homogenisation of persons and a homogensation of goods [Dumont, 1983] [Rosanvallon, 1989]. Persons become essentially individuals with di erent wants and desires, which are of an unrestricted diversity by their nature, in this they become equal. Their desires and wants become the basis of the human relationships. Goods become essentially bearers of an abstract value through which everything can be expressed and compared. They can be exchanged and can be substituted for each other. If you do that, I will do this. If you give up your right to X, I will give up my right to Y. Goods, services and rights too, become the subject of bargaining and contracts. They become to be seen as bearers of one basic value. The ideals of liberty, equality and brotherhood, developed within a particular conception of the world with the following elements: moral universalism, rules are necessary for moral judgements and these rules form the basis of the moral judgements, volitional individualism, people are seen essentially as persons having individual wants, desires and passions of unrestricted diversity; these determine which situations are better or worse, moral value monism, there is one value, goods are essentially seen as bearers of this value. Before proceeding, I have to make some terminological remarks, in order to avoid misunderstandings. First, with moral universalism used in this thesis is meant the view that rules are necessary for moral judgements and form the basis for those judgements. It is not referring to the idea that moral judgements are valid for all people and not merely valid or true 3
See also [Gough, 1957, p. 14].
CHAPTER 1. INTRODUCTION
6
relative to one's own community. However, the idea of non-relative validity is accepted in this thesis but it is captured by moral realism. The issue of relativism is discussed extensively in chapter 2 and 5. Second, moral value pluralism as it is used in this thesis indicates the view that there is just more than one value. It holds that not all what is valuable is covered by one value. It is not pointing to a plurality of ideas about what is valuable in a plural society. This plurality is not denied to exist, but is captured in this thesis by the term disagreement. Let me return, after these terminological remarks, to the discussion of the traditional framework of equality as a political ideal. From an historical point of view the modern ideal of equality assumes via the concept of a contract: moral universalism, volitional individualism, and moral value monism. But also seen from a systematic point of view, one could argue that the traditional background assumptions are related to equality. Without individualism there could be no relevant questions about distributions. Issues concerning distributions assume individuals with their own interests, desires, wants etc, otherwise these issues could not arise in a relevant way.4 Furthermore, the ideal of equality assumes moral universalism. If equality has some meaning at all it has to subscribe at least to the principle that similar persons in similar situations should perform similar actions. Finally, the idea of equality also assumes monism because it assumes there is one aspect according to which persons should be treated equally and this aspect is assumed to be the one and only value. On the other hand, one could argue too that these three elements together lead to the formation of equality as an ideal. Individualism and moral universalism, which make the idea of an agreement expressed by a contract intelligible, assume that all persons are alike as wanting and desiring individuals trying to reach a consensus. In order to arrive at such a consensus there has to be some equality in the distribution of goods. Without it there is hardly possible a consensus; one should be able to justify one's actions, one's demands etc. to others on grounds that are regocnised by others and cannot be reasonably rejected. This is held to be one of the most important reasons for the ideal of equality [Sen, 1992, p. 16-19] [O'Neill, 1985]. And because of this, the one and only value, whatever it is, has to be distributed equally. Moral universalism together with volitional individualism and moral value monism are the main background assumptions with which the ideal of equality is closely connected. However, moral universalism lost its attractiveness [Dancy, 1993]. That moral universalism is not very satisfactory is shown by phenomena such as moral con icts, and more important 4
See Chapter 4 p. 129 on the consequences of individualism for value monism.
1.2. THREE TRADITIONAL BACKGROUND ASSUMPTIONS
7
by the interpretation of what following a rule consists of as it is analysed by Wittgenstein in his Philosophical Investigations. The conception of persons as having desires and wants of an unrestricted diversity is dubious too [Nagel, 1978] [Dancy, 1993] [Davidson, 1980]. We cannot ascribe every desire to persons. Our ascription is bounded by what is objectively desirable. It is argued also that monism with respect to values is dubious or even better false [Dancy, 1993] [Stocker, 1990]. Here too moral con icts are pointed at to show its falseness. Should we conclude that because moral universalism, volitional individualism, and moral value monism are false, the ideal of equality has to be removed from the set of political ideals? Its historical grounds are mistakes, so this seems to be a natural conclusion. On the other hand, these three assumptions are also responsible for serious problems concerning the ideal of equality. By accepting moral universalism the development of a proper measure representing judgements according to the ideal of equality, is blocked. With a measure is meant an ordering that represents the moral seriousness of inequality of a distribution. The more serious the inequality, the more it should count in political decision making. But how much a particular inequality should count is also dependent on other ideals. For example Sen states as one of the reasons for a quasi-ordering of the seriousness of inequality instead of a complete ordering: .., even as normative indicators the inequality measures are best viewed as 'non-compulsive' judgements recommending something but not with absolutely compelling force. This has implications in terms of the treatment of inequality rankings as prima facie arguments and permitting situation-speci c considerations to be brought into the evaluation if such supplementation is needed. [Sen, 1973, p. 75] This dependency on other aspects leads to a quasi-ordering because of moral universalism. That this dependency does lead to a quasi-ordering is due to moral universalism can be seen as follows. The moral universalism I referred to holds that rules are necessary for moral judgements and form the basis for these judgements. Rules are of the form: Actions of kind or distributions of kind ~a are . The is a moral evaluative quali cation of the actions or distributions such as `right' or `wrong', `good' and `bad', or in a situation of choice between alternatives it can mean: `is to be preferred'. The reason why a
CHAPTER 1. INTRODUCTION
8
particular action or distribution has the moral evaluative quali cation for example right or better, is because of such a rule. In such rules the actions and distributions are described in non-moral terms because it should be possible to check whether the rule was followed or broken. The rules mean that if in evaluating two distributions from an egalitarian point of view, one is judged to be better or worse than the other; it is judged apart from all other aspects not speci ed in ~a or ~b, better or worse regarding inequality in all other situations.5 In using a formula one could state ~a I ~b , 8C((C;~a) I (C; ~b)) in which (C;~a); (C; ~b) are distributions under conditions C, I means: is worse regarding inequality, and the right part means that in the situations speci ed by C, whatever C is, holding all other aspects than those speci ed by ~a and ~b constant, distribution ~a is worse than distribution ~b regarding inequality. This will lead to an incomplete ordering if there is a set of conditions C such that (C;~a) I (C; ~b) and another set C such that (C ;~a) I (C ; ~b). It is not clear how to evaluate the distributions ~a and ~b leading to the incompleteness as Sen states.6 Frankfurt for example describes such a possibility [Frankfurt, 1987]. Suppose there are situations with a distribution of a good in which one person gets 10 units of some good and another 0 units and a distribution in which both get 5 units and suppose the former is worse than the latter from an egalitarian point of view. Suppose further there is a situation such that 10 units are necessary for survival, then it is not clear, even not from an egalitarian point of view, that the former distribution is worse than the latter. Of course, one could argue that, all things considered, the former distribution is better but still worse in one respect, namely regarding inequality, and consequently this would be not an example of di erent evaluations on the distribution. But there is a serious problem with this argument. It is not made clear why in case 10 units are necessary for survival, the distribution in which all receive 5 units is better in even one aspect than the one in which at least one person survives?7 An underlying 0
0
0
Moral universalism as used in this introduction brings with it separability of moral evaluative principles. See also chapter 2 p. 33, and chapter 7 p. 239 and chapter 8 p. 261 on speci city, i.e. separability with respect to equalisanda. Separability is recognised to be an element of moral universalism in [Dancy, 1993, p. 75] [Jackson, 1985] [Kagan, 1988]. 6 Together with the familiar de nitions of f g as f 6 g and f h as f 6 h I I I I and f 6 I h, we would have ~a I ~b, which is not necessarily an equivalence because it is not necessarily transitive. 7 It is discussed more extensively in chapter 6 in the sections on the levelling down objection. Although the discussion there seems to undermine this argument, in fact it 5
1.2. THREE TRADITIONAL BACKGROUND ASSUMPTIONS
9
assumption in the argument showing that the ordering regarding inequality is incomplete, is moral universalism. Volitional individualism also leads to a serious diculty for the ideal of equality, namely the problem of interpersonal comparisons. How should the satisfaction of wants of di erent individuals be compared? Without such an interpersonal comparability there cannot be a proper idea of equality. Without it one cannot state that persons are equally well o because such a comparison is nonsensical [Sen, 1973] [Roemer, 1985]. Finally, with value monism the desirability of the ideal of equality becomes a problem. For instance, is equality the only political ideal? What about ideals of perfection or political rights? These other ideals can con ict with equality, denying this is the idiocy Hare mentioned.8 Furthermore, what should the equalisandum be? The most serious arguments against a particular conception of equality are, as explained by Cohen, those based on the idea of equality itself but with a di erent equalisandum [Cohen, 1989]. For instance, the ideal of equality of wealth in terms of money is criticised because it leads to inequalities in want satisfaction, or welfare, or availability of resources or capabilities [Roemer, 1986] [Dworkin, 1981b] [Sen, 1992]. Which one is the proper equalisandum? Monism would imply equality to be the only relevant political ideal, which seems to be dangerous. Monism also would imply that there is one proper equalisandum, which seems to be counterproductive in the applicability of the ideal of equality, because this all encompassing equalisandum cannot be determined. Now we face a dilemma if moral universalism, volitional individualism and moral value monism are held to be the necessary assumptions for equality as a political ideal. Either moral universalism, volitional individualism and value monism are false, then we should abandon equality because its background assumptions are false, or we should accept these assumptions as right, but then the ideal of equality is damaged because of the problems concerning interpersonal comparisons, the desirability of equality and the equalisandum and nally a less informative quasi-ordering being likely instead of an ordering. This dilemma would be solved if the necessity of the background assumptions could be rejected. One could deny in a distributive way each one of the assumptions, moral universalism, volitional individualism, and moral value monism, while keeping the others and look for the possibilities of the articulation of an ideal of equality. In this thesis, I will follow another path for solving the dilemma. Because all three assumptions are to be rejected and because the assumptions are related to each other, as doesn't because it is based on a particularistic point of view, by which the reference is allowed to vary with the equalisandum. 8 See this chapter p. 3.
10
CHAPTER 1. INTRODUCTION
becomes clear in the next chapters, and because the opposites of the traditional assumptions, moral particularism, realistic individualism and moral value pluralism, can be covered by moral realism, the necessity of the traditional assumptions is challenged by the articulation of an ideal of equality within moral realism. I show that it is possible to develop an ideal of equality within moral realism that is built upon the arguments borrowed from the work of Wittgenstein, Davidson, McDowell, Dancy and Hurley. It is argued that equality is concerned with liberties, i.e. the possibility to enjoy or perform valuable actions without disadvantageous consequences, and that equality is based on a moral way of looking at oneself as one among other persons. The ideals of the triad at the beginning of the modern time, liberty and fraternity, can be recognised to support equality. In this thesis, the three questions 1. What is the proper equalisandum? 2. Why is equality desirable? 3. What is the ordering according to the ideal of equality? are answered as follows. The rst question, the one concerning the equalisandum, is answered by admitting several equalisanda. It is no longer assumed that there is just one equalisandum that is always the most important one. An equalisandum is taken to be that what makes goods valuable to people and consequently that what is worthwhile to be distributed equally. In this answer we can discern moral value pluralism, which is responsible for the plurality of equalisanda, moral particularism, which allows that in di erent situations di erent values can be urgent and moral realism for the explanation of what is meant by being valuable. Because in di erent situations di erent equalisanda can be urgent the meaning of equality and its measure are to be determined apart from a particular equalisandum. The second question, concerning the desirability of equality, is based on a moral way of looking, and seeing oneself as one among others. It is not derived from some set of moral rules. It is argued that inequality is bad because some are worse o than all could have been simultaneously. This answer is possible on the basis of moral realism and realistic individualism. The third question, the one concerning the ordering regarding the moral badness of inequality, is answered by a complete ordering in a distribution problem that can be represented by an index based on a summation over persons of the di erences between how well o a person could have been if the goods were distributed in a way such that all were equally well o and how well a person actually is, in so far the persons are worse o than all
1.3. OVERVIEW
11
could have been regarding a particular equalisandum. This answer is made possible because of moral pluralism and particularism. Moral pluralism and particularism back the idea that in di erent situations di erent equalisanda are important. So, the answer does not aim at an egalitarian general evaluation over all possible allocations based on one `overall' equalisandum. Moral pluralism and particularism enable a separation of the general moral problem of weighing several values from the one concerning the meaning of equality and consequently the measure of inequality. Realistic individualism is re ected by the summation over persons being worse o than all could have been. In the remaining part of this introduction, I present an overview of the chapters indicating how the answers stated above are arrived at.
1.3 Overview This book consists of three parts. In part I, containing the chapters 2,3,4 and 5, a new framework for the articulation of a political ideal of equality is introduced as an alternative to the three traditional background assumptions, moral universalism, volitional individualism and moral value monism. The alternative which is introduced is moral realism. Moral realism with realistic individualism and moral value pluralism will directly point to a plurality of equalisanda. An equalisandum is considered to be worthwhile to strive at and consequently worthwhile to be distributed equally. In this part the rst question: `What is the proper equalisandum?', is dealt with. In chapter 2, the rst of the traditional background assumptions of equality, moral universalism, is discussed. This assumption holds that moral rules are necessary for moral judgements and form the rational basis for moral judgements, including those concerning equality. It is shown that the main reason in favour of moral universalism that is based on the idea that moral universalism is saving moral judgements from the sceptical morass, is not convincing. The use of moral universalism does not solve moral problems and moral con icts as it was supposed to do. With the help of Wittgenstein's ideas exposed in his Philosophical Investigations on `similarity' and `to follow a rule', which are elaborated further by Davidson in his ideas on language and interpretation, it is argued that moral judgements, including those concerning equality, are true or false because the world is as it is and not because they are in accordance with a set of moral rules. Consequently, it is not to be expected that the ideal of equality is derived from a set of moral rules. The ideal of equality is related to a particular moral way of looking at oneself as one person among others. Because of its moral particularism moral realism undermines the argument
12
CHAPTER 1. INTRODUCTION
for the incompleteness of the ordering of distributions regarding the ideal of equality. Moral realism opens up the possibility of a complete ordering in a distribution problem. In chapter 3, the second of the traditional background assumptions, volitional individualism, is discussed. Volitional individualism, the view that situations are evaluated as better or worse on the basis of the preferences of persons which are considered as originating from individual choice, brings with it the problem of interpersonal comparisons of what goods mean to people. It is shown that two ways of circumventing this comparability problem in judgements concerning equality are not satisfactory. Acknowledging external e ects as reasons for redistributions is argued to be intelligible only if interpersonal comparisons are presupposed. It is shown that the non-envy analysis is not satisfactory either without allowing interpersonal comparisons. In discussing Dworkin's non-envy proposal a problem related to responsibility is met. It is a problem concerning the possibility of an evaluation of distributions without taking the causes of distributions into account, which was stated as the aim of this thesis. After moral pluralism is introduced, a solution for this problem is formulated in chapter 4 in discussing the equalisandum. In chapter 3, realistic individualism, being an aspect of the moral realism, is introduced. This realistic individualism in which it is held that the content of desires and wants has to be interpreted on the basis of what is desirable, can account for the possibility of interpersonal comparisons which are necessary for moral judgements concerning equality. It is shown how the problem of interpersonal comparisons is related to moral universalism. In chapter 4, the third background assumption, moral value monism, is dealt with. It is shown that the reasons for accepting moral value monism are not convincing. It is argued that monism cannot account for genuine choice and deliberation. It becomes clear that monism cannot account for an ideal of equality because the distinction between the value of some good and the value of the distribution of that good cannot be made. Moral value pluralism is argued to be a better alternative and leads to a solution to the rst question: `What is the proper equalisandum?'. Instead of one equalisandum a plurality of equalisanda is accepted. It is spelled out that equalisanda concern that what makes goods valuable to people, it is the possibility to enjoy or perform valuable actions without su ering disadvantageous consequences, they are called liberties. It is argued that the problem, met in chapter 3, concerning the evaluation of distributions without taking their history into account, is solved by arguing that responsibility should not be incorporated into the equalisandum, contrary to what is suggested by some authors as for example Dworkin, Cohen, and Van Parijs.
1.3. OVERVIEW
13
In each of the chapters 2,3 and 4, some aspects of moral realism based on the Wittgenstein-Davidson approach to language and interpretation, is introduced. The aspects are introduced as alternatives to respectively moral universalism, volitional individualismand moral value monism. After having shown in these chapters what moral realism will mean for an ideal of equality, it is possible and also necessary to evaluate the main objections against moral realism as a framework for a political ideal to equality. In chapter 5 this task is performed. The charge of conservatism, the idea that there are no reasons for change, is rebutted by showing that moral realism can allow for reasons for change. The charge of imperialism, the idea that nothing is recognised to be important for others if it is not recognised to be important for oneself, is met with help of the interpretation suggested by Baker and Hacker of Wittgenstein's ideas on following a rule instead of a sociological interpretation. It is shown that the the charge of imperialism is harmless. The claim that quasi-realism can better account for moral phenomena than moral realism is argued to be not valid. Finally, the charge that disagreement is not solved by moral realism is discussed. Disagreement is argued to be not a serious threat for moral realism as a framework for an ideal of equality. In part II, containing the chapters 6, 7 and 8, the essential meaning of a political ideal of equality is determined that shows what is wrong with inequality. The related measure re ecting the moral badness of inequality is developed too in this part. In chapter 6, the meaning of equality is articulated. It is advanced as an answer to the objections any ideal of equality has to deal with. The so-called levelling down objection holding that a situation or distribution is not better in any aspect if the better o become worse o and the worse o remain equally worse o , leads to the articulation of a realistic individualistic ideal of equality. The central idea is that inequality is wrong because some are worse o than they might have been if the distribuenda were distributed such that all are equally well o with respect to the equalisandum. This answers the second question: `Why is equality desirable?'This ideal of equality is not derived from moral rules, but has its roots in a moral way of looking at oneself as one among others. In the chapters 7 and 8, a measure of inequality is determined and the third question: `What is the ordering according to the ideal of equality?' is answered. In chapter 7, several measures that are proposed in the literature on measures of inequality are discussed and a list of properties is proposed which a proper measure should satisfy. They are: 1. A restricted Pigou Dalton transfer principle, meaning that a transfer from a poor person to a rich person will worsen inequality, under the
CHAPTER 1. INTRODUCTION
14
assumption that the poor person is worse o than all could have been. 2. Symmetry restricted to equalisanda, meaning that changing who gets which part of the equalisanda, or what comes to the same, take a permutation of persons in the distribution, will make no di erence for the index of inequality. 3. Restricted homogeneity, expressing the idea that the index is independent of the arbitrary units of measuring the amount of equalisanda. 4. Principle of independence of no complaints, meaning that adding persons who have no complaints and are not worse o than all could have been does not change the index. 5. Weak independence, meaning that the ordering of inequality based on the distribution of a subgroup cannot be reversed by the welfare of others Decomposability or additivity, speci city, and the so-called population principle, are missing from this list. It is explained why these are missing. In chapter 8, it is argued against the view of Temkin and Sen that a complete ordering of inequality representing the moral badness of the distribution in a distribution problem is possible. A measure is arrived at which represents this ordering. It shows that a complete ordering is possible. The measure or index I() which is developed is:
X(x
I(~x) = M(
n
ref
i=1
, minfxref ; xig)2 )
(in which i represents persons, ~x represents the distribution vector of an equalisandum, M() represents a monotone increasing function and xref represents the amount of the equalisandum equally available for all) Part III is the nal part of this thesis and consists of chapter 9, in which the meaning of the ideal of equality developed in the moral realistic framework in the former chapters is illustrated for the domain of health and health care. The common issues raised in the discussion on health care are commented upon. It is argued that questions concerning distributions should be answered on particularistic judgements concerning the need for health care. I suggest that these t in the way of working of general practitioners.
Part I
A new framework for equality
15
Chapter 2
Moral universalism 2.1 Introduction In the introductory chapter, it was explained that the modern ideal of equality is traditionally associated with a set of background assumptions. One of these assumption is, as was explained, moral universalism. With moral universalism I referred to the idea that rules are necessary for moral judgements and are the basis for these judgements. These rules were supposed to be of the form: Actions of kind or distributions of kind ~a are . Whenever one does , or whenever there is a distribution ~a it is ; is a moral quali cation for example `right' or `should be chosen'. The reason why a particular action is for example, right is that it can be subsumed under such a rule. In these rules the actions or distributions are described in non-moral terms in such a way that they can be recognised. In the previous chapter it was argued that moral universalism blocked a complete ordering that re ects how bad a distribution is regarding the ideal of equality. It blocked a complete ordering because of the in uence of aspects of circumstances beyond those described by the distribution on the moral seriousness of inequality.1 Against the argument that moral universalism leads to an incomplete ordering it can be argued that instead of rules of the form Actions of kind or distributions of kind ~a are . moral universalism refers to rules of the form: 1
See chapter 1 p. 7.
17
18
CHAPTER 2. MORAL UNIVERSALISM
In circumstances of kind C actions of kind or distributions of kind ~a are . meaning that whenever there is a situation of kind C, an action of kind or a distribution of kind ~a is . In such rules, the situation too is described in non-moral terms because they too should be accessible to observation. This could mean that moral universalism could be saved from the charge that it is responsible for an incomplete ordering, because the ordering is no longer on distributions simpliciter, but on distributions in circumstances of kind C. Consequently, moral universalism would be no longer inferior to moral particularism regarding the argument of incompleteness. The main issue in the discussion between moral universalists and moral particularists would become the question whether indeed all the relevant aspects can be speci ed by C.2 The issue would be whether all our moral judgements about for example `What to do?', can be represented by moral rules. In this chapter, the introduction of moral particularism, the alternative to moral universalism, does not focus on the question whether all our moral judgements can be represented by moral rules. Apart from the arguments that our moral judgements can be represented by rules, one of which is spelled out in the appendix of this chapter, there is another issue in the discussion between moral particularism and moral universalism, which is more fundamental. It concerns the second part of the characterisation of moral universalism, the basis of moral judgements. Moral particularism or moral realism that is proposed in this chapter denies that moral rules are necessary in order to determine what is right or wrong. It holds that we judge that something is right or wrong because the world is as it is and not because our judgements can be derived from a set of rules we prefer or we decide to. We cannot point to rules as the basis of what is right or wrong. This moral realistic alternative view means that the argument for an incomplete ordering regarding inequality is no longer valid, because by accepting particularism it is no longer a priori valid that if in a particular situation of a particular kind distribution of kind ~a is better than a distribution of kind ~b regarding inequality, it is also better in any other situation of that particular kind. This moral particularism also means that it is no longer expected that a political ideal of equality is based on the application of or deduction from a set of moral rules. In the rst part of this chapter, I discuss four ways of reasoning with moral universalism that claim to arrive at morally sound judgements and hold that moral judgements are not just expressions of taste or choice but are objectively sound. The four ways that are discussed are those proposed 2 See also the discussion on separability, or independence of irrelevant alternatives, or the sure thing principle in [Broome, 1991, p. 94 .] [Hurley, 1989, p. 64 .].
2.1. INTRODUCTION
19
by Hare, Singer, O'Neill and Gewirth; they capture the essentials of the main motive for adhering to moral universalism. Each of the four ways of reasoning provides an argument in favour of equality. It is shown that these four ways of reasoning with moral universalism do not satisfy the claim that it saves moral judgements from drowning in the sceptical morass. Hence, an important reason for moral universalism, namely that it leads to objective morally sound judgements, is undermined. Subsequently two other reasons for accepting moral universalism are discussed. One is based on supervenience, the other on the concept of reason. Both reasons are dependent on the concept of similarity. Recognising this leads by following the ideas of Wittgenstein's Philosophical Investigations to moral particularism. In the second part of this chapter, moral realism is introduced. It is based on Davidson's radical interpretation view which can be seen as a further elaboration of Wittgenstein's ideas. After a short introduction of Davidson's radical interpretation view on language, moral realism is explicated by contrasting moral realism with moral universalism as it is elaborated by Hare, Singer, O'Neill and Gewirth. The contrast between moral universalism and moral realism is made clear by the issues of why people are moved by moral reasons, why they are moved by a particular moral reason rather than their own prudential reasons, and how the moral statements can be justi ed or how they can be known. These questions are commonly considered as serious problems for moral realism to deal with. In confronting moral realism with moral universalism on these questions, the answers of moral realism appear as superior to those of moral universalism. In the third part, I turn to a provisional evaluation of what moral realism introduced in this chapter means for the development of an ideal of equality. Moral realism as a moral particularistic alternative to moral universalism opens up the possibility of a complete ordering because the argument for the incompleteness, as outlined in the former chapter, is undermined. In moral realism, separability being an element of moral universalism used in the argument, is clearly no longer a priori valid. Another consequence of taking moral realism as a framework for the development of an ideal of equality is that in moral realism equality is based on a moral way of looking at others and ourselves. Seeing ourselves as one among others is not just a result of reasoning and calculation, it is a moral way of looking. Not seeing oneself as one among others is not a failure of reasoning, it is a morally defective way of looking at this world. Thus equality does regain its moral status. The invocation of a moral way of looking instead of reasoning is not a drawback, it is acknowledging that moral reasoning cannot do anything without taking a moral point of view. The moral point of view cannot be circumvented by reasoning.
CHAPTER 2. MORAL UNIVERSALISM
20
Reasoning is not compulsive, and knockdown arguments are not available.3 The moral way of looking at ourselves as one among others appears to be closely connected with the relation between rst person statements like `I like (believe) such and such' and third person statements like `He likes (believes) such and such'. In the next two chapters, moral realism is elaborated further by proposing realistic individualism and moral value pluralism as alternatives to volitional individualism and moral value monism respectively. After these elaborations it is possible to state and answer the criticisms on taking moral realism as a framework for the development of an ideal of equality, therefore that is postponed to chapter 5.
2.2 Why moral universalism?
2.2.1 The use of moral universalism
Moral universalism is supposed to be of help in providing a foundation for moral judgements. It is argued that teaching and learning morality is not possible without moral universalism [Frankena, 1973]. In learning morality one is learning principles which state what is right and what is wrong. In general, if one learns something one does not merely learn a particular instance of performing something, but one learns something which can be used again in the future. One learns to perform a certain kind of actions in certain kinds of situations [Hare, 1952, p. 60]. Similarly, if one learns morality one learns for instance, what kinds of actions are right and what kinds of actions are wrong. Learning morality means learning moral principles. These principles are the basis of moral judgements. They are of the form: In situations of kind C it is right to do actions of kind . Teaching, learning and even rationality itself assumes there to be something to be learned or to be rational. It concerns so-called principles of morality. It will not be dicult to imagine that if there are several principles there will arise con icts between them. For example the dilemma Jephta called down on himself (Judges, 11:30-40) is a situation showing the con ict between the principle of keeping one's promises and not ordering someone to be killed. Because of such moral con icts it is argued that the principles of morality are merely rules of thumb which have a prima facie status such that we will act in most cases correctly if we follow them. 3
See also the story of Achilles and the Tortoise cited in chapter 5 p. 159.
2.2. WHY MORAL UNIVERSALISM?
21
However, moral universalism does not merely mean that following moral rules will lead to making right judgements in general, it holds that it is one of the principal elements of morality itself. Moral universalism is not primarily important for so-called rst aid in moral emergencies, in which, because we lack time and intelligence to discover all the subtle moral important aspects, we should rely on rough principles. It is held that moral universalism is important in arguing why those rough rst aid rules should be kept at all. Its role concerns more than being rules of thumb. Moral universalism expresses a view on morality in which it is assumed that there is something to learn that can be reasonably defended and argued for in an objective way. It tries to nd criteria for what is right and what is wrong in order to escape moral scepticism. By this scepticism is meant that there is nothing special about moral judgements. They are seen as exclamations of preferences, which cannot be rationally argued for. Some believe that this has to be done and others believe that something else has to be done, that is the end of the discussion. There is no hope for agreement, there are no valid arguments. The only way of coming to an agreement that can be laid down in a contract is by bargaining, or using force. There is no sense in rational argument because there is no standard to which parties can turn for solving their disagreement. However, those subscribing to moral universalism take this subjectivism to be wrong and try to formulate a basic reason with the help of which the content of a rational agreement in a contract can be determined. Moral universalism is claimed to be of help in determining the content of such a rational agreement and in determining what kind of actions are right and what kind of actions are wrong. There are several ways in which moral universalism is claimed to be of help. Four of them, namely those elaborated by Hare, Singer, O'Neill and Gewirth, are discussed. These four philosophers defend moral universalism as a central element in determining what is right and what is wrong and these authors represent the essentials of all other ways of using moral universalism. In the presentation, it is shown how equality is argued for within these elaborations of moral universalism. These arguments will be seen again in the development of an ideal of equality in this thesis.
`What if someone did it to you?'
One way of reasoning in which moral universalism has an important role in determining what to do, is by attending to the question: `What if someone else did it to you?'. In other words, one is invited to look at another action of kind that a ects oneself instead of someone else. In his book Freedom and reason R.M. Hare illustrates this with an example borrowed from a parable that is described in Matthew 18:23 [Hare, 1960, p. 90]. Person i
22
CHAPTER 2. MORAL UNIVERSALISM
owes money to person j and person j owes money to person k. Now suppose further that there is a legal system that makes it possible that someone can put someone else in prison to exact the debt owed to him. Suppose j is making up his mind. Can he maintain that he ought to imprison i? Universalism says he might, but in that case, he also has to allow that k puts him into prison.4 Because he does not want this latter action to be performed, he should not imprison i. So in general, if someone is making up his mind whether some action of kind that he is inclined to do ought to be performed, he should ask himself whether this inclination would remain if he were the one a ected by an action of kind . If it does not, he should not perform actions of kind . To be clear, the inclinations Hare points at are not just wants of a whimsical nature but concern also what are called by Hare ideals. For example an ideal in Hare could be that debtors should pay their debt or else be imprisoned, even if someone himself is a debtor. So, j having this ideal could admit that in wondering whether he ought to imprison i he has to be imprisoned by k too. His ideal is that debtors should pay whatever the consequences to himself. Ideals being pervasive can be so important that living without them could make life not worth living. That is one of the characteristics of central ideals, they can go against desires and direct inclinations for example, the reasoning of Sartre's Pierre in Les jeux sont faits. Pierre argued to Eve, if I did not help my comrades I would be worthless to you, you would live with a coward and our living together would be worthless. Although I put our life at risk, if I didn't I would not be the Pierre you love. Thus ideals can be pervasive and can go against what we want directly. Although ideals that are pervasive are not necessarily irrational, they can be based on all reasons available, they are dicult to discuss. Disagreement on them cannot be settled easily and as Hare admits, sometimes they cannot be settled at all [Hare, 1960, p. 136]. Ideals are not essentially consensual, and disagreement can remain. This undermines the idea of avoiding the sceptical morass of subjectivism of moral judgements. What would this reasoning with moral universalism mean for equality? Hare is explicit on this [Hare, 1960, p. 118]. Suppose three people have one bar of chocolate and they want to divide this bar. Suppose further there are no other characteristics such as sex, ownership etc. that are relevant. Then because of moral universalism and the absence of a di erence that shows that one should get more than another, the bar should be divided in three equal parts. In situations of kind C one should do actions of kind 4 The terms `debt, pay, legal system, put in prison' are used here in a purely descriptive way without invoking any moral attitude to them. Within moral universalismthis is held to be possible. Below it will appear that in moral realism this makes no sense.
2.2. WHY MORAL UNIVERSALISM?
23
, where is `give a particular part of the chocolate' and the situations of the three persons are of the same kind C. Who is a ected by should not make a di erence. If they received parts of di erent sizes, then the rule `whenever situation of kind C one should do actions of kind ' would be violated. Hence, moral universalism would not be satis ed. So, moral universalism used in this way can justify equality. Without relevant differentiating reasons, all should receive the same.5 We meet this reason for equality again but in a di erent way, not as an element of our reasoning towards moral judgements but as an element of our recognition that we are one person among other persons.6 Summarising, the way of using moral universalism in determining what is right and what is wrong, by directing one's attention to the question `What if someone did it to you?', is dependent on inclinations, wants and ideals people have. A thorough disagreement on wants and ideals cannot be settled rationally. If one indeed thinks it is right to be punished with death by committing adultery, even if one committed it oneself (as the people in Thomas More's Utopia are suggested to hold), then according to Hare's view one cannot argue about this with someone who has other preferences and ideals. So, it does not solve moral problems as it promised to do. Let me turn to another way of reasoning using moral universalism that is less dependent on wants, desires and ideals of the actor.
`What if everybody did?' Another way of using moral universalism is by attending to the question `What if everybody did?'. It is elaborated by M.G. Singer in the following line of reasoning known as the generalisation argument. If some kind of action is such that if it were performed by everybody it would have bad consequences, then not everybody should do that action of kind . Because of universalism, if not everybody should perform an action of kind , then no one should. If it were right that one or some did , then it would be right for everybody to do , because moral universalism holds that if something is right for one, it is right for all in similar circumstances. Whenever there is a situation of kind C it is right to do , was the rule. For example, if it would be disastrous if all robbed a bank, then no one should do. It can be the case that it would not do any harm if person i 5 One can argue that in receiving parts of di erent size it is not clear at all that di erent moral judgements are posed on the three people, it could be the result of some procedure which is applied to all three alike. In this view however the equalisandum is not the bar but the application of a procedure. This problem concerning the proper equalisandum will be discussed in the next chapters. 6 See chapter 6 p. 199
24
CHAPTER 2. MORAL UNIVERSALISM
performed a particular action, for example took a ower from the park, or walked along some area in which rare birds were breeding, but it would still be wrong. Because if there is a reason for i to do the action there is similarly a reason for all others to do the same and that would lead to disastrous consequences. It even does not matter if it happened to be the case or if it could be argued that it is unlikely that others would perform it and that i is the only person who will perform the actions, it still would be a reason for all others and because if all performing this action leads to harmful consequences, it is wrong. If it is right for one to perform an action under a particular description, it is right for all. All could reason similarly. Disastrous consequences could still follow, according to Singer. Of course a simple argument against this way of reasoning with moral universalism, discussed by Singer too, is that it is too coarse. One could argue that it is wrong to be a surgeon because if all were, the consequences would be disastrous. There would be no plumbers for clean water in the hospitals. There would even be no clean water outside the hospitals either. But also, it would be wrong not to be a surgeon. If nobody were it would be disastrous too, wouldn't it? So, the generalisation argument seems to lead to the conclusion that it would be wrong to be and wrong not to be a surgeon. The same goes for all professions and all important activities. Consequently, the generalisation argument seems to be worthless. However, these problems can be handled by admitting that indeed the argument is not applicable in cases in which the principle is invertible. If the generalisation principle is invertible the actions should be described in a more relevant way, for example: being a surgeon if one has the necessary capacities. It is not clear at all that if all who have this capacity become a surgeon would be disastrous. There is another complication, namely if some action is described in an unnecessarily restricted way, for example dining at 6 p.m. at a particular restaurant. If everybody did, it would be disastrous. The problem here is that the action is described with irrelevant details such that they become iterative. In this case the argument cannot be used either. So, in the case that there are iterative predicates used and in the case that the argument is invertible the principle is not applicable. However, as shown by Singer, by adjusting the description properly it can be used. Against the generalisation argument it is also argued that because moral universalism is trivial, its use cannot be but trivial too. For example, although it can be argued that if all robbed a bank, it would be disastrous and because of this it is right for no one to rob a bank, it can also be argued that it would not be disastrous if all called Lionel B. born 24-111964 robbed a bank. So it could be allowed for Lionel B. born 24-11-1964 to rob a bank. Because all actions can be described in universal terms with
2.2. WHY MORAL UNIVERSALISM?
25
only one person satisfying the description the principle would be useless. However, against this reasoning Singer replies that if being called Lionel B. born 24-11-1964 can be a proper relevant descriptive feature similarly we should admit that being Harold A. born 25-11-1964 to be a proper description. We would come to the idea that if names and dates of birth were relevant then again we arrive at disastrous consequences because all could claim to have a reason to rob a bank. Hence, we should not allow these arbitrary descriptions. The generalisation principle is not so easily dismissed as trivial. Equality can be argued for by this generalisation argument. If it is right for one to receive a certain amount of some good, it is equally right for another. If not all can get the same amount then nobody should get it. Thus it can be seen that equality is a consequence of the generalisation argument. The idea of availability for all is met again in the ideal of equality proposed in this thesis. It holds that inequality is wrong because some are worse o than all could have been simultaneously. The reference situation with respect to which it is evaluated how worse the worse o are and how serious the inequality is, is determined by what is available for all.7 How should we evaluate this way of reasoning using moral universalism in determining what is right and what is wrong? It turned out that the answer to the question `What if everybody did?' is dependent on consequences being disastrous. Without these judgements the argument is impotent. And about this, contrary to what Singer believes, there can be a permanent disagreement [Singer, 1971, p. 94]. Consequences seen as good by some are sometimes evaluated as disastrous by others. So, the whole argument for establishing a proper foundation for morality and a way of establishing what is right and what is wrong depends on this agreement. Although the generalisation argument seemed to be a way of determining the basis of morality, it failed because of its dependence on evaluating consequences and these are subject to very important moral disputes. Let me turn to another way of reasoning using moral universalism which does not depend on agreement on what counts as disastrous, but which uses inconsistency as an argument in showing that actions are wrong or are right.
`What if all others did it too?' The third way of reasoning with moral universalism, which is not dependent on an external evaluation of consequences and which consequently does not su er from the drawback of the previous ways of reasoning with moral universalism, is principally based on the idea of inconsistency and is borrowed 7
See chapter 6 p. 198.
26
CHAPTER 2. MORAL UNIVERSALISM
from Kant's categorical imperative. O. O'Neill explored it in her Acting on principle. Actions are best described by the maxim the agent holds in acting. Maxims are the most complete descriptions of what an agent is doing. In a maxim, the action is speci ed with the help of the intention and purpose of the action as it is seen by the agent himself. These descriptions have no moral content and are just descriptive; they receive their moral content by the following test. The test for determining the moral value of actions is whether it would be inconsistent to hold both: 1. I will if C in order to P . 2. Everyone will if C in order to P . Actions of kind under conditions of kind C with purposes of kind P are permitted if these two statements are not inconsistent. They should not be performed if they are inconsistent. Two kinds of contradictions should be distinguished: `contradiction in conception' and `contradiction in the will'. An example of a contradiction in conception can be illustrated by the maxim becoming a slaveholder. The maxim on its own is not contradictory, but together with its universalised version it leads to a contradiction. If everybody becomes a slaveholder then everybody would have property rights and nobody would be a slave. Consequently, there could be no slaveholders, and therefore satisfaction of the universalised version is conceptually impossible. Hence, one should not become a slaveholder. The other contradiction, contradiction in the will is more important. It arises if the maxim together with its universalised version cannot be maintained simultaneously; not because it is impossible to hold both, but because the universalistic part will lead to unfavourable conditions under which the success of the actions speci ed in the maxim becomes less likely. For example, it is consistent to hold that: 1. I neglect everything others need. 2. Everyone neglects everything others need. Strictly spoken these are not inconsistent. But if I am a normal rational agent, I want also the necessary and sucient means to success in order to reach some success of the action described by my maxim. So, if I am in need of help in order to succeed, then I also want help. But exactly this is denied by the universal counterpart; everyone neglects the needs of others. A similar argument should be given against governments who for example aim for a surplus in international trade. This is not universalisable, not only because it is impossible for all countries to have a trade surplus, but
2.2. WHY MORAL UNIVERSALISM?
27
because the universalised maxim will lead to conditions which make it less likely for the government to succeed. With this way of reasoning with moral universalism equality is to be argued for in the following way. The maxim of acting for inequality leads to an inconsistency: 1. I will act in order to get more than everyone else does. 2. Everyone will act in order to get more than everyone else does. First, it is impossible to have a situation in which everyone gets more than everyone else does. It is impossible and hence a contradiction in conception. But this contradiction could be circumvented by turning to competition and see the content of the maxim as trying to get more than others. Together with the universalised maxim there is now not a contradiction in conception. However, there still is a contradiction, a contradiction in the will. This contradiction in the will is the reason for condemning inequality. The desire to become better o than the others, compete with others, should not be accepted rationally because the universal variant will lead to situations which precludes my becoming better o than others. Others will try to block my attempts to succeed. So, there is contradiction in the will. It is not said that competition is never permitted, it is, in restricted areas, in sports and games for example. Sports and games have as the ultimate goal not winning but playing itself. However, competition in the economic sense does not have this element of a game. It concerns everything. One central idea of games is that one can refuse to play a game, but one cannot reject economics or politics. One cannot see real life as just a game, because there is no other domain to retreat to as is possible with games. There is nothing beyond living, but there is something beyond sport and music competitions. So, it is not rational to want competition in which everybody strives to become better o than all others. It could be possible to defend competition to be rational as means of improving all, but it remains irrational to see it as the only means [O'Neill, 1985]. That is the reason why inequality should not be striven at. This idea about equality, which is rather similar to the one connected with the generalisation argument above, is also met again in the idea of availability for all, which is a central element of equality as an ideal. The reasoning with moral universalism in questioning `What if everybody did it too?'and formulating the universalised maxim does not su er from the drawback of the foregoing methods, in which the dependency on evaluating consequences undermined the use of universalism as a way of escaping the subjectivism and the relativism of morality. In the reasoning presented here such an evaluation of the consequences is considered as part
CHAPTER 2. MORAL UNIVERSALISM
28
of the description of the actions by their maxims; the success of the actions is aimed at and is the ground for evaluating the consequences of the universalised maxim. It only uses inconsistency in conception or inconsistency in the will in determining whether something is right or wrong. The evaluation is internal to the test and not external as in those described in the former sections. This seems to be less arbitrary. However, its use in determining what should be done is highly dependent on the maxim of the agent performing the action. This dependency does make it less useful. For instance, it can be argued that the act of refusing bribes should be condemned, because the universal variant would lead to the extinguishing of the practice of bribery and as a result refusing bribes would become impossible, hence an inconsistency in the will. Consequently, refusing bribes would not be permitted.8 This line of reasoning however could be rejected by holding that the maxim is not stated in a relevant way; it is not `refusing bribes', but `refusing bribes in order not to disturb a fair procedure'. The main element of the maxim is `acting in such a way that fair procedures are not undermined', of which the universal variant is not impossible at all. But whatever the content of the maxim should be in this example, it shows the dependency of the precise content of the maxim. If refusing bribes is not seen in the light of the maxim of the agent furthering fair procedures, but the agent just refuses bribes because they are too risky to accept and nothing else, then refusing bribes would be not permitted. In other words, one should accept bribes. Hence, an action can easily be argued as a permitted one by the test, just because it can be seen in the light of a particular maxim that will not lead to an inconsistency. The test seems to be more appropriate to judge reasons and persons than actions and situations. To be fair to O'Neill it has to be mentioned here that she emphasises that too. In this study, I am interested in evaluating actions and situations rather than reasons and persons. The aim of this thesis is the development of equality as a political ideal with respect to which distributions can be evaluated. As in the discussing whether something is right or wrong, the content of the subjective maxim under which the action is performed will be the most debated issue, which is not easily settled, the objectivity is likely to be threatened. So, let me turn to the next way of reasoning with moral universalism, which is not dependent on the precise content of subjective maxims or external evaluations. 8
The example is due to Singer [Singer, 1971, p. 279].
2.2. WHY MORAL UNIVERSALISM?
29
`What is a right for you, is a right for another too.'
The previous uses of moral universalism did not show that moral universalism does overcome the threat of subjectivism. It was seen that moral judgements were dependent on ideals, evaluations of consequences as being disastrous, and subjective maxims. The fourth way of reasoning with moral universalism does not su er from these drawbacks. It is a sort of transcendental argument. The line of argument is: 1. 2. 3. 4. 5. 6. 7. 8.
Agents perform actions. Those actions have a purpose. In acting agents want to control the process in reaching this purpose. Consequently, as agents they see freedom as an important value for themselves Furthermore, as agents they want to be capable of performing actions. Hence, they see well-being as an important condition. Because they have a reason to claim a right to freedom and well-being and because agents are similar as performers of actions one should judge rationally all other agents to have those claims too. Consequently, all agents have a similar right to freedom and wellbeing.
The above reasoning exposed by A. Gewirth in Reason and morality, is not based on subjective wants of agents but on values that all agents as agents are bound to have. They have reasons for appreciating freedom and well-being for themselves. But because it is a reason for them it is a reason for others too (7). In this step moral universalism is recognised [Gewirth, 1978, p. 105]. So, if they are rational they see that other agents have those rights in a similar way. Thus equality as equality of freedom and well-being is a consequence of rationality. The idea `what is a reason for you is a reason for another too' is met again in the ideal of equality proposed in this study, not in a chain of reasoning but as an element of what it is to recognise oneself as one person among other persons.9 In general, this reasoning seems undeniably valid, but what comes of it if it is used in particular circumstances? What is the sort of rationality 9
See chapter 6 p. 199.
30
CHAPTER 2. MORAL UNIVERSALISM
used and why is moral universalism used in this way? Let me turn to the question of how it can be of help in solving moral problems and how it gives an answer to the question `What should I do?' One of the problems seen in the application of the above stated ideas is of course when well-being and freedom con ict, or when the freedom of one person is in con ict with the freedom of another, or if the well-being of one person is incompatible with the well-being of another. In general, how does it help in solving these common moral problems? It is argued by Gewirth that if there are moral con icts, they can be solved by determining how important the freedom for one person qua being an agent is in comparison with the freedom of the a ected other, or whether the well-being of one is more important for being an agent than the particular freedom. And of course similarly, the well-being of the one should be compared with the well-being of another in case there is a con ict. If the well-being of the rst is less important than the well-being of the latter, the well-being of the latter should prevail. As will be clear, the judgement on the importance for a particular freedom or well-being for being an agent is an evaluative matter on which all other judgements depend. So, the judgement itself is a moral statement about which there can be disagreement and many problems are of this nature. For instance, how much to spend on defence, or health care, or infrastructural improvements, which part of health care should be granted to all and which part can we accept as dependent on people's own choice? Whose well-being counts how much against whose freedom? It is not at all obvious that this fourth way of reasoning with moral universalism will be of help in solving moral problems in a way such that moral arguments get an objective status. Moral arguments remain dependent on evaluations, for example views about ideals, what is to count as more important, freedom or well-being; a dependency, which moral universalism wanted to circumvent.
Conclusion
Summarising, moral universalism was introduced as a way of lending more objectivity to moral judgements and as a help for arriving at moral solutions to moral problems. I discussed four ways of reasoning with moral universalism. I mentioned that each had its own view on equality, which we meet again as elements of the political ideal of equality proposed in this study. But the main issue here concerned the inadequacy of moral universalism to solve moral problems. Moral con icts remain unsolved because people have di erent ideals, di erent ideas on what is disastrous, di erent descriptions and maxims of actions, and di erent judgements on the importance of actions with respect to freedom and well-being. So, in applying these
2.2. WHY MORAL UNIVERSALISM?
31
proposals it became clear that they do not satisfy the promise of moral universalism to pull morality out of the morass of subjectivity by invoking rationality. Moral problems are not solved at all by moral universalism. It has been shown that one reason for accepting moral universalism to be not quite adequate. There is no sense in accepting moral universalism if it is motivated by the idea that it saves moral judgements from scepticism. The main motive for adopting moral universalism is undermined. Nevertheless it is still possible that some other arguments show moral universalism to be unavoidable. Let me turn to these.
2.2.2 Arguments for moral universalism, and moral particularism
In the previous section, it was shown that in the end none of the four ways of reasoning with moral universalism could solve moral problems. One of the reasons for trying to establish and accepting moral universalism, is undermined. But it is not yet shown that it is false or unreasonable. Before I can conclude that it is false, I have to examine the arguments for moral universalism and show that they are not valid. To this task I will turn now. There can be discerned two sorts of arguments for moral universalism. One is based on so-called supervenience and is advanced by Hare and the other is based on the concept of reason and is proposed by Gewirth and Singer.
Supervenience
One reason for holding that moral universalism is true is because moral predicates are supervenient on non-moral predicates. This means that if two actions, persons, or situations are judged to be morally di erent then there has to be a non-moral di erence between them, otherwise the use of the moral predicate is unintelligible. Moral predicates have a non-moral descriptive content in virtue of which actions, persons or situations receive their moral predicates. This view is argued for in the following way. The word `good' is learned as a word in situations of choice and is used for recommending an object or action. For instance, a good knife has some properties that makes it a good knife, it is sharp, easy to handle etc. So, because of some non-evaluative properties an object is evaluated as good. It would be unreasonable or even not understood, if two knives with similar properties were not both evaluated as either good or not good [Hare, 1952, p. 95]. Similarly, the moral use of the word `good' has descriptive content. Because an action has particular properties it is good and is recommended. A similar action, i.e. an action with the same non-moral properties, should also be evaluated as good, otherwise the meaning of the
CHAPTER 2. MORAL UNIVERSALISM
32
words `good', `right' etc. could not be learned properly and would not have any intelligible meaning at all. Moral universalism is a necessity for learning the meaning of moral predicates. It is derived from the necessity of supervenience of moral predicates on non-moral predicates. This supervenience has as a consequence that similar actions in similar situations, in which the similarity is stated in non-moral terms, should be evaluated equally. This view on moral language and moral universalism is advocated by Hare [Hare, 1952] [Hare, 1960]. If a certain non-moral description gives rise to a certain moral predicate, everything conforming to that non-moral description give rise to the same moral predicate. Similarity in non-moral predicates implies similarity of moral evaluation. The non-moral predicates are called standards of moral evaluations. This argument for moral universalism is dependent on the concept of similarity stated in non-moral terms. It is considered to be prior to similarity in moral evaluation. This view is questioned shortly, but rst the argument for moral universalism based on reasons is presented, because as is shown both arguments can be criticised by one and the same argument.
The idea of reasons
The other argument for moral universalism is based on the concept of a reason. Moral universalists such as Singer and Gewirth hold that if something is Y because of predicates P1; P2; P then it should be the case that always if something is P1; P2; P it is also Y. When it is not, the predicates P1 ; P2; P are not the reasons for Y, and there has to be some other predicate that is relevant for being Y. It is held that it is similar to the notion of cause. If something A is the cause of B then always: if A then also B; if A and not B, something di erent from A would be also part of the cause for B; thus the principal argument for moral universalism [Singer, 1971, p. 37][Gewirth, 1978, p. 105]. The above stated argument for moral universalism is dependent on a particular view on reasoning, namely as a sort of deduction in classical logic. But our reasoning is not covered properly by the system of classical logic. Some other possibilities are suggested as systems of reasoning, for example non-monotone systems as proposed by for instance, D. Lewis [Lewis, 1973] [Stalnaker, 1984]. These systems are called non-monotone because, in contrast to the classical systems, an increase in premises is not necessarily followed monotonically by an increase in conclusions. In this kind of logic, it can be the case that with premise P one can conclude Q but with P and some other premise P', one has to withdraw conclusion Q without holding that one cannot accept Q if P. So, P can be the reason for holding that Q, but not always, for instance, in the case that P' is also accepted. To follow
n
n
n
2.2. WHY MORAL UNIVERSALISM?
33
the favourite example of non-monotone reasoning: If Tweety is a bird then we can conclude it can y. But if we learn that Tweety is a penguin we have to withdraw the conclusion that Tweety can y. Adding the premise that Tweety is a penguin, leads to withdrawing a conclusion, namely that Tweety can y. The set of conclusions does not follow monotonically the set of premises. Similarly, we can state that the reason for someone going to the island Schiermonnikoog is that he has his work over there. But he would not travel to Schiermonnikoog if one of his legs was fractured, or if he came to know the ferry was too dangerous. But that none of his legs is fractured, or that the boat is not too dangerous, is not a reason for his travelling to Schiermonnikoog. The reason is just doing his work over there and not the absence of all sorts of imaginary conditions which would preclude his going. Non-monotone reasoning is more appropriate for modelling our reasoning than classical logic. By accepting non-monotone reasoning as more appropriate, the argument for universalism does not seem to hold any longer. It can be the case that doing his work on the island Schiermonnikoog is his reason for the journey, but not on all occasions he has that reason he will go. There can be defeaters.10 Hence, the argument based on what reasons are, in favour of moral universalism, is not convincing. Similarly, it can be held that making someone happy is a reason for an action that makes that person happy, until we learn it concerns killing foxes for pleasure or playing baseball by hitting simultaneously heads of cows. Being a lie is seen as a reason for not making the utterance, but in case you can save a life by lying it is di erent. In addition, borrowing a book is a reason for returning it to the one you borrowed it from, until you learned it was stolen by him. Something can be a reason without being a reason in other circumstances [Dancy, 1993, p. 79]. Reasons are not reasons on their own but only in combination with other premises, they are not separable. This phenomenon is accounted for in so-called non-monotone reasoning.11 Although moral universalism interpreted as separability of reasons is precluded by non-monotone reasoning, it is not denied that if because of P1; P2; P something is Y, it would be Y in similar circumstances in which P1; P2; P hold. In this interpretation one could argue moral universalism to be right. But now similarity turned out to be the main concept on which moral universalism is based just as it was in the argument based on supervenience. The two arguments have the dependency on similarity in common. Let me turn to this concept in order to determine whether the arguments for moral universalism are convincing.
n
n
See also [Dancy, 1993, p. 24]. A similar line of argument is found in [Brink, 1989, p. 160] in his exploration of so-called weak supervenience. 10 11
34
CHAPTER 2. MORAL UNIVERSALISM
Similarity; Wittgenstein's view and particularism
In the previous paragraphs it became clear that the intelligibility of moral universalism depends on the concept of similarity. This concept appeared to be central in the argument of supervenience and the argument of what a reason is, but what does similarity mean? To what could we point in order to show similarity between actions, persons and situations? A rst answer would be: a common characteristic. But what would that answer mean? Let me look at a simple case, similarity of length of persons. We can state that people are of an equal length if their length is equal, which means that as the result of performing a particular procedure, taking a measurement, the same number is given to persons of equal length. The procedure of taking a measurement could be described as the number of times a measuring rod can be put alongside a person in a lying position after that person slept well. But what should we say if one of the examiners broke o a little part of the measuring rod each time it was laid alongside the person and claimed to be measuring the length? To what could we point in order to show him he is wrong in his claim? If we said: `Look here two people of equal height which are not of equal length according to your measurement.', he could answer: `But those persons are not of equal length. They have, as I have just shown, a di erent length.' How could we convince him that he does not follow the prescribed procedure properly. Of course he does it in the wrong way, that is not the question, but how could we make him clear he is wrong. It is like the problem Wittgenstein posed in discussing what is it to follow a rule beyond the actions performed in following that rule [Wittgenstein, 1945, 185]. To what could we point if someone who claims to be `adding 2 repetitively' goes on with 1000, 1004, 1008. How could we make clear that he is doing something else than `adding 2 repetitively'? As was made clear by Wittgenstein, we could not point to some sort of programme because programmes can only be discovered by us in operation and how could we distinguish system failures from the programme? Wittgenstein's answer is clear, there is nothing to point to beyond the common practice of people, or the common form of life. These practices are learned by us. One of the practices is learning language, also moral language. We learn to use the words `good', `right' and `wrong'. We learn to apply these words to actions, persons and situations. From this it will be clear that the basic relevant similarities in actions and persons are their common moral evaluations and not their non-moral similarities. The predicates `good', `right' and `wrong' are learned directly x
2.2. WHY MORAL UNIVERSALISM?
35
in a common practice. In using moral language some common form of life and morality is assumed to exist. Morality is not intelligible beyond this common practice. Consequently, it is by learning what persons are good, what actions are right and what actions are wrong, we learn moral predicates. We group them together under these moral descriptions of common evaluations. Afterwards we discover non-moral predicates that go along with them. But rst moral evaluation, later the non-moral description. As McDowell argues, it is unlikely that we can group actions together in virtue of their non-moral predicates without already knowing what is good and right [McDowell, 1981] [McDowell, 1985]. Of course, we could learn this grouping together but not in a comprehensible way without already understanding the moral predicates `good' and `right'. For instance, in learning how to use the word rude, we can group together actions that can be seen as examples of rude actions and afterwards we discover the non-moral properties of these actions. These properties are not the basic grounds for describing the action as rude. The action can be described in that way because it is rude. To be clear, it is not denied that some moral evaluations can be learned by moral rules, but it is denied that they can be learned without having any idea of what good and right means. 12 It is not denied that some reasons can be given why this particular action is rude by pointing to some features of the action. But the principal ground for calling this action rude is because it is rude, it is what rude means. In short, the similarity which is relevant for moral predicates is dependent on the practice of moral evaluation, and not directly on the non-moral similarity. A supervenience relation happens to be true and is not a basic characteristic of moral terms. The non-moral terms are not the ground for applying a moral predicate. A moral predicate can be given because its application leads to a true judgement. The predicate `right', or `wrong', or `rude' can be applied because an action is right, or wrong, or rude. To be 12 Williams argues against this view that it is possible to use moral terms without accepting the moral judgement, namely in a quotational manner. For example, it is possible for someone to say: `He said rightly that that action is rude.' without accepting the evaluative practice of using this term by himself [Williams, 1985, p. 143]. This can be accepted but it is an argument against the view that it is always impossible to use the non-evaluative counterpart of moral terms. But this is not held by the radical interpretation view (See next section). In this radical interpretation view it is just denied that all moral terms can be used in this way. It is denied that someone is a total amoralist and does not accept any morality at all, but a partial amoralist remains possible. The holistic nature of the radical interpretationview admits a partial amoralist. So, the internalism of radical interpretation is immune against the argument of Brink in [Brink, 1989, p. 45]. (See below)
CHAPTER 2. MORAL UNIVERSALISM
36
clear, it is not denied that there are rules that represent our moral judgements or can even guide our judgements. But rules are no longer important in showing what should be done, or what is right in a particular situation. They are only derivatives of moral judgements. It can no longer be held that something is rude because it has such and such non-moral predicates. It is rude because that is what rude means. Those non-moral properties can be used as indications that some actions are likely to be rude but they are not the ground of the action being rude. Moral universalism does not any longer service to indicate which actions are right. Rules can be rules of thumb, nothing more. The ground for evaluating some particular action as right, wrong or rude, is the rightness, the wrongness or the rudeness of that particular action and not some application of standards. Those standards are secondary. The particular judgements are conceptually prior and are embedded in a common form of life. Hence, moral particularism is more appropriate than moral universalism. Summarising, in these sections it was shown that moral universalism did not solve moral con icts. One of its reasons namely that it would remove moral con icts and rescue morality from the morass of subjectivity, relativity and scepticism was undermined. Furthermore, it was shown that the reasons for moral universalism based on supervenience and what a reason is, were dependent on the concept of similarity. This concept was pivotal in the argument for moral particularism guided by Wittgenstein's view on rule following. The main discussion between moral universalism and moral particularism does not concern the question whether there can be rules such that by following them we will act morally right.13 The crucial issue is the role of rules for moral judgements. In moral universalism rules are basic and in moral particularism the particular judgements embedded in a common form of life are fundamental. This is an aspect of moral realism that is discussed in the next part of this chapter.
2.3 Moral realism
2.3.1 Davidson's radical interpretation
After undermining the reason for moral universalism by showing that it did not solve moral problems, it was also shown that the arguments for moral unversalism were not convincing. The arguments based on supervenience and the concept of reason depended on the interpretation of similarity. It was shown that similarity in moral matters is determined by the moral 13
See also the appendix of this chapter.
2.3. MORAL REALISM
37
evaluation itself directly and not by the non-moral predicates as is held by moral universalism. It was argued that the foundation of a particular judgement is its truth. It is embedded in the common practice of moral evaluation. In order to understand each other we have to assume some statements to be true. We interpret with help of these the meaning of statements that people express. The common practice has to be assumed. The truth of some statements has to be assumed. Without this assumption, we cannot interpret what others mean when they use language. This is the central idea of Davidson's theory of radical interpretation [Davidson, 1984] [Davidson, 1986] [Davidson, 1990]. It is called radical because no expressions have a meaning on their own, all expressions have to be interpreted. If we want to understand each other, we have to interpret each other. We meet each other and we try to determine what the other means, it is essentially radical interpretation. The content of beliefs and desires has to be interpreted. It was made clear by Davidson that we cannot understand what another means, believes, wants and does, separately. Beliefs, desires and actions are interrelated. We interpret another person by questioning him and by looking at what he is doing. We interpret actions and utterances. We do not ascribe rst beliefs about the world to people apart from ascribing desires to them. It is done simultaneously and the ascriptions are interdependent. Interpretation of what someone else means is only possible because we presuppose he has similar perceptions as we have, lives in the same world and is as rational as we are. The assumption that another is like us and lives in the same world as we do is called the principle of charity. If someone speaking an unfamiliar language sees a rabbit on the island Schiermonnikoog and utters: `Look a deer' , our interpretation will not be `Look a deer' (there are no deer on the island). We presuppose that he sees the same as we do, for example a rabbit. Our interpretation will be `Look a rabbit', with `deer ' he means a `rabbit'. The same goes for reasons for actions. We have to assume a large part of similarity in what is valuable to us and what we care about, in order to understand what another is doing. If someone is walking to the baker, we do not interpret this as a way of getting parsley to the moon. In interpreting another we have to presuppose a broad agreement. The agreement that is necessary for understanding, summarised in what is called the principle of charity, is not like an option for us. We cannot choose for it; it cannot be rejected. We just have to assume this principle in interpreting each other. We can of course argue about its content but the principle of charity itself remains transcendental to all understanding and deliberation. We have to assume a common world, without this assumption
38
CHAPTER 2. MORAL UNIVERSALISM
we cannot interpret each other, and the content of beliefs, desires and the meaning and actions cannot be determined. Utterances receive their meaning only by interpretation and this is only possible if general agreement is presupposed. This holds for utterances that refer to how the world is going, as well as for utterances that are evaluations of this going on. This radical interpretation is taken as the basis for moral realism that is referred to in this thesis. With moral realism I refer to the idea that moral judgements are taken to be true or not true independent of what we prefer or decide to be the case. In order to get this idea more clear I discuss it in contrast to moral universalism with respect to the issues of the explanation of the relation between moral beliefs and actions in general, why people are moved by particular moral reasons rather than prudential reasons and how moral statements can be justi ed or known? These are generally mentioned as serious problems for moral realism. I show that moral realism has plausible particularistic answers to these problems. It is the basis of the further elaborations in the next chapters, in which realistic individualism is introduced that can account for interpersonal comparisons, and in which moral value pluralism is introduced. But let me rst turn to the questions: 1. What is the explanation of the relation between moral beliefs and actions in general? 2. Why are people moved by particular moral reasons rather than prudential reasons? 3. How can moral statements be justi ed or known? in order to illuminate moral realism.
2.3.2 The relation between moral beliefs and actions.
In general, it is held that moral beliefs are linked up with actions and their motivation. Whether we read the work of Hare, Singer, O'Neill, or Gewirth, all are convinced that moral beliefs are inherently motivating and not because we want to, or we have a desire to be moral. The latter suggestion would lead immediately to the question `Where does this desire come from?' What should the answer be: `Evolution', or `Implanted by our creator'? But can we accept these answers and still hold that it is our desire? They would be instincts rather than desires. We would not be moral because we want to, but because the way we developed in evolution, or because the way we were created. It would be external to us. But what else could we say? We like to be moral, just as we like ice cream? But do
2.3. MORAL REALISM
39
we like to be moral? That is highly dubious. Invoking some desire to be moral will not be a satisfying answer to the question `Why be moral?' An internal link is more likely. An internal link of course has to be explained. Hare for example explains this link by his `prescriptivism'. He holds that moral terms such as `good' are learned in contexts of choice. One of the aspects of evaluating something as good is choosing it from alternatives that are less good. This choice is implied by the use of moral language. If one says that something is good but he does not act according to it, we doubt whether he really believes it is good. So, the question `Why be moral?' is answered by Hare by his prescriptivism according to which moral terms go together with prescriptions. Although it seems a plausible view, it is not a satisfactory answer, because it assumes that there is a strict separation between descriptive language and moral language. The former is used primarily as a way of picturing the world and the latter is primarily used as advice in situations of choice. Description is the primary use of the descriptive part of language and prescription is the primary use of moral language. But this separation is not convincing. We can use descriptions as prescriptions and the other way round. Shouting `Fire' is describing and informing that there is some re, but it is also some sort of advice, it can mean `Call the re brigade!'. Similarly, saying of Caesar that he was cruel, is giving a description. We cannot preclude that he became a ruler, we cannot in uence his actions anymore. It is more like a description than a prescription. And why should we take the predicate cruel not to be primarily descriptive and `Fire' primarily prescriptive? Why should we say that `Fire' means in fact `Call the re brigade!', why not the other way round and say that the real meaning of `Call the re brigade!' is `Fire' and the former expression is just a longer expression for the latter?14 Although Hare's explanation for the internal link between moral language and action seems to be plausible, it runs into problems because of its separation between prescriptive and descriptive language. A disadvantage the radical interpretation theory of Davidson does not su er from. But let me turn rst to the answer of Gewirth and Singer. The moral universalists Gewirth and Singer, answer the question about the relation between moral beliefs and actions by rationality. It is rational to follow moral beliefs. In denying moral claims to be relevant at all, one is irrational. But this answer triggers immediately the question: `But why be rational?'. One answer would be that it is self-defeating to claim there are reasons for not being rational. In holding that irrationality can be argued for, one is bound to rationality. So, being rational cannot be denied 14
See also [Wittgenstein, 1945, x19].
CHAPTER 2. MORAL UNIVERSALISM
40
rationally. But it can be denied arationally. If someone is irrational and does not claim to have reasons for his irrationality we have someone who accepts arational irrationalism, which is held to be a consistent possibility. Hence, we have still to answer the question `Why be rational?'. It is not answered by Gewirth and Singer. In the radical interpretation view the question `Why be rational?' is answered in a simple way. We are rational just because we are bound to assume we are. We cannot interpret each other as arational irrationals. The position of an arational irrationalist is a possible position, but cannot be lived up to in such a way that we should consider someone as being an arational irrationalist. In interpreting what others say, believe, want, and do, also the arational irrationalist, we have to suppose that people are rational like we are. Although it is a possible view, which is not inconsistent, it is not communicable. It is like the argument of Wittgenstein about solipsism, it is a consistent view but not communicable. The solipsist telling he is the only person in the world, is using the concept of external world in which he is the only object, which is not consistent, because the solipsist holds actually there is no external world, so he cannot express his belief properly [Wittgenstein, 1918, 5.62]. Before turning more extensively to the radical interpretation view on the relation between moral beliefs and actions, I have to mention the view of O'Neill. She would answer the question of how moral beliefs motivate for actions, that we could not live on our own. We are not autonomous separate persons who can do things on our own, we are social beings [O'Neill, 1985] [O'Neill, 1989]. But how this answer should be interpreted, is not yet clear to me. Sometimes it is stated as a prudential reason, but then what about those who in fact can do on their own. The prudential reason seems to be not satisfying because it would for example lead to the idea of caring only for the ones I may need and just forgetting all the others. It looks as if the original question is transformed into `Why follow particular moral reasons instead of prudential reasons?' which is taken up in the next section. In the radical interpretation theory of meaning, the link between moral beliefs and actions is explained as a general feature of language and not just of moral language as Hare did. As mentioned above, the meaning of utterances has to be interpreted. What somebody says, believes, wants, and does can only be interpreted. These elements are not interpreted separately. In interpreting what someone means, we listen and look at him, and we have to presuppose a common world we live in. If someone is behaving in a particular way, we will consider his behaviour as relevant information in determining the content of his beliefs. Actions and beliefs are not independent of each other. In the way beliefs in general are linked with x
2.3. MORAL REALISM
41
actions, moral beliefs are linked with actions as well. There is nothing special about moral terms linked with actions, contrary to what Hare assumed. To be clear, it is not said that a particular belief leads necessarily to a certain particular action. This is not, contrary to what Brink holds, one of the elements of internalism [Brink, 1989, p. 45]. He argues against internalism, the view that moral beliefs and actions are linked conceptually, that it would consider an amoralist who can use our moral terms but is not motivated by them, or does not care about them, as an impossibility, while actually such an amoralist is possible. The possibility of such an amoralist is not denied in the radical interpretation view, if with an amoralist is meant a partial amoralist who does not share all of our moral convictions. Consequently, Brink would consider the radical interpretation a form of externalism because there is no link between particular beliefs and particular actions. But because a total amoralist with no morality at all, is incomprehensible in the radical interpretation view, the radical interpretation view is still to be considered as some sort of internalism.15 In the radical interpretation view it is stated that it is a conceptual necessity that language and action are knitted to each other. Linguistic utterances have no meaning without this link, they would even not be recognised as such. Beliefs, whether moral or non-moral, can be considered to cause actions because the content of the beliefs is determined by the interpretations of these actions. In explaining and understanding actions of others, there is no demand for a particular natural law connecting this kind of beliefs with that kind of actions; the intellectual discovery is the other way round. We act, we interpret these actions. We interpret these actions as caused by beliefs of which the content itself is an interpretation partly determined by these actions. The purpose is not nding laws between beliefs and actions and predict actions from beliefs. We are simply obliged to understand and interpret beliefs. The general question `Why be moral?' is answered in the radical interpretation view by the platitude that we simply are, it is a condition of interpreting each other. The link between moral beliefs and actions is just like the link between other beliefs and actions. Its nature is illuminated by radical interpretation simply as the relation between language, beliefs and actions. It makes no sense to consider moral beliefs apart from actions. The content of moral beliefs is determined by interpretation for which actions are essential sources of information. There is by necessity a conceptual link between morality and actions. In the radical interpretation view the question `Why be moral?' is answered in a simple way. 15
See the footnote on p. 35.
CHAPTER 2. MORAL UNIVERSALISM
42
Summarising, the answer to the question about the explanation of the relation between moral beliefs and actions, is not properly answered by invoking a desire to be moral, because it is highly questionable whether we desire to be moral. The prescriptivism of Hare was neither a satisfying answer because it assumed a strict separation between descriptive and prescriptive language. Nor was the answer `Because we are rational' on its own satisfactory. It had to be supplied by the radical interpretation view in which we see ourselves as rational. Being an arational irrationalist is not a recognisable option for us. The explanation of the relation between moral beliefs and actions is just simply that beliefs and also moral beliefs have to be interpreted and are conceptually linked with actions and wants.
2.3.3 Why follow particular moral reasons rather than prudential reasons?
The second question to be considered in order to illuminate moral realism does concern the motivation for following a particular moral reason rather than a prudential reason. It is closer to the common sense question than the former abstract question about following moral reasons without considering the alternative reasons. The alternatives lure people away from acting morally right, hence, the relevance of this more practical question. In answering why following moral reasons rather than, let us say, prudential ones, we can expect answers like: `Because it is rational.' This answer would be given by Singer and Gewirth. Or we could expect `Because it is in everybody's interest.', or `Because it is according to my ideals.', as would be said by Hare, or `Because it is just right.', as would be claimed by following the ideas of Davidson. But are these answers satisfactory? The rst would mean that not following moral reasons and acting according to a prudential reason would be irrational. But would we say this? If somebody acts on his prudential reasons and not on his moral reasons, is he just irrational? Is the liar who knows that he never will be discovered as a liar, and who gathers himself a great fortune by his lie, just irrational?16 It is more proper to see such actions as immoral, and they may even be judged to be wicked. Such failing to act morally is not just like making a mistake in reasoning, it is immoral, wrong, and it can even be called sometimes wicked. Acting in an immoral way is acting in a way that shows a failure to see the alternatives in a particular morally right way. It is not merely irrational. 16
See also [Barry, 1989].
2.3. MORAL REALISM
43
There are no compelling reasons to act in a morally proper way. We are not forced too, even not by reason. We should not look for a compelling reason for acting morally right, because there are no such reasons.17 In the answers given above to the question `Why act morally right rather than prudent?', it is for example not settled who are the ones I have to concern about. It is not determined de nitely by rationality who my neighbour is. Who or what belongs to the `all' in the expression `in the interest of all'? Who are to be seen as agents in the reasoning of Gewirth, who want to have control of their actions and are similar like me? What about stones, houses and animals? How to discern that they are not like me? What about people outside the circle of my family or village? What about the people beyond the group I need for my own purposes? In the view of O'Neill, presented in the previous section, the people beyond those I need might be excluded from moral concern. How should I look at them, just as if they are stones with which I cannot identify and which do not belong to the `all' in `the interest of all', and consequently do not belong to the ones who have a right to freedom and well-being? Hare does not say who are the ones to identify with. He adds that identi cation with others is stimulated, extended and intensi ed by literature, novels, plays and poetry, but it is beyond rational control. In the end it depends on a particular way of seeing the world [Hare, 1960, p. 181]. In the radical interpretation view of Davidson the issue is not settled either, but it does not pretend to. In this view, it is made clear that all depends on a particular interpretation and seeing things or persons in a particular perspective. Morality is not just reasoning, it is perceiving the world in a particular way, in a moral way. Acting not in a morally sound way is not merely acting irrational, but it is showing some sort of moral blindness or moral insensibility. To act morally properly has its roots in seeing the world in a particular moral perspective [McDowell, 1981]. For example, accepting equality as an ideal can be apprehended as seeing the world in a particular moral way, namely seeing oneself as one among others. It is not based on a deduction like reasoning from abstract universal principles. Neglecting equality is not just a mistake in reasoning, but it shows lack of a particular moral way of perceiving. By acknowledging this moral perspective for equality, equality has regained its moral status and does not loose its rational basis. The same goes for many more ideals and moral judgements. A moral judgement cannot be forced upon oneself or upon another. One has to see or discern the point. One has to see what is worthwhile to care for and what not. 17 See section 2.2.2 on similarity and also the story of Lewis Caroll of Achilles and the Tortoise in chapter 5 p.159.
44
CHAPTER 2. MORAL UNIVERSALISM
Once the importance of a moral view is recognised, the next question is of course: `How do we know which view or insight is better?' It is the question concerning the justi cation of particular moral views. It is an epistemological question. How do we arrive at moral knowledge? What is the role of perceiving the world in a particular moral perspective? How do we know that one view is more sound than another? Let me look at these questions.
2.3.4 Justi cation of moral beliefs
So far I discussed the issue of why we are moral, but I did not discuss the question: `How we know a moral belief to be true?' This question has of course to be answered in line with ideas on how we arrive at knowledge in general; in other words with epistemology in general. Two issues dominate the epistemological discussion: the foundation of knowledge and the relation between our beliefs and the external world. A well-known view with respect to the former issue is foundationalism according to which we have some foundational beliefs from which others are inferred; without the former we would not have any knowledge at all. A common idea with respect to the latter issue, the relation between our beliefs and the external world, is that we arrive at knowledge because we are caused to. The world causes us to have particular beliefs. Let me examine these ideas more precisely in order to explain the possibility of moral knowledge.
Epistemology; a coherentist view
The foundationalist answer is inspired by the view that we have inferential and non-inferential knowledge. Knowledge that is justi ed is inferred, hence inferential knowledge. The possibility of true knowledge is dependent on the possibility of non-inferential knowledge which is basic [Dancy, 1985, p. 54]. Consequently, the basic knowledge has to be self-evident infallible knowledge. Sense data are suggested as this kind of knowledge. Although it seems attractive to see knowledge in this way, there are some serious problems with this view. Once sense data are postulated as basic knowledge, we have no escape from the inner world of the subject with sense data to the outside world. How do we know what we are experiencing, how can we use these sense data if these are just internal to us and only supposed to be transparent and clear to ourselves and not to someone else? How could the subject even be sure that he uses the sense data in a right way and that he gives the right names to the sense data if they are not comprehensible to others. To these questions Wittgenstein's private language argument is illuminating.
2.3. MORAL REALISM
45
Suppose, there are mental phenomena that are principally inaccessible to others, as is assumed to hold for sense data; only rst person judgements are possible. That person should be able to recognise the phenomenon. But there is only recognition if it is not the case that a subject thinking he recognises a phenomenon implies he recognises it. But how could he recognise the private phenomenon without it being true that thinking he recognises the phenomenon implies that he recognises the phenomenon? That is not possible, because he is the only one that can recognise the phenomenon. There is nothing beyond his experience of the phenomenon and nothing beyond his thinking he recognises the phenomenon. So, his thinking he recognises it, implies his recognition of the phenomenon. Let us imagine a table (something like a dictionary) that exists only in our imagination. A dictionary can be used to justify the translation of a word X by a word Y. But are we also to call it a justi cation if such a table is to be looked up only in the imagination? - `Well, yes; then it is a subjective justi cation.' But a justi cation consists in appealing to something independent. - `But surely I can appeal from one memory to another. For example, I don't know if I have remembered the time of departure of a train right and to check it I call to mind how a page of the time-table looked. Isn't it the same here?' - No; for this process has got to produce a memory which is actually correct. If the mental image of the time-table could not itself be tested for correctness, how could it con rm the correctness of the rst memory? (As if someone were to buy several copies of the morning paper to assure himself that what it said was true.) Looking up a table in the imaginationis no more looking up a table than the image of the result of an imagined experiment is the result of an experiment. [Wittgenstein, 1945, 265](Translation G.E.M. Anscombe) Hence, private experiences could not even be recognised by the person who has them. So, mental events which are inaccessible to others are even not accessible to the person himself, because he cannot recognise them [Wittgenstein, 1945, 244]. Consequently, the sense data have to be public accessible implying the possibility that someone is erring in experiencing. Sense data cannot provide the basis of infallible knowledge that was demanded. To be clear it is not denied that perceptions can contribute to knowledge but perceptions are not infallible. Wittgenstein's private language argument points to leaving foundationalism and accept the fallibility of perceptive knowledge. A view following this suggestion is coherentism. x
x
46
CHAPTER 2. MORAL UNIVERSALISM
In foundationalism there is a strict separation between basic beliefs and inferential beliefs. The latter are derived from the former. The former are supposed not to be altered, they are settled once and for all. Of course some new beliefs can be added and as a consequence the latter can be adapted but the former, the basic ones, do not change they only expand. By adding basic beliefs, the set of basic beliefs increases monotonically and none of them is retracted, but the set of inferential beliefs can alter and is not necessarily increasing monotonically. But is this a correct view? Why not accept that perceptual knowledge is not infallible, and accept that there are no basic beliefs that are infallible and accept an increasing coherence as a justi cation for a belief? Coherentism holds that a belief is justi ed if the set of beliefs is more coherent together with the belief than without it. If by accepting a belief, the beliefs in the set of beliefs mutually support each other better than without that belief, it is justi ed [Dancy, 1985, p. 111]. In a similar way, moral beliefs can be held to be justi ed if their acceptance in a set leads to a more coherent set. We could say that what is right is best tting in our theory, i.e. it ts best in the set of moral and non-moral beliefs we have, a view which is explained and explored by Hurley [Hurley, 1989, p. 193]. But what of the relation of beliefs with the actual world? How to account for the correctness of the beliefs, how do we know it is not all a coherent illusion? The answer is simply: `It cannot.' Strictly spoken it is not impossible that we all believe a coherent illusion, but it is not comprehensible. Beliefs belong not to a transparent kind of objects installed in us, but we interpret beliefs. We interpret beliefs of others and also those of ourselves. As explained in the introduction of the radical interpretation view, in interpreting what others believe we have to assume that others see the same things as we do, aspire the same things as we do, live in the same world, etc. There have to be common truths also moral truths, without it, there could not be any interpretation. Because of this necessity it is conceptually impossible that we are all believing a common all pervasive illusion made up for example by some demon. Radical interpretation means that most people are mostly right, not because it is an optimistic point of view, but because otherwise we could not interpret each other and determine what others believed. We would be merely solipsists who did not even know what we believed for ourselves.18 In accepting a coherent view of justi cation of knowledge, moral knowledge becomes easily comprehensible. As Hurley states, we could hold that good is what is best according to the theory tting best in our set of beliefs. 18 As stated above solipsism is conceivable but not communicable [Wittgenstein, 1918, x5.62].
2.3. MORAL REALISM
47
But what about moral perception as I proposed above, what about moral knowledge, do we perceive it directly as in so-called direct realism or via indirect perception and by inference as in so-called indirect realism? This latter seems to be the idea of Hare in his view on supervenience. His view can be explained by ascribing to him the idea that we perceive natural, non-moral qualities with our sense organs and infer from these with help of the standards we learned, or we constructed, moral qualities like rude, kind, good, right, evil, worthwhile etc., but we do not arrive at them directly. So far, I did not discuss perceptions, although they seem to have a special role in the acquisition of knowledge. A common view on perceptions is that they are caused by the world. A causal theory of perception means that perceiving a cat is believing there is cat which is caused by the existence of that cat. The question is, do we perceive the cat directly or indirectly through a sort of sense datum or idea [Dancy, 1985, p. 143]. This latter is argued for by pointing to illusions and hallucinations. We can have an experience of perceiving a cat without there being a cat. Because we cannot discriminate between the two experiences, one caused by a cat and another not caused by a cat, there has to be some common representation, one caused by a cat and one caused by for example some demon who made up an illusive environment. McDowell rebuts this argument by holding that we have no reason to accept the existence of one and the same representation which can either be caused by a cat or a demon [McDowell, 1982]. He denies it is one experience and holds there to be two di erent experiences, one caused by a cat and another caused by a demon. The fact that they are not indistinguishable to us, does not imply they are similar representations, which would be a reason for holding that we perceive the world through these representations. So, indirect realism does not follow from the possibility of illusions or hallucinations. But even not noticed by Dancy, McDowell overlooked the main point of the argument. If an experience is caused by something, let us say a cat on a mat, then because of the concept of causation it could be caused by something else. An event e can only be caused by an event f if e and f are di erent events. Event e is said to be di erent from f if e and f have di erent causes or di erent e ects. This means that it is said to be caused by event f only if it is possible that it could be caused by something else than f. Otherwise, the events would not be considered as two di erent events.19 We do not 19 This is a consequence of for example Davidson's criterion for identity of events. Events are identical if they have the same causes and the same e ects [Davidson, 1980, p. 179]. If e can only be caused by f, then it will have the same causes and e ects as f. It could be objected that f could have di erent e ects from those of e, namely in case there is no perception, but this objection is not valid, because we have already e i.e. a perception, which can only be caused by f, so it will not do to assume there is no e.
48
CHAPTER 2. MORAL UNIVERSALISM
say that someone's sitting in his chair caused his sitting, or the event of perceiving the white cat on the mat caused the event of perceiving the cat. The events in a causal relation are separate and their relation might have been di erent. The e ect might have been caused by some other event than the actual cause. The representation might be caused by something else than the actual cause, and as a consequence an indirect theory has to be accepted if a causal connection was the essential relation between the world and beliefs. But this indirect perception theory is not plausible, one form of it, namely the form in which the intermediate ideas are sense data, is already rejected. The other form runs also into problems as is seen next. In the sense data view on knowledge it was held that the external world was inferred to and declared to be unobservable. It was shown to run into problems because of the assumed privacy of the experiences. There is however some other form which does not hold the external world to be unobservable, but which invokes the indirect intermediate representations as explanation of what it is to observe the world [Dancy, 1985, p. 166]. It is held that there is a sort of double awareness. One is aware of the world by virtue of the representations, similar to being aware of something via a mirror and not directly. There are some problems with this view because of the relation between the external objects and the representations.20 What is their relation, is it one of similarity? But what is this similarity, is the representation of a triangular object itself triangular? It is dicult to arm this. Turning to causation for help will not do either because the similarity will be lost, a cause is not similar to its e ect. There is another possibility along Wittgenstein's line of reasoning, which is however not explored by Dancy. The similarity is similarity in a common practice and learned in a way just as other practices are learned. But this answer shows that the intermediate representation in the observation is super uous. Because now the intermediate representations are concluded to in the practice itself in a common world, they are interpreted as beliefs in Davidson's view. This common world and practice is a necessity for the interpretation of what others perceive and believe. The intermediate representations do not have a status apart from their relation to the common world learned in becoming acquainted with the practice of perceiving. Their special role in explaining errors is lost because errors can be explained without them. Errors are simply explained by the common practice of becoming acquainted with the world itself. This practice as a basis for knowledge implies also the possibility of error. That possibility is inherent to the concept of rule following in a common practice. 20 In [McDowell, 1994] this is the main problem discussed. The solution McDowell proposes is similar to the one proposed below in the main text here. See also chapter 3 p. 103.
2.3. MORAL REALISM
49
Because we can do without intermediate representations direct perception of the world is a more plausible view than indirect perception via sense data or ideas. This view called direct realism together with coherentism can account for perceptual errors, also moral illusions. There is no longer a necessity for basic infallible knowledge. Coherentism together with direct realism is a satisfactory position in epistemology that can account for knowledge and errors. The possibility of error is explained related to the common forms of life and the common knowledge, whether it concerns moral or non-moral knowledge. That some goods are valuable and consequently to be valuable to be distributed equally, can belong to such a corpus of knowledge. The coherence of the ideal of equality with our other moral and non-moral beliefs determines whether it is reasonable to accept this ideal or not.
Williams' argument against moral perception
In the previous paragraph, moral perception was argued to be possible just like other perceptions. But there is still a famous argument against moral perception to be considered. The argument is formulated by Williams [Williams, 1985, p. 145]. Williams argues against moral perception that there is no explanation for di erences in perception resulting from di erences of position. Such an explanation should make clear that di erent observators can have di erent perceptions of the same object. Such an explanation is lacking in moral matters but available for example in the case of seeing a broken stick while the stick is not broken at all but a part of it is under water, or seeing an ellipse while it is in fact a circle. It is agreed that one perceives an unbroken stick and a circle but that it looks like a broken stick and an ellipse. The images of the broken stick and the ellipse can be explained by the unbroken stick and some optics, by the circle and some optics, respectively. Because such phenomena are lacking in moral matters moral perception is not like visual perception. Williams' argument would be refuted by an example in moral matters in which there is a di erence between `as it looks' and `as it is'. Both, `as it looks' and `as it is', are necessary elements of perception. Peacock for example called them sensation and representation respectively [Peacock, 1983]. In order to arrive at such elements in moral matters we should nd a case in which a moral judgement as it appears is caused and dependent on the way the moral judgement in fact should be. Then this appearance is tracking the true moral judgement and can be called an element of perception. Cases of indirect self-defeatingness as for example presented by Par t seem to be proper candidates [Par t, 1984, p. 5].
50
CHAPTER 2. MORAL UNIVERSALISM
For example, suppose utilitarianism is the right moral theory then it could be possible to adhere to another theory, because adherence to the latter will lead to better actions evaluated by the utilitarian view itself. In this case a judgements based on this other theory is tracking the utilitarian theory which was supposed the right one. It looks to a person as if it is an action of certain kind but in fact, on the more fundamental level, it is an action furthering the greatest good of the greatest number. Although it seems to be an example we are looking for, it is not. In the cases of visual perception, a person would not see a circle if he did not have the image of an ellipse and if one had not the sensation of a broken stick he would not have the awareness of the unbroken stick either. The perception of the object is due to this positional image. This relationship is lacking in the former example, in which utilitarianism called for another moral theory. But what about the following story? Suppose, it is agreed that morality is learned in expanding circles, rst in the family, then in the village and so on. Suppose it is by learning via these expanding moral environments that utilitarianism is learned to be the proper moral theory. Suppose further, that in order to act better in utilitarian terms, we should restrict ourselves to our nearby moral environment. Here we have a case in which the morality restricted to our own family, is based on the proper theory which is utilitarianism and this latter theory is the right one, arrived at via the restricted morality. Without the restricted theory we would not have learned that utilitarianism is the right theory. The restricted view is necessary for the utilitarian view and the latter is a ground for adhering to a restrictive one. Although this seems to be an example against Williams, it is not quite similar to the broken and the unbroken stick and the ellipse and the circle. In the latter, a person can hold simultaneously, even after explaining the distortion of his image by positional factors, that he nevertheless has the image of a broken stick. It still looks as if it is a broken stick, although actually he also knows and sees an unbroken stick. But in the moral case the person with the restrictive view cannot hold his view to be the right one if he accepts also the utilitarian view to be the right theory. Accepting one excludes the other. So, the moral case is not similar to those of visual perception. Before proceeding, let me state what the relations are between the images of the object, A, and the object, B. We can state: 1. It looks A but it is B. 2. If it were not B, it would not look A; it looks A in virtue of being B.
2.3. MORAL REALISM
51
3. If it did not look A, it would not be B; B is not recognised but due to A. 4. Even if it is recognised as B, it remains to be seen as A. Fill in for A a broken stick and for B an unbroken one or for A an ellipse and for B a circle, then we have the cases of visual perception. Consider now some agent-relative value, enjoying some friends or doing philosophy oneself.21 Now there can be a situation in which enjoying my friends con icts with others enjoying their friends, or my doing philosophy interferes with others doing philosophy. This could be a situation in which it seems that enjoying my friends and doing philosophy myself should be done, although in fact it should not, because the alternatives are more proper. The rst condition, it looks A but it is B, is satis ed. If enjoying friends or doing philosophy would not be valuable at all it would not be valuable for me either. This is the second condition, if it were not B then it would not look A. In addition, if it were not valuable for me, it would not be valuable for others either. This interdependency is inherent in agent-relative values and their agent-neutral counterparts. It is to be seen as element of the principle of charity in the radical interpretation view of Davidson. This means that the third condition, if it did not look A it would not be B, is satis ed too. The fourth condition, even if recognised as B it remains to be seen as A, is also satis ed. Even after recognising that others should not be hindered one could still hold at the same time that one should do philosophy oneself or enjoy one's friends. If it would be given up it could be given up also by all others; hence, it has to be simultaneously held that although it would be right to give up enjoying your friends and philosophy, it still looks di erent to you. We have a case which is similar to visual perception. In the example the structure of positional objectivity is displayed.22 It shows that Williams' claim that di erent positional views which can be explained in visual perception by optics, but di erent positional view cannot be explained in moral perception, is refuted. Although the explanation concerning moral perception is not necessarily physical in nature this does not a ect the function of the example.23 See for a discussion on agent-relative and agent-neutral values [Dancy, 1993]. The term positional objectivity is borrowed from Sen in [Sen, 1993]. It is argued by Putnam that the absolute view, which is assumed by Williams as the locus of explanation, was not intelligible at all [Putnam, 1992]. Putnam takes the following properties as characteristics of the absolute conception: It consists of primary qualities and it is physical in nature. It lacks value terms. There will be in the end convergence on it. (This is challenged on its turn by 21 22 23
CHAPTER 2. MORAL UNIVERSALISM
52
The con ict and the relation between the rst-person perspective and the third-person perspective in this example is mediated by a central element of the ideal of equality proposed in this thesis, namely the idea that what is valuable for me is valuable for another too. This means that the situation of being simultaneously equally as well o as possible with respect to some value, could be seen as the situation in which the con ict of perspectives is dealt with in a proper way. The ideal of equality mediates in the con ict of perspectives. One is not asked to forego what is valuable if one has less than what is available for all, but one can agree with being asked to forego what is valuable if one has more than what is available for all with respect to that value.24 The relation between the rst-person perspective and the third-person perspective is responsible for the con ict between the rst-person view and third-person view, but it suggests also a way to handle it, namely by equality in which one can accept the rstperson perspective and the third-person perspective together as in, it looks A but it is B. By this example Williams' argument against moral perception is answered. Moral perception is a comprehensible possibility within coherentism and direct realism. Errors and their persistence can be accounted for. Perspectival elements of visual perception have their analogue in morality with for example agent-relative values and their agent-neutral counterparts. So, moral perception can be seen as a way of arriving at moral knowledge. Summarising, the issue of moral knowledge is dealt with as knowledge in general. Knowledge can be accounted for by coherentism in which it is held that beliefs are justi ed if their acceptance lead to a more coherent set of beliefs, which is a set in which the beliefs support each other better, and direct realism holding that one perceives the world and the moral aspects directly. The foundationalist answer was rejected because it lead to an imRescher [Rescher, 1993]. A challenge not accepted of course by Putnam, it is one Putnam's characteristics of objectivity and truth) It contains explanations of locality, say positional objectivity. The problems with this absolute view are according to Putnam: the relation between ordinary concepts and the absolute scienti c view. Both use the same words but how should they be translated into each other? These translations are not determined (Quine). They will contain evaluations based on interests. the way this absolute world is pictured. If explanations are given in this absolute world, it will also contain counterfactuals, but these are exhibiting interests and contain evaluations, otherwise they could not be understand and the absolute world would not be plausible at all, it could certainly not give explanations. 24
See chapter 6 p. 199.
2.4. MORAL REALISM, ITS PARTICULARISM AND EQUALITY 53 plausible sense data theory which cannot account for an external world and was refuted by Wittgenstein's private language argument. Furthermore, it was argued that the causal theory of perception was implausible because it lead to an indirect perception theory of which it was shown that the intermediate representations were super uous and did not have the role they were supposed to have. Coherentism does not su er from these problems. Perceptive errors without intermediate representations can be explained. Coherentism explains how direct perception, and also moral perception is possible. In moral realism as it is presented here, perceiving the world in a moral way was proposed as an answer to the question why people are moved by particular moral reasons rather than prudential reasons. Moral realism also explained the relation between moral beliefs and actions. Moral beliefs are interpreted beliefs, in which actions are considered to be important. There is a conceptual link between moral beliefs and actions. So, moral realism developed from the radical interpretation view of Davidson, in which moral particularism is included, can be considered to be a plausible alternative to moral universalism. After this brief introduction of moral realism as an alternative to moral universalism as background assumption for an ideal of equality, it is time to evaluate what it means for the development of the ideal of equality so far.
2.4 Moral realism, its particularism and equality What is gained so far for the development of an ideal of equality by turning to moral realism as an alternative to moral universalism? First, the moral particularism of moral realism undermined the argument for an incomplete ordering regarding inequality. Although it might be possible for moral universalism to answer this argument in a formal way, moral particularism is a more radical answer. Second, moral realism answers questions concerning the foundation of morality in a more appropriate way than moral universalism. Moral judgements and ideals, including an ideal of equality, are not to be expected to be derived from a set of moral principles. There is no set of compulsive principles and reasons for a moral ideal. As explained by the coherence theory of knowledge, moral beliefs should be related in a coherent way with our other moral and non-moral beliefs. Moral beliefs and ideals have their roots in a particular moral way of perceiving. The ideal of equality has its roots in a particular way of looking at others. In interpreting what is going
54
CHAPTER 2. MORAL UNIVERSALISM
on in this world and what other people do, desire and believe, one has to assume that one lives in a common world with a lot of agreement on what is valuable. What is valuable for oneself has also assumed to be valuable for another in this one common world. In interpreting another one sees oneself as one among others. It is recognising that how well o another person is regarding a particular value matters as much as how well o oneself is. The rst-person perspective and the third-person perspective are assumed to be dependent on each other. The ideal of equality proposed in this thesis is closely related to these ideas on a particular moral way of seeing others and seeing oneself as one among others. It holds that all should be equally as well o regarding a particular value as is possible for all. Moral realism is a promising background for the ideal of equality to receive its proper place. But what about the argument against moral realism as a background for equality, because its moral particularism denies what equality arms? In the introductory chapter it was mentioned that if equality has any meaning at all then it has at least to subscribe to `like cases should be treated alike', which is subscribing to a rule.2526 This aspect of equality is said to be denied by moral particularism. So, how could moral particularism be a fundament of ethics in which equality has a role? Accepting equality and so accepting the principle `treat like cases alike' is contrary to moral particularism. Moral particularism and equality cannot go together, thus the argument. Although it seems to be a plausible argument it is not valid. Accepting equality is not accepting moral universalism and in moral particularism it is not denied that `like cases should be treated alike'. Moral particularism was introduced as the denial of moral universalism and moral universalism was the view that the reason for something being good or being right is to be found in rules. It would hold that one should accept `like cases should be treated alike' because that is what rules mean. But this is not the essential meaning of equality. Just using rules and treating like cases alike is not acting on behalf of the principle of equality. Very often this has nothing to do with equality at all. For example, if a physician treats a patient su ering a trichophyte between his toes with a cream containing miconazol and he treats every patient su ering the same disease with the same cream, he does not treat them alike because of the ideal of equality. He treats them alike because they su er the same pathology, a trichophyte, for which the same treatment is e ective. It would be di erent if there was scarcity of antimycotic therapeutics, but this scarcity does not play a signi cant role in decisions on therapies for See chapter 1 p. 6. See also de Beus in [De Beus, 1993] who developsan ideal of equality from procedural equality with the help of humanism. 25 26
2.4. MORAL REALISM, ITS PARTICULARISM AND EQUALITY 55 rather simple diseases like a trichophyte. So, treating like cases alike is not the same as acting on behalf of the ideal of equality. The foregoing example does not show that equality does not imply moral universalism, it is merely argued that they are not the same. But the example suggests what the crucial di erence is between them. Moral universalism was a basis for moral judgements, not a moral one, but a rational one. In moral universalism it is impossible to deny the principle `treat like cases alike', it is even seen as the basis of all judgements. But if the ideal of equality is interpreted as moral universalism it has been stripped from its moral aspect. Something would not be any longer judged according to the moral principle of equality, but just because of what making a judgement means. So, equality interpreted as moral universalism does lack an essential aspect, namely the moral aspect. This moral aspect is given back to the ideal of equality by moral realism by pointing to the moral way of looking at oneself as one among others as the basis for the ideal of equality. But what about the idea that equality implies moral universalism, is this a correct idea? No, because accepting equality is not holding that something is good because of such and such a rule, as is claimed in moral universalism. It will mean that something is right or wrong for example because it exhibits equality or inequality. This is how moral ideals are used, or how they are expressed. Equality does not imply moral universalism. The idea that equality implies moral universalism is likely based on the belief that if a situation is in accord with the principle of equality then it is also in accord with moral universalism, as in every model satisfying equality is also satisfying moral universalism means equality implies moral universalism. But this reasoning missed the essence of moral universalism and the essence of the principle of equality. Satisfying rules is not the same as being good because of those rules. The idea of `because of' is not captured by the idea of a model satisfying some theory.27 Following equality is merely apparently like accepting moral universalism. The reasons for judgements are di erent. In judging something to be good with respect to equality, one is judging the equality exhibited to be good because it is good. In moral universalism something is judged to be good on behalf of a rule. The argument that equality is inconsistent with moral particularism is presumably based on the wrong assumption that within moral particularism no moral rules can be accepted. If that was a proper assumption, equality indeed would be inconsistent with moral particularism. But rules can be accepted within moral particularism, but they are not to be considered as 27 This is illustrate also by the awkward dictator in Arrow's theorem, who can lack any in uence on the social decision function, but who just happens to be in accord with it. See chapter 3 p. 66.
CHAPTER 2. MORAL UNIVERSALISM
56
the foundation of moral judgements. The fundament of something being right or wrong is just its being right or wrong. That something is wrong because it is not according the ideal of equality does not mean it is wrong because the rule `treat similar cases similarly' is violated, but because it is violated in a particular way, in an inegalitarian way, namely that some are worse o than all could have been simultaneously. The crucial di erence between moral universalism and moral particularism is not whether or not there are some rules that could be followed in order to be sure to make proper judgements, the issue concerns the status of those rules. Moral universalism claims that rules are the ground for something being good, and moral particularism denies just this. Within moral particularism the supervenience of right and wrong on other non-moral aspects is not fundamental but secondary, it is considered as happening that it is supervenient on those aspects.28 So, there is nothing wrong in accepting equality within moral particularism. Summarising, it is shown that moral particularism is not inconsistent with the ideal of equality. Contrary to what the argument against moral realism because of its moral particularism states, moral realism gives back the moral status to the ideal of equality which was covered over by the idea of what judgements mean in moral universalism. In moral realism the roots of equality are be found in the moral way of looking at oneself as one among others. Moral particularism undermines the argument for the incompleteness of the ordering regarding the moral seriousness of inequality. So vis-a-vis moral universalism moral realism is seen to be a promising alternative as a framework for an ideal of equality.
2.5 Summary In this chapter, I introduced moral realism as an alternative to moral universalism as background for an ideal of equality. The reasons for moral universalism that hold that rules are necessary for moral judgements and form the ground for moral judgements were discussed. First, four ways of reasoning with moral universalism as a foundation for moral reasoning were discussed. The reasons for establishing some objective ground for moral reasoning appeared to be not convincing, because of the dependency of moral reasoning on subjective evaluation as in Hare's universalistic prescriptivism and in Singer's use of generalisation argument, or because of the dependency on subjective maxims as in O'Neill's Kantian ethics, or because the unsolved moral con icts in Gewirth's good reason approach. Although the reasons for establishing a sound moral basis failed, their views 28
See also the discussion on supervenience in chapter 5 p. 170.
2.5. SUMMARY
57
on equality were not rejected. The idea, emphasised by Hare and Gewirth, that what is a reason for one is a reason for another too, is an element of the ideal of equality which is developed in this thesis. The idea seen in the approach of Singer and O'Neill, namely equal collective availability of some good for all, is also a central element in the ideal of equality proposed in this thesis. Subsequently, it was shown that the reasons for moral universalism based on supervenience and the concept of reason were dependent on the concept of similarity. This concept led by following Wittgenstein's ideas on `similarity' and `to follow a rule' directly to particularism. Rules do not have any meaning beyond human practices. The Davidson's radical interpretation view on language was introduced. Three questions were discussed with the help of which moral realism was contrasted with moral universalism in order to illuminate the former. In moral realism, the question why people are moved by moral beliefs at all is answered by an internal conceptual link between beliefs, including moral beliefs and actions. Beliefs and also moral beliefs, have to be interpreted; they are not interpreted apart from desires and actions. In discussing the question why people are moved by a particular moral reason rather than a prudential one, it was explained that in moral realism particular moral beliefs come about by perceiving the world in a particular moral way. In moral realism it is held that not acting according to morality is not merely irrational but can be immoral and bad. Finally, it was shown how the epistemological questions concerning moral knowledge were answered in moral realism. Instead of foundationalism, which was argued to lead to the problem of infallible knowledge, coherentism is accepted in which a belief is justi ed if the set of beliefs is more coherent with that belief than without it. Direct realism, the view that we are perceiving the world directly and not through some sort of mediating images, provides satisfying answers to problems concerning knowledge by perception. It was argued that with coherentism and direct realism we can account for illusions and persistent misperceptions also on moral matters. The argument of Williams against the possibility of moral perception was answered by an example exhibiting interdepency of the rst-person perspective and a third-person perspective. In this chapter, moral realism and its related moral particularism was introduced as an alternative to moral universalism as a background for an ideal of equality. By this alternative the argument for the incompleteness of the ordering re ecting the moral seriousness of the deviation from the ideal of equality was undermined. It opened up the possibility of a complete ordering in a distribution problem. It was argued that moral particularism is not inconsistent with equality. It was shown that equality does not imply moral universalism, and particularism, although it denies the status of rules
58
CHAPTER 2. MORAL UNIVERSALISM
as necessary for the foundation of moral judgements, it does not deny the possibility of rules representing our moral judgements. It was argued that moral realism underlines the moral aspect to the ideal of equality. In moral realism the roots of equality lie in a moral way of perceiving this world and looking at oneself as one among others. The ideal of equality proposed in this thesis, in which how well o all simultaneously could have been with respect to some value determines what can be asked for in the name of equality, is related to this way of perceiving. Moral realism explains this way of perceiving in which the mutual dependency of the rst-person perspective and the third-person perspective is acknowledged. In the following chapters, the other traditional background assumptions of the modern idea of equality, volitional individualism and moral value pluralism will be discussed. In these chapters the further consequences of moral realism for equality will be explored.
2.6. APPENDIX 1
59
2.6 Appendix 1 It can be proved that if a moral theory leads to action-guiding judgements in a simple form that can be formulated in a quite usual formal language, i.e. rst order predicate logic, and it happens to be that in each situation similar actions are judged similarly and in similar situations one judges actions similarly because of supervenience, than there is a nite set of action guiding rules which prescribe those actions which are held to be good according to that morality. The proof is based on the use of the compactness theorem for rst order predicate logic, and is borrowed from an argument given by G. Hellman with respect to some ideas about the mind/body-relation.29 With a morality is meant a body of knowledge according to which actions in their so-called basic action descriptions are judged to be good in particular situations. Basic action descriptions are such that those descriptions cannot be further analysed by other actions, they are performed directly. For example, switching on the light is not a basic action description but moving your ngers in a particular manner such that that the light will be switched on, is. Action descriptions are seen by Davidson as predicates on actions. For example Jones buttered the toast in the bathroom with a knife at midnight. is analysed as: x(Butter(x; jones; toast) In(x; bathroom) With(x; knife) At(x; midnight)) [Davidson, 1980, p. 105]. 9
^
^
^
^
The statement that in each situation S in which a moral theory is applicable, which can be described by T, there is an indication of a good action, can be interpreted as: in every situation S there is a local (situation bounded) de nition of the predicate GOOD() in terms of a basic action description (). So in those situations S in which T is satis ed S = T, there is a local de nition of GOOD() satis ed too: S = x(GOOD(x) (x)). So, we have a list, i
i
i
i j
8
i j
i
j
S1 = x(GOOD(x) (x)) S2 = x(GOOD(x) (x)) S3 = x(GOOD(x) (x)) .. . j
8
j
j
8
k
j
8
l
Because there are about in nite situations S , it is possible that there are in nite local de nitions x(GOOD(x) (x)) abbreviated by of which at least one is true in a model for T. i
8
i
i
29 The argument of Hellman is elaborated in a syllabus by J. van Benthem on Model theory.
CHAPTER 2. MORAL UNIVERSALISM
60
It will be shown that there is a nite set of basic action descriptions which is coextensive with GOOD(). This can be shown with the help of the compactness theorem. Subsequently, it will be shown that because of supervenience there is a nite set of action-guiding principles such that acting according them will be acting morally sound. It has to be shown that there is a nite set of formulas, say 1; 2; 3; (n N) for which T = 1 2 3 : Suppose, there is not such a nite set then for every nite subset of the in nite set of local de nitions it is valid that 1 ; 2; 3; T= 1 2 3 : f
j
_
f
6j
_
_
__
ng
2
n
ng
_
__
n
Hence, it is not true that in every situation in which T is valid such a nite disjunction is valid. Therefore, there is a situation in which T is valid and ( 1 2 3 ) is valid. There is a model for T ; . And there is such a model for every nite 1; 2; 2; subset of local de nitions, because we assumed there was not a nite subset of local de nitions de ning globally what is good. Consequently there is a model for T ; ; ; (= the in nite set of local de nitions). This model exists because of the compactness theorem: A set of formulas R has a model i every nite subset of R has a model. But now there is a model in which no local de nition is true, which is contrary to what was assumed. So far, it has been shown that a nite set of basic action description will be sucient for determining the morally correct actions. Now it will be shown with a similar argument using the compactness theorem that there is a nite set of action-guiding rules such that following them will result in acting morally correct. As was shown in sections 2.2.2 it happened to be the case that in situations that are similar in relevant aspects one should perform the same actions. These relevant aspects of the situation can be seen as properties of actions. The rightness of an action parallels its properties. This might be called supervenience of rightness of an action on its properties. I suggest the following formalisation. In every situation S in which morality is applicable, in which T is valid, S = T, for every action x which has the following set of properties C 1(x); C 2(x); it is valid that (x) and you should do x. So, we have the following list of rules: :
[ f:
:
_
:
[ f:
_
i
:
:
j
:
_ _
n
ng
k
g
f
i
i
i j
f
i j;
i j
g
i
2.6. APPENDIX 1 T T .. . T T .. . T T .. .
61
C11 1(x); C11 2(x); C11 3(x); C21 1(x); C21 2(x); C21 3(x);
[f
;
[f
;
;
;
= 1(x) GOOD(x) = 1(x) GOOD(x)
;
g j
;
g j
C12 1(x); C12 2(x); C12 3(x); C22 1(x); C22 2(x); C22 3(x);
= 2(x) GOOD(x) = 2(x) GOOD(x)
[f
;
;
;
g j
[f
;
;
;
g j
C1 1(x); C1 2(x); C1 3(x); C2 1(x); C2 2(x); C2 3(x);
[f [f
n ; n ;
n ; n ;
n ; n ;
= (x) GOOD(x) = (x) GOOD(x)
g j
n
g j
n
Now I have to show that this list of action-guiding principles is a nite list. The set of properties in an antecedent is nite because of compactness of rst order predicate logic. So, for each rule there is a nite set of properties C 1(x); ; C (x) such that: T C 1(x); ; C (x) = (x) GOOD(x) Now again with the use of compactness I show there is a nite list of those action-guiding rules. Because I stated that GOOD() is coextensive with some non-moral properties, one of the list of properties has to be true of a good action. So we can state for each model satisfying T; (x); GOOD(x) one of the properties C (x)meaning C 1(x) C (x) has to be true. With the same argument with compactness as above there is for each equivalence a nite set of C (x)'s. Because of that, there is a nite list of actionguiding rules which covers the appropriate actions according moral theory T of the form: IF C (x) THEN PERFORM (x) WHICH IS (x) ! exactly what was to be shown. Consequently, the discussion between moral universalism and moral particularism is not about the possibility of a nite list of action-guiding rules, both can accept it. The crucial issue is the status of moral rules. In moral universalism these are the reasons for evaluating an action as right or good. In moral particularism an action is evaluated as good because it is good. This reasoning outlined above can be extended to the comparative form of rules. can mean the description of the alternatives to be compared, (x) could mean: a distribution x being a instead of b and GOOD() can be taken the alternative to be chosen, or BETTER or EQUIVALENT ALTERNATIVE(). If we take two alternative kinds of distributions a and b which can be ordered vis-a-vis each other in each situation, meaning that one of the two is allowed to be chosen, then the conditions under which a has to be chosen and those under which b has to be chosen, can be speci ed by a nite list of predicates. f
[f
i j;
i j;
i j;l i j;l
g
g j
i
^
i
i j
i j;
^ ^
i j
i j
i
i j;l
62
CHAPTER 2. MORAL UNIVERSALISM
Chapter 3
Volitional individualism 3.1 Introduction In the previous chapter, moral realism was introduced as an alternative to moral universalism, the rst of the three traditional background assumptions of equality. In this chapter realistic individualism is introduced as an alternative to volitional individualism, the second of the traditional background assumptions. Volitional individualismrefers to the idea that people are seen as persons with individual wants, desires and passions of an unrestricted diversity originating from a volitional part of persons, which should be the basis of the evaluation of allocations.1 The alternative introduced here, realistic individualism, is the view that people have individual wants, desires and passions, which indicate what a ects individuals and which should be the basis of the evaluations of allocations. The di erence between realistic individualism and volitional individualism is, that in realistic individualism these wants, desires and passions are not considered as originating from a well of volitions. Their content should be determined by interpretations and is restricted by our way of interpreting. In moral realism which was introduced in the previous chapter, it was explained that what is seen or recognised to be desired is dependent on what is desirable. Realistic individualism is introduced in this chapter because it allows and explains the possibility of interpersonal comparisons of how well o people are, while this poses a serious diculty for volitional individualism.2 1 The taste model of Grin is an element of this volitional individualism [Grin, 1991]. 2 It will turn out in chapter 6 that realistic individualism enables the development of an ideal of equality to which an important argument, the so-called levelling down
63
64
CHAPTER 3. VOLITIONAL INDIVIDUALISM
This chapter consists of three parts. In the rst part, I formulate a dilemma for equality, namely that it needs the possibility of interpersonal comparisons of how well o people are while these comparisons are impossible. In the second part, the rst horn of the dilemma, the necessity, is challenged. Two proposals are discussed which attempt to solve distribution problems in an egalitarian spirit without employing interpersonal comparisons. They are the underlying ideas of several economists who argue for a more egalitarian redistribution that is consistent with the Pareto principle. This Pareto principle, re ecting volitional individualism, holds that two alternatives are equally good if everyone is indi erent between them, and the rst of two alternatives better than the second if everyone either prefers the rst to the second or is indi erent between the two and someone de nitely prefers the rst [Broome, 1991, p. 152]. Decisions that do not con ict with this principle are decisions to which all could consent, it represents the idea of freedom of choice. Therefore, the proposals for redistributions consistent with the Pareto principle are seen as ways of resolving the common stated con ict between equality and liberty. The rst proposal that is discussed is based on ideas of Hochman and Rodgers. It is proposed that distributionproblems can be dealt with by taking external e ects of goods and commodities into account. These commodities are not held to be important merely for the individual who enjoy these commodities for himself, but they are also of in uence through the well-being of others. It is acknowledged that distributions a ect persons beyond what all receive for themselves. However, it appears that this approach presumes interpersonal comparisons in order to be intelligible. The second proposal discussed is based on the idea that a morally proper distribution is such that no one has a complaint that another person is better o in a way such that he would like to receive what the other person receives in a distribution. It is known under the name of the non-envy approach. It is discussed rather extensively because it is a complex and ingenious approach about which there is some misunderstanding. Although it may be super uous for those familiar with the non-envy approach, I guess, it is illuminating for those who are not trained in economic reasoning. I hope those too will be able to understand what the essentials of this second proposal are. The discussion concerns several non-envy analyses of Varian, Pazner and Schmeidler, Dworkin, and the criticism of Roemer on them. Dworkin's proposal, a market of insurances and lotteries, points to a serious threat to the whole project of this thesis, which is concerned with the evaluation of a distribution without taking into consideration the objection does not apply.
3.2. THE PROBLEM OF INTERPERSONAL COMPARISONS
65
way it came into existence. This issue arises because the evaluations of distributions with regard to equality seem to be dependent on a history of individual choices. In the next chapter, after the introduction of moral value pluralism, this question about the dependency of judgements of equality on responsibility is discussed further. Here I show that the non-envy proposals su er from problems because of the attempts to propose an egalitarian ideal without interpersonal comparisons. The non-envy approach is not criticised in itself, but it is argued that it runs into problems in so far it remains connected with volitional individualism without interpersonal comparisons. The challenge of the rst horn of the dilemma, the necessity of interpersonal comparisons, turns out not to be successful. Finally, in the third part of this chapter the second horn of the dilemma for equality, the impossibility of interpersonal comparisons, is dealt with. Realistic individualism, which can account for interpersonal comparisons, is proposed as an alternative to volitional individualism. It is based on the idea already explained in the previous chapter in the introduction of moral realism that the content of desires and wants has to be interpreted. This successful challenge of the second horn solves the dilemma for equality.
3.2 The problem of interpersonal comparisons The problem of interpersonal comparisons is that it is argued that they are necessary for an ideal of equality, and it is argued that they are impossible. I turn to the arguments subsequently.
3.2.1 Necessity of interpersonal comparisons
The argument for the necessity of interpersonal comparisons of how well o people are is rather simple. Statements concerning equality are not comprehensible without interpersonal comparisons. If someone has a certain amount of a good that is equal to the amount someone else has, then it is inevitable that some kind of comparison of the former person to the latter concerning that good, the equalisandum, has to be made. It simply belongs to the meaning of `an equal amount'. It is similarly with expressions like `less' or `more'. It is just because of what `equal', `less' and `more' mean, that some sort of comparison is necessary. They are comparatives. It will be clear that this necessity will not be denied by anybody. But what is denied is that there is a common evaluation or a common ground for these comparisons. It is not denied that each person can make such comparisons for himself, but persons can disagree on such comparisons. It is possible that some hold that A is more than B and some hold that B is more than
66
CHAPTER 3. VOLITIONAL INDIVIDUALISM
A. Some hold that Jane getting eardrops for a light external otitis is getting more than John receiving some ointment with menthol for itchy patches of his skin, but others can judge di erently. So, it is not the impossibility of judgements but the impossibility of common judgements, that is the issue. But why are such common judgements necessary? In an article Why ethical measures of inequality need interpersonal comparisons, written by Hammond, we would expect, because of its title, to nd an answer. Although as is explained shortly, the article does not give a convincing answer, the problem becomes clear in proper way. Hammond shows that the absence of interpersonal comparisons does imply the existence of a dictator as is described in Arrow's impossibility theorem. A dictator in this sense is a person whose preferences about whatever set of alternatives, are identical to those of society [Arrow, 1951]. The assumptions made by Hammond are: 1. there is some function, call it a utility function, from goods to the equalisandum for each individual such that one is better o if one receives more of some good and this utility function is dependent on what oneself receives and not on what another person receives, 2. based on such utility functions there is some complete social ordering on allocations dependent on these functions (Because it is a social ordering re ecting a social judgement integrating individual judgements, this ordering is called an `ethical measure' of inequality.), 3. this social ordering is consistent with the Pareto principle, which means that two alternatives are equally good if everyone is indi erent between them, and the rst of two alternatives is better than the second, if everyone either prefers the rst to the second or is indi erent between the two and someone de nitely prefers the rst. Accepting these assumptions lead by the reasoning similar to that in Arrow's impossibility theorem to the conclusion that if interpersonal comparability is excluded, then there is some dictator. It seems that we have here a reason why ethical measures of inequality need interpersonal comparisons. But, as explained in the following paragraphs, it is not a very convincing argument.3 3 It could be argued that a dictator in the sense of Arrow is not disastrous. This dictator happens to have an ordering similar to the social ordering, but it is not necessarily in uencing or causing this ordering. It might be that we know there is some dictator, as a conclusion of Arrow's theorem, but that nobody knows who the dictator is. Even the dictator himself can be ignorant of being a dictator. Arrow's result can also be interpreted as a hopeful and optimistic result. At least one person has the same ordering as the social ordering, which is welcome of course if this social ordering is precisely the ordering we are looking for.
3.2. THE PROBLEM OF INTERPERSONAL COMPARISONS
67
In this chapter, one proposal is discussed in which the rst assumption is denied and in which the utility function is dependent also on what another has. In this proposal so-called external e ects of goods are taken into account. It is based on ideas of Hochman and Rodgers. The second assumption is questioned too. A proposal in which this assumption is denied is discussed; it is the so-called non-envy analysis. As Varian explains, this analysis is not meant to give a complete social ordering but merely one acceptable allocation [Varian, 1974, p. 65].4 Finally, the third assumption, the Pareto principle, can be questioned. The idea behind this principle, i.e. volitional individualism is rejected and realistic individualism is introduced. But in contrast to the rejection of the former two assumptions, it is not suggested as a method of circumventing interpersonal comparisons. Realistic individualism shows that these comparisons are possible. Hence, the argument based on the implication of the existence of an Arrowian dictator is not as convincing as it was assumed to be. On the other hand it shows the problem clearly by pointing to the similarities between the problem concerning interpersonal comparability and the problem of integrating di erent preferences and evaluations of people into one social ordering and arriving at a common simple measure of inequality. We cannot expect that there is one acceptable measure of inequality if it makes no sense to say that one has more than another. This way of stating the problem is already suggesting the answer which is given in the previous chapter by moral realism. In moral realism, judgements are true or false beyond our opinions and preferences. It is argued in this chapter that indeed the Wittgenstein-Davidson approach to language and interpretation, which supported moral realism, also supports realistic individualism and can account for interpersonal comparisons of how well o people are. Let me turn rst to the other part of the problem of interpersonal comparability, namely its impossibility.
3.2.2 Impossibility of interpersonal comparisons
That interpersonal comparisons are not merely necessary but do form indeed an important problem will be clear if one turns to questions concerning the distribution of income to which economists often restrict themselves in discussions on the issue of equality. Income may seem to be a clear concept and consequently a comparison of income may seem to be simple. But some 4 By both proposals, taking external e ects into account, and the non-envy analysis, the condition of separability of the social ordering on how others fair (one of the assumptions of Hammond [Hammond, 1976, p. 73]) is not satis ed, which is in a sense not a surprise for concerns regarding equality. See also [Broome, 1989].
68
CHAPTER 3. VOLITIONAL INDIVIDUALISM
second thoughts about such comparisons point to a series of questions one has to answer. Who are the income recipients, individuals, families, households? What is meant with income, only primary income, consisting of incomes that are earned from productive activity, or secondary income which is primary income and direct cash transfers such as taxes and social security payments, or tertiary income which is secondary income together with the modi cations by public sector activities such as regulations of prices and the provision of public goods? Furthermore, what period one is looking at, is it one month, or a whole life, or 5 years? And once these questions are answered one should think about measuring incomes. For example not all goods are begotten from productive activity but may be the result of gifts, or from services and goods which are not traded on the market etc. All this information is dicult to recover [De Beer, 1993] [Sundrum, 1990] [Tinbergen, 1979]. So income seemed to be a clear and distinct concept but a little re ection shows it is not clear at all. The aforementioned questions on income are not merely of taxonomical interest for comparative statistical research. They re ect the idea that income as such is not relevant at all, but that what income means for people is relevant. Consequently, this meaning should count in comparing incomes. A certain amount of money means something di erent if it belongs to one individual, or a family, or even an extended family. The questions concerning the recipient, the time span and the di erences between primary, secondary and tertiary income re ect that what is to be measured is not income itself but what it means to people. In general, one can say goods are relevant because they do something to people. What a good means to people is relevant and not the good itself. [Sen, 1980] [Cohen, 1989]. And precisely because of that, diculties with interpersonal comparisons arise. On the other hand it should be noted here that not only the idea that goods do something to people is a problem for comparisons and so for equality, but its counterpart, that people do something with goods, is also a problem for the ideal of equality. This latter problem is postponed to the next chapter, in which the role of responsibility in the equalisandum is discussed. Here the problem of interpersonal comparisons is examined. That goods do something to people is a principal part of the problem concerning interpersonal comparisons, because what these goods do to different people has to be compared. How to measure what one good does to person A and satis es him, and how to nd out whether this is more or less than a di erent amount of the same good does to another person B? Robbins who is held to be one of the rst authors starting the discussion on this subject, argued that such comparisons cannot be made scienti cally, they are not empirical or objective.
3.2. THE PROBLEM OF INTERPERSONAL COMPARISONS
69
There is no means of testing the magnitude of A's satisfaction as compared with B's. [Robbins, 1948, p. 140] Tinbergen makes the problem clear with the following example. Suppose we have two persons, a sailor and a shoemaker. The sailor earns f 20,{ a week and the shoemaker f 30,{ a week. Suppose we want to make these people equally happy and we agree that the sailor may earn a little bit more. Each week we increase his wages with one guilder and ask him to warn us if he is as happy as the shoemaker is. Tinbergen remarks that we cannot expect to get this warning if the sailor is wise [Tinbergen, 1953, p. 58]. He concludes that we lack a way of making comparisons. Both views can be interpreted as based on volitional individualism in which what goods mean to people is determined by their volitional part. It is dependent on their wants, desires and passions which spring from a mysterious well of volitions that is beyond any restriction, with the consequence that the satisfaction of wants and desires of di erent people cannot be compared in a scienti c way. Besides the problems of interpersonal comparability which are necessary for an ideal of equality, this idea of Robbins and Tinbergen, taken as inspired by volitional individualism, means for example also that we cannot say in a scienti c way that there is economic growth. If we cannot compare the satisfactions of two people who are neighbours, we certainly cannot compare the lives of people living at di erent times. Consequently, we can never point to some political action or decision as necessary for economic growth, because the content of the latter cannot be determined scienti cally. To be clear, Robbins did not deny interpersonal comparisons to be possible but denied them a scienti c objective status. Robbins considered such comparisons as conventional or evaluative judgements. In a later section I return to the issue of the status of interpersonal comparisons. But whatever the status of interpersonal comparisons, whether it is evaluative or scienti c, the ideas of Robbins and Tinbergen in 1953 show that it might be convenient if distribution problems could be solved without such comparisons, because there seems to be no simple means available of testing the magnitude of satisfaction if volitional individualism is accepted. In the next section, two proposals are discussed.5 5 Below it will become clear that the problem is not interpersonal comparisons of what goods do mean to people per se, but the assumption that there is one general good for each person which is most important in all situations, which can be recognised as the assumption of moral value monism and moral universalism.
70
CHAPTER 3. VOLITIONAL INDIVIDUALISM
3.3 Necessity challenged The argument of Hammond for the necessity of interpersonal comparisons can be challenged in two ways. As was mentioned above, it can be opposed by admitting external e ects and by suggesting that only one alternative should be socially decided to, which does not require a complete social ordering. These suggest two ways of circumventing interpersonal comparisons and still endorse egalitarian reasons for redistributions. Both ways can be recognised in the public debate [Hartog, 1981] [Hennipman, 1981] [Dworkin, 1994]. They are interesting because their evaluation is consistent with the Pareto principle. According to the Pareto principle two alternatives are equally good if everyone is indi erent between them, and the rst of two alternatives is better than the second if everyone either prefers the rst of the two alternatives to the second or is indi erent between the two and someone de nitely prefers the rst. Within volitional individualism it is held that everyone uses his own evaluation of the allocation according to his own preferences, whatever they are. This suggests that the tension is decreased between liberty, each evaluates distributions according to his own preferences, a change of an allocation is not accepted if one person dislikes this change, and the idea behind equality, nobody has a complaint. It is suggested that the free competitive market is a way of arriving at such acceptable distributions. So, it will be illuminating to discuss these proposals. One of the two proposals uses the idea that commodities have external e ects and the other uses the idea of placing oneself in the position of another by imagining that one has the commodities of another person. These are discussed subsequently.
3.3.1 External e ects as reasons for redistribution
It is commonly assumed in presentations of evaluations of allocations of goods, that a redistribution in the sense of a transfer of one good from one person to another person is beyond the scope of Pareto like evaluations. Allocations are evaluated as Pareto better if it can be said that no person is worse o and at least one person is better o , each one using his own wants and desires for determining what is better for himself. Pareto optimal allocations are those allocations such that there is no allocation in which at least one is better o without someone else being worse o . In these evaluations it is commonly held that how well o one is, is dependent only on the goods one has for oneself and not on the goods another has. Because of this assumption transfers of a good from one person to another cannot be considered as Pareto better or worse. A transfer of a certain amount of good ak from R(ich) to P (oor) will have as a consequence that R will have
3.3. NECESSITY CHALLENGED
71
less ak and so will be worse o than before the transfer and P will be better o . Consequently, the allocations before and after the transfer cannot be ordered by the Pareto principle. It can be di erent if one admits that the goods someone else has a ects one's own satisfaction. It is likely that other person's incomes a ect how well o one is. For example it is possible that one has sympathy for another so the goods another has will a ect how much one enjoys the goods. Less altruistic motives are possible too, for instance if one does not like beggars around one's garden, and hates to see paupers, then this can be a reason for a transfer of goods. In international a airs it is suggested, for example by Tinbergen, that developmental aid to less developed countries should be given in order to stop persons coming from those countries to the more developed countries [Tinbergen, 1991].6 If one takes external e ects of goods into consideration, then it can happen that a transfer leads to an improvement consistent with the Pareto principle. To be more explicit. Suppose that how well o someone is, is represented by wr which is not merely dependent on his own commodity bundle, ~ar (of which akr is a commodity ak belonging to r) but also on the bundles others have, which can be represented by a matrix ~a1~a2::~ar ::: = A. Instead of wr (~ar ) we have wr (A). It can happen that a di erent allocation of bundles, say A0, is Pareto superior to the original allocation as a result of a transfer. For instance if the well-being of person R, wr , increases if for example the income of the poorest 10 % increases, then a transfer of some part of his own income to these poor will be a Pareto improvement if the loss of well-being because of his loosing income is less than the gain because of the poor getting more. In general if dwr (A)=dakr < dwr (A)=dakp . A transfer from R to P will be a Pareto improvement, at least if it is assumed that dwp (A)=dakp > 0 which is, I guess, not the most dubious assumption. This idea of taking the external e ects of income, or commodities, into account is proposed as a proper way of solving distribution problems by Hochman and Rodgers, Hartog, Hennipman and recently by van Wijck [Hochman & Rodgers, 1969] [Hochman & Rodgers, 1974] [Hartog, 1981] [Hennipman, 1981] [Van Wijck, 1992]. 6 External e ects can be based, as is exposed in the main text, on the so-called micro judgements, what others have because of what they do or are. But external e ects are also seen in so-called macro judgements, the ideas on which distribution of goods one generally would like to see. This distinction between macro and micro judgements concerning distributions is commonly made in social psychological studies. In these they are seen to be more or less conceptually independent and merely correlated as in [Van Wijck, 1992]. But as will be clear on a little re ection, those judgements cannot be independent, and micro judgements will determine the macro judgements. I will not distinguish these two kinds of judgements.
CHAPTER 3. VOLITIONAL INDIVIDUALISM
72
This idea is criticised among others by Mishan because it will not necessarily lead to equal allocations in which all have the same commodity bundle [Hochman & Rodgers, 1974]. It could even lead to very unequal distributions, for example if one is envious to someone who has sympathy for you such that dwi(A)=dakj < 0 and dwj (A)=dakj < dj (A)=daki and aki > akj . This argument is rebutted by Hochman and Rodgers. They argue that it merely shows that the commodity bundles on their own are not important for the evaluation of a distribution, but the well-being related to that income is relevant. So, there is nothing wrong or curious with this unequal distribution it even shows the plausibility of their analysis according to Hochman and Rodgers [Hochman & Rodgers, 1974]. The above stated method of taking external e ects into account seems to be a proper way of circumventing the problem of interpersonal comparisons. There is now a reason for redistributions and transfers without using such comparisons, which is based on just the wants, desires and passions of each individual. However, upon further re ection it appears that the interpersonal comparisons are not avoided at all. They are used in an implicit way in taking external e ects into account. Just as Hochman and Rodgers argue against Mishan, indeed the commodities in themselves are not important but what they do to people. Precisely that makes clear that the method suggested by them is not a way out of the problem of interpersonal comparisons. This becomes clear if one turns to the explanation of those external e ects of goods. The explanation of the external e ects is not that goods that belong to others do a ect the well-being of somebody, but that what those goods mean to those others a ect one's own well-being. Sympathy for example, does not mean that one wants the other person to have more of the goods, but that the other person will have more of those other goods because one believes that the other is better o with more of those goods. Similarly, disliking the behaviour of paupers and beggars can be a reason for a transfer of some goods to them because of what those goods do to them. If they become better o and have enough they will have no longer an incentive to beg and hang around your garden. One wants to induce some change in behaviour which is mediated by what those goods do to other people. One does not just want them to have more of some commodities. What goods do to people has to be known and compared in order to explain those external e ects of goods otherwise, the external e ects remain void and without any intelligibility.7 Knowing what another wants, desires and prefers, means also that one can compare this with wants, desires and 7
See also for example [Hurley, 1989, p. 84-101].
3.3. NECESSITY CHALLENGED
73
preferences of someone else or of oneself.8 If one could not, then a decision on who should receive more could not be answered at all. Within volitional individualism in which wants, desires and preferences are unrestricted and have their origin in a volitional part of persons which is beyond the domain of knowledge, these external e ects cannot be explained.9 It has to be assumed that what the goods mean to other people, seen from their point of view, can be known for the explanation of the external e ects of goods. These external e ects of goods and commodities assume the possibility of interpersonal comparisons. It is part of knowledge of what goods do to people, i.e. what people want and prefer. The analysis cannot do without it. It is not a way out of the problem of interpersonal comparisons. To be fair to Hochman and Rodgers, it was not their aim to circumvent the problem of interpersonal comparability. But some economists, for example Hennipman, and Hartog, to mention just two Dutch authors who deny the possibility of interpersonal comparisons advocate this analysis as a way of circumventing interpersonal comparisons [Hennipman, 1981] [Hartog, 1981]. In the previous paragraphs, it was argued that in order to explain external e ects one has to assume that interpersonal comparisons are possible. Nevertheless, they might still be taken as subjective evaluations. It is not yet shown that the explanation assumes that interpersonal comparisons that can be made in an empirical way. One can accept subjective interpersonal comparisons which can be formulated as wi being dependent on wj and wh of other persons j and h. It seems that the analysis can avoid the assumption of objective interpersonal comparisons. But this is merely appearance, in fact it does not avoid interpersonal comparisons that reach beyond the subjective evaluations. Although the discussion about the status of interpersonal comparisons is postponed to a later section of this chapter, I point here to the interpretation of wi (~ai ::wj (::); wh(::)::).10 It can only mean that in order to understand what a good does to other persons one has to place oneself in the position of those other persons and use also the evaluations of these other persons. In this it is assumed that there is something to be known namely, what a good does to another person. Of course one can disagree See also in this chapter p. 101. There is of course a similar way of reasoning for the external e ects because of socalled macro judgements, they are not intelligible at all beyond some aesthetic pleasure distributions might give. Some reason has to be given, and what goods do to people, will be an essential part of it. 10 Some interesting puzzles arise if one writes this function more explicit, what would be the arguments of wj () and wh ()? Would it be A or (~aj ; :::;wi ();::) and (~ah; :::;wi ();::) respectively. Is this circularity and should it be seen in the light of xed point theorems or just recursion? Taking external e ects into account is less simple as it seemed to be. 8 9
74
CHAPTER 3. VOLITIONAL INDIVIDUALISM
with another person on what it means for him, but it has to be assumed that these statements are true or false, depending on what the good means to that person. This is denied if one takes interpersonal comparisons to be subjective evaluations, or conventions, which are taken to be neither true nor false. In a later section of this chapter, it appears that in Davidson's theory of radical interpretation, those judgements on what goods do to people are just like other statements about the world.11 Summarising, taking external e ects of goods into consideration, one can argue for transfers of one person to another, on the ground that it is a Pareto improvement. Although this seems to be a possibility to argue for a redistribution without admitting interpersonal comparability, it appears that it is not. Without interpersonal comparisons, the external e ects cannot be explained. The explanation of external e ects refers to the meaning of goods to other people and knowledge about their wants and desires. So, this way of solving distribution problems without using interpersonal comparisons will not do, let me turn to the second proposal.
3.3.2 Non-envy analysis
The other way of circumventing interpersonal comparisons is based on the idea of non-envy. The idea that persons are not envious of each other's commodity bundles, is an element of an egalitarian distribution. According to the non-envy analysis an allocation is considered as good if it satis es not merely the condition of being Pareto optimal, which means that there is no allocation in which some are better o and nobody worse o , but also satis es the condition that nobody envies the commodity bundle of another person, by which is meant that no one would like to have the commodity bundle of someone else instead of his own. In an allocation that satis es these conditions there is no one who has a complaint about the distribution such that he would like to change place with someone else in the allocation. Actually, as will be clear of this latter description, `non-envy' is a misnomer. The issue is not concerning some emotional response called envy, but the idea of equality that nobody can complain about an allocation that another is better o such that he would like to change place with in that allocation. However, in the literature this approach goes by this name and I will conform to the use of the term `envy' as used in that literature, meaning preferring the position of another in an allocation, or preferring that what another has to one's own position. Allocations satisfying these conditions of being Pareto optimal and envy-free in the sense explained, are so-called fair.12 An allocation A is fair i : 11 12
See for example [Hurley, 1989, p. 84-122 ]. They are also called superfair [Baumol, 1986]
3.3. NECESSITY CHALLENGED
75
A is Pareto optimal, 8B 8i wi (~ai ) wi (~bi ) , and A is envy-free, 8i; j wi(~ai ) wi(~aj ) The second condition of a fair allocation, the non-envy condition, was introduced by Tinbergen, who attributed it to the physicist Ehrenfest, as a way of making judgements about allocations; for example, the judgement that some allocations are better from an egalitarian point of view than others; without using interpersonal comparison of well-being.13 Tinbergen introduced it as a way out of the problem of interpersonal comparisons of satisfaction and saw it as a practical alternative [Tinbergen, 1953, p. 59]. The non-envy analysis is introduced as an important element of the idea of equality, namely some sort of symmetry that in judging allocations the person to whom a good is allocated is not relevant in itself. The non-envy condition is based on an evaluation of permutations of distributions in which the commodity bundles change owner. This re ects the idea that justice and equality should be blind to whom it operates. All should be treated alike. Nobody should have a complaint that another is better o such that he would like to change place with another person. An envy-free distribution is one in which all prefer their own commodity bundle to, or are indi erent to, a bundle someone else receives. All prefer the allocation to, or are indi erent to, an allocation in which the commodity bundles change of owner, i.e. a permutation of the allocation. In formula, an allocation A is envy-free: 88i wi(~ai ) wi(~a(i) ), in which stands for permutations [Kolm, 1972, p. 25].14 This analysis will be discussed rather extensively, because it is an intriguing analysis according to which eciency in the sense of Pareto eciency and the liberty of people incorporated in the Pareto principle are held to be consistent with egalitarian ideas. Procedures are suggested to arrive at allocations that are Pareto optimal and envy-free. In these, the choice of persons themselves get an important role. This will threaten the whole project of this thesis, because now it seems to be impossible to evaluate distributions regarding equality without looking at their developmental history, which was, as was stated in the introduction, the aim of this study.15 First, the existence of an envy-free Pareto optimal distribution is shown in case production is not taken into account. Next, production is taken into account and several arising problems are described. Dworkin's 13 Kolm writes that the original question Ehrenfest answered was: `What is a fair wage?' He answered: `The wage such that each individual prefers his wage with his job to the wage-job pair of others.' [Kolm, 1996] The issue of just prices will return below. 14 See appendix 1 for a note on the relation between the non-envyconditionand Suppes' grading principle. 15 See chapter 1 p. 1.
CHAPTER 3. VOLITIONAL INDIVIDUALISM
76
alj B Eq al A ali G J
Oj
I
q
q
q
q
Oi
aki
ak
akj
Figure 3.1: Edgworth diagram with indi erence curves proposals is presented, which also exhibits serious problems because of the introduction of insurances. It is shown that problems are not just arising because production is taken into account, but contrary to what seems to be generally assumed, that they were already discernible in the rst results. The non-envy analysis cannot do really without interpersonal comparisons of what goods do to people, without running into serious problems.
Non-envy, a rst result In order to explain the idea of allocations being fair, i.e. being Pareto optimal and envy-free, I will follow Kolm and Baumol by using so-called Edgworth boxes. For reasons of simplicity, I restrict the discussion to two persons, two goods situations. The main points of these situations can be quite easily extended to situations with more goods and more persons. In an Edgworth diagram ( gure 3.1) the horizontal line starting with Oi till ak represents the amount of ak that person i receives. The remaining part from ak on that line (in the direction of Oj ) belongs to j. The length of the line starting from ak till the end represents the amount person j has. In a similar way the vertical line represents from Oi on how much i receives of a commodity al and the remaining part on that line the amount j receives of a commodity al . The content of this box represents all possible allocations concerning the goods ak ; al and persons i; j. The centre of it,
3.3. NECESSITY CHALLENGED
77
Fi c = 0 G A0 cc cc BB ccTBEq B 0 BcTB c B B cc B G c1IB cc IA A
Oj
q
q
q
q
q
q
q
Oi
Figure 3.2: The non-envy boundary Fi Eq, represents the allocation in which both of them get an equal amount of both goods. In this box we can represent the so-called indi erence curves for i and j. For example, I represents an indi erence curve for i, which means that i is indi erent between receiving the bundle represented by B and G on I. J represents an indi erence curve for j. Now looking at an allocation A in gure 3.2, on an indi erence curve IA in which i is receiving aki and ali , i will not envy j, if he does not judge the allocation A0 , the symmetrical counterpart of A in which the commodity bundles changed owner, to be worse than A. So, person i does not envy j if A0 is on or below IA . Now we can de ne the non-envy boundary Fi as those allocations, B, of which the symmetrical counterparts, B 0 , are on the same indi erence curve IB . In all the allocations below this boundary, for example G, i prefers the bundle j has, so i is envious of what j has. But in all allocations on or above this border i does not prefer j's bundle. In a similar way the non-envy boundary of j can be represented, and the allocations below this area are such that j does not prefer the bundle of i to his own. Combining these boundaries in one diagram, we can see in gure 3.3 that Eq belongs to both boundaries. In this diagram we can also see that there is an area N between the boundaries Fi and Fj in which no person prefers the commodity bundle of the other. In this area of envy-free allocations there exist an area representing allocations which are Pareto superior to Eq, P . One of these is Pareto optimal and envy-free
CHAPTER 3. VOLITIONAL INDIVIDUALISM
78
N Eq P J ) * N I > Fj A
Oj
Fi
q
q
Oi
Figure 3.3: Pareto ecient and envy-free allocations and consequently, it is a `fair' allocation. To give a less abstract story of fairness. Suppose there are two persons, one likes potatoes much more than rice and another likes rice more than potatoes. If both of them receive an equal amount of potatoes and an equal amount of rice, then both of them would be better o by trading potatoes for rice. The equal division, Eq, of the commodities is, although not Pareto optimal, envy-free because no one prefers the bundle of another. In order to get a clear idea of the meaning of envy-free allocations the allocation A in the area N , P is illuminating. Although the person preferring potatoes is having less than he would have in allocation Eq, he nevertheless does not prefer the commodity bundle of his neighbour j, because then he would have even less potatoes. It is shown by Kolm that the indi erence curve of i (and also of j) through Eq is not crossing the non-envy boundary of i, and all other points of the non-envy boundary than Eq are below the indi erence curve through Eq. This can be easily seen with the help of gure 3.4. If some indi erence curve IA for i that is further away from the Oi , representing one in which i is better o than he is in allocation Eq, it will have no allocations on it which are on the non-envy boundary. This will be so because all symmetrical counterparts of A on this indi erence curve will be below this indi erence curve and never be on this curve. IA will not have any points in common with the non-envy boundary. Only the indi erence
3.3. NECESSITY CHALLENGED
79
XX XEqX A0 X XXX A Fi
Oj
IEq IA
q
q
q
Oi
Figure 3.4: Indi erence curves above the one with Eq do not intersect the non-envy boundary curve through Eq will have a symmetrical point on the same indi erence curve, which is Eq itself. Consequently, all other points of the non-envy boundary belong to indi erence curves representing a lower well-being of i. It will be clear that it is assumed that indi erence curves of one person do not intersect and are convex, representing the idea that if you have already much of one good you are more likely to change it for another. This means that the area between the indi erence curves through Eq from i and j lay within the non-envy area, as shown in gure 3.3. Consequently, there is always a Pareto optimal envy-free allocation. Eq belongs to the area P and N. So N is not empty. But in N there is a Pareto optimal distribution which can be seen as follows. Start with Eq and let the persons exchange their commodities. Let them trade their goods in a free market. The resulting allocation will be in P, they only trade if the resulting allocation is Pareto superior to Eq, and they will stop if a Pareto optimal distribution is reached. This distribution will be in N and in P i.e Pareto optimal and envy free, hence it will be fair. In a situation with a xed amount of goods and a xed number of persons, there is always a fair allocation. This was made intuitively clear for situations with two persons and two goods. It can be shown that it also holds for cases of more than two goods and more than two persons.16 16
A proof for the case of more goods and more persons is given in appendix 2
80
CHAPTER 3. VOLITIONAL INDIVIDUALISM
Summarising this discussion. In a situation with a xed amount of goods and a xed number of persons there exists a fair allocation, i.e. an allocation which is Pareto optimal and envy-free. This allocation can be seen as a result of an equal division of each good among the persons who will exchange their commodities in a perfect competitive market. The result is Pareto optimal and envy-free. No one prefers the commodities of another to his own when it starts, because all have an equal part of everything, and via exchange on a perfect competitive market this cannot be altered, otherwise the exchange would not have occurred. The existence of a fair distribution is not only established, but a procedure to arrive at one is suggested too. This idea is elaborated further by Dworkin in proposing equality of resources. But before turning to that elaboration, I will rst pay attention to what will happen if production of the goods is taken into account. That will facilitate a critical discussion on Dworkin's ideas and the non-envy approach in general.
Non-envy analysis and production
Although the result of the former section is convenient, its applicability is limited. In real world cases we don't have a xed amount of goods because it is dependent on human production. It is natural to wonder whether in those cases in which we take production into account there is also such a result. The answer is: `No.' It has been shown by Pazner and Schmeidler that in case production is taken into account there is not always a Pareto optimal envy-free allocation.17 The original example is given in appendix 3. Here I will illustrate it by a gure borrowed from [Boadway & Bruce, 1984, p. 173]. Production is introduced by considering leisure as a good. Suppose now there is some good a, which is produced, and leisure. The production of some good a costs leisure. Leisure is used and transformed in some amount of good. The amount of a is not xed but is dependent on the leisure which is transformed. Suppose both persons i and j receive the same amount of leisure, say l, but their productive capabilities di er, and so their real wages, the amount of goods they can produce, will di er. The more able person i will receive more than j in exchange for his leisure. The amount of a which is received by i and j in exchange for their leisure is represented on the vertical axis and the amount of leisure on the horizontal. The indi erence curves for i and j are I; J respectively. It can be seen in gure 3.5 that in the allocation A j prefers i's bundle, but that in transferring an amount t 17 Below it will become clear that this presentation of the problem is not correct, there are always Pareto ecient envy-free allocations; they are characterised by equality of how well o people are and so depend on interpersonal comparisons .
3.3. NECESSITY CHALLENGED
81
6
a
,, , IA , ai Z , Z Ai , ai , t Z ZZ , ZZ A0iZZ,, Z Z aj + t h hhhh,Zh,ZhZhZAZ0j Z aj h hJAh,h,hhhhZhZhhZhZhhhJA h hZhIAhZhZh Aj hZ ,, Z Z q
q
q
q
O
0
0
l
L
-
Figure 3.5: Budget lines of i and j for and after a transfer t of good a from i to j of the good a from i to j, it is possible that i starts preferring j's position before j stops preferring that what i has. So no transfer will lead to an envy-free allocation. The negative result if production is taken into account brought Varian to the idea of some other concepts of envy with their corresponding ideas of fairness, namely wealth fairness and income fairness. An allocation is fair with respect to wealth if an allocation is Pareto optimal and no one prefers the combination of the goods of another and the amount of work he has to do in order to produce those goods. So complaining about one's possessions being less than those of a dentist is relevant only if one prefer his commodity bundle if one is prepared to produce as much as the dentist in order to acquire this commodity bundle. So, suppose everybody has a well-being function wi with the arguments the commodity bundle ~ai and his remaining leisure after working an amount of ri , l , ri , so wi(~ai ; l , ri ), then his wealth can be considered as the compound of commodities and leisure, (~ai ; l , ri). Suppose furthermore that in order to produce the bundle ~aj by agent i Ri(~aj ) units of work are needed, (8i Ri(~ai ) = ri ). An allocation which is envy-free with respect to wealth can now be de ned. An allocation is envy-free with respect to wealth if 8i; j wi (~ai ; l , ri) wi(~aj ; l , Ri(~aj ))
82
CHAPTER 3. VOLITIONAL INDIVIDUALISM
Varian has shown that the non-envy condition with respect to wealth is consistent with Pareto eciency in production economies [Varian, 1974]. We should realise that in allocations that satisfy wealth fairness, those persons less able in producing are denied to have a reason to complain about the amount the more able persons receive, because the less able ones would loose too much of their leisure in producing the same amount of goods as the more able persons do. They do not prefer the combination of receiving as much as the able ones and producing as much as those. To put it more clearly, persons born with a handicap in production will not be compensated if this principle of wealth fairness determines the allocation. The concept of wealth fairness is, because of this, very suspicious and allocations being fair with respect to wealth are not because of being fair morally right. Let me turn to the other concept of fairness, which is income fairness. An allocation is income fair if it is Pareto optimal and the income, the compound of commodities and leisure, of everyone is equal. If ~p is the price vector of the commodities ~a and w is the wage-rate, then envy-free with respect to income can be de ned. An allocation is envy-free with respect to income if 8i; j (~p; w)(~ai; l , ri) (~p; w)(~aj ; l , rj ) If every agent has the same income, to which leisure is reckoned too, there is no preference for someone else's income. It can be proved that the result of an exchange on a perfect competitive market (a Walrasian equilibrium18) starting with a distribution in which everybody has an equal amount of goods Na and everybody has the same amount of leisure l is Pareto ecient and is envy-free with respect to income [Varian, 1974]. In an income fair allocation the more able agents work for the less able persons and the result is that they can be worse o than those unable persons. Because the leisure of an able person is more expensive than the leisure of the disabled and because of Pareto optimality, every agent has to buy his own leisure, (leisure of a person cannot be enjoyed by another and so it is wasted if it is not bought back by that person) the more able agents have less to spend on the other commodities. In a sense, the more able ones are punished for their capabilities or as Dworkin states, they are enslaved by their abilities [Dworkin, 1981b, p. 311].19 Suppose one has the ability to earn a high income for example as a hangwoman, because of the fact that one has, just as Lizenka Tachec in Pavel Kohout's novel 18 A Walrasian equilibrium is de ned as a price commodity matrix such that the markets are cleared; there is no excess supply or excess demand. 19 The plea for a tax on capability by Tinbergen could be seen in this light [Tinbergen, 1979, p. 135]. But Tinbergen is not using any longer a non-envy analysis in his work Income Distribution, he uses equality of welfare as the main objective [Tinbergen, 1979, p. 130]. The tax on capability is an instrument for reaching that equality which is based on the correlation found by him between capability and primary income.
3.3. NECESSITY CHALLENGED
83
The hangwoman, `a face one would like to meet at the dentist'. It would
be costly not to perform this work as a hangwoman if this idea of income fairness was accepted. Morality sometimes demands us to give up some of our own ambitions to further the interests of others. But should welfare patterns be reversed by morality, as is demanded by the concept of income fairness, even if one has serious objections to the way one can `produce something valuable for example as a hangwoman'? An armative answer is highly questionable. Both concepts of fairness, income fairness and wealth fairness, treat agents, even if they have similar preferences, di erently, depending on their abilities. They are treated di erently because of properties beyond their responsibility. This is the reason why Pazner and Schmeidler introduced another idea about allocations that could be acceptable from an egalitarian point of view [Pazner & Schmeidler, 1978]. They introduced the concept of a Pareto ecient egalitarian equivalent allocation. They hold an allocation admissible if it is for each agent equivalent to an allocation in a (hypothetical) economy in which there is an equal distribution of all resources. Pazner and Schmeidler did show that in the standard economy such a Pareto ecient egalitarian equivalent allocation exists. An allocation that is Pareto ecient and egalitarian equivalent is or should be considered to be envyfree. There is no reason to complain, because everyone receives a commodity bundle equivalent to a commodity bundle that is equal for all.20 An example due to Daniel shows the strange consequences of using Pareto ecient equivalent allocation as a principle for distributions. They are unacceptable from a moral point of view [Daniel, 1978]. The example he described is an exchange economy with four agents 1,2,3 and 4, and two goods ak and al represented in gure 3.6. Eq is the egalitarian allocation in a hypothetical economy, I,II,III, IV are the indi erence curves of agents 1,2,3 and 4 respectively, and G is the egalitarian equivalent allocation. The Pareto ecient egalitarian equivalent allocation. G leaves agents 1 and 2 preferring the bundles of agents 3 and 4. The justi cation for this allocation is that agents 1 and 2 should accept their allocations and should not have complaints having less than , i.e. preferring the goods of, agents 3 and 4, because what they receive is equivalent in a hypothetical economy to an allocation in which they get as much as agents 3 and 4. Of course this kind of reasoning is unacceptable if there is in the actual economy an envy-free Pareto ecient allocation such as is the case in this example, namely E. 20 The trick of their proof is to start with an egalitarian distribution say Eq in which ai = e for all persons i, Eq is an egalitarian distribution. Now there exists a real number 0 < r < 1 such that for r:Eq there is an equivalent allocation which is Pareto ecient [Pazner & Schmeidler, 1974].
CHAPTER 3. VOLITIONAL INDIVIDUALISM
84
III
,, @@ G , E@; @ , @ G@ @ , 6 @I @ , Good al @,@,@II @Eq , @ @ , @@E @; @G , G , O Good ak r
3
r
r
1
13
r
r
2
2 r4
r
4
IV
Figure 3.6: Pareto ecient egalitarian equivalent allocation Eq and a Pareto ecient envy-free allocation E Production seems to threaten the whole idea of fairness. The problem resulted from introducing talents and other non-transferable goods like leisure. It is illuminating to look at what Dworkin proposes as a solution to this problem. He introduces a market of insurances. In this way Dworkin intends to circumvent interpersonal comparison of well-being and to avoid, as he believes, the above mentioned exploitation of the talented [Dworkin, 1981b, p. 320 .]. Although it is an interesting instrument, the introduction of a market of lotteries, leads to particular complications concerning the information needed to determine whether a distribution is acceptable. Complications which seem to be overlooked by many authors.21 Let me turn to his ideas on equality.
Dworkin's proposal As was seen in the previous paragraph, the treatment of non-transferable goods like ability and handicaps gives rise to problems. Income fairness and wealth fairness proposed by Varian were not proper solutions, neither was the idea of a Pareto ecient egalitarian equivalent allocation proposed 21 G. Lock seems to be an exception in his critical discussion on Dworkin's proposals with his example of the indolent indigent person [Lock, 1989]. Arnsperger in [Arnsperger, 1996] is also aware there is something wrong with Dworkin's proposal.
3.3. NECESSITY CHALLENGED
85
by Pazner and Schmeidler. The general idea behind the discussion on the proposals is that handicaps and lack of talents, or in general properties beyond the responsibility of the agent, should not lead in themselves to di erences in distributions. This idea is also expressed by Dworkin who holds that disadvantages for which a person himself is responsible, should be treated di erently and may lead to di erences in distributions. It would be unfair and contradictory to the principle of equality if a modest person who had been saving a lot had to hand over his savings to a wasteful person. This element of responsibility is one of Dworkin's main reasons for rejecting equality of welfare theories. Persons are held responsible for their preferences by Dworkin [Dworkin, 1981a] [Dworkin, 1981b]. It seems that according to Dworkin preferences originate in the will or the volitional part as is held in volitional individualism. Dworkin's proposal is discussed here as a solution to the problem of envy-free allocations in cases of production; a solution which threatens the project of this study by his introduction of an insurance market, because the aim was to arrive at evaluations of distributions regarding equality without taking their historical development into account. Taking responsibility into account is the principal reason why Dworkin distinguishes brute luck from option luck. Brute luck happens to a person and is not the result of a decision, as in for example accepting a gamble, or a lottery. Brute luck is imposed on someone. An instance of brute bad luck is a storm blowing o the roof of one's house. Option luck on the contrary, is the result of a freely undertaken gamble or lottery. An instance of option bad luck is loosing in the National Lottery. By entering the lottery one freely accepts the risk of loosing. Brute luck should be distributed equally; it is a sort of negative resource for which persons are not responsible. Option luck is dependent on decisions of agents and as a consequence it should not be simply distributed equally. It is not fair to ask from the winner of a lottery to give his prize to the losers, the essential element of a lottery would be lost, thus is argued by Dworkin. Dworkin emphasises the distinction between brute luck and option luck, it fades away once one realises the possibility of insurances. The mere possibility of insurances transforms brute luck into option luck. If you are struck by brute bad luck the results are due to your own decision because you could have insured against it. So, by not having insured against this bad luck you decided to have a particular bet in a gamble and because of that you are of course responsible.22 In this way brute luck is eliminated 22 You could hold that not all of the bad luck could be compensated because some damage cannot be restored and no bene t can be compensating. But by taking this point of view the whole project of Dworkin collapses because he assumes all goods to be intersubstitutable, otherwise his insurance method for having low earning capabilities
86
CHAPTER 3. VOLITIONAL INDIVIDUALISM
and prevented from being considered as a resource that has to be divided equally. Following the idea of transforming brute luck into option luck by an insurance scheme, Dworkin introduces an insurance against having a lack of talents. He illustrates his proposal by using the idea of an original situation. Although Dworkin rejects original positions approaches as justi cations of social principles, he admits they can be used as illustrations of such principles. This function however cannot be satis ed by original positions if they are inhabited by bare individuals without anything known about their own character as for example in Rawls' theory. How could they have preferences at all? In Dworkin's amended original position people know their talents but they lack the information about the income that they can get by these in the particular economy they will enter. In this original position a person can decide to insure for an income that he could have received in case he could earn more with his talents. The premium for this insurance is paid in case one can earn this maximum. Being a little bit more formal and following Roemer [Roemer, 1985], Dworkin's insurance in the original position guarantees at least an income Emax . If such an insurance scheme is to be considered as a rational option for the persons in the original position it has to be the case that X pmax :(Emax , Premium) + (1 , pmax ):Emax pu :Eu u in which pu is the probability of the ability of earnings Eu. The insurance in which Eu is supplied up to Emax , unless u = max in which case the
Premium has to be paid, should be feasible. This means that the bene ts Emax , Eu are covered P by the paid Premium, the feasibility condition. So pmax :Premium u pu:(Emax , Eu), which is: Pu pu:(Emax , Eu) Premium pmax It shows that because pmax is small, one has to pay a large premium if one happens to be capable of earnings Emax . By lowering the wage for which you insure, the premium can be decreased as is seen by the feasibility condition. The p in the denominator will increase and the numerator will decrease. By using some abbreviations and realising we have to reckon with what goods mean to people, W(), and not the earnings themselves we get as the rst condition: X p :W(E ,Premium)+X p :W (E ) X p :W(E )+X p :W(E ) s s u ins s s u u
s u s u would fail completely. In his story the clamshells and money represent the universal standard commodities.
3.3. NECESSITY CHALLENGED
87
in which ps denote the probabilities of the situations s in which the earnings Es are above the earnings Eins for which one is insured, and pu denote the probabilities of the situations in which Eu < Eins. By abbreviating ,Premium by Ts and Eins , Eu by Tu , and letting these be referred to by Tv , this condition becomes:
X p :W(E v
v + Tv )
v
X p :W(E ) v
v
v
Essentially we have here the problem of nding a Tv which maximises:
X p :W(E v
constrained by
v
v + Tv )
X p :T v
v v0
which is the feasibility condition; the bene ts have to be paid by the premiums.23 The next step in Dworkin's proposal is to replace this hypothetical insurance market, in which all insure against earning less than a particular icome, by a tax system that is an approximation of this market. The tax is meant to be the premium people would pay if they can earn more than the particular income insured against. Those earning less than that income get a bene t up to that income. The premium is determined by taking the average income as the coverage income Eins which is considered by Dworkin to be the most reasonable one. So the tax system mimics the insurance market in which the talents are more or less equalised, because the probabilities of getting a particular income are considered the same for all 23 This analysis is criticised by Van Parijs [Van Parijs, 1990]. He holds that it would be more rational to maximise the minimum of the possible utilities, minfW (Ev + Tv )g, rather than to maximise expected utility, v pv :W (Ev + Tv ). This proposal however will lead to equality of welfare, which was rejected by Dworkin. Therefore, it is highly questionable whether this was meant by Dworkin. Suppose agents choose such a scheme of Tv 's that will maximise minfW (Ev + Tv )g. Suppose furthermore that they realise some insurance which guaranteesa certain welfare level W (Eins), but that there are some di erences in utilities W (Ev + Tv ) in the di erent situations v left, 9vW (Ev + Tv ) , W (Eins ) > 0, then the scheme of Tv could be adjusted in such a way that the minimum utility U (Eins) could be increased further, namely by taking a higher premium in those situation which result in utilities above the minimum. These adjustments will lead to equality of welfare.(To be clear this reasoning has nothing to do with using preferences as resources which is discussed in the same section by Van Parijs, it follows merely from his proposal of maximising the minimum [Van Parijs, 1990, p. 337].) As a consequence the critique by Van Parijs on Roemer's example against Dworkin's insurance proposal is not correct. Roemer's example is presented in appendix 4 of this chapter.
P
88
CHAPTER 3. VOLITIONAL INDIVIDUALISM
in the original position. In this system of an insurance market there is no room for complaints that another is better o such that one would like to have had changed places, because di erences are due to option luck. In this way Dworkin argues the distribution under the tax system is an envy-free allocation.24 This last step however is a suspicious one for Dworkin. This shift from the hypothetical insurance market in the original position to the tax system can be made if he accepts his own proposal as a pattern principle. A pattern principle does not take into account how a distribution came about but looks at the actual patterns of distribution. So, this step could be argued for because Dworkin's proposal is considered as one of the non-envy approaches which are in fact pattern principles. These non-envy proposals are essentially pattern principles, according to them a distribution is considered to be fair because of its distribution pattern. Knowledge about other things than the pattern of distribution and the preferences of people is not necessary. The actual historical process of exchanges is not directly important in determining whether a particular distribution is just. In the sections above the essential key in determining the acceptability of a distribution, was its being envy-free. It has been shown that such distributions exist in particular economies. As is clear from the proofs for the existence of envy-free distributions, the allocations resulting from an initial equal distribution followed by a process of exchanges on a free and competitive market happen to be such envy-free allocations.25 But a Pareto ecient envy-free allocation might be arrived at in another way. The distribution pattern is the reason why an allocation is just, that it resulted from an equal start followed by trades on a free and competitive market is not decisive. Consequently, the last step in Dworkin's reasoning from a hypothetical market of insurances to a tax system could be made because Dworkin considers his proposal as a pattern principle like other non-envy proposals. However there is a relevant di erence. By the introduction of a market of insurances, or lotteries and gambles, something changed because advantages or disadvantages resulting from these are to be seen as option luck or option bad luck. One could have forgone this lottery. But now it is clear that the actual distribution pattern is not enough for determining the acceptability of the allocation. Someone can have much more than another because of option luck, which as Dworkin argues, must not be transferred to the losers who accepted the risk of bad luck. 24 A similar way of reasoning is used by Dworkin in his proposal for some basic insurances in health care [Dworkin, 1994]. 25 See gure 3.3 in this chapter on p. 78, and the footnote on p. 80 in this chapter, and the discussion on Roemer's example in this chapter on p. 93.
3.3. NECESSITY CHALLENGED
89
The whole idea of option luck being relevant for a distribution implies that one has to consider not only the actual distributionpattern but also the choices made by the agents. In order to determine whether a distribution is fair one has to consider the value of the gambles that people accepted. So, one should know the alternatives which turned out not to be instantiated. I admit one can take as the value of a gamble its expected value, but persons do not receive expected values. Values do not remain expected; some will become actual. Just because of this aspect Dworkin's proposal is not a pattern principle. A story by Borges The Lottery in Babylon can illustrate this point even more clearly. Borges writes about a society in which people enjoy gambling very much. Gambling evolves into a basic right, which should not be denied to those without money, and gambling is expanded to all areas of life. The drawings are accomplished by a secret society which also sees to the execution of the `prizes" of the drawings. In short, everything is subject to gambles, housing facilities, dead penalties, taxes etc. In this Babylonian society the actions undertaken, need not to be di erent from those in a `rational' society. Because of the uncertainty pervading the outcome of every action, together with the necessity of not compensating option luck for which a person himself is responsible, which can be seen as a gamble, Dworkin's proposal is not a pattern principle but has procedural aspects. Every distribution pattern can be the result of freely chosen actions, so its developmental history is decisive.26 The step from the hypothetical insurance market to the system of taxes is illegitimate because in this step agents are denied to choose freely which risks to accept and which kinds of option luck to accept.27 In general, we can conclude that in every analysis of equality in which the responsibility of agents is taken into account, pattern principles are excluded. Because of the uncertainty of the outcomes of actions performed by agents, the way the distribution came about is essential. An evaluation of a distribution regarding equality, without taking into account how it came about, which is the aim of this thesis28 , would become impossible if responsibility of agents played this role and had to be incorporated in the ideal of equality. In the next chapter, I will argue that there is no reason to give responsibility that role. 26 Arnsperger mentioned that Dworkin's non-envy should apply to preferences induced by the outcomes of the resulting economic transformation process from resources to consumption goods [Arnsperger, 1997]. This induction already was impossible, if, what all others could simultaneously do, was not taken into account. Because of that he suggests that what all others could do should be used in order to arrive at the induced preferences. But if a stochastic process is admitted this will not do, as the story of Borges illustrates. 27 This was noticed in a sense by G. Lock in [Lock, 1989]. 28 See chapter 1 p. 1.
90
CHAPTER 3. VOLITIONAL INDIVIDUALISM
Summarising the discussion so far, the proposals that are based on the non-envy analyses seem acceptable ways of circumventing interpersonal comparisons of what goods do to people. The rst proposal, fairness, was only applicable in case production was not taken into account. If one considers production to be relevant the whole picture changes. Several competing proposals, Varian's wealth fairness and income fairness and Pazner and Schmeidler's Pareto ecient egalitarian equivalent allocation were presented of which none was a convincing analysis of the essentials of equality. Dworkin's idea of introducing insurances as an instrument of dealing with non-transferable goods leads to problems because his proposal is no longer pattern based and consequently, his argument for taxation is invalid. That levying taxes is mimicking an insurance scheme would be a valid argument only if his allocation principle was a pattern principle. Hence, the non-envy proposals discussed so far were not showing that interpersonal comparisons could be avoided, although the core of the problems is not yet clear. In the next section, I turn to that core and argue that the non-envy approach leads to problems if it embraces volitional individualism and denies interpersonal comparisons.
The evaluation of the non-envy analysis
It is time to make up our mind about the non-envy approach and try to discern the idea that is responsible for the problems with this approach. In this section, I discuss three issues in order to indicate that volitional individualism is responsible for the problems the non-envy approach su ers from. First, I discuss proposals that turn to weaker conditions than the non-envy condition. Next, I discuss an argument of Roemer against the non-envy proposals. Finally, I turn to the monotonicity conditions, notably population monotonicity and resource monotonicity, about which it is argued that they are in con ict with the non-envy approach. These three issues will illustrate that the non-envy approach by itself does not su er from the problems that are spelled out, but that volitional individualism is responsible. As was made clear in discussing cases in which production is taken into account, envy-free Pareto ecient allocations do not always exist. Because of this lack, related to so-called non-transferable goods like leisure or production capability, some alternatives are suggested for the non-envy condition [Fleurbay, 1994] [Arnsperger, 1996]. One group of proposals suggests that if envy-free allocations are not possible within the set of Pareto ecient allocations, one should aim at minimising the preferences for the position of someone else. It is suggested for example that one should look for allocations that minimise the maximum of the intensity of these en-
3.3. NECESSITY CHALLENGED
91
vious preferences [Fleurbay, 1994][Arnsperger, 1996]. Although it can be shown that such allocations exist, it is highly questionable whether they are acceptable. The idea of individualism seems to be not respected. In the envy-minimisation approach only complaints that are based on preferences for someone else's bundle beyond a minimum level are reasons to reject an allocation as unjust, the others are neglected. Not everyone's individual complaint about an allocation does count but only those whose intensity of preference for the bundle of someone else is above an arbitrary particular level does. So, from the point of view of volitional individualism, which is the reason for looking at the non-envy analysis in this chapter, these proposal are not satisfactory. Another group of proposals too turns to a weaker condition of non-envy, but in a di erent way. Let me explain one of these, namely undominated diversity advocated by Van Parijs. The variants of this condition su er from similar problems. According to the non-envy approach an allocation is not acceptable if there is even one individual who would like to change place with another person in the allocation using his own preferences. It is suggested that the non-envy condition is demanding too much. Because if a person who is unable to play the piano because of missing a nger wants compensation for lack of this ability to such a level that he would no longer prefer to change place with a pianist, the non-envy condition would demand such a compensation. It is too much demanding. It is too dependent on the unrestricted diversity of wants and preferences of people. Volitional individualism seems to pose a problem here. It is suggested that undominated diversity is required. The idea of undominated diversity is that the preferences of just one person is not enough to establish the unacceptability of an allocation, or a reason for compensation. In order there to be a reason for compensation, also other persons following their preferences should agree that a person would be better o if he changed place with the person he prefers to change place with. If they agree, then there is a reason to reject the allocation, or there is a reason for compensation. So, the person missing his nger would only receive a compensation if all others agreed that he would be better o if he changed place with the pianist. An allocation is considered unjust as long as there are two persons such that everyone in the society agrees that one of these two persons would be better o in the position of the other person. Van Parijs advocates this criterion which he borrows from Ackerman [Van Parijs, 1990][Van Parijs, 1995, p. 59-88]. In other words, the requirement of undominated diversity holds that in an allocation satisfying undominated diversity for all persons i and j there is at least one person k according to whose preferences the commodity bundle and non-transferable goods that i has is not preferred to those that j has.
92
CHAPTER 3. VOLITIONAL INDIVIDUALISM
In formula: 8i; j 9k wk (~ai ) wk (~aj ). It has been established that there exist Pareto ecient allocations that satisfy the condition of undominated diversity. It seems a perfect modi cation repairing the defect due to volitional individualism of the non-envy condition and the problem of taking production into account. But it is merely an appearance. Undominated diversity means that if in case of a Pareto ecient allocation, one person prefers the position of another to his own, and all but one agree with him, the allocation is acceptable. So the preference of this one exceptional person, whatever his preference is, can make that an allocation is satisfying undominated diversity. Someone could be prevented getting compensation because of a whimsical preference of a member of society. So, here too volitional individualism leads just as in the non-envy case to unacceptable results and has to be replaced. Van Parijs, aware of this problem, restricts the preferences that should be taken into account to those of persons who understand what it means to have a particular bundle of commodities, or handicaps, or talents. But this restrictions means leaving volitional individualism. Not all preferences whatever they are, are taken into account. So, we conclude that the proposals with the weaker conditions than being envy-free, are not satisfactory. The modi cation of non-envy to envy-minimisation was not a solution because of its con ict with volitional individualism and undominated diversity did not do either because of volitional individualism. Let me turn to the second issue to illustrate that volitional individualism is responsible for problems related to non-envy approach. Roemer argues against the non-envy analysis that it leads to strange conclusions. As is argued shortly this problem does not occur only because of non-transferable goods as Roemer suggests. It is shown that they are already present in case production and non-transferable goods are left out of consideration. Roemer has shown that in case di erences in well-being are considered to be determined by hidden resources, the person with having less of these hidden resources will be even worse o if these hidden resources were recognised and distributed equally, while one should expect he would be compensated for having less of it. Roemer's example is as follows: Suppose there are two types of persons, t1 and t2 each having a utility function wt1 and wt2. Furthermore there is one good and for all amounts b of that good the well-being of t1 type of people is higher than the well-being of t2 people, 8b wt1(b) > wt2(b). Now it appears that the di erence between the utility functions of t1 and t2 can be explained completely by another commodity let us say e, which is not transferable; for example the amount
3.3. NECESSITY CHALLENGED
93
of endorphins a person has. So 8b; e wt1(b; e) = wt2(b; e) where b is the amount of the transferable good and e is the amount of non-transferable good. The reason why wt1 is higher than wt2 given a particular amount of good b, is that t1 has more of the non-transferable good e than t2. So far the description of the example. Now starting from an equal division of resources, each agent gets half the amount b of the good available, 12 b, and half the amount of endorphins t1 has, 12 et1 , and half the amount t2 has, 12 et2. By supposition there is more et1 than et2, so et2 is more expensive than et1 . Because in a Pareto ecient allocation each agent will have bought all of his own non-transferable good (otherwise it would have been spilt) persons of type t2 end up with less to buy of the transferable good than persons of type t1. So they will be even worse o than without recognising and dividing equally the hidden resources e. This example of Roemer showing the strange consequences of the nonenvy proposals, is not dependent on acknowledging hidden resources it can be easily transformed into an example with three goods a, b, and c, and two persons 1; 2 in a situation with a xed amount of resources. Suppose person 1 who is a little bit deaf, can get a variety of hearing aids with di erent ampli cation possibilities; the more it ampli es, the more he can hear, and person 2, who is good hearing, so a hearing aid is not of any help to him, but who is a little bit myopic, for which a negative lens would be helpful and within a certain range, the more negative dioptrics the better for him. Such a negative lens will not be of any help for 1. Both of them love going to the theatre. For 1 it holds that he will enjoy the theatre more if he has a hearing aid, and similarly for 2 if he can see what the actors do. Compare a, let us say, the tickets for the theatre, to the good represented by b in the example of Roemer above, b, the hearing aids amplifying power, to et1, and c, the dioptrics of the lenses, to et2 . Now, contrary to Roemer's reasoning, it is not simply shown that 1 is worse o than 2 if there is less of b than c. Because how should we determine that there is less of good b than good c in a relevant way? The only way to make such remarks is by comparing the e ects of these on persons. In order to state that there is less ampli cation available for 1 than dioptrics for 2, we have to compare what the e ect of these instruments are for 1 and 2 respectively. It will not be very precise but in a rough manner we are able to do this. If with some ampli cation 1 can hear normal conversation, but 2 does not have a visus better than 0,25 with help of his negative lens, while it could be improved to 1.0 by some other lens, then we can
94
CHAPTER 3. VOLITIONAL INDIVIDUALISM
say that 1 is better o with his hearing aid in the theatre than 2 with his negative lens. So within certain limits we can make reliable judgements on who is better o . This kind of judgements makes it possible to criticise the solutions presented as envy-free and Pareto optimal not because 1 is worse o than 2, but because according to the non-envy approach there are more allocations allowed than we should accept, for example the one resulting from an equal split of the goods a; b and c followed by transfers on a competitive market. Similarly, it should not be possible that recognising that someone is worse o because of having less o some good than someone else, that someone could end up even worse o if this good is recognised to be a resource and divided equally. Someone should be compensated for having less. The story of the persons with a visual and auditory handicap indicates that determining prices as Roemer does for non-transferable goods (the endorphins) is highly questionable. Kolm considers these even as nonsensical [Kolm, 1996]. How could the amounts of endorphins for i and j be traded.29 Both of the examples point to an old economic subject: what should be considered as just prices?[Robinson, 1962, p. 47] In the traditional view in the beginnings of this era, prices are seen as re ecting what is a just or fair distribution in which interpersonal comparisons of what goods mean to people are made, such as enjoying the theatre or feeling pain, instead of prices being determined by the processes of the market. This is not only restricted to non-transferable goods but applies also to other goods as illustrated by the example with hearing aids and glasses. That it is not restricted to non-transferable goods, is of course to be expected. There is much trade on the market where there is no transfer at all, for example the trade in land, mines and mountains. Also other transferable goods are often not transferred, they remain were they are. What is traded, are rights to use or enjoy something as it was made clear for example by Ross' analysis of property rights[Benn, 1967]. Such rights can of course also concern more abstract matters such as capabilities. These rights can even be enforced. So, what counts is the question what is a morally right allocation, for which interpersonal comparisons are needed. The problem pointed at by the example of Roemer is that the non-envy analysis by accepting volitional individualism allows more allocations than we should accept. It is a problem of which it was seen that it is not restricted to cases in which production is taken into account. 29 The argument here is in line with Scanlon's remarks on Roemer's approach [Scanlon, 1986].
3.3. NECESSITY CHALLENGED
95
Finally, I will turn to the third issue in order to illuminate that volitional individualism is responsible for the serious problems related to the nonenvy approach, namely the so-called monotonicity conditions. One concerns an increase in the number of persons, the other concerns an increase in resources. Following the population monotonicity means that if one turns to an economy in which there is just one more person while the resources remain xed, then one should expect that all would decrease in what they receive. But it is argued that this is not always the case [Arnsperger, 1996]. The preferences of the added person can change the allocation in such a way that some persons become better o . This con ict between the nonenvy approach and the population monotonicity is related to accepting unrestricted diversity of preferences, in other words volitional individualism. But if one turns to realistic individualism one could hold there is just one type of preference and consequently this problem will not arise. What about the increase of resources? Resource monotonicity means that one should expect that no one will become worse o . The following example due to Moulin presented by Young shows that it might turn out di erently in [Young, 1994, p. 155] in the non-envy approach. Suppose there are two persons i and j and two sorts of goods gin and vermouth of which there are 600 one-litre bottles of each. Both persons get 300 bottles of each. Person i takes the liquor only in a mix of 3 parts gin on 1 part vermouth. Person j enjoys the liquor in a mix of 1 part gin on 2 parts vermouth. The result of the competitive market starting with each having 300 bottles of each liquor will be that i ends up with 360 bottles of gin and 120 bottles of vermouth and j ends up with 240 bottles of gin and 480 bottles of vermouth. The following equations should be solved: ai + aj = 600 (concerns the amounts of vermouth); bi + bj = 600 (concerns the gin), ab = 13 (the mix of i); ab = 2 (the mix of j). But now, consider the case in which there are 900 bottles of gin instead of 600. Solving the above equations with bi + bj = 900 replacing bi + bj = 600, results in i having 720 bottles of gin and 240 bottles of vermouth and j having 180 bottles of gin and 360 bottles of vermouth. Hence, j ends up worse o than in case there were 300 bottles of gin less. So this solution is not consistent with resource monotonicity. Although it is suggested that the allocation arrived at above is the only envy-free Pareto ecient solution, it is not, there are more of them. And similarly as in the case of the hearing aids and glasses, it should not be stated that the non-envy approach leads to solutions not consistent with resource monotonicity, but that some allocations are accepted that are not consistent with resource monotonicity. There are also envy-free Pareto ecient allocations that are consistent with resource monotonicity. I will i i
j
j
96
CHAPTER 3. VOLITIONAL INDIVIDUALISM
present one shortly. In order to get the problem clearer one should ask why inconsistency with resource monotonicity matters? The problem as before is volitional individualism and not accepting interpersonal comparisons. This becomes clear if one wonders whether it would be wrong if the one ending up worse o by the increase of resources was originally the better o person. Otherwise stated, why should the better-o pro t from an increase of resources? If one takes for example the total amount of litres of liquor as the measure to be compared to each other,30 then in the case of 600 bottles of gin i has 480 litres and j 720 litres j is clearly better o . In case there are 900 bottles of gin i has 960 litres and j 540, now j is the worse o .31 Although both allocations are unequal there is no reason why monotonicity poses a problem for non-envy proposals if interpersonal comparisons are accepted. A con ict with monotonicity is a problem if one does not accept such comparisons, i.e. if one accepts volitional individualism. If one maximises for instance the amount of liquor such that both of them have an equal amount of litres in the proportions they like, then monotonicity is not violated and an increase of resources will not lead to a decrease of what one person receives. If the total amount of liquor to be divided equally, is maximised, then the following system should be solved. Maximise ai the amount of litres for i restricted to ai + 13 ai = aj + 2:aj and ai + aj 600 and 13 ai + 2aj 600. Now calculations show that the rst inequality is the restricting one. Handle it as an equality and one arrives at i receiving c.a. 415 bottles of gin and 138 bottles of vermouth, and j receiving 184 bottles of gin and 369 bottles of vermouth. Because the rst inequality, the one concerning vermouth, was the e ective restriction, increasing the amount of gin will not change the distribution. One cannot say that this distribution is not ecient because it is assumed that the bottles are valuable only if they can be consumed in their preferred mixes. So gin without vermouth has no value and can be disposed without any e ect on the total amount of satisfaction of both persons. The allocation is also envy-free. Here we have an envy-free Pareto ecient allocation which does not con ict with resource monotonicity. Again one could argue that the problem, as in case of the hearing aids and glasses, is a problem of just prices. What are prices such that the equal distribution will result given the starting point in which each person has an equal part of the resources? It is the question with which 30 The total amount of litres of liquor could be taken to be re ecting the time one can enjoy drinking alcoholic drinks, another possibility could be taking the intoxicating e ect of alcohol as the equalisandum. 31 According to the index to be developed in the later chapters, the former allocation is more unequal than the latter, the inequality was diminished.
3.3. NECESSITY CHALLENGED
97
the whole idea of non-envy approach started, i.e. the question of the just price for labour [Kolm, 1996, p. 202]. A dicult project if one remains committed to volitional individualism . Summarisingthis section, it was pointed out that the non-envy approach becomes a problem if it remains connected with volitional individualism. The two suggested weaker conditions of non-envy as a way out of there being no envy-free Pareto ecient allocations, were not satisfactory. The envyminimisation suggestion was rejected because it violated the individualism. Undominated diversity led to unacceptable results because of volitional individualism; the preferences taken into account should be restricted to those of persons who understand what it is to have a particular bundle of goods or to su er from a handicap. Next, the argument of Roemer was presented and it was shown that the problem with the non-envy proposals is not directly connected with production or non-transferables but with the lack of interpersonal comparisons. It cannot be said that the non-envy approach leads to unacceptable allocations, but the approach might accept unacceptable allocations. The main issue in the examples which is hidden is the old economic subject of just prices for which interpersonal comparisons are needed. The same was indicated by the problems of non-envy and the monotonicity conditions. It is time now to accept the necessity of interpersonal comparisons of what goods mean to people in order to arrive at plausible judgements about allocations that are concerned with equality.
Summary and conclusion
The discussion on several non-envy proposals i.e. fairness, income fairness, wealth fairness, Pareto ecient egalitarian equivalent allocations and the insurance market, showed they were not appropriate. Actually the discussion indicates that interpersonal comparisons are possible. This is assumed. It is the background of the problems with respect to which the proposals are discussed. The possibility of interpersonal comparisons of how well o people are is acknowledged by the authors and inspires the discussion on the di erent proposals. Varian for example writes in discussing wealth fairness: Admittedly this de nition is not entirely ethically satisfactory. Perverse cases arise when one agent is the sole producer of some good, since in that case no complaint against him could be justi ed. It is especially bad if this agent is the sole producer of some good that gives utility only to him. [Varian, 1974, p. 73] So, Varian assumes that there is a method of describing that only one person gets some utility of some good and others not. What this good
98
CHAPTER 3. VOLITIONAL INDIVIDUALISM
does to people is assumed to be comparable. Furthermore, why is it wrong that the less able should get less than others as is admitted to be the case by Varian [Varian, 1975, p. 116], in what sense is the able favoured at the expense of the less able in wealth fairness, and in what sense is the able exploited by the less able? This is probably based on ideas how well o they are and so it is assumed they can be compared. Pazner and Schmeidler also admit that it is possible to compare preferences of persons, otherwise they could not state that people with the same preferences should not be treated di erently on the ground of aspects beyond their responsibility. The modi cations of the approach by weakening the non-envy condition to envy-minimisation or undominated diversity, in order to solve the problem of the lack of Pareto ecient envy-free allocations if production is taken into account, appeared not to be acceptable. The former violated volitional individualism, the latter ran into problems because of volitional individualism. The discussion on the example of Roemer that he brought up to show that the non-envy approach was not acceptable did also show that the problems with the non-envy approach are not due to production or non-transferables, but are due to the lack of interpersonal comparisons. It was not argued that the non-envy approach leads to unacceptable allocations but it was argued that it might accept such allocations. The main issue in the examples, which is hidden, is the old economic subject of just prices for which interpersonal comparisons are needed. The same goes for the problems of non-envy and monotonicity. The possibility of interpersonal comparisons of what goods do to people arises out of the development of the non-envy proposals itself. First, preferring the part of someone else to one's own was determined with respect to commodities only. It was held that it should be that allocation A is envyfree if 8i; j wi(~ai ) wi(~aj ) in which ~ai was a commodity bundle. Person i had to place himself in the position of having the commodities of another. Later, non-transferables were taken into account, for example talents. But talents are only talents in a particular economy. This was made clear by Dworkin in his exposition of an original position in which all persons know who and what they are, but in which they do not know what they can make out of it and consequently what their economic position will be. So, by taking talents into account in the non-envy condition, one does not point to talents per se but also to what one can do with one's talents in a particular society. In the non-envy analysis reckoning with talents, one had to compare oneself with what another receives and his position in the society. So we get as one of the conditions 8i; j wi (~ai ; ti) wi (~aj ; tj ) in which ~ai is again the commodity bundle and ti are the talents and capabilities, and the position of a person.
3.3. NECESSITY CHALLENGED
99
A further development32 of taking into account more di erences between people besides commodities and positions or talents is suggested by realising that it is possible that one does not prefer the commodity bundle but sometimes the `wealth' of someone.33 Consider again gure 3.3 on p. 78 with two persons again, one a potato lover and the other preferring rice. In allocation A in the area N , P the person preferring potatoes is having less than he would have in allocation Eq. Nevertheless, he will not prefer the commodity bundle of his neighbour, because then he would have even less potatoes. Although the lover of potatoes does not want to change the commodity with his neighbour's bundle, he can prefer in a certain way the `wealth' of another to his own. In this kind of preferring the part of someone else, one takes the preferences and ideals of another into account. But once preferences and volitions, v, are taken into account, we arrive at comparisons of wi(~ai ; ti; vi ) and wi(~aj ; tj ; vj ) which express interpersonal comparisons. The next question is of course what is the nature of these wi . It is argued for example by Kolm that these wi are similar for all. All di erences in evaluations are captured by di erences in ~a, t or v. There is no reason why wi would be di erent from wj . The di erences are `externalised'. These preferences are so-called fundamental preferences [Kolm, 1972, p. 79]. In the next section it will be indicated how such an `externalisation' is possible at all. So, the non-envy analysis although trying to avoid interpersonal comparisons, actually show a tendency to such comparisons. The discussion on the di erent proposals assume these to play an important role. Summarising, we can conclude that neither taking external e ects into account nor the non-envy analysis were successful in circumventing interpersonal comparisons. The former failed because the explanation of the external e ects assumed these comparisons to be possible. The latter failed because of problems of acceptability if the proposals are linked up with volitional individualism . En passant the introduction of a market of lotteries or insurances by Dworkin illustrated that if responsibility is incorporated in the ideal of equality, the evaluations of a distribution cannot be according to a pattern principle because the evaluations will depend on the way the distribution came about. In the next chapter, the issue of responsibility will be taken up, here I will continue with a discussion on the impossibility of interpersonal comparisons, because one part of the dilemma of equality, the necessity of interpersonal comparisons, was not challenged successfully. 32 This spelling out the di erences between people can be seen as the development of the stages of universalisations described by Mackie [Mackie, 1977] [Mackie, 1985]. An extension suggests itself, based on the third stage of universalisation, in which preferences and valuations of another person are taken into account. 33 This is pointed to by Baumol, see [Baumol, 1986, p. 25].
100
CHAPTER 3. VOLITIONAL INDIVIDUALISM
3.4 Impossibility rejected; realistic individualism The two ways of avoiding interpersonal comparisons by challenging the assumptions of Hammond's argument for the necessity of such comparisons were not successfull. Taking external e ects into account was not possible without accepting interpersonal comparisons. And giving up a social ordering inspiring the non-envy proposals was not satisfactory either in so far these proposals were linked up with volitional individualism. Volitional individualism according to which allocations should be evaluated on the basis of the individual preferences, wants, and desires springing from a well of volitions in persons, inspired the third assumption of Hammond's argument, the Pareto principle,. As volitional individualism was pointed to as the main reason for the impossibility of interpersonal comparisons, the solution for the problem of interpersonal comparisons is to be found in abandoning volitional individualism. An alternative is introduced here. It is called realistic individualism. One common held reason for the view that interpersonal comparisons are beyond the domain of knowledge is because we cannot know what is going on in the mind of someone else. It is a sort of scepticism about other minds [Hare, 1981, p. 118] [Little, 1950]. This can take two forms: either there is no knowledge possible because there are no other minds and one is a solipsist, or one admits there are other minds but one denies to have knowledge about them because these other minds are essentially hidden. The former will not be dealt with explicitly. Economists are not solipsists. But even more important, reasons for rejecting the latter form of scepticism are also reasons that show the former to be not very convincing. So let me turn to how this scepticism can be overcome and look at arguments for the possibility of making statements about other minds. One argument is that in every day life we normally make judgements on how others fare. It is not considered nonsense if we say that someone su ers from a headache. A physician might be punished if he neglects the utterances of someone who tells him he su ers an uncomfortable feeling in his chest, some sort of `heavy compression on his chest', radiating to his cheek and his right arm. This information calls for action by a physician. Working as a physician is only possible by taking these statements about feelings and su ering seriously. Without being able to understand them he could not work. One could wonder how a physician can be held liable to punishment by neglecting statements on other minds while some economists advising the same government which make him liable to punishment, deny the possibility
3.4. IMPOSSIBILITY REJECTED; REALISTIC INDIVIDUALISM 101 of such statements? What would these economists say if a physician denied them too if they entered his surgery with their discomfortable feelings? Should the physician be paternalistic and deny their convictions on the impossibility of knowledge about other minds, or should he follow them, and reply they can inform him about their uncomfortable feelings but he wonders whether they have these feelings really? Some physicians did get warnings from the medical council for less. It is puzzling indeed, and either physicians should stop or change their work and stop listening to and talk about feeling pain, and not only physicians should do but we all should stop listening to such expressions, or these economists should review their assumptions. I will argue for the latter. The view holding that we cannot know about other minds is based on the idea that we can only know something about our own minds. It is commonly held that our own minds are accessible to ourselves but not to others. It means that what we refer to by our language in talking about our own mind is not communicable, it is not generalisable. Consequently, the idea, for example adhered to by Hare, that we `know' about other minds because of analogy between what we experience by ourselves and we believe others have as experiences, as in extrapolation [Hare, 1981, p. 119], is false. This was made clear by Wittgenstein in paragraph 293 of Philosophical Investigations: If I say of myself that it is only from my own case that I know what the word `pain' means - must I not say the same of other people too? Moreover, how can I generalise the one case so irresponsibly? Now someone tells me that he knows what pain is only from his case! - Suppose everyone had a box with something in it: we call it a `beetle'. No one can look into anyone else's box, and everyone says he knows what a beetle is only by looking at his beetle. - Here it would be quite possible for everyone to have something di erent in this box. One might even imagine such a thing constantly changing. - But suppose the word `beetle' had a use in these people's language? - If so it would not be used as the name of a thing. The thing in the box has no place in the language-game at all; not even as a `something': for the box might even be empty.- No, one can `divide through' by the thing in the box; it cancels out, whatever it is. That is to say: if we construe the grammar of the expression of sensation on the model of `object and designation' the object drops out of consideration as irrelevant. [Wittgenstein, 1945, x293] (Translation G.E.M Anscombe.)
102
CHAPTER 3. VOLITIONAL INDIVIDUALISM
By holding that we have access only to our own minds it is as if our language is private in the sense that no one but oneself can know what it refers to. It is a sort of private language, not because it is only spoken to oneself but it is essentially impossible to communicate in this language with others. It is impossible because they have no access to its references. But the possibility of such a private language is denied by Wittgenstein by his private language argument mentioned in the previous chapter.34 Suppose there is such a private language, then it has only sense if it can be used correctly. But this is not possible because it is even unthinkable that it is used incorrectly, hence to speak of correct and incorrect use has no sense and the language itself looses its sense. Thinking one is experiencing a headache would be the same as feeling a headache. There would be nothing one could turn to in order to refute that it is a headache. It would be necessarily so that if one believes it is a headache it is a headache, one cannot misname. Because misnaming is not possible in this language it is senseless. With this language one cannot make distinctions between experiences. Because there is no way to distinguish experiences there is no sense in this [Wittgenstein, 1945, x258 .]. The main reason for the idea that we have only access to our own mind is that there is supposed to be a separation between actions and behaviour on one side and what is going on in the mind on the other [Dancy, 1985, p. 71]. But this supposition is wrong. As was explained by the introduction of moral realism in the previous chapter, there is a conceptual link between the meaning of words, also words about the inner mind, and behaviour. What someone says and does is linked conceptually to the content of mental events. It is a misleading picture that there is something going on in the mind and which we later give a name. That it is misleading is illustrated by the passage cited form Wittgenstein's Philosophical Investigations about the box and the beetle, and his private language argument. Instead of suggesting there is something going on in a person's mind which gets a name, it is held that persons act and speak. Because of that we interpret their words in such a way that we ascribe to them ideas, beliefs and desires. That is meant with realistic in realistic individualism. An individualism in which wants, desires, passions and references are not considered as having their origin in a well of volitions, but are interpreted and so restricted by interpretation.35 Realistic individualism does not back the Pareto principle as volitional individualism did, but it suggests principles in which allocations are evaluated that are based on how the life of individual persons is a ected by the allocations beyond the individual desires as in the Pareto 34 35
See p. 45. See also for example, [Davidson, 1986] [Hurley, 1989, p. 84-122].
3.4. IMPOSSIBILITY REJECTED; REALISTIC INDIVIDUALISM 103 principle.36 In Davidson's theory it was stated that we can only interpret and communicate with others if we assume to live in the same world, to perceive the same things and assume with the principle of charity, that other persons are as rational as we are. We understand another by interpreting what he does and says. These interpretations are only possible by assuming agreement about a vast amount of statements, whether these are about our beliefs or our desires. One could say that interpersonal comparisons cannot do without intrapersonal comparisons and deny in fact the former an independent status. Intra-and interpersonal comparisons are knitted to each other. As illustrated by Wittgenstein's arguments, the latter cannot do without the former.37 A similar point is made clear by McDowell: We can understand an impression of `inner sense' in which the concept of, say, pain, is drawn into play as an awareness of the circumstances that the subject is in pain. The structure of awareness and object is appropriately in place just because the subject does not conceive what it is for her to be in pain - the circumstance that is the object of her awareness - exclusively in terms of the `inner' or rst-person angle on that circumstance that constitutes the awareness of it. She understands that the very same circumstance is thinkable - by someone else, or by herself at di erent times - otherwise than in thought expressive of `inner experience'. This gives the idea of the circumstance an independence from the awareness of it. [McDowell, 1994, p. 38] It is besides the truth that one can consistently hold, as Hennipman seems to do, that interpersonal comparisons of utility or welfare are not possible and that such comparisons are either intrapersonal comparisons [Hennipman, 1977, p. 161], or evaluations that are based on justi ed interests. Intrapersonal comparisons are shown to be indispensable for interpersonal comparisons, but the latter are also necessary for the former. Also, evaluations of justi ed needs arise from a basic common world. Justi ed interests which are interpreted as rational recognisable interests, are the basis for the interpretations of the content of the beliefs and desires of other persons. Intrapersonal comparisons and evaluations are two elements of the same radical interpretation theory of Davidson. One could argue, we indeed have some knowledge about other minds and we can make statements about other minds, about what others feel etc., 36 We will meet this idea below in chapter 6 p. 194 under the name of person a ectingness. 37 See p. 101.
104
CHAPTER 3. VOLITIONAL INDIVIDUALISM
but as mentioned before that was not denied to be possible. Economists like Robbins did not deny them to be possible but denied them a status as objective empirical statements because we lack a method of veri cation or even falsi cation. Also in the example Tinbergen gave it was not shown that interpersonal comparisons are not possible.38 On the contrary, the example assumes them to be possible, otherwise this story could not be stated. But it shows that the method in the example is not a correct way to determine equal happiness because of the possibility of misleading the research worker in the experiment. The phenomenon known as manipulability does play a role here, meaning that one can become better o by pretending having other preferences than one actually has.39 It does not mean that such interpersonal comparisons cannot be made. The example given by Tinbergen exhibits the idea that in order to these comparisons having any sense at all, there has to be some rule or procedure to establish its truth. But this idea can be considered to be wrong as shown by Wittgenstein's remarks on rules. Suppose we have such a method, it would be a rule of the form: If in this kind of circumstances C we observe this kind of phenomena , then we can conclude to these kinds of things M. (In which C and are described in non-mental observable terms, while M is an evaluative term from the languages of desires passions and preferences.) But what would this rule mean? As stated in the previous chapter such rules do not have any meaning beyond their particular applications and are only intelligible in a common practice, or a common way of life.40 So these rules will not help us. According to Davidson's radical interpretation theory, empirical and objective knowledge about other minds is not impossible; it is even a necessity for interpretation. We can admit that there is no simple general way of determining whether one is better o than another is. Such a method is not necessary; it is not of help in determining objectivity. Whether or not there is a method is just an empirical question. Such a method is not necessary for knowledge about other minds. We can admit that there is for example no known general method to establish whether the near deaf can hear as bad as the near blind can see, although we can say in particular circumstances that the near deaf is helped better by his hearing aid than the near blind is helped by his glasses. Whether that is the case, is dependent on the context, for example whether they visit a concert, a ballet, a musical, a play, or a pantomime. Determining whether persons are equally well-o in general will be dicult, but this does not mean we cannot See p. 69. The non-envy approach is not free from these problems either. They are known under the name of manipulability[Thomson, 1987]. 40 See p. 34. 38 39
3.5. SUMMARY
105
make interpersonal comparisons of what goods do to people in particular circumstances. The reason why we judge these circumstances important is because we assume such interpersonal comparisons to be possible. The realism introduced in the previous chapter makes clear how interpersonal comparisons about what goods mean to people can be made. Hence, realistic individualism is more promising for the development of an ideal of equality.
3.5 Summary In this chapter realistic individualism was introduced as an alternative to volitional individualism. Realistic individualism is the view that is based on moral realism, which holds that individual wants, desires and preferences do not have their origin in a well of volitions and are unrestricted in their diversity. They have to be interpreted, an interpretation that is bounded by what is desirable in this world. Realistic individualism holds that the evaluations of allocations should be based on how the life of individual persons is a ected by the allocations. The problem of interpersonal comparisons of how well o people are was, that such comparisons are necessary but impossible. The rst part of the problem, the necessity, was challenged by denying two assumptions of an argument spelled out by Hammond. It held that denying interpersonal comparability would lead to a dictator in the sense of Arrow's impossibility theorem. Two challenges for this argument adopted the Pareto principle representing volitional individualism and both challenges were advocated as ways of reducing the common stated con ict between liberty and equality. One way of challenging the necessity of interpersonal comparisons was based on ideas of Hochman and Rodgers in which external e ects were taken into account. But it appeared that in order to explain the external e ects interpersonal comparisons were assumed to be possible. The other challenge was the non-envy approach. It exhibited the idea that a complete social ordering is not needed and that just one acceptable allocation would suce in social decision problems. In the non-envy proposals the condition of non-envy was proposed as an egalitarian element. This condition holds that nobody has a complaint about an allocation that another is better o such that he would like to change place with him in the allocation. First, the possibility of Pareto ecient envy-free allocations was established while production was not taken into account. Second, production was taken into account and problems arose because of non-transferable goods such as talents. A simple method of arriving at Pareto ecient envyfree allocations like an equal starting point followed by an exchange on the
106
CHAPTER 3. VOLITIONAL INDIVIDUALISM
competitive market, was not available. The proposal of Dworkin, the introduction of a market of insurances, was not satisfactory because by introducing option luck and option bad luck, the history of the distribution becomes essentially relevant. An evaluation is no longer merely based on the distribution pattern. This dependency on the history of the allocation undermines Dworkin's step from hypothetical insurance markets to a tax system. Next it was shown that the problems the non-envy approach su ered from, had their origin in volitional individualism. Two proposals with weaker conditions than non-envy namely envy-minimising , and undominated diversity, were argued to be unsatisfactory within the framework of volitional individualism. An argument of Roemer was presented showing that the non-envy proposals could lead to unacceptable allocations. It happened not only to be the case if production was taken into account; it was shown that this problem was already present in the cases with xed amount of goods. It became clear that not the non-transferable goods posed the problem for the non-envy approach, but volitional individualism. The discussion on population monotonicity and resource monotonicity indicated the same origin for problems. The problems for the none-envy approach were due to volitional individualism. It appeared that the discussion on the development of the proposals itself showed a tendency to abandoning volitional individualism by accepting the possibility for interpersonal comparisons. The advocates of the non-envy approach themselves actually assumed the possibility of interpersonal comparisons in discussing their proposals. After the failure of the challenges of the necessity of interpersonal comparisons, the impossibility was questioned. The Wittgenstein-Davidson approach, notably Wittgenstein's remarks on mental terms, private language and rules, made clear how knowledge about other minds is possible. Moral realism as it was introduced in the previous chapter was seen to form a basis for realistic individualism. It is the view that contrary to volitional individualism, individual wants, desires and preferences do not have their origin in some well of volitions, but have to be interpreted and this interpretation is bound by what is desirable in this world. This common world makes interpersonal comparisons on what goods do to people possible in particular circumstances. As was made clear in the beginning of this chapter such comparisons were necessary for an ideal of equality. That there can be disagreement on particular comparisons is not a serious threat to the ideal of equality. Equality can still have a meaning in this common background and can be interpreted even if there is disagreement. Realistic individualism was introduced as an alternative to volitional individualism. This individualism suggests evaluations that are based on
3.5. SUMMARY
107
how the life of persons is a ected by allocations beyond the individual desires mentioned in the Pareto principle. In chapter 6 we will meet this individualism again in discussing the so-called levelling down objection and person a ectingness. But rst the third of the three traditional background assumption, moral value monism is discussed in the next chapter. There I take up again the issue of the meaning of responsibility for the ideal of equality; an issue which is, apart for the aim of this thesis, also relevant for the common stated con ict between liberty and equality. In the next chapter I arrive at a further determination of what should be distributed equally and end up with equalisanda that could be subsumed under the name `liberties'.
108
CHAPTER 3. VOLITIONAL INDIVIDUALISM
3.6 Appendix 1 By the formulation of Kolm of non-envy in terms of permutations a natural question is: what is the relation is between this envy-free condition and another principle of justice which uses the concept of permutation, namely Suppes' grading principle? [Sen, 1979a][Suppes, 1966] Suppes' grading principle is also based on the idea that it is not important to whom a commodity bundle belongs, and permutations are used in evaluating allocations. According to Suppes someone i judges an allocation A more just than another allocation B if there is some allocation such that the goods in B changed owner, i.e. there is a permutation , such that 8j wi (~aj !j ) wi (~b(j )!(j )) ^ 9j wi(~aj !j ) > wi(~b(j )!(j ) ) (wi (~aj !j ) meaning i's evaluation of j receiving the bundle ~aj ) So in a two person case person i considers allocation A to be more just than B if he considers receiving ~ai for him to be better than ~ai and j receiving ~aj to be better than ~bj , or he considers receiving himself ~ai to be better than j receiving ~bj and j receiving ~aj to be better than receiving ~bi himself. This grading principle was also meant not to use interpersonal comparisons of well-being. But for the reasons explained in the discussion on external e ects, the evaluations that are based on what another receives cannot do without interpersonal comparisons. What about the relation between Suppes' grading principle and the nonenvy condition? This is a rather ambiguous one. The non-envy condition implies that there is no person for which there is an alternative allocation in which the commodity bundles changed of owner is prefered. This means that the second part of the condition for being more just according to the grading principle cannot be full lled, provided that all have a similar evaluation about what another person receives as that person himself, i.e. 8i; j wj (~aj !j ) wj (~bj !j ) ) wi (~aj !j ) wi (~bj !j ) If this last condition on the preferences is not satis ed then it can be possible that there is some allocation considered as more just by all but in which there is no envy, namely when the persons are indi erent to what they receive themselves but consider the allocation more just because they evaluate it better for the other; an evaluation which can be di erent from what the persons think about it for themselves. For example, wi (~ai!i ) = wi (~aj !i); wi (~ai!j ) > wi(~aj !j ); wj (~aj !j ) = wj (~ai!j ); wj (~aj !i) > wj (~ai!i )
3.6. APPENDIX 1
109
Distribution ~ai!j and ~aj !i being a permutation of A is seen as more just than A by i and j, although A is envy-free. What about the reverse? Is the non-envy condition satis ed, if there is for no one an alternative allocation which is like the original one but with commodity bundles changed of owner, i.e. is a permutation, more just? Suppose the non-envy condition is not satis ed, does this imply that there is a person according to which an alternative allocation in which the commodity bundles change of owner is more just according to the grading principle? The following example is clear in itself. Suppose the allocation of some good of which both of them desire more than less, and of which i receives 10 units and j 5 units. Now the non-envy condition is not satis ed because j prefers the commodities of i. But according to the grading principle this allocation is not more just in the view of those persons than its alternative in which i receives 5 units and j 10. So the relation between Suppes' grading principle of justice and nonenvy is not a simple one. An allocation in which all get an equal part of an equally desired good is not judged as more just for all than an allocation in which some get more than others.41 It seems that the non- envy condition is more promising, it recognises that there is a di erence between these allocations. In the allocation in which all get an equal amount no one will prefer the part of someone else, in the other there will.
41
See [Sen, 1979a, p. 149 .]
110
CHAPTER 3. VOLITIONAL INDIVIDUALISM
3.7 Appendix 2 It has been shown, by Foley and by Kolm, that in a pure exchange economy in which production is excluded, and in which no agent is satieted (every agent always thinks more is better) and, the pair of prices and commodity bundles (~p; A) is a Walrasian equilibrium under the condition of equal incomes, A is envy-free and Pareto-ecient.42 This statement implies that in an exchange economy there exists an allocation which is envy-free and Pareto ecient, namely the one which is the allocation A of the Walrasian equilibrium resulting from equal incomes. In order to get a better idea of the meaning of this statement I present a sketch of Varian's way of showing this. [Varian, 1974]. Pareto eciency Equal income means 8i; j ~p:~ai = p~:~aj . Suppose there is a B such that for some agents j 9i wj (~bj ) > wj (~ai ) and for the other agents h wh (~bh ) =Pwh (~ah ) P Then because of the increasing property43 of wi, i ~bi > i ~ai which is excluded because we are reasoning about a pure exchange economy in which production is excluded from consideration; the total amount of goods to be distributed is taken to be constant. non-envy condition Suppose there is an agent i preferring the part of someone else j to his own, wi(~aj ) > wi(~ai ). Since (~p; A) is a Walrasian equilibrium and each agent is using his budget to the limit, we have p~:~aj > ~p:~ai, which is impossible because we assumed that all had an equal income 8i; j p~:~aj = ~p:~ai . In an exchange economy the Walrasian equilibrium resulting from equal incomes is envy-free and Pareto ecient. Because it has been proved in classical economics that there exists a Walrasian equilibrium in an exchange economy [Takayama, 1985, p. 273] there exists an envy-free Pareto ecient allocation, namely the Walrasian equilibrium resulting from equal incomes.
42 ~ p is a price vector and A is a matrix in which each row denotes the commodity bundles of a person. A Walrasian equilibrium is de ned as a price commodity matrix such that the markets are cleared; there is no excess supply or excess demand. 43 A function f is monotone increasing if f (x) f (y ) , x y
3.8. APPENDIX 3
111
3.8 Appendix 3 The example of Pazner and Schmeidler is as follows [Pazner & Schmeidler, 1974]. w1(a1 ; 1 , r1) = 11 :a + 1 , r1 10 1 w2(a2 ; 1 , r2) = 2:a2 + 1 , r2 a1 + a2 = 101 r2 + r1 0 ri 1; i = f1; 2g wi: the utility function of agent i with the arguments, the commodity bundle and the leisure of agent i ai : the commodity bundle for agent i ri: the amount of labor of agent i, which if it is subtracted from the time available yields the amount of leisure.
Explanation Pareto eciency demands: 1. r = 1. Suppose r = 1 , "; (" > 0), then 1 could make his utility level higher by changing this " into a This would mean an improvement of :" , :" = ", without changing the 1
1
11 10
10 10
1 10
1
utility level of agent 2, which contradicts Pareto eciency. 2. r2 = 0. Suppose r2 = " > 0, then by exchanging some a2 = 101 " for " leisure agent 2 could improve without lowering the utility level of agent 1. w2(a2 , 101 "; 1 , 0) is higher than w2 (a2; 1 , "). The di erence is 108 ", contradictory to Pareto eciency. Because of the non-envy condition: a 11 a + 1. 1. w1 (a1; 1 , r1 ) w1(a2 ; 1 , r2) which means 11 10 1 10 2 2. w2 (a2; 1 , r2 ) w2(a1 ; 1 , r1) which means 2a2 + 1 2a1 . Furthermore a1 + a2 = 101 :0+1 = 1. Substituting 1 , a1 for a2 yields, 3 a1 21 which is a contradiction. 4 22
CHAPTER 3. VOLITIONAL INDIVIDUALISM
112
3.9 Appendix 4 The example of Roemer against Dworkin's proposal is illuminating; it is like the one he presented against fairness simpliciter [Roemer, 1985, p. 161]. Suppose in an economy there are two sorts of people t1 and t2 with two di erent strict concave utility functions say wt1 and wt2. Suppose utility function wt1 lies above wt2 and is the steeper one. Suppose they are di erentiable and the wt0 1 is also higher than wt0 2. So the t1-people derive more utility from their earnings and have a higher marginal utility than the t2-people. The t1-people try to nd a T w 1 such that
X p :w
t
v t1(Ev + Tvwt1 )
v
is maximised subject to
Xp
w :Twwt1
w
0
The t2-people of course, try to nd the T w 2 which maximises t
X p :w w
subject to
w t2(Ew + Twwt2 )
Xp
w :Twwt2
w
0
Now because of the veil of ignorance in Dworkin's original situation it is unknown to which kind of people one belongs. This has its counterpart of course in the amended version of Dworkin's original position. Suppose the probability to belong to one kind is 12 , one half of the population is of the t1-type and the other half is of the t2-type. The maximisation problem is to maximise 1 X p :w (E + T w 1 ) + 1 X p :w (E + T w 2 ) 2 u u t1 u u 2 u u t2 u u t
subject to
X p :(T w u
v
Solving this problem yields
w u t 1 + T u t2 ) 0
t
3.9. APPENDIX 4
113
wt0 2(Ew ) + Tww 2 ) = wt0 2(Ev + Tvw 2 ) = = wt0 1(Ev + Tvw 1 ) = wt0 1(Ew + Tww 2 ) t
t
t
t
This condition has to be satis ed otherwise one could increase one's expected utility by exchanging some of the Tv with the lower marginal utility for some Tw with the higher marginal utilities. Unless marginal utilities were equal across types, there would be a possible improvement. Now because wt0 1 lies above wt0 2 and the marginal equalities have to be equal the t1-type should get more post insurance income than the t2-type. Only a higher income for the t1-types is consistent with equal marginal utilities because of the concavity of the utility functions; in those marginal utilities decrease by an increase in income. It means that those with a low utility function get even less than those who were already better o . The example shows an unacceptable consequence of Dworkin's proposal. It would be even unacceptable to a resourcist as Dworkin considers himself.
114
CHAPTER 3. VOLITIONAL INDIVIDUALISM
Chapter 4
Moral value monism 4.1 Introduction So far, I dealt with two of the three main traditional background assumptions of the ideal of equality: moral universalism and volitional individualism. Two alternatives were introduced: moral realism and realistic individualism. Both are based on the Wittgenstein-Davidson approach to language and interpretation. In this chapter, the third assumption, moral value monism, is discussed. Value monism was held to be one of the traditional background assumptions of the ideal of equality because of the idea of a contract. In contracts the goods for people are compared to each other and in contracts it is agreed that the loss of one good is compensated by some other good. If you give up your right to that which will be advantageous for me, then I will give up my right to something else in return. Such a trade-o suggests the idea that there is one value with respect to which all goods have to be compared. This moral value monism can also be discerned as the assumption behind the discussion on the proper equalisandum; it seems that it is assumed that there is just one equalisandum. In this chapter moral value pluralism is introduced as an alternative to moral value monism. It leads to the idea that there are several equalisanda instead of just one. This will have in its turn as a consequence that the articulation of the ideal of equality has to be independent of a particular equalisandum. This chapter consists of two parts. In the rst part, the reasons for accepting moral value monism are examined and moral value pluralism is introduced. I follow mainly Stocker's line of argument against monism [Stocker, 1990]. Two main reasons for value monism are discussed. One is 115
CHAPTER 4. MORAL VALUE MONISM
116
based on the idea that the main concern of morality is that it should be action-guiding and the other on the idea of compensation. Of both it is shown that they are not valid. Subsequently I present an argument against monism that is based on a discussion on the phenomenon of akrasia or weakness of will. It is shown that although in spite of what is commonly held, monism can account for akrasia. It is argued that monism cannot account for all common sorts of con icts and genuine choice. At rst sight it seems that an explanation for moral con icts and genuine choices calls for moral universalism, but as is shown, moral particularism within a moral realistic framework can account for these con icts and genuine choices too. In the second part, it is argued that there are several equalisanda instead of just one. I arrive at equalisanda by discussing an issue touched upon in the previous chapter, namely the role of responsibility in the ideal of equality and the proper equalisandum. As was argued in the previous chapter, incorporating responsibility in the equalisandum threatens the project of this study because evaluations of distributions without taking their history into account become impossible.1 Several recent proposals for an equalisandum, notably those of Rawls' primary goods, Dworkin's resources and Cohen's advantages, are discussed. Scanlon's idea on the equalisandum exposed in his Preferences and urgency [Scanlon, 1975] and his idea on responsibility explained in The sigini cance of choice [Scanlon, 1988] are presented. They are of help in arguing that there are several equalisanda and that responsibility is not to be incorporated in the ideal of equality. Thus the threat to the aim of this study, the development of an evaluation of distributions that is pattern like, is neutralised. In this discussion on the proper equalisandum and responsibility, I arrive at a plurality of equalisanda. They can be subsumed under the name of `liberties', referring to real possibilities for individuals to act or enjoy situations that are valuable, without su ering from disadvantages that others do not su er from.
4.2 Monism versus pluralism 4.2.1 Reasons for value monism
One of the main reasons for accepting monism is that it is held to be necessary for a de nite answer to the question: `What to do?', to which the answer is of course: `Choose the best option!'. A complete ordering of available options is a guarantee that there is a best option that should be chosen. It is held that such a complete ordering assumes just one value, which is the value according to which the options are evaluated. If more values 1
Chapter 3 p. 89.
4.2. MONISM VERSUS PLURALISM
117
have to be taken into account then a best option is not any longer guaranteed and some arbitrariness is held to be inevitable and morality looses its de nite action-guiding character. Once taking morality as essentially action-guiding, one value has to be assumed otherwise it could happen that morality ended up with arbitrary choices, thus the argument. This value is taken to be the central value of morality. In history it can be seen that several suggestions have been made for this value, happiness, eudaimonia, welfare, to mention only a few. Action-guidingness as the central idea of morality is given as one reason for monism. The second reason for value monism is more or less particular for the discussion related to equality. It is based essentially on the same assumption about comparing options and goods as the one of the rst reason. In using the ideal of equality the idea of compensation is a central one. It is held that it has to be possible that lack of some sort of good can be compensated by some other good in order to arrive at equality. Comparing several goods in order to compensate, assumes one value with respect to which the goods and lacks are compared. If there existed more values then there would be no guarantee that the lack can be compensated properly by some good. Wiggins for example formulates this idea as the principle of compensation in kind: ..., if course x is better in respect of eudaimonia than course y, then there is no important disadvantage that x has in comparison with y, or no desirable feature that y o ers that x does not o er too, by way of an equal or greater degree of that very feature [Wiggins, 1982b, p. 259] Such a principle can be seen as a reason too for value monism. Both reasons are discussed subsequently.
De niteness of morality As mentioned above, one reason for monism is the idea that morality should give a de nite answer to the question: `What to do?' A complete ordering of options will secure a de nite answer, because a complete ordering always has a best element. This element, the best option, is of course the answer. But it is not clear at all that the only concern of morality is to give de nite answers to the question: `What to do?' And even if it is, it is not clear that the best has always to be chosen and even if the best has to be chosen it is not clear why a complete ordering on one scale means value monism. It could be possible that there is a complete ordering without there being one supervalue which is responsible for this ordering.
118
CHAPTER 4. MORAL VALUE MONISM
Let me rst turn to the idea that ethics is concerned with de nite answers to the question: `What to do?'. It can be admitted that ethics is concerned with this question, but it is not its only focus. Morality is concerned also with judging persons, virtues, or aspects of actions instead of the action itself, or that it is concerned with what is worthwhile to strive at. It is not merely the particular action which is of concern to morality but also the question: `What if the circumstances were di erent?' For example suppose as a derivative of Rawls' theory that the one and only value to be taken into account is the well-being of the worst-o person.2 Now it can be argued, and in fact some do, by pointing to incentives, that we should divide goods not equally, because then the worst-o would be even worse o . It is better even for the worst-o that goods are not divided equally, because equality would lead to a decrease of the incentive to work and so to a diminished total amount of goods to divide, resulting in the worst-o receiving less. Thus is argued that there would be less to divide because the incentive for production would be far less with an equal distribution. Against this reasoning, one can hold that persons should not be moved by their own income but they should be moved by the well-being of the worst-o . If that value were the incentive one could and should distribute goods equally. The incentive argument would not be any longer a reason for di erences. On the other hand, one knows that people should, but in fact are not motivated to work for the well-being of the worst-o and so in fact the worst-o will actually end up worse o in case goods are distributed equally and the distribution should not be equal. In this example one could say that you should divide the goods equally and you could say that you should not divide the goods equally. The former gives the answer to the question: `What has to be done idealiter?', the latter: `What has to be done given that people inhabiting this world are not perfect?' But even if one admits that given the feasible options, goods should not be divided equally, it still makes sense to remark that idealiter goods should be distributed equally. An equal distribution is not simply worse than an unequal distribution. It is worse because we are not perfect. That is an important supplement to the judgement that an unequal distribution is better. The judgement that it is better not to divide goods equally, is not complete. One aspect is missing by neglecting namely the judgement that idealiter equality would be better. Not all moral judgements are action-guiding, they also concern judgements. Admitting this, one reason for the requirement of monism in ethics is undermined, although it is not yet shown to be false. 2 This example is derived from Cohen's interpretation of Rawls' di erence principle in which the role of incentives cannot be used as a phenomenon to which the di erence principle can be applied if it is cited by those whose incentives it concerns [Cohen, 1993].
4.2. MONISM VERSUS PLURALISM
119
Let me for the sake of the argument, admit that morality is concerned exclusively with the question: `What to do?'. After all in this study I am concerned with evaluating options i.e. distributions, with respect to the ideal of equality. The next question then will be: `Is monism the appropriate way of arriving at de nite answers?' It is clear that if one has a complete ordering of available options there is at least one maximal element that is best. A common de nite answer to the question: `What to do?' is: `The best available option.' Although maximisation of some good and striving for the best is seemingly obviously the right method, in fact it is not the only right one. It is argued against the maximisation view, according to which the option with most of the value should be chosen, that it can lead to worse results. For example, perfectionism leading to spending an in nite amount of time on some task in order to ful l it perfectly well, will lead in the end to nothing. Maximisation is, to borrow the concept from Par t, indirectly self-defeating. It learns that in order to arrive at the best we should not strive for it [Par t, 1984, p. 5]. It would disturb in a sense the relationship with ourselves. We could not describe directly to ourselves what we care about i.e. the best. The relationship with other persons would be disturbed too. For example, giving up friendship just because one can develop a better one undermines the whole idea of friendship. This is at least as serious as indirect self-defeatingness. Best can be the enemy of good. Beyond these problems for the maximisation view, supererogation is mentioned against this view. With supererogatory acts are meant acts that are not obligatory but nevertheless very good. Such acts are performed by moral saints. The argument states that the distinction between what is very good and what should be chosen cannot be made in the maximisation view. The phenomena cited above could be accounted for if not only the best should be striven at but `good enough' would be enough and should be aimed at, i.e. some sort of satis cing theory. In this view perfectionism is not required, it does not require to give up some friendship for a slightly better one and supererogatory actions can be seen as those actions being better than just good enough. The maximisation view can o er two arguments to rebut these attacks. It can argue that maximisation is still not to be rejected because of the examples cited to show maximisation is wrong, because the examples can be accounted for in the maximisation view. The other argument takes the satis cing theory to focus on the wrong value that is assumed to be central. These arguments are discussed subsequently. The examples cited above can be accounted for in the maximisation view by considering the actions which have to be chosen according to the satis cing theory, namely to be the best of the available options. Not being
120
CHAPTER 4. MORAL VALUE MONISM
a perfectionist is the best option available to us. Not changing friends just because one person would be a slightly better friend than another, is the best available option. Similarly, the alternative which is not supererogatory is the best available option given the fact we are not saints, although of course it would be better if we were saints; of course we can be ashamed of not meeting these high standards. Supererogatory acts can be accounted for in the maximisation view as actions of which it is a pity we cannot perform them because we are not perfect, just as in the case of admitting that the option of distributing goods equally should not be chosen, because it is not the best option given the fact we are not motivated as we ideally should be. The conclusion is that the examples cited to show the maximisation view is not covering moral phenomena, are not conclusive, the phenomena can be accounted for in the maximisation view. But is this answer sucient to save the maximisation view? It can be argued that the maximisation view is still not saved, because this view, in which the option which should or could be chosen is considered to be the best option, is still defective in a certain sense. The satis cing theory can make more and ner discriminations than the maximisation view. If for example, an option which has enough of some value F in order to be chosen, then according to the maximisation view this option would be maximal, and it would not make any sense to say that another option has more value. In the satis cing view one can say that this option has enough, but some other option also having enough value in order to be chosen, has even more value. In the maximisation view, these two options would have an equal amount of value, one could not discriminate between them. In the satis cing theory one can. It allows for more and ner discriminations and makes this view preferable, because it ts better in our ways of reasoning, in which we make these ner descriptions. It could be argued that these ner discriminations could be made also by the maximisation view. It could be held that beyond what is best for doing, an option can have more of some other value than the value `to be performedness' than the other option. But by this response value monism is left and value pluralism is turned to and the motivation for the maximisation view is undermined. Maximisation is confronted with some problems even if the examples of self-defeatingness of perfectionism, friendship and supererogatory acts can be accounted for. Let me turn to the other argument that is mentioned in defence of the maximisation view. Against the non-maximisation view cited to question the maximisation view, it is argued that if some option with less of the good than the maximum is better than the option with the maximum, then there is some other value with regard to which options are judged and not just the good which
4.2. MONISM VERSUS PLURALISM
121
was meant to ful l that role. To be more formal, if an option with a certain amount say n of the value F is better than the option with the most amount of this value say m, then this betterness of that option assumes some other value V so that V (nF) V (mF) even if m > n. So the wrong value was taken to be the only one, it should be V instead of F. By taking the right value V instead of F the maximisation theory appears to be the right one. Although the reasoning of the maximisers seems to be quite convincing, there is still a problem. This reasoning shows too that maximisation is not conceptually true either as is claimed by the maximisers, because saying that some option is better than some other option because it has some more value V , also assumes some other value W in which this `betterness' of the option is captured. The maximisation view leads to an in nite regress. However contrary to Stocker's opinion, this is not the main problem, for the regress argument can be rebutted by reinterpreting the value of the maximisers. The value meant by the maximisers can be taken such that it is necessarily maximised. Its maximisation is appropriate by de nition, and thus the in nite regress is stopped.3 The main problem is that maximising this value will not help us in decisions, it represents the end of deliberation about what to do. The possibility of a satis cing theory, just be happy with enough and not strive always to the maximum, can be interpreted in such a way that the maximisers can stick to their idea that there is just one value which is to be maximised. But this will sacri ce the meaning of this value for choices, it will represent the outcome of deliberation and monism is no longer leading to a de nite choice, it is the result of deliberations and does not have any role in that deliberation. It would not make any sense that without monism we cannot determine what we should do, because what we should do is not determined by this monistic value but by some other process of evaluation. Of course there is a sense in which one can say without monism no de nite solution, but this is only a strict logical sense similar to the way in which the dictator in Arrow's theorem is a dictator in a special sense. He might even not know that he is a dictator, or that he is in uencing the social ordering in a direct way; the orderings of society and the dictator happen to be congruent.4 In a similar way the option that is maximal on the ordering is decided to, but not because it is maximal; it could have been a di erent one that would then be maximal on the ordering. So it is argued that the maximisation view could be saved but only at the expense of making the monistic value dependent on the deliberation itself. It does not help in the deliberation and thus the main reason for the 3 4
See also [Broome, 1991, p. 17] against Philippa Foot See chapter 3 p. 66
122
CHAPTER 4. MORAL VALUE MONISM
maximisation view is undermined. The reasoning shows that one can stick to monism and the view that all options can be ordered along a complete ordering, but the value which should exhibit this ordering is not helping in decisions; it does not o er reasons for choice; it represents choices. One could wonder whether it is a value at all for which we care. We can choose upon other values as in the pluralistic variant by deliberation. This deliberation can result in a complete ordering without there being any supervalue at all, but just an ordering about what to do. It is not shown that this ordering is based on merely one supervalue, it is just re ecting the result of deliberation and decision. Actually, it are the other values we care about and with the help of these, we try to evaluate the options in particular circumstances. Summarising, the reason for monism that is based on the idea that morality is concerned exclusively with action-guiding judgements, and because of this, de nite answers have to be given, namely being a maximum of a complete ordering of options, whereby the ordering exhibits one value, is not quite a valid reason. Morality is not concerned merely with actionguiding judgements, as is shown by the possibility of moral con icts in an imperfect world. But even if it were, the maximisation view is not the only view possible, the satis cing theory could be o ered as an alternative. The maximisation view could be maintained but at the cost of loosing its relevance for deliberation and choice and the reason for monism that it is of help in answering the question: `What to do?' would be undermined. A complete ordering of options does not imply monism. There can be a complete ordering without there being one value being responsible for this ordering. Of course one could give the ordering a name for example `to be performedness' but such a `value' would barely have the status of value because we do not care about it. We care about the other values which have to be weighted in particular circumstances; the supposed supervalue `to be performedness' does not help as it supposed to be. Let me turn to the other reason o ered for monism that is rather speci c for distributive contexts.
Comparing and one supervalue
The second reason for monism is rather particular for contexts concerning equality. In these contexts the concept of compensation is important. It is held that lack of some good can be compensated by some other good. These comparisons of lacks and compensations call for one value that is responsible for these comparisons. Without this value, such comparisons are argued to be impossible. As a consequence of this thought, it is seen
4.2. MONISM VERSUS PLURALISM
123
that the discussion about the equalisandum, what has to be distributed equally, is essentially about the nature of this value. Mainly two ways of comparing are in uential, one is in terms of monetary value, the other is more directly in terms of welfare. In the previous chapter, we met these already in discussing the non-envy analysis as resources and income which are held to be easily comparable and welfare which is held to be non-comparable.5 According to the rst, money is the good to which all others are compared. The value of all goods is measured in terms of money, it is the price of the goods and that price is determined by the free and competitive market. It represents how much the people care about a good. The value is determined by the alternative uses of this good i.e. the monetary value. The value is measured by what one could do by trading that good for some other good. It shows the purchase possibilities, which on their turn show the possibilities of choice. It will be clear that this view is associated with a resourcist theory of distribution according to which the equalisandum are the resources. We met this kind of theories in the previous chapter. So, one proposal for the central value in distribution problems is money. According to this proposal the prices settled by the market determine the just compensation.6 The other way of comparing lack and compensation is not through money but by some rather abstract value. Compensation and lack are compared on some scale such that the lack is considered to be compensated if and only if the option with lack together with the compensation for this lack, is equivalent on that scale to the option without this lack and without its compensation. This comparison is more directly linked to what some good or lack means to someone, it is not directly inspired by the idea of alternative uses as in the previous method. One could call this value welfare. The question to which we have to turn here is: is this reason for comparisons sucient for monism to be accepted? Let me turn rst to the method which is mediated by the free and competitive market and subsequently to the more direct method of comparison. The rst method of comparing lack and its compensation by the marketprices is criticised on the ground that this way of measuring is too much in uenced by irrelevant in uences i.e. by for example how other persons evaluate some good.7 If almost nobody cares about something, for instance See chapter 3 p. 67. This idea of compensation is for example also seen in Nozick's idea of compensation [Nozick, 1974, p. 63]. Kolm argued that this idea ends in egalitarianism [Kolm, 1985, p. 162 .]. 7 See [Kolm, 1984, p. 60 .] for some problems of the idea commonly advocated in 5 6
124
CHAPTER 4. MORAL VALUE MONISM
philosophical ideas, the demand for it is negligible and consequently almost worthless. Furthermore, prices are determined not only by the demand in particular circumstances but also by the supply. Water in the desert is much more valuable than in Holland nowadays. This means, that because supply and demand are not intrinsic features of goods, the prices are not intrinsic features of these goods. Goods cannot have prices in general but only in particular circumstances. The prices are dependent on the particular situations, which are external to the good. Hence this way of measuring the value of something is not re ecting the value of the good itself. A response to this objection of prices being too much arbitrary could be that goods have to be particularised to a particular time and situation, meaning that water in the desert is some other good than water in Holland on one of its rainy days. This will account for some of the external in uences, but the in uence of how others care about something is not neutralised. For example, loosing a photograph having merely value to you but to nobody else would mean that you lost nothing valuable. Such a loss has not to be compensated. But apart from the problem that the value of something that has only some value to you would be determined by what others want tot give for it, monism is not yet argued to be reasonable. Suppose one could in a reasonable way compare goods mediated by money, whether or not via market prices, would this mean that money was the only value? The answer can be clear, monism would not be implied. Money is just a way of measuring a certain amount and it does not mean it is the only value. Instead of money chickens could have been chosen, which does not mean that only chickens matter in this world. Even if all values can be represented by money, then it is not implied that money is the only value. So, the method of comparing goods through prices on the market is not a reason for adhering to monism. Let me turn now to the second method of comparing lacks and compensations. By the second method the lack of some good and the compensation for it is more or less directly compared. If the option with lack together with compensation for this lack and the option without lack and without its compensation are equivalent, then the compensation is sucient for the lack. A complete ordering will, in combination with some sort of continuity property for compensation, ensure that there exists for each lack a sucient compensation. This complete ordering is, just as in the previous section, re ecting a value. As in the previous section, we can argue that this value is not the reason for the compensation to be sucient for the lack.8 The liberal environments that the value of a good is determined by the market prices. 8 It is tempting to take this constructed `value ' such as `to be performedness' as
4.2. MONISM VERSUS PLURALISM
125
compensation is not mediated by asking for equivalenve regarding this value but by considering the lack and compensation directly. It is not asked how much of this value is lacking and thus such and such compensation is right. The deliberation is performed on the ground of evaluating the lack of some good and the possible compensations in a particular circumstance. Here again the idea that a complete ordering implies there to be just one value, is used to conclude to monism. But as stated already, in order there to be a complete ordering in a particular situation of options, the assumption of there being just one supervalue is not a necessary one. It is not implied that all options from all situations can be arranged in a complete ordering on the basis of one value as would be held by monists. It is not valid to conclude from the idea that for all situations there is a complete ordering or scale of comparing based characterised by a particular value or good to the idea there is one ordering or scale for all situations characterised by one value. It is quite intelligible that in di erent situations, goods are di erently evaluated vis-a-vis each other. Actually, in the previous chapter this was illustrated. It was shown that the evaluation of dioptrics against ampli cation of acoustic signals is dependent on the particular situation, i.e. whether the visit to the theatre concerns a pantomime, a concert or a play.9 The reason for monism that is based on comparing and determining compensation for lack of goods is not a convincing reason. It is even more serious for monism. It is not only that monism does not follow from the possibility of comparing di erent goods, it makes even talking about lacks and compensations impossible. In monism in which only one value is acknowledged there is hardly lack and compensation, there is just less, or more, or an equal amount of one value. The above-cited principle of compensation in kind recognised by Wiggins, assumes that there is one value relevant, which implies that there is no important distinction to be made between the option with lack together with compensation and the option without lack and without compensation, they just have an equal amount of the one and only value and that is all.10 The whole idea of compensation for some particular lack is not intelligible, there are just amounts of value, whether these are instantiated by a particular combination of goods or some other combination, is the equalisandum. However as is clear from the main text here and below it is an empty equalisandum hiding what we care for. It does not make intelligible why it should be divided equally beyond the values we care for that form the basis of the constructed arti cial `value'. It does not add anything and it hides the `desiderative structure discussed on p. 129 . in this chapter 9 See chapter 3 p. 104. 10 See the principle of compensation in kind cited at p. 117.
126
CHAPTER 4. MORAL VALUE MONISM
not relevant at all. All goods are only to be seen as instantiations of this singular value. And in case of compensation there is actually no reason to speak of compensation of some lack. By these compensations in kind there will remain no trace of the lack as we normally recognise there to be. For instance, being compensated with money for having no longer personal photographs, does not exclude regretting having no photographs. This regret is not intelligible if one is a monist. A monist can only regret having less of the one and only value than one could have, but not that it consists of this composition rather than that; only the amount of the value is relevant; its way of instantiating is not relevant at all. This means that, although until now I undermined the reasons for monism, I touch here upon an argument against monism namely that monism cannot allow for our way of talking in a sensible way about lack and compensation. This argument is turned to in the next section.
4.2.2 Reasons for moral pluralism
In the previous sections, I have shown that the reasons for monism were not convincing. Pluralism is still an option. It was not yet shown that monism is less satisfactory than pluralism. In the last section, one argument against monism was touched upon, namely the fact that monism cannot account for lack and compensation. In this section I discuss this kind of argument against monism more extensively. The introduction to this eld of arguments against monism is the phenomenon of akrasia, weakness of will, which consists of doing something of which it is clear, also to the actor, that it is not rational to do. For example staying in bed while it is clear that getting up is much better. This phenomenon is cited in order to show that monism is wrong, for example by Wiggins [Wiggins, 1982b, p. 262 .]. Davidson stated a view on akrasia, which also accounts for moral con icts in general [Davidson, 1970]. But as is shown shortly this view of Davidson elaborated by Jackson lures us to moral universalism [Jackson, 1985]. So, with akrasia we meet two theoretical problems: monism versus pluralism and moral universalism versus particularism. The main problem now is to articulate a view that can account for moral particularism, moral pluralism and moral realism, of which it was argued that it would be convenient for the development of the ideal of equality. Let me turn to the phenomenon of akrasia for the discussion monism versus pluralism.
4.2. MONISM VERSUS PLURALISM
127
Akrasia and moral pluralism In akratic actions showing weakness of will, for example remaining in bed while it is known to be better to get up, the actor is acting irrationally. On the other hand, there is something in the akratic action which makes it attractive; it is not just irrational. Remaining in bed gives some pleasing bodily sensation of warmth and rest that is disturbed by getting up. But if there was merely one value then it is not explicable why the actor remained in bed; there would be nothing attractive to that action. The akratic action has less value than getting up, so there would not even be a single reason for the actor to choose for the akratic action; it would be highly unintelligible. But akratic actions are known to be possible and also tempting and not merely irrational. It seems that only intervention of some other value can account for akratic actions. Getting up is better, but the bodily sensations interfere. Brie y, this is the reasoning against monism based on akrasia. This argument based on akrasia is not valid in this form. Some comments have to be made in order to get the problem of akrasia clear. It can be admitted that something has to intervene between seeing what the best option is and choosing the akratic action. But it is not necessarily so that some other value has to intervene. Akratic actions although understandable, are nevertheless irrational actions, one could for example, point to some rather common psychic defect. It is not without reason that it is weakness of will, which is in a particular way irrational and not to be admired. Hence it is not clear at all what akrasia has to do with monism. The real problem will become clear if it is recognised that within monism there is never genuine choice at all. With genuine choice is meant a choice between alternatives which are both attractive in their own way. Was there merely one value, a maximiser could not describe choices as choices at all. Only the best would be attractive, there would not be any genuine choice. The alternatives bearing less than the maximum of value are not really alternative options, and if the alternative options are equivalent the di erences don't matter and there is no genuine choice either. So the problem pointed to by akrasia is not akrasia and its irrationality, but concerns differences in the attractiveness of actions. It is even more clearly seen in far more common situations of choice and deliberation which do not concern irrational actions. The reason why akratic actions call for pluralism also show why the more common cases of genuine choice, even if there is no con ict of reason, call for pluralism. So far, it is still not yet shown that akratic actions require pluralism. That is not by chance, because it cannot be shown. If some other way than maximisation of the monistic value is agreed to be possible, for example a satis cing theory, then akratic actions can be accounted for within
CHAPTER 4. MORAL VALUE MONISM
128
monism. The akratic action could be considered to be just satisfying while an alternative would be judged to be better or even best. Deviating from the maximisation view will not require pluralism for explaining akrasia. And as made clear by Stocker, sticking to the maximisation view is not of any help in making clear how within pluralism akrasia is explained [Stocker, 1990, p. 242 .]. Suppose one is a pluralist and a maximiser, then if the values in the akratic action and the alternative action that is better are not comparable, then the statement that one option is better than the akratic cannot be made seriously. On the other hand were the values comparable than the akratic action would loose its attractiveness vis-a-vis its alternative in the maximisation view. So, akrasia is not explained by pluralism either. The call for pluralism is not dependent on the phenomenon of akrasia.
Moral con ict, genuine choice and moral pluralism
Instead of arguing in favour of pluralism because of akrasia one can better argue for pluralism because we recognise that there are moral con icts and genuine choice. Although as stated already, it is not true that all forms of moral con ict are impossible within monism, a limited sort of con icts is possible. The con icts which can be accounted for in monism, as was stated above, are those for example in which it is regretted that there were no other alternatives available than there actually were because of the imperfection of the people inhabiting this world.11 But we recognise there to be other forms of con ict that can be detected, even in normal reasoning in which there is genuine choice. Precisely this was seen to be impossible in monism. Monists could argue that some more con icts than those because of imperfect people, are possible. If there is just one value then it is not excluded that there can be a con ict about who is getting that value, or to whom it belongs, or applies. As for example in the argument to answer Williams' argument against the perception view of ethics in which doing philosophy was the only value, there was still a con ict possible, because doing philosophy is not something merely abstractly in the air, but is linked with persons doing philosophy and it can be the case that one person doing philosophy excludes another doing the same.12 Con icts can arise because the values can be linked with di erent people, thus a monist. Although it seems to be a nice defence for monists, it is not a good one. It is even showing monism to be wrong. The indexicality of some value with respect to persons points immediately to the issue of a fair distribution. The question of the distribution has to be solved by invoking 11 12
See p. 118 in this chapter. See chapter 2 p. 49.
4.2. MONISM VERSUS PLURALISM
129
some other value than the indexed one. The indexed value cannot function itself as a canon for its own distribution, just because it is indexed. Some other value has to be invoked, for example equality or some other distribution relevant notion. It can even be stated that the whole idea of equality presupposes value pluralism, namely beyond the value `equality', some value that has to distributed equally.13 In this light it is even strange that the whole discussion about the equalisandum seems to assume there to be just one equalisandum, while monism has already to be abandoned for acknowledging the ideal of equality. Returning to the possibility of con icts for monists, it is concluded here that monists can not account for con icts which came into existence through indexing the value. Similarly, indexing some value with respect to time and place will give the same problems as indexing with respect to persons.14 Summarising monism can allow for some con icts but not for genuine choice and moral con icts resulting from con icts of reasons. But what do these con icts and genuine choices show? How should they be described? Re ecting on what monism is missing, namely the capacity of making moral con icts and genuine choice intelligible, leads to realising there is lack of so-called desiderative structure, a term used by Pettit. Pettit uses this term in order to indicate some defects in decision theory as a full theory of explanation of actions. Like decision theory in which that action is to be chosen which maximises expected utility, monism explains that an action has to be chosen by citing: `It is the best'. But this explanation is hardly an explanation because an option is necessarily taken to be best if it is chosen, just as in decision theory it is necessarily the option with the maximum of expected utility. The reason why remains obscure. Within momismactions are seen as bearing one value and all other properties are not relevant for deliberation and choice. Only being a bearer of a particular amount of that value is interesting. So, any explanation in virtue of what an action is seen as good is missing. But this reasoning is defective because in our daily life we do deliberate on actions and evaluate them because of their particular properties. They are right or wrong in virtue of This recognised by Par t in [Par t, 1989]. Stocker argues that di erences in time are mostly di erences in the composition of some good and so contradicting value monism according to which composition does not matter at all. It matters much if someone has a desert, say ice, after dinner or between the aperitif and the soup. The whole meal changes, it is not merely a matter of time. In this way time does matter and a change in the ordering of time changes the whole good and the whole value of some complex good. Here too indexicality will involve pluralism, because of the change of the whole meal by permutating the parts of a meal in a di erent order, the value is changed. There is not one good indexed to times but there is some value consisting of the whole complex, just as was seen to be the case in the issue on distribution resulting from indexing the value to persons [Stocker, 1990, p. 250-260]. 13 14
130
CHAPTER 4. MORAL VALUE MONISM
something. And this something is the reason for considering the action as desirable or not. In virtue of these properties actions are further deliberated about or not. Those properties matter and are not merely the bearer of the single value re ecting for example the value `to be performedness'. Apart from repairing the lack of reasons for choice by this deliberative structure Pettit argues also in favour of this structure because it can explain why in attitudinal propositional contexts di erent descriptions of the same action cannot simply be substituted without altering its truthvalue. For example if telling a certain lie in a particular situation happens to be the same action as saving someone's life, then the description of saving a life cannot be substituted simply by the description telling a lie. `I am happy to have saved a life' is not similar to `I am happy to have told a lie'. These cases of non-substitutability show properties of actions to be relevant. Another argument of Pettit for this desiderative structure is that choices which seem to be irrational are by considering them more precisely and acknowledging moral relevant properties, appear to be perfectly rational. For example one of the axioms for rationality is the sure thing principle meaning that if some alternative is preferred just because of the discerning properties of the alternative it is also in another situation in which it is also the discerning property. For example if you prefer a bet 1 to 2: 20% chance: 80% chance: 1: an ice cream of strawberry liquorice 2: an ice cream of vanilla liquorice then it would be strange if you did not prefer bet 3 to 4: 20% chance: 80% chance: 3: an ice cream of strawberry a bar of chocolate 4: an ice cream of vanilla a bar of chocolate Of course independent of what is won in the 80% chance case, you prefer strawberry ice to vanilla. So the sure thing principle is formulated as a rule of rationality. Consider now the following example from Diamond, in which you prefer distributing some good by lottery to giving it directly to one of to persons Mary or John. This would be irrational. Because if you prefer bet 1 to 2: head: tail: 1: Mary wins John wins 2: John wins John wins (John gets it whatsoever) then you should, following the sure thing principle, prefer 3 to 4:
4.2. MONISM VERSUS PLURALISM
131
head: tail: 3: Mary wins Mary wins (Mary gets it whatsoever) 4: John wins Mary wins But because of fairness, you introduced the lottery and you just prefer 1 to 2 and 4 to 3. The properties of the options are relevant, not only the bare options characterised by their abstract formal structure. Desiderative structure in which properties of options play an important role, is proposed by Pettit to complement decision theory. In the desiderative structure, actions are judged in virtue of their properties, which means generally being good qua action being of a particular type say instead of good simpliciter. This makes much more con icts possible than allowed for by monism. It also makes genuine choice possible. In this view, an action can be called prima facie desirable because it has some features. Con icts can arise if an action is simultaneously prima facie desirable and prima facie undesirable, i.e. if there is a reason why it would be good to perform the action and a reason why it would be good to forego the action. But this should be carefully stated. What does prima facie good mean? An action of type being prima facie right could mean that if about some action it is only known to be of type , it should be chosen. Similarly, prima facie wrong because of being of type means that if the action is only known to be of type , it should not be chosen. This is a hint to the right meaning of prima facie but it is not capturing the idea of prima facie correctly. Stated in this way it is an epistemological concept. Against this epistemological interpretation Jackson states that if one knows that a certain action is wrong in general, and chooses that action and in the particular circumstances foreseen by the agent with some degree of reliability, it turned out to be right then there is no moral con ict. There are no traces which we are familiar with in con icts, although this action was considered as prima facie wrong, because the knowledge of the badness in general and ignorance of the particular circumstances should have lead to abstaining from that action. The situation is more like being told that some action is wrong and someone else telling you it is right. But this epistemological problem is di erent from the genuine moral problem, which is not merely epistemological, because in these some traces remain if an action has to be done which is bad in some way. It is not like the reasoning about Tweety that in learning that Tweety is a bird and a penguin something of the capacity of ying remains, the whole idea of the capacity of
ying is cancelled by learning Tweety is a penguin. It is di erent with moral con icts, there traces do remain. If you save a life by lying, it still remains true although the actions was right, that it had something regrettable,
132
CHAPTER 4. MORAL VALUE MONISM
namely that it was a lie.15 Thus compensation of the lack of some good in some option is compensation and not just compensation in kind as in the principle of compensation in kind.16 But if this epistemological meaning of prima facie is not right, how should prima facie be explained? One could argue that some action is called prima facie right (or wrong), if it is right (wrong) if all other things considered the same, the action would be better, (worse). The action would be considered prima facie better in virtue of some property if in comparing the state of the world in which all other valuable aspects remain independently the same, the option with would make it better. In itself this cannot be of help to explain moral con icts, because there is nothing said about the relation between the judgements based on a property in situations in which the other aspects are kept the same and judgements of situations in which these others aspects are varied too. Moral universalism as was seen in the introduction, would be a natural answer. being better than : means that under all conditions C in which all other aspects were kept constant (C; ) is better than (C; : ). Thus in every context an action which lacks lacks something which an action with does not lack. This reasoning using separability of values as in moral universalism, can help to account for moral con ict and genuine choice. But now we have a problem because this account contradicts moral particularism. Moral con icts are now accounted for, but in a way not consistent with moral particularism. So far, it has been argued that moral con ict and genuine choice demand moral pluralism. Without pluralism there would be a lack of what Pettit called desiderative structure. In this structure actions are judged by their properties that refer to the idea of being prima facie right or wrong. But the idea of being prima facie right or wrong seems to be based on moral universalism. So, the question is now: how to account for moral con icts and genuine choice accepting moral pluralism in a way that is consistent with moral particularism?
Moral con ict, genuine choice and moral particularism
At a rst glance it seemed that particularism would be consistent with pluralism which can account for moral con ict. By indexing the values to the particular situations pluralism seemed to be the result; choosing not in the 15 This di erence between epistemological con ict and moral con ict is described by Hurley in [Hurley, 1989, p. 130 .]. Pro tanto reasons do not cancel each other completely because they grew from di erent reasoning systems each with its own central value. Also Dancy argues that there is a di erence between moral con icts and epistemological con icts in his argument against induction as a way of arriving at moral knowledge [Dancy, 1983]. 16 See p. 117.
4.2. MONISM VERSUS PLURALISM
133
light of more of some abstract value called `friendship', but between this or that friendship. Because of this indexing to particular situations the desiderative structure dissolves and only the value seen in the relation `better to be performedness' remains, which cannot allow for con icts. Actions were merely bare bearers of this value `to be performedness'. It appeared subsequently that other reasons for actions invoke moral universalism by taking the judgements in situations in which all other aspects were kept constant but the property , to be relevant for situations in which these other aspects C are di erent. But moral particularism denied this reasoning to be valid. It is denied that judgements in situations in which other aspects were held constant can be applied directly to situations in which they are di erent. There seems to be a dilemma: either being a pluralist and being capable of accounting for moral con icts but accepting moral universalism, or being a particularist but not being capable to account for con icts. How to solve this? Moral particularism denies that if something is a reason, or is valuable in one situation it is necessarily a reason or valuable in any other situation. But moral particularism cannot of course, on the cost of becoming senseless, deny that what is important in one situation can be important in an other. Judgements in di erent situations are of course related to each other. If this was denied, then a disastrous fragmentation of value and meaning was implied. Nothing intelligible was possible any longer. Even a simple description of situations would become impossible. By describing situations, situations are compared to each other. If the terms used in descriptions would have di erent meanings in di erent situations, we would loose sense of meaning, whether it concerns moral or non-moral terms. Terms, whether they are moral or non-moral, have to be interpreted as is stated by Davidson's radical interpretation view on language. They get their meaning through interpretation for which the principle of charity was indispensable. We assume we live in the same world. Similarly, we can see this principle to be important for interpreting what is said in di erent situations by ourselves or by others. We can consider each speaker to be particularised to situations. Once this is realised, the principle of charity will care for the relation between moral judgements in di erent situations. The principle of charity means that in interpreting we should take the other or the indexed other as similar as ourselves in our situations, assuming we see the same things, we care about similar things etc. Otherwise, we could not understand what was said. This was not a choice up to us, but we do, otherwise we would be solipsists not capable of understanding others, not even ourselves. Does this mean that radical interpretation and the principle of charity do imply moral universalism? No, it does not.
134
CHAPTER 4. MORAL VALUE MONISM
Moral universalism holds that a reason in one situation is necessarily a reason in any other too.17 Moral particularism does not deny that a reason can be a reason in some other situation, but it is denied to be necessarily so. It could have been di erent, it all depends on the situation. In one it is a reason, in another it is not. The principle of charity does not imply that in all interpretations there should be the same similarities as basis for interpretation. They can be di erent. In for example contexts of politics, social inequality is seen to be an important value in another for example in dangerous situations, safety will be a most important value. Some situations are seen as mainly determined by the context of politics and some by the context of danger, rescue and survival. Here the idea of moral vision can be of help. Moral particularism will not deny there is some relation between judgements in di erent situations, but it denies the necessary character of it. It is not said that one can do without common reasons. It can be held that in every comparison or description there are some common values or reasons, but it can be denied that there is some set of reasons or values necessarily the same in all. So, it is admitted that for all limited sets of situations there is a set of common reasons and values, but not that there is one set of common rules for all sets of situations.18 The recourse to moral universalism in order to account for moral con icts was premature. Judgements in one situation can be relevant for other situations, and this is sucient. Moral universalism was not necessary. It is sucient that it is acknowledged that there are relations between judgements in di erent situations. In contrast to the moral universalist, who does not even recognise the problem, the particularist has of course to answer the question how we know in which situations some reasons are relevant and in which not. The answer has to be found in epistemology. In the second chapter a coherentist view was argued for, according to which something is knowledge if it ts in more properly with other things already known than its alternative, whether it concerns moral or non-moral knowledge. Moral perception will play a role. By the question `How do we know when 17 A rather popular idea that the adjustment that the importance of the reasons can vary is inconsistent. If reasons are important in all situations than that is also valid for the reasons for a particular weighing of these reasons. Consequently, di erent weights are not possible, because that would require di erent reasons applying to the situations. Non-monotone reasoning could be a way out but it does not save moral universalism as was argued in chapter 2 p.33. 18 In order to preclude some misunderstandings namely that all situations are to be considered as similar, we should bear in mind that this is not implied, just as the principle of charity does not imply there are no di erences between people. It is acknowledged that there are di erences and that people have di erent wants, but in order to discern these di erences, one has to assume a lot of similarity. Similarly, situations di er but in order to discern the di erences it has to be assumed that they have a lot in common.
4.3. EQUALISANDA INSTEAD OF ONE EQUALISANDUM
135
reasons are relevant and not?', it should not expected to nd a general rule which can be applied as a kind of algorithm. As was explained in the previous chapters such an algorithm is not likely to be found as explained by coherentism and particularism.19 Summarising, moral particularism can account for con icts by accepting that judgements in di erent situations can be related to each other, i.e. one reason in one situation is relevant for another. Although it is admitted that it is necessarily so that some judgements are related to each other in di erent situations, without these relationships there would even no sensible descriptions possible, it is denied that these judgements are necessarily related in this way. The dilemma, either accounting for con icts but losing moral particularism or keeping particularism but not being able to account for moral con ict, can be solved by realising it is not a genuine dilemma. We can accept moral pluralism within a particularistic framework.
4.3 Equalisanda instead of one equalisandum So far, I argued in favour of moral pluralism meaning that there are more values than one. I have shown that the arguments for monism based on the narrow view on morality, that it has to be action-guiding, and the idea that comparability implies one value, are not valid. Furthermore, I argued that monism is defective because it cannot account for moral con icts and even not for genuine choice. It was shown that monism lacks a desiderative structure. So moral value pluralism is a more promising starting point in developing ideas about the equalisandum or even better, equalisanda. Value pluralism suggests already that there are more equalisanda than just one, contrary to what is assumed in the discussion on the proper equalisandum. In the discussion on the proper equalisandum, ingenious examples are constructed to show that some particular equalisandum is not the proper one. For example, in arguing that well-being cannot be the proper equalisandum, it is argued that it would lead to a strange distribution if some person was only happy by eating caviar and living in a palace, and someone else was as well-o as the former in his best days, by just sitting in front of his little house, some piece of straw between his lips, looking at the sea extending before him. The former would get too much of the resources than would be fair, and it is argued consequently that well-being cannot be the proper equalisandum but that resources are. Against the latter view that being a resource is the de ning characteristic of the equalisandum some other examples can be given. As will become clear, pluralism is of help in solving this discussion about the equalisandum. 19
See chapter 2 p. 46 and chapter 3 p. 104.
CHAPTER 4. MORAL VALUE MONISM
136
This discussion on the proper equalisandum concerns for its most important part the incorporation of responsibility in the equalisandum. As was argued in the previous chapter, incorporating responsibility is a serious complication for the project of this thesis, i.e. an evaluation of distributions without taking into account the way they came about. Because a pattern of distribution could be the result of di erent choices of individuals and even by gambling, as was illustrated by the Babylonian lottery, responsibility cannot play a role in pattern principles. Responsibility poses a problem for the ideal of equality as treated in this study. In the subsequent section, I discuss the proposals of Rawls, Dworkin and Cohen. They suggest an equalisandum in which responsibility is incorporated because as they argue, neglecting responsibility will lead to an unequal distribution. The situation in which somebody pays continually for the pleasures of his lazy incontinent neighbour represents such an unequal distribution. Thus it is argued that egalitarian ideas themselves are the basis for taking responsibility into account. These ideas can also be discovered in the debate in Holland on the health insurance, which is aimed at equal care for all. It is suggested for example by the Minister of V.W.S., Borst, that persons who are responsible for their injuries by doing sports should pay for their own treatment by physiotherapists, in order to make as much available for all as possible.20 I argue against these proposals in which responsibility is densely interwoven in the equalisandum. My argument is based on Scanlon's ideas on what should be distributed equally and his ideas on responsibility. I arrive at equalisanda. This is possible because of moral value pluralism. The consequence is that the meaning of the ideal of equality has to be articulated independent from a particular equalisandum.
4.3.1 Subjective welfare as equalisandum
A well-known idea of equality is equality of welfare in which a distribution is considered to be right if everyone has reached an equal level of welfare whereby the level of welfare is determined from the point of view of each person's own tastes and preferences. This principle assumes a subjective criterion for welfare levels. The level of welfare of a person is determined by that person himself. This idea of equality has some attractive features. In the rst place, it takes seriously the idea we met in the previous chapter, that goods are not goods simpliciter but that goods do mean something to people.21 And because goods do mean something to people, dealing with 20 21
Algemeen Dagblad, 7-3-1995 See chapter 3 p. 67.
4.3. EQUALISANDA INSTEAD OF ONE EQUALISANDUM
137
goods on their own is not sucient for determining a fair or equal distribution. In the second place, it takes into account a variety of tastes, interests and preferences. This feature is particularly important in a society in which there is a great diversity of values to which people are committed, as is taken to be the case in our society. It is commonly held that by turning to equality of subjective welfare, totalitarian intervention in personal development in order to reach consensual preferences can be avoided. It is held that freedom and autonomy are not necessarily overthrown by the ideal of equality of subjective welfare. Although equality of subjective welfare is an appealing idea there are several problems. Apart from the problem of interpersonal comparisons, with which I dealt in the previous chapter, I mention two that are particularly important in the coming dispute on equalisanda: o ensive tastes expensive tastes The rst objection against subjective welfare as the equalisandum holds that there is no way to exclude claims of sadists and racists for compensation if they endure for example lack of sadistic or racist pleasure. Such claims that are based on o ensive tastes should not count for establishing an equal distribution. However, adherents of equality of subjective welfare have no reason to exclude these claims. The mere possibility of these offensive tastes form an objection to subjective welfare as the equalisandum. The second objection concerns expensive tastes. It holds that it is not right according to the idea of equality, to transfer commodities from relatively poor but content people to rich people who are discontent because of their re ned expensive unsatis ed desires. Again it is argued that in the idea of equality of subjective welfare there is no reason for excluding such expensive tastes and no reason to condemn such transfers. The recent discussion on the proper equalisandum was focussed on these two objections to subjective welfare.
4.3.2 O ensive and expensive tastes excluded by social ideals
One reason why o ensive tastes should not be taken into account by establishing an equal distribution is that such tastes are contrary to the ideals by which they give rise to claims of compensation. O ensive tastes are excluded by the social ideals of justice. They should not be given any weight in distribution problems. It is even argued that utilitarianism based on
138
CHAPTER 4. MORAL VALUE MONISM
subjective welfare can exclude o ensive tastes because such tastes are contrary to utilitarianism itself. It is better that no one has those tastes in a society considered from an utilitarian point of view [Rawls, 1971, p. 3031]. Generalising this line of reasoning, one arrives at the idea that an ideal P using subjective welfare, can exclude tastes and preferences which are contrary to P itself. So it can restrict itself to ino ensive tastes. This formulation has a circular appearance. How to determine whether tastes and preferences are contrary to the ideal if it is not clear which tastes to include? This question however, shows merely an apparent objection. It is possible to compare the situation in which the ideal P is satis ed and in which all tastes are included with the one in which o ensive tastes are not present. The latter could be better according to the ideal P, because satisfaction of those o ensive tastes harm others. For some ideals this might be the way to exclude o ensive tastes, but for the ideal of equality of subjective welfare I am sceptical. Let me use the above stated method to circumvent circularity in the case of the ideal of equality of subjective welfare. Let me compare two situations. In one situation a sadist is satis ed in his sadistic preferences and his level of subjective welfare equals that of all the others let us say n units. I assume that such a comparison is possible just for the sake of the argument. In the other situation the sadist's preferences are neglected and not satis ed altogether and his welfare is m units equally for all. Is the former worse according to equality of welfare? Of course all might be worse o , n < m, because of compensation to the victims, but that is not relevant here. The question here is whether regarding the ideal of equality the former is worse? The answer is: `No it is not.' Both situations can be considered as equivalent. In the latter the ideal of equality is just restricted to taking only ino ensive tastes into account. It could be worse for the sadist, but only in a way which is not relevant, because his sadistic preferences should be neglected. Circularity is not avoided by the comparison method in case of utilitarianism. However this answer is not nal. If it is acknowledged that the ideal of equality of welfare is just one among other social ideals, it is possible to exclude o ensive tastes; not by the ideal of equality itself but by the combination of ideals in the way the comparison method indicates. Ideal P refers in this case to a combination of ideals. Suppose we have a combination of utilitarianism or perfectionism and equality of subjective welfare.22 It is possible that o ensive tastes should be neglected by the ideals themselves. It is possible that by compensating the sadist for his lack of sadistic pleasures, the total amount of 22
See [Nagel, 1991] for what is meant here by perfectionism.
4.3. EQUALISANDA INSTEAD OF ONE EQUALISANDUM
139
welfare is lower in which case there is a reason for neglecting those tastes because it is contrary to the ideals particularly utilitarianism. Of course, in theory it remains possible that it is better to take o ensive tastes into account in case the pleasures generated by o ensive tastes outweigh the costs in terms of welfare of others. So o ensive tastes are not de nitely excluded but I believe the sting is removed from the objection that subjective welfare theories lack any reason for excluding o ensive tastes. It has been shown that there can be a reason for the exclusion of o ensive tastes. O ensive tastes can be excluded by social ideals. Rawls' theory of justice, particularly his theory on primary goods can be referred to as an example. The primary goods he proposes as candidates for being the equalisandum are not a ected by the problem of o ensive tastes. Although the primary goods can be seen as derived from subjective welfare, o ensive tastes are excluded by his social ideals even before they can get any weigh. This is demanded by his method of constructing social principles which is essentially based on reasonable morally inspired people [Rawls, 1971] [Rawls, 1985]. What about expensive tastes? The reasons why they should be excluded could be that such tastes fall within the domain of volition. Although it is not said as in volitional individualism that all preferences have their origin in a well of volitions, it is stated that moral persons have some part in forming and cultivating their nal ends and preferences. It is likely the case that those with less expensive tastes have adjusted their lives and their nal ends to their income or wealth that they could reasonably expect. Consequently, it is regarded as unfair that they should receive less in order to spare others from the consequences of their lack of foresight or self-discipline [Rawls, 1985, p. 168 -169]. It is because of responsibility that expensive tastes should be neglected in determining equal distributions. This idea too is incorporated in Rawls' theory on primary goods. Being a primary good as de ning characteristic of the equalisandum proposed by Rawls does not su er the problems of o ensive and expensive tastes.23
4.3.3 Objective welfare as equalisandum
Are the aforementioned reasons for excluding o ensive and expensive tastes because of their anti-social content and their voluntary nature respectively 23 To prevent misunderstandings on my description of Rawls, these objections to equality of subjective welfare are not the decisive reasons for Rawls to condemn subjective welfare theories; they are just consequences of his argumentation. His main argument for primary goods is based on the idea that primary goods are necessary conditions for realising the powers of moral persons and are all-purpose means for a suciently wide range of nal ends of reasonable morally inspired people. These people determine the basic ideas of social justices according to Rawls.
140
CHAPTER 4. MORAL VALUE MONISM
indeed the appropriate reasons for exclusion? The former can be questioned because not all o ensive tastes are anti-social; morality demands more of us then just conformity to the social moral rules.24 For example, the person who does not want to have any moral feelings at all has an o ensive preference for being inhuman, although it is not a preference which is necessarily contrary to social moral rules. The latter can be questioned because reasons for excluding o ensive tastes can be also applicable to expensive tastes and could put aside the reason based on responsibility for one's expensive tastes. As is said above, morality demands more from us than mere conformity to social moral rules. It suggests that o ensive tastes have to be excluded directly. They are not worth to be taken into account. So, there is invoked a system of ideas by which tastes and preferences are judged as valuable or worthless. It means that the idea of equality of subjective welfare has to be abandoned and that we should use objective judgements about the importance of what goods do to people. Objective here means the contrary of subjective and refers to the basis of interpretation of preferences as explained in the previous chapters in radical interpretation and realistic individualism. Objective judgements on the level of welfare of a person means that these judgements are independent of this person's tastes and preferences; they may even con ict with those personal tastes. Subjective judgements are not to be seen as unimportant altogether, but their value is determined by objective judgements on for example the value of freedom of choice and the value of the freedom of making mistakes. Subjective judgements are not any longer directly relevant for determining equal allocations they are relevant only via an objective evaluation. This view is plausibly illustrated by Scanlon in taking it for granted that we judge a claim for help for building a religious monument by somebody who foregoes a decent diet for it, not a priori as strong as a claim for aid in obtaining enough to eat [Scanlon, 1975]. Someone's own judgement on the importance of claims is not decisive. To be clear, it is not held by Scanlon that both claims, the one for help for the building and the one for obtaining food, should not count at the same degree of urgency, but it is not a priori necessarily so because of their being subjective evaluations. The degree of urgency depends on the objective value of such a religious building. According to the view that objective welfare is the proper equalisandum o ensive tastes are excluded directly by objective evaluations. Furthermore expensive tastes can be treated similarly. On the basis of those objective judgements, they do not have a high priority in a distribution problem. Contrary to Rawls' proposal, they should not be excluded because people 24
See for example P.F. Strawson's article: Social morality and individual ideal.
4.3. EQUALISANDA INSTEAD OF ONE EQUALISANDUM
141
can be held responsible for them, but because they are less urgent. That people can develop some preferences and can be held responsible for them, can be taken to mean that people can do without them; they are not necessities and so loose some of their urgency. Responsibility for preferences in itself is not a reason for rejecting claims, it is at most a sign of being not very urgent if one considers objective welfare as the equalisandum. In short, Scanlon holds contrary to Rawls that responsibility is not densely interwoven in the tissue of egalitarian ideas, he keeps them disconnected. Scanlon's objective judgement view is simpler. It uses only one reason for excluding o ensive and expensive tastes and explains the intuition on which the former proposal for exclusion was based. Of course, this proposal has to be supplied with a theory that accounts for objective judgements. Scanlon mentions two sources for such objective judgements, ethical naturalism and conventionalism. The latter he thinks the most plausible one because in that one consensus has a morally proper place, which is denied in the former. In the preceding chapters an alternative view was presented, namely moral realism developed within a radical interpretation framework. This view can account for objective judgements, i.e. moral judgements being true just because they are true and not because someone is convinced that they are true. What about this idea of objective welfare being the proper equalisandum?
4.3.4 Resources as equalisandum
The idea presented by Scanlon immediately suggests the question: `How can we determine on the basis of these objective judgements whether a distribution is equal?' This question is not easily answered. How should one compare the welfare of the optimist with the welfare of the pessimist? What about the comparison of the loss of welfare or regret because of not leading the ideal life of an ambitious person who judges his actual life hardly valuable with the regret of a less ambitious person who judges his actual life as quite reasonable. Should the ambitious man be compensated for regretting his actual life more than the less ambitious person? It is reasonable that compensation should be given, proportional to the degree of di erence between people's regret leading the actual life instead of a life which could be reasonably expected. The determination however of this reasonable expectation is impossible without assuming some distribution of resources to be reasonable. Because what kind of life is reasonable to be expected is dependent on the resources one can reasonably expect to have. So, an idea of entitlements to resources is assumed. Without such an idea no meaningful judgement on the equality of welfare is possible. This does not only hold for subjective welfare but also for objective welfare as
142
CHAPTER 4. MORAL VALUE MONISM
it appeared in Scanlon's proposal. Dworkin states: Any pertinent test of what someone should regret about the life he is in fact leading, even on the best rather than his own theory about what gives value to life, must rely on assumptions about what resources an individual is entitled to have at his disposal in leading any life at all [Dworkin, 1981a, p. 225]. On account of these arguments, equality of welfare, whether it concerns subjective or objective welfare, is dismissed as an adequate theory of distribution by Dworkin. The only acceptable theory of objective welfare for Dworkin is not a theory about what is valuable but consists of a list of goods including for example physical and mental competence, education and opportunities as well as material resources etc. Such a theory is a resourcist theory in welfare language. The objective judgements which satisfy the requirement that on the basis of them we can determine whether a distribution is equal, are judgements on resources, thus Dworkin. In a resourcist view, o ensive tastes and expensive tastes are not excluded because they are not judged as valuable, but because tastes and preferences should not be taken into account at all, only resources should. The example of Dworkin concerning expensive tastes is illustrative. Suppose Louis lives in a society in which there is realised a state of equality of welfare. Now Louis is developing and growing older and happens to get the belief that it would be better for him if he adjusted his tastes and he developed some other more re ned tastes which happen to be more expensive to be satis ed and he is aware of that. If they were satis ed his welfare should raise considerable but if they were not satis ed his welfare would become lower. For these new insights he can hardly be held responsible. To hold that one is responsible for what one considers to be true or valuable is at least highly questionable. Although Louis cannot be held responsible for his new insights, he is responsible for acting upon these believes i.c. developing these expensive tastes. Louis has a choice between two options: stay in the position he is now in and refrain from developing the expensive tastes develop the expensive tastes but get less enjoyment than in the status quo; so accept a lower level of welfare. It is wrong according to Dworkin to give Louis a third option in which he develops his tastes at the costs of other persons, because this would diminish other person's shares of resources [Dworkin, 1981a, p. 237]. Expensive tastes are excluded with an appeal to responsibility in a resourcist theory
4.3. EQUALISANDA INSTEAD OF ONE EQUALISANDUM
143
of distribution. It suggests a tight connection between responsibility and the idea of equality of resources, because if Louis had no choice it would be appropriate to judge that the distribution was not egalitarian at all and therefore he should receive some more of the resources. In this resourcist theory responsibility is densely interwoven with the idea of equality of resources. It becomes clearer if we follow Dworkin in his distinction between brute luck and option luck; a distinction we already met in the previous chapter.25 Brute luck happens to somebody, it is not the result of a choice. An example of brute bad luck is a storm blowing o the roof of somebody's house or lung cancer developing in the course of a normal life. Option luck, to the contrary, is the result of a voluntary undertaken action. In case of for example the loss of money in a lottery, or lung cancer after a life of heavy smoking, one has chosen an option that turned out to be an unsuccessful gamble. Brute bad luck should be compensated, optional disadvantages should not. Goods desired by one person can also be valuable to another. To keep a distribution right from an egalitarian point of view people should pay the price of the goods they voluntary desire. They should pay the price for the life they have decided to lead. This price is measured in terms of what others give up in order the former can do as they want. This idea is the basic element of equality of resources, which explains why optional disadvantages should not be compensated, but brute bad luck should [Dworkin, 1981b]. Responsibility is considered to be essential in the ideal of equality of resources. By neglecting responsibility unequal distributions will result, thus is argued by Dworkin. In this resourcist theory responsibility is densely interwoven in the idea of equality.
4.3.5 Access to advantages as equalisandum
We could wonder whether this interpretation of how well-o people are in resourcist terms, is appropriate; there are good reasons for doubt. In a recent article Cohen shows that neither welfare nor resources are the proper equalisandum but that the judgements on which the distribution should be based concern access to advantages [Cohen, 1989]. That welfare is not a proper equalisandum is shown by an example of a very well to do person who has many opportunities to welfare but su ers from a pair of paralysed legs. According to equality of welfare theories his high level of welfare is a reason for rejecting his claim for help in the form of a wheelchair. But as Cohen holds, his level of welfare has nothing to do 25 In the previous chapter I pointed already to the problems with this view, here Dworkin's theory is cited because of the relation between responsibility and the equalisandum.
144
CHAPTER 4. MORAL VALUE MONISM
with his paralysis. It is a disadvantage such that it is right to compensate for it. So, equality of welfare is not the proper equalisandum according to Cohen. In a similar way it is shown by an example that being a resource is not the de ning characteristic of the equalisandum either. The example consists of a person who does not lack any capacities and has enough resources who can move his arms very well, but who su ers from a terrible pain after moving his arms. According to equality of resource theories this man has no claim to a medicine which could relief his pain, he does not lack any resources for which he should be compensated. To describe the person as a person lacking the capacity of moving his arms without pain is welfare talk in resourcist language and will not do. But not lacking any resources he really su ers from a disadvantage and so according to Cohen he has a claim to the medicine. Cohen concludes that the equalisandum should not be characterised by resources but by access to advantages. The proposed characterisation of the equalisandum, access to advantages, does not su er from the o ensive tastes objection. Unsatis ed offensive preferences are not disadvantages in themselves. But contrary to the resourcist theory such tastes could count and should count if they are disadvantages like cravings and if a person would accept for example psychotherapy to get rid of them. For the costs of such a therapy he should be compensated. How about expensive tastes? About these it is argued that they should be taken into account to the degree they are involuntary. If they are not developed freely these expensive preferences should be taken into account. In Cohen's view there is a reason for helping or for giving compensation such that expensive tastes can be satis ed but only if these tastes are not the result of a choice. If the preferences result from a choice no compensation should be given. If Paul likes photography and John likes shing and photography happens to be more expensive then shing, Paul should receive some compensation to the degree he is not responsible for his preferences and not responsible for the fact that satisfying these preferences is expensive. Responsibility cancels any claim on compensation. According to Cohen it is unjust and contrary to the idea of equality that people should be exploited for the pleasure of others. This would be the case if expensive tastes for which a person himself is responsible, were compensated. Exploitation is contrary to the idea of equality, responsibility is important, thus Cohen. In the resourcist view sketched above as proposed by Dworkin, the distinction between resources and welfare, or context and person, or actual circumstances of life and ideas about what makes life valuable, is the morally important distinction. That what is the result from particular
4.3. EQUALISANDA INSTEAD OF ONE EQUALISANDUM
145
ideas about what makes life valuable should not be compensated, only lack of resources or particular circumstances of life, or in other words the extra personal elements of the context should be compensated for. Tastes do not count, resources do. But according to Cohen this distinction is besides the point. The more fundamental distinction is between involuntary disadvantages and disadvantages which one begot as a result of one's own choices. Dworkin's reason for rejecting claims referring to welfare; after all these claims are based on ideas about what makes life valuable and thus on intrapersonal characteristics; are plausible in case these claims were voluntary developed. If the pessimistic and ambitious nature were voluntary begotten they should not count, but if this nature is forced upon persons they should count, thus Cohen. In short, Cohen holds that we should take only involuntary disadvantages into account. As a consequence it is not against equality to help a person building a religious building if he is not responsible for having this particular belief. It is not against equality even if we do not endorse the value of the particular religious believe. It is however a di erent matter if the person because of this particular religious belief has to su er for building the monument and that su ering is part of his religious commission. In that case the su ering is so intrinsically connected with his religious project that it is hardly defensible that he should be compensated for his su ering. Cohen holds that the equalisandum is access to advantages, which demands compensation for those disadvantages which are not due to the subject's choice and which the subject would not choose to su er. The last part of the circumscription is because of the example in which the su ering of the person is an inherent element of his project by which he su ers and which should not be compensated. In Cohen's line of reasoning responsibility is decisive for the compensation demanded by equality. It is given a central role in the idea of equality, but is it a right role? Isn't it strange that egalitarianism should lean so heavily on such a complex notion as responsibility by which metaphysics about free will and determinism is introduced in questions of political philosophy? Cohen does not believe this to be a reason for rejecting his proposal. If equality is complex we should not deny that complexity. Furthermore he assures us that the question of free will or choice is a matter of degree. In a footnote he refers to Scanlon's Tanner lecture: The signi cance of choice, in which free will and choice is treated in a non-metaphysical way. He wonders whether Scanlon's approach can be used to improve a theory of distribution a la Dworkin. As will be made clear below, it cannot be used as such. On the contrary, it o ers further support for theories of equality in which responsibility is not densely interwoven.
146
CHAPTER 4. MORAL VALUE MONISM
4.3.6 Equalisanda The idea that responsibility determines what should be compensated is illustrated in the discussions by examples about big spenders, lazy persons, people with irrational or even immoral desires, in short examples in which the actions are not very praiseworthy. Bad actions, or irrational actions, do raise the question of responsibility in a particular way, but one should not forget that if the notion of responsibility is applicable at all, it is certainly applicable to rational actions, actions which are performed because of good reasons [Strawson, 1986, p. 33]. Actions that are morally right belong to this category and if we are responsible at all, we are at least responsible for morally right actions. By shifting our attention from examples with irrational and immoral actions to actions which are morally praiseworthy, it is illuminated that responsibility is loosing its important role in determining the equalisandum. Let us look at an example with a morally right action with disadvantages for its performer in which, in my opinion, the claim on compensation for the disadvantage is not forfeited because of responsibility. Suppose you are walking alongside a canal in an expensive suit that you lent from a neighbour because of some important event. A child playing near the canal falls into the water and it appears the child is not able to swim. You jump into the water knowing your suit will be spoilt for ever and you should pay for that. The reason you jumped into the canal is part of a particular morality in which saving children from drowning has some place. This morality was not installed in you; you developed it voluntary. Your jump was not a re ex. You jumped into the water to save the child in full awareness the suit would be spoilt; you realised you could not have jumped into it, but then the child would have been drowned, a result which is considered to be awful in the morality you developed for yourself. As the disadvantage of spoiling the suit is clearly traceable to a choice; the disadvantage itself was of course not chosen, it was not part of the project of saving the child; compensation is not demanded by equality according to Cohen's proposal. The same goes for Dworkin's proposal in which the distinction between option luck and brute luck is relevant here. However I do not believe these judgements are reasonable. Should this saviour's disadvantage not even be partly compensated, merely because of his choice to jump? One could argue that compensation is reasonable not because of the idea of equality but because morally right actions should be awarded and not be punished. Morally right actions should be stimulated by awards. By refusing any compensation one will probably destroy the willingness to perform morally right actions. And thus the example above does not show that compensation is required by the idea of equality. This view,
4.3. EQUALISANDA INSTEAD OF ONE EQUALISANDUM
147
which I call the award view, has to compete with another view, which I call the possibility to a normal life view, which can be considered to be nally more in the line of ideas of those who argue in favour of the central role of responsibility in the ideal of equality. The latter view is based on arguments for compensation because of equality that is embedded within moral realism. The man in the example, who jumps in the water, is like the man who is able to move his arms but su ers from pain afterwards. The man in the latter case can freely move his arms so he could hold them relaxed in which case he does not su er pain. In this case too the disadvantage is traceable to a subject's choice, not in the sense that he chose the pain but in the sense that he could avoid it. Similarly, a morally right action with consequential disadvantages is voluntary undertaken although the disadvantages are not chosen at all. Having pain after moving one's arm is a disadvantage just as the bad consequences after a morally right action is a disadvantage. It suggests that we should care about capabilities that people have. Some of them should be cared for in the sense that they should be possible for all equally. Some activities such as morally right ones we see as valuable and should not be denied to some people because the risk of disadvantages following such actions is too costly for them to bear. Some activities should be possible for all. If it is admitted for morally valuable actions why not for many more? There are many more actions that can be seen as valuable that bring with them a risk of disadvantages. These disadvantageous consequences are not chosen. The probability distribution on the possible consequences is not under control of agents. They would of course prefer doing the action without any risk of disadvantages.26 But in all those cases the disadvantages are seen as the result of voluntary undertaken actions and should according to the ideas in which responsibility determines the equalisanda, not be compensated. The example of saving the child shows that it is not right to deny compensation for disadvantages resulting from voluntary undertaken actions because they are voluntary begotten. Responsibility does not have the role it has been given. This is also illuminated by the example of the person suffering pain after moving his arms and the person su ering lung cancer which develops in the course of his normal life. The lung cancer in the course of a normal life is considered by Dworkin as brute bad luck in contrast to a life in which one smoked heavily which is considered to be optional bad luck. In other words, it is the result of an unsuccessful gamble. What count as a 26 A similar line of reasoning can be discerned in Van Parijs' argument for real liberalism, in which the essence of liberalism has to be concentrated on protecting real possibilities instead of formal opportunities.[Van Parijs, 1991b, p. 156,184] [Van Parijs, 1995].
148
CHAPTER 4. MORAL VALUE MONISM
normal life is of course determined by what we consider to make life valuable. The ideas on what is valuable are necessarily presupposed. They are the more fundamental ones. Without these, the distinction between bad luck and option luck is impossible to understand and similarly the point of Cohen's examples is dicult to apprehend. The examples which are meant to show the importance of responsibility for equality are apparently convincing because the actions undertaken are not praiseworthy or valuable or necessary for the development of a normal life. They are irrational, bad and they do not deserve to be cared for. It would be even better if they were not done at all. It is certainly not according the idea of equality to care for an equal possibility to do these bad actions. For many other actions it is di erent and they are components of a valuable way of life. The actions which are a component of a valuable way of life, should be the subject of our concern. These should be available for all because they are worthwhile. That what is valuable for people is also worthwhile to be distributed equally. These consideration lead me to the suggestion that possibilities to perform actions, to e ectuate choices, or to enjoy situations are the proper candidates for equalisanda. One could call these liberties. These should be equalised. Liberties are those choices, actions and possibilities that are considered to be valuable; they are the normal elements of a person's life and so in a sense necessities for a normal way of life. That they are valuable is the reason that they should be possible for all.27 Because of that, they are subsumed under the name of liberties and not the other way round. We should care that they are equalised. Something is an equalisandum in virtue of being valuable and not because of a special characteristic. Because of moral value pluralism we can conclude there are equalisanda instead of one equalisandum. A consequence of the view that what is valuable is also an equalisandum is that responsibility has some role in equality but merely a secondary role. It is not a decisive role, its role is dependent on the judgement on the value of choice and responsibility and on some idea about in what degree someone should su er from his wrong actions as a kind of punishment in order to discourage these actions. But why prefer this view to the alternative which I called the award view in which disadvantages that are morally praiseworthy should be compensated because such praiseworthy actions should not be discouraged? There are some arguments in favour of the former. As was stated above, all the interpretations of the examples that are meant to point to the importance of responsibility given by Cohen and 27 See chapter 6 p. 199 for the idea in accord with moral realism: what is valuable for me is valuable for you too.
4.3. EQUALISANDA INSTEAD OF ONE EQUALISANDUM
149
Dworkin need some assumption on what is valuable in life and what counts as a normal life. One element of a normal life is of course the possibility of performing right actions. Compensation should be given not because good action should be awarded but because they are an element of a normal life in this world. This was the possibility to a normal life view. In the award view responsibility forfeits any claim on compensation for disadvantages due to one's own choices but this does not comply with the idea that one wrong choice should not have such harsh consequences. By applying the award interpretation, the natural jungle would be replaced by the social jungle. One liberty that looks important to me is the liberty to make some mistakes. A world in which everybody is neurotically anxious avoiding mistakes and in which every mistake results in a disadvantage, is awful. There should be some liberty for performing mistakes. How much depends on our view on that matter. The idea that responsibility cancels compensation for disadvantages is not at all an adequate view on equality. The nal argument in favour of the possibility to a normal life view against the award view is derived from Scanlon's treatment of responsibility in his Tanner lecture: On the signi cance of choice. An example mainly due to Scanlon can be used to make the point clear. Suppose some area is seriously polluted. The material is planned to be removed and some precautions have to be taken because during this work some material will be dispersed into the open air. If somebody inhales the material he will get pulmonary problems. So, everybody should stay inside his home; the doors and windows should be closed. Those living in the neighbourhood are warned by television, radio and newspapers. Of course it is inevitable that some people are a ected by a pulmonary disease because they were outside their home during the cleaning of the area. One person did not receive any message about the danger; he didn't know anything about the removal. On the evening of the dangerous work he left his home as he did every evening. Another person is very curious and did not belief the seriousness of the warnings. He left his home in order to see how this piece of work was done. A third person went outside after he weighed the reasons of pro and con thereby taking the risk of a pulmonary disease into account and judged his going outside to be rational. A fourth was used to go for a walk but had just forgotten the warnings. The fth saw the other persons and ran outside to warn them and probably prevented some more pulmonary damage to the victims.
150
CHAPTER 4. MORAL VALUE MONISM
All those described above did leave their home voluntary by their own choice, no one was forced. The pulmonary problems are traceable to choice Is this really the reason of forfeiture of compensation if there is one. Should they pay for their own treatment really? Why do we not argue that we should acknowledge that we are small minded, strangely curious and very often distracted as in the case of the rst, second and fourth person? Should we really think these features as worthless? Should it not be reasonable to pay for the treatment then? But consider the rational and the heroic types, should not we compensate for their disadvantages? Are rational and virtuous actions not valuable? Of course they are. The conclusion must be that responsibility does not forfeit the right to compensation. If that would be the case, all persons should pay for their own treatment. All should bear the costs of their choice alike. That is, I believe, contrary to the idea of equality in which that what is valuable called liberties, should be equalised and the compensation for disadvantages resulting from these liberties form claims to compensations of which the urgency depends on the liberty that was executed and the liberty that as a result of that action is threatened. To be clear, I will not deny that the history of liberties exhibiting the way a particular situation came about is not important, it is. It shows the other liberties, which if we should not compensate, would be threatened otherwise. But to what extend its history and responsibility should be in uencing the claims on compensation depends on the value of all the liberties at hand. Looking once again at the lazy neighbour for whom all the pleasures were paid by the work of others, indeed there is inequality. However it is not simply restored by refusing to pay for the pleasure of that lazy man. There is inequality in the relevant liberties, which are: the liberty to enjoy pleasure and the liberty to enjoy leisure. Which inequality is more important depends on the urgency of these liberties. The pleasure of the neighbour to which is pointed, is a too abstract term for a sensible judgement on the urgency. All these liberties do not make equality simpler but if equality is complex in this way. let it be so. To acknowledge this complexity is better than using simple slogans in which an appeal to responsibility is used to prevent people to be compensated and to prevent care for their liberties to which they have according to egalitarian ideas a reasonable claim. Summarising, it was argued that responsibility is not densely interwoven with the idea of equality. The reason that it seemed to be came from one-sided examples in which only irrational actions were the causes of disadvantages. But once one turns to rational actions with disadvantageous consequences it becomes clear that responsibility for a disadvantage does not forfeit a claim to compensation for it.
4.4. SUMMARY
151
It was suggested that what is valuable should be available for all are the equalisanda. These are called liberties, they are liberties to certain actions or activities or enjoying situations. The importance of these liberties determines how much they count in distribution problems. Because of a plurality of values there is a plurality of liberties, consistent with moral value pluralism. There is a plurality of equalisanda. It will give rise to the idea of several ideals of equality each for its own equalisandum or liberty. Equality concerning the more urgent liberties weigh of course more than those of the less urgent.
4.4 Summary In this chapter, the third of the three background assumptions of the ideal of equality was discussed. It was argued that monism was not tenable. First it was seen that the reasons for monism were not sound. The assumption of ethics to be concerned solely with action-guiding judgements was not right. Furthermore, a complete ordering does not imply that there is one supervalue. And even if such a value was postulated, this value was seen to be not of any help in deciding what to do, because it is representing the results of deliberation. It was mentioned that the whole idea of equality actually presupposes moral pluralism, because there is a di erence between the value of a distribution and the value distributed in that distribution. Subsequently, it was shown that monism is unsatisfactory because it cannot account for moral con ict and genuine choice. Pluralism could account for these, but in a more complex way than one expect at rst sight. Particularism had to be stated in a way such that it does not mean that no judgement is related to any other in other situations as would be held in the most extreme denial of moral universalism. Pluralism could account for con icts and genuine choice within a particularistic framework if it is acknowledged that there are relationships between judgements in di erent situations. These judgements are not necessarily so related in all situations. After arguing in favour of moral pluralism within a moral particularistic framework, I turned to the problem of the equalisandum. Pluralism was seen to be of help in establishing equalisanda which could be subsumed under the name of liberties. It was shown that the reasons for incorporating responsibility into the equalisandum were based on a myopic view, only examples in which morally wrong actions were looked at in arguing that some good had not to be given to someone. Once these morally wrong actions were substituted by morally right actions the examples were loosing their force, indicating that responsibility was not the main reason for denying compensation. Real possibilities for valuable actions or experiences with-
CHAPTER 4. MORAL VALUE MONISM
152
out bearing disadvantages that others do not su er from were introduced as equalisanda. They were called liberties. Consistent with particularism one could say that in a particular situation with a distribution problem particular liberties of several persons are important. For example in political contexts this will be political power, in situations of acute danger this will be chances of being rescued and survive, instead of political power. A consequence of this view on equalisanda especially its denial of the inherent link with responsibility is that evaluations of distributions remain applicable without taking their history into account. It was argued in the previous chapter, that in case responsibility had to be incorporated, procedural evaluations or historically based principles were inevitable.28 By taking these liberties as equalisanda one can end the discussion on the proper equalisandum. The end-state evaluations as in pattern principles of distributions remain possible. How to evaluate distributions regarding the ideal of equality is discussed in subsequent chapters. So far, I have only introduced moral realism, realistic individualism and moral value pluralism as alternatives to the traditional background assumptions. Proceeding with the articulation of the ideal of equality in the second part of this study is possible after I end this part on the new framework in the next chapter by answering the criticisms to moral realism as basis for the ideal of equality.
28
See chapter 3 p. 89.
Chapter 5
The new framework and its criticisms 5.1 Introduction In the previous chapters, I introduced moral realism, realistic individualism and moral pluralism, as alternatives to the traditional framework for an ideal of equality, which is characterised by moral universalism, volitional individualism and moral value monism. Now it is time to discuss the criticisms of moral realism as a framework for an ideal of equality. These hold that the Wittgenstein-Davidson approach undermines the reasons for taking moral realism as a proper basis for the development of an ideal of equality. I show that the charges against moral realism as it was introduced in the previous chapters, can be answered properly. This discussion will open up the way to the next part of this thesis in which an ideal of equality and its measure is articulated. Let me rst recite the main reasons why moral realism was argued to be an appropriate starting-point for the analysis of an ideal of equality. Moral particularism that is based on moral realism, is the view that moral rules are not the basis of moral judgements. It was introduced because it was more satisfactory from a general philosophical point of view. In moral particularism moral judgements, were considered to be judgements about the external world which can be known to us. The knowledge could be accounted for by epistemological coherentism. This moral realism implied that the question on the proper measure of inequality cannot be answered simply by some stipulative measure for example a statistical measure like variance and declare this to be the measure of moral badness of inequality. 153
154 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS A stipulative measure is not enough. A proper measure of inequality should be tracking the extent to which the inequality of distributions is morally wrong and should not simply follow our reactions or our ways of describing these distributions. This tracking is precisely what is to be expected from a measure within moral realism. Furthermore, it was argued in the introduction that moral universalism prevented the possibility of a simple measure of inequality which represents a complete ordering regarding the moral badness of inequality, moral particularism avoids this disadvantage. Finally, the moral realism turned out to be a proper basis for the alternatives to the other traditional background assumptions, realistic individualism and moral value pluralism. Realistic individualism, the view that holds that individuals have wants and desires which have to be interpreted and hence are restricted by this interpretation, appeared to be a promising view for the development of an ideal of equality because it could account for interpersonal comparisons of what goods mean to people. It was argued that such comparisons were necessary for determining the meaning of an ideal of equality. The proposals of circumventing interpersonal comparisons failed. It was seen that volitional individualism prevented the possibility of such interpersonal comparisons and hence hindered the development of an ideal of equality. Moral value pluralism appeared helpful because the discussion on determining the proper equalisandum could be answered in a more satisfactory way, namely by acknowledging that there are several equalisanda. It was seen that pluralism was already assumed by the whole idea of an ideal of equality because of the di erence between the value of what is to be distributed and the value of the distribution itself. The reasons for holding that moral realism is an appropriate background for the development of an ideal of equality are argued to be undermined by several criticisms. One is that moral realism is a form of conservatism. The approach is held to be conservative because interpreting and understanding other persons is held to be based on commonly held practices and takes society to be of superior importance. It is argued that this conservatism does not allow any role of moral or political ideals, hence not a role of an ideal of equality either, for guiding rational changes of a society. I rebut that the criticism that holds that conservatism cannot allow for any reasonable change, is not correct. It appears that another charge will become more acute, namely the problem known under the name of imperialism. Moral realism as introduced in this thesis is accused of so-called imperialism of the rst-person perspective in understanding other persons. What a person cannot understand for himself he cannot understand in another. One's own knowledge determines the interpretations. If one meets someone making sounds in a way not to be recognised as having any regularity, no
5.1. INTRODUCTION
155
coherent understanding can follow, it remains just noise. It might be that one did not recognise these sounds as music, for example because one did not learn enough about music. Knowledge determines what you recognise. Thus it is argued that radical interpretation gives an undue weight to rst person statements, other person's ideas, wants and ideals are moulded by the interpreter's knowledge. The truth of moral judgements becomes highly dependent on intrapersonal characteristics of the evaluating person. It is a threat to equality in which it is a leading idea that other person's views have to be taken seriously. This would be undermined by imperialism. I answer this charge by arguing that moral realism does not imply imperialism, it allows for acknowledging several points of view that are accessible to us. The charges of conservatism and imperialism are answered by turning to a non-sociological interpretation of Wittgenstein's ideas on rules and concepts, as for example clearly exposed by Baker and Hacker and Hurley [Baker & Hacker, 1984] [Baker & Hacker, 1985] [Hurley, 1989] [Hurley, 1992]. The importance of moral realism is challenged further by arguing that there is a viable alternative to moral realism as it was introduced here, namely quasi-realism. It is argued that quasi-realism advocated by Blackburn, accounts for most phenomena that realism claimed to be explaining and can account for even more [Blackburn, 1984] [Blackburn, 1993]. This quasi-realism is anti-realistic because it denies there to be moral facts in the world. It adheres to taking moral statements to be literally true or false, there is moral knowledge and even truth or falsity of statements being mind-independent. Contrary to moral realism it holds that morality is projected by us into the world, a world which consists merely of natural facts according to quasi-realists. Blackburn argues that quasi-realism can account for supervenience of moral judgements on non-moral judgements, i.e. the idea that a moral property cannot be di erent apart from a di erence in natural properties, while moral realism cannot. I rebut this criticism and show that moral realism is not inferior to quasi-realism and can account for supervenience. The nal criticism to be discussed here is the one concerning moral disagreement. It is argued that moral realism does not solve moral disagreement. But agreement was in discussing the necessity of interpersonal comparisons mentioned as essential for the ideal of equality.1 This charge is answered by pointing out that disagreement is indeed not eradicated but that the disagreement presupposes a common world in which the ideal of equality has its foundation. Arguments that even such a common world is not presupposed are answered. 1
See chapter 3 p. 65.
156 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS These criticisms on taking moral realism as a proper basis for an ideal of equality are discussed subsequently. This discussion will help clearing up misunderstandings about moral realism which is taken as the new framework for the articulation of an ideal of equality and its measure in the second part of this study.
5.2 Conservatism In this section the charge of conservatism against moral realism is discussed. I argue that conservatism allows for change but that the main problem will be the charge of imperialism in which one's own view is necessarily the central one. The charge of imperialism is discussed in the next section. In chapter 2 moral universalism was criticised with the argument that rules do not extend beyond their applications in common practices and common forms of life.2 Mediated by the private language argument, such common practices were referred to in arguing against sense data and the idea that interpersonal comparisons are projections of an inner private process.3 Such a common background is also seen to be essential in radical interpretation as it was introduced in chapter 2. It was said, that in order to understand what another does, says and believes, he has to be interpreted. It was explained that interpretation was only possible against a background of similarity, i.e. that persons live in a common world, have similar perceptions and are similarly rational. This assumption was called the principle of charity. It has to be assumed that persons have one common background on which the interpretations are based. Someone going the baker is not interpreted as someone trying to get parsley to the moon, or as someone getting a sausage of mud, because the latter intentions are not intelligible. It is not denied that it might appear later that they wanted indeed a sausage of mud, but this requires an explanation with help of other common reasons and wants.4 The common background is functioning as what is taken to be true. On this common background the meaning of sentences is based. The interpretation of a sentence is bounded by the truth conditions of that sentence. Interpretations are built on statements of the form, belonging to what is known as Tarski's convention T: `Sentence `p' is true i p' whereby p is taken to be the unre ective common world wherein persons live. Together with some restrictions for grammar and a theory of lanSee chapter 2 p. 34. See chapter 3 p. 101. So, desires are restricted as was argued before, see also p. 101 . and [Norman, 1971, p. 55,65]. 2 3 4
5.2. CONSERVATISM
157
guage, sentences get their meaning.5 This is brie y Davidson's programme [Davidson, 1984]. His programme on the meaning of sentences extends to the content of beliefs, desires and the meaning of actions. In all these the unre ective common background functions as a point of reference for truth and for statements of truth as in the above-mentioned statement about the truth of sentences. If the common background is the reference for truth, then a society which functions as that common background, is the reference for truth. Now society becomes the measure for truth and the idea that moral truths cannot go against society gives rise to the idea that society is the most important value, which should always prevail.6 The idea that the unre ective common background is the measure of truth, also of moral and political truth, suggests conservatism. For example Scruton in his outline of the conservative system of beliefs and attitudes writes: The rst axiom is the simple principle that, lacking an overmastering ideal (.....), conservatism must necessarily take many forms. Solon, asked what is the best form of government, replied `For whom? And at what time?' It is a particular country, a particular history, a particular form of life that commands the conservative's respect and energy, and while he may have an imaginative grasp of other real or ideal arrangements, he is not immersed in them as he is immersed in the society that is his own. No utopian vision will have force for him compared to the force of present practice, for while the former is abstract and incomplete, the latter is concrete quali ed by familiar complexities that may be understood without describing them. .. . It is rather to emphasise that political understanding, as a form of practical judgement, does not readily translate itself into universal principles, and that conservatism, which takes its main inspiration from what is, has little competence to meddle in what is merely possible. [Scruton, 1984, p. 36] These remarks t strikingly well into a moral particularistic view. Because the common background is the reference for truth, it cannot itself be criticised; a view immediately leading to conservatism. There is no moral truth beyond that common background and no change can be argued 5 Here it is not the proper place to discuss these restrictions,some of them are discussed in [Blackburn, 1984, p. 261 .]. 6 Norman ascribes this kind of reasoning to Durkheim [Norman, 1971, p. 113 .].
158 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS for on the basis of this background. This leads to the criticism that moral realism as it is introduced in this study, does not allow any change on the basis of rational criticism, because the common background is taken to be the measure for truth. It is held to be impossible that there are ideals on the base of which this common background can be criticised. For example, a racist society cannot change from within on the basis of rational criticism based on ideals, because racism is held to belong to that common background. Similarly, a communist society cannot change on the basis of rational criticism because the communist ideas are presumed to be the common background. All change is beyond rational evaluation and consequently cannot be seen as an improvement. Notwithstanding all this we judge the recent changes in South Africa and those in former Eastern European countries as a start of an improvement, not only in our Western European view, but also in the eyes of the members of those societies themselves. So, something has to be wrong with the argument that a common background cannot lead to improvement of itself. And indeed, there is something wrong. One way of allowing rational change on account of moral criticism is suggested by Norman in citing Durkheim: It is possible that, as a result of some passing upheaval, some fundamental moral principle is hidden for a time from the public conscience which, not feeling it, denies that it is there (theoretically and explicitly, or practically as in action; it does not matter). The science of morals can appeal from this temporarily troubled moral condition to that which pre-existed. [Norman, 1971, p. 144] The common background is held to be ambiguous and lled with several moral ideas laying there waiting for exploration.7 But reality, or the common world, is then full of contrasting ideas. Such a complexity is not ruled out a priori, it will be examined in the next section. An other defence is more simple and is more in line with conservatists ideas as explained and defended by Scruton: It [the constitution, SFH] may change and develop in accordance with its own inner logic - the logic of precedent, practice and judicial abstraction. The conservative instinct is not to prevent that change- since it is the vital motion of the state- but to guard the essence which survives it, and which enables us to say that its various stages are stages in the life of one state or 7
A similar view is expressed by Walzer in [Walzer, 1988]
5.2. CONSERVATISM
159
nation. And the constitutional essence guards in its turn the social essence. Here then, is the conservative cause in politics. [Scruton, 1984, p. 52,53] So one could point to the practices which determine or re ect what counts as legitimised in political and moral a airs. Consequently, society can be criticised rationally on the basis of these practices. One could argue against this defence that change is only possible within an inconsistent common background, consisting of the particular beliefs and those resulting from reasoning according to legitimisingor rational practices according to which those former particular beliefs are wrong. For example one could criticise the capitalist belief system in a society on the ground of the practice of reasoning holding those capitalist beliefs to be wrong, but it is like believing `A' , and `If B then C' together with `B', so also `C', which happens to be similar to `not-A'. The above stated argument is not threatening the above stated conservative view on change, it is even backing it. This can be made clear by referring to Lewis Caroll's famous dialogue between Achilles and the Tortoise in which Achilles tries to force the Tortoise to accept a conclusion `Z' from the propositions `A' and `B' from which `Z' follows logically. Achilles is asked by the Tortoise to add to the premises `A' and `B' the proposition `C'(=`If A and B then accept Z'). Achilles adds `C' but then the Tortoise replies that Achilles has to add also to the premisses `D'(=`If A,B and C then accept Z') in order to accept `Z' by reasoning. But after this addition, this is not enough for the Tortoise, he asks that Achilles adds to the premises `E'(=`if A,B,C,D then accept Z'). And of course even this will not be enough for the Tortoise. The story of Achilles and the Tortoise can be seen as an illustration of the idea that reasoning itself, such as drawing a conclusion from premises, is a practice of proper reasoning according to a practice of drawing conclusions. Drawing conclusions from premises cannot be justi ed beyond that practice.8 Consequently what counts as an inconsistency is determined by the practice of rational reasoning itself in that society, it is not something which can be determined beyond that society's practice of reasoning as is assumed to be the case in the argument against conservatism. Society's own practices should be used and not the reasoning of the outsider.9 Thus See [Winch, 1958, p. 55-57]. This last remark should be taken carefully. An observer has to interpret the beliefs and attitudes of that society and in interpreting he has to have a similar way of reasoning, otherwise he could not understand that society's beliefs and attitudes. This latter possibility is not excluded and its consequences are discussed below in the next section on imperialism. With respect to the possibility of understanding and interpreting by outsiders, it should be clear that it is not held that reasoning is something beyond the 8 9
160 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS it can still be held that the charge of inconsistency against the possibility of rational change is not valid. Inconsistency is directly determined by the practices of that society, so criticism of the background is possible together with that background, without leading to inconsistencies. This defence of moral realism based on radical interpretation will immediately trigger the argument that it is radically relativistic and could not be properly called realism.10 Each society could have its own form of reasoning which might be not intelligible to another society. This would lead to some awkward conclusions. Justi cation of foreign aid for example would be beyond the possibilities, because what is needed by those other nations and societies could turn out not to be understandable by the donor countries. Questions of justice, equality and inequality would be necessarily restricted to one society. This charge of being relativistic is a serious one [O'Neill, 1993]. The charge could be a little bit softened by arguing that foreign aid can be justi ed on the ground of the norms and values of the donor society giving foreign aid. This is in a sense imperialistic. Aid is not given because others are in need but because they are judged by the donor country to be in need, even if the receiving country does not understand why aid is given.11 On the other hand, it should be acknowledged that the donor country has to act on its own knowledge and evaluations. It has nothing else to turn to. Within conservatism foreign aid is not excluded, although it has to deal with the charge of imperialism. This charge is dealt with in the next section. Another argument against the defence of conservatism that it can allow for rational criticisms and change could be that its concept of rationality can change and so it is relativistic by missing any good measure of what is rational and irrational. What is rational now can be irrational later. This charge too can be rebutted by the conservatives. The argument is even dicult to understand for stubborn conservatives. The argument exhibits some sort of requirement that rationality should be similar over time. But this requirement does not have any sense because of course rationality will remain the same, it does not change, it remains similar according to the common practice of reasoning. In order to detect or imagine that some change of this concept of rationality is possible, there should be some sort of point of reference beyond this society, which is not to be found. The whole concept of rationality is not oating. This idea would suggest that there is something persistent relative to which it is oating and to which it practice in that society. 10 See for example [Nozick, 1993, p. 154]. 11 See for a similar argument against the idea that moral subjectivism implies indi erence [Willams, 1972, p. 40 .].
5.2. CONSERVATISM
161
can be anchored. But this is not to be found. There is no such anchoring to something beyond society. To require such a possibility is asking for arguments being knockdown arguments. The story of Achilles and the Tortoise shows that there are no knockdown arguments. Arguments just lure but do not force. I conclude that in so far moral realism is like conservatism it can deal with the most pressing arguments of admitting rational change, international justice and relativism. Although the charge of imperialism remains. Before I turn to that problem, I present as a preparation for the next section another view on rule following. A view which does make more sense than the one presumed in the conservative view. So far, rule following and language was interpreted in a rather sociological way. The practice referred to was seen as a practice in an existing community. For example, Kripke interpreted Wittgenstein's ideas on rule following and his private language argument `as it struck him', in this way. For example, he wrote that in our interpretation of Robinson Crusoe on a desert island making sounds and some gestures as following a rule we take him as a member of our community [Kripke, 1982, p. 110]. We apply our criteria of rule following. But this sociological view is highly susceptible of communal solipsism [Williams, 1974] and can be questioned because of its imperialism. As was mentioned in the former chapters, rule following was seen to be possible by making sense of the distinction between to follow a rule and merely thinking to follow a rule.12 There should be something beyond the thinking of the rule following subject that can account for this di erence, something out there beyond the inner world, objective and independent of the thinking.13 The community view holds that the existing others that are `out there' with their reactions and dispositions for reactions should account for the di erence between merely thinking to follow a rule and to follow a rule. Their reactions account for the possibility of following a rule. But the fact that it are others who follow that rule or a di erent rule, has nothing to do with it. It is just something out there beyond the subjective idea of a practice that matters. The fact that it are others who react, adds nothing. How can one person acting under an illusion help preventing another person acting under an illusion? Is a mistake commonly made not any longer a mistake? The community view is the wrong way of accounting for rule following, as Baker and Hacker argue. See chapter 2 p. 45. See for a further explanation of the idea of something out there, for example [Hurley, 1994]. 12 13
162 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS A rule is, as was stated above, not something apart from its applications outside the mind. The link between the instances of the rule following out there and the rule is direct, it is not mediated by the reactions of other persons as is suggested by the community view. There is nothing more to be said than if someone is following the rule `add 2' proceeds after writing out there 1004 with 1006, it is just what `add 2' is. If he proceeds with 1005, he can acknowledge he is making a mistake. Behaviour is not in accordance to that rule because others approve of going on in that way. No, it is in accordance to the rule just because acting in that way is just what the rule means. Other persons have nothing to do with that. To follow a rule is essentially dependent on the possibility of making sense of the distinction between to follow a rule and merely thinking to follow a rule. It is dependent on the possibility of making mistakes. This will mean that it has to be something beyond the private phenomenological inner world. It is objective, something out there, and as a consequence it is in principle accessible to others [Baker & Hacker, 1985, p. 179]. That it is accessible to others does not mean that the criteria of rule following are the actual or hypothetical or counterfactual responses of members of a community [Baker & Hacker, 1984, p. 42]. That it is something out there, means that following a rule could be explained to others and is not essentially private. It is possible that one follows a rule on one's own without there actually being any other person or community understanding that rule, contrary to what the community view suggests. Contrary to what Kripke states, Robinson Crusoe on his desert island can follow a rule without assuming that there is a society in which he is admitted. What could be meant by admitting him in our community? Do we consider the cat as member of our community if we say that the cat is hunting the mouse, if we apply our criteria of hunting? Similarly, do we take Robinson Crusoe to be a member of our community if we apply our criteria of rule following?14 It is just the objective possibility of Robinson Crusoe making mistakes, i.e. the regularity of his behaviour showing a complexity which yields normativity. It is not because we believe Robinson Crusoe's behaviour falls within the criteria of following a rule that he is following a rule, as the community view suggests, but just because he is following a rule [Baker & Hacker, 1984, p. 40]. This view on rules and rule following has its counterpart in moral realism. The common background which is supposed to exist is the common world which is objective beyond the private phenomenology just as the possibility of making mistakes is something beyond the private phe14
See for similar remarks [Blackburn, 1984, p. 85].
5.3. IMPERIALISM
163
nomenological domain. The world is not merely accessible to a restricted group of persons or community, but it is essentially accessible to all. As will be clear, conservatism, in which rule following is accessible only to the members of a society, is left behind by this interpretation. But does it help to escape the charge of imperialism? This question is discussed in the next section.
5.3 Imperialism The charge of imperialism is that the other person's views and the ways he is a ected by goods are moulded into the view of the interpreter. It threatens an important aspect of the ideal of equality, namely that how another is a ected by goods should be the basis of an evaluation of an allocation, as laid down in the assumption of individualism. The charge of imperialism is seen for example in the interpretations of Nagel and Williams of Wittgenstein's ideas on rule following and Davidson's radical interpretation. They argue that what someone can understand is dependent on one's own knowledge [Nagel, 1986] [Williams, 1974]. The central part of the argument is that it is taken to be the case that rst-person judgements as `I believe that...' are the grounds for third-person judgements as `He believes that ...'. The rst person judgements are projected onto others. What another does or believes is interpreted along the lines prepared by one's own beliefs. Of course, in a sense this is inevitable, because it are my opinions and beliefs about someone and not those of someone else. This is just like saying that my sensations are not felt by someone else, this is simply true, but it has nothing to do with imperialism. With the charge of imperialism it is meant that one's interpretation of someone else is restricted such that the other person's beliefs could not be detected because of one's own blind spot. Lacking any sense of colour, one can lack understanding of another who does discriminate between colours [Nagel, 1986], or being a mediaeval peasant one can lack any understanding of the sounds which are now considered as music, or unable to consider Schrodingers equation as a scienti c formula. In each of these cases one's restricted knowledge or belief, is restricting the interpretation of another. Other person's belief systems are moulded into one's own nite and fallible beliefs helped by the principle of charity according to which we see others as similar to ourselves. And this charge throws radical interpretation back into the bag of idealism contrary to its realistic pretensions and threatens others being taken seriously. At the end of the former section I introduced the Baker Hacker interpretation on Wittgenstein's idea of rule following. By this interpretation
164 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS one way of imperialism was circumvented. Whether one is following a rule correctly is not any longer dependent on the reactions or likely reactions of others including those of an interpreter. In contrast with the community view, my rules are not determining the rules of others. Whether someone is following a rule correctly is something beyond the private mind of the inner world, something out there, and so accessible to others; their reactions are not essential. This idea of an objective world out there, accessible to others, will be the basis of the answer against the charge of imperialism concerning radical interpretation. The issue of the argument of imperialism against radical interpretation concerns the possibility of excluding mistakenly another person or another person's actions as intelligible while he could be understood by others. Seeing colours and the actions based on this visual understanding, hearing music and not just natural sounds, are supposed to be not understood to the observer if the observer lacks the abilities or knowledge concerning colours, music, but it is not excluded that they could be understood. To be clear one should distinguish between being not understood and being not understandable or being not intelligible. The former is not the real concern, our interpretations are of course fallible and so is our knowledge. Davidson does not claim that all others are understood by us. There can be wide cultural di erences, it is not denied that this is possible. But the claim of the charge of imperialism is that radical interpretations means that all others are understandable and are intelligible to us, i.e., there can be no other belief systems beyond those we can possibly understand [Davidson, 1974], while in fact there can be others that are not understood by us. The charge of imperialism however, would be right if knowledge or belief could not change, or could not be di erent from what it is or was; if it was static and consisted of eternal truths. There would not be a di erence between understanding and being able to understand. But people can learn and their abilities can change and their beliefs and desires change, so they could become able to understand the music, the colour inspired actions, and Schrodingers equation. The claim that all others are understandable and intelligible to me is not necessarily seen to be imperialistic if I acknowledge that I could have di erent knowledge or could change. Understandable, or intelligible, need not be the same as here and now understood, but can be interpreted as possible understanding, meaning: not necessarily hidden for others. By acknowledging the possibility of change of di erent beliefs and abilities, one could say in a sense that I incorporate in my belief system the possibility of all other subjects and conceptual schemes. It is precisely the principle of charity, which seemed to be the ground for imperialism, which is responsible for this incorporation, for example illustrated by `If I were in
5.3. IMPERIALISM
165
his position I would believe...', expressing the understanding of the other person. Of course it is me who is incorporating these other possibilities, but this is not what is meant by the charge of imperialism. With imperialism is meant an undue restriction. That is not the case. My belief system is just expanded to acknowledging other possible belief systems which I actually don't have but which I could have.15 It is excluded that there is a person living in a world, possibly far away, and of which we are not aware, who is not merely understood by us or others but who is not intelligible to us. This possibility is excluded because the principle of charity takes care for the understanding by hinting at what we would believe if we were there. One could challenge this defence of radical interpretation with the remark that this principle cannot be turned to if persons do not exist in that far away world and hence we could not understand that world. Let me formulate the issue of imperialism di erently, which does not rely on this use of the principle of charity. Is it not possible there to be aspects of our world of which we cannot know or have any understanding because it is beyond our grasp of intelligibility? If it is possible, the charge of imperialism would be correct and serious, but if it is not, then radical interpretation is not leading to undue restrictions that are dependent on our beliefs. Let me distinguish di erent cases of lack of access to that part of reality of which we cannot make sense. In one form, it is the old sceptical issue, is it not possible we all are living in an all pervasive illusion and that the common world, which is so basic for interpretation, is just the beetle in the box which might be not a beetle or might be even missing.16 In the other, 15 Hurley argues, also essentially based on the principle of charity [Hurley, 1992, p. 99], that it is excluded that for example persons from England can understand Plutonians and Plutonians can understand Saturnians, but the English cannot understand the Saturnians. Intelligibility or translatability is transitive. If the English can understand the Plutonians and also their beliefs about the Saturnians, then of course they can understand the Saturnians. Similarly, if possibilities are introduced, intelligibility is transitive. If it is possible that the second understandsthe third, then the second would have beliefs, etc., of the third under some possible circumstances even if the second does not, as things stand as present, understand the third. If it is possible that the rst understands the second, then it is possible, given what has already be said about di erent possible circumstances (above and in note 24 below), for the rst largely to understand a set of attitudes that includes not just the actual beliefs of the seconds, but also what the second would believe under a variety of possible conditions. [Hurley, 1992, p. 104] (note 24 is explaining the principle of charity as mentioned by David Lewis in [Lewis, 1983](SFH)) 16
See chapter 3 p. 101.
166 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS it is stated as a world not known to us with which we do not interact, it is like an unknown world somewhere out there, which we cannot understand. First, the sceptical illusion. Indeed, there are some beetles which might be missing, but these would not be the aspects of the world but the experiences considered as the private sense data of them. These private experiences are the beetles Wittgenstein is talking about. The world itself cannot be like a beetle, the missing world would imply solipsism with its necessary but meaningless privacy. But even if it is acknowledged that there is a world out there, could it not be possible that it is totally di erent from the world we think we live in? If it could, imperialism would be a real charge. But it cannot. We cannot err totally about the world we live in, because that world is connected in a simple way with our beliefs and desires. Our beliefs and desires and the interpretations of what others belief, desire is directly dependent on the truth of parts of our knowledge. Without this there would be no understanding. There is no special feature in the world which makes our beliefs generally true. Was the world di erent, then we would mean di erent things by what we believe now. If we lived in a world in which ` `water is wet' is true' but in fact water is not wet, we would mean something di erent with wet in contrast with what we now mean by it. The sceptical possibility of there being a world without being understandable for us is excluded.17 This reason for imperialism is answered. Let me turn to the other one, the far away world were we do not live in. Suppose there is a world far away which we do not understand and cannot understand and about which it is necessarily impossible to have any understanding, then the charge of imperialism would be a valid reason for considering radical interpretation as a form of idealism. But this possibility is excluded too. The far away world out there could not be something necessarily inaccessible to us, because of what we mean by `something out there'. Being out there is just being accessible to others, and if it is accessible to some, it is accessible to us. The reasons for the inaccessibility cannot be conceptual reasons, they could be practical as for example being too far away to interact with. But then we could formulate that this world would be understood under other circumstances which are not metaphysically impossible. It should be realised that metaphysical impossibilities are grammatical impossibilities. Squared circles or princes appearing simultaneously as prince and as frog, are not possible. Similarly, the impossibility of understanding the outer world would be pointing to something ungrammatical. Even heaven although not understood by most of us here and now, is not metaphysically inaccessible.18 God could not make a world 17 18
See also chapter 2 p. 46 It is said to be understandable by God's love and his grace.
5.4. QUASI-REALISM
167
metaphysically inaccessible to us. It is used to be said that God could create anything except what would be contrary to the laws of logic.-The truth is that we could not say what an 'illogical' world would look like. [Wittgenstein, 1918, x3.031] (transl. Pears & McGuinness,1974) Summarising, the charge of imperialism that would undermin moral realism as a proper basis for an ideal of equality is not valid. The rst-person perspective is not a threat to understanding others. They do not pose undue restrictions to interpretation of what others a ect. Other persons views can be recognised. Although it can happen they are not understood, there is no impossibility that they can be understood as the charge of imperialism holds. After all, everything in the world is accessible to all.
5.4 Quasi-realism The two arguments dealt with in the former sections, conservatism and imperialism, were indirect arguments against moral realism, they concerned its basis namely radical interpretation. In this section, a more direct argument is discussed. It was argued that radical interpretation accounted for moral realism, i.e. the view that moral judgements are true or false, and are objective, meaning that they are pointing to something beyond our own attitudes. Moral realism is threatened by an alternative: quasi-realism, or also called projectivism, advocated by Blackburn. Quasi-realism holds that there is on one side a natural world and on the other side our reactions to this world which are projected and spread into this world. There are no moral facts in virtue of which moral statements are true or false, there are just natural facts and the projections which account for the truth or falsehood of moral judgements. There is just the natural world on the one hand, and on the other hand, our attitudes, our actions and functions which account for the other propositions beyond the natural ones [Blackburn, 1984, p. 181 .]. It is argued that this quasi-realism is more appropriate because of its economy of metaphysics. No other facts than natural facts have to assumed, so no moral facts or mathematical facts or extra natural facts have to postulated in order to account for, or explaining, our use of making statements and judgements [Blackburn, 1984, p. 182]. The second argument held by Blackburn in favour of quasi-realism is that it can account for supervenience, the idea already passed on in chapter 2, that a moral property cannot be di erent apart from a di erence in natural properties. Moral realism is argued to fail in giving an explanation for supervenience
168 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS [Blackburn, 1984, p. 182 .] [Blackburn, 1993, p. 111-129]. The third argument is that quasi-realism can account for the relation between moral judgements or moral beliefs and actions or desires. In quasi-realism they are projected attitudes and thus the connections between them and actions and desires are accounted for because all of them belong to the same domain of the passions, the conative or motivational part of persons. Moral realism locates the moral beliefs in the cognitive representational part and so there has to be a special explanation for moral beliefs to be motivating according to Blackburn [Blackburn, 1984, p. 187 .]. These arguments are discussed in order to nd out whether they are serious charges against the moral realism. I argue here not that moral realism is to preferred to quasi-realism but my aim is at showing that it is at least as adequate as quasi-realism, meaning there are no convincing reasons to leave moral realism and embrace quasi-realism. The third argument against moral realism stating that moral realism has to invoke a special mechanism for moral beliefs to be motivating, while in quasi-realism such a mechanism is not necessary, is not a real charge against the moral realism based on radical interpretation. In the chapter 2 this argument has already been dealt with.19 There is an intrinsic or conceptual relation between beliefs, whether they concern moral beliefs or not, and actions. It is not necessary to postulate some mechanism in order to explain that moral beliefs are motivating. They just are because of what beliefs and also moral beliefs consist of. Beliefs are interpreted beliefs, they are not there just waiting to be connected with a desire. There are acts to be explained, the actor's beliefs are interpreted with the help of these acts. There is no reason for an apart mechanism to account for beliefs being motivating. It is a rather at conceptual relation. There is nothing mysterious or deeply metaphysical about human beings. The di erence with respect to the relation between beliefs and actions between moral realism based on radical interpretation and quasi-realism is not in favour of quasi-realism. The rst argument in favour of quasi-realism, that it needs less metaphysical categories of facts, is just a bad argument. Whether some view is more or less economical with respect to kinds of facts or aspects is beside the point, what matters is whether a view is correct or not. Furthermore in the radical interpretation view it is just assumed there is one world in which we live, not a natural world and a moral world and a mind world or whatever world there could be postulated, but just one. Whether it is natural or non-natural is not relevant at all. First there is this one world with people in a Babylonianlike situation. After interpretation there will 19
See chapter 2 p. 38.
5.4. QUASI-REALISM
169
arise a distinction between natural and non-natural propositions, it is not a priori given, as the argument against moral realism presumes. The second argument holding that moral realism cannot account for supervenience, is an argument which deserves more attention. It has not been dealt with in the previous chapters and it seems to be a serious argument. Supervenience of moral properties M on natural properties N means: Something cannot have a particular M-property and particular N-properties without there being something else having a different M-property while having the same N-properties as the former. Supervenience is not rejected by moral realists, it is acknowledged that it applies to our moral judgements. If something is judged as right or good and something else is judged as wrong, then there should be some di erences which can be characterised by natural properties. But moral realists hold also that: There are some moral propositions that are true, but whose truth is not entailed by any naturalistic facts about their subject. Some take non-entailment of moral properties from non-moral properties to be right for all moral propositions, but Blackburn argues with this less strong claim which strengthens his argument against moral realism. Supervenience and non-entailment cannot go together, that is the basis of Blackburn's argument against moral realism [Blackburn, 1993, p. 111-129]. One way of trying to arrive at consistency of supervenience and nonentailment is by interpreting supervenience as holding that it could be otherwise. So instead of a particular M-property say M1 being coextensive with a particular set of N-properties, it could have been the case that another M-property say M2 was coextensive with this set of N-properties. Supervenience just means that once a particular coextensive relation exists for one object, it is valid for all objects, although it could have been di erent. In other words, there are worlds in which an object is M1 and also satisfying N-properties and all other objects which are M1 satisfy these N-properties, and there are worlds in which an object is M2 and satis es the N-properties which were in the former world coextensive with M1, and all other objects being M2 satisfy those N-properties. Supervenience is satis ed and also non-entailment, natural propositions do not entail moral predicates, because the latter are dependent also on the particular supervenience relation, the conclusion might have been di erent. So, non-entailment is consistent with supervenience, and there seems to be no problem for moral realists.
170 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS The problem for the moral realists pointed to by Blackburn is precisely this defence to the charge of being inconsistent by accepting supervenience and non-entailment. It illustrates what supervenience in this interpretation precludes, namely so-called mixed worlds. Worlds in which some objects are M1 satisfying N-properties and some objects M2 satisfying this same Nproperties. The ban on mixed worlds cannot be explained by moral realists. So, either they have to account for this ban, or abandon supervenience or non-entailment. It is shown in the following paragraphs that moral realism based on radical interpretation can be defended by all these three means. It can account for this ban, supervenience in a particular sense will be abandoned and non-entailment in a particular sense is abandoned. This will become clear by discussing the ban on mixed worlds. Why should mixed worlds be banned? Within quasi-realism the ban is argued for because of constraints on projections. The purpose of projecting value predicates demands that supervenience and so those mixed worlds are banned [Blackburn, 1984, p. 186]. According to Blackburn supervenience can be explained as follows by an anti-realist or quasi-realist. There can be no question that we often choose, admire, commend, or desire, objects because of their naturalistic properties. Now it is not possible to hold an attitude to a thing because of its possessing certain properties and, at the same time not hold that attitude to another thing that is believed to have the same properties. The non-existence of the attitude in the second case shows that it is not because of the shared properties that I hold it in the rst case. Now, moral attitudes are to be held towards things because of their naturalistic properties. Therefore it is not possible to hold a moral attitude to one thing, believe a second to be exactly alike, yet at the same time not hold the same attitude to the second thing. Anybody who appears to do this is convicted of misidentifying a caprice as a moral opinion [Blackburn, 1993, p. 122]. Although it seems a reasonable explanation of supervenience, the moral realist can account for supervenience too and can even account for the explanation of the quasi-realist. The constraint is not a constraint on projection, but a constraint on interpreting beliefs. We hold moral beliefs, or we keep persons holding moral beliefs, concerning moral predicates about something that are coextensive with naturalistic properties, because the world is as it is. It is just the way our beliefs are constrained [Blackburn, 1993, p. 122]. But supervenience is not merely a constraint on what we believe or ascribe as belief, in contrast with a constraint on the truth as a kind of a metaphysical fact, it is just what morality means, what the practice of moralising
5.4. QUASI-REALISM
171
consists of. Although it can acknowledge all this, the anti-realist explanation given, that naturalistic properties are the reasons for the value of an object, is not acceptable by moral realism as explained by McDowell and Dancy.20 But the ban on mixed worlds can be explained in moral realism by pointing to what the practice of moralising is. Moral realism can account for the constraints on the projections in quasi-realism. Within the radical interpretation view, the projections spoken of by Blackburn, are attitudes projected into this world. But attitudes are not simply there to be projected, they have to be interpreted. So it is understandable that those projections follow the constraint which ban the mixed worlds. It was built in the attitude already according to the radical interpretation view. The problem of mixed worlds which Blackburn considered to be a charge to moral realism, can be seen as a consequence of atomism, in which each proposition on its own can be true or false, whether it concerns moral or naturalistic propositions. Nothing precludes a combination of naturalistic propositions each with a particular truthvalue with a moral proposition being true and a combination with similar truthvalues of the naturalistic proposition as in the former one but the moral proposition being false, both combinations are possible. So, atomism can be held responsible for the problem of mixed worlds. But atomism is not only a problem for moral realism regarding supervenience of moral terms on non-moral terms, but for all conceptual relations between propositions. Also lawlike or causal statements treated as atoms pose serious problems to realism. `A causes B' could be either true or false distinct from `A is the case' and `B is the case'. The answer of quasi-realism is, as was to be expected, to consider statements about causality as projections [Blackburn, 1984, p. 211] [Blackburn, 1993, p. 103]. But what about the statements of `John is bald' and `John does not exist', which proposition is the naturalistic one and which the projected one? The solution which suggests itself is just leaving atomism. Atomism is rejected in the radical interpretation view. One can can conclude that the problem of mixed world is solved in moral realism which is based on radical interpretation. Whole sets of propositions or beliefs and desires are interpreted in a common world and not with respect to a set of individual facts. The world is seen as just one whole world. So far, the charge based on mixed worlds is answered by showing they pose no problem for moral realism in radical interpretation, and even the reason for supervenience within quasi-realism can be explained within radical interpretation. But what about the explanation o ered by quasi-realism 20
See chapter 2 p. 34.
172 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS on the ban on mixed worlds evaluated at its own merits, is it a better explanation? The reason which was given, was that we react to something on the basis of certain naturalistic properties, and mixed worlds are not consistent with this. But what is meant with `on the bases of' or with `because of'? Does it mean that only if the natural properties are di erent we can react di erently? But in a sense this is vacuously satis ed. If our reactions are to be described in a natural vocabulary, what prohibits these to be seen as naturalistic propositions about that object? Then of course, if our reactions di er, the naturalistic propositions would di er too. So, supervenience would be vacuously satis ed, it would be impossible not to be in accordance with supervenience. If supervenience could be infringed, meaning that we can react di erently to objects without there being a difference in naturalistic properties, the only explanation which could be given is that it is not exhibiting a moral attitude but something beyond attitudes. But this answer is not di erent from the one given by moral realism in the radical interpretation view. The reason given by quasi-realism is not superior. Quasi-realism does not pose a threat to moral realism with respect to supervenience. The arguments in favour of quasi-realism compared to moral realism do not show quasi-realism to be superior to moral realism that is based on radical interpretation. Whether quasi-realism is a plausible view in general is a question beyond the purpose of this thesis. Concluding, it was shown that moral realism is not inferior to quasirealism. It can account for moral beliefs being motivating. That quasirealism is more economical is just beside the point. The problem posed by supervenience and non-entailment is solved by accounting for a ban on mixed worlds by considering supervenience as a constraint on our beliefs re ecting what moralising is in this common world in which we live and interpret one another. However supervenience seen as the view that natural properties are the reasons for moral properties is rejected. Supervenience is just a constraint, and not something fundamental about the origins of values. Also non-entailment is accepted as meaning that natural properties are not the reason for the moral properties. It is not rejected that natural properties can entail moral properties in the sense of being material equivalent or being coexistent. But such an equivalence is not to be seen as a matter of priority of judgements. Some moral judgements and some natural judgements could happen to be equivalent, this is not denied by moral realism.21 Furthermore, it was explained that the problem posed for moral realism that it could not explain supervenience, had its origin in atomism which is not assumed by the moral realism based on radical interpretation. 21 This was also the lesson to be learned from the discussion in chapter 2 and its appendix.
5.5. DISAGREEMENT
173
5.5 Disagreement Moral realism based on radical interpretation was introduced as a way of saving evaluative judgements from the morass of subjectivity and scepticism. This was necessary because without objectivity of these judgements, interpersonal comparisons were prevented and judgements on equality become dicult as was gured out in foregoing chapters.22 In the foregoing chapters, it appeared that the di erences of opinion prevented a uniform answer to questions concerning equality i.e. interpersonal comparisons and the proper equalisandum. These di erences of opinion are not avoided by moral realism. Merely objectivity in the sense of referring to the world is established. It is established that such judgements concern something beyond the subjective self and are, as a consequence, intelligible to others. Although there is assumed a broad range of agreement for the possibility of interpretation, di erences of opinion are not denied to be possible. Di erences in judgements are still possible against this background of agreement. Di erences cannot be expected to be extinguished. In other domains too there remain unsettled questions although we know there to be right answers. The ideal of equality is not threatened by the di erences of opinion, because the ideal of equality can still be located in the domain of the objective background and the general agreement. It would be di erent if this general agreement was questioned. And actually the general agreement is questioned. The general agreement is questioned by two kinds of arguments. These arguments threaten the role of radical interpretation as a sound basis for moral realism. One kind of arguments turns around the issue of indeterminacy of meaning and the other argues that general agreement is not a necessity for interpretation and mutual understanding, for example in case of three mutually understanding persons or communities. Let me turn rst to indeterminacy. In radical interpretation it is assumed that we hold as much beliefs to be true and rational as we can, we assume we live in the same world and assuming the other to be like us. This was called the principle of charity. But now it is argued that although one can agree on the truth of a proposition one can still have a di erent interpretation and live in a di erent world. Blackburn mentions the Lowenheim Skolem theorem as the basis for this argument [Blackburn, 1984, p. 278]. This theorem says that if a set of propositions has a model, in which those propositions are interpreted to be true, then there is a model which is nite or enumerable in nite. It means that if you have a model with a domain with more than enumerable 22
See especially chapter 3 p. 65.
174 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS elements about which the propositions of the set are true, then there is also a smaller model with nite or enumerable in nite elements which makes the same set of propositions true. So, there can be two di erent models for the same set of propositions. Agreement on the truth of beliefs does not mean having similar interpretations, or living in the same world in which the meanings of the propositions do mean the same. The propositions do not determine the model or the interpretations.23 This argument can be answered in several ways. First the Lowenheim Skolem theorem is not valid for all languages. It is a property of rst order predicate logic but not for example for second order logic. And it is not clear at all that our natural language has to be seen as rst order predicate logic.24 The problem could also be approached along a di erent route, which is less formalistic. It is assumed that the indeterminacy arises because there can be several distinguishable models, or what is the same, distinguishable interpretations of a language, although the propositions believed in that language are agreed on to be true. The models or interpretations stand for the world and the connection with the world, respectively. But what if beliefs about interpretations belong to the beliefs agreed upon? Then similarity in beliefs cannot be based any longer on di erent interpretations. It looks as if the aforementioned argument of Blackburn is based on a di erence between beliefs or propositions that are essentially about non-psychological objects and propositions or beliefs like those about interpretations and about psychological objects. The former are held to be the basic ones, they are held to determine the world we live in. But this distinction, which is a central aspect of quasi-realism, is not any longer important in radical interpretation. In radical interpretation such a distinction is not assumed to be basic. There are just beliefs, there is no fundamental distinction between the several kinds of beliefs like physical or psychological beliefs. This means that beliefs with a psychological content, for example those about meaning, can belong to the corpus of beliefs used to interpret the language.25 They do not even determine these models up to isomorphism. For second order logic, compactness is not valid either, so this answer would invalidate the proof given in the appendix of chapter 2 based on compactness. However, there are some logics for which compactness is valid but the Lowenheim Skolem theorem is not, for example the logic with a quantor meaning `uncountable many' [Barwise, 1977, p. 44]. 25 The language now does not distinguish between language and meta-language as is the case in rst order predicate logic and its formal semantics. This lack of distinction can result in for example incompleteness, invalidity of the Lowenheim Skolem theorem and compactness being not valid any longer. It is however not excluded that compactness is valid for a part of the language, for example the one about moral theory. It is not excluded that although the whole language is not a rst order predicate logic, some parts of it are. So for moral theory compactness could be valid and the argument in the 23 24
5.5. DISAGREEMENT
175
If interpretations are partially determined by beliefs about interpretation i.e. psychological beliefs, Blackburn points to another problem of radical interpretation [Blackburn, 1984, p. 280]. Davidson's radical interpretation was meant to be letting whole sentences or better, whole sets of sentences to be the subject of interpretation. The interpretation of parts or interpretation on the subsentential level are considered to be abstractions. Parts of sentences cannot be true on themselves without the whole sentence being understood. But by letting in beliefs concerning psychological aspects and consequently ideas about intentions, desires and will, and the principle of charity, there will be a way open to interpretation rst on the subsentential parts. For example if there is a football match and you hear something in a crowd near that football match without hearing exactly what is said, you will infer it is about that match and so parts of the sentences are interpreted before the whole sentence is interpreted, thus Blackburn. This argument of Blackburn against radical interpretation can be dealt with easily to be illustrated by the example of the football match. In concluding that the subject of the sentences vaguely heard in the crowd without being exactly interpreted, is the football match, will not lead to the idea that parts of those sentences are referring to the football match like establishing what a part of sentences means. They are interpretations of utterances resulting from beliefs having a sentential content. For example the belief that there is a football match, it is an interesting football match, a happening which arouses me etc. There is no necessity for interpreting rst a part of a sentence. A subject is not necessarily part of the sentences, unless one is mixing up subject as a grammatical term and subject in the sense of subject matter. The argument against moral realism due to indeterminacy based on the Lowenheim Skolem theorem can be answered by taking also non-physical beliefs into account. Let me turn to the second kind of arguments. There is a second kind of arguments for questioning the general agreement being a threat to the objectivity of moral judgements. It is based on an example exhibiting intransitivity of translation or intelligibility. Suppose a language Saturnian can be translated in English and a language Plutonian can be translated in Saturnian but not in English. Plutonian cannot be translated in English because the Plutonians and the English do not have enough in common to understand each other, otherwise they could understand each other and translate each other's language. It will be clear that there is a problem. What are the objective moral judgements, those in the basis of the understanding between the English and the Saturnians, or appendix of chapter 2 could remain valid, although the Lowenheim Skolem theorem is not valid for the whole corpus of the language.
176 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS those in the basis of the understanding between the Saturnians and Plutonians? The content of those moral objective judgements is indeterminate, objectivity seems to be lost. Moral judgements di er in di erent locations of understanding. The answer of course has to be found in the solution which blocks intransitivity of intelligibility as is suggested by Hurley.26 Intelligibility and translatability is transitive if one turns to capacity to understand instead by sticking to abilities for understanding. In sticking to the latter one denies that persons can learn, or could have di erent abilities from those they happen to have. By realising one could have di erent abilities, the above indeterminacy is just apparent. By extension of abilities the English will be able to understand the Saturnians and there could be again one common objective basis of judgements including moral judgements. The foregoing argument suggests that if three persons or communities understand each other they share a common basis. But does mutual understanding imply there is just one common basis, a basis common for all three, which is the basis for their understanding? A Condorcet like paradox could happen. Suppose the English understand the Saturnians and vice versa, the Saturnians understand the Plutonians and vice versa, and furthermore the English understand the Plutonians and vice versa. Now it could happen that if all there could be agreement on is divided in three, rouhgly spoken, equal parts A,B and C. Suppose further that the basis for the mutual understanding between the English and the Saturnians is agreement on A and B, between the Saturnians and the Plutonians on B and C and between the English and Plutonians on C and A. Now there is no common agreement, only bilateral agreement and indeterminacy seems to be the result, and radical interpretation leaves us with nothing objective. A better look at the example however shows how it can be answered. In trying to ascribe whether the community accepts A or the negation of its content, it appears that the English in understanding the Saturnians have to accept A, but in understanding the Plutonians deny A, otherwise there would be at least one part upon which there was common agreement namely A. So the English alternate in accepting and denying A dependent on who they interpret, they cannot simultaneously understand the Saturnians and the Plutonians, so talking to them simultaneously or in one setting, would exhibit contradictions in their belief system and would of course lead to adjustments. This could also be formulated in another way. The English could interpret the Plutonians directly because of their mutual agreement on A and C, but also indirectly via the interpretations of the interpretations 26
See p. 165 and [Hurley, 1993].
5.5. DISAGREEMENT
177
of the Saturnians of the Plutonians. This possibility is the basis of the argument Hurley gave against the charge of imperialism. But now there will arise con icts between what they interpret directly and indirectly of the Plutonians. These con icts will lead to adjustments and a basic agreement will be the result. Beside the foregoing answer the Condorcet like paradox could be refuted simpler by pointing out that they agree at least that they understand each other. Or could it be that the English held about the Saturnians they cannot understand the Plutonians. But here it should be mentioned that it is not the question of really understanding but the capacity of understanding, which counts. Understanding has to be learned and can change, it is not to be seen as something static, unchangeable and eternal. This brings immediately the problem of the indeterminacy of moral judgements again into the discussion, but now it is not a metaphysical problem but an epistemological problem. And this I have already dealt with. There is no assured way of arriving at knowledge. In the coherence account of knowledge we are never sure about anything. This means that there will probably be disagreement. It will arise in knowledge about facts and about moral matters, but this is not the same as saying there are no objective moral judgements, because there are, despite the lack infallible knowledge. So, radical interpretation does not solve moral dispute, but establishes the dispute by holding that there are objective moral judgements, although we cannot claim rockbottom, infallible knowledge of it; we are searching. This result could be judged as very poor, but what else was to be expected? There could only be another conclusion namely a claim that infallible knowledge is possible for us to arrive at, but that is rather dicult to defend. Whether this result has consequences, should be judged by how it is used. In Hurley's Natural Reasons it is used to invalidate the use of Arrow's impossibility theorem as a reason for moral scepticism, by pointing to the diculty of interpreting two axioms for the proof of that theorem namely the universality axiom and the axiom of independence of irrelevant alternatives. In an article of Hurley Justice without Constitutive luck it is used as a background for a reason for equality. A cognitive approach to morals and politics holding that there is something to be known, leads to the idea of the possibility of distortion of arriving at knowledge which is a bias. Inequality is likely to lead to bias in moral and political a airs. So in order to arrive at reliable, although not infallible judgements, one should strive for equality. The result, despite not establishing infallible ways of arriving at agreement, can mean something. Beside what exactly do we hope for when we hope for world wide agreement, do we strive for that really?
178 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS Summarising, moral realism is not solving disagreement. But the ideal of equality could be formulated within the domain of the objective and the general background agreement of moral di erences. The argument that questions this general agreement based on the Lowenheim Skolem theorem could be answered by pointing to other propositions than those about the natural facts of the world, but for instance those about meaning. These restrict the interpretations even more. The argument based on a Condorcet like paradox turned out to be mistaken, because it was based on an inconsistent belief system, but realising inconsistencies will lead to adjustments of one's belief system.
5.6 Summary The arguments against taking moral realism based on the WittgensteinDavidson approach as a background for an articulation for an ideal of equality, were dealt with. The charge of conservatism was rebutted. First by showing that within conservatism there could be a role for ideals and rational change. It appeared that the threat was actually imperialism of the interpreter's own view. This imperialism would mean a kind of idealism, which would undermine the moral realism itself and in particular the idea of taking views of other persons seriously. But moral realism could be saved from this charge by turning to a view di erent from a sociological or community view on rules and rule following for example a view explained by Baker and Hacker. In this non-sociological view a rule is considered not to be a rule because it is recognised as such by others of one's own community, but is considered to be a rule because it is something beyond the subjective phenomenology of the inner world, something `out there' and consequently, recognisable for others. It appeared that instead of being responsible for imperialism, the principle of charity saved radical interpretation from being imperialistic. By this principle all other beliefs than we accept for ourselves can be recognised by us as beliefs. Other person's views can be recognised. It was shown that the alternative to realism, quasi-realism was not superior to the moral realism based on radical interpretation. It was argued that within radical interpretation, moral realism can account for the ban on mixed worlds, worlds in which supervenience is not satis ed. This ban re ects what the practice of moralising is, just what quasi-realism mentions as reason. It can accept supervenience but it will not accept non-moral properties as the reason or as the ground for the moral properties, although it can accept they are implied by them because they happen to be coextensive.
5.6. SUMMARY
179
The arguments against radical interpretation holding that there could be lacking a common background on which there is agreement, and so lacking objectivity, was rebutted by pointing to the restricted validity of the Lowenheim Skolem theorem, and to the holism accepted in radical interpretation. There is no a priori distinction between several kinds of propositions, they all have to be interpreted similarly. A Condorcet-like argument was shown to be invalid. The argument of the possibility of disagreement is recognised to be harmless once one points to the impossibility of infallible knowledge. Within the coherence account of knowledge, there is no place for infallible knowledge and there will be room for some disagreement. The ideal of equality can be located in the domain of the objective. The conclusion of this chapter is that moral realism that is based on radical interpretation is not undermining the reasons for it being a proper framework for the development of an ideal of equality. And so I will proceed in the next part with the determination of the meaning of equality within a moral realistic framework underlying moral particularism, realistic individualism and moral value pluralism.
180 CHAPTER 5. THE NEW FRAMEWORK AND ITS CRITICISMS
Part II
Equality
181
Chapter 6
The meaning of the ideal of equality 6.1 Introduction After the preliminary analysis in the foregoing chapters in which a moral realistic framework together with moral particularism, realistic individualism and moral value pluralism, was introduced as an alternative to the traditional framework, the meaning of an ideal of equality is discussed in this part. In this chapter, I argue for an ideal of equality such that it is clear that it is indeed an ideal. As already stated in the introduction, I do not discuss the policies that should be chosen in order to arrive at a more egalitarian world and consequently, I do not discuss the question who is responsible for inequality and who should do something about it. I analyse the structure of the ideal of equality, such that it will become clear what the arguments invoking the ideal of equality are pointing to, and how distributions are to be evaluated with respect to the ideal of equality. In order to develop the ideal of equality two methods are available: 1. look directly for reasons why equality is desirable, 2. look at the arguments that are raised against equality and determine from these what equality should mean. In the second chapter on moral universalism, some arguments for equality were mentioned, for example the conceptual relation between judgements from a rst-person perspective and a third-person perspective, but the content of the ideal of equality remained still hidden. By elaborating 183
184 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY the arguments for equality directly, one runs the risk of ending up with a stipulative arbitrary de nition which is beside the political philosophical discussion. This danger is less by following the second method, because in that case one is already participating in the political philosophical dialogue. Although it is less direct, I will follow this second method and start in the next section with the examination of some arguments that are raised against equality. These arguments against equality have to be answered by any ideal of equality which makes sense. In section 2 the arguments against an ideal of equality are presented. It is argued that moral value pluralism of the new framework can answer the objections against equality which are based on the idea that there are more ideals than equality that are important and based on the idea that there is one equalisandum. The arguments against the ideal of equality stating that egalitarians have something important to say but just miss the point, receive most attention. These arguments hold that a situation is unacceptable not because of inequality, but for instance the fact that some have less than the bare minimum. In these indirect arguments against equality the so-called levelling down objection has a central role. This objection holds that nothing is gained when the better-o become worse o and the worse-o do not improve at all, while according to equality this would be an improvement in at least one aspect. The levelling down objection is discussed in the sections 6.3 and 6.4. Its background is discussed. From this discussion the meaning of equality as an ideal arises. This meaning is explained more extensively in section 6.5. The main idea is that the ideal of equality is a situation in which the distribuenda are distributed such that all are equally as well o as is possible for all with respect to an equalisandum. Inequality is considered to be bad for some persons, namely the worse-o , because they are worse o than all might have been simultaneously if the distribuenda were distributed di erently. In section 6.6 some problems concerning this view are discussed, namely those concerning the determination of the distribution such that all are as well o as all could have been simultaneously. Finally, in section 6.7 it is shown that the rival views which inspired the indirect arguments against equality, can be accounted for in the view on equality developed in this chapter. In the following chapters this meaning of the ideal of equality is the basis for the construction of a measure of inequality.
6.2. ARGUMENTS AN IDEAL OF EQUALITY HAS TO ANSWER 185
6.2 Arguments an ideal of equality has to answer In discussing the arguments any ideal of equality has to answer, I con ne myself to the arguments against equality as an ideal in itself. I do not consider the arguments that deny equality to have the desirable consequences it is claimed to have, such as for instance a higher total welfare. This consequence of equality is for example questioned by arguing that the presupposition of the law of diminishing returns in utility of resources is not valid.1 This type of arguments concerning the consequences are not discussed by me. The arguments I turn to concern equality as a desirable political ideal in itself more or less independent of its consequences, they concern the essentials of equality as an ideal. Two di erent kinds of arguments against equality can be distinguished, namely those holding that: 1. equality is not always the most important ideal, 2. equality is not an ideal at all. The arguments belonging to the former kind do not deny that equality is desirable but they deny that equality always overrides other political ideals such as eciency, liberty, or perfection etc. Many situations are pointed to in showing that equality is not the most important ideal. For example it is mentioned that it is not defensible that we should all become blind because some are, or that we should let our legs be amputated because some are missing these extremities. These arguments are indeed serious arguments against radical egalitarians who hold equality the most important ideal, a view dicult to defend. It might be argued that there might be situations in which radical egalitarianism should be claimed to be the right view, because deviating from this idea would give too much ammunition for those resisting every change towards a more egalitarian world. This consideration however is of a di erent kind than I want to discuss and is beyond the scope of my exploration. I do not concern myself with strategic considerations for arriving at a more equal world in political bargaining, in which indeed it could sometimes be justi ed to present a less di erentiated version of one's ideals. Here I am interested in the structure and meaning of the ideal of equality which determines the meaning of arguments invoking equality as an ideal. By admitting that equality is not the most important ideal and that it can be overridden, it is not held that equality has no value at all. Consequently, the arguments that equality will have sometimes undesirable 1
See for this argument [Frankfurt, 1987].
186 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY consequences, as is held by Lucas and Frankfurt, do not show equality to be undesirable, but merely that equality is not always the most important ideal and can be overridden by other ideals that show why the consequences are undesirable [Lucas, 1965] [Lucas, 1977] [Frankfurt, 1987]. It might even happen that equality in fact never overrides any other ideal in the actual course of this world, but that it remains nevertheless, valuable. The di erence between this idea and the idea that the ideal of equality is not important at all, is that in the former view it merely happens to be the case that it is never overriding, in the latter it is not merely contingently so. In the latter, equality is denied to be valuable at all. The former holds that it is possible to imagine a situation in which equality is rightly acted upon, the latter does not. The consequences of this di erence turn up in the descriptions of the political choices. In holding the former view alternatives are described as more or less unequal. By accepting the latter the whole issue of inequality is irrelevant and consequently it will not be a relevant part of the description of alternatives. The arguments against radical egalitarianism are not a threat to the ideal of equality considered as one ideal among others. In chapter 3 moral pluralism was argued for and so radical egalitarianism was already argued against. The whole idea of equality was argued to be only reasonable within a pluralistic framework, because it presumes that there are at least two kinds of values namely those concerning the distribution and the value of what has to be distributed. The arguments against radical egalitarianism are not a threat to the ideal of equality within a moral pluralistic framework in which it is considered to be one ideal among others as moderate egalitarians hold.2 The arguments mentioned so far against radical egalitarianism do not touch the moderate egalitarians. But there are also arguments against moderate egalitarianism. Two kinds of arguments can be distinguished: 1. arguments questioning the proper equalisandum, 2. arguments questioning the proper reasons for the badness of unequal distributions. The former kind denies that a particular ideal of equality in which some particular good should be equalised, is desirable. Every particular ideal of equality proposed is argued against in the name of equality itself, but with a di erent equalisandum, i.e. the good which should be distributed equally. According to Cohen and Sen these are the most important arguments against a particular ideal of equality [Cohen, 1989] [Sen, 1992]. It 2
See chapter 4 on indexed values p. 129.
6.2. ARGUMENTS AN IDEAL OF EQUALITY HAS TO ANSWER 187 is admitted that equality itself is an important political ideal but every suggested equalisandum is denied to be the right one. Against each proposal one can nd some examples which show the equalisandum not to be the proper one. For example as was already discussed in the foregoing chapters against the idea that welfare is the correct equalisandum it is argued that resources should be the equalisandum, which in its turn is criticised because opportunities or capacities, or liberties should be the equalisandum etc.[Dworkin, 1981a] [Dworkin, 1981b] [Cohen, 1989] [Sen, 1992, p. 15].3 Against each particular proposal one can nd some examples showing that it is not the right one. This lead Lucas to the argument against the ideal of equality, that equality with respect to one good will introduce inequality with respect to another [Lucas, 1965]. Lucas holds that moderate egalitarianism is restricted by acknowledging multiple equalisanda. He presents his plurality of equalisanda as a way of compensation; inequality with respect to one good can be compensated by inequality with respect to another good. According to Lucas the role of the ideal of equality is diminished by this. Although I agree with Lucas on the recognition of multiple equalisanda, I disagree with him on his view that a plurality of equalisanda is restricting moderate egalitarianism itself. It merely shows that no particular equalisandum is the most important. The idea of multiple equalisanda follows in a natural way from moral value pluralism. Besides this answer to Lucas, it also seems that Lucas neglects the actual correlation between different forms of inequality. Being worse o in one respect is highly correlated in this world with being worse o in another respect; it is not compensated as Lucas is suggesting. For example inequality with respect to mortality and morbidity is correlated with social economic inequality [Townsend & Davidson, 1982].4 The idea that one sort of inequality compensates for another cannot be pointed to in order to devaluate the ideal of equality. Within pluralism it is not compensation, which is the reason for multiple equalisanda, but the recognition that there is more than just one general good.5 See especially chapter 4 p. 135 . Walzer is sensitive to this trespassing of the domains of particular equalisanda. It is remarkable that although nationality is one of the most important determinants of possession of other goods, as is clear by looking at the tables in the World Development Reports of the World Bank, Walzer is not discussing this correlation in the same critical ways as the other links. [Walzer, 1983] 5 The fact that their distributions are highly correlated can be seen as an indication that in a society indeed one good is held to be the most important one, which calls for a criticism to that apparent monism. 3 4
188 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY Summarising, the arguments which criticise equality because it takes the wrong equalisandum are only arguments against equality in so far they are linked with the idea that there is just one equalisandum. Once pluralism is turned to, these arguments do not touch any longer the ideal of equality. The arguments turn into arguments showing that equality is important with respect to several equalisanda. Let me discuss next the other kind of arguments.
6.3 The levelling down objection The second kind of arguments against the ideal of equality as one ideal among others are those arguments questioning the proper reasons for the badness of unequal distributions. The basic idea is that egalitarian ideals run parallel partially with other political ideals. This parallelism turns into an argument against equality. It is argued that inequality does not matter at all because its role can better be taken over by the other ideals. Situations exhibiting inequality are presented and it is questioned what is wrong in it. Critics of equality argue that it is not the inequality that is wrong but something else, for example that some are below a certain minimum level of welfare, or do not have enough goods to lead a human life, or that the worse-o are denied priority [Lucas, 1965] [Lucas, 1977] [Frankfurt, 1987] [Par t, 1989]. These arguments with di erent interpretations of what is wrong in the situations in which there is inequality have one common element. This common element is what Par t calls `the levelling down objection' [Par t, 1989]. On the ground of this objection they claim that their own reasons why there is something wrong in situations in which there is inequality, are more plausible than inequality. The levelling down objection runs as follows, illustrated by gure 6.1. Suppose in situation A everybody is equally well o with a certain equal standard of living and in situation B half of the people are on the same standard of living as in A, but the other half is better o . Moderate egalitarians, who admit also the importance of other ideals, can judge B to be better than A `all things considered', but worse in one aspect, namely worse with respect to equality. According to the minimum view advocated by Lucas, B is better than A and worse in no aspect, assuming that all are above the minimum. In Frankfurt's opinion too, situation B is better than A and not worse with respect to any ideal, assuming of course that the people in A and B have enough. Par t, who argues for the so-called `priority view', according to which bene ting people matters more the worse o these people are, also judges B to be better than A and worse in no aspect. The principal di erence between moderate egalitarianism and
6.3. THE LEVELLING DOWN OBJECTION
A
189
B
Figure 6.1: Illustration of the levelling down objection one of the three views mentioned, is a di erence in evaluating the situation A versus B . The moderate egalitarians can judge B all things considered better than A but still worse with respect to the ideal of equality. The others judge B better than A without there being even one aspect in which B is worse than A. The main problem now is of course: `Is it possible that B is worse than A with respect to an distributive ideal?' In order to answer this question, I turn to the background of the view that B cannot be worse than A in any distributive aspect. The assumption is the idea that B cannot be worse in any such aspect, because no one is worse o . The view that a situation like B is in no distributive aspect worse than A is inspired by the idea that something can only be worse if it is worse for someone in some aspect. Because according to the levelling down objection in B no one is worse o in any aspect than in A, moderate egalitarianism is refuted. This idea that nothing can be worse if it is not worse for someone is seen to be expressed in two ways: one is the Pareto principle, the other is called by Broome the principle of personal good [Broome, 1991, p. 36, 165].6 Broome de nes these: 1. Pareto principle: Two alternatives are equally good if everyone is indi erent between them, and if everyone either prefers the rst of two alternatives to the second or is indi erent between the two and someone de nitely prefers the rst, then the rst is better than the second. 6
See also chapter 3 p. 64 .
190 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY 2. Principle of personal good: Two alternatives are equally good if they are equally good for each person, and if one alternative is at least as good as another for everyone and de nitely better for someone, it is better. The rst was met already in chapter 3 where it was seen to be possible, following the idea of Hochman and Rodgers, to argue for a redistribution of goods in accordance with the Pareto principle once one acknowledges external e ects, i.e. the in uences of how others fare on someone's evaluation of the situation.7 There is however an important problem with invoking the Pareto principle as a ground for the levelling down objection. It is its restriction to preferences, which was the reason it could be seen as representing volitional individualism. Because, what is meant by preferences, actual preferences or preferences which are rational, or mere morally acceptable preferences or also bad or anti-social preferences? It is precisely the referral to preferences which makes the Pareto principle implausible. What is good is not simply based on preferences. Something can be preferred by someone although it is not good for him or her. Similarly, something can be good for someone without being preferred by someone, precisely what the Pareto principle denies.8 Furthermore as was argued in the former chapters, the Pareto principle need not be accepted because a pressing reason for accepting it, namely volitional individualism, was argued to be better replaced by realistic individualism based on moral realism. This opens up the possibility to other principles than the Pareto principle. The Pareto principle is not the proper basis for the levelling down objection, what about the other principle, the principle of personal good? The principle of personal good is more promising than the Pareto principle because it starts directly with what is good and not with what is preferred by people. The way the levelling down objection is to be answered will depend on one's view on equality. Is inequality bad for someone, or is it bad in general not for someone in particular? Broome calls the view that inequality is an individual harm to people individualistic egalitarianism. If inequality is seen as something bad in general and not bad for someone, it is called communal egalitarianism [Broome, 1991, p. 177-182]. Communal egalitarianism is directly threatened by the levelling down objection. But if inequality is a personal harm the levelling down objection looses its force, B is worse for some with respect to equality. The levelling down objection is See chapter 3 p. 70 . This is the core of Temkin's argument against the desire ful lment theory as basis of what he calls `the slogan' underlying the levelling down objection [Temkin, 1993, p. 264-272]. 7 8
6.3. THE LEVELLING DOWN OBJECTION
191
circumvented in individualistic egalitarianism unless one turns to a general good as equalisandum in which the individual harm of inequality is already incorporated. In order to get a clear view on the issue it is useful to distinguish between the goods that can be distributed directly, the distribuenda, and the goods that should be distributed equally, the equalisanda. The distribuenda should be distributed in such a way that the equalisanda are distributed equally. In general, the equalisanda cannot directly be distributed, for example welfare or transport capability; they will be distributed indirectly by a particular distribution of other goods for example income. Looking at the descriptions of situations like A and B in gure 6.1, one should realise to what one is looking, do they refer to the distribuenda or to the equalisanda? In general, when one is looking at the descriptions of situations like A and B in gure 6.1, one is looking at how well people fare with respect to distribuenda. With respect to this, the individual badness due to inequality is not yet incorporated in the gures and inequality can be represented as an individual harm. But once one has turned to an equalisandum as comprehensive as a general good as Broome does, there is nothing left but some communal egalitarian view if one wants to point to morally relevant inequality in situation B . The badness of the individual harm of inequality is then already incorporated and reckoned with. So, if one restricts oneself to particular distribuenda or particular equalisanda, one can answer the levelling down objection by arguing that situation B is worse for some with respect to one ideal namely equality. Are the situations interpreted as concerning the general good in which the individual harm of inequality is incorporated, this cannot be the answer to the levelling down objection and one should turn to some communal interpretation of equality, which has to be argued to be reasonable vis-a-vis the levelling down objection. Considering just one general good is like turning to monism, which was argued in chapter 4 to be highly suspicious. It is dubious whether there is something like the general good which should be equally distributed. Because if in this general good the individual harm of inequality is already incorporated, it is suggested that the individual harm of inequality can be compensated for by something else. But I do not understand how the individual harm of inequality can be compensated for. Compensation is a particular good in a particular situation for lack of a particular other good. Could it be the case that one should be compensated for inequality with respect to mobility by money to enable a connection to for example internet? Here the equalisandum could be considered to be receiving information. Lack of ways of receiving information can be compensated for but not the inequality. How should I understand compensation if the general good, which incorporates everything including equality, is the equalisandum? I
192 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY cannot answer that question. It makes sense to restrict oneself to particular equalisanda and reject the description which Broome is suggesting, that in the situations referred to by A and B the individual harm of inequality is already incorporated in the amount of general good. Rejecting this general good opens up the possibility of an individualistic interpretation of equality, without the threat of a return of the levelling down objection in which the individual harm is incorporated in the equalisandum. The question whether such an individualistic interpretation is intelligible is dealt with in the next section. The answer is arrived at by looking more precisely to the meaning of the principle of personal good which suggests an interpretation of equality that is congruent with this meaning.
6.4 Person a ectingness In order to arrive at an individualistic interpretation of equality such that the levelling down objection can be answered, I have to discuss rst the meaning of the principle of personal good. This principle expressing the idea that something can only be worse or good if it is worse or good for someone is too much linked with the idea of there being an all encompassing good for individuals. The idea that something can only be worse or good if it is worse or good for someone can better be stated as so-called person a ectingness of ideals or goods. Person a ectingness ts better into the moral pluralistic framework. It does not suggest that there is one general good. The meaning of this person a ectingness can be made more clear by looking at the so-called non-identity problem and out of the discussion on that problem, the contours of an ideal of equality arise in which inequality is considered to be bad for someone and not bad in general, for which, as was argued, the levelling down objection is no threat.9 9 Following Temkin, Par t discusses an example showing that we accept that a situation is worse while it is not worse for someone [Temkin, 1993, p. 260] [Par t, 1989]. Suppose there are two situations: a situation P in which some criminal has a certain standard of living in which he did not su er any reduction of welfare due to his wrongdoing, and a situation Q in which he is worse o than in P. Suppose further there are no further e ects on the behaviour of other would be criminals. Nobody but the criminal himself knows he is living. According to Temkin we could still judge situation P worse than Q although nobody is worse o in P. So it seems that in a retributive view on punishment the person a ectingness is violated and so it is not a universal characteristic of ideals. The example is not convincing. In a retributive view on punishment the way the criminal became worse o is not irrelevant. Situation P is not judged to be worse than Q, it is regretted that the criminal is not punished in P, but the same should be said of Q. That the criminal is worse o in Q does not give any compensation for the lack
6.4. PERSON AFFECTINGNESS
193
The non-identity problem is a problem introduced by Par t in his Reasons and Persons, in the debate on justice and future generations [Par t, 1984, p. 366-371].The person a ecting view would deny any importance of non-existent persons. It is refuted by Par t by an example in which there is no di erence in moral evaluation between two medical programs although in one the individual to which harm is prevented can be pointed to and in the other not, because `the individual' does not yet exist. In short the example is: Suppose there exists a condition of pregnant women Ct, which will cause the child she is carrying, to become handicapped in a certain way. By a simple treatment this condition can be cured and the handicap of the child will be prevented. The condition Ct can be detected by a particular test. Suppose further there is another condition Cp which will result in conceiving a child with the same handicap as is the consequence of condition Ct if the women become pregnant. There is no treatment for this condition but it is known to disappear within two months. By detecting the condition and postponing pregnancy the handicap can be prevented. Two medical programs are possible. One consists of testing pregnant women for condition Ct and if the test is positive start the treatment. The other is testing women for condition Cp if they intend to become pregnant. In both programs the same number of handicaps are prevented. The programs are equally expensive and only one can be implemented. I agree with Par t that the two programs are equally worthwhile. Person a ectingness however will lead to a di erent judgement. In the program that is based on the detection of condition Ct harm is prevented of known and individuated children. Particular children are bene ting from the program, they are a ected by the program. In the other program no child is of punishment. That he fares worse in Q than in P is not relevant unless he fares worse because he acted wrongly. Only if there is a connection between the wrongdoing and being worse o , there is something good according to the retributive view. Any disadvantages other than e ects of punishment are not important at all. Punishments in a retributive view are essentially of a deontic or procedural nature. Not the fact that the criminal is worse o is important, but the way he became worse o is decisive. Any deontic principle or ideal implies of course that something can be wrong although it is not wrong for somebody, because not only the consequences count but also who or how it was brought about. This means that the idea that person a ectingness holds for consequentialistic ideals, is not yet refuted and it is this refutation that I am looking for in the discussion about equality. My main concern in this study is with the ideal of equality with respect to which situations are to be evaluated and not who is responsible or how it was brought about.
194 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY bene ted because they do not yet exist. According to the person a ecting view the program based on the detection of condition Ct is better than the program directed to condition Cp . Accepting that both programs are equally good, implies the rejection of the person a ectingness as a universal characteristic of moral judgements. It seems to be a counterexample to the person a ectingness of values and goods. It seems to be established that person a ectingness is not a universal characteristic of goods and ideals seemingly undermining the background of the levelling down objection. But before we conclude the levelling down objection to be undermined we have to explain why person a ectingness does not t our moral intuitions in the example. An explanation of why person a ectingness is refuted in this non-identity case is that it can be interpreted as a case showing interpersonal incommensurability of goods. Person a ectingness works as if we cannot compare in a relevant way the welfare of the persons before and after a decision or action if the consequences of these decisions concern situations inhabited with different people. Person a ectingness behaves like the idea of impossibility of interpersonal comparison, comparisons are held to be only possible if they concern one and the same person. In the program directed at Cp there is done no good according to person a ectingness because we cannot point to individuals which are bene ted, there are no individuals before and after the execution of the program which can be compared as is the case in the program directed at Ct. But once interpersonal comparisons are recognised to be possible and especially the reason how interpersonal comparisons are possible, this reasoning will not be correct any more. In chapter 3 it was argued in introducing realistic individualism, that interpersonal comparisons are possible and are a necessity for interpreting each other, indicated by the principle of charity [Davidson, 1980] [Davidson, 1984] [Davidson, 1986]. It could be held that we judge how others fare, by comparing them to ourselves and not just to the persons themselves. This can be seen to be an explanation why person a ectingness runs into troubles in the non-identity cases, because it is like denying comparisons between di erent people. Although person a ectingness as it was formulated is refuted by the non-identity case, it remains possible to amend it in such a way that interpersonal comparability is not denied. It could be read as: A situation A cannot be worse than a situation B if there is no person in A worse o than a person in situation B . This amended version is congruent with our intuition in the above nonidentity problem and saves the essential idea of the principle of personal
6.4. PERSON AFFECTINGNESS
195
good and person a ectingness, tting nicely in realistic individualism.10'11 Person a ectingness was introduced as a way to preclude the consequence of utilitarianism that we should make as much happy people as possible until the marginal total utility of persons becomes zero [Par t, 1984]. The amended version of the person a ectingness can still be used to preclude this interpretation and block the utilitarian procreative demands. The amended version however, can also still be used as the background for the levelling down objection. Let me turn to this objection again. It can be argued that the situation B in which half the people are better o than the other half is overall considered better compared to the situation A in which all are as well o as the worse-o in situation B . Moderate egalitarians can agree with this judgement but they hold that there is an ideal with respect to which B is worse than A. This would be not possible if the amended person a ectingness was valid. But let me look to what precisely is worse in B . Of course it will be told that it is worse with respect to equality. But what does that mean? An answer is that B is worse because it is regrettable for the worse-o that they are not as well o as would be the case if the distribuenda were distributed in such a way that all are equal with respect to the equalisandum at hand. Because of comparing B with what might have been the case, there is something regrettable for some persons which explains why B is worse in one aspect than A in which there is nothing regrettable. That some are worse o than all could have been, can be taken as the basis for the ideal of equality. But more has to be said. One could question this kind of reasoning, it would imply that there are no situations unambiguously better, because one can always imagine that it might have been still better. So, even the situation in which everybody being equally well o is not ambiguously good because it might have been still better if everyone got still more of the good than they actually have. Consequently, there can always be argued that there is something regrettable. An improvement for all can still be worse in one aspect, namely once improved it might show up that there is a situation that is still better for all. This criticism however will loose its force once we look a little more into the interpretation of the counterfactuals which are responsible for the content of what is regrettable. 10 This amended version is more or less the meaning of the `grading principle' of Suppes in combination with the `identity principle' [Sen, 1979a, p. 150-156]. 11 The amended version of person a ectingness is much weaker than the same number quality claim or Q in [Par t, 1984, p. 360], but it is consistent with it, and also with the so-called theory X. It is remarkable that Par t does not explicitly note that in his examples the principal questions are those concerning equality as in the Jack and Jill example in [Par t, 1984, p. 395].
196 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY In describing or interpreting counterfactuals and imaginary examples we need to take some elements, which are not mentioned at all, to be similar as in the actual situation and some di erent. Which elements should be taken as similar and which not, depend on the context of the inquiry [Rescher, 1964] [Putnam, 1992]. For example it is highly ambiguous what is meant by the counterfactual question: `Suppose I had been born blind what sorts of jobs would be available for me now?' Does the counterfactual also include that my education would have been di erent, that my parents would have had other attitudes towards me, teached me other things, or does it merely imply having no sight and no concepts related to it but by hearsay? If we are discussing practical and technical handicaps in some jobs, the second interpretation will be meant, if we discuss psychological and emotional characteristics, the second is not capturing the relevant issues and the rst was meant. The same goes for the counterfactual situations taken as basis for what is regrettable in inequality. The meaning of the ideal of equality is linked to the interpretation of counterfactuals. Interpreting the latter can give insight to the meaning of the ideal of equality. This interpretation is discussed in the next sections after summarising the discussion so far. In this section and the previous section, the arguments against equality with the levelling down objection as their common element were discussed. Pointing to the Pareto principle as representing the essential idea behind this objection will not do, because of the di erence between preferences and what is good. Something bad can be preferred and something good might not be preferred. The Pareto principle cannot be invoked to back up the levelling down objection. The other representation based on the principle of personal good seemed to be more promising. This principle however was linked too much to the idea of one all encompassing general good into which inequality as individual harm is considered to be possibly compensated for. Therefore it was not a plausible backing of the levelling down objection either. Inequality cannot be seen as something which can be compensated for. Compensation is a response to lack of a particular good in the form of another particular good in a particular situation. Inequality can be interpreted as bad for someone, but the inequality one su ers from is itself not lack of some particular good called equality that can be compensated for by some other good. It is better to consider the levelling down objection as a consequence of so-called person a ectingness. Although it was shown that person a ectingness was refuted by the nonidentity cases, an amended version of it in which interpersonal comparisons of goods are accounted for, consistent with realistic individualism, turned out to be still a good interpretation for the levelling down objection. To this objection an individualistic interpretation of the ideal of equality would be an answer. In such an interpretation inequality is considered to be bad
6.5. THE MEANING OF EQUALITY
197
for some people and not bad in general. It was suggested that inequality can be seen as an individual harm for some people namely for those worse o than all could have been simultaneously.
6.5 The meaning of equality The answer to the levelling down objection to the ideal of equality suggests an individualistic interpretation of the ideal of equality according to which inequality is something bad for some individuals. It is bad for those persons who are worse o than they might have been in a situation which would be ideal with respect to equality. Persons who are worse o than they might have been in this ideal situation can have a complaint. It is not meant that they should complain or that if they do not complain there is nothing wrong. With complaints of the worse-o is meant here merely the fact that they are worse o than they could have been if the distribuenda were distributed in such a way that all are equal with respect to the equalisandum. These worse-o are the focus of the ideal of equality.12 It is not denied that the better-o can be concerned with inequality, but they have nothing to complain, the worse-o have something to complain. In several measures of inequality, which will be described more extensively in the next chapter, this idea is subscribed to. The complaints are relative to an ideal reference situation. Sometimes the average of the distribuendum is suggested as a reference, sometimes in an aggregate way the so-called `equal equivalent distribution' (which is an allocation such that all have an equal amount of distribuenda such that the aggregation of equalisandum is equal to the aggregation of the equalisandum of the actual distribution) is proposed as a reference in combination with the average [Kolm, 1968, p. 186].13 But neither the average of the distribuenda, nor the equal equivalent are the proper references. The former not because what matters is not equality of the distribuenda but equality of the equalisandum. It is not goods but what goods do to people which is relevant. The latter will not do because the situations in which there is equality are not restricted by the aggregate of what goods do to people but are restricted by the distribuenda. The restricted availability of distribuenda is the reason of the restriction to how well people fare with respect to a particular equalisandum. A more plausible reference is that situation such that all persons are simultaneously equally well o with respect to the equalisandum, it is the best all can collectively get. 12 This idea is seen in a rough form in [Honderich, 1980, p. 23, p. [Vermeersch, 1988] and [Temkin, 1993, p. 19,44,45]. 13 See also chapter 7 p. 226.
138]
198 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY The ideal to which is referred by the counterfactual is such that by redistribution of the distribuenda all are equally well o . To be more formal. Let A be the allocation P ofPthe distribuenda then the reference is that allocation A such that A A and the equalisandum xi for a person i which is dependent on the distribuenda is maximalsuch that 8i; j xi(A ) = xj (A ).14 The equalisandum should be equally available for all simultaneously. The ideal of equality is concerned with some having less than what could have been possible for all. Although inequality is an individual harm, equality is not good because it is good for someone, but because of the absence of individual complaints. It is not important that some have more than others, but that some have less than might have been the case for all. The idea of collective availability is explained in a clear way by Cohen in an analysis of proletarian unfreedom [Cohen, 1983]. He argues that a proletarian is unfree, not because no individual proletarian can escape the fate of selling his labour, but because they cannot escape this fate collectively. Just as the case in which some goods are available to some but not to all, it is available to each one but not to all collectively. Cohen illuminates the idea of collective availability by an example of 100 guests on a tour in some interesting town. The tour operator organised a special trip with a bus with only 30 places. If the company just predicts that only 30 would enjoy this trip and so would send another bus if there happened to be more people interested, then there is equal collective availability. It is di erent if there is just sent one bus with 30 places, then there is collective unfreedom and inequality The idea of collective availability was touched upon already in the ideas of the universalists in chapter 2, especially in the presentation of O'Neill's interpretation of Kant. There it was stated that all actions should be performed under the idea that all can act according to that maxim on behalf of which was acted. It was argued to be a reason for equality.15 Here I propose a similar idea of equality but with an an important di erence in its function. There it was introduced as a universal principle leading to an objective morality. Here it is an illumination of a political ideal namely equality, which is one ideal among others. It could be said that here universalism did recover its moral character, which it threatened to loose in rationality. 0
0
0
0
14 The question whether the equalisandum x for some person i is dependent on the i whole distribution A or only on what i receives ai is only a matter of representation. The inegalitarian consequences of external preferences as put forward for example by Rakowski following Dworkin, are no longer possible here because as argued in the text, particularly in chapter 4 p. 135 ., equality does not concern preferences but what is good for people [Dworkin, 1981a] [Rakowksi, 1991]. 15 See chapter 2 p. 27.
6.5. THE MEANING OF EQUALITY
199
The principle of equality just invokes the idea that we are all equal, meaning that what is a reason for me is a reason for you too. Of course there can be di erences in our reasons for acting or holding ideas etc., but in general what is important for me is important for you too. We cannot do without it. It is assumed in interpreting each others actions and utterances. It is as Davidson argues, a necessity for communication. We are bound to assume similarity. Without it we could not compare ourselves with others, we could not even state that we di er from others. Similarity is the basis of it. It is assumed in the principle of charity, the principle which enables radical interpretation. We see ourselves as one person among others. We assume other persons to see and feel the same as we ourselves do. Of course there can be di erences, but they can be explained only on the basis of general similarity. The necessity of assuming, what is important for me is important for you too, is the basis for social equality as a political ideal. Rejecting this equality is seeing oneself as a practical solipsist.16 Although it is not impossible, a person who thinks himself to be the only one in this world is, as explained in the former chapters a being, without language without rationality, in short who is not a human person at all but something else.17 Accepting the underlying idea of equality means that for example in judging philosophy being important, one believes doing philosophy oneself is good but also doing philosophy by others is good. If there show up some di erences of opinion about what is valuable, then it should be recognised that these di erences can only be stated by some other similarity to which one could turn. But further, this background of equality indicates the source and the solution of the con ict between agent-relative and agentneutral evaluations which was mentioned in answering the argument of Williams against moral peception.18 Equality favours the agent neutral view. Turning to particular equalisanda makes also clear that the idea that the ideal of equality is limited by agent relative values, as was stated by Nagel in holding that equality interferes too much with the personal life of persons, is not correct.19 Nagel's view is convincing because it was stated in terms of a general good or preference satisfaction, then one could tell in a plausible way that equality does interfere with one's own projects. It is a plausible story because the projects of di erent people can di er widely, hiding the similarity because of its abstract character. But if one turns to This term is borrowed from [Nagel, 1978]. See chapter 2, p. 37 . and chapter 3 p. 101 . Chapter 2 p. 49 .. This view of equality being limited by claiming some room for someone's personal projects is stated by Nagel [Nagel, 1991, p. 50,75-80]. 16 17 18 19
200 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY particular actions or values it is loosing its plausibility. It is dicult to state that doing philosophy oneself has to take priority to doing philosophy of someone else. And to these particular actions and values one should turn, because as was argued in the sections above, the general good or preference satisfaction in general are not proper equalisanda. The underlying idea of equality shows that it is the cement or ground of society, not contingently but essentially. It is recognised to be the cement in the idea that one should justify one's needs, one's theories, and judgements. This is only possibly on behalf of recognition of similarity. Equality is an important aspect, hard to reject. Sen writes: It is also of considerable pragmatic interest to note that impartiality and equal concern, in some form or other, provide a shared background to all the major ethical and political proposals in this eld that continue to receive argued support and reasoned defence. [Sen, 1992, p. 19] For example Wiggins states that a moral system S which is and can be lived by, will recognise an abstract claim right or entitlement to x under condition C just where x is something the denial or removal of which under conditions C (and can be seen as giving) gives the person denied or deprived part or all of a reason, and a reason that is avowable and publicly sustainable within S, to reconsider his adherence to the norms for reciprocity and co-operation sustained by S. [Wiggins, 1982a, p. 32] These claim rights can only be interpreted against the background of similarity of people, i.e. what is a reason for you is a reason or another too. The same can be said about contract theories of for example those of Scanlon or Kolm [Scanlon, 1982][Kolm, 1985]. These too point to what is common to all, they presuppose similarity; the idea of unanimity can be seen as re ecting what is simultaneously equally possible for all. Not only in political philosophy but also in the political arena itself, is the idea expressed that equality refers to a collective possibility of receiving some good. It is for example seen in the Queen's speech of 1994 on the actions to be taken by the Dutch government concerning medical services:
6.6. THE REFERENCE
201
In order to keep indispensable health care accessible for all, a tight control on the development of volume and costs is necessary.20 This expresses the importance of collective availability of medical services of all. If there is no control about volume and costs, then the collective availability is threatened. This analysis shows clearly what is wrong in situations exhibiting inequality. It is not directly the comparative inequality which is wrong, but that some are worse o than they might have been in a reference situation in which everyone is equally as well o as possible. It is not relevant that others are better o , but it is relevant that a redistribution of distribuenda could improve the worse-o . This means that egalitarian ideas are not just expressing feelings of envy, being worse o than some others simpliciter, but being worse o than might have been the case if the distribuenda were redistributed such that all were equally well o . The inequality is to be seen in relation to the reference. This reference situation, which is determined by what is collectively equally available for all, is based on the underlying idea: what is important for me is important for you too. This idea is an essential central element of society. To be clear that equality is locked into other ideas about society as for example contract theories is not a defence of equality, but only explaining why equality matters and showing how it is interconnected with the other ideas about society. So far, only the contours of the individualistic interpretation of the ideal of equality are drawn. In order to ll in these contours several questions have to be answered. The rst is of course, how is this reference determined? The second is, is this ideal of equality indeed answering the levelling down objection? And nally, how does this idea of equality transform into a measure of inequality and how is this measure related to the other measures. The former two questions will be answered in this chapter. The last question will be answered in the following chapters.
6.6 The reference The reason why inequality is an individual harm is that those worse-o can complain that they might have been better o in a reference situation. But how is this reference of the ideal of equality determined? In the reference all are simultaneously equally as well o as possible with respect to the particular equalisandum. There is no situation possible in which all are 20 `Om essenti ele gezondheidszorg voor iedereen toegankelijk te houden is het nodig de ontwikkeling van volume en kosten strak te beheersen.' [NRC handelsblad 1994, 20 september, p. 23]
202 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY
A
B
Figure 6.2: Illustration of the vanishing di erence between the ideal of equality and the ideal of perfection equally better o with respect to the equalisandum. But what is meant with `possible'? That is not yet made clear. What is the relation between the situation which is to be evaluated and the reference? Is it meant with `possible', somewhere in the future after more economic development due to development of production and skills. If this is meant it could be argued that all are below the reference, all could have a complaint that they might have been better o taking this reference in the future into account. Indeed this argument can be discovered in the form of an argument against instead of in favour of equality. Consider for instance situations A and B in gure 6.2. If there is a way to transform situation A into that of B , in other words if B is an alternative to A, then one could argue that B should be the reference, because that is how well o people might have been. The argument of eciency or perfectionism against equality can be detected here. It can be seen in a utilitarian perspective but it can also be seen in the Marxist view on the development of production forces which will lead to fundamental real equality in which there is enough for all. If B is taken as the reference and if the complaints are the basis for the wrongness of inequality, then an improvement of the best-o in situation A would be an improvement according to equality. But this is far away from our judgements regarding inequality. Of course it could be an improvement considered overall, but it can hardly be seen as an improvement with respect to equality.
6.6. THE REFERENCE
C
203
D
E
Figure 6.3: Illustration of the transformation of judgements regarding the ideal of equality to those regarding the ideal of perfection Consider for example the situations in gure 6.3. If the view just stated on the reference was right then the situations in order to illustrate the levelling down objection, C and D, could even turn out to lead to the judgement that situation D was better even with respect to equality than C . If E was the reference for C and also for D then D would be better than C with respect to equality, and there would be no levelling down objection. But reading equality in this way is too much deviating from our judgements about equality. It shows equality to transform into some other ideal and indicates that it would become super uous. Because the purpose of this study is the development of an ideal of equality, which is one ideal among other ideals, it should not be super uous and should not be just another name of the ideal of eciency or economic growth. It is not plausible to accept this way of determining the reference. There is, as is already indicated, another way. Instead of looking for all possible alternatives to the situations to be relevant in order to determine the reference, it should be restricted to those in which the distribuenda are redistributed but not augmented. The distribuenda are the restrictions of the total amount of the equalisandum. The complaints referring to the possibility of being better o if the distribuenda were distributed otherwise, namely such that all were equally well o with respect to the equalisandum, form the basis for the moral evaluation inequality. In di erent problems and situations there will be di erent equalisanda and consequently di erent references, one for each situation. For example,
204 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY in case the distribution concerns rights to walk in a park belonging to one family and in which access is allowed only to the members of that family, the reference would be the situation in which the park is open to the public, under the assumption that enjoying wandering round the park would not be destroyed by this public access to the park. Here the complaint of the worse-o concerns a complaint vis-a-vis the best-o , those who have already access to the park. But in for example a situation in which some medicaments have to be distributed, the reference situation will be the one in which all have that part of the medicaments such that their medical health status is equal. Here, in contrast with the previous one, the besto can end up worse o in the reference compared to the situation to be evaluated, because they turn in some of their medicaments. The examples show that in di erent situations there will be di erent references. So far it was assumed that distributing the distribuenda lead to a redistribution of the equalisandum which follows the amount of the distribuenda. The more of the distribuenda, the more of the equalisanda. But what if this relation is disturbed? How to evaluate situations if the distribuenda cannot be redistributed in such a way that the equalisandum, equally available for all, is following positively the total amount of distribuenda? Suppose the distribuenda in a particular situation are allocated in a way represented by the matrix A, then the ideal reference situation can be represented by the matrix A and A is representing the allocation P such that P maximum of the equalisandum xref (A ) is allocated equally and A A . Now it is not excluded that xref () is dependent on the allocation of the total amount of distribuenda in a way such that its value is not higher the more there is of the distribuenda. There will arise problems with our judgements regarding inequality. We can distinguish three kinds of situations in which problems will arise which have to be discussed: 0
0
0
0
1. the redistribution of distribuenda will lead to a higher possible equalisandum available for all collectively such that it is even better for the already better-o , 2. the redistribution of the distribuenda will lead to a diminished equalisandum that is even worse for the worse-o , 3. the collective available amount of equalisandum is not dependent on the total amount of distribuenda in case of non-transferable goods. These will be discussed subsequently. The rst kind of situations poses a problem because in these it could be the case that all have a complaint with regard to the reference, implying that improving the best-o would mean an improvement concerning equality, which would mean a judgement
6.6. THE REFERENCE
205
which is far away from our ordinary ideas on equality in order to count as a judgement regarding the ideal of equality. As the discussion on the ideas of the non-envy analysis has shown, a redistribution could lead to an improvement for all.21 For example a redistribution of hearing aids and glasses under the theatre public could mean an improvement for all, for example if the deaf had the glasses and the myopics had the hearing aids. A redistribution such that all improved could lead to a situation in which all could enjoy the theatre equally. So, all would have a complaint vis-a-vis this reference because they might have been better o by the redistribution of the distribuenda. Improving the best-o would mean an improvement with respect to equality because the complaints would diminish. But this is contradicting our ideas about equality. The possibility of an improvement for all by a redistribution of distribuenda, seems to be a threat to the development of the proposal of equality that is based on individual complaints vis-a-vis a reference. But it turns out to be not harmful. In the example mentioned above one should distinguish between a part of the improvement in the reference due to the exchange of distribuenda, i.e. glasses and hearing aids and pure redistribution. In considering the situation in this way one should not look at hearing aids and glasses as such but one should look at them as bearers of an exchange value. This exchange value has to be seen as the distribuendum, an exchange between glasses and hearing aids in case one pair of glasses is worth one hearing aid is not to be counted as a redistribution. If all have either a hearing aid or a pair of glasses, all were already equally well o with respect to enjoyment of the theatre. The improvement is not due to the redistribution of the distribuendum but to a mutual advantageous exchange. The mutual advantageous exchanges, which is the productive aspect of the redistribution, should be distinguished from the redistributive aspect interpreted in a narrower sense. Evaluations of inequality come in if the redistribution is no longer mutual advantageous with respect to the equalisandum.22 If See chapter 3 p. 93. This productive part can be taken as the mutual pro t regarding the equalisandum. It can be taken to be the improvement that is equally for all that is at least the result of an exchange; it equals the minimum of the improvements belonging to persons i that are worst o in the allocation in the set of possible allocations in a contract set resulting from exchange. In order not to live on the credit of a not actual improvement regarding equality, the inequality of the distribution can be taken as the worst inequality of those of the allocations in which person i is worst-o . In a formula: Let CS be the contract set, the possible allocations if there were a free exchange without transaction costs. Let A be the actual allocation. The inequality of A, I (A) = I (B ) in which I () is an inequality index and B is the allocation such that 8B 2 MPCS : I (B ) I (B ) and MPCS is 21 22
206 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY there were a di erence in possession of the distribuendum then of course there would be inequality with respect to the enjoyment or possibility of enjoyment and the reference would not be such that all improved by a redistribution of the distribuenda. Of course one should look with suspicious eyes to the use of the phrase `possibility of enjoyment' instead of enjoyment simpliciter. But it is not a real problem here because with `possible' is meant here collectively possible and not merely distributively possible for which suspicion is a right attitude. If one does not consider the exchange value as distribuendum, but one considers the commodities themselves as distribuenda, one should look at the redistributions of the hearing aids and glasses separately. The reference with respect to the hearing aids will not be such that all improve, and the reference based on the redistribution of the glasses will not be such that all improve by the redistribution of the glasses. This will no be without serious problems because the complaints will be dependent on the redistribution of the other distribuendum.23 The former method, taking into account the exchange value and consider this to be the distribuendum is more appropriate here.24 So, the rst kind of situations can be analysed in such a way that the evaluations regarding the ideal equality do not imply that an improvement of the best-o has to be considered as an improvement regarding equality. The second kind of situations, in which the equalisandum would be less in a more equal distribution than in case of the actual situation, could the set of allocations in which the equal production part of the redistribution is minimal:
fB 2 CS j8B 0 2 CS; 9i; 8j : (xi (B ) , xi (A)) (xj (B 0) , xj (A))g in which xi (A) is the equalisandum belonging to person i in allocation A. Of course, this will only make sense
after the measure of inequality has been developed in chapter 8. It seems that by mutual advantageous exchanges budgetsets are introduced as the distribuenda, and that this will undermine the view on the analysis of for example the example brought forward by Cohen (chapter 4 p. 144) of the well to do with pain after moving his arms. However this is not implied. The particularism will prevent this because the particular aspects of markets and the actual functioning of these have to be reckoned with. Here, it is merely stated that one can analyse situations in which all improve by a redistribution in a way such that judgments concerning equality can be made. 23 See also the discussion on speci city in chapter 7 p. 239 and chapter 8 p. 261. 24 A similar analysis is possible in cases with one distribuendum in which a redistribution will lead to an improvement for all because of external e ects. We met this in chapter 3 in the discussion on the ideas of Hochman and Rodgers. As pointed out, external e ects have to be explained and interpreted, for example by altruistic feelings. This means that there is actually another equalisandum, for example instead of purchasing power enjoyment of co-operation in recreational activities. But now it will be appear that there is just more than one distribuendum, one that can be held responsible for the external e ect, for example the power to withhold co-operation. So, the analysis in the main text can be followed.
6.6. THE REFERENCE
207
arise if the production will be less because those being well o will produce less due to a lack of incentives. This possibility is extensively discussed in the literature related to the di erence principle of Rawls.25 But it is not a threat to the proposal here. Here I am concerned with complaints vis-a-vis the ideal reference, how well o people might have been relative to that ideal. It remains a possibility that all produce as much as they did under inequality. Consequently, the complaints with respect to the reference belong to those relevant to inequality. How well o the worse-o are is directly related to how well o the better-o are. What matters for judgements on equality is that some are worse o than they might have been if there was another distribution. One could rebut that not all situations which might have been the case, such as for example mentioned above those situations after a further economic development, are relevant for the ideal. The point is that merely distribution e ects should count in so far they directly a ect the distribution of the equalisandum. In the case of economic development the result is more or less external to the distribution, just as in case people are producing less as reaction to more equality. A lesser production is not inherent to the redistribution of the distribuenda and to the determination of the allocation of the equalisandum and can be disregarded for the determination of the reference. Another way these second kind of situations, in which all end up worse if the equalisandum is equally divided, could occur is, if inequality is an element of the value of equalisandum, for example, if something receives its value because of its exclusivity. This will be the case in a context in which the good or the possession of the good is seen as way of expressing oneself and showing one's individuality. Once this is understood, the equalisandum has to be changed to something like `possibility of expressing one's individuality', which is collectively possible although it will not be possible by way of the same good. So, also these second kind of situations causing problems for the idea of inequality being based on complaints with respect to the reference can be answered by turning to the proper equalisandum. Let me turn to the third kind of situations that seem to pose problems. The third kind of situations concerns those in which the total amount of collective available equalisandum is not dependent on the total amount of distribuenda, i.e. the distribuenda are not transferable. The situations A and B in gure 6.4, illustrating the levelling down objection, considered so far, concerned transferable distribuenda. Consequently, a redistribution being a transfer could lead to a situation in which all are equally well o with respect to the equalisandum. The reason why B was worse with 25
See chapter 4 p. 118 and [Cohen, 1993].
208 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY
A
B
Figure 6.4: Illustration of the levelling down objection respect to inequality than A was that in A no one could complain that he might have been better o in the referential equal situation while in B the worse-o could complain about that. This kind of reasoning is not possible if one considers non-transferable goods. An example of a non-transferable distribuendum shows up in Par t's example of the levelling down objection, taking the light out of the eyes of those who can see if others are blind. A normal function of the visual system cannot be transferred from one person to another. Par t argues against equality as an ideal by stating that destroying a normal functioning visual system cannot be seen as an improvement, in no way [Par t, 1989]. How is this to be seen in the light of the above developed idea of equality? Will there be an improvement in one aspect in destroying the visual system or is there none with respect to which it will mean an improvement. The answer is: `It depends.' If vision is seen as the equalisandum then the ideal reference in both situations, the one in which some can see and the one in which no one can see are equivalent with respect to equality. In the former situation there will be no complaint of those who cannot see, there is no one worse o than the reference. So with respect to equality there is no improvement by destroying the visual system of those who can see. This means that the levelling down objection is no longer an objection to the ideal of equality, because the evaluations of situations A and B are similar, A is in no way better than B . They are from an egalitarian point of view equivalent.
6.6. THE REFERENCE
209
The view that A and B are equivalent from an egalitarian point of view, so might be argued, undermines the whole idea of equality because it leaves a meaningless and impotent ideal of equality. In most situations one could point to worse-o who would not improve by a redistribution of distribuenda. So the reference would be such that all are equally well o as the worse-o with respect to the equalisandum. Consequently, there would be no complaints with respect to the reference, so there would be no relevant inequality. This way of looking at the levelling down objection seems to be defeating the ideal of equality as a relevant ideal. But as indicated above in the answer how to look at the levelling down objection `It depends.' leaves open an alternative. In Par t's example the visual system was seen as a distribuendum which was non-transferable and the possibility of vision was the equalisandum. But there is another way of looking at this situation. One can take something other than vision as an equalisandum, for example capability of earning one's own income, or enjoying the theatre or museum. By acknowledging this, one can turn to the idea of compensation for the loss of vision. The loss of a normal functioning visual system can be compensated for. Consequently, a change of the equalisandum by accepting compensation also brings with it a change of distribuenda such that a transfer of distribuenda becomes possible. For example if not just vision but earning capability is seen as the equalisandum then destroying the visual system of others could mean an improvement with respect to equality, it would restore for example a fair competition. To be clear it is not said that it would be a redistribution which is morally acceptable. Other reasons and ideals, for example perfectionism, can override the ideal of equality in which all are equally blind. But it could be argued that being equally blind is good in just this egalitarian aspect. So the concept of compensation brings with it a change in equalisandum and distribuenda in such a way that transfers become possible and the levelling down objection is answered as was shown above in case there are transferable goods, namely there is something regrettable in the situation of some being blind, namely that the blind might have been better o , not with respect to vision, but with respect to the equalisandum that will turn up if one looks for compensation of lack of vision. In chapter 9 this issue of non-transferable distribuenda will be met again in an even more distressing way in the context of health and health care and will be discussed again.26 Summarising, the situations posing problems concerning the individualistic interpretation of equality as it was developed here, based on complaints with respect to a reference, can be dealt with. It was shown that by taking 26
See chapter 9 p. 292.
210 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY the right distribuendum and equalisandum the reference can be determined, the problematic judgements disappeared. The ideal of equality can meet the levelling down objection. The situation B is worse than A in one aspect, because in B there are some complaints with regard to the reference in which the distribuenda are redistributed such that all simultaneously receive equally as much of the equalisandum as possible. If the distribuenda are not transferable then the levelling down does not pose a problem at all, because then B is not seen to be worse than A.
6.7 The other views revisited Before I proceed with the measures of inequality I have to answer the question, how this idea of equality is related to the other views, such as `the minimumview' of Lucas, `the view of enough' of Frankfurt and `the priority view' of Par t. Lucas' reason for holding that situations exhibiting inequality are wrong is that some have less than a certain minimum. If all are above this minimum an appeal to an ideal of equality is no longer a very strong one [Lucas, 1965] [Lucas, 1977]. This evaluation can be accounted for in the view on the ideal of equality as it is developed in this chapter. We can accept some sort of hierarchy of goods with respect to which there should be equality. Once equality is reached on the most important goods, some less important goods enter the conscience concerning distributions. Lucas does not discuss this possibility and rejects the political ideal of social equality too easily. According to Frankfurt's view the situations in which there is inequality are wrong because some have less than enough. This view can be accounted for too. Frankfurt explains that persons have enough if they do not actively strive for more of a good, although they do not deny to prefer more of this good to less [Frankfurt, 1987]. But one of the reasons why persons do not strive actively for more is precisely that they realise they should not have more than is available for all. The reasons for thinking otherwise are precisely anti-egalitarian reasons. Par t nally states that in general, situations which exhibit inequality are bad because the worse-o are denied priority. He argues the priority view to be plausible because it is reprehensible that people are worse o than they might have been [Par t, 1989, p. II-6]. In this explanation of the priority view it is left undetermined how well o people might have been. The reference is undetermined. The reference of equality proposed in this chapter is an answer to this lack of determination; it determines how well o all people might have been. The ideal of equality as it is developed here
6.8. SUMMARY
211
is in accordance with Par t's reason for the priority view. So the three competitors, `the minimum view' `the view of enough' and `the priority view', can be explained to be plausible in the view on equality developed in this chapter. This makes it a promising base for the development of a measure of inequality to which we will turn in the next chapters.
6.8 Summary In this chapter the meaning of the ideal of equality was determined. It was inspired by the arguments against equality. First, it was shown that a large category of arguments against equality could be rebutted because of the wrong presupposition that equality is the only ideal or that there is just one equalisandum. Moral value pluralism was the answer to the arguments that equality will lead to bad consequences, or that equality with respect to one good is not valuable because some other good matters more and is claimed to be the proper equalisandum. Within moral pluralism it can still be held that equality is an ideal among others although not always the overriding one. The most intriguing arguments were the ones stating that equality runs partially parallel with another ideal but is considered to be wrong and super uous. These arguments did show a common objection against equality namely the so-called levelling down objection. The levelling down objection holding that well-o becoming less wello without an improvement of the worse-o will not be better with respect to any ideal, was discussed. It was argued that it is an expression of socalled person a ectingness, meaning that a situation cannot be worse or better if it is not worse or better for someone. The person a ectingness was amended in line with realistic individualism in order to account for non-identity cases in which person a ectingness as it was originally stated did function as interpersonal incomparability of what goods do to people. Finally, it was argued that the levelling down objection could be answered by an individualistic interpretation of equality in which complaints of those who are worse o than they might have been in an ideal reference situation have a central role. This ideal reference is determined by a redistribution of distribuenda, such that all are collectively equally well o with respect to the equalisandum. The problems with determining the reference such as all having a complaint with regard to this reference, or all becoming worse o in this reference or in case of non-transferable goods, could be solved by turning to the proper distribuenda, and equalisanda. This individualistic interpretation of equality within moral realism is the starting point for a simple measure of inequality to be developed in the next chapters.
212 CHAPTER 6. THE MEANING OF THE IDEAL OF EQUALITY
Chapter 7
Measures of inequality 7.1 Introduction In the previous chapter I argued for an individualistic interpretation of the ideal of equality in which the complaints of those who are worse o than all could have been simultaneously, have a central role. This interpretation will of course have consequences for a measure of inequality which is meant to be an index for the moral seriousness of inequality. These consequences will be examined in this chapter and in the next chapter. These chapters make use of some mathematical reasoning. Although the details of the reasoning can be skipped by the reader, it is nevertheless explained rather extensively in order to make the reasoning accessible for those who are not mathematical experts. Several measures have been suggested. They are formulated all under the supposition of one xed equalisandum for which there is a measure.1 In this chapter, I discuss some of them, because the critical remarks that have been stated let appear the properties that an adequate measure of inequality should satisfy. An example is illustrative. The badness of inequality could be taken to be represented by the di erence between the best-o and the worst-o i.e. xmax , xmin , the range (xi is the amount of x person i has, max indicates the best-o and min indicates the worst-o ). This measure, although attractive because of its simplicity, is inadequate. It leaves out of consideration all persons which are in between the worst-o and best-o . A distribution represented by A in gure 7.1 will be considered to be as bad regarding inequality as distribution B. 1 In the next chapter I will have some remarks on the measure of the equalisanda.
213
CHAPTER 7. MEASURES OF INEQUALITY
214
A
B
Figure 7.1: Illustration of the range not being a proper measure
x
A
B
Figure 7.2: Illustration of the mean deviation not being a proper measure This example shows that a measure should take into account not merely how well o the best-o and the worst-o are but also those in between. A measure which takes into account also how well o those others are, is for example, the mean deviation, the sum P of nthe di erences of how well peoplePfare with respect to the mean i.e. i=1 jx , xi j, (x is the mean i.e. n1 ni=1 xi , n is the number of persons). But this measure is not adequate either, because now a distribution represented by A in gure 7.2 will be seen to be as unequal as distribution B. This critical remark exhibits the idea that transfers from a rich person to a poor person should make a di erence to the inequality.2 2 Whether all transfers should do, is however an open question. I return to it below
7.2. LORENZ DOMINANCE
215
Like the examples of measures mentioned above some other measures are discussed in this chapter in order to arrive at a list of properties a measure should satisfy. I start with those measures that follow the ordering of Lorenz dominance, meaning roughly that the seriousness of the inequality of a distribution ~x 0 is greater than that of another ~x, ~x I ~x 0 , if the Lorenz curve of ~x is above the Lorenz curve of ~x 0 , i.e. ~x LD ~x 0 . The Lorenz curve, which is explained in section 7.2, is a widely used way of representing the inequality of a distribution. In that section, some further properties of Lorenz curves and Lorenz dominance are discussed. The relation of Lorenz dominance has a serious shortcoming: it is incomplete in its ordering of distributions. It does not indicate how to evaluate the inequality if the Lorenz curves do intersect each other. One response is to look at extensions. These are examined in section 7.3.1. It appears that the measures considered there are not consistent with our moral ideas on equality. In that section some important properties arise out of the discussions. The lack of a satisfactory extension suggests to attend to another response. This other response is accepting incompleteness and look for explanations of incompleteness. These explanations can be seen as reasons against the possibility of a complete ordering. The explanations are discussed in section 7.3.2. It is argued that they are not acceptable or are not leading to judgements parallel to Lorenz dominance. Therefore, the characteristic properties of the Lorenz dominance have to be amended. This amendment is the subject of section 7.4. Finally, in section 7.5 I end up with some properties which an adequate measure should satisfy. In the next chapter a simple measure is presented which satis es these properties, it is the nal argument against the view that there is no complete ordering regarding inequality in distribution problems.
7.2 Lorenz dominance In newspapers and economic literature, it is usual to represent the inequality of distributions by citing how many of the persons possess a particular part of the wealth, or income [Samuelson, 1984, p. 80]. For example, the income distribution in the Netherlands represented by a diagram as in gure 7.4 in this chapter. In the example here, it is clear that a transfer should make a di erence. But is it also in case the transfer takes place in the upper part of the gure and a less rich person receives some of the good from an even more rich person? Does that make any di erence in the seriousness of inequality? I am not sure about that. Sen choose his examples carefully in order to be convincing in [Sen, 1973, p. 26].
CHAPTER 7. MEASURES OF INEQUALITY
216 20% 15% 10% 5% 0%
?
0%-10%
?
90%-100%
Figure 7.3: Income distribution in the Netherlands 1995 was presented in an article in the Volkskrant.3 On the horizontal axis are put the groups ordered from poor to rich, 1:::i:::n, on the vertical axis nxxi . The Lorenz curve is the result of the cumulation these fractions. The income of the rst percentile will be similar to those in the former curve. The second percentile is added to the rst and the total amount they possess is represented on the vertical axis. On the horizontal axis we see the percentage of the total population ordered from poor P to rich represented by the percentiles and on the vertical axis we see i1 nxxi for the group up to the ith percentile. Connecting the points will result in the Lorenz curve as in gure 7.4 The diagonal line is representing an equal distribution in which 10% of the poorest people possess 10% of the total amount of goods, 30% possess 30%, 50%, possess 50% etc. A Lorenz curve of ~x 0 below another curve of ~x means that ~x 0 is worse regarding inequality than ~x, it is Lorenz dominated, ~x LD ~x 0 . Apart from the fact it is a common practice to represent the inequality of distribution by a Lorenz curve, there are some other reasons for Lorenz dominance being an attractive ordering regarding inequality. Lorenz dominance is implied by a social welfare function which is Schur concave and monotone increasing in its arguments. The idea is that the judgements `all things considered' can be represented by a welfare function W(~x) of the distribution ~x, the higher the value of this function the better. Schur concavity does mean that W(B~x) W (~x) in which B is a bistochastic matrix 3 Volkskrant 9 december 1995 an article concerning the inegalitarian e ects of the reformation of welfare state by the Dutch governing coalition in 1995.
7.2. LORENZ DOMINANCE 100%
217
, , , , , , , , , , , ,, , ,, , ,"", !,! ,
q
q
q
50%
q
q
q
q
q
q
0%
q
1 10
5 10
10 10
Figure 7.4: Lorenz curve of income distribution in the Netherlands 1995 in which all the elements bji are : 0 bji 1 and the sum of the elements of a row is 1 and the sum of the elements of a column is 1. Such a matrix can be considered to represent transfers from rich to poor. Monotonicity in its arguments does mean that W(~x) is non-decreasing if xi increases, representing the Pareto principle or the principle of personal good or even better person a ectingness.4 It has been proved that: (x = x 0 & ~x LD ~x 0 ) =) W(~x) W(~x 0 ) [Dasgupta & Sen, & Starlett, 1973] So Schur concave monotone welfare functions W are consistent with Lorenz dominance under the condition of an equal total amount of what is to be allocated.5 Because those indices are representing the judgements on how well a society is `all things considered', Lorenz domination is seen to be an attractive ordering of inequality. It is common to many measures or indices of inequality and social welfare [Kolm, 1968, p. 149].
P
4 See chapter 6 p. 194. 5 An example of such a welfare function is utilitarianism W (~x) = n w(xi ) in which i=1 w is increasing (the more of xi the higher w (xi )) and concave, meaning that the more of xi the less it will add to w , i.e. the law of diminishing returns. Equalising will result in
an increase in , the popular argument for equality within utilitarianism. See for some critical remarks p. 226 of this chapter. W
218
CHAPTER 7. MEASURES OF INEQUALITY
Beside the reason mentioned, which relies heavily concave social welfare functions, there is another reason for the attractiveness of Lorenz dominance as an ordering indicating the moral badness of inequality. It can be shown that Lorenz domination will result if symmetrical distributions, distributions in which persons changed place with respect to the distribution, in other words, permutations of the distribution, form an indi erence class in combination with an amended version of the person a ectingness of chapter 6 of which the best-o are excluded. This can be shown by following the line of argument of Kolm in [Kolm, 1972, p. 101 . ]. The echo of Suppes' grading principle and the non-envy condition can be heard. Furthermore the idea that the best-o are excluded from those to be reckoned with, does re ect the idea of equality as it was developed in the previous chapter, together with the idea that if the best o are equally well o the second best o should not count in determining how good a distribution is, and so on. Let me follow Kolm's argument. If a distribution ~x has a similar total amount of goods as another distribution ~y, and if the former is better than the latter, , then the Lorenz-curve of the former is everywhere above the latter. The explanation is quite simple and instructive. = (x1 ; x2:::; xn) and ~y = (y1 ; y2 :::; yn) such that Pni=1Takexi =distributions Pni=1 yi.6 ~xTake a permutation of ~x say ~x 0 such that for all 0 0 i : xi xi+1, similarly take for ~y; ~y 0 such that for all i : yi0 yi0+1 . Now if for all i < m : x0i yi0 (m being the maximum number for which xm 6= ym ), then distribution ~x 0 is better or equal to distribution ~y 0 , ~x 0 ~y 0, because of the amended version of person a ectingness the best-o , i = n are excluded from consideration, or if these are equally well o in all distributions the second-best-o are excluded, and so on. Because of symmetry we have ~x ' ~x 0 and ~y 0 ' ~y, so ~x ' ~x 0 ~y 0 ' ~y . But if ~x 0 ~y 0 , then the amended version of person a ectingness implies that the Lorenz curve of ~x 0 , or what comes to the same, the Lorenz curve of ~x is nowhere below the Lorenz curve of ~y because:
8i < m : x0i yi0 =) 8i :
Xi x0 Xi y0 ; k=1
k
k=1
k
in which m is the maximal number such that x0i 6= yi0 . So far Kolm's argument. It is also valid that if the Lorenz curve of ~x is nowhere under the Lorenz curve of ~y then ~x ~y. Consider x0m and ym0 in which m is the maximal 6 It would be according the common rules to state = ( 1 2 n )T ()T meaning the transpose of (). Because of typographical convenience and because it does not lead to great misunderstandings I sometimes use the vertical representation of vectors and sometimes the horizontal. x ~
x ; x ; ::x
;
7.2. LORENZ DOMINANCE
219
number such that x0i 6= yi0 . Make a transfer from ym0 to the m , 1 person in such a way that in the resulting distribution ~y 00, x0m = ym00 , without disturbing the ordering of the persons.PiThis is possible because the Lorenz 0 Pi y0 . It means that curve of x ~ is not under y ~ i.e. 8 i : x k=1 k 00 k=10 k Pn x0i Pn yi0 so ym0 x0m. Furthermore ~y ~y , because ym00 > ym0 i=m i=m and i m can be excluded from consideration. But also ~x 0 LD ~y 00 . Repeat this procedure p times until m = 2. Now it can be concluded that if m = 2 the worst-o and the second-worst-o person can di er and all others are equally well o . The second-worst-o can be deleted from considerations so, because the total amount of goods is not changed, this means that x1 y1 p . Consequently if ~x LD ~y then ~x ' ~x 0 ~y p ~y 0 ' ~y . So, symmetry and the amended version of person a ectingness excluding the best-o from consideration and if they are equally well o excluding the second-best-o etc., is equivalent to the ordering of Lorenz dominance, provided the total amount of the goods to be distributed is constant. Actually as shown by [Foster, 1985, p. 51], the class of indices I which are consistent with Lorenz dominance meaning that the index does not contradict the ordering of the Lozenz dominance, in formula: x~ LD ~y =) ~x I ~y can be characterised by four properties: 1. Pigou Dalton principle, meaning that a transfer from a poor person to a rich person will result in an increase of inequality. 2. Symmetry, or anonymity principle, based on the idea of impartiality, meaning that taking a permutation of the distribution, will make no di erence for the value of the index of inequality. 3. Population principle, meaning that adding a population with the same distribution will result in a value of the index of inequality that is similar to the values of the index of the populations taken apart. 4. Homogeneity, meaning that the index of inequality is invariant and will not change if the amount of goods received by all is multiplied by the same factor.7 All measures that satisfy these four properties are consistent with Lorenz dominance and all measures consistent with Lorenz dominance have these properties. The latter is shown in a clear way by Foster. Lorenz dominance follows these properties. Take ~x LD ~y, then from the construction of 7 I return to this property in the next chapter. This property can be argued for because there should be no di erence between measuring income in dollars or in dollarcents.
220
CHAPTER 7. MEASURES OF INEQUALITY
the Lorenz curve it is clear that symmetry is satis ed. Changing the persons that receive a particular amount will not change the Lorenz curve, the Lorenz curve is determined by the ordering from poor to rich. The Lorenz curve of ~x and of ~x will be P similar because is seen in both the numerator and the denominator of n:xix . Furthermore, the Lorenz curve of ~x is similar to ~y if ~y is consisting of several replications of ~x because the percentiles do not change, the fraction of the population possessing a particular fraction of the total amount does not change. Finally, the Pigou Dalton principle can be shown by turning to the contraposition. If distribution ~x is the result of a transfer of a rich person j to a poor person i starting ~y , then for the m < i and m > j nothing changes Pfrommk=1,1a xdistribution k = Pm,1 yk . For i m j we can note that the percentile to k=1 x x which i belongs will receive more because some persons P of the Pi + 1-the percentile will now belong to the i-th percentile so mk=1 xxk mk=1 yxk , resulting in ~x LD ~y . Now the other direction: if an index I satis es the four properties it is consistent with Lorenz domination. It can be shown as follows [Foster, 1985, p. 51]. Suppose an index I satis es these properties and let ~x and ~y be two distributions with dimension n and m representing the numbers of persons. De ne ~x 0 as m replications of x1 ~x and ~y 0 as n replications of y1 ~y. Now because of the argument in the former paragraph ~x 'LD ~x 0 and also ~y 'LD ~y 0 . If ~x 'LD ~y then ~x 0 = ~y 0 because ~x 0 and ~y 0 have the same total amount of goods and the same number of persons. Consequently x~0 'I ~y 0 and so ~x 'I ~x 0 'I ~y 0 'I ~y because of symmetry, homogeneity and the population principle of I accounting for the other outermost equivalencies. Suppose next that ~x LD ~y, then because of the previous paragraph ~x 0 LD ~y 0. So, also because of the previous paragraph ~x 0 can be obtained from ~y 0 by transfers from rich persons to poor persons. Because I satis es the Pigou Dalton principle ~x 0 I ~y 0 and because of homogeneity and the population principle and symmetry we have ~x 'I ~x 0 I ~y 0 'I ~y. So, there are several reasons for using Lorenz dominance as basis for judgements on inequality. But there is a serious problem, namely what to do if Lorenz curves do intersect? It is just a partial ordering.
7.3 Incompleteness As mentioned, the Lorenz dominance as an indicator of inequality does not lead to a complete ordering. There are two answers to this incompleteness. One is developing a complete extension of the ordering of the Lorenz dominance, the other is accepting incompleteness as an inherent feature of judgements of inequality. These two answers are discussed respectively.
7.3. INCOMPLETENESS
221
, , , , , B , ,A , , , , , , , C , , , , 1
n
1
n
1
n
x3: n:1x x2: n:1x x1: n:1x
1
n
Figure 7.5: Illustration of the GINI coecient
7.3.1 Extensions of Lorenz dominance GINI coecient
One natural way of extending the ordering of Lorenz dominance is by taking the areas under Lorenz curve as a representation of inequality. One commonly used index is known as the GINI coecient. It is the ratio of the area enclosed by the Lorenz curve and the diagonal, A, and the total areaA , or what comes to the same area under or above the diagonal B, G = areaB one minus twice the area C under the Lorenz curve, G = 1 , 2C. The greater the area A, the greater the inequality. The GINI coecient can be expressed as: G(~x) = 1 + n1 , n22 x (nx1 + (n , 1)x2 + + 2xn,1 + xn) This can be made clear by the following reasoning. The area under the Lorenz curve, C() is: x1 + 1 x1 + 1 1 x2 + = C(~x) = 12 n1 n x n nx 2 n nx 1 1 1 (x + x ) + 1 1 ((n , 1)x + (n , 2)x + ) = 1 2 2 n nx 1 2 n nx 11 1 1 1 2 n nx nx + n nx ((n , 1)x1 + (n , 2)x2 ) =
222
Finally:
CHAPTER 7. MEASURES OF INEQUALITY 1 1 1 1 1 2n + n nx (nx1 + (n , 1)x2 + ) , n nx (x1 + x2 + ) = 1 1 1 2n + n2 x (nx1 + (n , 1)x2 + ) , n = 1 + 1 (nx + (n , 1)x + ) , 2n 2 n2x 1 G(~x) = 1 , 2C(~x) = 1 + n1 , n22 x (nx1 + (n , 1)x2 + )
There is also another formulation of the GINI coecient which is cited more often: Xn Xn jx , x j G(~x) = 2n12x i j i=1 j =1
This formula indicates that the di erence between each pair is the central idea of a measure of inequality.8 Although the GINI coecient is complete and of course consistent with Lorenz dominance, it is not according to our moral intuitions. For example a transfer from a rich person to a poor person will lower the GINI coecient, but the amount of how much is dependent on how much groups there are between these rich and poor persons and not merely on the di erence of how well o they are. This can be seen from the rst formulation of the GINI coecient. It could happen that a transfer of a rich person to a poor person with a small di erence but with many groups between them, is lowering the index more than a transfer from a rich person to a poor person where there are a few groups between them but where the di erence is much larger. For example, a transfer T from group 2 to 1 with a di erence of 100 without changing the groups, is lowering the GINI coecient with 2 n2 x (nT , (n , 1)T). A similar transfer T from a group n to group n-100 with a di erence of say, 10 between them, will lower the GINI coecient by n22 x (100T , T). This indicates that the GINI coecient leads to some odd evaluations [Sen, 1973, p. 32]. The argument against the GINI coecient makes clear, that for an evaluation of an index of inequality, the issue of how much a transfer from rich to poor changes the index, is considered to be relevant. To be more precise, one idea about our evaluations of inequality is that a transfer from a rich to a poor person matters more if the transfer is in the lower part of the distribution. This principle is called the principle of diminishing 8 It is often said that it can be easily shown that both formulations are similar, but it is seldom explained. The explanation is in appendix 1 of this chapter.
7.3. INCOMPLETENESS
223
transfers. As was shown above the GINI coecient can be in con ict with this principle. The transfer T in the upper part of the distribution can lower the GINI coecient more than a similar transfer in the lower part. Confronted with these odd evaluations of the GINI coecient due to arbitrariness of some evaluations according to this index, one wonders if there is a systematic way of extending the ordering of the Lorenz dominance. One way for example would be adding more restrictions and wonder whether the class of indices is restricted such that the orderings are complete. Let me turn to that possibility.
Decomposable indices
Decomposability is argued to be an attractive property for a measure because it enables to determine the inequality of the population on the bases of the inequality of parts of this population. If one knows the inequality of the subgroups, Ig , the number of persons, ng , belonging to that group, and how well o these subgroups are, xg , then it is possible to determine the inequality of the whole group. Decomposability means that the inequality of the whole group is the weighted sum of the inequality within the subgroups and the inequality between the subgroups. To be more formal. I(~x; n) =
XG w I(~x ; n ) + I(x ; x x g=1
g
g g
1
2
G ; G)
Shorrocks demonstrated in a good accessible way that the class of indices that are i) symmetric, ii) follow the population principle, iii) are homogeneous and iv) have values of zero if there is equality and v) are greater than zero if there is inequality, vi) are continuous and vii) have continuous rst order and second order derivatives, is:
Xn Ic (~x; n) = n1 c(c 1, 1) (( xxi )c , 1); c 6= 0; 1 i=1
Xn Ic (~x; n) = n1 log xx ; c = 0 i=1
i
Xn Ic (~x; n) = n1 xxi log xxi ; c = 1 i=1
[Shorrocks, 1980].
224
CHAPTER 7. MEASURES OF INEQUALITY
The index with c=2 corresponds to the square of the coecient of variation, the ones with c=1 and c=0 are suggested by Theil as measures of inequality. The coecient of variation is: r P(x , x )2 1 i CV (~x) = x n This coecient is criticised by Sen, because it is in con ict with the principle of diminishing transfers [Sen, 1973, p. 28]. Transfers in the upper part of the distribution count as much as transfers in the lower part. For example, take an in nitesimal transfer from xi to xi,t, then the in nitesimal dCVxi,t xi change in the coecient of variation will be , dCV dxi + dxi,t . Because it is a mean preserving transfer the change in pn1x2 is zero, the dCV dxi is P for all pn1x2 21 pP(1x,x )2 2 i(x , xi )(n , 1). Hence, it is equally for all i [Atkinson, 1972]. Apart from this issue of the principle of diminishing transfers meaning that transfers should have less e ect on the inequality in the upper regions in case c=2, there remains still something arbitrary on these measures.9 For example how to choose between the values of c. Furthermore, it is not clear at all whether decomposability is consistent with the idea of equality which was argued for in the previous chapter. Following the meaning of the ideal of equality as it was developed in the previous chapter namely that inequality is bad because some persons are worse o than all could have been, means that the inequality within the group above the reference does not in uence the inequality of the total population.10 This means that lowering the inequality within a group which is better o than in the ideal reference will not a ect the inequality of the whole group. For example decreasing inequality in the western prosperous world will probably have no consequences for the inequality on a global scale. So an improvement in the former will not have any e ect on the inequality on the global scale. And indeed by considering the problems on a more global scale, these local problems seem to evaporate. It is not said that inequality in local application is of no importance at all for the inequality. Political decisions leading to a decrease in inequality
9 The argument of Sen that the coecient is highly suspicious because only di erences with respect tot the mean are reckoned with, can be rebutted by pointing to the lemma 5.12 stated by Kakwani in [Kakwani, 1980, p. 86]. It reads: If, in any pairwise comparison, a person with lower income su ers some depression proportional to the square of the di erence in incomes, the average of all such depressions in all possible pairwise comparisons leads to the coecient of variation. 10 See chapter 6 p. 197.
7.3. INCOMPLETENESS
225
on a local scale is not necessarily constitutive of a decrease in inequality on a global scale, but it could in the end result in an improvement by starting a process of actions leading to less inequality for example, by promoting and extending sympathy. This is a process in time and concerns the question: how to arrive at a more egalitarian world? This issue of how to promote equality on a global scale is not the subject of this study. Here, I am concerned with the content of the ordering of inequality as part of an analysis of equality as an ideal. The ordering at a particular moment is at stake and its dependence on the orderings in and between groups. And within this analysis of properties of indices, a reduction of inequality within the group above the ideal reference has no consequences for the moral badness of inequality. An intuition which is plausible and can be explained by the ideal of inequality developed in the previous chapter. Decomposability is not a property an index of inequality should have. There is still another method for arriving at extensions of Lorenz dominance circumventing the arbitrariness of measures considered so far, namely by starting with the in uence of the distribution on the social evaluation `all things considered'. Measures which start from these social evaluations `all things considered' do not su er from the arbitrariness that the above extensions su ered from. Let me look at these.
Kolm-Atkinson measures
The former extensions of the ordering of the Lorenz dominance su ered from arbitrariness; there is no connection between the measure and the reason why inequality is wrong. The so-called normative measures start with the in uence of distributions on the evaluations of those distributions `all things considered'. The moral badness of the inequality is measured directly by the loss of welfare or what represents this welfare, for example money as is suggested by Kolm [Kolm, 1968]. Dalton takes the loss of welfare because of inequality directly as the measure of inequality. His index is corresponding to ID (~a) = W(a~u) , W(~a) in which ~u = (u1; u2; : : :) = (1; 1; : : :).11 This index represents how serious inequality is in terms of social welfare. Relevant indices should be invariant 11 Dalton's measure is mostly presented in a relativised form
1,
For
W
the utilitarian interpretation
P
() ( ) ( i ) is commonly used. W ~ a
W a~ u
w a
226
CHAPTER 7. MEASURES OF INEQUALITY
under certain transformations of W and not be in uenced by idiosyncratic properties of W. Usually such a W is only determined up to certain transformations. A utilitarian welfare function is determined up to linear transformations meaning that W and W 0 are equivalent if W 0 = 1 W + 2 (1 and 2 are numbers and 1 ; 2 > 0). For the essential properties it should not make any di erence whether one should take W or W 0 . However, Dalton's measure is dependent on these particular idiosyncratic properties [Atkinson, 1972]. A possibility which does not su er from the dependency on irrelevant characteristics is measuring the inequality in terms of the resources which are spoiled because of an unequal distribution. The measure is in terms of resources instead of in terms of welfare. The leading concept introduced by Kolm, which is used later by Atkinson, is the equal equivalent distribution, ~ae [Kolm, 1968] [Atkinson, 1972]. This equal equivalent distribution is the distribution in which all receive the same amount of the resources ae such that W(ae:~u) = W (~ae). The inequality is measured by a , ae , or by its relativised form 1 , aa . If the equal equivalent distribution is a:~u then the index is 0. The welfare function chosen by Atkinson is of a utilitarian form in which all have the same utility function, which is concave, meaning it satis es the law of diminishing returns. Although there is a diculty of all having the same utility function there is even a more serious problem pointed to by Hansson [Hansson, 1977]. The measure would mean for example that if the utilitarian function becomes more concave then the inequality would in fact be less, but the e ect on the inequality index is di erent. It is illustrated by gure 7.6. Suppose we have a utility function w and a utility function w0 which is in the lower regions equal to w but in the higher ones below w. Consider two distributions xp and ar in the lower and higher regions respectively, then in case of w the di erence between w(ar ) and w(ap ) is greater than the di erence between w0 (ar ) and w0(ap ), which is representing the di erence between what goods do to people. But 21 (w0(ar ) + w0(ap )) is lower than 21 (w(ar ) + w(ap )), so a0e ae . Consequently a , a0e a , ae meaning the index of inequality is greater, although the di erence between what the goods do to people is less. Apart from the suggestion of Hansson that one should measure inequality directly in terms of equalisanda and not in terms of distribuenda, the example discloses even more. It shows that one cannot take the loss of social welfare to be an index of inequality [Sen, 1973] [Sen, 1978]. Inequality is just one aspect of social welfare and is not represented by it. Stating that they are similar is mixing up the meanings in an unclear way. Social welfare is one thing and inequality another.
7.3. INCOMPLETENESS
227
w(ar ) w(ar )+u(ap ) 2 0 w (ar )+w0 (ap )
q
w0 (ar )
q q
w0
q
2
w(ap ) = w0(ap )
w
q
ap
a0e ae
ar
Figure 7.6: Hansson's problem Stated di erently, once one has a social welfare index representing judgements `all things considered', one does not need any longer an index of inequality [Osmani, 1982]. An index of inequality is useful as a component of the social evaluation in order to arrive at a judgement `all things considered'. But if one wants to use these judgements `all things considered' in order to arrive at an index of inequality representing the way inequality affects the social evaluation function, one has already in mind what is meant by inequality and one has already an idea of an index apart from the social evaluation. So, although the social welfare approach in order to arrive at a normative measure of inequality seemed to be a promising way to avoid arbitrary measures, it is not satisfactory because of an essential di erence between equality and social evaluations `all things considered'. Summarising, extending the ordering of the Lorenz dominance is not successful because it led to measures with some odd consequences in case of the GINI coecient and to arbitrary judgements in the case of decomposable indices, or in the case of the normative approaches they turn out to be not adequate at all. The discussion learned that an index should satisfy the principle of diminishing transfers, i.e. it should be more in uenced by transfers in the lower region than in the higher one. Furthermore, the argument of Hansson indicated that one should not turn to distribuenda or goods simpliciter but to the equalisanda, what goods do to people, in order to measure how serious inequality is. This failure to arrive at satisfying
228
CHAPTER 7. MEASURES OF INEQUALITY
extensions of Lorenz dominance, suggests the other answer to incompleteness of the ordering according to Lorenz dominance, namely acceptance of incompleteness. Let me look at that answer to incompleteness.
7.3.2 Acceptance of incompleteness and its explanation
One answer to the incompleteness of the Lorenz dominance is accepting the incompleteness of the ordering of inequality. Incompleteness could be considered to be expected in normative questions re ecting complexity. Realising this complexity, one could take for example as a measure of inequality what all the measures about which one is certain in a particular domain have in common. The nal ordering would be the intersection of these orderings. The ordering of Lorenz dominance could be interpreted in such a way. If all indices satisfy the principle of Pigou Dalton, principle of population, homogeneity and symmetry, then the Lorenz dominance is something they all have in common. Consequently, incompleteness is nothing mysterious and is just what was to be expected. Sen is suggesting this as part of an explanation of incompleteness [Sen, 1973, p. 72] [Sen, 1977]. This explanation of incompleteness is criticised by Temkin. He argues, that if one is certain about the adequacy of a measure in a particular domain, one can just rely on that measure; con rmation by other measurements is super uous [Temkin, 1993, p. 141 .]. If one of the measures is reliable in a particular kind of situations, then the fact that other measures are unclear or unreliable in this domain is not a reason for withdrawing the judgement that one distribution is more unequal than another. Divergence of opinion of the di erent measures is no reason to abstain from a judgement. One trustworthy measurement is just enough. Consequently, the intersection approach is not an appropriate explanatory reason for the incompleteness of the ordering regarding inequality. Although Temkin criticises the intersection approach suggested by Sen, he accepts incompleteness and underlines the complexity of equality.12 He argues that the idea of equality is exhibiting several principles. Temkin argues for these di erent principles underlying the idea of inequality because 12 Temkin writes:
What we need, it might be claimed, is to arrive at a measure of inequality that accurately captures each of the aspects involved in that notion, according them each their due weight. Such a measure would give us a way of accurately comparing many, though perhaps not all, situations regarding inequality. [Temkin, 1993, p. 52]
7.3. INCOMPLETENESS
229
,!
,!
A
B
C
Figure 7.7: Temkin's series they explain according to him the di erent judgements in cases like the following series in gure 7.7. In the rst situation all except one have a reasonable amount of some good. In the next all except two have this same amount of good and the two exceptions have the lesser amount. Continuing, we arrive at the situation in which all except one have the lesser amount of the good. The question is how to rank these situations according to the moral badness of inequality. Temkin argues several judgements to be reasonable. Some argue that the situations in the series become better and better, because in the rst there is one person victimised exhibiting discrimination. As the series progresses there are more people worse o and consequently this victimisation is less. Others argue the situations become worse and worse, because there are more and more worse o people, consequently the total amount of complaints of the worse-o is increasing. Di erent principles can explain the di erent judgements.13 This series used as illustration of di erences of judgement regarding inequality are situations with intersecting Lorenz curves.14 Suppose the
13 In the next chapter I return to the judgements on this series again on p. 259. 14 The other examples he presents in order to show that there are several di er-
ent principles behind the idea of equality are also situations in which the Lorenz curves intersect. These situations described in chapter 3 of his book Inequality , consist of four groups which from poor to rich receive a,b,c,d of some good, respectively, call e=a+b+c+d. By considering the Lorenz curves one could look at ve points, (0 0); ( 41 ae ); ( 42 a+e b ); ( 43 a+eb+c ); (1 a+b+e c+d ) Now it can be easily checked that the situations Temkin points to in order to argue for di erent principles, are situations with intersecting Lorenz curves. See [Temkin, 1993, p. 61, 63, 73, 75, 78]. ;
;
;
;
;
230
CHAPTER 7. MEASURES OF INEQUALITY
,, ,, , , ? , , ,, , , 6 , ,, Figure 7.8: Lorenz curves of Temkin's series level of the best-o is xmax and the level of the worst-o is xmin and the number of people is n. The amount of available goods can be represented by (n , m):xmax + m:xmin in which m is the number of worse o people. The fraction of the goods available to the poorest n1 fraction, Fxmin ; n1 is xmin 1 (n,m):xmax +m:xmin . The fraction of the goods available to the best-o n x max fraction is (n,m)xmax +m:xmin . So a point on the Lorenz curve linked to the all but the best-o fraction, is the fraction of the goods owned by max the people minus the richest fraction n1 , Fxmax ; n1 is 1 , (n,m)xxmax +m:xmin . In the series Temkin considers n is constant. By increasing m from 1 min to n , 1 we can see that Fxmin ; n1 = (n,m):xxmax +m:xmin is increasing and x max Fxmax ; n1 = 1 , (n,m)xmax +m:xmin is decreasing so the Lorenz curves of the situations in the series intersect, illustrated in gure 7.8. The incompleteness of the ordering regarding inequality is explained by Temkin by the di erent principles behind the idea of equality. Unanimity of the principles in their evaluation of situations nds its counterpart in unambiguous judgements. It does not mean that if unanimity is lacking judgements are not possible; it might be that in some situations one principle might prevail above another; this is the di erence between this approach and the intersection approach. However, it is not excluded that
7.3. INCOMPLETENESS
231
there might be con icts of evaluation which are not solved so easily. The situations pointed to by Temkin to clarify his arguments are examples of these con icts. The explanation based on the di erences of principles behind the idea of equality is attractive, however it presupposes a view on ethics in which principles constitute the basis of moral judgements. In a particularistic framework argued for in this study, it is not acceptable. Principles are not the basis of our moral judgements, so they cannot be the basis for incompleteness of judgements. In order to arrive at an explanation we should turn to those situations again and see what precludes unambiguous judgements. Judgements regarding inequality are concerned with distributions and redistributions. It is likely that the main diculty is that of comparing the advantage of one person with the disadvantage of another. For example, in the series cited above it is not clear how the disadvantage of the worse-o should be compared to the advantage of the besto vis-a-vis a reference situation. Or in other words, one is not clear how a change of the position of one person is to be compared to the change of the position of another person. The compensation for losses is not clear. This diculty of comparing advantages and disadvantages explains in a more direct way why judgements regarding inequality are incomplete. After all incompleteness in judgements is not uncommon in normative matters. How is this diculty of comparison to be explained? An explanation due to the impossibility of interpersonal comparisons is dealt with in chapter 3, and is excluded. There is no a priori argument for holding that what goods do to people cannot be compared. What could be possible is uncertainty about how much disadvantage of one is set o by the advantage of another. This could be framed in terms of a moral con ict like those of dilemmas, but in the context of equality in which we assume a fundamental similarity and comparability of people, this is excluded. The diculty is rather one of uncertainty than a dilemma. But uncertainty can be dealt with by invoking the vocabulary of probability and decision making under uncertainty. Can this uncertainty explain the incompleteness of the ordering of inequality represented by Lorenz dominance? Suppose intersecting Lorenz curves represent uncertainty about how much disadvantage can be set o against the advantage of someone else. Compare two distributions A and B with the Lorenz curves LA , and LB which intersects LA . Following the above explanation of incompleteness the intersection of LA and LB means that there is uncertainty in judgements regarding inequality. B could be seen as distribution A followed by a transfer from a rich person to a poor person in the upper part TrR1!p1 and a transfer from
232
CHAPTER 7. MEASURES OF INEQUALITY
,, ,,
, , , , BN ,, L B , L A , , @R LB0 ,,
Figure 7.9: Illustration of uncertainty is not an explanation for the incompleteness of Lorenz dominance a poor person to a rich person in the lower part TpP2!r2 . The uncertainty is based on the uncertainty in comparing TrR1!p1 to TpP2!r2 . Suppose further there is a range 4TpR1!r1 in which the transfers TpR1!r1 result in uncertainty of judgements between A and B. If the transfer from a poor to a rich person is greater than those included in this range, the resulting distribution is worse than A and transfers below those in this range, will result in a distributions better than A. By lowering the transfer TpR1!r1 and approaching zero there will be somewhere a distribution B 0 better than A. If TpR1!r1 = 0, this B 0 is certainly better than A because the LB0 is above LA because the upper part is equal to LA but the lower part is above LA due to the transfer TrP2!p2 . But it is highly unlikely that there is no TpR1!r1 such that B 0 is as unequal as A. But if there is such a transfer, then there is an intersection of Lorenz curves without uncertainty, namely LB0 and LA . So the intersections of Lorenz curves do not mean uncertainty on judgements regarding inequality. One has to forego the explanation of incompleteness of the ordering regarding inequality by uncertainty of comparing transfers of advantages and disadvantages, or one has to turn away from Lorenz dominance as the nal expression of an unambiguous partial ordering. Because the rst is in line with the particularistic conception of moral judgements and dicult to refute, I turn to Lorenz dominance and look more critically to its four de ning characteristics.
7.4. PROPERTIES OF LORENZ DOMINANCE REVISITED
233
Summarising, the Lorenz dominance, which is a well-known ordering of situations regarding inequality, su ers from incompleteness. Extending the ordering to a complete one either appeared to introduce unacceptable parts of the ordering, for example the GINI coecient or some arbitrary parts as in the decomposable ones, or just inadequate measures such as the normative measures. Accepting incompleteness however failed an explanation why a partial ordering is the one according Lorenz dominance and why the situations with intersecting Lorenz curves pose problems. So it is natural to revisit Lorenz dominance in order to arrive at an adequate measure of inequality.
7.4 Properties of Lorenz dominance revisited As was mentioned above it was shown by Foster that indices consistent with Lorenz dominance can be characterised by four principles: 1. Pigou Dalton principle 2. Symmetry 3. Homogeneity 4. Population principle If Lorenz dominance is left then also these principles should be left. But also if these four principles are not satis ed, the Lorenz dominance should be left. The Pigou Dalton principle seems to be indisputable, a transfer from rich persons to poor persons will lessen the inequality. But as appeared in the discussion on the extensions of Lorenz dominance notably the GINI coecient and the coecient of variation, a transfer should be more important in the lower region than in the higher. For an ordinal ordering in which one is only establishing whether or not a situation is worse than another regarding inequality, and not how much better or how much worse, the principle of diminishing transfers could mean only that there is a region in the upper part of the distribution in which transfers from rich to poor persons will not have any e ect.15 This is consistent with the idea developed in the previous chapter in which the complaints of those below the reference level should count. If the transfer is between those above the level there will be no in uence on how bad the inequality is. Although there is no reason to withdraw this principle categorically, it has to be amended 15 Temkin also points to a limited importance of the Pigou Dalton principle in the higher regions [Temkin, 1993, p. 79].
234
CHAPTER 7. MEASURES OF INEQUALITY
by adding that not all transfers have an impact but only those a ecting persons who are worse o than all could have been. Symmetry can be questioned if it is applied to distribuenda or resources and not to equalisanda, because what goods do to people is not for all people similar. Sen points to the con ict between the Pareto principle and Suppes' grading principle, showing a permutation does make a di erence [Sen, 1979a, p. 149]. His example is as follows. There are two persons a Hindu and a Muslim, and two situations: a) in which the Hindu is receives 0 units of pork and 2 units of beef and the Muslim 2 units of pork and 0 units of beef, and a situation b) in which the Hindu receives 1 unit of pork and 0 units of beef and the Muslim 0 units of pork and 1 unit of beef. Now according to the Pareto principle situation b should be preferred to a because the Muslim is not interested in pork at all and the Hindu not in beef, but according to the grading principle i.e. taking the permutation of the distribution, the evaluation is reversed. A permutation does make a di erence. However if one turns to the equalisanda in which these di erence are accounted for, the permutations will be equivalent. One should be clear what one understands by symmetry. By symmetry one can understand that in case there are two persons of which the characteristics are represented by an ordered pair, where xi is the resource part and yi represents the other characteristics what goods do to people, ((x1 ; y1 ); (x2; y2)) '(I ) ((x2 ; y1); (x1; y2 )). Symmetry interpreted in this way, as symmetry of resources, is not acceptable as was shown by Sen's example. But symmetry regarding what goods do to people represented by ((x1 ; y1); (x2; y2 )) '(I ) ((x2; y2 ); (x1; y1)) is dicult to deny. So, symmetry is acceptable but again it should be applied carefully, namely to the equalisanda and not to distribuenda. Homogeneity is a principle, which is also dicult to deny if it is meant to express invariance to the units of measurement. But a proportionally increase in what all have, for example an increase of 200%, is not directly clear to have no in uence on the inequality. If all improve, it could be that there should be a change of equalisandum.16 If all can a ord plenty of food the distribution problem is di erent and is no longer one concerning bare survival and it is transformed into the distribution problem of for example more luxury goods. This change of equalisanda from more urgent equalisanda to less urgent equalisanda could be an explanation for the idea presented by Temkin that the moral badness should be discounted the more auent the worst-o is [Temkin, 1993, p. 185]. The principle of homogeneity should be used with care. If it is meant to account for a change of units it is undeniable valid, but if it is used in situations in which all 16 See chapter 6 p. 210.
7.4. PROPERTIES OF LORENZ DOMINANCE REVISITED
235
improve proportionally, it is not acceptable at all. Increasing the equalisandum for all could mean a call for a change in equalisandum because of the change of the distribution problem at hand. So, its use is valid but dependent on a clear determination of the equalisandum. One should realise that one should compare primarily the equalisanda and not distribuenda. Homogeneity can be accepted but carefully.17 The population principle is the most implausible of the four principles. If a distribution is morally wrong then two such distributions are even worse. One situation with an unequal distribution is already worse than what could be ideally the case, but two is of course even worse. One could defend the principle by arguing that the badness of the inequality per person is not changed, but then one turned to a relativised version of the badness of inequality. This way of looking might be useful if one is interested in the mechanism of (re)production of inequality. In that case proportional numbers might be more informative than absolute gures. Although the study and understanding of how inequality is (re)produced is not unimportant, it is not the subject of this study here. Here I am concerned with the moral seriousness of inequality. It is plausible to hold that the more replicas of an unequal distribution the worse. This view however could be questioned by the following argument discussed by Temkin [Temkin, 1993, p. 218]. Suppose the more replicas or the more people with the same unequal distribution, the worse the inequality, then the magnitude of the di erence of the best-o and the worst-o could be traded o against the number of persons in the distribution. It could be possible that a large gap between the worst-o and the best-o for only a few persons could be even at least as good regarding inequality, as a situation with a small gap between the best-o and the worst-o while the worst-o are even better-o than the worst-o in the former situation as represented in gure 7.10. So, it seems to be possible that whatever the gap is, there is always a number n whatever the gap is, such that A and B are equally unequal. This consequence, called by Temkin the repellant conclusion, of the idea the more replicas of a distribution the worse, is dicult to accept. The argument is rather convincing but not unavoidable. In the argument it is assumed that all di erences can be traded o by numbers independent of how small this gap is. But this is not clear to be necessarily so. One can argue that the gap between the worst-o and the best-o is of such an important nature, or it concerns such an urgent equalisandum that it cannot be traded o by a lesser gap concerning a less urgent equalisandum distributed unequally among a greater number of people. This method of blocking the repellant conclusion is quite natural and ts in nicely within 17 In the next chapter I return to the use of homogeneity on p. 247.
CHAPTER 7. MEASURES OF INEQUALITY
236
f
f
n
A
B
n
Figure 7.10: Illustration of Temkin's repellant conclusion the particularistic and pluralistic framework which is used in this study. The di erence of urgency of the equalisanda explains the way the repellant conclusion can be blocked.18 The more persons with complaints, the worse. But it is also natural that more persons with no complaint will not in uence the badness of the inequality, it does not change the extent the complaints of those worse o than they could have been. Adding persons without complaints does not matter. This will be called the principle of independence of no complaints. Summarising, the Lorenz dominance has lost some of its attractiveness. The characterising principles are not acceptable without amendments. The Pigou Dalton principle is not necessarily applicable in the higher regions of the distribution; it is restricted to those a ected by the transfer which are worse o than all could have been simultaneously. It can be considered as the principle of diminishing transfers in an ordinal environment. The principle of symmetry is acceptable if it is applied to equalisanda, but not acceptable for distribuenda. Because of di erences in urgency of equalisanda the principle of homogeneity is only valid if it is meant to correct
P
18 Formally one could state that each replica makes the inequality worse but in a !1 g g ( g ), the g are such that the sum is convergent diminishing way. So in Gg=1 and bounded so that increasing the number of people will not give rise to an inequality of unbounded badness. This idea is discussed by Temkin, but is lacking an explanation, which makes him hesitating to accept this way of reasoning [Temkin, 1993, p. 223 .]. Instead of this formal solution the argument based on the di erences in urgency of equalisanda is more natural. a I
x ~
a
7.5. PROPERTIES OF A MEASURE OF INEQUALITY
237
for di erent units. These di erences between the urgency of equalisanda are also the reason why one can deny the population principle without accepting the repellant conclusion. It is plausible to accept the principle of independence of no complaints. So far the four properties are discussed and show that the Lorenz dominance is not the ordering with which an appropriate measure has to be consistent. It is however not excluded that Lorenz dominance can be seen to be a guiding ordering in situations in which there is the same number of people and there is the same total amount of the equalisandum. The most problematic properties, homogeneity and the population principle play no role in such cases. What about intersecting Lorenz curves in these cases? They are still not explained and Lorenz dominance is not excluded. May be the ordering can be extended in these cases as is suggested by some authors [Lambert, 1993, p. 73 .] [Shorrocks & Foster, 1987]. But such an extension should not be turned to without considering the general question of the ordering regarding inequality. The danger of suggesting an ad hoc solution is reduced by turning to this general problem. In order to arrive at a possible extension of the Lorenz dominance in the cases of same number, same total amount of equalisandum, but also in order to arrive at a satisfying ordering of inequality in general, I discuss some more properties which are suggested to apply to an index of inequality.
7.5 Properties of a measure of inequality Beside the characterising properties of the Lorenz dominance some other properties an index of inequality should satisfy are suggested. One idea is that an index should be zero if there is equality and greater than zero if there is inequality. If one is looking for an index representing an ordinal ordering then these properties formulated in this way are not useful, because the representation of an ordinal ordering is a whole class of indices up to monotone increasing transformations. There is no di erence between an index and its monotone increasing transformation, they both represent this ordering. This will mean that if the monotone increasing transfer is for example the addition of a constant greater than zero, then there is an index with a value greater than zero for the equal distribution namely, zero plus this constant. And although it is an adequate index it is not satisfying the above mentioned property. Of course it is not excluded that one could choose an index with zero as value if there is equality, but this choice is not representing an intrinsic feature of the ordering of inequality. Nevertheless, these properties stated together have some value. They indicate that if there is inequality the index should have a value greater than the value
238
CHAPTER 7. MEASURES OF INEQUALITY
of the index for an equal distribution. This property is easily seen to be implied by the restricted Pigou Dalton principle. It is also suggested that an index of inequality could or should be independent, meaning that the ordering of the inequality of a subgroup is independent from how others fare. Formally, for any group g and the rest :g: (~xg ; ~x:g ) I ( I )('I )(~x 0g ;~x:g ) () (~xg ; ~y:g ) I ( I )('I )(~x 0g ; y~:g ) Although a priori it would be a mathematically convenient property it should not be required from an index of inequality. In the previous chapter it was explained that how well o people fare above the reference, did not a ect the measure of inequality.19 Suppose we have rst two situations, say (~xg ; ~x:g ) and (~x 0g ;~x:g ) in which all are better o than the subgroup g so the inequality between the members of g does matter, and next a pair, say (~xg ; ~y:g ) and (~x 0g ; y~:g ), in which others not belonging to group g are worse o such that all of group g are above the reference, then it will be clear that the index will not depend at all on the inequality of this subgroup g all together and it could be the case that (~xg ;~x:g ) I (~x 0g ;~x:g ) & (~xg ; y~:g ) 'I (~x 0g ; ~y:g ) So the above property of independence is not valid for an index of inequality. The index should be dependent only on the people worse o than the reference. A weaker version of independence could be accepted, because it is clear from the discussion that the in uence of the inequality of a group will not be reversed, it is restricted to a change from some e ect to no e ect and from no e ect to some e ect. If one restricts oneself to I and I one could hold that (~xg ;~x:g ) I (I )(~x 0g ;~x:g ) () (~xg ; ~y:g ) I (I )(~x 0g ; ~y:g ) On the other hand if one restricts oneself to situations in which the reference is xed there will be no problem and the inequality varies with the complaints of the individuals or groups of individuals independent from how well o others are. This independence is also in line with the individualistic interpretation of equality argued for in the previous chapter.20 21 19 See chapter 6 p. 197. 20 See chapter 6 p. 190. 21 A version of decomposability restricted to xed references will not do so easily as
in the case of independence. Below in the Appendix 1 of the next chapter, I will argue it is not acceptable. The main problem concerns the di erent ideal references in the subgroups.
7.6. SUMMARY
239
There is another property that is sometimes suggested, namely speci city. This means that the inequality regarding some good should be considered apart from the inequality of some other good, i.e. separability with respect to equalisanda [Kolm, 1977]. With an argument like the one Kolm presents against partial equity there is also introduced some doubt whether speci city should be a property for an inequality index [Kolm, 1972, p. 81] . Take for example some index w representing how well o someone is, a similar function for all. Now the amount of w is dependent on the distribution of some other goods or characteristics, which can be considered to be the combination of a particular partition of the goods and characteristics, for example the goods and characteristics xi and some other goods and characteristics yi . Now if one is looking for equality one could argue that there is equality if w1(x1 ; y1) = w2(x2; y2 ). Suppose furthermore there is equality with respect to xi, so x1 = x2 = x so w1(x; y1) = w2(x; y2). Meaning that if y1 = y2 there will be equality, because the w are similar for all. But if there is no equality concerning xi there is likely to be no equality if y1 = y2 . So the inequality of xi and yi are not independent of each other. This argument is rather convincing if one is considering distributions of resources, but once one is restricting oneself to equalisanda it is not any longer compelling. The argument above stated that equality with respect to a distribuendum is dependent on equality of another distribuendum. But speci city as meant here should not mean: orderings concerning the distribuenda or resources are independent of each other. Speci city here should be stated in terms of equalisanda and should mean that one can consider the orderings of inequality with respect to an equalisandum independent of each other. Each problem suggests its own equalisandum. One could rebut that it is still wrong to consider the equalisanda apart. In the argument against speci city in terms of distribuenda, one can also interpret w x and y all as equalisanda. Now the argument is an argument against speci city in terms of equalisanda. The issue here concerns the aggregation of measures of inequality each with its own equalisandum into one overall index. In the next chapter, after the presentation of plausible index of inequality, I will argue that speci city is not acceptable for an index. The argument is better to be stated after an index of inequality is more clearly determined in the next chapter.
7.6 Summary In order to arrive at some properties an adequate index of inequality should satisfy, I started with the Lorenz dominance as an ordering representing
240
CHAPTER 7. MEASURES OF INEQUALITY
how bad inequality is. This ordering was incomplete, because in the case of intersecting Lorenz curves one has no indication how to order the distributions vis-a-vis each other. Extensions such as the Gini coecient, the decomposable indices such as the coecient of variation and Theil measures turned out to contain some odd and arbitrary judgements concerning inequality. The normative approach as a way of arriving at an extension was not satisfactory either, because equality is an element of social welfare and cannot be equated with social welfare. The lack of a satisfactory extension suggested the other answer, accepting incompleteness. This incompleteness required an explanation. It was argued that the explanations of Sen and Temkin were not convincing. Another possible explanation was to be found in the uncertainty of comparing advantages and disadvantages of di erent persons. It appeared that the incompleteness explained by uncertainty did not parallel the incompleteness of the Lorenz dominance, i.e. the intersecting Lorenz curves. Consequently, the four properties that characterise Lorenz dominance could not be accepted, although their revised versions could. Together with a discussion on some other properties we arrived at the following list of properties that an adequate index of inequality should satisfy: 1. A restricted Pigou Dalton transfer principle, meaning that a transfer from a poor person to a rich person will worsen inequality under the assumption that the poor person is worse o than all could have been simultaneously. 2. Symmetry restricted to equalisanda, meaning that changing who gets which part the equalisanda, or what comes to the same, take a permutation of the distribution, will make no di erence for the index of inequality. 3. Restricted homogeneity, expressing the idea that the index is independent of the arbitrary units of measuring the amount of equalisanda. 4. Principle of independence of no complaints , adding persons who have no complaints, who are not worse o than all could have been, does not change the index. 5. Weak independence, meaning that the ordering of inequality based on the distribution of a subgroup cannot be reversed by the welfare of others: (~xg ;~x:g ) I (I )(~x 0g ;~x:g ) () (~xg ; ~y:g ) I (I )(~x 0g ; y~:g ), which can be strengthened to strong independence if the reference representing how well o all could have been is xed. It was argued that decomposability or additivity of the index meaning that it is determined by the inequality within subgroups and inequality
7.6. SUMMARY
241
between the subgroups, is not acceptable. The discussion on speci city is postponed to the next chapter. With this list of properties which a measure of inequality should satisfy together with the meaning of the ideal of equality argued for in the previous chapter, a measure of inequality can be developed that represents a complete ordering regarding inequality in distribution problems.
CHAPTER 7. MEASURES OF INEQUALITY
242
7.7 Appendix 1 It is seldom explained that G(~x) = 1 + n1 , n22 x (nx1 + (n , 1)x2 + + 2xn,1 + xn) is equivalent to: Xn Xn jx , x j G(~x) = 2n12x i j i=1 j =1
In these formulations the fractions are ordered such that i is worse o than j i i < j. Here follows an explanation. We have to show that both formulations are equal: n X n 1 2 1 X 2n2x i=1 j =1 jxi , xj j = 1 + n , n2x (nx1 + (n , 1)x2 + ) () n X n n X 1 1 X 2 2n2 x i=1 j =1 jxi , xj j = n2 x (n x + nx , 2 i=1 (n , i + 1)xi ) ()
n X i n Xn x , 2 Xn (n , i + 1)x ) () 1 X 1 (X (x , x ) = nx + i n2x i=1 j =1 i j n2x i=1 i i=1 i i=1
Xn ix , Xn (n , i + 1)x = Xn (n + 1)x , 2 Xn (n , i + 1)x () i=1
i
i=1
i
i
i=1
Xn ix + Xn (n , i + 1)x = Xn (n + 1)x i=1
i
i=1
i
i
i=1
i=1
i
Chapter 8
A simple measure of inequality 8.1 Introduction In the previous chapter I arrived at a list of properties that a satisfactory measure re ecting the moral badness of inequality should satisfy. They are: 1. A restricted Pigou Dalton transfer principle, meaning that a transfer from a poor person to a rich person will worsen inequality, under the assumption that the poor is worse o than all could have been simultaneously. 2. Symmetry restricted to equalisanda, meaning that changing who gets which part of the equalisanda, or what comes to the same, take a permutation of the distribution, will make no di erence for the index of inequality. 3. Restricted homogeneity, expressing the idea that the index is independent of the arbitrary units of measuring the amount of equalisanda 4. Principle of independence of no complaints, adding persons who have no complaints, who are not worse o than they all could have been, does not change the index.
243
244
CHAPTER 8. A SIMPLE MEASURE OF INEQUALITY
5. Weak independence, meaning that the ordering of inequality based on the distribution of a subgroup cannot be reversed by the welfare of others: (~xg ;~x:g ) I (I )(~x 0g ;~x:g ) () (~xg ; ~y:g ) I (I )(~x 0g ; y~:g ), which can be strengthened to independence if the reference, representing how well of all could have been, is xed. Constructions of a satisfactory index or measure I(~x) such that it is greater if and only if distribution ~x is worse with respect to inequality, so far discussed, failed. This failure did raise some doubts about the possibility of a complete ordering of inequality in a distribution problem. The failure had to be explained and these explanations turned into arguments against a simple measure representing a complete ordering in distribution problems. This line of reasoning is blocked if we can determine a simple measure. In this chapter such a simple measure I(~x) is suggested which is determined up to a monotone increasing transformation. It is sucient to support the idea that in distribution problems there is a complete ordering regarding inequality. I start the determination of the measure of inequality by discussing the measure of equalisanda in section 8.2.1. A proper measure of equalisanda is a prerequisite for a satisfactory measure of inequality. Problems concerning the measure of equalisanda will infect a measure of inequality. In section 8.2.2, I will determine the form of the measure of inequality. The properties: independence, the restricted transfer principle and symmetry, and the invariance to units of a measurent of the equalisandum, lead to a measure which is additive separable with respect to the amount of equalisandum allocated to persons. By attending to iso-inequality curves the measure will be determined further. In section 8.2.3 it will be argued that from this class of functions characterised in the previous section one can choose one particular class that has some convenient properties the others lack. It is the class of functions that are monotone transformations of the Euclidean distance from a distribution to the ideal reference, restricted to those being worse o than they could have been in the ideal reference. Next, I show in section 8.3, how this measure meets the arguments of Sen and Temkin against simple measures. It is argued that on a closer examination the di erent judgements on the series that Temkin presents as evidence against a simple measure, support the idea of a simple measure. Finally, in section 8.4, I have some remarks on the aggregation of the measures of inequality concerning the diverse equalisanda.
8.2. THE CONSTRUCTION OF A SIMPLE MEASURE
245
8.2 The construction of a simple measure 8.2.1 The measure of equalisanda
One part of a proper measure of inequality concerns the measure of equalisanda. A measure of inequality presupposes that the equalisanda can be ordered and measured. Problems with a measure of the equalisanda will lead to problems in the ordering regarding inequality. If the ordering with respect to the equalisanda is not clear then an ordering of inequality will of course be problematic too. In chapter 3 the possibility of interpersonal comparability was already explained. Denying interpersonal comparability was like being a strange creature, a solipsist.1 By the argument there, comparability of levels of equalisanda was established. Having a certain amount of distribuenda leads to a certain amount of equalisandum which can be compared to the amount of equalisandum another enjoys. But it was not established that di erences of amounts of equalisanda could be compared between persons. The latter is of course a desirable property for a measure of inequality.2 But as was made clear by Ng it is plausible that if interpersonal comparability of levels is possible then interpersonal comparability of di erences in equalisanda is possible [Ng, 1984]. Comparability of di erences will follow from level comparability if some conditions are satis ed. These are: 1. existence of overlapping individuals, meaning that there are distributions such that a person is better o regarding the equalisandum with one distribution than another person is with another distribution, and there are distributions such that it is the other way around. Furthermore, persons are not equally well o with regard to the equalisandum in all these distributions. 2. semi-connectedness and continuity, meaning that between any pair of distributions in which a person is di erently well o with respect to the equalisandum, there is a continuum of distributions connecting the indi erence curve of one distribution with the other. The rst condition is satis ed in contexts in which inequality has some relevance. If the condition was not satis ed, some persons would be, whatever the distributions, worse o than all other persons. But then there would be no inequality with respect to these distributions. Because all but See chapter 3 p. 101. If comparing the equalisanda di erent persons enjoy was not possible the orderings would be restricted to those of maximising the minima [Sen, 1973, p. 44]. Improving the second worst-o would not mean an improvement with respect to inequality which is not always according to our intuitions. 1 2
246
CHAPTER 8. A SIMPLE MEASURE OF INEQUALITY
x~ 0
q
~s ~z
~x q
q q q
~q
Figure 8.1: Indi erence curves for i in Ng's argument the worst-o would be better o than the reference and the latter would not be below the ideal reference, there would not be any relevant inequality with respect to this equalisandum. The second condition is also seen to be satis ed in the situations in which there is some inequality. In case of discrete equalisanda while the second condition cannot be satis ed, it is a `have or have not' situation like for example having or lacking a normal physiological apparatus for vision. In such cases the amount of the equalisandum one person enjoys can be taken arbitrarily and level comparability is directly di erence comparability. Ng's reasoning is as follows, it is illustrated with the help of the gures 8.1 and 8.2. Take for example two overlapping individuals i and j and distributions ~x;~x 0 ; y~; ~y 0 such that in ~x person i is better o than person j is in ~y with respect to the equalisandum, xi > yj , and in ~x 0 i is worse o than j in ~y 0 ; x0i < yj0 , and furthermore xi > x0i and yj0 > yj . This is possible because of the rst condition, the existence of overlapping individuals. Because of semi-connectedness and continuity there exists a distribution ~z such that zi = x0i, and a continuity of distributions between xi and x0i , whether or not via zi . Similarly, there exists a continuity of distributions between yj0 and yj whether or not via a distribution w~ such that wj = yj . Take some ~s such that yi < si < xi than there is distribution ~t because of semi-connectedness and continuity such that yj < tj < yj0 and si = tj . But because of continuity there is also a distribution ~q such that x0i < qi < si and also a distribution ~r such that qi = rj < tj . Now because we have si = tj and si > qi = rj < tj , we can conclude that si , qi = tj , rj which
8.2. THE CONSTRUCTION OF A SIMPLE MEASURE
247
y~0 ~t ~r q q
~y
q q
w~
q
Figure 8.2: Indi erence curves for j in Ng's argument is interpersonal comparison of di erences of equalisanda. This reasoning is applicable to all persons, consequently we have level comparability and comparability of di erences. Because for all persons one could easily accept that there is one and the same distribution leading to zero of an equalisandum, namely having zero of the distribuenda, this implies a complete ordering of an interpersonal comparable equalisandum given the equalisandum is intrapersonally a complete ordering, which can be represented with a ratio scale. In such a scale the measure of the equalisanda is determined up to multiplication by a positive real number.3 The units of measuring the amount of equalisanda can be chosen arbitrarily. This was the reason for homogeneity as one of the properties a measure of inequality should have. The structure of the measure of inequality should be invariant to the units chosen.4 So far the measurement of the equalisandum is discussed, let me turn to the measure of inequality.
8.2.2 Independence
An index of inequality is of course dependent on the distribution x1 ; : : :; xn or ~x of equalisanda in which xi represents the amount of the equalisandum allocated to person i = 1; : : :; n. Such an index can be represented by I(~x). The ordering regarding inequality is ordinal in which only better or worse See also [Roberts, 1977, p. 23]. It will appear in the next section that the measure of inequality is a homothetic function 3
4
248
CHAPTER 8. A SIMPLE MEASURE OF INEQUALITY
with respect to inequality counts, so the measure I() will be determined up to monotone increasing transformations. In the previous chapters, it was also made clear that what matters regarding inequality is that some are worse o than all simultaneously could have been. That some are better o is not important at all. The moral badness of inequality is dependent on the complaints of those worse o than they could have been. So the index of inequality is dependent on xref , xi for those persons i for which xi < xref , xref represents how well o with respect to the equalisandum all could have been, i.e. the ideal reference. If xi > xref , more or less xi will have no in uence. So the relevant complaints of person i can be de ned by ci = xref , minfxi ; xref g. The complaints of those persons having more than in the reference, is zero. The index of inequality is dependent on complaints, so it can be represented by I(~c). One of the properties a measure should satisfy is independence. It was rst restricted to weak independence because whether an increase in the equalisanda resulted in a decrease of inequality was not independent of how well o the others were.5 It was argued to be dependent on how well o the others were, because the reference could be di erent. But once the reference is xed, one can accept independence. This means that if the complaints of one group g increase or decrease, then independent of the complaints of the others who belong to the group :g, the badness of the inequality varies with the complaints in group g. This will mean that a certain increase in the complaint of one person in group g can be equivalent to an increase of the complaint of another person in group g independent how well o those in :g are. With equivalence is meant here that the inequality is the same. The index does not change if one interchanges or substitutes the increase of the complaint of one person with an increase of the complaint of another one. In other words an increase of a complaint of one person which can be o set by the decrease of another such that the moral seriousness remains the same, is independent of how well o others are. How they should be compared is independent of how well o the persons in :g are. This property called independence has as a consequence that the index of P inequality is of the separate additive form M( ni=1 fi (ci )) (M() being a monotone increasing transformation function). That this separate additive form is implied can be seen in the following way. Suppose the index can be di erentiated twice with respect to the complaints; an assumption which is quite acceptable. The rate of substitution of complaints between two individuals is independent on how well o others are. More formally: 5
See p. 238 chapter 7.
8.2. THE CONSTRUCTION OF A SIMPLE MEASURE
249
d( dIdc(~ic) = dIdc(~jc) ) = 0; i; j 6= k dck By a theorem of Leontief it can be shown that I() will be of the separate additive form Kolm[Kolm, 1977, p. 10].6 An example with three individuals will suce to give the idea of how the result is reached. With three individuals there will be three equations of the following form arrived at by the di erentiation rule for quotients: d2I : 1 , d2I : dI = 0 dc1dc3 dcdI2 dc2 dc3 dc1
8i; j; k :
d2I : 1 , d2I : dI = 0 dc1dc2 dcdI3 dc3 dc2 dc1 d2I : 1 , d2I : dI = 0 dc2dc1 dcdI3 dc3 dc1 dc2 Because dcdi2dcI j = dcdj2dcI i , else I() would not be di erentiable twice [Apostol2, 1961, p. 278] the equations are of the form: dI = 0 a: dI1 , b: dc 1 dc2 dI = 0 k: dI1 , b: dc 1 dc3 dI = 0 k: dI1 , a: dc 2 dc3 while 2 2 2 I a = dcd dcI ; b = dcd dcI ; k = dcd dc 1 3 2 3 1 2 After substituting the results from the second and the third equation in the rst equation one arrives at: dI ) = 0 a:( dI1 , dc 2 dc2 6 This property is also used in the informal proof given by Broome in [Broome, 1991, p. 83]. There are also some other proofs see for example [Fishburn, 1970].
250
CHAPTER 8. A SIMPLE MEASURE OF INEQUALITY
Meaning a = 0 resulting in b = 0 and k = 0, because the other factors cannot be zero, I() is varying with the complaints. The other solution dI dc2 = 1 means also that b = 0, and consequently a = 0. So, we can state that the partial derivatives are functions fi0 of those components belonging to i only. So the gradient of M(~c) is: f10 (c1 )~e1 + f20 (c2)~e2 + f30 (c3 )~e3 (~ei is a unit vector on axis i) This means that the index is of the separate additive form. That can be seen by integration via the path (0; 0; 0); (c1; 0; 0); (c1; c2; 0); (c1; c2; c3) P 3 which is i=1 fi (ci ) in which fi0 is the derivative of fi .7 The minimum of the index is reached if the complaints ci are zero for all. The greater the value of this sum of fi 's, the greater the inequality. Because we are after a representation of an ordinal P ordering the index of inequality will be of the separate additive form M( ni=1 fi (ci )) in which M is a monotone increasing transformation function. So far, we did not use all the properties enumerated above and a natural question is what can be said more about the index because of these properties beyond that it is of a separate additive form. Symmetry with respect to the equalisanda will have as a consequence that the index will not be di erent if the positions in the tuple ~x describing the distribution change. The dependence of fi on i, i.e. the position, is not allowed by symmetry,Pson 8i; j; c : fi (c) = fj (c) = f(c). Hence, the index will be of the form: M( i=1 f(xref , minfxi ; xref g)) Furthermore, the principle of transfers of Pigou Dalton, will lead to the restriction that the possible functions f are convex.8 If it was not convex, a transfer from a relatively well o person below the reference to another worse o person below the reference, would not lead to a decrease of the index of inequality. But can we say still more about that function f? By di erent f's we have di erent forms of the inequality equivalents, or iso-inequality curves in the space of distributions X n . Two distributions on such a curve, de ned by I(~c) = a and a is constant, are equivalent regarding inequality. The form of this curve P can be derived by the gradient because the direction of the gradient is: ni=1 f 0 (ci )~ei .9 (f 0 is the derivative of f), is normal to this curve [Apostol2, 1961, p. 265]. The direction of the gradient This line integral is independent of the path, because ( n ) can be taken as a 7
0
f ;:::;f
1
0
continuous gradient eld [Apostol2, 1961, p. 339] and so the index of ( 1 2 3 ) is well de ned by this integral. 8 A function is convex if all the points of a line between two points belonging to the graph of the function, are above the graph. n dM ( f (ci )) n 9 The gradient is i=1 d( ni=1f (c )) ( i ) i but the direction of the gradient is x ;x ;x
P
P P
i=1
i
f
0
c
~ e
8.2. THE CONSTRUCTION OF A SIMPLE MEASURE
251
xref
, , , - ,, , x6 , 6 , , ,, xref x2
1
Figure 8.3: Iso-inequality curves expressing almost the maximin measure is the direction in which a variation of inequality is maximal. Some gures can be illuminating. Suppose three persons of which person 3 has more than the reference. The iso-inequality curves could for example be as in gure 8.3. It shows an example in which the greatest change in the index of inequality is made by changing the complaints of the worst-o . Figure 8.4 is an example in which the greatest change in the inequality is made by an equal change of the complaints of all. Figure 8.55 shows the case in which the direction of the greatest change is in the direction of the reference, i.e. a change of complaints proportional to the complaints persons have. This class of measures can be characterised more precisely.10 Because the equalisandum is measured on a ratio scale the structure of the measure of inequality, the form of the iso-inequality curves, should be invariant to the units used to measure the equalisandum. Because the inequality is measured up to a monotone increasing transformation, the direction of the gradient should be similar on each ray through the reference. They are not dependent on the length of the line between the distribution and the reference, but only on the ratios of the complaints, because the length of the line on a ray has no absolute meaning at all. It could mean a particular amount of inequality, but also any other amount of inequality. The direction
P P
similar nto the one mentioned in the text because one can divide by the common factor dM ( i=1 f (ci )) . d( ni=1 f (ci)) 10 The idea of the following explanation is derived from the argument of Hicks in [Hicks, 1965, p. 336].
252
CHAPTER 8. A SIMPLE MEASURE OF INEQUALITY
xref
@@,, @@ , ,
@@ , , x6 @ @, @,@,@@ , ,, @@ @@ x2
1
xref
Figure 8.4: Iso-inequality curves expressing almost the total deviation from the reference as measure
xref
, , , , * , , x6 , , , 7 , xref x2
1
Figure 8.5: Iso-inequality curves expressing the Euclidean distance to the reference as measure
8.2. THE CONSTRUCTION OF A SIMPLE MEASURE
253
of gradient should not be dependent on the monotone transformation and the particular units used in measuring the equalisandum. In other words the ratios of the parts of gradient should be independent of the units. This means for the gradient: 0(Pn f(ci )):f 0 (ci ) M 0(Pn f(:ci )):f 0 (:ci) M 8ci ; cj ; : M 0(Pni=1 f(c )):f 0 (c ) = M 0(Pni=1 f(:c )):f 0 (:c ) j i j i=1 i i=1 Hence, 0 (ci ) f 0 (cj ) = k() 8ci ; cj ; : ff0 (:c = ) f 0 (:c ) i
j
k() is a value dependent on but independent of ci . Di erentiating with respect to gives: 0 (ci ):f 00 (:ci ):ci 0 , (ff0 (:c = k0 () i )):(f 0 (:ci )) This is similar to: 0() f 00 (:ci ):ci = f 0 (:ci ): , 1: kk() Take = 1, we have: f 00 (ci ):ci = f 0 (ci ):a (a is constant with respect to ci) The solution to this is: f 0 (c) = a1cp + a2 This means for f that it is of the form:11 f(c) = b1cq + b2 :c + b3 Because of ordinality the constant b1 can be incorporated in the monotone transformation M() of the index of inequality, we arrive at: f(c) = cq + d1 :c + d2 Because adding more people which have no complaint does not a ect the inequality, (the principle of independence of no complaints) implies f(0) = 0, so d2 = 0. 11 Strictly spoken, one should add by convention + 1 + 2 if = ,1. This is however excluded because should be convex because of the restricted Pigou Dalton principle. This form violates this condition. logc
f
d :c
d
p
254
CHAPTER 8. A SIMPLE MEASURE OF INEQUALITY
Once again using the fact that the equalisanda are determined up to a ratio scale it will be clear that d1 = 0. q:(ci)q,1 + d1 = q::(ci)q,1 + d1 8ci ; : q:(c j )q,1 + d1 q::(cj )q,1 + d1 But this is only possible if d1 = 0, (q > 1). So the class of indices can be described as: Xn M( (ci )q ) i=1
In which q > 1, because of convexity to f. If q ! 1 gure 8.3 will result [Varian, 1992, p. 20 .]. All functions belonging to this class are possible. The question is whether the form of the iso-inequalities can be determined further, or what comes to the same, whether f can be determined.
8.2.3 Euclidean distance as basis The indices of inequality are of the form:
Xn
M( (ci )q ); (q > 1) i=1
An index is determined by a value of q. The question is: can the indeterminacy of the iso-inequality curves be resolved? This indeterminacy of the iso-inequalities was the reason for the gaps in the evaluations, whether a distribution is more or less unequal than another, discussed in the previous chapter.12 The indeterminacy can be seen to represent the uncertainty about how much an increase in complaints of one person can be set o against a decrease in complaints of another without changing the seriousness of the inequality. By determining this form of iso- inequalities the uncertainty seems to disappear. But this disappearing uncertainty is contrary to our experience of uncertainty. It could be argued that a further speci cation of the form of the iso-inequalities is not to be expected and even not desirable, because if it was determined precisely, the index would not represent the judgements adequately by denying this uncertainty. Although this reasoning seems rather convincing, it is not valid. Our experience of uncertainty is not contradicted by a further determination of the iso-inequalities. It is denied that this indeterminacy and uncertainty is an inherent feature of inequality. Uncertainty can mean that although, 12
See chapter 7 p. 231.
8.2. THE CONSTRUCTION OF A SIMPLE MEASURE
255
some distribution is worse than another it is not seen by us to be so, for example because of lack of knowledge. This can be the case for example in case we are not sure about the amount of equalisanda someone has. We can be uncertain about the exact dependency of the equalisandum on the distribuenda.13 It is also possible that the reference is not determined precisely. In that case, the direction of the gradient is not determined q ref ,xi ) varies with xref . It can be concluded that a exactly because ((xxref ,xj )q further determination of f is not contradicting the possibility of lack of a clear evaluation in a particular case, uncertainty remains possible. Is there a way of specifying or singling out one of the indices? Of course one could choose a value of q and indeed this particular feature is separating an index from the rest. But this is like using names or dates of birth as relevant features in moral matters.14 It does not meet the argument of arbitrariness. If particular values of q can be used as singling out the index from the rest, then any particular value could be used as such and indeterminacy remains. Looking at the gures above it is remarkable that all the indices but the one with q = 2 have the feature that a change of inequality in the direction of the gradient, results in a change of the direction of the gradient of the distribution arrived at. This change of the direction of the gradient is dependent on the former change in inequality as is illustrated in gure 8.6. But this dependency will, whatever the way it is constructed, arbitrary. A problem the index with q = 2 does not su er from, because there is no such dependency. The direction of the gradient does not change if there is a change of inequality in the direction of the gradient. The index with q = 2 will not su er from arbitrary dependency of the direction of the gradient on former changes in the direction of the gradient. The index with q = 2 is also the only one such that the direction of the maximal change, i.e. the gradient, at any point is directed to the ideal reference, a property that is likely to be expected from a measure. This index is the only one such that the distribution of the total amount of equalisandum necessary to reach the reference, is following the direction of the maximal change of inequality. This can be seen as follows. The total amount P to make all the comP necessary P of the equalisandum plaints zero is ni=1 4xi such that ni=1 4xi + ni=1 ci = 0. Suppose now the distribution representedPby ~ of thisPamountP is in the direction of the P f (ci ) n n n n i gradient, j = f (cj ) and i i = 1, ( i=1 i ( i=1 4xi ) = i=1 4xi) 0
0
13 14
See for example the footnote on p. 205 in chapter 6. See chapter 2 p. 24.
256
CHAPTER 8. A SIMPLE MEASURE OF INEQUALITY xref
,, ,, , , x6 , , ,, xref x2
1
Figure 8.6: Illustration of the change of gradients dependent on former changes in inequality in the direction of gradients
P
Because 8i : ci + i( ni=1 4xi) = 0 we can state: P f 0 (ci ) = i = i: ii=1 4xi = ,ci = ci f 0 (cj ) j j : Pii=1 4xi ,cj cj
In other words f c(ici ) = k (k is a constant independent from the complaints ci ). So f 0 (ci ) = k:ci. Integrating this equality shows f(c) is of the form c2 . If the total amount of equalisanda necessary to reach egalitarian reference is distributed along the gradient then the reference will not be reached unless q = 2. If q 6= 2 the reference will not be reached because of wasting the equalisandum to those already better o than the reference. Only the function with q = 2 has the property that an improvement along the gradient at a point will result in reaching the reference. Because of these properties, a change along the gradient will lead to the reference, and the change of the direction of the gradient is not dependent on the change of the inequality, the index with q = 2 can be given priority to the other indices. So, the index re ecting the moral badness of inequality in a proper way is: 0
Xn (x
I(~x) = M(
i=1
ref , minfxref ; xig)2 )
The inequality is worse, the greater the Euclidean distance between the reference and the distribution restricted to those that are worse o than all could have been.
8.3. THE ARGUMENTS FOR COMPLEXITY REVISITED
257
,, , , , , , , , , : , , x6 @, , @@, ,@ , , @ xref x-
xref
2
1
Figure 8.7: Illustration of changes in inequality in the direction of gradients This index has some resemblance to the variance and the coecient of variation. There is some relation with the descriptive statistical measures. It would of course be very surprising if there was no relation at all. After all, all the measures that are suggested try to capture some aspect of inequality. Their disadvantage however, was that they lacked the proper connection with what is morally wrong with inequality. The index presented here has the advantage that it is not arbitrary, it is based on the meaning of inequality re ecting what is morally wrong with inequality, namely that some persons are worse o than all could have been simultaneously. On the other hand because there is such a close connection with the coecient of variation we should be careful and check whether it does not su er the drawbacks of the this coecient, in other words whether it holds against the arguments against the measures discussed in the previous chapters. Because if it could not, there would be a powerful reductio ad absurdum argument against the possibility of a simple measure of inequality.
8.3 The arguments for complexity revisited It could easily be argued that one important reason for the complexity of inequality has its origin in the assumption that there is one equalisandum. This assumption is responsible for some problems about determining the meaning of equality and also about the measure of inequality. If it is assumed that there is one equalisandum it seems that equality is complex while in fact the problem is located at the attempt to search for `the most
258
CHAPTER 8. A SIMPLE MEASURE OF INEQUALITY
important equalisandum' in all situations. If there is no ordering with respect to this equalisandum, there will not be an ordering of inequality either. But as will be clear, it is not the measure of inequality which is complex but the ordering regarding this `most important equalisandum'. Once moral value pluralism is accepted, the problem or the lack of the ordering of this `most important equalisandum' can be bypassed and it is no longer an argument for the complexity of the measure of inequality. But even if such a plurality is accepted and one turns to one among several equalisanda, even then there are some arguments against a simple measure. These were discussed in the previous chapter.15 So it is checked whether the index presented here, which has some resemblance with the variance or coecient of variation, does not su er from the disadvantages of these measures. Sen mentions against the coecient of variation that it is arbitrary and the principle of diminishing transfers is violated, furthermore only comparisons with the mean count.16 These arguments are not valid against the measure presented here. First, although it could already be questioned whether the variance was based on a comparison with the mean only, for the index presented here it is certainly not valid.17 There is no comparison with the mean but with a reference which represents the situation in which all are simultaneously equally as well o as could be with respect to the equalisandum. Sometimes this could happen to be the mean, but it is not necessarily so. This argument of Sen is not valid against the index presented here. Second, Sen argued that the principle of diminishing transfers is violated by the coecient of variation. Again, this argument is not valid for the index presented here, because as was argued, the only way in which an ordinal scale could exhibit this principle of diminishing transfers is by denying that in the higher regions of the distribution, above the reference, a transfer has no e ect at all.18 More or less in uence cannot be described in another way than either some in uence or no in uence at all, because di erences in inequality cannot be compared. This idea is incorporated in the index by only measuring the gaps between those who are worse o than they might have been in the reference. Those better o do not count at all. Third, Sen argues that taking the square of the gap in order to satisfy the principle of transfers of Pigou Dalton is just arbitrary, why could not another convex function be taken. But this argument too is answered by this index. It was shown that taking the square was not arbitrary at all, it In section 7.3.2. See chapter 7 p. 224. 17 See the remark by Kakwani cited in footnote on p. 224 in chapter 7. 18 See p. 233 of chapter 7. 15 16
8.3. THE ARGUMENTS FOR COMPLEXITY REVISITED
,! A
259
,! B
C
Figure 8.8: Temkin's series was the only one with the property that an improvement along the gradient at a point would result in reaching the reference. Of course, it could happen to appear that another index can be singled out, but for the time being there is no such reason and the most plausible index is the one following the Euclidean distance. So Sen's arguments can be answered. What about the ones presented by Temkin? The arguments presented by Temkin against the coecient of variation or the variance are similar to those of Sen [Temkin, 1993, p. 123 .], which have been dealt with in the previous paragraph, but they are embedded in a more general argument for complexity, namely his claim that there are several di erent aspects of inequality, which is a thread to the idea of a simple measure. His argument is mainly based on di erences in evaluations of a series of situations in which there become more and more people worse o illustrated in gure 8.8. There are the following evaluations with regard to inequality if the number of the poor increases in the series above: 1. better and better 2. worse and worse 3. all equivalent 4. rst worse, then better It is remarkable that the evaluation: ` rst better, then worse', is missing. The possibility of these di erent judgements is explained by Temkin by
260
CHAPTER 8. A SIMPLE MEASURE OF INEQUALITY
di erent principles. Now we should check whether the index presented here can account for these di erent judgements. How is the series to be evaluated by this index? In the series, there are two groups, the rich and the poor people, receiving xrich and xpoor respectively. Further the reference xref is equal or below what the rich receive.19 Suppose the number of the poor is m. Further, the reference will generally be dependent on m so, xref = g(m) while g0 (m) 0, (g0 (m) is the derivative of g(m)) because if m increases the reference will decrease or remain constant. An increasing reference is not plausible. The inequality index will be a monotone increasing transformation of m(g(m) , xpoor )2 . The judgement `better and better' means that this index decreases, in other words, the derivative is negative and remains negative. So let us turn to the derivative. This is: (g(m) , xpoor )2 + 2m(g(m) , xpoor )g0 (m) = (g(m) , xpoor )(g(m) , xpoor + 2mg0 (m)). The judgement better and better means that g(m) , xpoor + 2mg0 (m) < 0 for all m which is possible. The judgement `worse and worse' means g(m) , xpoor + 2mg0 (m) > 0 for all m which is also possible. The judgement that all would be equivalent would be the case if g(m) , xpoor1 +2mg0 (m) = 0 for all m, this would be the case if g(m) is of the form m, 2 , xpoor :m + k or if g(m) = xpoor . `First worse, then better' would be the case if there was a change of sign from positive to negative. If the sign of the second derivative is negative, this will not be a problem. The sign of the second derivative is negative, i.e. g0 (m) + 2g0(m) + 2mg00 (m) < 0 if g00(m) < 0, in other words if g(m) is concave. g(m) being concave means that the changes in the reference will be less and less as m, the number of the poor, increases. Hence the judgement of ` rst worse, then better' will be possible under quite reasonable assumptions.20 What about the judgement ` rst better, then worse'? Although this is possible by the index presented here, the function will behave rather peculiar. This can be seen if the series is extended with the extremes on both sides, one in which all are equally well o as the rich people and one in which all are equally well o as the poor people. Because the extremes show no inequality, the judgement `better and better'would be ` rst worse, then better'. The judgements `worse and worse' would be ` rst worse, then better' at the end. The judgement `all equivalent' would remain all equivalent, if all could have been not better o than the See the discussion on this in chapter 6 p. 204. The arguments are similar if the index was determined by 6= 2 1. In the formulas in the text the 2's can be replaced by . 19 20
q
q
>
8.4. SPECIFICITY AND AGGREGATION
261
worst-o , or it would be rst worse then equivalent and subsequently better, as in an on-o situation. The judgement ` rst worse, then better' is just ` rst worse, then better'. The evaluation ` rst better, then worse' would be in fact, ` rst worse, then better, then worse again and nally better'. This is a rather peculiar behaviour of an index of inequality. This strange behaviour is consistent with the lack of the judgement ` rst better, then worse'. The lack of this judgement can be accounted for by assuming that a proper index gives the underlying judgement ` rst worse, then better'. So, an index of inequality as presented in this chapter can account for the judgements which Temkin cites as evidence for complexity. But it does something more, it can account also for the lack of judgement ` rst better, then worse'. In Temkin's account of the judgements, the missing of the judgement ` rst better, then worse' just happens. Actually the acceptance of an index is more in line with the judgements on the series than the account of Temkin himself. The judgements on the series do support the idea that there is a simple measure instead of the idea that (in)equality is complex. Summarising, the assumption that there is an index satisfying the properties an index of inequality should satisfy led to an index which follows the Euclidean distance of a distribution to the ideal reference. In this it has a resemblance with the well-known measures as the variance and the coecient of variation. It did not exhibit the drawbacks these su ered from. It could account for the judgements of the series Temkin presented as evidence for the complexity of inequality, it could even account for the implausibiliy of a particular judgement, ` rst better then worse'. This judgement would mean that the judgements regarding inequality would vary in a very peculiar way in Temkin's series.
8.4 Speci city and aggregation Finally, one issue about the index of inequality has to be looked at. In the introduction of this study, I wondered whether it is possible to accept the importance of equality as an ideal within a particularistic framework. I argued that the simple index appeared to be possible because of this framework and not despite of it. The moral universalism, implying that like cases should a priori be treated similar, or what came to the same, if something is worthwhile in one situation it is a priori also in another, although the way and extent in which can di er, blocked a simple measure of inequality. This idea that increase in one value will mean an increase in the total value, led to a quasi-ordering.21 Sen argued that the extent to which inequality 21
See chapter 1 p.7.
262
CHAPTER 8. A SIMPLE MEASURE OF INEQUALITY
matters is not independent from the other values and principles [Sen, 1973, p. 75]. For example, the situation in which there is an equal distribution of one good or equalisandum which precludes the survival of at least one person, is not preferable to the situation in which one survives but in which the distribution of the equalisandum is unequal. It is not preferable even not from an egalitarian point of view. In this way moral universalism led to a quasi-ordering instead of an ordering. By accepting moral particularism and moral value pluralism, it becomes possible to analyse the situation di erently. By accepting the plurality of equalisanda, equality is not linked to one equalisandum or one good which was supposed to be an aggregation of several goods. It became possible to analyse a situation such that although a distribution is better regarding inequality with respect to one equalisandum it is not so regarding equality with respect to another equalisandum. Precisely this possibility blocks speci city of equality with respect to one equalisandum. Kolm proved that if inequality exhibited speci city, i.e. separability with respect to equalisanda, there would be a general index of inequality I(X) = F(h1(I 1 (~x1)) + : : : + hm (I m (~xm ))) (in which the superscripts indicate a particular equalisandum), which is separate additive with respect to the inequality indices each concerning one equalisandum. Now it can be shown that speci city does not apply.22 If speci city was valid, it would mean, because of the separate additive form of aggregation, that less inequality with respect to one aspect, would be better regarding inequality in general, disregarding the amount of inequality of the other equalisanda. But the example concerning survival mentioned above, shows it does not. In the situation described above the value of inequality with respect to the other equalisanda does not count, because in this case it could be argued that survival should be the proper equalisandum. Because only one person can survive, equality has nothing to order. In this respect, the survival of only one person is not worse regarding inequality than no survival at all; the reference is no survival at all. Equality with respect to the distribution of the other equalisanda does not make it better regarding inequality. The functions hj could not be adjusted so that the foregoing situation is treated properly by this separate form of aggregation, namely, that equality with respect to the other equalisanda has nothing to contribute to the equality.23 An aggregation in this way is Because of individualism and speci city this index would be possible [Kolm, 1977]. The function j could be adjusted in such a way that the total inequality did not allow the repellent conclusion, described on p. 235 in chapter 7, such that inequality with respect to one equalisandum no matter how large, could be set o against the inequality of an equalisandum which became so large because of the numbers it concerned, no 22
23
h
8.5. SUMMARY
263
excluded and because of that, speci city is excluded. Lack of speci city is of course no surprise. That an aggregation in a separate additive way of indices of equalisanda into one index of inequality is not possible, was to be expected. One serious problematic aspect of the traditional views on equality concerned the assumption that there is just one equalisandum. This equalisandum was of course the supervalue or the common comparing value which incorporated all evaluations. The problems with this supervalue, which is, either not leading to an ordering of the equalisandum, or is of no help in determining the content of equality, infected the measure of inequality.24 Because of these problems with the equalisandum, it seemed that equality was complex. But once multiple equalisanda are allowed, it is possible to formulate a simple measure of inequality.25 A simple separate additive aggregation of these idices however is not to be expected. Because moral universalism is left it is not any longer a priori possible. The only possibility remaining, is that it could happen to be possible. But the argument given by Sen, although given in order to question the possibility of a simple index of inequality, shows that such a simple aggregation does not happen to exist.26
8.5 Summary In this chapter, I argued contrary to Sen's and Temkin's opinion that a simple measure that represents in a distribution problem a complete ordering regarding inequality is possible. Their argument was based mainly on the lack of a measure which incorporated the most basic ideas of inequality. The measures they considered were not satisfactory, they did not represent the moral badness of inequality in a proper way. That led them to the idea matter how small the di erence is between the better-o and the worse o . 24 One set of problems is for example seen in determining freedom in general as the equalisandum. The budgetset is suggested as an index of freedom. A greater budgetset means more freedom. However there are problems with this, because of the possibility of intersection of budgetsets instead of inclusion. See for example gure 3.5 in chapter 3. A problem pointed to by Sen and LeGrand and recently by Vallentyne against the suggestion by van Parijs [Sen, 1973, p. 67 .] [LeGrand, 1991] [Vallentyne, 1996] [Van Parijs, 1995]. 25 See also p.146 chapter 4. 26 It is not excluded that one could construct a complete ordering regarding equality over all situations. One could think of such an ordering as a construction of the orderings of the situations in the distribution problems whereby the weighings of the values of the equalisanda take care for the ordering between these problems. It is also not excluded that such an ordering will be incomplete, but the incompleteness will be due to the general problem of weighing di erent values and will not be located in the complexity of the ideal of equality as was assumed by Temkin en Sen. It is however highly dubious whether such an ordering over all situations makes sense.
264
CHAPTER 8. A SIMPLE MEASURE OF INEQUALITY
that a simple measure was not possible and proposed some arguments to explain this idea. In this chapter, I have shown that a measure is possible. I assumed the existence of a measure and I have shown on the basis of the properties a measure should have, that it is of a separate additive form of which the terms are a function of the complaints of those who are worse o than all could have been regarding the equalisandum. Next, I have shown that, because the equalisandum can be represented on a ratio scale, the function with the complaints as arguments was a simple power function. Of this class, it was possible to point at a particular one to be the index, because it was the only one such that an improvement in the direction of the gradient, the direction of maximal change in inequality, was in the direction of the reference. The index constructed in this chapter is of the form: Xn I(~x) = M( (xref , minfxref ; xig)2 ) i=1
(in which M is a monotone increasing transformation function) I considered the arguments of Sen and Temkin again, to check whether this index could handle the arguments they stated for their view that a simple measure was not possible. It appeared that the index was immune to the arguments of Sen and could better account for the judgements Temkin based his view on, than Temkin's own explanation. Finally, at the end of this chapter it is possible to answer the three questions posed in the introduction: 1. What is the equalisandum? There are several equalisanda instead of one, each with a di erent urgency in a particular problem. 2. Why is equality desirable? Inequality matters because some are being worse o than all could have been simultaneously. Denying equality to have any value at all is only possible by practical solipsists, although not impossible, dicult to defend. 3. What is the ordering according to the ideal of equality? The ordering in a distribution problem is one represented by
Xn
I(~x) = M( (xref , minfxref ; xig)2 ) i=1
(in which M is a monotone increasing function)
8.5. SUMMARY
265
These answers became possible in a moral realistic framework which is characterised further by moral particularism, realistic individualism and moral pluralism, based on the Wittgenstein-Davidson approach on language and interpretation. This framework was argued to be a fruitful alternative to the traditional background of moral universalism, volitional individualism and moral monism for the development of an ideal of equality. In the next chapter, I will illuminate some consequences of the answers for current political issues, such as questions concerning health and the distribution of health care.
266
CHAPTER 8. A SIMPLE MEASURE OF INEQUALITY
8.6 Appendix 1 In the previous chapter, I argued that decomposability was not acceptable, because the inequality in a group which is better o than all could have been, would not add to the inequality. Strictly spoken, additivity is not consistent with the idea of equality as it is developed in this study. But just as in case of independence, one could keep the ideal reference xed for the whole group and admit that indeed inequality in the group of those better o than the reference, does not count, because there is no relevant inequality for that group. So, the reference used in the subgroups for the determination of the inequality within the groups is not above the reference of the whole group. But what should the references be in the subgroups? It should not be such that improving the best-o in such a subgroup would improve the inequality within the subgroup. The most likely reference is the level of the best-o in the subgroup. All in that subgroup could be as well o as the best-o in a subgroup unless the reference of the whole group is lower than the best-o of the subgroup. What about the inequality between the groups? This inequality should be determined with the reference of the whole group, and something which represents how well o the groups are. That could be the level of how well o all could have been in that group. Decomposability might be de ned in a way that is consistent with the idea of why inequality is wrong. The question is whether the measure as it was developed in this chapter is decomposable. The answer is no, simply because the measure is determined up to monotone transformations, consequently the addition of values of inequality has no meaning. The next question would be, is there a monotone increasing transformation such that index is decomposable?27 But also this question is to be answered negatively. Suppose there would be an index which is decomposable. Decompose the whole group in such a way that all strictly worse o than the reference are in one group and the others in another. If the index is decomposable the iso-inequalities in the non-decomposed index should be equal to the iso-inequalities in the decomposed index. If the index is
Xn (x
I(~x) = M(
27
i=1
ref , minfxref ; xig)2 )
This question is more in line with [Shorrocks, 1984].
8.6. APPENDIX 1
267
If x2 is the maximum of those worse o than xref we have also n X I(~x) = M( (x g
i=1
2
Xn
, xi )2) + 0 + M( (xref , x2)2 ) i=1
The iso-inequalities should be equal of these indices. This means that the direction of the gradient should be equal. In other words: (xref , x1) = Mg0 :2:(x2 , x1 ) + 0 (xref , x3) Mg0 :2:(x2 , x3 ) + 0 (M rst derivative with respect to its argument which is Pnig=10g is(xthe 2 2 , xi ) ) This relation should be valid for all xi < x2 . This will be satis ed only if: (xref , xi ) = k (x , x ) 2
i
This will not be true in general. So there is no decomposable index.28
28
Taking the mean as representant will not do either.
268
CHAPTER 8. A SIMPLE MEASURE OF INEQUALITY
Part III
A practical consequence
269
Chapter 9
Equality of health 9.1 Introduction In the previous part, an ideal of equality was articulated that led to a simple measure re ecting the moral badness of inequality. It was based on the idea that it matters that some are worse o than all could have been simultaneously. Although traditionally equality as an ideal was developed within a contract theory with moral universalism, moral monism and volitional individualism as background assumptions, it was shown that it was promising to develop equality within a moral realistic framework with moral particularism, realistic individualism and moral value pluralism, because the three traditional background assumptions are not the proper assumptions for moral theorising in general and because they hindered the development of a comprehensive ideal of equality in particular. In this chapter, some practical consequences of the idea of equality as it was developed in the former chapters are explored so that the theoretical ideas exposed in the previous chapters become clearer. After all let moral philosophy be evaluated by its particular consequences. The eld of health, health care and health care policies is taken as the eld to show some of the consequences. It is a eld where almost everybody has some familiarity with. Furthermore, there are some tendencies in the health care system in the Netherlands that deserve critical discussion from an egalitarian point of view. To mention two of them: the call for a free market of health care products and the development of Dutch College of General Practitioners guidelines (NHG-standaarden) for the treatment by general practitioners [Kool, 1995] [Rutten & Thomas, 1993, p. 2,11].1 As 1
These are not to be seen independent of each other. The Dutch Organisation of
271
272
CHAPTER 9. EQUALITY OF HEALTH
will be shown below the ideal of equality as it was developed in the previous chapters is a reason for being reluctant to embrace these tendencies with enthusiasm. It is not to be expected in this chapter to nd solutions to all distribution problems with respect to health, but it can be expected to nd indications for ways of dealing with these problems. It will extend the debate on the issue beyond the question of `Who determines what?', which is the question that dominates the public debate. In this chapter the equalisandum is taken to be health itself and not health care or access to health care. Ultimately health care matters because it can contribute to health, so I will concentrate mainly on the latter. Distribution problems concerning health and health care are discussed on two levels. There is the macro level, which concerns health care policies directed at groups of people. And there is the micro level concerning choices about particular patients. First, I turn to the running discussion on the decisions on the macro level in section 9.2. It is argued that the common assumptions behind this discussion are not acceptable, mainly because they distract from what the proper view of the ideal of equality aims at. Next, the discussion of the micro level is turned to in section 9.3. The current discussion in this domain is criticised because of its assumption of moral universalism, monism or volitional individualism. After the criticisms on the proposals in the running debate, I indicate in section 9.4, what the ideal of equality can mean in daily practice. The problem of non-transferable goods will appear here in a most distressing way, namely if there is no cure for a serious condition. It is shown that it can be overcome by transforming the hidden interpretative principle of charity on the basis of which we are able to understand each other, into an overt basic moral guide. Finally, I show that the ideal of equality developed in this thesis, in the context of health, ts in nicely with what is held to be the main function of general practitioners.
9.2 Macro level decisions On the macro level, distribution problems concerning health and health care are for example questions on how much to spend on health care and how much on other sectors of society. Should it be 10% of the GNP or should it be less? These questions concern the weighing of di erent values such as education, safety, economic growth and also health. These evaluative problems are not the main concern in this chapter. They are not without relevance, but they do not show clearly the particular problems for the ideal General Practitioners (L.H.V.) started to de ne their products by standards, a condition for an ecient market [L.H.V., 1988]. See p.276 below.
9.2. MACRO LEVEL DECISIONS
273
of equality, they refer to the general problem of evaluating di erent values. I restrict myself to those problems in which equality of health is held to be the main issue. On the macro level two elds of interest can be discerned with respect to the problem of distribution of health and health care: 1. socio-economic related di erences in health, 2. choices between di erent health care policies concerning di erent groups of patients. These are discussed subsequently.
9.2.1 Socio-economic related di erences in health
Socio-economic related di erences in health is a rather popular eld to discuss. It was a main subject of the Annually Meeting of the Royal Dutch Medical Association in 1996. In Britain there was the famous Black Report which showed that lower social economic classes have a higher morbidity and mortality.2 People belonging to those lower classes are less healthy than those in higher social economic classes. On a global scale it is even more distressing, to mention only maternal mortality in labour, which was per 1,000,000 live births in 1980 in Chad: 700; in The Netherlands: 5 [World Bank, 1992], not even mentioning the morbidity of urinary stulas after labour. In The Netherlands di erences related to social economic status are con rmed [Mackenbach, 1994] [Stronks et. al., 1995]. Although in the publications on this subject, theories of justice are mentioned and sometimes discussed, it remains rather dicult to discern what the relation between socio-economic status and health has to do with equality of health.3 Socio-economic status is primarily seen as one of the determinants of health and it is not clear how it is related to evaluating the badness of inequality of health. There seem to be two ways of these socio-economic di erence being related to the evaluation of di erences of health: 1. The relation between socio-economic status and health is not wrong, but the di erences in socio-economic status are wrong. By abolishing these di erences equality will be the result. 2 `..., men and women in occupational class V had a two-and-a-half times greater chance of dying before reaching retirement age than their professional counterparts in occupational class' [Townsend & Davidson, 1982, p. 51] The mortality di erences related to socio-economic status increased in the last 20 years in the U.K.. In the higher classes mortality declined with 36% from 438 to 282 per 100,000; in the lowest class it increased with 2% from 798 to 816 [Medisch Contact, 1997]. 3 See for example [De Jong, 1986] [Stronks & Gunning-Schepers, 1993].
274
CHAPTER 9. EQUALITY OF HEALTH
2. The relation between socio-economic status and health itself is wrong and should be abolished, di erences in socio-economic status should not be related to health. These will be discussed.
Acceptance of the relation between socio-economic status and health
The idea of acceptance of the relation between socio-economic status and health is likely to be inspired by the idea that once there is socio-economic equality, di erences in health are no longer relevant. Health or its determinants are considered to be as ordinary commodities on the market. In a recent study it is argued that because diseases are no longer a public bad; they do not concern public dangers as infections did, diseases are individualised; people can buy individually treatments or insurances on the health care market [Kool, 1995]. Also Arrow explored the idea of a competitive market in health care [Arrow, 1971]. The idea that the relation between socio-economic status and health is acceptable is intelligible within the idea that under the assumption that there is equality of resources, differences in commodities, among them one's health or its in uence on one's functioning, are due to one's own choice on the free competitive market and because of that should not be corrected.4 An idea we met in chapter 3 in discussing the non-envy proposals, notably Dworkin's proposals, according to which distributions are acceptable if no one would be better o if he owned the commodities of another and no one could improve without someone else becoming worse o . It led to the idea of equality of resources. The inequalities are acceptable because they are considered to be due to a free choice on the market. This market is a means of arriving at allocations that are held to be acceptable in these non-envy and resource theories.5 The idea that the relation between socio-economic di erences and health is acceptable, can be considered to be intelligible within these non-envy and resource theories.6 4 The ideas in [Stronks & Gunning-Schepers, 1993] could be interpretedas an example of this line of thought. 5 See [Dworkin, 1993] [Dworkin, 1994] for his explicit ideas on health care. See also [LeGrand, 1991, p. 103-126] for an example. 6 The idea that health care and other health providing products are to be regulated by the market, is a background of what is called the economic valuation of life. In this economic valuationthe value of life is expressed in some good, for examplemoney. Several proposals exist [Jones-Lee, 1976] [Mooney, 1977] [Johansson, 1995]. The main idea is that it is determined how much people are prepared to pay in order to avoid risks, or how much people are prepared to pay for a treatmentof a disease or illness. It is suggested that the value of a statistical life ranges from 0,1 to 15,6 million 1990 dollars [Johansson, 1995,
9.2. MACRO LEVEL DECISIONS
275
In the chapters 3 and 4, it was argued that these non-envy theories and resource theories ran into problems in their attempt to avoid interpersonal comparisons of goods. It was shown that volitional individualism was responsible for these problems. Volitional individualism, in which interpersonal comparisons of what goods do to people are not possible, had to be replaced by realistic individualism, in which such comparisons are possible. This had as a consequence that the solution suggested by the non-envy proposals, an equal start and the rest up to free choice on the free competitive market, became clear to be no longer acceptable, because it could accept inegalitarian results. For example in the case of persons su ering from myopia being in the possession of hearing aids and the persons su ering from loss of hearing being in the possession of glasses, who enjoy going to the theatre, the non-envy proposals could accept inegalitarian allocations because the prices of glasses and hearing aids would not necessarily be just prices.7 Furthermore, in chapter 4 it was shown that the idea behind the nonenvy proposals and Dworkin's resource theory, namely the incorporation of responsibility in the equalisandum, was wrong. Responsibility in itself does not matter. What matters is the value of actions and events in uencing the health risks, i.e. the content of one's own choices. This means that if actions or activities are considered to be valuable, these should be possible to be performed without serious consequences. If someone su ers from a serious health condition due to himself, for instance because he saved a child, he should be helped and compensated for the costs of treatment, despite the fact he brought the serious condition onto himself, because saving a child is a valuable element of life. It seems as if the advocates of incorporating responsibility into the equalisandum have persons in mind who bring diseases onto themselves by performing actions in a silly activity in the hope that others will care for them, for example playing in sports in a dangerous and accident-prone way. Instead of denying them medical help or compensation for the costs of that help, it could be argued they should be helped in overcoming this accident prone behaviour. It could be that they behave dangerously because they have too much con dence in the repairing powers of the art of medicine. It is also possible they have other reasons for liking being injured and next being helped by the medical profession, as for example described in the Munchhausen syndrome. These too should be helped in discovering what they want and how to arrive at that, without using the indirect p. 94]. These proposals based on a health care market will be discussed more extensively below. Its universalistic, monistic and volitional individualistic assumptions will be pointed to. 7 See chapter 3 p. 93.
276
CHAPTER 9. EQUALITY OF HEALTH
way of injury and medical attention. Of course, it could be possible that someone behaves again and again in a reckless way without there being reasons that should be corrected. Compensation could be denied to such a person on grounds of equality because it could be argued that his activity, doing sports in such a reckless way, is less valuable than doing sports in a less reckless way, and enjoying the latter for others is being threatened by his activity. But as was explained above the value of activities is central and not responsibility as it is held to be in Dworkin's resource theory.8 In short, in the previous chapters, notably 3 and 4, it was shown that the theories, in which the relation between socio-economic status and health is intelligible to accept, i.e. the non-envy analysis and resource theories, were not tenable. Beyond the problems with the theories behind the acceptability of the connection between socio-economic status and health there are some others too. Arrow pointed to these problems, which are reasons for being reluctant to the acceptance of a market in health care whether or not mediated by insurance companies [Arrow, 1971]. They concern the marketability of health care products. 1. there is a monopoly of medical knowledge on the side of the health care providers and insurance companies, which precludes a free market, 2. it is not clear what a good medical treatment is, the product is not clearly delineated,9 3. medical action is called for on conditions that are not predictable, and can be of a nature, because of irreversible e ects for example death, such that choices between several ways of treatment are not easy to be made. These arguments on their own can be rebutted. For example one can turn the rst argument into an argument against monopoly in favour of a free market in which there is more competition.10 But the other problems See chapter 4 p. 146 . The guidelines of the N.H.G. are an attempt to repair for this lack [Rutten & Thomas, 1993]. 10 Competition on the other hand will have some disadvantages too. There is for example the problem of skimming of the market described clearly by Milner [Milner, 1980]. This will be acute for health care distribution, which concerns highly particularised demands. The following example based on an example due to Faulhaber can be illuminating [Moulin, 1995]. Suppose three consumers, three eastern northern islands in the Netherlands among them Schiermonnikoog, demand a particular good, say a helicopter ambulance service with the following cost table because of geographical reasons: units 1 2 3 Suppose that the customers of the services are prepared to pay 7 costs 6 9 14 8 9
9.2. MACRO LEVEL DECISIONS
277
will become more serious. How to delineate the products, how to choose between the several options? That similar problems are also valid for many other sectors, for example the sector of legal advice, or more generally, sectors of important services by specialists, is not an argument for introducing the free market as a solution for health care but should be seen as a criticism on the market for these other sectors. The above mentioned problems concerning the acceptance of the relation between socio-economic status and health can be stated in still another way. With the suggestion that equalising the socio-economic status will be sucient for equality, it is neglected that we are concerned about equality of health and not with equality of access to health care. Even if socio-economic di erences are non-existent, there will still be di erences in health. Because in the ideal of equality of health it matters that some people are less healthy than all could be, the socio-economic equalisation cannot be its main goal. So, let me turn to the other view in which socio-economic di erences are considered to be important for the evaluation of di erences of health.
Rejection of the relation between socio-economic status and health
The second view in which socio-economic di erences matter in evaluating inequality in health is based on the idea that the relation between socioeconomic status and health should be rejected, or that it should be changed. Two ideas can be distinguished: 1. there should be no relation at all between socio-economic di erences and health, 2. di erences with respect to health should be compensated for by socioeconomic di erences, socio-economic status should not be correlated positively with health but negatively. The former view holding that socio-economic status should not have any in uence on health, and that the populations of di erent social economic classes should not be possible to be di erentiated with respect to health, morbidity and mortality, can be seen as advocated by Walzer [Walzer, 1983]. for one unit of service. Than the surplus if all three want one unit is 7*3-14=7. But any two consumers can generate a surplus of 7*2-9=5 which is per consumer 25 instead of 73 . If one rm services all three customers then it should demand a price of at least 4.66. If a rm o ers his service to two it can o er the service for 4.5. So by the free market the rst option is not the most prosperous one to the rms and will be removed by the market, an unstable situation will be the result. So free entrance and competition will cause instability and ineciency.
CHAPTER 9. EQUALITY OF HEALTH
278
He states: No social good x should be distributed to men and women who possess some other good y merely because they posses y and without regard to the meaning of x. [Walzer, 1983, p. 20] This view advocated by Walzer, with whose views I sympathise, should be looked at critically. The quotation states that a distribution should not be based on irrelevant characteristics. Walzer's idea about what is held to be irrelevant is based on so-called social meanings. But this should be criticised because meanings determined by a society are, if they are not charged by conservativism, at least subject to the charge of imperialism as was explained in chapter 5. It was explained that the Wittgenstein's conception of rules, or forms of life, of which Walzer's view is an example, lead to relativism and subjectivism hence to imperialism by determining in a relativistic way what a meaning is.11 More important than the above critical remark on social meaning is that it should be realised that equality does not concern statistical correlations between socio-economic status and health, it does not concern the possibility of di erentiating the populations of di erent socio-economic status with respect to morbidity and mortality. An egalitarian in the eld of health is concerned about some persons being worse o regarding health than all could have been. So, until now, we have not seen a proper reason to turn one's attention to socio-economic related di erences in health. Let me look at the other view. The second view holding that the correlation between socio-economic status and health should be reversed could be considered to be advocated by Lucas [Lucas, 1965] [Lucas, 1977].12 Lucas states that there should be admitted to be a plurality of equalisanda, which makes it possible that inequality with respect to one equalisandum could be cancelled by inequality with respect to another equalisandum. This view should be criticised too. First, it was made clear that inequality is not to be cancelled or compensated. Lack of some good can be compensated, compensations for inequality make no sense.13 What is wrong about inequality is not some feeling or dissatisfaction concerning equality in itself, but being worse o than all could have been. Second, the view would also mean that the lower the socio-economic status, the better should be the health. Turning to two groups of persons su ering the same disease, the socio-economic worse-o should get a better medical treatment 11 12 13
See chapter 5 p. 156 . See chapter 6 p. 187. See chapter 6 p. 191.
9.2. MACRO LEVEL DECISIONS
279
in a way such that their health would improve more than of those of the socio-economic better o group. I cannot but consider this an odd result. The view based on Lucas' ideas, on the importance of socio-economic related di erences in health does not help us in determining egalitarian concerns with respect to health. Summarising, neither accepting the relation of socio-economic di erences and health, nor the rejection of it, form proper reasons to attend to this relation. Contrary to what is mainly thought about inequality in health, looking at socio-economic related health di erences is not the most helpful way to look from an egalitarian point of view at di erences in health. Turning to socio-economic related di erences in health is distracting from the main issue, namely, that some are worse o than all could have been. I will not deny that socio-economic related di erences in health should not be realised to exist, after all social economic status happens to be a determinant of health. It should not be denied, that knowledge can be of help, but it does not concern the main issue of equality of health. In section 9.4 below, I indicate how it can be of help in an egalitarian perspective. Let me turn now to the other macro level discussion, namely choices between health care policies.
9.2.2 Choices in health care policies
On the macro level, decisions are made concerning the allocation of resources to di erent activities in health care, for example: how much should be given to heart surgery, or to the treatment of aids, or the prevention of aids, or activities in sports medicine? These choices concern distribution issues on groups of patients. In order to handle these problems more clearly, a cost-bene t analysis or decision theoretic analysis, is suggested. This method is developed in several ways, one is in the direction of direct money measures, another as so-called quality adjusted life years, and still another, as a correction of quality adjusted life years, namely as healthy years equivalents. In a cost-bene t analysis it is spelled out what is the gain and the loss of a particular policy. On the macro level it concerns groups of people and that deserves suspicion. As Broome pointed out against the costbene t approach, the bene t or cost is given in terms of a probability of an e ect or outcome of a treatment [Broome, 1978]. This probability can hide a more precise evaluation, if it could be known who will su er the cost and who will get the bene t of a policy. This knowledge will change the evaluation in a dramatic way for some people. There is a di erence between a policy presented as a choice between less or more probability of success of a treatment, or as a choice between certainly no e ect or
280
CHAPTER 9. EQUALITY OF HEALTH
a reasonable probability of a cure. It could be that for some people the latter description is more appropriate, but remain hidden in the general description of e ects of policies because they are about groups of patients in which the great di erences within these groups remain hidden. Not all HIV infected patients are similar and not all persons su ering a sport injury su er in a similar way. One should not con ne oneself to groups of persons which are not homogeneous. It is not correct to present a choice as a choice between more or less probability while more information can be known. Equality does not concern equality between groups, the main concern is that some individuals are worse o than all could have been. Macro level problems should be considered as aggregates of micro level problems. In the process of aggregation important information will be hidden, or lost, or deliberately not taken into account. But it should be realised that the main problem is the micro level decision. Once that is solved, the construction of the macro level decision can be made on the basis of considerations what information has to be taken into account and what to consider as being less relevant.14 The macro level decisions on policies refer to, or are based on, micro level issues. These are turned to in the next section. Summarising, on the macro level two issues are seen to be discussed, socio-economic related di erences in health and choices on health care policies concerning groups of patients. The rst was not directly relevant for an egalitarian view on health. Equality does not concern the absence of socioeconomic related di erences in health. It does not matter whether there are no such di erences either because there are no socio-economic di erences, or whether the socio-economic groups cannot be di erentiated with respect to mortality or morbidity. Equality is concerned with some persons being worse o than all could have been. Absence of socio-economic related di erences of health will not imply that there are no persons worse o regarding health than all could have been. The second issue concerning health care policies was argued to be referring to the micro level decisions. The di erences in health within groups should not remain hidden. The macro level debate is not the domain of getting a clear view of what equality with respect to health means.
9.3 Micro level decisions On the micro level, choices and decisions have to be made about which particular patient has to be treated given the scarcity of resources. A 14 This way of looking remains particularistic even if one turns nally to groups on the macro level. Moral unversalism is not assumed because the reasons for choice are the particular problems with particular people.
9.3. MICRO LEVEL DECISIONS
281
paradigmatic example is the following. Suppose there is one proper therapy for a disease for which there are two doctors who can provide this therapy. The question is: `Who from the following list su ering the disease should be treated, or should be treated rst?': a woman without a husband and children, a woman with four children dependent on her, one of the two physicians, a severely multiple handicapped child, a child with only a prospect of living one year, a man who brought the condition on himself. Such examples are used in trying to develop criteria on which decisions should be based. One criterion could be eciency in treatment capacity, meaning that the physician should be treated rst, because that would expand the possibilities for treatment. One could hold in general that those who enable more possibilities for treatment by their contribution to the social good, should be treated rst, or even more general, those who mean a lot to others, for example either by caring for dependants or by contributing to economic growth. This criterion would be unacceptable because for example the severely handicapped people, would probably lose on all aspects. Another criterion could be responsibility, meaning that the man who brought the condition by his own actions on himself is excluded from treatment. As was argued in the chapters above, this criterion too is not without its problems because there are many di erent ways in which conditions of health are consequences of actions. This was made clear by the example due to Scanlon of cleaning a polluted area which was a dangerous activity and because of that people were warned not to leave their houses during the cleaning activity, but despite the warning some did leave their house.15 It was shown that the evaluation of responsibility depends primarily on the value of the action undertaken and not on responsibility. It was shown that the criterion of responsibility was not acceptable. Instead of eciency and responsibility, the criterion rst in, rst helped, as was for example the case in Bethesda, is proposed.16 By following this criterion those most in need will not be helped. Only those being less ill, or having already some compensation in the form of help from family or 15 16
See chapter 4 p. 149. See John 5:1-18.
282
CHAPTER 9. EQUALITY OF HEALTH
friends, are capable of reaching the healing well in Bethesda. Should we turn then to a lottery assuming there are no reasons to decide upon? This is not reasonable either. A general problem underlying the criteria of selecting patients is that they are meant to be criteria that should be used as criteria in all situations. The generalised versions turn out to pose the problems. Sometimes a lottery is useful indeed, if there is no other reason to turn to [Goodwin, 1992]. Similarly sometimes the most ecient solution should be chosen, and a mostly needed physician should be treated rst. But if this becomes a rule such that others are excluded, a serious problem arises. What is asked for, are particular reasons in particular situations.
9.3.1 Maximising qalys, hyes and lyars
One way to arrive at a decision in particular circumstances is on decision theoretic grounds. On the basis of decision theory, several instruments are developed: 1. quality adjusted life years (qalys), 2. healthy years equivalents (hyes), 3. money values of health risks, based on willingness to be compensated by or pay for changes in health risks (they can be called life year adjusted repayments, lyars). These approaches were mentioned in the former section as cost-bene t analysis about policies, here they will be discussed as ways of arriving at particular decisions in particular situations. The quality adjusted life years, qalys, were introduced as a correction of the idea that in evaluating medical treatments only the number of years of individuals count. The correction consists of multiplyingthe life expectancy by an adjustment factor for the quality of life. This factor is determined by questionnaires in which an evaluation is asked from subjects about living in a particular condition. In some questionnaires it is assumed that persons qaly maximisers, which means they will maximimise q = PareTt=1discounted dt,1q(ct) where d 1 is a discount factor in time, `later matters not more than now' and t is the unit of time [Johansson, 1995, p. 154].The qualities are normalised so that the worst condition cw has quality q(cw ) = 0, and the best condition, being healthy ch , has quality q(ch ) = 1. This normalisation re ects that the relevant way of measuring quality of life is on a ratio scale. In order to construct this ratio scale it is sometimes asked `How many times more ill is a person in state c1 compared to state c2?' The
9.3. MICRO LEVEL DECISIONS
283
subjects should assume that the descriptions of the states relate to a young to middle aged adult, and all states have the same prognosis and could be cured if the patient is treated and if left untreated, the patient's condition would remain static until some other condition supervenes. If a condition c has a quality such that q(c) = 6 and the time to live in this condition is 10 years and d = 1 the quality adjusted life years are 0; 6 10 = 6 [Kind et. al., 1990, p. 60]. This idea of quality adjusted life years is criticised by Mehrez and Gafni because it could happen that the maximum of quality adjusted life years is not similar to the maximum utility, if utility is interpreted as a representation of choices [Mehrez & Gafni, 1989] [Mehrez & Gafni, 1991]. They introduced healthy years equivalents (hyes) which is the number of years in a healthy condition, T , such that its utility is equivalent to the life to be evaluated, U(chT ) = U(cT ). Because of this de nition there cannot be a discrepancy between a choice based on utility and on healthy equivalent years. In this reasoning, it is assumed that the utility of each healthy life year is positive, i.e. the more, the better. An example of discrepancy between hyes and qalys is the following. Suppose a person faces an option: live 5 years in a state c1 and then die, and in the other option he will live 10 years in a state c2 less healthy than c1, so q(c1) > q(c2 ). Assume that the utility of living 5 years in state c1 followed by death U(c1 ; 5) = 0; 65 and living 10 years in state c2 and then die U(c2 ; 10) = 0; 5. The person's choice would be the former option, live 5 years in a healthy state. Using the qaly method would lead to the following result. The former option living 5 years and then die would be worth 0; 65 5 = 3; 25 quality adjusted life years. The latter option would be worth 0; 5 10 = 5 quality adjusted life years. So, the latter would be preferred on the basis of quality adjusted life years while the healthy years equivalent following direct the preferences, would show the former to be preferred. This is a serious disadvantage for the quality adjusted life years, they lost their proper connection with the decision theoretic framework. So, hyes are to be preferred to qalys because the former represents in a better way the advantages of treatments to people. Both, qalys and hyes are criticised by Johansson because they neglect that utility options that concern health risks, should incorporate their nancial consequences too. How a state is to be evaluated by a person is also dependent on his income or in other words his socio-economic position in that state [Johansson, 1995, p. 157]. Using this information we can arrive at what could be called life year adjusted repayments (lyars). Or more often called compensation variation, meaning the amount of money CV such that w(a , CV; c2 ) = w(a; c1) the compensation for losing some of the health from going from condition c1 to a worse condition of health c2 ,
284
CHAPTER 9. EQUALITY OF HEALTH
or alternatively, equivalent variation, the amount of money EV such that w(a+EV; c1 ) = w(a; c2) if the change from the state c1 to a better one c2 is foregone. This method of evaluating health status, or changes in health status, is more according to the complexities of choice than the healthy years equivalent methodology in which the question of socio-economic status is hidden. But incorporating the socio-economic consequences in health care decisions on its turn is criticised by for example Culyer, because medical treatments should not be evaluated in terms of some overall measure as money or income, but by what he calls some characteristics particularly relevant for representing health, in order to avoid what Sen calls welfarism [Culyer, 1990] [Sen, 1979b]. And so he argues against the lyars but for qalys. The former incorporate elements which have nothing to do with health and health care. How should these di erent measures or indices be used? The above instruments were presented as a method of help in order to arrive at a decision concerning a selection problem. On the basis of the measures qalys, hyes, or lyars, decisions are proposed. The main approach to the question: `What to do?' is modelled as the question: `What treatment should be chosen?' by incorporating the characteristics of patients in the description of the treatments, or treatment protocols. In the case described above one treatment which could be chosen could be: if the patient is a single woman do 1, if the patient is a woman with four dependent children dependent on her do 2 etc. An alternative treatment protocol could be: if the patient is a single woman do 2, if the patient is a woman with four dependent children dependent on her do 1 etc. The options are described like a treatment, which is for example proposed for uncomplicated urinary tract infections: if the patient is female and not pregnant then once a day 300 mg trimethoprim for three days and if the patient is pregnant three times a day 375 mg amoxicilline for 3 till 7 days.17 A treatment or treatment protocol, or Dutch College of General Practitioners guideline,which has to be compared to other treatments or protocols. The selection problem is now formulated as: `Which treatment to choose?' The answer to the question: `What treatment should be chosen?' is of course: `The one with the highest pay o .' with the pay o in terms of the goal of health care represented by the indices qalys, hyes and lyars. But this way of modelling the problem hides an important aspect of the selection problem. One should distinguish between two di erent questions: 1. Which treatment would be best for this patient? 17
See [Rutten & Thomas, 1993, p. 308].
9.3. MICRO LEVEL DECISIONS
285
2. Which patient should receive this treatment? These are two di erent selection problems: one concerns a choice between treatments, the other a choice between patients. The distinction can be hidden, as is shown above by describing the selection problem in general terms in which the characteristics of the patients are incorporated in the action or the treatment, the rst question becomes the model of the problem. The latter question is covered up or remains hidden, and consequently egalitarian concerns will be neglected. Although it can be admitted that egalitarian concerns are not always the most important considerations, they deserve some place in the selection of patients. This means that the indices mentioned, which are possibly appropriate for answering the question what treatment a patient should receive, are not directly appropriate for the selection of patients. As Broome states the indices, such as qalys, hyes and lyars, are not developed in order to arrive at fair solutions [Broome, 1988]. So, maximisingqalys, hyes or lyars is not the proper solution to the selection problem of patients.
9.3.2 Are qalys, hyes and lyars proper equalisanda?
Although the indices are not meant to answer the problem of the selection of patients, they could be relevant in another way, namely as the equalisandum. Equality of health could be interpreted in such a way that it means that in the ideal of equality the indices should be equal for all. The answer to the selection problem would be from an egalitarian point of view: the patient whose treatment will diminish the inequality most. Is this interpretation of equality in health with these indices as equalisanda, be an acceptable one? I argue it is not. Apart from the question to which index we should turn, there are some arguments against taking these indices as the equalisandum because all of them have unacceptable assumptions. They all assume moral universalism in the sense of additivity, or also called separability, of life years, money and quality of life. They tend to monism. And most important they are likely to assume elements of volitional individualism. These three assumptions were argued in the rst part of this study to be highly dubious. It was argued that moral universalism is better to be replaced by moral particularism, moral monism by moral pluralism and volitional individualism by realistic individualism. Moral universalism in the sense that if in one situation more of one value is better it is also in any other situation, known as separability or additivity, is an assumption of the decision theoretic framework of the indices. It is assumed that more life years is better irrespective of the quality. The same
286
CHAPTER 9. EQUALITY OF HEALTH
is held to be valid for money or other valuable aspects of life. But this idea was argued to be wrong in chapter 2, on moral universalism, and chapter 4, on monism. If something is valuable it does not mean that more of it is always better irrespective of the other values. This is also valid for the domain of health and health care. The quality of life is not always independent from the length of life for example, if one should nish a project or could repair a misunderstanding before dying. Two hours will be an improvement of one's life if one has a chance of clearing up misunderstandings, or nish something which is less valuable without this nal contribution. It could be a reason to endure pain and refuse analgesics that disturb consciousness and facilitate death. But also, it is not necessarily the case that dying two hours later is better irrespective of one's quality of life. If all is said and done, or there is no chance of clearing up the misunderstandings, these two hours are not better. More time to live can block the quality of one's life in particular circumstances. But it can be valuable if there is for example the possibility of repairing something wrong because some of one's intimi can still arrive. Receiving more or less life years is not better irrespective of one's quality, or what one takes to be valuable. Apart from this example there are many more examples available to illustrate that additivity is not valid. Receiving more money is not always better. One's life could be upset in a disastrous way by for example losing one's close friends. Of course it is not always disastrous but several stories illustrating these possibilities are intelligible. For example in the story of Roald Dahl Hendrik Meier he makes clear that being rich can make one become more concerned with losing one's property. The story also shows that it is possible that someone changes his view of life by gaining more money or by what enables him to make more money. Another example against additivity of for example ability to see is La Symphonie pastorale by Andre Gide. Quality of life for Gertrude did not seem to be improved by the succesful operation of her eyes. This non-additivity is so clear that it is a miracle that additivity is still taken to be assumed in a decision theoretic framework. The idea of separability or additivity can be defended against the charges made by the examples described above by redescribing the situation in such a way that separability is plausible. But it will lead in the end to descriptions in which what is valuable is relativised to particular situations. Although it seems to be an excellent defence, it undermines the whole idea in decision theory which is based on the idea that a consequence can occur in di erent situations. This is precluded in the redescriptions if consequences are described as for example, gain of life years in such and such a situation instead of life years simpliciter, or as receiving money in such and
9.3. MICRO LEVEL DECISIONS
287
such a situation instead of gaining money simpliciter. One will end up with descriptions of the action or option, together with the consequences, in a way such that they become one whole object as elements of a preference relation without any (desiderative) structure. Lack of this desiderative structure was seen to be reason for Pettit to criticise decision theory [Pettit, 1989].18'19 The assumption of separability in itself is defensible but at the cost of another basic assumption about the structure in decision problems. This link between assumptions is easily overlooked and could explain why additivity is so stubbornly assumed in decision theory. But whatever the explanation of using this additivity, it is simply not valid. As all three indices, qalys, hyes and lyars are based on this additivity, these indices are not valid representations of the meaning of the value of a life. Apart from the issue of additivity the indices have another characteristic which is criticised in the foregoing chapters namely monism. It is assumed that there is one good that matters, namely the good represented by the index. All what matters in health and health care is assumed to be possible to be expressed in time i.e. years of life, or money. But actually there is no essential way to answer the question: `What it is the currency of health?'. The alternative to monism is that one abolishes this question as irrelevant. Whether one takes a time measure or a money measure is not of any relevance at all, one could even take roses or chickens or any other good as units of the index of health. Within a pluralist framework this would be acceptable, the index is interpreted as a representation of a multidimensional good without claiming the units of the index representing the essential value of health or health care. The indices are interpreted as representations of individual choices and not the value of health or health care per se. In this way monism could be avoided. The element of choice of individuals can be seen as a way of expressing autonomy against paternalism of the idea `doctor knows best' [Mooney, 1977]. In this view, the health status is subjectively judged by each person him- or herself. But apart from the peculiarity of the health status being by de nition determined by subjective evaluations, it also brings with it the See chapter 4 p.129 in which is referred to this lack of structure. Beyond the criticism based on lack of desiderative structure, there is another criticism. For example one basic structural assumption of decision theory, the distinction between actions and consequences, would be blurred with the relativised redescriptions. It is a distinction made in the assumption P3 in [Savage, 1972, p. 25]. Consequently, by redescribing in the way indicated in the main text, one of the axioms in decision theory is not valid any longer. It seems to be the case that one can save additivity by rejecting another axiom the idea of the strict distinction between acts and consequences. See for another defence of additivity Broome who gives up the rectangular assumption in [Broome, 1991]. 18 19
288
CHAPTER 9. EQUALITY OF HEALTH
problem of interpersonal comparability. How should individual choices be compared to each other? Does seven qalys mean the same for each person? One way of answering these problems is turning to the idea that the choices are not subjective by de nition but approximations of an objective idea of health. This is the assumption on behalf of which the view held by several authors who consider a health questionnaire or a poll relevant, becomes intelligible. This approach is taken by for example van Praag, and Culyer [Culyer, 1990] [Van Praag, 1993]. In this approach there is some discussion on who should be interviewed, patients, potential patients, or the professionals and in case of the latter who of the professionals, the physicians or the nurses. The evaluations of the latter tend to be more in accord with those of patients than those of the physicians [Kind et. al., 1990, p. 61]. The purpose of all questionnaires is to arrive at an evaluation close to the truth and the answers should be as unbiased as possible. There is however a problem with the view that what health means can be distilled from health questionnaires. What exactly is the relation between answers to questionnaires representing choices or hypothetical choices of persons whether or not professionals, and health status? Are they constitutive of what health means as in the autonomy approach seems to be assumed, or are they conceptually independent as van Praag and Culyer seem to assume. As argued, the former leads to the problem of interpersonal comparisons. But the latter is not without its problems either. If independence is assumed, then the answers to the questionnaires are not constitutive of what health means, it is just the result of a test, the result of an experiment. They are like symptoms of a disease. The expressions are seen as some evidence for the judgements concerning health. But these questionnaires are not necessarily valid approaches. It might be that instead of vocal expressions other indicators are more valid, for example faecal expressions. But could this be acceptable? Is there no conceptual relation between what counts as health and what people prefer or what they see for themselves as healthy? What people consider as health for themselves is not independent from what health means [Nussbaum, 1993, p. 13-47]. It is not acceptable that what people prefer and think for themselves has nothing to do with health. But it is not clear that their ideas are constitutive of it, as is likely to be assumed. The polls of patients, potential patients, professionals, may be seen as indications and may form a starting point but they are merely that and should not count as a nal judgement. It can function as a rst word, but not as a nal word. What a nal word should be is not expected to be procedurally accessible. It is, following the Wittgenstein-Davidson approach to language and interpretation, an element of the common background and belongs to the domain of particular judgements. The idea that it should be
9.3. MICRO LEVEL DECISIONS
289
determined by a procedure is a trace of universalism, which was also seen in chapter 3 to be an underlying problem of interpersonal comparability.20 Once this trace is left, interpersonal comparisons are no longer a problem because the basis of what matters in health is not constituted by a choice, either a choice by normal people or by health care professionals, but it is just a fact belonging to the common objective world. Health questionnaires of people or professionals can be rst approximations but cannot be seen as a nal word. Because qalys, hyes and lyars are based on representations of choices, or hypothetical choices, they should not be taken to be the proper equalisanda. They are merely indications and not more than rst approximations. Only particular judgements on health should count. What is important concerning the allocation regarding health is how ill or how worse o a person is with regarding health. This is essentially a particular judgement, which is not to be constituted by a procedure that was assumed to be in the construction of qalys, hyes and lyars. What matters is not the amount of qalys, hyes or lyars, but how serious one's condition of health is. The worse, the more should be allocated to the person su ering this condition. Equality of health demands that rst the most seriously ill or handicapped person is helped, or at least given so much that he is as well o as the second worse-o , next this second worse-o should be helped, etc. This does not mean that the measure of inequality does follow the maximin principle. The index is Xn I(~x) = M( (xref , minfxref ; xig)2 ) i=1
It is the concavity of the inequality index which asks for this.21 That this function is the sum of squares is less important than its concavity, because the precise function will be drowned in the imprecision of the judgements concerning the equalisandum i.e. health. How this measure of inequality a ects the meaning of equality in health care is the subject of the remaining sections. Summarising, concerning the micro level discussions several criteria were suggested in order to select people for treatments. It turned out that they were unacceptable because of their universalistic nature. A method based on decision theory was proposed as a particularistic solution, but it turned out that these methods also su ered from a moral universalistic assumption in the form of separability, and its presumed monism. And if it it did not presume monism it su ered from problems similar to those of volitional individualism concerning interpersonal comparisons. Health judge20 21
See chapter 3 p.104. See p. 256 in chapter 8
290
CHAPTER 9. EQUALITY OF HEALTH
ments are not to be seen as constituted by subjective judgements. Neither should it be assumed that there is no relation at all between what people think for themselves what matters in health and what health means. The Wittgenstein-Davidson approach to language and interpretation can account for a conceptual relation between what people think for themselves and what health means. It is accepted that judgements of health are particularistic judgements, which are not to be determined by a procedure. They are the common background, which explains why there is a relation between what people believe health is and what health is. The polls do not deliver the nal answers but should function as merely a rst approximation.
9.4 Equality of health in medical practice In the previous sections it was argued that what matters for equality in health, is care according to need, i.e. rst the most seriously ill or handicapped. A principle subscribed to by practising physicians as for example general practitioners. But what if the disease someone is su ering from cannot be treated once and for all, what if there is no cure and the prognosis is infaust? It should be remembered that with regard to equality one is not concerned with goods themselves but with what goods do to people and what a particular condition means to people. This suggests that if there is no cure for a disease one should look for ways of compensating the de ciency and for means to alleviate the impairment or disadvantage concerning a particular function. The key for compensations of lack of health are these functions as contexts of symptoms and diseases. Such functions are considered to be valuable for human life. Moral particularism and pluralism will acknowledge that there is a plurality of them. For example: the ability to enjoy a beautiful mathematical proof, enjoy a visit to the theatre, enjoy mobility, or giving sense and meaning to one's life in one's family or society. This way of looking at diseases in which one concentrates on the meaning of complaints, diseases or handicaps to patients, is considered to be an important aspect of working as a general practitioner, who is supposed to give personal continuous integral care [Hodgkin, 78] [L.H.V., 1985, p. III.11]. In a dialogue or interview, traditionally called anamnesis, also the meaning of complaints should become clear [Van Es, 1984]. This way of looking at problems of health, con rms the idea that there is a plurality of functions to look at in discovering what bothers a patient. Health should be considered to be relevant in a multidimensional functional space.
9.4. EQUALITY OF HEALTH IN MEDICAL PRACTICE
291
Attempts to construct one central idea of health measured by qalys, hyes and lyars failed because of the problems discussed in the previous section. The subjective element of choice in these measures was a reason to restrict the relevance of the health questionnaires. Turning to instruments that do not su er from being representations of subjective choices in order to measure the impairment of functions, for example the Nottingham Health Pro le or the Sickness Impact Pro le, will not do either. Many actions performed by general practitioners do not improve health as measured by these questionnaires [Schayck et. al., 1995] [Jacobs & Touw-Otten 1995] [Sobel & Winters, 1996, p. 75 .]. Many treatments that are considered to be worthwhile by patients and general practitioners, will not show any e ect on measured health. Continuous integral personal care demands looking at the multiple aspects of health. Covering them by one index will result in the so-called problem of `bandwidth versus delity' [Streiner & Norman, 1995, p. 23,24]. This problem is acute in general practice, because of comorbidity, complaints are not independent from each other. Arthritic pain may mask shortness of breath. So, relieving pain may result in complaints of shortness of breath.22 23 That unidimensional measures do not suce can be illustrated further by the fact that lowering the budget in NHS did not result in a lowering of health status measured by gross methods such as mortality etc. Here again additivity, or separability is a problem. Looking for example at the Dutch Sickness Impact Pro le for example, a fragile patient su ering a cardio-vascular problem resulting in orthostatic hypotension that lead to impossibility of sitting and fainting while sitting, but now improving and able to sit in a wheelchair realising more and more he is dependent now on help of others and disturbed by the fact he cannot do as he likes, would have a higher score on the Dutch SIP, although in fact he is in better health than in the condition by which he was forced to stay in bed. On the category `sleep and rest', a change from laying in bed (1) to sitting (2) would not mean an improvement. On the category `emotional behaviour', his realising being dependent on others makes him feel a burden to others feel and being worthless (1) and would mean an increase of the index by 5*1 points. On the category of `body care and movement' there is an improvement because the patient is no longer most of the day in bed (11) meaning a decrease with 20*1 points. On the category `mobility' there is an improvement because the patient is no longer most in bed (4) meaning a decrease of 7.2*1 points. On the category `social interaction' there is a change by realising his worthlessness such that he is irritable (4), meaning an increase of 14.5*1 points. On the category `ambulation' the patient now arms he is moving in a wheel chair (1), (while laying in bed he did not) meaning an increase of 8.4*1 points. On the category `alertness behaviour' he now arms he does not nish the tasks he started (4), (while laying in bed he did not even started), meaning an increase of 7.7*1 points. On the other categories there are no changes. This would mean an increase of 8.5 points on the Dutch SIP which represents a worsening of condition, while in fact there is an improvement. The numbers in parenthesis refer to the number of the items in the Dutch SIP. See for the Dutch SIP [Jacobs, 1993]. 22 23
292
CHAPTER 9. EQUALITY OF HEALTH
[NRC handelsblad, 9-11-1996]. The relevance of individual treatment is also questioned, or even shown to contribute little to improvement of health by McKeown [McKeown, 1979].24 The view in which health is considered to be one unidimensional aspect cannot be held to be the leading idea among health care professionals. Accepting the multidimensionality in a particularistic framework is more appropriate. Some of the dimensions of health are: absence from pain, life expectancy, absence from shortness of breath. In short, complaints which could be alleviated by health care. It should be realised that such complaints are also embedded in functions. This is acknowledged in the education of general practitioners. They are trained to ask for the meaning of symptoms for patients. It is part of what is called `working methodical'. One should not treat laboratory results but patients. One should not treat diseases but people. One should turn to functions in order to evaluate whether a treatment of the disease or symptoms is to be advised or that a treatment will have too much disadvantages. These elements of health and health care are considered to be one of the basic ideas of the way of working of general practitioners [Van Es, 1984]. In what way is the ideal of equality relevant in this multidimensional context of health and health care?
9.4.1 What if there is no cure?
A rst reaction could be that the ideal of equality is not relevant at all because, what to say about severely handicapped people whose handicap cannot be cured, what should be redistributed in order to alleviated their condition? Whatever one redistributes their handicaps will not be changed. Consequently, the reference will be of a level such that the idea of equality loses any sense because nobody is below this level of the reference. The same could be pointed to if one turns for example one's attention to children of two years old who su er an incurable fatal disease. Whatever one redistributes it will not change their condition. The reference as de ned by the ideal of equality will turn out to indicate a level such that the ideal of equality loses its relevance. In case of particular restricted handicaps such as loss of vision one could turn to some form of compensation for example regarding the capacity to earn some income for oneself.25 But in case of multiple severe handicaps or a fatal disease with an infaust prognosis 24 The remarks in the main text would lead to abolishing health care if the proposals of the Dutch Committee `Keuzen in de zorg' are followed. In the so-called funnel of Dunning all treatments that are not e ective are rejected as treatments to be collectively paid for [Dunning, 1991]. The main text indicates that individual health care would not pass this lter of the funnel. 25 See chapter 6 p. 209.
9.4. EQUALITY OF HEALTH IN MEDICAL PRACTICE
293
regarding to what could there be compensation? These examples make a serious threat to the ideal of equality as it was developed here. This problem of compensations of severely handicapped people and fatal diseases in young children is not merely a problem particular for the ideal of equality as it was developed here. Severely handicaps form problems for several conceptions of the ideal of equality, for example: equality of welfare, the maximin criterion and the idea of van Parijs. Van Parijs de nes a just distribution to be the one in which all receive an equal basic income which is adjusted for handicaps in such a way that there is at least one person who would not mind change place with the handicapped person.26 The height of the basic income is determined by what makes this distribution possible [Van Parijs, 1995]. Consequently, if there are handicapped persons for whom compensation is not possible, the basic income will become very low. That there are severely handicapped people does form a problem for other proposals too, but there is a di erence with the approach developed in this study. The conceptions mentioned, showed equality of health to be not feasible, while the ideal of equality itself remains intact. For the ideal of equality as it is developed here, it is not just feasibility which is at stake but the ideal of equality itself. Because the impossibility of compensation lowers the reference in such a way that the whole idea of equality loses its meaning, nobody will be below the reference and consequently there is no relevant inequality. The conclusion that there is no inequality with respect to health because of severe and multiple handicaps and fatal illnesses of children, can be avoided in two ways. The rst holds that every disease or handicap is essentially avoidable. Health is possible for all. One is not ill because another is healthy. So the reference could be health for all. This response will not be followed here, because it refers to health apart from other functions and isolates health from distribution e ects on functioning. The reference is determined without any reference to redistribution. It will lead to a view like utilitarianism because all are less healthy than all could possibly be. So improving health also of the most healthy ones, would diminish equality. This is not according to the intuitions on equality. The other response is taking the idea of compensation more serious and give more attention to what is still possible to compensate for. It should be realised that there are important similarities between `healthy' people and handicapped people or seriously ill children. For example pain should be relieved for both. A handicapped person can su er from diseases any other can su er from too, which requires of course examination and treatment, possibly with even a higher priority because it hinders the functioning of a 26
See also chapter 3 p. 91 for the discussion on undominated diversity.
294
CHAPTER 9. EQUALITY OF HEALTH
handicapped more than of another.27 But apart from su ering from pain and other discomforting conditions, there are some functions which remain important and should be available for all. One is for example, recognition by others, expressing oneself whether or not directly or mediated by others for example parents. Seriously ill children should be given a voice, a way of expressing themselves and their experience should be let worthwhile. This means support for parents which is more than just the medical care. It requires sharing of experiences. In this way compensation could be possible. Although this idea of compensation in case there is no cure seems a clear indisputable requirement, it is not satis ed. Handicapped people and their relatives can tell in a convincing way that these conditions are not yet satis ed. The answer given above, namely considering more ways of compensating, should not be considered to be a return to welfare in general. Its abstract character might suggest it is. But in contrast to welfare in general, the abstract domain with respect to which compensation is asked for is particular and not general like welfare. It is not like the welfare based on subjective choices but concerns what really matters in particular circumstances. It is to be interpreted in a moral realistic framework.28 It is better to be seen as the implicit principle of charity necessary for understanding each other in radical interpretation, becoming more explicit and turning into a practical moral challenge.29 Because of the abstract character of the equalisandum, like expressing oneself, it could be argued that the equalisandum loses its relation with the distribuenda. Expressing oneself can in numerous ways and more or less dependent on distribuenda. Because of this lose connection it can be argued that redistribution is no longer relevant in determining the reference, consequently the whole idea of equality as it was developed here, is undermined. This argument can be met. The ways in which one can satisfy or enjoy the functions, even de ned in an abstract way, still remain dependent on the distribuenda. Of course, the particular complexity of this connection can be abused by those holding the power in the allocation process, echoing the idea that one can be happy without being rich. But actually, the dependency of valuable functions on distribuenda will remain, and the ideal of equality is not undermined. Apart from the problem concerning the reference because of diculties with the possibilities of compensation, there is a problem for the reference because it is not independent from references with respect to other equal27 For example a large part of mentally handicapped persons appeared to su er from unrecognised oesophagitis [Bohmer, 1996]. 28 See chapter 3 p. 104. 29 See chapter 2 p. 37.
9.4. EQUALITY OF HEALTH IN MEDICAL PRACTICE
295
isanda. Indeed there are several functions that are valuable, of which one function can be more important than another in a particular problem. The references linked with these functions should be compared relative to their importance. References should be determined in relation to the references of the other equalisanda. This is a problem that is more general than just for the ideal of equality. The weighing of di erent values is a problem for all proposals on moral grounds, it is one of the most essential issues of moral debates, it does not show the ideal of equality to be unintelligible. The discussion above can be illuminated with a little more formal approach. The equalisandum being something valuable, x, is a function of the condition of health c, commodities on the market m and non-marketable conditions for example care or attention a. In the reference for the ideal of equality it is valid that:
8i; j xi (ci ; mi; ai ) = xj (cj ; mj ; aj ) This requirement for the reference means for example on a global scale that the value of the function with a similar condition should be equal. In other words an aids patient here should have the same functional possibilities as an aids patient in another part of the world. If the value of this function x is not possible to be equal some other function should be looked for in order to arrive at the proper compensation. For example, if one is unable to enjoy a mathematical proof one should look for what could be like enjoying such a proof. For multiple handicapped people this means care instead of locking up because of limited resources or personnel.30 This way of looking discloses that it is not just a question of what counts as proper care, but that it is a distribution problem. How one lives or can express oneself or is cared for, is dependent on how other persons live. The shortage in care for handicapped people is not just scarcity, it is a distribution issue, for which the market of determinants of health has shown itself not to be the most appropriate instrument. The ideal of equality concerning health demands priority for handicapped people for treatment of disease because they su er more seriously from impairments. Their function is already impaired and a disease will mean even more impairment. Eciency however will guide us to other decisions. For reasons of eciency one should treat those who can contribute more to others or will pay more for their treatment. Furthermore the position on the market, the demands, and contributions are considered to be representing someone's own choice and so expressions of autonomy. Despite 30 In Holland there was the case of Jolanda Venema a mentally retarded child who was chained in her room in an institute as was argued, due to lack of resources for a decent treatment.
296
CHAPTER 9. EQUALITY OF HEALTH
these claimed advantages, which are dubious, the market mechanism is not to be seen as a nice instrument of regulating health care services from an egalitarian point of view. Summarising, health is multidimensional in which di erent functions and values do play a role. The attempts to reduce this multitude of functions to one overall function are not succesful. This multidimensionality is compatible with the ideal of equality of health. The problems with the compensation for incurable serious handicaps or diseases can be approached by the recognition of value plurality. The principle of charity implicit in the radical interpretation view, turns into an explicit moral challenge as the basis for compensation. The particularistic approach can be recognised as a characteristic of the method of general practitioners. General practitioners being the guards of health care make particularistic judgements on the health status and are inspired by the idea of care according to need in a multidimensional way, precisely what is asked for by the ideal of equality. A rather remarkable conclusion because public health care is mostly seen as an equalising instrument instead of the individual health care provided by general practitioners.
9.4.2 The general practitioner in an egalitarian perspective It is remarkable that instead of public health decisions, the decisions in general practice can be seen to tend to be according ideas of the ideal of equality of health. As explained, in the latter particular evaluations of health are made and particular aspects of impairment of health are considered and attempted to be resolved. In it, the idea: care according to need, is considered to be important. Public health measures in The Netherlands are implemented for example by the consultation oces for mother- and childcare. They are developed to give health care for all, mainly early diagnosis and prevention. It could be seen as a way of preventing socio-economic di erences with respect to for example knowledge of caring for children. But even if it is an e ective way of improving public health, it is not exactly what is asked for by the ideal of equality. Public health measures are restricted by their general nature, they concern groups of people. For example one can give more attention to or care for people in risk groups, but it remain groups to which one attends, while the individual health status should be the concern for equality of health. One could turn to groups in order to change the life expectancy in some groups. But to whatever group one belongs death is serious, and group oriented programs should not receive a priori priority. They might be ecient but they are not dir-
9.4. EQUALITY OF HEALTH IN MEDICAL PRACTICE
297
ectly inspired according the ideas of equality of health. For the latter, the general practice with its particularism could be more appropriate. General practice can follow the ideal of equality in a more clear way than public health, an advantage that is threatened by the introduction of so-called Dutch College of General Practitioners guidelines for treatment in which for each class of complaints or reasons for encounter for example, complaints of the shoulder, the actions to be undertaken are determined. These guidelines tend to become the rules by which general practitioners are evaluated. They are seen as de nitions of the products which general practitioners can supply [Rutten & Thomas, 1993]. Apart from the queer content of some of these standards, for example one can have the acute painful frequent miction syndrome without su ering from acute painful frequent miction but just excretion of cloudy urine [Rutten & Thomas, 1993, p. 306], and the way of development of guidelines similar to the `Delphi method' which was shown to be highly dubious [Cooke, 1991], it threatens the particularistic nature of the work of the general practitioner, because the rules should be followed, and are seen as constitutive of the product of general practitioners. Although general practitioners themselves believe that particular circumstances are important, it is highly questionable whether the insurance companies will accept this particularism. But what about the data of di erences in life expectancy and socioeconomic di erences in health, are these not relevant at all? They could help in re ecting about one's work as a general practitioner, they could expose the bias in one's way of working and confront one with one's tendencies of action for example that one is more likely to help patients of one's own socio-economic group, probably because identi cation is easier. Knowledge of these data could lead to a more critical evaluation of one's own way of working. In this way the data are relevant indeed as eye opener. But this material is not appropriate to de ne egalitarian goals. The aim is not to lessen just the di erences between groups. What matters is that all should be helped according to need. The data can disclose that in fact patients receive help according to some other criterion for example `easier to identify with'. Because general practitioners are responsible for the allocation of health care, they de facto guard the entrance to health care, and because their work is particularistic oriented and can accept the idea help according to need, their work could be centred according the ideal of equality. On the other hand it is often heard that allocation problems should be undertaken by politicians and not by practising physicians, otherwise the relation between patient and physician will be disturbed. It is suggested that physicians are only concerned with eciency, doing their best for their individual patients without looking to the needs of other patients. Daniels for example
298
CHAPTER 9. EQUALITY OF HEALTH
writes that restriction of budgets does not threaten the relation between the patient and physician because the physician can still advice what is actually the best for the patient under the condition of scarcity of resources [Daniels, 1985, p. 135-138]. Although Daniels' remarks might be a defence against the charge that budget constraints threaten the relation between the physician and the patient, this defence is distorting this relation. It expresses a bureaucratic attitude in which patients hear again and again they should turn to another to protest. Responsibility of choices in health care are di used in a way suggesting that physician have nothing to say on that issue. But The Royal Dutch Medical Association, has stated in her rst articles that in conformity with her name `Koninklijke Nederlandse Maatschappij tot Bevordering der Geneeskunst' she wants to promote, advance and further the art of healing. It means that physicians commit themselves to promoting health care in general. This means they cannot work without considering e ects of their treatment on others, or stated in a di erent way, they cannot be blind for the e ect of their work now on their work on other occasions. The article could mean that they take the idea of advising in the best of their patients not merely in an agent-(and patient-)relative way but also in an agent(and patient-)neutral way, meaning not destroying but even furthering the conditions such that he or another colleague can do the same on other occasions and so for other patients. In chapter 2 it was suggested that the ideal of equality can be seen as way of incorporating the relative and neutral reasons to act. A patient will be helped in so far as others are not harmed by that treatment.31 Taking the agent-(patient-)-neutral interpretation of acting for the best for the patient, is seen to be recognised in general practice by for example remaining on call if one is visiting a patient.32 It is also accepted by patients that one breaks o a treatment or interview in order to help other patients that are more seriously ill. The relationship would even break down if the interest of other patients would not have any in uence on one's advice. It would destroy the sensitivity for di erences in seriousness of complaints, all complaints would become of utmost importance undermining one's profesSee for the relation between agent relativity and agent neutrality and equality chapter 2 p. 51. 32 In fact it should have more consequences. If it is realised that merely 12% of persons with complaints evaluated by physicians to be reasons for examination because they signal serious diseases and so being good reasons for visiting a physician, actually visit a general practitioner, it will be clear there is something seriously wrong either with the organisation of health care, or the medical theory leading to these evaluations of complaints. (At least the latter because no such serious diseases appeared clearly in the follow up.) These gures published in 1983 seem to be forgotten in discussions on health care [Huygen et. al., 1983]. 31
9.5. SUMMARY
299
sional competence on which a patient has to base his thrust. Comparisons of the health status of patients are and should be made. In the ideal of equality developed in this thesis, the reference is a central idea. A general practitioner is in the position to pay attention to the distance between someone's health status and the reference and act accordingly. The particular context of patients is considered to be of central importance. For example glasses and hearing aids are more important for someone su ering dementia. The sensory inputs should be as clear as possible in order to enhance ways of communication. This can be easily recognised in general practice. But also it is to be recognised that people above the reference with minor complaints can be explained that they do not need to consult a general practitioner, but can consult other sources of information. Working according the ideal of equality could be considered as a leading aspect of the work of general practitioners. Their work could be evaluated by this. It could even become their most important public function. If this ideal is not subscribed to and eciency becomes the main goal the position of general practitioners will be undermined. It might be argued that because of eciency they can be better incorporated in public health care, organised in the way of consultation oces governed by general rules or alternatively that they can be better transformed into little shopkeepers of health care products providing the most pro table products. The result will be that most needed medical services are lost.
9.5 Summary In this chapter, it was explored what the ideal of equality meant for the discussion on equality of health. It was argued that from the issues on the macro level, the one concerning socio-economic di erences of health was likely to hide the main concern of egalitarians, i.e. that some are less healthy than all could be. That there are no socio-economic related di erences in health, either because there are no socio-economic di erences at all, or there is no positive correlation between socio-economic status and health, is not what matters regarding the ideal of equality of health. Equalising the socio-economic status is not the main concern of equality. The idea that it is, could be intelligible if one holds a non-envy conception of equality or a resource theory of equality, but both were rejected in the former chapters because of the queer consequences these conceptions have. The idea that socio-economic classes should not be di erent in their health status could be considered to follow from the idea of Walzer on equality. This was rejected because it was based on social meanings and
300
CHAPTER 9. EQUALITY OF HEALTH
fell in the trap of relativism or imperialism. This is a disadvantage, which can be avoided by the more realistic interpretation of Wittgenstein's conception of rule following, the basis of the ideal of equality developed in this thesis. But even more important is that equality of health does not concern populations of di erent socio-economic status being indiscernible from each other regarding health, but it does concern that some are worse o than all could have been. The idea that socio-economic status should be negatively correlated with health could be considered to follow from the ideas of Lucas. But this too was rejected because it would imply to give the individual with a lower income a better treatment than the one with the higher income although they have the same condition of health. Socio-economic di erences of health are not declared to be not important, socio-economic status cannot be denied to be a determinant of health, but the ideal of equality of health is that there are no di erences in health at all, independent from their relation to socio-economic di erences. The other macro level issue concerning choices between groups of patients was reduced to micro level decisions on choices between people. The micro level decisions should determine how the macro level decisions should be made. The micro level decisions are attempted to be solved by criteria which turned out to run into problems because of their assumed generality. As a way out of those general methods the decision theoretical approach was suggested. The instruments, qalys, hyes and lyars, were introduced. It was questioned whether these could function as equalisanda. It turned out that these instruments were based on separability or additivity, an element of moral universalism. Additivity was not acceptable. It could only be saved at the expense of undermining other essential parts of decision theory. Furthermore, they turned out have monistic tendencies, and if they could escape these monistic tendencies they turned out to be based on subjective choices in line with volitional individualism, with the consequence that interpersonal comparisons of health status became a problem. In the alternative, in which health is considered to be something objective, and is approached by taking polls and health questionnaires as the method of determining health, the conceptual relationship between health and what people believe about health was seen to be neglected. These polls could not be taken as the last words on health but merely as the rst ones. It was argued that health should be considered in a multidimensional way which is acknowledged by general practitioners who are educated to look at the meaning of diseases and impairments for their patients. Further it was made clear that following the ideal of equality developed in this study,
9.5. SUMMARY
301
which can be represented by the index:
Xn
I(~x) = M( (xref , minfxref ; xig)2 ) i=1
means that the worst-o should be helped rst, then the second worsto , etc. This is re ected by the concavity of the index. It demands in a particularistic way health care according to need. A way of working which can be considered together with pluralism, the basis of the work of a general practitioner, who is considered to give continuous personal and integral care and who is considered to be the guard of the entrance to the health care system. It was argued that the conception of equality was threatened by considering severely handicapped people and children with a fatal disease with an infaust prognosis, because the level of the reference will become so low that nobody would be below the level. But this problem could be met. The principle of charity, necessary for understanding in the Wittgenstein-Davidson approach to language and interpretation, transformed into an explicit moral challenge. It is the basis on which there will remain possibilities for compensation. These could be worked out in general practice. In general practice in which continuous personal and integral care should be given, the ideal of equality could be incorporated rather easily. The idea that care should be given according to need is already acknowledged in general practice. The ideal of equality as it was developed here, could be considered to become the main characteristic of general practice. I suggested that if eciency is aimed at, it could be argued that the general practice would lose its special place and that it would be better to be incorporated in another structure like the consultation oces, working according to protocols. And if the market is embraced, the general practice will easily evoluate into a shop of health care products which provide the most pro table products with the consequence that health care for the most needy is lost. The ideal of equality could become the rationale for the discipline of general practice, or family medicine. Within the particularistic pluralistic framework proposed in this thesis there will become a eld of research beyond epidemiological studies and the development of treatment protocols. A eld for which interpretative methods will be more appropriate in order to recover meanings of conditions. A eld which seems to be forgotten but worthwhile to remember and to explore. An important project following from a viable conception of equality within a moral realistic framework.
302
CHAPTER 9. EQUALITY OF HEALTH
Bibliography [Apostol1, 1961] T.M. Apostol, Calculus vol 1. NewYork, 1967. [Apostol2, 1961] T.M. Apostol, Calculus vol 2. New York, 1967. [Arnsperger, 1994] C. Arnsperger, Envy-freeness and distributive justice. In: Journal of Economic Surveys, 8, 1994, p. 155-186. [Arnsperger, 1996] C. Arnsperger, Envy-freeness and distributive justice: A reply to Kolm's comment. In: Journal of Economic Surveys, 10, 1996, p. 217-223. [Arnsperger, 1997] C. Arnsperger, Refromulating equality of resources. In: Economics and Philosophy, 13, 1997, p. 61-77. [Arrow, 1951] K.J. Arrow, Social choice and individual values. New York, 1951. [Arrow, 1971]
K.J. Arrow, Uncertainty and the welfare economics of medical care. In: Essays in the theory of risk bearing, Dordrecht, 1971, p. 177-211.
[Atkinson, 1972]
A.B. Atkinson, On the measurement of inequality. In: Journal of Economic Theory, 2, 1972, p. 244-263. [Baker & Hacker, 1984] G.P. Baker & P.M.S. Hacker, Scepticism, rules and language. Oxford, 1984. [Baker & Hacker, 1985] G.P. Baker & P.M.S. Hacker, Rules, grammar and necessity. Oxford, 1985. [Barry, 1989] B. Barry, Theories of justice. Berkeley, 1989. [Barry, 1995] B. Barry, Justice as impartiality. Oxford, 1995. 303
BIBLIOGRAPHY
304 [Barwise, 1977]
J. Barwise (ed.), Handbook of mathematical logic. Amsterdam, 1977.
[Baumol, 1986]
W.J. Baumol, Superfariness. Cambridge, Massachusetts, 1986.
[Benn, 1967]
S. Benn, Property. In: Encyclopaedia of Philosophy, P. Edwards (ed.), vol 6, 1967, p. 491-494.
[Blackburn, 1984] S. Blackburn, Spreading the word. Oxford, 1984. [Blackburn, 1993] S. Blackburn, Essays in quasi-realism. Oxford, 1993. [Boadway & Bruce, 1984] R.W. Boadway & N. Bruce, Welfare economics. Oxford, 1984. [Bohmer, 1996]
C.J.M. Bohmer, Gastro-oesophageal re ux disease in intellectually disabled individuals: epidemiological and clinical studies in institutes in the Netherlands and Belgium. Thesis, Free University Amsterdam, Amsterdam, 1996
[Borges]
J.L. Borges, De Loterij van Babylonie. From Ficiones. Dutch translation is: De Zahir, Amsterdam, 1986, p. 116-124.
[Brink, 1989]
D.O. Brink, MOral realism and the foundations of ethics. Cambridge, 1989.
[Broome, 1978]
J. Broome, Trying to value a life. In: Journal of Public Economics, 9, 1978, p. 91-100.
[Broome, 1979]
J. Broome, Trying to value a life, a reply. In: Journal of Public Economics, 12, 1979, p. 259-262.
[Broome, 1988]
J. Broome, Good, fairness and QALY's. In: Philosophy and medical welfare, J.M. Bell & S. Mendus (eds.), Cambridge, 1988, p. 57-73.
[Broome, 1989]
J. Broome, What's the good of equality? In: Current issues in Microeconomics, J. Hey (ed.), Cambridge, 1989, p. 236-288.
[Broome, 1991]
J. Broome, Weighing goods. Oxford, 1991.
BIBLIOGRAPHY [Carens, 1981] [Cohen, 1983] [Cohen, 1989] [Cohen, 1993] [Cooke, 1991] [Culyer, 1990]
305 J. H. Carens, Equality, moral incentives and the market: an essay in utopian politico-economic theory. Chicago, 1981. G.A. Cohen, The structure of proletarian unfreedom. In: Philosophy and Public A airs, 12, 1983, p. 3-33. G.A. Cohen, On the currency of egalitarian justice. In: Ethics, 99, 1989, p. 906-944. G.A. Cohen, Incentives. draft, Oxford, 1993. R.M. Cooke, Experts in uncertainty. Oxford, 1991. A.J. Culyer, Commodities, characteristics of commodities, characteristics of people, utilities, and the quality of life. In: Quality of life, S. Baldwin & C. Godfrey &
C. Propper (eds.), London, 1990, p. 9-27. [Dancy, 1983] J. Dancy, Ethical particularism and morally relevant properties. In: Mind, 92, 1983, p. 530-547. [Dancy, 1985] J. Dancy, Introduction to contemporary epistemology. Oxford, 1985. [Dancy, 1993] J. Dancy, Moral reasons. Oxford, 1993. [Daniel, 1978] T.E. Daniel, Pittfalls in the theory of fairness. In: Journal of Economic Theory, 19, 1978, p. 561-564. [Daniels, 1985] N. Daniels, Just health care. Oxford, 1985. [Dasgupta & Sen, & Starlett, 1973] P. Dasgupta & A.K. Sen & D. Starlett, Notes on the measurement of inequality. In: Journal of Economic Theory, 6, 1973, p. 180-187. [Davidson, 1970] D. Davidson, em How is weakness of the will possible? In: Essays on actions and events, Oxford, 1980, p. 2142. [Davidson, 1974] D. Davidson, em On the very idea of a conceptual scheme. In: Inquiries into truth and interpretation, Oxford, 1984, p. 183-198. [Davidson, 1980] D. Davidson, em Essays on actions and events. Oxford, 1980.
306 [Davidson, 1984] [Davidson, 1986] [Davidson, 1990] [Davies, 1994]
[De Beer, 1993] [De Beus, 1993] [De Jong, 1986] [Dunning, 1991] [Dumont, 1983] [Dworkin, 1981a] [Dworkin, 1981b] [Dworkin, 1993] [Dworkin, 1994] [Fishburn, 1970]
BIBLIOGRAPHY D. Davidson, em Inquiries into truth and interpretation. Oxford, 1984. D. Davidson, em Judging interpersonal interest. In: Foundations of social choice theory, J. Elster (ed.), Cambridge, 1986, p. 195-211. D. Davidson, The structure and content of truth. In: Journal of Philosophy, 87, 1990, p. 279-328. M. Davies, Simulation v.s. theory theory: what is at issue? In: Objectivity, simulation and the unity of consciousness, C. Peacocke (ed.), Oxford, 1994, p. 99127. P. de Beer, Het verdiende inkomen. Deventer, 1993. J. de Beus, Economische gelijkheid en het goede leven. Amsterdam, 1993. G.A. de Jong, Ongelijheid in gezondheid, sociaal medisch onderzoek en beleid. Thesis, Erasmus University Rotterdam, 1986. A.J. Dunning (ed.), Kiezen en delen. Den Haag, 1991. L. Dumont, Essais sur l'individualisme. Editions du Seuil, 1983. R. Dworkin, What is equality? Part 1: Equality of welfare. In: Philosophy and Public A airs, 10, 1981, p. 185-246. R. Dworkin, What is equality? Part 2: Equality of resources. In: Philosophy and Public A airs, 10, 1981, p. 283-345. R. Dworkin, Justice in the distribution of health care. In: McGill Law Journal, 38, 1993, p. 883-898. R. Dworkin, Will Clinton's plan be fair? In: New York Review of Books, 41, 13 january 1994, p. 20-25. P.C. Fishburn, Utility theory for decision making. London, 1977.
BIBLIOGRAPHY [Fishkin, 1988] [Fleurbay, 1994] [Foster, 1983] [Foster, 1985] [Frankena, 1973] [Frankfurt, 1987]
307 J.S. Fishkin, The complexity of simple justice. In: Ethics, 98, 1988, p. 464-471. M. Fleurbay, On fair compensation. In: Theory and Decision, 36, 1994, p. 277-307. J.E. Foster, An axiomatic characterization of the Theil measure of income inequality. In: Journal of Economic Theory, 31, 1983, p. 105-121. J.E. Foster, Inequality measurement. In: Proceedings of Symposia in Applied Mathematics, H.P. Young (ed.) 1985, p. 31-69. W.K. Frankena, Ethics. New Jersey, 1973.
H. Frankfurt, Equality as a moral ideal. In: Ethics, 98, 1987, p. 21-43. [Gewirth, 1978] A. Gewirth, Reasons and morality. London, 1978. [Goodwin, 1992] B. Goodwin, Justice by lottery. New York, 1992. [Gough, 1957] J.W. Gough, The social contract. Oxford, 1957. [Grin, 1991] J. Grin, Against the taste model. In: Interpersonal comparisons of well-being, J. Elster & J. Roemer (eds.), Cambridge, 1991, p. 45-69. [Hammond, 1976] P.J. Hammond, Why ethical measures of inequality need interpersonal comparisons. In: Theory and Decision, 7, 1976, p. 263-274. [Hansson, 1977] B. Hansson, The measurement of social inequality. In: Foundational problems in teh special sciences, Butts & Hintikka (eds.), Dordrecht, 1977, p. 303-312. [Hare, 1952] R.M. Hare, The language of morals. Oxford, 1952. [Hare, 1960] R.M. Hare, Freedom and reason. Oxford, 1960. [Hare, 1975] R.M. Hare, Abortion and the golden rule. In: Philosophy and Public A airs, 4, 1975, p. 201-222. [Hare, 1981] R.M. Hare, Moral thinking. Oxford, 1981.
308 [Hare, 1989]
BIBLIOGRAPHY R.M. Hare, Liberty, equality and fraternity in South Africa? In: Essays on political and morality, Oxford,
1989, p. 167-181. [Hartog, 1981] F. Hartog, Openbare nancien en overleg democratie. In: Inkomensverdeling en openbare nancien, P.J. Eijgelshoven & L.J. van Gemerden (eds.), Utrecht, 1981, p. 111-127. [Hennipman, 1977] P. Hennipman, Welvaartstheori en economische politiek. Alphen a/d Rijn, 1977. [Hennipman, 1981] , De verdeling in de paretiaanse welvaartstheorie. In: Inkomensverdeling en openbare nancien, P.J. Eijgelshoven & L.J. van Gemerden (eds.), Utrecht, 1981, p. 128-170. [Hicks, 1965] J. Hicks, Capital and growth Oxford, 1965. [Hochman & Rodgers, 1969] H.M. Hochman & J.D. Rodgers, Pareto optimal redistribution. In: The American Economic Review, 1969, p. 542-557. [Hochman & Rodgers, 1974] H.M. Hochman & J.D. Rodgers, Redistribution and the Pareto criterion. In: The American Economic Review, 1974, p. 752-757. [Hodgkin, 78] K. Hodgkin, Towards earlier diagnosis in primary care. Edinburgh, 1978. [Honderich, 1980] T. Honderich, Violence for equality., Middlesex, 1980. [Hurley, 1989] S.L. Hurley, Natural Reasons. Oxford, 1989. [Hurley, 1992] S.L. Hurley, Intelligibility, imperialism and conceptual scheme. In: Midwest Studies in Philosophy XVII, P.A. French & T.E. Uehling & H. Wettstein (eds.), Notre Dame, 1992, p. 89-108. [Hurley, 1993] S.L. Hurley, Justice without constitutive luck. In: Ethics; Royal Institue of Philosophy Supplement 35, A. Phillips Griths (ed.), Oxford, 1993, p. 179-212. [Hurley, 1994] S.L. Hurley, Unity and objectivity. In: Objectivity, simulation and the unity of consciousness, C. Peacocke (ed.), Oxford, 1994, p. 49-77.
BIBLIOGRAPHY
309
[Huygen et. al., 1983] F.J.A. Huygen & H. v.d. Hoogen & W.J. Neefs, Gezondheid en ziekte; een onderzoek van gezinnen. In: Nederlands Tijdschrift voor Geneeskunde, 127, 1983, p. 1612-1619. [Jackson, 1985] F. Jackson, Davidson on moral con ict. In: Actions and events, E. Le Pore & B.P. McLaughin (eds.), Oxford, 1985, p. 104-115. [Jacobs, 1993] H.M. Jacobs, Health status measurement in family medicine research. Thesis, Univesity of Utrecht, Utrecht, 1993. [Jacobs & Touw-Otten 1995] H.M. Jacobs & F.W.M.M. Touw-Otten, De Nederlands versie van de Sickness Impact Pro le in huisartsgeneeskundig onderzoek. In: Huisarts en
Wetenschap, 38, 1995, p. 122-128. [Johansson, 1995] P.O. Johansson, Evaluating health risks. Cambridge, 1995. [Jones-Lee, 1976] M.W. Jones-Lee, The value of life. London, 1976. [Johnston, 1989] P. Johnston, Wittgenstein and moral philosophy. London, 1989. [Kagan, 1988] S. Kagan, The additive fallacy. In: Ethics, 99, 1988, p. 5-31. [Kakwani, 1980] N. Kakwani, Income, inequality and poverty. Oxford, 1980. [Kind et. al., 1990] P. Kind & C Gudex & C. Godfroy, Introduction: What are QALY's? In: Quality of life, S. Baldwin & C. Godfroy, C. Propper (eds.), London, 1990, p. 57-62. [Kohout] P. Kohout, De beulse. Utrecht, 1979, (Dutch translation), the original title is "Katyne". [Kolm, 1968] S.Ch. Kolm, The optimal production of social justice. In: Public economics, J. Margolis (ed.), London, 1969, p. 145-200. [Kolm, 1972] S.Ch. Kolm, Justice et equite. Paris, 1972. [Kolm, 1976a] S.Ch. Kolm, Unequal inequalities I. In: Journal of Economic Theory, 12, p. 416-442.
310 [Kolm, 1976b] [Kolm, 1977] [Kolm, 1984] [Kolm, 1985] [Kolm, 1996]
[Kool, 1995] [Kripke, 1982] [Lambert, 1993] [LeGrand, 1991] [Lewis, 1973] [Lewis, 1983] [L.H.V., 1985] [L.H.V., 1988] [Little, 1950] [Lock, 1989] [Lucas, 1965]
BIBLIOGRAPHY S.CH. Kolm, Unequal inequalities II. In: Journal of Economic Theory, 13, 1976, p. 82-111. S.Ch. Kolm, Multidimensional egalitarianism. In: The Quarterly Journal of Economic Theory, 91, 1977, p. 1-13. S.Ch. Kolm, Le liberalisme moderne. Paris, 1984. S.Ch. Kolm, Le contrat social leberal. Paris, 1985. S.Ch. Kolm, Playing fair with fairness (a comment
on Arnsperger's "Envy-freeness and distributive justice. In: Journal of Economic Surveys, 10, 1996, p.
197-215. T. Kool, Naar een gezondheidsmarkt? Thesis, Amsterdam, 1995. S. Kripke, Wittgenstein: on rules and private language. Oxford, 1982. P.J. Lambert, The distribution and redistribution of income. Manchester, 1993. J. LeGrand, Equity and choice. New York, 1991. D. Lewis, Counterfactuals. Oxford, 1973. D. Lewis, Radical interpretation. In: Philosophical papers vol 1, New York, 1983. L.H.V. Vademecum voor de huisarts. L.H.V. Utrecht, 1985. L.H.V. De positie van de huisarts in de toekomst. Discussienota, L.H.V. Utrecht, 1988. I.M.D. Little, A critique of welfare economics. Oxford, 1950. G.E. Lock, Self-ownership, equality of resources and the case of the indolent indigent. In: R. & R., 18, 1989, p. 35-47. J.R. Lucas, Against equality. In: Philosophy, 40, 1965, p. 296-307.
BIBLIOGRAPHY
311
[Lucas, 1977]
J.R. Lucas, Against equality again. In Philosophy, 52, 1977, p. 255-280. [Mackenbach, 1994] J.P. Mackenbach, Ongezonde verschillen. Assen, 1994. [Mackie, 1977] J.L. Mackie, Ethics. Middlesex, 1977. [Mackie, 1985] J.L. Mackie, The three satges of universalization. In: Persons and values 2, Oxford, 1985, p. 170-183. [McDowell, 1981] J.M. McDowell, Non-cognitivism and rule-following. In: Wittgenstein: to follw a rule, Holtzman & Leich (eds.), London, 1981, p. 141-161. [McDowell, 1982] J.M. McDowell, Criteria, defeasibility and knowledge. In: Perceptual knowledge, J. Dancy (ed.) Oxford, 1988, p. 209-219. [McDowell, 1985] J.McDowell, Values and secondary qualities. In: Morality and objectivity, T. Honderich (ed.), London, 1985, p. 110-129. [McDowell, 1994] J. McDowell, Mind and world. Oxford, 1994. [McKeown, 1979] T. McKeown, The role of medicine. Oxford, 1979. [Medisch Contact, 1997] K.N.M.G. Sterfteverschillen tussen arm en rijk. In: Medisch Contact, 52, 1997, p. 51. [Mehrez & Gafni, 1989] A. Mehrez & A. Gafni, Quality adjusted life years, utility hteory and healthy years equivalents. In: Medical Decision Making, 9, 1989, p. 142-149. [Mehrez & Gafni, 1991] A. Mehrez & A. Gafni, The healthy years equivalents: How to measure them using the standard gamble approach. In: Medical Decision Making, 11, 1991, p.
[Milner, 1980] [Mooney, 1977] [Moulin, 1995] [Nagel, 1978]
140-146. M. Milner, Unequal care. New York, 1980. G.H. Mooney, The evaluation of life. Oxford, 1977. H. Moulin, An assessment of cooperative game theory. Paper for the coference Paris Cergy, 1995. T. Nagel, The possibility of altruism. Princeton, 1978.
312
BIBLIOGRAPHY
[Nagel, 1986] [Nagel, 1991] [Ng, 1984]
T. Nagel, The view from nowhere. Oxford, 1986. T. Nagel, Equality and partiality. Oxford, 1991. Y.K. Ng, Interpersonal level comparability implies comparability of utility di erences. In: Theory and Decision, 17, 194, p. 141-147. [Norman, 1971] R. Norman, Reasons for actions. Oxford, 1971. [Nozick, 1974] R. Nozick, State anarchy and utopia. Oxford, 1974. [Nozick, 1993] R. Nozick, The nature of rationality. Princeton, 1993. [NRC handelsblad 1994] Troonrede, NRC Handelsblad 20 september 1994, p. 23. [NRC handelsblad, 9-11-1996] [Nussbaum, 1993] M.C. Nussbaum, Therapy of desire. Oxford, 1993. [O'Neill, 1975] O. O'Neill, Acting on principle. New York, 1975. [O'Neill, 1985] O. O'Neill, Consistency in action. In: Morality and universality, N. Potter & M. Timmons (eds.), Dordrecht, 1985, p. 159-186. [O'Neill, 1989] O. O'Neill, Constructions of reasons. New York, 1989. [O'Neill, 1993] O. O'Neill, Justice, gender, and international boundaries. In: The quality of life, M.C. Nussbaum & A.K. Sen (eds.), Oxford, 1993, p. 303-335. [Osmani, 1982] S.R. Osman, Economic inequality and group welfare. Oxford, 1982. [Par t, 1984] D. Par t, Reasons and persons. Oxford, 1984. [Par t, 1989] D. Par t, On giving priority to the worse-o . Unpublished Manuscript, Oxford, 1989. [Pazner & Schmeidler, 1974] E. Pazner & D. Schmeidler, A diculty in the concept of fairness. In: Review of Economic Studies, 41, 1974, p. 441-443. [Pazner & Schmeidler, 1978] E. Pazner & D. Schmeidler, Egalitarian equivalent allocations: A new concept of economic equity.
In: Quarterly Journal of Economics, 92, 1978, p. 1-45.
BIBLIOGRAPHY [Peacock, 1983] [Pettit, 1989]
313
C. Peacock, Sense and content. Oxford, 1983. P. Pettit, Decision theory and folk psychology. In: Foundations of decision theory, S.L. Hurley & M. Bacharach, Oxford, 1989, p. 147-175. [Putnam, 1992] H. Putnam, Renewing philosophy. Cambridge, Massachusetts, 1992. [Rakowksi, 1991] E. Rakowski, Equal justice. Oxford, 1991. [Rawls, 1971] J. Rawls, A theory of justice. Oxford, 1971. [Rawls, 1985] J. Rawls, Social unity and primary goods. In: Utilitarianism and beyond, A.K. Sen & B. Williams (eds.), Cambridge, 1985, p. 159-185. [Rescher, 1964] N. Rescher, Hypothetical reasoning. Amsterdam, 1964. [Rescher, 1993] N. Rescher, Pluralism. Oxford, 1993. [Roberts, 1977] K.W.S. Roberts, Welfare theoretical social choice. Thesis, Oxford, 1977. [Robbins, 1948] L. Robbins, An essay on the nature and signi cance of economic science. London, 1984. [Robinson, 1962] J. Robinson, Economic philosophy. Middlesex, 1962. [Roemer, 1985] J.E. Roemer, Equality of talent. In: Economics and philosophy, 1, 1985, p. 151-187. [Roemer, 1986] J.E. Roemer, Equality of resources implies equality of welfare. In: Quarterly Journal of Economics, 100, 1986, p. 751-784. [Rosanvallon, 1989] P. Rosanvallon, Le liberalisme economique. Editions du Seuil, 1989, (in 1979 printed as: Le capitalisme utopique). [Rutten & Thomas, 1993] G.E.H.M. Rutten & S. Thomas, NHG standaarden voor huisartsen. Utrecht, 1993. [Sabine & Thorson, 1973] G.H. Sabine & T.L. Thorson, A history of political theory. Ilinois, 1973. [Samuelson, 1984] P.A. Samuelson, Economics. London, 1984.
314
BIBLIOGRAPHY
[Savage, 1972]
L.J. Savage, The foundations of statistics. New York, 1972. [Scanlon, 1975] T.M. Scanlon, Preferences and urgency. In: The Journal of Philosophy, 72, 1975, p. 655-669. [Scanlon, 1982] T.M. Scanlon, Contractualism and utilitarianism. In: Utilitarianism and beyond, A.K. Sen & B. Williams (eds.), Cambridge, 1982, p. 103-128. [Scanlon, 1986] T.M. Scanlon, Equality of resources and equality of welfare; a forced marriage? In: Ethics, 87, 1986, p. 111-118. [Scanlon, 1988] T.M. Scanlon, The signi cance of choice. In Tanner lectures on human values, C.M. Murrin (ed.), Cambridge, 1988, p. 149-216. [Scanlon, 1991] T.M. Scanlon, The moral basis of interpersonal comparisons. In: Interpersonal comparisons of well-being, J. Elster & J.E. Roemer (eds.), Cambridge, 1991, p. 17-44. [Schayck et. al., 1995] C.P. van Schayck et. al., Longfunctie en ervaren gezondheid bij patienten met astma of chronische obstructieve longziekten in longitudinaal onderzoek. In:
[Scruton, 1984] [Sen, 1973] [Sen, 1977] [Sen, 1978] [Sen, 1979a] [Sen, 1979b]
Huisarts en Wetenschap, 38, 1995, p. 132-138. R. Scruton, The meaning of conservatism. London, 1984. A.K. Sen, On economic inequality. Oxford, 1973. A.K. Sen, Welfare inequalities and Rawlsian axiomatics. In: Foundational problems in the special sciences, Butts & Hintikka (eds.), Dordrecht, 1977, p. 271-292. A.K. Sen, em Ethical measurements of inequality; some diculties. Repr. in: Choice welfare and measurement, Oxford, 1982, p. 416-431. A.K. Sen, Collective choice and social welfare. Amsterdam, 1979. A.K. Sen, Utilitarianism and welfarism. In: The Journal of Philosophy, 76, 1979, p. 463-489.
BIBLIOGRAPHY [Sen, 1980]
315
A.K. Sen, Equality of what? Repr. in: Choice welfare and measurement, Oxford, 1982, p. 353-369. [Sen, 1981] A.K. Sen, Plural utility. In: Proceedings of the Aristotelian Society, 1981, p. 191-215. [Sen, 1992] A.K. Sen, Inequality reexamined. Oxford, 1992. [Sen, 1993] A.K. Sen, Positional objectivity. In: Philosophy and Public A airs, 22, 1993, p. 126-145. [Shorrocks, 1980] A.F. Shorrocks, The class of additive decomposable inequality measures. In: Econometrica, 48, 1980, p. 613623. [Shorrocks, 1983] A.F. Shorrocks, Ranking income distributions. In: Economica, 50, 1983, p. 3-17. [Shorrocks, 1984] A.F. Shorrocks, Inequality decomposition by population subgroups. In: Econometrica, 52, 1984, p. 1369-1385. [Shorrocks & Foster, 1987] A.F. Shorrocks & J.E. Foster, Transfer sensitive measures of inequality. In: Review of Economic Studies, 54, 1987, p. 484-497. [Singer, 1971] M.G. Singer, Generalization in ethics. New York, 1971. [Singer, 1985] M.G. Singer, Universalizability and the generalization principle. In: Morality and universality, N. Potter & M. Timmons (eds.), Dordrecht, 1985, p. 47-73. [Sobel & Winters, 1996] J.S. Sobel & J.C. Winters, Shoulder complaints in general practice. Groningen, 1996. [Stalnaker, 1984] R. Stalnaker, A note on non-monotonic modal logic. Unpublished paper, 1984. [Stocker, 1990] M. Stocker, Plural and con icting values. Oxford, 1990. [Strawson, 1986] G. Strawson, Freedom and belief. Oxford, 1986. [Strawson, 1961] P.F. Strawson, Social morality and individual ideal. In: Freedom and resentment and other essays, London, 1974, p. 26-44.
316
BIBLIOGRAPHY
[Streiner & Norman, 1995] D.L. Streiner & G.R. Norman, Health measurement scales. Oxford, 1995. [Stronks & Gunning-Schepers, 1993] K. Stronks & L.J. Gunning-Schepers, Should equity in health be target number 1? In: European Journal of Public Health, 3, 1993, p. 104-111. [Stronks et. al., 1995] K. Stronks & H. van Mheen & J.P. Mackkenbach, Socioeconomic status of the Dutch population. Report nr. 431501010, National Institute of Pulbic Health and Environmental Protection, Bilthoven, Erasmus University Rotterdam, 1995. [Sundrum, 1990] R.M. Sundrum, Income distribution in less developed countries. London, 1990. [Suppes, 1966] P. Suppes, Some formal models of grading principles. In: Synthese, 1966, p. 284-306. [Temkin, 1993] L.S. Temkin, Inequality. Oxford, 1993. [Takayama, 1985] A. Takayama, Mathematical economics. Cambridge, 1985. [Thomson, 1987] W. Thomson, The vulnerability to manipulative behav-
ior of resource allocation mechanisms designed to select equitable and ecient outcomes. In: Information,
incentives and economic mechanisms, T. Groves & R. Radner & S. Reiter, Minneapolis, 1987, p. 375-396. [Tinbergen, 1953] J. Tinbergen, Redelijke inkomensverdeling. Haarlem, 1953. [Tinbergen, 1979] J. Tinbergen, Income distribution. Dordrecht, 1979. [Tinbergen, 1991] J. Tinbergen, Armen komen de welvaart bij ons halen. In: NRC Handelsblad, 23 May 1991, p. 8. [Townsend & Davidson, 1982] P. Townsend & N. Davidson, Inequalities in health. The Black report. Middlesex, 1982. [Vallentyne, 1996] P. Vallentyne, Self-ownership and equality. Workshop University Warwick, 4 May 1996. [Van Es, 1984] J.C. van Es, Patient en huisarts. Utrecht, 1984.
BIBLIOGRAPHY
317
[Van Parijs, 1990] P. van Parijs, Equal endowments as undominated diversity. In: Recherches Economique de Louvain, 56, 1990, p. 327-356. [Van Parijs, 1991] P. van Parijs, Qu'est-ce que qu'une societe juste? Paris, 1991. [Van Parijs, 1991b] P. van Parijs, Why surfers should be fed: The liberal case for an unconditional basic income. In: Philosophy and Public A airs, 20, 1991, p. 1-32. [Van Parijs, 1995] P. van Parijs, Real freedom for all. Oxford, 1995. [Van Praag, 1993] B.M.S. van Praag, The relativity of the welfare concept. In: The quality of life, M.C. Nussbaum & A.K. Sen, (eds.), Oxford, 1993, p. 362-385. [Van Wijck, 1992] P. van Wijck, Inkomensverdeling en beleid. Thesis, Rotterdam, 1992. [Varian, 1974] H.R. Varian, Equity, envy and eciency. In: Journal of Economic Theory, 9, 1974, p. 63-91. [Varian, 1975] H.R. Varian, Distributive justice, welfare economics, and the theory of fairness. In: Philosophy and Public A airs, 4, 1975, p. 223-247. [Varian, 1992] H.R. Varian, Microeconomc analysis. London, 1992. [Vermeersch, 1988] E. Vermeersch, De ogen van de panda. Antwerpen, 1988. [Walzer, 1983] M. Walzer, Spheres of justice. New York, 1983. [Walzer, 1988] M. Walzer, Interpretation and social criticism. In: Tanner lectures on human values, C.M. Murrin (ed.) Cambridge, 1988, p. 1-80. [Wiggins, 1982a] D. Wiggins, Claims of need. In: Needs, values, truth, Oxford, 1982, p. 1-57. [Wiggins, 1982b] D. Wiggins, Needsl, values, truth. Oxford, 1982. [Willams, 1972] B. Williams, Morality: An introduction to ethics. Cambridge, 1972.
318 [Williams, 1974]
BIBLIOGRAPHY
B. Williams, Wittgenstein and idealism. In: Moral luck, Cambridge, 1981, p. 144-163. [Williams, 1985] B. Williams, Ethics and the limits of philosophy. London, 1985. [Winch, 1958] P. Winch, The idea of a social science. London, 1958. [Wittgenstein, 1918] L. Wittgenstein, Tractatus logico philosophicus. Frankfurt aM, 1984. [Wittgenstein, 1945] L. Wittgenstein, Philosophische Untersuchungen. Frankfurt aM, 1984. [World Bank, 1992] World Bank, World development report 1992. Oxford, 1992. [Young, 1994] H.P. Young, Equity: In theory and practice. Princeton, 1994.
Index Ackerman, B.A., 91
Blackburn, S., 155, 157n, 167{171,
173{175
action description basic, 59 relation with belief, 38{42, 53, 57, 102, 168 action-guidingness, 59, 116{118, 122, 151 agent-neutral value, 51, 52, 199, 298 agent-relative value, 51, 52, 199, 298 akrasia, 116, 126{128 amoralist partial, 35n, 41 total, 35n, 41 Anscombe, G.E.M., 45, 101 Apostol, T.M., 249, 250, 250n arational irrationalist, 40, 42 Arnsperger, C., 84n, 89n, 90, 95 Arrow, K.J., 55n, 66, 66n, 105, 121, 177, 274, 276 Atkinson, A.B., 224, 226 atomism, 171 award view, 147, 148
Boadway, R.W., 80 Bohmer, C.J.M., 294n Borges, J.L., 89, 89n Borst, E., 136 Brink, D.O., 33n, 35n, 41 Broome, J., 18n, 64, 67n, 121n,
189{192, 249n, 279, 285, 287n brotherhood, see fraternity Bruce, N., 80 care
continuous integral personal, 290, 291, 296, 301 public health, 296, 299 Carens, J.H, 2n Caroll, L., 43n, 159 categorical imperative, 26 causality, 47, 171 choice genuine, 127{129, 132, 135, 151 circularity, 73n claim rights, 200 coecient of variation, see measure of inequality Cohen, G.A., 9, 12, 68, 116, 118n, 136, 143{146, 148, 186, 187, 198, 207n coherentism, 46, 49, 52, 57, 134, 153, 177
Baker, G.P., 13, 155, 161{163, 179
ban on mixed worlds, 170{172, 179 Barry, B., 2, 42n Barwise, J., 174 Baumol, W.J., 74n, 76, 99n Benn, S.I., 94 bias, 177, 297 319
320 collective availability, 25, 27, 52, 57, 198, 201 common practice, 34, 37, 48, 104, 154, 156, 288 communalegalitarianism, see equality meaning communal compactness theorem, 59{61, 174n compensation, 116, 117, 122, 125, 126, 290 in kind principle of, 117, 125, 132 compensation variation, see measure of health competition, 27 Condorcet paradox, 176, 177, 179, 180 conservatism, 13, 154, 156{161, 163, 179, 278 continuity, 245 contract theory, 4, 6, 115, 200 contradiction in conception, 26 contradiction in the will, 26, 27 convention T, 156 convex, 250 Cooke, R.M., 297 counterfactual, 52n, 118, 195 Culyer, A.J., 284, 288 Dahl, R., 286 Dalton, H., 225 Dancy, J., 1, 6, 7, 8n, 10, 33,
33n, 44, 46{48, 51n, 102, 132n, 171 Daniel, T.E., 83 Daniels, N., 297 Dasgupta, P., 217 Davidson, D., 1, 7, 10, 11, 13, 19, 37, 39, 42, 43, 47n, 48, 51, 53, 57, 59, 67, 74, 102n, 103, 104, 106, 115, 126, 133, 153, 157, 163,
INDEX 164, 175, 179, 194, 199, 288, 290, 301 Davidson, N., 273n De Beer, P., 68 De Beus, J.W., 54n De Jong, G.A., 273n decision theory, 279, 282, 283, 285{ 287, 289, 300 decomposability, 14, 223, 240, 266 Delphi method, 297 desiderative structure, 129, 131, 132, 135, 287 desire ful lment theory, 190n Diamond, P.A., 130 dictator, 66, 66n, 105, 121 di erence principle, 118n, 207 disagreement, 13, 65, 106, 173{ 179 moral, 6, 155 discounting inequality by auence, 234 distribuendum, 191 double awareness, 48 Dumont, L., 3, 5 Dunning, A.J., 292n Durkheim, E., 157n, 158 Dutch College of General Practitioners guidelines, 271, 276n, 284, 297 Dworkin, R., 1, 9, 12, 64, 70, 75, 80, 82, 84{87, 87n, 88{ 90, 98, 99, 106, 112, 113, 116, 136, 142{147, 149, 187, 198n, 274, 274n, 275, 276 economic growth, 69, 203 Edgworth box, 76 eciency, 75, 185, 202, 203, 281, 295, 299
INDEX egalitarianism moderate, 186, 188 radical, 185, 186 Ehrenfest, P., 75, 75n envy-minimisation, see non-envy analysis epistemology, 44, 49 equal equivalent distribution, 197, 226 equalisandum, 2, 9, 10, 12, 135{ 151, 247 equality complexity, see also incompleteness of ordering2, 228, 257, 261 global and local, 224 meaning of ideal, 10, 13, 19, 23, 25, 27, 29, 43, 52, 54{58, 183, 184, 196{257 communal, 190, 191 individualistic, 190{192, 197, 201, 209, 238 of health, 272, 273, 277 of resources, 80, 123, 135, 142, 143, 187, 274{276, 299 of welfare, 82n, 85, 87n, 135, 136, 138, 144, 187, 293 equivalent variation, see measure of health Euclidean distance, see measure of inequality example story Achilles and the Tortoise, 20n, 43n, 159, 161 allocation of potatoes and rice, 78, 99 beetle, 101, 102, 165 beggar, 71, 72 Bethesda, 281 blindness, 185, 208 bribery, 28
321 building a religious monument, 140 cleaning a polluted area, 149, 281 endorphins, 93 football match, 175 gin and vermouth, 95 hangwoman, 82 hearing aids and glasses, 93{ 96, 104, 125, 205, 275, 299 hidden resources, 92 Hindu and Muslim with pork and beef, 234 indolent indigent person, 84n infaust prognosis, 290, 292{ 296, 301 music, 155, 163, 164 painful movement, 144, 147, 206n paralysis, 143 penguin, 33, 131 photography and shing, 144 Plutonians and Saturnians, 165n, 175 private park, 204 Robinson Crusoe, 161, 162 rude, 35, 47 sailor and shoemaker, 69 saving a child, 146, 147 situation of survival, 8, 134, 152, 262 slavery, 26 strawberry and ice, 130 talents, 84 treatment of a trichophyte, 54 treatment vs prevention of unborn children, 193 expected utility, 129 expensive tastes, 137, 139, 140, 144 development of, 142
INDEX
322 exploitation, 144 external e ects, 12, 64, 67, 70, 71, 74, 99, 100, 105, 108, 190, 206n explanation, 72{74, 99, 105 externalisation, 99 externalism, 41 fair
with respect to income, 82{ 84, 90, 97 with respect to wealth, 81{ 84, 90, 97 fair wage, 75n Faulhaber, G., 276n feasibility condition, 86, 87 rst in, rst helped, 281 rst-person perspective, 52, 54, 57, 58, 154, 163, 167, 183 Fishburn, P.C., 249n xed point theorem, 73n Fleurbay, M., 90 Foley, J., 110 Foot, P., 121n foreign aid, 160 Foster, J.E., 219, 220, 233, 237 foundationalism, 44, 45, 52, 57 Frankena, W.K., 20 Frankfurt, H., 8, 185n, 186, 188, 210 fraternity, 3, 10 friendship, 119, 133 fundamental preferences, 99 Gafni, A., 283 Gelasius I, 4
general practitioner, 14, 296, 297, 299, 300 generalisation argument, 23, 27, 56 Gewirth, A., 19, 21, 29, 31, 32, 38, 39, 42, 43, 56
Gide, A., 286
GINI coecient, see measure of inequality Goodwin, B., 282 Gough, J.W., 5n gradient, 250, 251, 255, 256, 259, 264, 267 grading principle, 75n, 108, 195n, 218, 234 Grin, J., 63n Gunning-Schepers, L.J., 273n Hacker, P.M.S., 13, 155, 161{163,
179
Hammond, P.J., 66, 67n, 70, 100,
105 handicap, 82, 84, 97, 281, 292 Hansson, B., 226, 227 Hare, R.M., 3, 9, 19{23, 31, 32, 38{43, 47, 56, 100, 101 Hartog, F., 70, 71, 73 health, 14 determinants, 273, 274 socio-economic related di erences, 187, 273{280, 297, 299 health questionaire, see measure of health health, disease public matter, 274 Hellman, G., 59 Hennipman, P., 70, 71, 73, 103 Hicks, J., 251n history of distribution, 75, 88, 89, 106, 116 Hochman, H.M., 64, 67, 71{73, 105, 190, 206n Hodgkin, K., 290 homogeneity, 14, 219, 233, 234, 236, 240, 243, 247 homogenisation, 5 Honderich, T., 197n
INDEX Hurley, S.L., 1, 10, 18n, 46, 72n,
74n, 102n, 132n, 155, 161n, 165n, 176, 177 Huygen, F.J.A., 298n ideal in Hare, 22, 42 ideal reference, 197, 201{210, 248, 255, 258, 261, 299 identity of events, 47n identity principle, 195n illogical world, 167 illusion, 47 all pervasive, 46, 165 imperialism, 13, 154, 155, 159n, 160, 161, 163{167, 177, 179, 278, 300 incentive, 72, 118, 207 income distribution, 67 in the Netherlands, 215 primary, 68 secondary, 68 tertiary, 68 incompleteness of ordering, 7, 8, 12, 17, 53, 56, 57, 215, 220, 233, 240 explanation of, 228{233, 240, 244 independence, 14, 238, 240, 248, 266 indeterminacy of meaning, 173{ 175 indi erence curve, 76{79 indirect self-defeatingness, 49, 119 individualism realistic, 10, 12, 38, 63, 95, 100, 102, 105, 140, 152, 154, 190, 195, 196, 211, 275 spiritual, 4 volitional, 5{12, 63, 65, 67, 69, 70, 73, 85, 90, 92,
323 94{97, 99, 100, 102, 105, 106, 139, 190, 272, 275, 285, 289, 300 individualistic egalitarianism, see equality meaning individualistic inequality compensation for, 187, 191, 196, 278 measure of, see measure of inequality inner world, 44, 161, 162, 164, 179 insurance, 85, 86 integration of di erent preferences, 67 intermediate representation, 48, 49, 53 internalism, 41 interpersonal comparability, 9, 12, 38, 63{67, 69, 74, 75, 80, 90, 94, 96{98, 103, 123, 154, 156, 173, 194, 211, 245, 275, 288, 289, 300 di erence, 245, 247 level, 245, 247 interpretative methods, 301 intersection of orderings, 228, 230 intrapersonal comparability, 103 invertible principle, 24 iso-inequality curve, 244, 250, 251, 254, 266 iterative predicate, 24 Jackson, F., 8n, 126, 131 Jacobs, H.M., 291 Jephta, 20 Johansson, P.O., 274n, 282, 283 Jones-Lee, M.W., 274n
just prices, 75n, 94, 96, 98 Kagan, S., 8n Kakwani, N., 224n, 258n
INDEX
324 Kant, I., 26, 198 Kind, P., 283, 288
knockdown arguments, 20 knowledge, 44 inferential, 44 infallible, 44, 45, 49, 57 moral, 38, 44, 46, 52, 57 Kohout, P., 82 Kolm, S.Ch., 75, 75n, 76, 78, 94, 97, 99, 108, 110, 123n, 197, 200, 217, 218, 225, 226, 239, 249, 262 Kool, T., 271, 274 Kripke, S.A., 161, 162 Lambert, P.J., 237
law of diminishing returns, 185, 217n, 226 LeGrand, J., 263n, 274n leisure as a good, 80, 84, 90 levelling down objection, 8n, 13, 63n, 107, 184, 188{197, 203, 207, 208, 211 Lewis, D., 32, 165n liberalism real, 147n liberties, 10, 12, 107, 116, 148, 150, 151 to make mistakes, 149 liberty, 3, 10, 64, 70, 75, 105, 107, 137, 185 Little, I.M.D., 100 local de nition, 59, 60 Lock, G.E., 84n, 89n logic classical, 32, 33 rst order predicate, 59, 174 second order, 174 Lorenz curve, 215, 216 intersection, 215, 220, 229, 237, 240
Lorenz dominance, 215{220, 222, 227, 231, 233, 236, 239 lottery, 282 Babylonian, 89, 136 Lucas, J.L., 186{188, 210, 278, 279, 300 luck brute, 85, 143, 146 option, 85, 88, 106, 143, 146, 148 option bad, 106 Mackenbach, J.P., 273 Mackie, J.L., 99n
macro judgements, 71n, 73n manipulability, 104 market, 70, 79, 80, 82, 88, 94, 95, 106, 123, 275, 301 condition monopolyof knowledge, 276 product determination, 276, 297 unpredictable irreversible outcome, 276 of determinants of health, 271, 274, 295 of insurances, 64, 76, 84, 85, 87, 89, 97, 99, 106 Marxism, 202 maxim, 26, 28, 56 maximisation view, 119{122, 128 maximise the minimum, 87n, 289, 293 McDowell, J., 1, 10, 35, 43, 47, 48n, 103, 171 McGuinness, B.F., 167 McKeown, T., 292 mean deviation, see measure of inequality meaning of a good, 68, 69, 72, 73, 86, 90, 94, 98, 105, 136, 290
INDEX measure chickens, 124, 287 measure of equalisanda, 245 measure of health compensation variation, 283 Dutch Sickness Impact Pro le, 291n equivalent variation, 284 health questionnaire, 288, 289, 291, 300 healthy years equivalent, 279, 283, 285, 287, 289, 291, 300 life year adjusted repayments, 283, 285, 287, 289, 291, 300 multidimensional,291, 296, 300 quality adjusted life years, 279, 282, 285, 287, 289, 291, 300 measure of inequality, 13, 67, 154, 213, 243{265 coecient of variation, 224, 233, 240, 257{259, 261 Euclidean distance, 244, 251, 252, 256, 261 GINI coecient, 221{223, 227, 233 maximin, 251, see also maximise the minimum251 mean deviation, 214 normative, 225, 227, 233, 240 overall equalisandum, 1, 11, 125n, 263 range, 213 Theil measure, 224, 240 total deviation from the reference, 251, 252 measure problem bandwith versus delitiy, 291 measuring length, 34 Mehrez, A., 283
325 micro judgements, 71n Milner, M., 276n minimum view, 188, 210 Mishan, E.J., 72 monetary value, 123 monotone increasing, 110, 216 monotone increasing transformation, 237, 248, 251 monotonicity with respect to population, 90, 95, 106 with respect to resources, 90, 95, 96, 106 Mooney, G.H., 274n, 287 moral con ict, 6, 11, 20, 30, 36, 56, 126, 128, 129, 131, 132, 135, 151, 231 moral particularism, 10, 11, 18, 36, 53{55, 126, 132{134, 151, 153, 157, 231, 236, 261, 280n, 290 moral universalism, 5{11, 17{36, 53{56, 69n, 126, 132{134, 156, 261, 272, 285, 289, 300 moral value monism, 5{12, 69n, 107, 115{135, 151, 191, 272, 285, 287, 289, 300 moral value pluralism, 6, 10, 38, 115, 120, 126{136, 148, 151, 152, 154, 184, 186, 187, 211, 236, 258, 262, 278, 290 moral way of looking, 19, 43, 53, 55{58, 134 More. T., 23 Moulin, H., 95, 276n Munchhausen syndrome, 275 Nagel, T., 2, 7, 138n, 163, 199,
199n
Ng, Y.K., 245, 246
326 non-entailment, 169, 170, 172 non-envy analysis, 12, 64, 67, 74{ 100, 104n, 105, 123, 205, 218, 274{276, 299 envy-minimisation, 91, 92, 97, 98, 106 undominated diversity, 91, 92, 97, 98, 106, 293n non-envy boundary, 77, 78 non-identity problem, 192{194, 196, 211 non-monotone reasoning, 32, 33, 46, 134n non-moral terms, 8, 17, 18, 31, 32, 35, 47, 169, 170, 179 non-transferable goods, 84, 90, 92{ 94, 97, 98, 105, 106, 204, 208, 209, 211, 272 Norman, R., 156, 157n, 158, 291 Nozick, R., 123n, 160n Nussbaum, M.C., 288 O'Neill, O., 6, 19, 21, 26{28, 38,
40, 43, 56, 160, 198 o ensive tastes, 137, 140, 144 ordinal ordering, 233, 236, 237, 247, 253, 258 original position, 86, 98, 112 Osmani, S.R., 227 other minds, 100, 102{104, 106 overlapping individuals, 245 Pareto ecient egalitarian equivalent allocation, 83, 84, 90, 97 Pareto optimal, see Pareto principle Pareto principle, 64, 67, 70, 75, 77, 79, 82, 100, 102, 105, 189, 190, 196, 217, 234 Par t, D., 49, 119, 129n, 188, 192n, 193, 195, 195n, 208{211
INDEX paternalism, 101, 287 pattern principle, 88, 89, 99, 136, 152 Pazner, E., 64, 80, 83, 83n, 85, 90, 98, 111 Peacock, C., 49 Pears, D.F., 167 perception causal theory, 47, 53 moral, 47, 49, 51, 57, 128 perfectionism, 119, 138, 185, 202 permutation, 75, 108 person a ectingness, 103n, 107, 192{ 197, 211, 217, 218 Pettit, P., 129, 131, 132, 287 Pigou Dalton principle, 13, 219, 233, 236, 238, 240, 243, 250, 258 population principle, 14, 219, 233, 235 positional objectivity, 51 prescriptivism, 39, 42, 56 prices just, 275 prima facie, 131, 132 primary goods, 116, 139 principle of charity, 37, 51, 103, 133, 134, 156, 163{165, 173, 179, 194, 199, 272, 294, 296, 301 principle of diminishing transfers, 223, 224, 227, 233, 236, 258 principle of independence of no complaints, 14, 236, 237, 240, 243, 253 principle of personal good, 189, 190, 192, 196, 217 priority view, 188, 210, 211 private language argument, 44, 45, 53, 102, 106, 156, 161 pro tanto, 132n
INDEX property right, 94 propositional contexts, 130 prudential reason, 38, 40, 42, 53, 57 punishment retributive view, 192n Putnam, H., 51n, 196 quasi-ordering, see incompleteness of ordering quasi-realism, 13, 155, 167, 173, 174, 179 Queen's speech, 200 Quine, W.V.O., 52n radical interpretation view, 19, 35n, 37, 40, 41, 43, 46, 51, 53, 57, 74, 103, 133, 140, 155, 156, 160, 163{165, 167, 171{175 Rakowksi, E., 198n range, see measure of inequality ratio scale, 247, 251, 254, 264, 282 rationality anchor of, 161 Rawls, J., 1, 86, 116, 118, 136, 138{141, 207 realism direct, 47, 49, 52, 57 indirect, 47, 48 moral, 1, 6, 10, 11, 18, 19, 38{54, 56, 57, 63, 65, 67, 102, 105, 106, 141, 147, 148n, 152{180, 294, 300 reason concept of, 31, 32, 34, 36, 57 reasoning common practice of, 159, 160 recursion, 73n redistribution better for the bettero , 204
327 redistribution worse for the worseo , 204, 207 relativism, 6, 160, 278 representation, 49 Rescher, N., 52n, 196 responsibility, 12, 65, 85, 89, 98, 99, 107, 116, 136, 139{ 141, 143{151, 275, 276, 281 Robbins, L., 68, 69, 104 Roberts, K.W.S., 247n Robinson, J., 94 Rodgers, J.D., 64, 67, 71{73, 105, 190, 206n Roemer, J.E., 1, 9, 64, 86, 87, 88n, 90, 92{94, 97, 98, 106, 112 Rosanvallon, P., 4, 5 Ross, A., 94 Royal Dutch Medical Association, 298 rule following, 34, 36, 48, 57, 104, 106, 155, 161{163, 179, 300 communityview, 161, 162, 164, 179, 278 Rutten, G., 271, 276n, 284n, 297 Sabine, G., 4 Samuelson, P.A., 215 Sartre, J.P., 22
satis cing theory, 119{122, 127 Savage, L.J., 287n Scanlon, T.M., 94n, 116, 136, 140, 141, 145, 149, 200, 281 scepticism, 100 moral, 21, 22, 36, 173, 177 Schayk, C.P., 291 Schiermonnikoog, 33, 37, 276n Schmeidler, D., 64, 80, 83, 83n, 85, 90, 98, 111 Schur concavity, 216
INDEX
328 Scruton, R., 157, 158
semi-connectedness, 245 Sen, A.K., 2, 6{9, 14, 51n, 68, 108, 109n, 186, 187, 195n, 200, 215n, 217, 222, 224, 224n, 226, 228, 234, 240, 244, 245n, 258, 259, 261, 263, 263n, 264, 284 sensation, 49 sense data, 44, 45, 48, 53, 156, 166 separability, 8n, 19, 33, 67n, 132, 239, 262, 285{287, 289, 300 separate additive form, 248, 250, 262, 264 Shorrocks, A.F., 223, 237, 266n similarity concept of, 32{34, 36, 43, 48, 57 Singer, M.G., 19, 21, 23{25, 28n, 31, 32, 38, 39, 42, 56 situation of survival, 262 Sobel, J.S., 291 society cement of, 200 solipsism, 40, 46, 46n, 100, 133, 166, 245 communal, 161 practical, 199 speci city, 14, 206n, 239, 241, 261{ 263 Stalnaker, R., 32 standard, 32, 36, 47 Starlett, D., 217 Stocker, M., 7, 115, 121, 129n Strawson, G., 146 Strawson, P.F., 140n Streiner, D.L., 291 Stronks, K., 273, 273n subjectivism, 278 Sundrum, R.M., 68
supererogation, 119, 120 superfair, 74n supervenience, 31, 32, 34{36, 47, 56, 57, 59, 155, 167, 169{ 172, 179 Suppes, P., 75n, 108, 195n, 218, 234 sure thing principle, 130 symmetry, 14, 75, 218, 219, 233, 234, 236, 240, 243, 250 Takayama, A., 110 Tarski, A., 156
taste model, 63n tax, 88, 89, 106 on capability, 82 Temkin, L.S., 2, 14, 190n, 192n, 197n, 228, 230, 231, 233n, 234, 235, 240, 244, 259, 261, 263, 264 Temkin's repellant conclusion, 235, 236, 262n Temkin's series, 229, 230, 244, 259, 261 rst better, then worse, 259, 260 the absolute view, 51n Theil measure, see measure of inequality theorem of Arrow, 55n, 66, 105, 121, 177 theorem of Lowenheim Skolem, 173{ 175, 179, 180 theorem of Leontief, 249 third-person perspective, 52, 54, 57, 58, 163, 183 Thomas, S., 271, 276n, 284n, 297 Thomson, W., 104n Thorson, T.L., 4 Tinbergen, J., 1, 68, 69, 71, 75, 82n, 104
INDEX total deviation from the reference, see measure of inequality Touw-Otten, T., 291 Townsend, P., 187, 273n trade surplus, 26 traditional framework, 1, 3{9, 63, 107, 115, 152, 153, 183, 271 transcendental argument, 29 translation intransitivity, 175 treat like cases alike, 54{56 uncertainty hidden knowledge, 279 uncertainty in comparisons, 231, 240, 254 undominated diversity, see nonenvy analysis universalisations, 99n utilitarianism, 50, 137, 138, 195, 202, 217n, 226, 293 Vallentyne, P., 263n
value expressing one's individuality, 207 indexed, 129, 132, 186n possibility of a normal life, 147, 149 to be performed, 120, 122, 130, 133 value of life economic, snee also measure of health274 Van Benthem, J., 59n Van Es, J.C., 290, 292 Van Parijs, P., 12, 87n, 91, 92, 147n, 263n, 293 Van Praag, B.M.S., 288 Van Wijck, P., 71, 71n
329 Varian, H.R., 1, 64, 67, 81, 82,
82n, 84, 90, 97, 98, 110, 254 Vermeersch, E., 197n view of enough, 188, 210 Walrasian equilibrium, 82n, 110 Walzer, M., 158n, 187n, 277, 278, 299 weighing values or reasons, 11, 134n, 272, 273, 295 welfare, 123 welfare function, 216, 226 welfarism, 284 Wiggins, D., 117, 125, 126, 200 William of Ockham, 4 Williams, B., 35n, 49, 50, 52, 57, 128, 160n, 161, 163, 199 Winch, P., 159n Winters, J.C., 291 Wittgenstein, L., 1, 7, 10, 11, 13, 19, 34, 36, 39n, 40, 44, 45, 46n, 48, 53, 57, 67, 101{104, 106, 115, 153, 155, 161, 163, 166, 167, 179, 278, 288, 290, 300, 301 Young, H.P., 95
330
INDEX
Samenvatting Gelijkheid; een moreel realistisch gezichtspunt
(Op weg naar een eenvoudige maat voor ongelijkheid)
In deze studie wordt een ideaal van gelijkheid geformuleerd binnen een moreel realistisch kader, dat wil zeggen een kader waarbinnen morele oordelen zoals "die handeling is goed" of "die verdeling is onrechtvaardig" waar of onwaar zijn. Dit moreel realistische kader wordt gekarakteriseerd door moreel particularisme, realistisch individualisme en waardenpluralisme. Het is gebaseerd op de ideeen van Wittgenstein, Davidson, McDowell, Dancy en Hurley. Traditioneel gezien, is het idee van gelijkheid ontwikkeld binnen een kader dat gekenmerkt wordt door: het moreel universalisme, het idee dat regels noodzakelijk zijn voor morele oordelen en de basis vormen voor morele oordelen. het wilsgeorienteerde individualisme, het idee dat een toestand in de wereld beoordeeld moet worden op grond van de mate waarin de wensen van individuen vervuld zijn waarbij wordt aangenomen dat deze wensen ontstaan in de wil van individuen en een onbeperkte verscheidenheid laten zien. het moreel waardenmonisme, het idee dat een waarde de belangrijkste is waarbij alles als drager van deze ene waarde wordt beschouwd. Hoewel het tegenovergestelde van deze drie traditionele achtergrondassumpties namelijk het moreel particularisme, het realistisch individualisme en het moreel waardenpluralisme, verenigd in het moreel realisme, beter overeenkomt met het denken over morele oordelen, ontbreekt een formulering van een ideaal van gelijkheid binnen dit moreel realisme. Dit geI
II
SAMENVATTING
brek is des te opmerkelijker daar de traditionele achtergrondassumpties een zinvolle formulering van een ideaal van gelijkheid verhinderen. Deze studie laat zien hoe een ideaal van gelijkheid geformuleerd kan worden binnen een moreel realistisch kader. De vraag hoe en door wie ongelijkheid ontstaat en in stand wordt gehouden en de daarmee samenhangende vraag over hoe meer gelijkheid bewerkstelligd kan worden, behoren niet tot het onderwerp van deze studie maar kunnen dienen als onderwerp van een vervolg op deze studie. Er zijn drie vragen aan de hand waarvan een ideaal van gelijkheid wordt ontwikkeld, te weten: 1. Wat is het aspect dat gelijk verdeeld zou moeten zijn? 2. Waarom is gelijkheid wenselijk? 3. Hoe kan de ordening van situaties naar de ernst van ongelijkheid gekarakteriseerd worden? Deze worden als volgt beantwoord binnen een moreel realistisch kader. Het antwoord op de eerste vraag, de vraag naar het equalisandum, dat wil zeggen datgene dat gelijk verdeeld moet zijn, is dat er niet een equalisandum is maar dat er meerdere zijn. Datgene dat waardevol is, is daarom ook waardevol om gelijk verdeeld te worden. Elke waarde heeft zijn eigen belangrijkheid afhankelijk van het verdelingsprobleem. Het waardenpluralisme en het moreel particularisme maken dit antwoord aannemelijk. Het antwoord op de tweede vraag betre ende de wenselijkheid van gelijkheid dan wel de onwenselijkheid van ongelijkheid is, dat ongelijkheid onwenselijk wordt geacht, niet omdat mensen minder hebben dan anderen, maar omdat sommigen minder hebben dan mogelijk is voor iedereen als de goederen anders verdeeld waren. Gelijkheid is wenselijk omdat ontkennen van de basis ervan, namelijk dat datgene dat waardevol is voor iemand zelf ook waardevol is voor een ander, praktisch solipsisme zou betekenen. Hoewel dit solipsisme niet onmogelijk is, is de wenselijkheid ervan niet inzichtelijk te maken. Aan dit antwoord ligt het realistisch individualisme ten grondslag. Het antwoord op de derde vraag luidt dat de ordening naar de ernst van ongelijkheid in een verdelingsprobleem kan worden weergegeven met een volledige ordening volgens een index die is gebaseerd op een summatie van het verschil tussen wat mensen zouden kunnen hebben als er een gelijke verdeling was geweest en wat mensen feitelijk hebben, voor zover ze minder hebben dan mogelijk is bij een gelijke verdeling. Het realistisch individualisme wordt gerepresenteerd door de summatie over de verschillen. In de inleiding wordt uitgelegd op welke wijze de drie assumpties: het moreel universalisme, het wilsgeorienteerde individualisme en het moreel
SAMENVATTING
III
waardenmonisme, als traditionele achtergrondassumpties kunnen worden beschouwd van een ideaal van gelijkheid en hoe deze de ontwikkeling van een ideaal van gelijkheid blokkeren. Het moreel universalisme blokkeert een volledige ordening van situaties volgens een ideaal van gelijkheid. Als een situatie gekarakteriseerd wordt middels een verdeling A van een bepaald goed, dan is volgens het moreel universalisme die situatie slechter qua gelijkheid dan een situatie met een andere verdeling B als ook andere situaties gekarakteriseerd door diezelfde verdeling A moreel slechter zijn dan die welke worden gekarakteriseerd door B. Maar, zoals Sen uitlegt, niet alleen een verdeling bepaalt hoe slecht die verdeling is qua gelijkheid, omdat ook andere idealen een rol spelen. Een verdeling kan qua gelijkheid slechter zijn dan een andere, bijvoorbeeld in het geval dat een persoon alles krijgt en de ander niets, vergeleken met een verdeling waarbij ieder evenveel van de goederen krijgt, terwijl het in een andere situatie zo kan zijn dat de verdeling waarbij een alles krijgt en de ander niets, niet slechter is qua gelijkheid, bijvoorbeeld in een situatie waarin bij een ongelijke verdeling van de goederen in ieder geval iemand overleeft en bij gelijke verdeling niemand. Vanwege het moreel universalisme geeft het ideaal van gelijkheid dan ook geen volledige ordening van verdelingen in verdelingsproblemen. Het wilsgeorienteerde individualisme verhindert de ontwikkeling van een ideaal van gelijkheid omdat niet duidelijk is wanneer er nu sprake is van gelijkheid, want hoe moet de wensbevrediging van de een vergeleken worden met die van de ander? Binnen het wilsgeorienteerde individualisme, waarin de wensen van individuen hun oorsprong vinden in de individuele wil, is zo'n vergelijking tussen personen niet mogelijk. Het wilsgeorienteerde individualisme brengt het probleem van interpersoonlijke vergelijking met zich mee. Ook het moreel waardenmonisme blokkeert de ontwikkeling van een ideaal van gelijkheid. Dit waardenmonisme vormt namelijk de basis voor de discussie over wat nu eigenlijk gelijk verdeeld moet worden. Het ideaal van gelijkheid dat inhoudt dat welzijn gelijk verdeeld moet zijn, wordt tegengesproken door het ideaal van gelijkheid zelf maar dan met een ander aspect dat gelijk verdeeld moet worden zoals bijvoorbeeld hulpbronnen. Tegenover elke formulering van het ideaal van gelijkheid kan een ander geformuleerd worden met een ander equalisandum. Kortom, de traditionele assumpties van het ideaal van gelijkheid verhinderen de ontwikkeling van een ideaal van gelijkheid. Een nieuw kader voor de ontwikkeling van een ideaal van gelijkheid ligt dan ook voor de hand. In het eerste deel van deze studie, bestaande uit de hoofdstukken 2, 3, 4 en 5, wordt dit nieuwe kader, het moreel realisme, ontwikkeld als alternatief voor de traditionele achtergrondassumpties van een ideaal van gelijkheid. In het tweede deel, bestaande uit de hoofdstukken 6, 7 en 8,
IV
SAMENVATTING
wordt een ideaal van gelijkheid ontwikkeld zodanig dat het de belangrijkste argumenten tegen een ideaal van gelijkheid kan weerleggen, met de bijbehorende ordening die de mate van ernst van de ongelijkheid weergeeft. Ten slotte wordt in het derde deel, bestaande uit hoofdstuk 9, besproken wat het geformuleerde ideaal van gelijkheid binnen een moreel realistisch kader betekent voor verdelingsvragen op het terrein van gezondheid en gezondheidszorg. In hoofdstuk 2, wordt de eerste van de traditionele achtergrondassumpties voor gelijkheid, namelijk het moreel universalisme, besproken. Binnen het moreel universalisme worden regels beschouwd als noodzakelijk voor morele oordelen. Ze worden beschouwd als de redenen voor de juistheid of onjuistheid van morele oordelen, ook voor de morele oordelen betreffende gelijkheid. Er wordt aangetoond dat de belangrijkste reden voor het moreel universalisme namelijk dat het morele oordelen kan behoeden voor het scepticisme, niet toereikend is. Moreel universalisme lost namelijk de morele problemen en con icten niet op, hetgeen er wel van werd verwacht. Met behulp van Wittgensteins ideeen over "het volgen van een regel" en "is gelijk aan" zoals hij ze liet zien in zijn Filoso sche Onderzoekingen, die verder zijn uitgewerkt door Davidson, wordt beargumenteerd dat morele oordelen, waaronder ook die over gelijkheid, waar of onwaar zijn op grond van de toestand in deze wereld en niet op grond van het feit dat ze in overeenstemming zijn met een verzameling morele regels of principes. Het gevolg is dat niet verwacht kan worden dat een ideaal van gelijkheid afgeleid moet worden uit een dergelijke verzameling regels of principes. Een ideaal van gelijkheid zal gerelateerd zijn aan een bepaalde wijze van observeren namelijk de wijze waarop iemand zichzelf ziet, als een onder vele anderen. Vanwege het moreel particularisme ondermijnt het moreel realisme het argument voor de onvolledigheid van de ordening die de mate van de ernst van de ongelijkheid weergeeft. Het moreel realisme opent dan ook de weg naar een volledige ordening. In hoofdstuk 3 wordt de tweede van de traditionele achtergrondassumpties, het wilsgeorienteerde individualisme besproken. Het wilsgeorienteerde individualisme houdt in dat situaties beoordeeld moeten worden op grond van de mate waarin de wensen van individuen vervuld worden waarbij wordt aangenomen dat deze wensen ontstaan in de wil van individuen en een onbeperkte verscheidenheid laten zien. Dit wilsgeorienteerde individualisme brengt het probleem met zich mee van de interpersoonlijke vergelijking van de mate van vervulling van die wensen. Algemener gezegd, het wilsgeorienteerde individualisme heeft te kampen met het probleem van de interpersoonlijke vergelijking van wat goederen voor mensen betekenen. Het wordt duidelijk gemaakt dat de twee manieren om dit probleem van interpersoonlijke vergelijking te omzeilen, namelijk het erkennen van externe
SAMENVATTING
V
e ecten van bezit van goederen en het streven naar afwezigheid van afgunst, geen oplossing geven voor de ontwikkeling van een ideaal van gelijkheid. Het erkennen namelijk van zogenaamde externe e ecten op grond waarvan men een herverdeling zou kunnen rechtvaardigen, zoals bijvoorbeeld Hochman en Rodgers voorstellen en hier te lande wordt overgenomen door Hennipman en Hartog, is alleen aannemelijk als men de mogelijkheid van interpersoonlijke vergelijking veronderstelt. De analyses die gebaseerd zijn op het idee dat gelijkheid afwezigheid van afgunst betekent, zoals bijvoorbeeld voorgesteld door Tinbergen, Varian en recentelijk ook door Van Parijs, blijken evenmin toereikend. In de bespreking van het voorstel van Dworkin dat ook is gebaseerd op afwezigheid van afgunst, komt een probleem aan het licht dat gerelateerd is aan verantwoordelijkheid voor het eigen handelen. Het betreft het probleem betre ende de mogelijkheid tot een beoordeling van een verdeling zonder dat de wijze van ontstaan van die verdeling erbij wordt betrokken. Een oplossing hiervoor wordt gegeven in hoofdstuk 4 als het equalisandum wordt besproken. In hoofdstuk 3 wordt het realistisch individualisme, een aspect van het moreel realisme, gentroduceerd. Binnen dit realistisch individualisme worden wensen genterpreteerd op basis van wat wenselijk en nastrevenswaardig is. Interpersoonlijke vergelijkingen die noodzakelijk zijn voor morele oordelen betre ende gelijkheid zijn binnen dit realistisch individualisme geen probleem. Het blijkt dat het probleem van interpersoonlijke vergelijking samenhangt met het moreel universalisme. In hoofdstuk 4 wordt de derde traditionele achtergrondassumptie, het moreel waardenmonisme, besproken. De redenen voor het aannemen van moreel waardenmonisme worden ontzenuwd. Het wordt duidelijk gemaakt dat het waardenmonisme geen mogelijkheidheeft om het bestaan van keuzen en de afwegingen over die keuzen te verklaren. Tevens is er binnen het waardenmonisme geen ruimte voor een ideaal van gelijkheid, omdat de daarvoor noodzakelijk onderscheiding van de waarde van wat verdeeld wordt en de waarde van een verdeling niet gemaakt kan worden. Waardenpluralisme, zo wordt duidelijk gemaakt, is een beter alternatief voor het ontwikkelen van een ideaal van gelijkheid. De vraag namelijk naar: "Wat is het aspect dat gelijk verdeeld moet zijn?", of anders: "Wat is het juiste equalisandum?" kan met behulp van het waardenpluralisme worden opgelost. In plaats van een equalisandum is namelijk een veelheid van equalisanda aannemelijk. Het wordt uitgelegd dat equalisanda in het algemeen datgene betre en dat goederen waardevol maakt voor mensen en dat ze de mogelijkheden betre en om van waardevolle handelingen te genieten of om ze uit te voeren zonder de nadelige consequenties hiervan te hoeven dragen. Deze mogelijkheden worden "vrijheden" genoemd. Er wordt duidelijk gemaakt dat het probleem dat aan het licht kwam in het vorige hoofdstuk bij de bespreking van Dworkins voorstel betre ende de rol van verantwoordelijk-
VI
SAMENVATTING
heid voor het eigen handelen voor een evaluatie van een verdeling, wordt opgelost. Het wordt namelijk aannemelijk gemaakt dat verantwoordelijkheid niet gencorporeerd dient te worden in het equalisandum, in tegenstelling tot wat wordt voorgesteld door bijvoorbeeld Dworkin, Cohen en Van Parijs. In elk van de hoofdstukken 2,3 en 4 werd een aspect van het moreel realisme dat gebaseerd is op de Wittgenstein-Davidson benadering van taal en interpretatie, gentroduceerd. Deze hoofdstukken betro en het moreel particularisme versus het moreel universalisme, het realistisch individualisme versus het wilsgeorienteerde individualisme en het moreel waardenpluralisme versus het moreel waardenmonisme. Nadat duidelijk is geworden in die hoofdstukken wat het moreel realisme kan betekenen voor de formulering van een ideaal van gelijkheid is het mogelijk om de belangrijkste tegenwerpingen tegen het moreel realistische kader voor gelijkheid te beoordelen. Dit is het onderwerp van hoofdstuk 5. Eerst wordt de tegenwerping weerlegd dat het moreel realisme dat conservatieve tendensen heeft waardoor het geen redelijke verandering toe kan laten. Daarna wordt het bezwaar weerlegd dat het moreel realisme imperialistisch is, waarmee wordt bedoeld dat iets niet als waardevol herkend kan worden voor een ander als het niet waardevol is voor de beoordelaar. De weerlegging van dit bezwaar van imperialisme is gebaseerd op de Baker-Hacker interpretatie van Wittgensteins ideeen over "het volgen van een regel", als alternatief voor een meer sociologische interpretatie zoals die van bij voorbeeld Winch of Kripke. Vervolgens wordt de stelling weerlegd dat de morele fenomenen beter door het quasirealisme zoals gentroduceerd door Blackburn, verklaard kunnen worden dan door het moreel realisme. Tenslotte wordt de tegenwerping besproken dat de verschillen in morele oordelen niet verdwijnen door het moreel realisme. In het tweede deel, dat bestaat uit de hoofdstukken 6, 7 en 8, wordt nader bepaald wat gelijkheid als ideaal inhoudt. Daarbij wordt ook een bijbehorende index voor ongelijkheid voorgesteld. In hoofdstuk 6 wordt een ideaal van gelijkheid ontwikkeld zodanig, dat de argumenten die veelal ingebracht worden tegen gelijkheid, kunnen worden beantwoord. Dit wordt mogelijk door het alternatieve moreel realistische kader, dat verder gekarakteriseerd wordt door moreel particularisme, realistisch individualisme en moreel waardenpluralisme. Het moreel waardenpluralisme geeft de mogelijkheid van de erkenning dat gelijkheid een ideaal is binnen een verzameling van meerdere relevante idealen. Veel kritiek betreft namelijk de opvatting dat gelijkheid het enige ideaal is. Tevens maakt het pluralisme tezamen met het particularisme mogelijk dat er verschillende equalisanda erkend worden en er niet gestreefd hoeft te worden naar de ontwikkeling van een equalisandum. Het realistisch individualisme maakt mogelijk dat het zogenaamde
SAMENVATTING
VII
argument van het aftoppen van alles wat boven het maaiveld uitkomt, de zogenaamde "levelling down objection", niet van toepassing is. Het centrale idee van het ideaal van gelijkheid is dat ongelijkheid onwenselijk is omdat sommige mensen slechter af zijn dan voor iedereen mogelijk zou zijn als er een andere verdeling was geweest. Ongelijkheid is dus niet onwenselijk omdat sommigen beter af zijn. Dat gelijkheid in principe wenselijk is, dat wil zeggen dat iedereen zo goed af is als mogelijk is voor iedereen, is inzichtelijk te maken met de basis van het moreel realisme zoals dat in deze studie is bedoeld. Het ideaal wordt niet afgeleid uit een verzameling regels maar is gebaseerd op een morele wijze van zien waarbij men zichzelf als een onder vele anderen ziet. In hoofdstuk 7 wordt aan de hand van een bespreking van bekende voorgestelde maten voor ongelijkheid de eigenschappen vastgesteld waaraan een maat voor ongelijkheid moet voldoen. Deze zijn: 1. Het Pigou-Dalton principe, althans een variant ervan: dit houdt in dat een overdracht van een goed van iemand die slechter af is naar iemand die beter af is, de ernst van de ongelijkheid doet toenemen, mits degene die slechter af is, slechter af is dan iedereen had kunnen zijn. 2. Symmetrie beperkt tot equalisanda: dit houdt in dat een verwisseling van wie wat krijgt, of anders gezegd een permutatie van de verdeling, geen e ect heeft op de ernst van de ongelijkheid, mits deze het equalisandum betreft. 3. Beperkte homogeniteit: dit houdt in dat een verandering van eenheid waarin het equalisandum gemeten wordt, bijvoorbeeld centen of guldens, geen invloed heeft op de ernst van de ongelijkheid. 4. Het principe van onafhankelijkheid van mensen zonder klacht over ongelijkheid: dit houdt in dat de ernst van de ongelijkheid niet verandert als er aan de groep mensen personen worden toegevoegd die niet slechter af zijn dan in de ideale referentie. 5. Zwakke separabiliteit: dit houdt in dat de ordening van situaties naar de ernst van ongelijkheid gebaseerd op de verdeling van een subgroep niet in zijn tegendeel kan omslaan door het nivo van de individuen die niet behoren tot de subgroep. In formule: (~xg ;~x g ) I (I )(~x g ;~x g ) () (~xg ; ~y g ) I (I )(~x g ; ~y g ), hetgeen sterke separabiliteit kan worden als de referentie die representeert hoe goed iedereen af had kunnen zijn, ge xeerd blijft. :
:
0
:
0
:
SAMENVATTING
VIII
De eigenschap van decompositie of additiviteit, die inhoudt dat de index van de ernst van de ongelijkheid bestaat uit een som van de indices die de ernst van de ongelijkheid in de subgroepen representeren, blijkt niet acceptabel te zijn. Eveneens ontbreekt het populatie principe, dat inhoudt dat de ernst van de ongelijkheid niet verandert als er aan een groep een andere groep met dezelfde ongelijke verdeling wordt toegevoegd. Dit principe is vervangen door het principe van onafhankelijkheid van mensen zonder klacht over ongelijkheid. In hoofdstuk 8 wordt op grond van de eigenschappen die besproken zijn in hoofdstuk 7 een index geconstrueerd die in een verdelingsprobleem een volledige ordening naar de ernst van ongelijkheid representeert. Hiermee is aangetoond dat in tegenstelling tot wat Sen en Temkin beweren, een volledige ordening naar de ernst van ongelijkheid mogelijk is, die kan worden weergegeven door een index. De index is vanwege de separabiliteit, P dat het individualisme weergeeft, van een separatieve additieve vorm , een theorema van Leontief maakt dit duidelijk, waarvan de termen vanwege het Pigou-Dalton principe een convexe functie f is, die vanwege de symmetrie voor ieder individu gelijk is, met als argument het verschil tussen wat mensen i hebben, xi , en wat ze zouden hebben als er een verdeling zou zijn waarin iedereen gelijk af zou zijn, xref P, voor zover ze niet beter af zijn dan in die referentie, minfxi ; xref g, dus ni=1 f(xref , minfxi ; xref g). Omdat het equalisandum kan worden weergegeven op een ratio schaal op grond van een argument van Ng over vergelijkbaarheid van verschillen, namelijk dat interpersoonlijke vergelijkbaarheid van nivo ook interpersoonlijke vergelijkbaarheid van verschillen met zich meebrengt, is de functie f een machtsfunctie, een argument ontleend aan Hicks maakt dat duidelijk. Aangezien een van deze functies, namelijk die met de macht 2, de eigenschap heeft dat de richting van de maximale verandering, de gradient, gericht is op de ideale referentie situatie is de index die wordt voorgesteld:
X(x
I(~x) = M(
n
ref
i=1
, minfxref ; xig)2 )
waarin M een monotoon stijgende functie is. Het betreft tenslotte een ordinale schaal van ernst van ongelijkheid. Vervolgens worden de argumenten van Sen en Temkin tegen een volledige ordening besproken. Er wordt geconcludeerd dat een volledige ordening en de daarbij ontwikkelde relatief simpele index de verschillende oordelen over de groep situaties die Temkin presenteert om juist de complexiteit van gelijkheid aan te geven, beter verklaart dan de complexiteit van gelijkheid die Temkin aanvoert als verklaring voor die verschillende oordelen.
SAMENVATTING
IX
In hoofdstuk 9 volgt een bespreking van de gevolgen van het ideaal van gelijkheid zoals dat is ontwikkeld in deze studie binnen het moreel realistische kader, voor het terrein van gezondheid en gezondheidszorg. De gangbare discussies over gelijkheid en gezondheid worden becommentarieerd op grond van het ontwikkelde ideaal van gelijkheid. Verdelingsvraagstukken in de gezondheidszorg moeten worden beantwoord op grond van particularistische oordelen betre ende de behoefte aan zorg. Tenslotte wordt aangegeven dat dit juist goed kan passen binnen de werkwijze van huisartsen.
